From Particular to General Linguistics: Selected Essays, 1965-1978 (Studies in Language Companion Series, V. 3)

FROM GENERAL TO PARTICULAR LINGUISTICS STUDIES IN LANGUAGE COMPANION SERIES (SLCS) The SLCS series has been establish...

Author: Yakov Malkiel

38 downloads 1536 Views 61MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

FROM GENERAL TO PARTICULAR LINGUISTICS

STUDIES IN LANGUAGE COMPANION SERIES (SLCS) The SLCS series has been established as a companion series to STUDIES IN LANGUAGE, International Journal, sponsored by the Foundation "Foundations of Language". Series Editors:

John W. M. Verhaar Gonzaga University Spokane, WA

Werner Abraham & University of Groningen The Netherlands

Volume 3 Yakov Malkiel From Particular to General Linguistics

FROM PARTICULAR TO GENERAL LINGUISTICS Selected Essays 1965-1978 by

YAKOV MALKIEL University of California at Berkeley

JOHN BENJAMINS PUBLISHING COMPANY Amsterdam/Philadelphia 1983

© Copyright 1983 - John Benjamins B.V. ISSN 0165 7763 / ISBN 90 272 3002 1 No part of this book may be reproduced in any form, by print, photoprint, microfilm or any other means, without written permission from the publisher.

AUTHOR'S ACKNOWLEDGMENTS

It gives me keen pleasure to thank here various parties without whose help and understanding this exacting project might not have crystallized. The publishers, John and Claire Benjamins, have probably shown greater patience with me, over a period of at least four years punctuated by crises, than I actually deserved. Acting as talent scout, Professor J. W. M. Verhaar obligingly established the initial contact with Amsterdam and, being an indefatigable world traveler, visited me twice in Berkeley instead of forcing me to make a pilgrimage to Jakarta. Among several young ladies who cheerfully assumed the responsibility for digital operations, Angela lovino and Anne Donaker underwent with a smile the torture of typing indexes .. . My friends and colleagues Use Lehiste, J. P. Maher, and Edward F. Tuttle authorized me to include their meaty spontaneous comments made during the discus sion period, on the occasion of the oral delivery, at a meeting, of one paper here in cluded. Without a single exception, the editors or business managers or society secretary-treasurers of those journals or transaction volumes in which the pieces here assembled had originally appeared, obligingly gave the new publisher and my self the requisite authorization to reprint the items, sometimes with minor changes and practically always with a small quota of necessary corrections. And this short, selective list certainly does not exhaust the full measure of my indebtedness to the many who in the first place encouraged or helped me to write — five, ten, or fifteen years ago — the articles here collected, or who made it possible for me to channel sufficient time, energy, and imagination into my research to produce papers which, in the opinion of several competent and responsible persons, deserved the honor of original publication and the even greater honor of inclusion in a miscellany of this size and slant. Berkeley, CA, March 9, 1983

TABLE OF CONTENTS

Author's Acknowledgments

v

Table of Contents

vii

Guide to Abbreviations

xi

FROM PARTICULAR TO GENERAL LINGUISTICS Introduction

3

A. GENETIC LINGUISTICS 1. Linguistics as a Genetic Science Presidential Address, Linguistic Society of America, December 1965). Language, XLIII:1 (1967), 223-245.

23

B. HISTORY OF LINGUISTICS 2. History and Histories of Linguistics

49

Romance Philology, XXII:4 (1969), 530-566, 573f. C. CLUES AS TO DATING 3. Range of Variation as a Clue to Dating

87

Romance Philology, XXI:4 (1968), 463-501 ("Emanuel S. Georges Memorial"). D. CONVERGENCE AND DIVERGENCE 4. Factors in the Unity of ROMANIA

129

Romance Notes, XVIII:2 (1977), 263-271 E. DIFFUSION 5. Lexical Borrowing in the Romance Languages

137

Language, LI:4 (1975), 962-976. F. LANGUAGE VS. THE REAL WORLD 6. Gender, Sex, and Size, as Reflected in the Romance Languages Studies in Romance Linguistics (Michigan Meeting), ed. M.P. Hagiwara, pp. 254-277.

155

VIII

TABLE OF CONTENTS

G. THE SOCIAL COMPONENT OF CHANGE 7. The Social Matrix of Palaeo-Romance Postverbal Nouns

179

Romance Philology, XXX:1 (1977), 54-90. H. LEXICAL INDEPENDENCE VS. GRAMMATICAL CONSTRAINT 8. Each Word Has a History of its Own Glossa: A Journal of Linguistics, 1:2 (1967), 137-149.

217

I. MULTIPLE CAUSATION 9. Multi-Conditioned Sound Change and the Impact of Morphology on Pho nology Language, LII:4 (1976), 757-778. 10. Multiple Versus Simple Causation in Linguistic Change To Honor Roman Jakobson: Essays on the Occasion of his Seventieth Birthday (The Hague & Paris: Mouton, 1967), pp. 1228-46. 11. The Five Sources of Epenthetic / j / in Western Hispano-Romance: A Study in Multiple Causation Hispanic Review, XXXVII:2 (1969), 239-275. 12. On Hierarchizing the Components of Multiple Causation Studies in Language, 1:1 (1977), 81-108.

229 251

269 297

J. ACCENTOLOGY AND PHONOLOGY 13. Conflicting Prosodic Inferences from Ascoli's and Darmesteter's Laws?. . . Romance Philology, XXVIII:4 (1975), 483-520. 14. Etiological Studies in Romance Diachronic Phonology Acta Linguistica Hafniensia, XIV:2 (1973 [-74]), 201-242.

323 361

K. AFFIXAL DERIVATION 15. One Characteristic Derivational Suffix of Literary Italian: ֊(t)aggine 399 Archivio Glottologico Italiano, LXI: 1-2 (1976), 130-145 {Mélanges Bruno Migliorini). 16. The Double Affixation in Old French gens-es-or, bel-ez-or, Old Provençal bel-az-or 411 Studia Neophilologica, XLV:2 (1973-[74]), 217-225. 17. The Rise of the Nominal Augments in Romance: Graeco-Latin and Tuscan Clues to the Prehistory of Hispano-Romance 419 Romance Philology, XXVI (1972), 306-334. L. ETYMOLOGY 18. Identification of Origin and Justification of Spread in Etymological Analysis Romance Philology, XXII:3 (1969), 259-280.

451

TABLE OF CONTENTS

ix

19. Primary, Secondary, and Tertiary Etymologies: The Three Lexical Ker nels of Hispanic saña, ensañar, sañudo 473 Hispanic Review, XLII:1 (1974), 1-32. 20. Etymology and Modern Linguistics 497 Lingua, XXXVI:2-3 (1975), 101-120. 21. The Interlocking of Etymology and Historical Grammar (Exemplified with the Analysis of Spanish desleír) 513 Proceedings of the Second International Conference of Historical Linguistics {Tucson, January 1976) (Amsterdam: North Holland Publishing Co., 1976), pp. 285-312. 22. Contacts Between BLASPHĒMĀRE and AESTIMĀRE 543 Romance Philology, XXX :1 (1976), 102-117 ("Jean Frappier Memorial). Supplement

559

Index of Names

593

Selective Index of Key Terms

607

GUIDE TO ABBREVIATIONS A. Languages and Dialects Abruzz. AFrz. Alav. Aient. Alg. Alj(am). Am.-E. Am.-Sp., Am(er). And. Apul. Ar. Arag. Arán V. Arom. Ast.

Abruzzese Altfranzösisch Alavés Alentejano Algarve (in Southern Portugal) Aljamiado American English American Spanish Andalusian Apulian Arabic Aragonese Aran Valley (= Upper Garonne Valley) Aromunian, or Macedo-Rumanian Asturian

. Bal. Béarn. Beir. Bergam. BÜb. Bog. Bol. Braz.-Ptg. Bret. Burg. Byz.

(a) Basque; (b) Berciano (subdialect of Asturo-Leonese, used in El Bierzo) Balearic (insular variety of Catalan) Béarnais Beirão (in Southern Portugal) Bergamasque Bilbaíno Bogotá City dialect Bolognese Brazilian-Portuguese Breton Burgalés (dialect of Burgos) Byzantine Greek

 Cac. Cal. Camp(id). Cast. .-Ast. Cat. C.-E.-Ast.

Cabraniego (subdialect of Central Asturian) Cacereño Calabrese Campidanese Castilian (dialects of Old and New Castile) Central Asturian Catalan Central and Eastern Asturian

XII

GUIDE TO ABBREVIATIONS

Celt. Ch.-Gr. Ch.-L. Col. Comasc. .-Sant. Cub. Cymr.

Celtic Church Greek Church Latin (a) Colunga subdialect (Eastern Asturian); (b) Colombian Comasco Central Santanderino Cuban Spanish Cymric (=Old Welsh)

Dalm. Dan. Dauph.

Dalmatian Danish Dauphinois

E.-Ast. Emil. E(ng). Eng(ad). esp. Extr.

Eastern Asturian Emilian English Engadin Valley espagnol Extremeño, i.e., dialect of Extremadura in SW Spain

Flor. Foréz. Fr. franç. Fr(an)k. Friul. frz.

Florentine Forézien French français Frankish Friulan französisch

Gael. Gal. Gasc. Gaul. Gen. G(er). Gk. or Gr. Gmc. Goth. Gris. Gr.-Lat. Guip.

Gaelic Galician Gascon Gaulish Genoese German Greek Germanic Gothic Grisons (= Graubünden) Graeco-Latin Guipuzcoan

Hebr. Hisp. Hisp.-Ar. Hisp .-Lat. Hisp.-Rom.

Hebrew Hispanic (= Spanish + Portuguese) Hispano-Arabic Hispano-Latin Hispano-Romance


xiii

IE, I.-E. Imp. L. Istr. It.

Indo-European Imperial Latin Istrian Italian

Jud.-Sp. L(at). Leon. Lim. LL Log(ud). Longob. Low G. Low L. Lucch. Lyonn.

Judaeo-Spanish Latin Leonese Limousin Low (or Late) Latin Logudorese Longobardo (= Lombard) Low German Low Latin Lucchese Lyonnais

M. M(ed.)Lat. Megl. Mi. Fr. MÜ. Mir. MIt. Moden. Morv. Moz(ar). MPtg. Murc.

Mérida subdialect of Extremeño Medieval Latín Meglo-Rumanian (Mount Olympus) Middle French Milanese Mirandese (Miranda del Duero subdialect of Western Leonese) Modern Italian Modenese Morvan dialect (East-Central France) Mozarabic Modern Portuguese Murcian (= SE Spanish)

Neap. NIr. N.It.

Neapolitan New Irish North Italian

OArag. OAst. Obwald. OCat. Occ(it). ODac.-Rum. OF(r.) OGal. OGen. OHG OIr. OIt.

Old Aragonese Old Asturian Obwaldisch (= Surselvano) Old Catalan Occitan Old Daco-Rumanian Old French Old Galician Old Genoese Old High German Old Irish Old Italian

xiv


OJud.-Sp. OLeon. OLomb. ONav. ON (0)Prov. OPtg. ORioj. OSay. OSp. OTrev. OVal. OVen. OW.-Fr.

Old Judaeo-Spanish Old Leonese Old Lombard Old Navarrese Old Norse Old Proven9al Old Portuguese Old Riojan Old Sayagues (variety of Zamoran) Old Spanish Old Trevisano Old Valencian Old Venetian Old Western French

Pad. Parm. Piedm. Pis. Poit. port. pre-Rom. Prov. Ptg. R.-Rom. Rum. R(uss).

Padovan Parmigiano Piedmontese Pisan Poitevin portugais pre-Romance Provencal, Occitan Portuguese R(h)aeto-Romance Rumanian Russian

Sal. Sant. Sard. SEFr. Seg. Sen. SFr. Sic. SIt. Sp. Surs. Sw.

Salamancan, or Salmantino Santanderino (=Montanes) Sardinian Southeastern French Segovian Senese (subdialect of Tuscan) Southern French Sicilian Southern Italian Spanish Surselvan (= Obwaldisch) Swiss

ted. Tortos. Transylv. Trasm. Tusc.

tedesco Tortosan (Catalan dialect) Transylvanian Trasmontano (in Northern Potugal) Tuscan


Upp. Arag.

Upper Aragonese (=alto aragonés)

Val. V.Anz. Vegl. Ven. V.Lat. (VLat.) W.Ast.

Valencian Dialect of the (Alpine) Valle Anzasca Vegliote Venetian Vulgar Latin Western Asturian

Zam.

Zamoran

xv

B. Standard Reference Works AFW BDE BIL CGL CIL CTL DCE(LC) DCR DEEH DEI DÉLF DELL DH(ist) EDMP ELM EWDS EWFS EWRS EWUG FEW

Altfranzösisches Wörterbuch, by A. Tobler & E. Lommatzsch Breve diccionario etimològico de la lengua castellana, by J. Coraminas (various editions) Bibliography of Italian Linguistics, by Robert A. Hall, Jr. Corpus Glossariorum Latinorum, comp. G. Loewe & G. Goetz Corpus Inscriptionum Latinorum, edd. Theodor Mommsen et al. Cunent Trends in Linguistics, ed. Thomas A. Sebeok Diccionario crítico etimológico de la lengua castellana, by J. Coraminas Diccionario de construcción y régimen de la lengua castellana (A-E), by Rufino José Cuervo Diccionario etimológico español e hispánico, by V. García de Diego Dizionario Etimologico Italiano, by C. Battisti and G. Alessio Dictionnaire étymologique de la langue française, by 0 . Bloch & W. von Wartburg (several editions) Dictionnaire étymologique de la langue latine, by A. Meillet & A. Ernout (four editions) Diccionario histórico de la lengua española (A-Ce), by the Spanish Academy (1933-36) Estudios dedicados a Menéndez Pidal (7 vols, in 8) Enciclopedia de lingüística hispánica Etymologisches Wörterbuch der deutschen Sprache, by F. Kluge, rev. by W. Mitzka (several editions) Etymologisches Wörterbuch der französischen Sprache, by E. Gamillscheg (two editions) Etymologisches Wörterbuch der romanischen Sprachen, by F. Diez (five editions) Etymologisches Wörterbuch der unteritalienischen Gräzität, by G. Rohlfs Französisches etymologisches Wörterbuch, by W. von Wartburg (et alii)

xvi GHLC GHLF GRS

HGFS HGIS  LEW LRW MGHE NDEH OSR REW RMP VGKS


Gramática histórica de la lengua castellana, by F. Hanssen Grammaire historique de la langue française, by Kr. Nyrop (6 vols.) Grammatik der romanischen Sprachen, by (a) F. Diez (five editions), and (b) W. Meyer-Lübke, each comprising 3 vols, plus index volume Historische Grammatik der französischen Sprache, by W. MeyerLübke, 2 vols, (four and two eds., respectively) Historische Grammatik der italienischen Sprache und ihrer Mundarten, by G. Rohlfs, in 3 vols. Homenaje a Ramón Menéndez Pidal (3 vols.) Lateinisches etymologisches Wörterbuch, by Α. Walde & J.B. Hofmann Lateinisch-romanisches Wörterbuch, by G. Körting (three editions) Manual de gramática histórica española, by R. Menéndez Pidal (six basic editions) Nouveau dictionnaire étymologique et historique, by A. Dauzat, J. Dubois, & H. Mitterrand Old Spanish Readings (1911 and later eds.), by J.D.M. Ford Romanisches etymologisches Wörterbuch, by W. Meyer-Lübke (1st and rev. 3rd eds.) Ramón Menéndez Pidal, El dialecto leonés (1906) Vergleichende Grammatik der keltischen Sprachen, by H. Pedersen (2 vols.)

C. Publication Series BAE BAR BBRPh

Biblioteca de Autores Españoles Biblioteca áelVArchivum Romanicum Berliner Beiträge zur romanischen Philologie (dir. E. Gamillscheg) BDHA Biblioteca de dialectología hispano-americana, dir. Amado Alonso ÈÉ Bibliothèque de l'École pratique des Hautes Études BH Biblioteca ffispánica (Madrid-Barcelona) Docum. inéd. Colección de documentos inéditos (Archivo general de la Corona de Aragón) EM Elliott Monographs in the Romance Languages and Literatures, ed.E.C. Armstrong Esp. Sagr. España Sagrada: Teatro geográfico-histórico de la iglesia de España GRL Gesellschaft für romanische Literatur (Dresden) NBA(A)E(E) Nueva Biblioteca de Autores Españoles Publ. of the Philol. Publications of the Philological Society (London) Soc.


RH RW SPRF TPS UCPL UPSRLL UUÅ

xvii

Romanica Helvetica (founding eds. J. Jud & Α. Steiger) Romanistische Versuche und Vorarbeiten (Bonn) Société de Publications Romanes et Françaises Transactions of the Philological Society (London) University of California Publications in Linguistics University of Pennsylvania Studies in Romance Languages and Literatures Uppsala Universitets Årsskrift

D. Scholarly Journals AGI ALH ALLG AnL AR Arch. ArL AS ASNP ASNS

Archivio Glottologico Italiano Acta Linguistica Hafniensia Archiv für lateinische Lexikographie und Grammatik A n thropological Լ inguistics Archivum Romanicum Archivum (Oviedo) Archivum Linguisticum American Speech Annali della Scuola Normale Superiore di Pisa Archiv für das Studium der neueren Sprachen (und Literaturen)

BBMP BF BDR BH(i) BHS BICC BIFCh BL BRAE BRAH BRPh BSC BSLP CA CEG CFS CL Fil FM FS HR HSN ID

Boletín de la Biblioteca de Menéndez Pelayo Boletim de Filologia (Lisboa) Bulletin de dialectologie romane Bulletin hispanique Bulletin of Hispanic Studies Boletín del Instituto Caro y Cuervo (= Thesaurus) Boletín del Instituto de Filología de Chile Bulletin linguistique Boletín de la Real Academia Española Boletín de la Real Academia de la Historia Beiträge zur romanischen Philologie Boletim da Segunda Classe (da Academia das Sciências, Lisboa) Bulletin de la Société de Linguistique de Paris Cunen է An thropology Cuadernos de estudios gallegos Cahiers Ferdinand de Saussure Comparative Literature Filología Le Français Moderne French Studies Hispanic Review [Harvard] Studies and Notes in Philology Լ Italia Dialettale

XVIII

IJAL IJSLP IMU JMRS JRESL JRL Krat. LeS Lg., Lang. LGRPh LIs LP MaR MLF MLN MLR MPh MSL(P) NJKA NRFH Phon. PhQ PMLA RABM RBPhH RDTP Rev. de 0cc.2 RF RFE RFH RH RL RL(a)R RLiR Rom. RP R(om.) Ph(iloi.) RPF RR RRL RVF SCL Sc.-Sl. Sp(ec) SP(h)


International Journal of American Linguistics International Journal of Slavic Linguistics and Philology Italia Medioevale e Umanistica Journal of Medieval and Renaissance Studies Jahrbuch für romanische und englische Sprache und Literatur Kratylos Lingua e Stile Language Literaturblatt für germanische und romanische Philologie Linguistics A Lingua Portuguesa Modern Language Forum Modern Language Notes Modern Language Review Modern Philology Mémoires de la Société de Linguistique de Paris Neue Jahrbücher für das klassische Altertum, Geschichte und deutsche Literatur Nueva Revista de Filologia Hispánica Phonetica Philological Quarterly Publications of the Modern Language Association of America Revista de Archivos, Bibliotecas y Museos (Third Series) Revue belge de philologie et d'histoire Revista de dialectología y tradiciones populares Revista de Occidente (Second series) Romanische Forschungen Revista de Filología Española Revista de Filología Hispánica Revue hispanique Revista Lusitana Revue des langues romanes Revue de linguistique romane Romania Revista de Portugal Romance Philology Revista Portuguesa de Filologia Romanic Review Revue roumaine de linguistique Revista Valenciana de Filología Studi şi cercetari Scando-Slavica Speculum Studies in Philology


SSL StN TLL VKR VR(om.) WS YWMLS ZRPh ZVS

Studi e saggi linguistici Studia Neophilologica Travaux de linguistique et de littérature Volkstum und Kultur der Romanen Vox Romanica Wörter und Sachen Year's Work in Modern Language Study Zeitschrift für romanische Philologie Zeitschrift für vergleichende Sprachforschung (= "Kuhn's Zeitschrift")

E. Miscellaneous Bibliographic Hints Acad. anon. Bibl. Bk. Bull. Canc. Canc. Vatic. Ch(ap). Cl./Kl. Cod. Alcob. col. Col.-Branc. comp. Congr. Dic. diss. doc(um). ed. Esc. Fac. fasc. f(n). f(ol). FS F.yL. G1. Hofbibl. inéd. Int. Intr.

Լ. M.

xix

Academy (Académie, Academia, etc.) anonymous Biblioteca Book Bulletin Cancione(i)ro Cancioneiro da Vaticana Chapter Klasse Codex Alcobacensis column Cancioneiro Colocci-Brancuti (= da Biblioteca Nacional) compiled, -er Congress Dicionário dissertation document edition; editor Escorial (Library) Faculty, Faculté, etc. fascicle footnote folio Festschrift Filosofía y Letras Glossary Hofbibliothek (= Imperial Library, Wien) inédito International Introito (='Prelude') Leipzig Madrid

XX

MIT Nac. P. Pal. PCG Ph.-H. Philol. Proc. Prov. Publ. qu(atr). repr. rev. Ser. Sitz.-ber. Soc. st. Suppl. Tol. tr. Univ. unpubl. UP or U.P. Verh. Wtb.


Massachusetts Institute of Technology Nacional Paris Palacio (Biblioteca del) Primera Crònica general, ed. R. Menéndez Pidal Philologisch-historische Klasse Philology, -ical Proceedings Province publish(er); publication (of); Publications quatrain reprinted revised; revision; review (of. ..); reviewed (by . . . ) Series Sitzungsberichte der Akademie der Wissenschaften Society, Société, etc. stanza Supplement Toledo translated by University, Université, (Universitaire), etc. unpublished University Press Verhandlungen Wörterbuch

F. Grammatical, Linguistic, Lexicographic, Stylistic Qualifiers abl. adj. adv. ambigen. anthr. arch. art. augm.  = cons. Cl(ass). coll. comp(ar). cond. def. deriv. derog. dial.

ablative adjective adverb ambigeneric anthroponym archaic article augmentative consonant Classical colloquial comparative (degree) conditional definite derived, dérivé, derivado, etc. derogatory dialectal


dim(in). eccl. epigr. f(em). fig. FS fut. gen. hydr. ichth. imper. impers. impf. ind(ic). inf(in). iron. joc. lit. m(asc). mod. n. neg. nom. obl. obs(ol). onom. orn. perf. p(er)s. pl. pleb. poet. pop. postv. p. ptc. prep. pres. pret. pron. prov. refl. sg. stand. subj. subst. top(on).

diminutive ecclesiastic epigraphy feminine fìgurative(ly) folk speech future (tense) genitive hydronym ichthyonym, or fish name imperative impersonal imperfect (tense) indicative infinitive ironical(ly) jocular literal(ly) masculine modern (a) noun, (b) neuter negative nominative oblique (case) obsolete onomatopoeia ornithonym, bird name perfectum person(al) plural plebeian poetic popular, -aire (= 'vernacular') postverbal past participle preposition present (tense) preterit(e) pronounced provincial, regional reflexive singular standard (a) subjunctive; (b) subject (case) substantive, -ivated toponym

xxii V V.

var. voc.


vowel verb variant vocative

G. Realms of Nature and of Human Endeavors archit. bot. geol. med. min(er). naut. vet.

architecture botany geology medicine mineralogy nautical veterinary science

H. Residual  Dépt. esp. fam. Jh. Ling. LSA NE NW orig.

century Département especially family Jahrhundert Linguistics Linguistic Society of America Northeast Northwest original(ly)

FROM PARTICULAR TO GENERAL LINGUISTICS

INTRODUCTION

I. The title for this collection of articles, From Particular to General Historical Linguistics, has been chosen not, or not only, on account of the amenity of the programmatic formula, but chiefly because this tag characterizes, as no alternative label would do, the slant of the miscellany as a whole and of most of the individual papers represented in it. Moreover, this particular wording captures a bit of the "Zeitgeist", as I shall attempt to expound later. The short and middle-sized writings assembled here were originally published either in learned journals or in special book-length ventures (e.g., testimonial vol umes) through the 'sixties and the 'seventies. Now it has been observed by reliable witnesses endowed with special sensitivity for "moods" gripping a community of intellectuals (including linguistic scholars, notoriously subject to such periodic spells) that, starting with the mid 'fifties, the most alert among the younger practi tioners of historical linguistics all of a sudden felt, at gatherings of societies and on other appropriate occasions, that things could not simply continue rolling as they had done before, rather smoothly at that, over a period of perhaps a half century. The older schools of spatio-temporal linguistics gave the impression of having spent their energies, through a kind of exhaustion calling for quick rejuvenation, by way of effective remedy, which seems to be the blessing or the curse — depending on the observer's vantage point — in most, if not all, such undertakings. Until then, I had been doing advanced work, for a segment of twenty years or so, almost exclusively in the domain of Romance Linguistics with perceptible em phasis on older Spanish and Portuguese. In all likelihood, the crisis alluded to above hit my particular circle of friends either a few years earlier or a few years later than it did comparable contingents, not necessarily "teams", of historical linguists. There was a variety of obvious and not so obvious circumstances that determined the actual dates of such emergencies — the disappearance from the scene of powerful figures, who, by sheer personal magnetism, could for a while slow down or deflect the waves of change; the success, or dismal failure, of a specific congress; the sudden availability, or unexpected collapse, of an influential research institute, a learned journal, or a monograph series, and the like. As a result, Slavists, Germanists, Anglists, Orientalists, and other groups similarly self-defined went through basically the same category of experiences at short distances from each other, as it were. Very much depended, in each case, on whether the individual caught up and hit by this wave was already a mature scholar, with numerous prior ideological commit ments; or a very young person, surprised by the impact of events in mid-training and consequently still able to extricate himself, if he so decided, from some earlier

4


orientation and both quickly and painlessly to adjust himself to a radically new con stellation of circumstances. It so happened that I was neither young enough nor particularly old at the critical juncture. Members of my own generation went through a veritable crise de conscience, which lasted a full decade. Some regarded themselves as duty-bound, not least as teachers and trainers of younger apprentices, to re-design energetically all their previous blueprints for research — either reserved for themselves or to be recom mended to their disciples. Clearly, it cannot be my responsibility here to determine whether an element of sheer opportunism also entered into certain decisions. In extreme cases, a scholar would publicly abjure his earlier practices and associations and strike out anew, playing the controversial role of a reborn (or, at least, reformed) linguist. Human nature being what it is, a group of scholars more conservative by training or temperament (or else by both) responded to the challenge of modernism under any guise by quickly moving, more stubbornly than ever, in the opposite direction and thus cutting off themselves and their followers from any avenue of approach to rejuvenation. Since "atomism" seemed to be at the opposite end of "structur alism", "functionalism", and similar movements, scholars stricken by this recal citrant mood, as a rule, looked for a shelter in such reputedly safe havens as ety mology or toponymy. And since reformers in all fields, from time immemorial, have been colorful, flamboyant personalities, not immune to criticism for their unconventional modus operandi, the hard-core conservatives, before long, discoverd the vulnerability of their opponents and justified their own negative decisions on moral grounds, as a refusal to condone superficiality or to engage in fashionable charlatanism. Such an evasive attitude, let me admit from the start, has never seemed judicious to me, not least because it threatens to lead to sterility. The only alternative to dramatically espousing such forms of extremism seemed to be, at first glance, the choice of a middle road. But a lukewarm compromise, under any circumstances, appears singularly unattractive, and its prospect certainly lacked any appeal at that crucial point. Instead of joining one of the camps, I decided to examine, rather unhurriedly, just what loomed as "good" or as "bad" in the older style and in the newer styles of historical linguistics. Of course, I realized that, like any maverick scholar, I might have to pay a steep price for my refusal to join un conditionally any neatly silhouetted group of linguists; but the beckoning reward of independence seemed worth that risk.

II. The learning process to which I was exposed at a major Central European uni versity in the early and mid 'thirties presupposed a very different segmentation of knowledge from the one which was to underlie eventually my own teaching and research activities in the New World. Within the broad, officially recognized frame of the Philosophical Faculty and the narrower frame — defined more psycholog ically than administratively — of the language-and-literature disciplines, the big nuclei were isolated in terms of the respective language families: Graeco-Latin

INTRODUCTION

5

("Classics"), Romance, Germanic, Celtic, Semitic, Finno-Ugric, and the like. Within such a kernel, two semi-independent halves typically asserted themselves, which could be labeled as the "linguistic emphasis" and the "literary emphasis". Students aiming primarily at a career in secondary teaching strove to maintain, as best they could, a balance between the two emphases; the small minority endowed with very sharply pronounced intellectual tastes (and aversions), including those few who were, optimistically, toying with the remote possibility of a university career in the future, tended to favor one emphasis at the expense of the other. Even so, specialization was not allowed, still less encouraged, from the start; a budding lin guist, to qualify as a serious philologist, was expected to acquire a solid factual knowledge of older and more recent belles lettres (and of the most important in quiries that had a bearing on them) and vice versa. This state of affairs strongly colored the acquisition and assimilation of knowledge and co-determined the fledgling scholar's first attempts to test his ability for independent research. Since the general principles of linguistic and literary investigation did not figure in the curriculum as separate, easily recognizable units, it took even alert and sophisti cated students a considerable period of time to find out that certain scholars, not necessarily those whose lecture courses were readily available to them, did experi ment with possibilities of establishing new schools of thought for linguistic and literary analysis. The general climate of feeling was that all the important broad insights could easily be gleaned from a book by then already more than fifty years old, Paul's Prinzipien der Sprachgeschichte. Further advances hinged on the gradual accumulation and neat sifting-out of fresh collections of data — culled from longneglected texts or from direct observation and elicitation through field-work; and on periodic syntheses of such stray data in manageable reference works, such as histo rical grammars and etymological dictionaries. Another thrust oft-mentioned in conversations that I overheard in those years was the steady seeking-out of banefully overlooked transition zones (Grenzgebiete) requiring dual or triple expertise. Joint command of Slavic and Eastern Romance, or Arabic and Hispano-Romance, etc. would guarantee an ambitious, hard-working newcomer a treasure-trove of discove ries and accelerated recognition by grateful spokesmen for the chosen discipline. The basic flaw in this widespread reasoning has by now been sufficiently exposed on both sides of the Atlantic. To limit myself to the linguistic facet of the argument, it is clearly unhelpful to become immersed in the particulars of certain diphthongal, metaphonic, or apophonic processes without the privilege of a preliminary over view — i.e., without adequate foreknowledge of the commonness and socio-spatiotemporal distribution of such phenomena on a wide, if not necessarily global, scale and on their essence. Once you have learned, from Edward Sapir (say), that there exist, in all, six ways for arriving from a given word x to a newly-minted word y (through affixation, composition, reduplication, stress-shift, and the like), then the specifics of such processes, as observable in a particular corpus of data, fall into place at once. Consequently, today's revised curricular conditions, which require the beginner to go, first and foremost, through an experience of broad orientations, as provided by the various departments of linguistics, and only then to plunge into specialized, necessarily very technical research, unquestionably represent a major step forward.

6


There can be no excusable return to the earlier practice. But, having made this fairly unoriginal statement, I hasten to add (and this ela boration is far less frequently encountered) that the reform, welcome and overdue as it may have been, was not lacking in several decidedly harmful side-effects. While a bit of theoretical clarification as a tone-setting prelude to the actual investigative operations has indeed turned out to be a boon and a necessity, it carries with it the risk, especially for practitioners of historical linguistics, that the writer may become infatuated with the peculiar charms of theoretical elucidation, to the extent that the actual spadework loses much, or even most, of its attraction for him. What ever my possible shortsightedness along other lines of endeavor, I almost instantly realized that any reckless espousal of a seductively elegant form of "modernism" in linguistic analysis carried with it the strong temptation to veer away from diachronic work in the direction of straight description. It almost seems as if, for better or worse, historical reconstruction does not smoothly lend itself to ever higher degrees of neatness; that it has an inherent com plexity, displaying, again and again, alliances of causes in action and counteraction; and that, as one advances and learns to subsume certain scatterings of processes under common denominators — a skill conducive to tighter organization and thus representing, scientifically, a step in the right direction — such progress is counter balanced by the discovery of additional, long-hidden factors, so that there actually occurs no diminution of intricacy in the problems envisaged, unless of course one deliberately chooses to disregard some of their less welcome facets — those that do not readily lend themselves to smooth analysis, to raising to a higher plateau of abstraction. One must simply learn to control such "disappointments". I also could not help noticing that some members of my generation, including a few of those who engaged in formal recantation of earlier practice as well as others, who tacitly repudiated the sort of knowledge and commitment that had been inculcated in them by their teachers, very seldom grew in stature as regards originality and pro ductivity even though they may have gained other benefits from their conversions. From these, in part painful, experiences and observations I drew the conclusion that, at least for my own research, the most satisfactory solution might consist in remaining open-minded to innovating theories, but highly selective and delibe rate in adopting them, and extraordinarily wary about abandoning, under flimsy pretexts, such patches of territory as had turned out to be demonstrably promising under an earlier ideological regime. From particular to general, in historical linguist ics, began to loom as a motto not for gradual shedding of particulars for the sake of esthetically pleasing generalizations, but for the selection of a platform allowing, in one direction, the free view of particulars and, in the other, the observation of their shading-over into universals.

III. During my years of university study in Europe I, rightly or wrongly, gained the impression that one should, at the beginning of his independent research activities, preferably concentrate on a single language family, while keeping abreast — for the

INTRODUCTION

7

sake of perspective — of work carried out in other domains; that its richly docu mented, rather than sorely lacunary record, spreading over more that two millen nia, made the Romance family a splendid training ground for this purpose; and that, within the Romance domain almost as a whole, in certain provinces (such as Old French) the soil had been almost too vigorously tilled, while others (principally Spanish and Portuguese) had been badly neglected and thus clamored for imme diate attention. In addition to these purely scholarly considerations, certain world events, around 1940, also counseled temporary concentration on Luso- and HispanoRomance, despite the fact that my doctoral dissertation, completed by the end of 1937, had dealt with French almost exclusively. The fact that, through a twist of circumstances, I found my first niche in a Hispanic department of the University of California's Berkeley campus, rather than in a unified Romance department in some other, comparable graduate school, pushed me still further in that direction. However, this was a temporary concession to the all-pervasive trend toward increased specialization. Only a few weeks after my arrival in this country I made a point of applying for membership in the Linguistic Society of America, and the very first paper that I had the exceptional luck of submitting to the editor of Lan guage and of having published by that journal in 1941 was slanted in the pan-Romanic direction, without any preferential attention to French or Spanish. The 'forties were marked by strong dissensions among several groups, not to say factions, of linguists of various backgrounds and vintages then residing in the United States, and I was rather disinclined to enter the fray, and do not, in retrospect, regret my hesitation at that inopportune moment to have voiced my own credo. It seemed far more advisable, for a while, to concentrate on the extension of new techniques to fields that had been left lying fallow. Consequently, I dedicated several years to experiments in Hispanic etymology, using as my model such research as had before been successfully conducted with respect to French (but, surprisingly, only seldom in France). Nevertheless, even during those years of unabashed concentration on narrowlycircumscribed problems, concern with broader issues, inviting a clearer statement of one's theoretical position and transcending the history of a single language, made itself felt, especially during the years 1949-1951. Thus, the lengthy paper "Studies in the Hispanic infix -y-" (1949), though operating overtly with the wellestablished category of infixes, by implication laid the foundation for future research in a newly defined category of derivational affixes, namely the "interfixes" (1959). The following year, the article "The hypothetical base in Romance etymology" — the outgrowth of an evening lecture delivered before the Linguistic Circle of New York - represented a first attempt to draw certain innovative theoretical conclusions from a variety of concrete experiments. Finally, "Lexical polarization in Romance" (1951) marked an early effort to strike out in a direction which was later to prove extraordinarily fruitful, namely toward the domain of hypercharacterization, in this particular instance lexico-semantic rather than grammatical. A certain harmful inhibition or self-limitation was thus broken by this triad of bolder explorations.

8


IV. As a result of the cicumstance that implications of highly specialized research for general diachronic linguistics had been a matter of occasional concern for me long before the "crisis" of the mid-'fifties, that particular storm, fortunately, by passed me well-nigh completely. I saw no need at all to abjure and condemn my earlier lines of curiosity; but, to keep close to the mainstream of events, I hence forth became distinctly more selective in targeting certain problems of reconstruc tion for an immediate attack. Narrowly circumscribed issues of etymological identifi cation were at no time discarded, but gradually ceased to be typical goals in them selves. The shifting place of etymology within the total edifice of linguistic knowl edge began to occupy stage center, and the task of categorizing etymological solu tions, of preparing typological surveys of lexical spadework, and even of analyzing into its irreducible constituents the organization of entire etymological and historical dictionaries became more and more seductive. Wherever possible, individual etymo logical issues were interwoven with much broader questions of change in language, as when it was shown, in connection with a Luso-Hispanic preposition characteristi cally difficult on the genetic side, namely Sp. hasta (OSp. ata~fata), Ptg. ate 'until' that it is unfruitful to focus attention on some of its variants separately, as had been done before; that a much sharper picture emerges once hasta (and its satellites) are paired off with their antonyms des(de) 'from' as well as with their near-synonyms and near-homonyms OSp. faza~hacia 'toward'; and that the entire problem, rich in all sorts of ramifications, acquires unexpected depth when viewed as a particularly noteworthy manifestation of a widespread evolutionary trend away from the littleexplored hazards of near-homonymy. To put it differently: While traditionally ety mological notes and articles were so planned as to fill small slots in diachronically tilted dictionaries and in language histories (or historical grammars), the new crop of such papers produced in my own experimental station aimed more and more at elucidating vexing problems of general linguistics: effects of polarization, serializa tion, hypo- and hypercharacterization, and the like.

V. The deliberate widening of the scope, in the direction of broad-gauged method ology, was aimed at not only within the domains of etymology or, speaking in more general terms, lexicology. Serious diachronic research in these twin disciplines cannot be undertaken or pursued without constant reference to historical grammar, as a whole, or to the contiguous, frequently overlapping provinces of phonology and morphology, in particular. I was aware of this state of affairs from the start; for reasons that need not be gone into, my imagination and initiative, at the outset, worked better when brought to bear on morphology (specifically, on affixal deriva tion) than when challenged to confront unsolved phonological problems. By the early 'forties, e.g., it dawned upon me that one of the safest and, at the same time, most promising avenues of approach to etymology was to establish a dossier for one of the less thoroughly investigated derivational suffixes, of which there is a generous

INTRODUCTION

9

supply in Romance, and then to rank the individual formations so developed on the basis of their relative structural transparency. At the bottom of such a list one almost invariably stands an excellent chance of detecting a sprinkling of derivatives displaying an inadequate "fit", a conspicuous extension of meaning, an obscure radical, or some other anomaly which, upon further inspection, arouses the suspicion that some unusual process has here been at work and was eventually covered up or camouflaged by the speech community. Thus, upon methodically studying the Old Spanish formations in -encia, most of them diaphanous, I began to recognize certain long-hidden connections, as when pendencia 'hassle, feud, row', counter to initial expectation, turned out to be related to POENITENTIA 'penance' (cf. Old Portuguese peendenga), though subsequent adjustments gave it the appearance of a mere prong of the PENDERE/PENDERE ('to hang, weigh') family — via litiga tion; or as when primencia 'taxes levied', despite its prima facie resemblance to prim(er)o 'first' (hence 'first fruits'?), was actually found to pertain to the obsoles cent verb premer/primir 'to extort, exert pressure, squeeze out, i.e., levy (taxes)'. Activation of my long-dormant phonological curiosity came at a distinctly later date. It took me a certain effort to realize that, counter to the teaching to which I had been exposed im my years of apprenticeship, the so-called regular phonetic correspondences, for all their usefulness to the student of the lexicon in quickly checking operations, on balance were less readily conducive to elegant and innovative etymological solutions than the more hazily defined sporadic sound changes or, for that matter, synchronic phonological phenomena. After 1960 this, to me new, line of analytical reasoning became more sharply profiled, as is shown, e.g., by the short piece "Weak phonetic change, spontaneous sound shift, lexical contamination", which appeared in the 1962 Testimonial for A. W. de Groot. From here a relatively smooth road led, in part toward the in-depth scouting of those unexpected situations where straight morphological conditions — e.g., the aberrancy of a paradigm — produced a climate in which a sound change eventually took shape ("The inflectional paradigm as an occasional determinant of sound change", 1969), in part toward the exploration of multiple causation, marked by a pattern of interlocking ("Form versus meaning in etymological analysis", 1966; "Multiple versus simple causation in linguistic change", 1967). Thus the elements of theory were brought in, but never at the expense of concrete experimentation.

VI. Against the background just sketched out, it becomes understandable why, among the studies included in this collection, practically each one aims at two goals very seldom bracketed in present-day linguistics (and, in certain quarters, held to be irreconcilable almost by definition). The one direction that the typical study absorbed into this book takes is the continuation and, one hopes, successful conclusion of a circumstantial glottohistorical inquiry initially undertaken by the pioneers and/or classics of the old and venerable historico-comparative school. As a result, the reader must be prepared to find here mention (with approval, qualifi cation, or rejection) of many earlier conjectures or explanations advanced by Diez,

10


Meyer-Lübke, Cuervo, Menéndez Pidal, and the other protagonists of the familiar dramatic development of Romance diachronic linguistics. The other, possibly more "relevant", direction that these studies take is the attempted enrichment — through discussion and exemplification — of linguistic theory and methodology, so far as diachrony is concerned. This pervasive bifocal approach has made it recommendable, if not mandatory, to have recourse, on several occasions, to both titles and subtitles, so as to make the possible usefulness of the papers concerned immediately clear to two rarely integrated groups of readers: the practitioners of Romance linguistics bent on continuing and further refining a preëxistent, consecrated tradition of re search; and the theoretically-inclined explorers of general linguistics, who are pre pared to strike out along lines previously overlooked, in an innovative mood, but do not mind taking into account such serious and imaginative research as has been conducted in Romance headquartes for over a century and a half. Let me exemplify these gropings: (a) "Linguistics as a genetic science" (1965), being a formal presidential address, respects certain constraints peculiar to that genre of conventional scholarly allocu tion. In essence, it attempts to bracket historical linguistics as a whole with other evolutionary sciences ֊ by collocating it with geology, paleontology, and paleo botany — at the heavy cost of loosening its connections with other varieties of lin guistic inquiry. While the nucleus of the paper contains, as announced in its title, a very broad-gauged discussion, the four particular "problems" appended to it by way of illustration entail a return to the Romance fold and involve fine-meshed analyses of issues which have grown out of earlier dialogues between specialists in this domain and which, all four, revolve around multiple causation (the sources of -io as a Spanish adjectival final segment; the widespread fluctuation between [δ ] and [r] in Hispano-Romance ; hypercharacterization of gender, chiefly in zoönyms; and consonant dissimilation at a distance). The concluding pages lead the reader back to a problem of distinctly broader implications, namely a tentative differentia tion of primary (or major) and secondary (or minor) causes. (b) "History and histories of linguistics" (1969) — a piece to which my friend and former student Margaret Langdon wrote a short sequel, to show how European observers tend to view the development of linguistics in this country — stands apart, topically, from all the other papers here collected, in as much as it steers clear, by definition, of any attack on concrete problems of language growth. It busies itself with the special difficulties besetting the path of those trying to acquire, and demonstrate their command of, a dual expertise: first, in a technical scientific field, and second, in the historiographic art of piecing together a close-knit, persuasively phrased account out of many loose shreds of bibliographic information. Theoretical ly, it is entirely defensible to aim at a history of schools of descriptive (synchronic) linguistics; in fact, a few such experiments have already been conducted. Since my own article concerns itself, preeminently, with a pyramid of earlier studies in histor ical (diachronic) linguistics, it is actually geared to two time axes. (c) "Range of variation as a clue to dating" (1968) is, strictly speaking, only the opening part of a far more ambitious venture; I have postponed, but by no means abandoned the hope for, its completion. Upon closer inspection, critics will un doubtedly discover a certain affinity to glottochronological thinking, except that

INTRODUCTION

11

the technique proposed is discernibly more cautious and the conclusions drawn from the evidence less sweeping and far more easily reconcilable with traditional patterns of investigation. The underlying line of thought runs, roughly, thus: If, in a language otherwise known for the tightness of its organization and a parsimoniousness of forms in its paradigms, the researcher nevertheless discovers certain scattered spots marked by a striking, not to say gratuitous, accumulation of variants, the chances are that the speech community at issue simply has not had the time to shed the overflow. A further inference is (and here the ingredient of dating comes to the fore) that the clusters of variants cannot be old, that they must have arisen not very long before the beginnings of actual record of the given language. By way of illustration, the paper offers a minute spatio-temporal analysis of the preterites of the verb TRAHERE 'to draw' in Luso and Hispano-Romance. Reduced to a formula, the proposal advocates the use of erratic polymorphism as a device for relative dating in reconstruction. (d) "Factors in the unity of ROMANIA" (1977) is not a technical paper, but the text of a short communication read before a group of humanists, for the most part innocent of any curiosity about, let alone any grasp of, advanced linguistic research. The talk represents an attempt to show to sophisticated laymen how linguists and literary scholars may jointly go about setting off a closely coherent cultural area. (e) "Lexical borrowing in the Romance languages" (1975) is a review article devoted to T. E. Hope's important dissertation on the sort of osmosis that has, for centuries, prevailed in the lexical relations between French and Italian. While much of the discussion is deliberately technical and addresses itself to Romanists, certain insights may, it is hoped, appeal to a wide range of students of lexical diffu sion, especially those concentrating on the interplay of infiltrations, in different directions, within the frame of close cognates. (f) "Gender, sex, and size, as reflected in the Romance languages" (1977) is something of a linguistic divertissement, though certainly not along the axis of pru rience. It is, essentially, a narrow-meshed discussion of one of the classic contro versies in the annals of Romance scholarship — a dispute that started in a low key almost a century ago, gradually grew in scope and intensity and, at its peak, about thirty years ago, attracted the attention of, and even engulfed, the spokesmen for several conflicting schools of thought. Even after one subtracts the purely anec dotal and incidental elements of that debate, there remains enough to feed the imagination of several future generations of linguists. (g) "The social matrix of palaeo-Romance postverbal nouns" (1977) marks a return to a patch of territory that has been lying fallow for three quarters of a centu ry. The distance that separates us from the days of A. Darmesteter, G. Paris, A. Tobler, and the Scandinavian pioneers of Romance linguistics enables us to try out an almost totally new approach to another "classic problem" of Romance scholarship. A quick look at this type of derivation in reverse, or clipping — highly atypical of Roman folk speech, which delighted in toying with chains of suffixes forewarns us that the discovery of a very special context, or social matrix, could alone justify the rise and subsequent flowering of what initially must have been a very bizarre grammatical model for verbal abstracts. Closer inspection of the Latin

12


corpus confirms that the ground-breaking words (LǓCTA 'struggle, wrestling', PUGNA 'fistfìght') pertained to military life and to the related domain of organized athletics. The nouns so shaped may very well have served as command words; and in the issuing of commands brevity is of the essence. In the twilight hour between Antiquity and the Middle Ages, the model was extended, still within the same province of real Ufe, to the designation of soldiers assigned special tasks ('spy', 'guide', 'sentry', and the like). Here, in all likelihood, a case of Romano-Germanic symbiosis is involved. (h) For a variety of reasons, including its relative brevity and the lightness of the documentation adduced, the purely analytical essay "Each word has a history of its own" (1967) stands apart from the other papers here gathered. This old dic tum — I have heard its authorship attributed to Jakob Grimm — in the end was picked up by, of all twentieth-century linguists, Leonard Bloomfìeld (1933), who almost desperately tried to reconcile it with the, to him dear, dogma of the regularity of sound change. However, there remain, upon close inspection of the arguments, certain irreducible incompatibilities between the two creeds, which both the student of straight historical grammar and the less austere word biographer must heed and should not expect to be allowed to explain away. The real lesson is that the theorist of language change, and any balanced practitioner of advanced research in this exact ing domain, must each strive to place himself above sheer "skeletal" grammar and above the entertaining attempts to flesh out etymologies as well. (i) "Multi-conditioned sound change and the impact of morphology on phonology" (1976) represents an attempt to draw up a decennial balance sheet, after years of rather intensive research which started with my contribution to a Texas Symposium ("The inflectional paradigm as an occasional determinant of sound change" [1968; oral presentation April 1966]). It turned out, in light of subsequent findings, that the claims originally staked out were — counter to the usual sequence of events in such forward-looking developments — too weak rather than too strong. Of the two characteristic processes focused upon in 1966-68, it is only the second, namely the sporadic monophthongization in late Old Spanish of the two rising diphthongs ie /je/ and ue /we/ to і and e, respectively, that receives a second hearing. In the meantime, a microscopic inquiry into the vicissitudes of the Old Spanish preterites has laid bare the long rivalry between the two models (a) 4, -iste,. . . ֊imos, -istes,. .. and (b) -i, -ieste,. . . -iemos, -iestes,. . . which ended with the almost complete victory of the former. The ultimate defeat of the segments -iem-, -iest- at the hands of -im-, -ist- would, for the first time, explain why the adjacency of bilabial (and, in their wake, of labiodental) consonants and of clusters shaped like /st/ (including /sp/ and /sk/) exerted such an unexpectedly strong influence on the monophthongization of ie (and, by way of a "sympathetic" or near-sympa thetic evolution, also of ue). For all the progress achieved along this line, and along several related front lines, the paper here included is only a midway report on what promises to amount to a break-through. A considerably more detailed analysis of the shift ue > e, long held to be enigmatic, was presented to the September 1978 convention ("Fachtagung"),held under the auspices of Vienna University, of the Indogermanische Gesellschaft and is to appear shortly in its transactions, under the title "Etymology as a challenge to

INTRODUCTION

13

phonology: The case of Romance linguistics" (where it shares the platform with the cases of one heavily overlaid sound change, one falsely inferred sound change, and one change still indeterminate — a rather motley array). Another prong pointed in an unexplored direction has been ue > dial. u, beside the more familiar ue > e, just as ie > e and ie > і for a while coexisted under more neatly delimited sets of condi tions. Justice is done to this newly emphasized dimension in the middle-sized paper (awaiting publication in Romance Philology): "The fluctuating intensity of a 'sound law' ". Finally research is under way on Old Sp. (con)migo 'with me', coñusco 'with us' — as against Old Ptg. mego, migo, Old It. meco, etc. to show that the change which, in the past, has usually been laid at the door of metaphony (i.e., in the last analysis, of a phonological process) actually represents the ultimate reverberation of the victory of mi < (dat.) MIHI over me < (acc, abl.) ME, etc., i.e., a purely mor phological shift rooted in a leveling process. From all these studies — in part parallel, in part interlocking, in part consecutive — the reader is encouraged to infer the message that morphology no longer deserves to be treated as the handmaiden of phonology. (j) "Multiple versus simple causation in linguistic change" (1967) starts out from the disappointing discovery that many once promising discussions have ended in a blind alley, as a result of a stalemate between opposing views or of the complete abandonment of a long-beckoning solution of a problem, mainly because "simple" answers to a challenging issue were reputed to be more elegant than "complex" answers, or ideologically more cogent. To the extent that this paper, on the side of exemplification, busies itself with the — often obscure — Spanish adjectives in -io, it slightly overlaps with one of the excursuses attached to (a). Further relevant ideas on Sp. -io < -IDU have been formulated in Steven N. Dworkin's Berkeley dissertation, which I had the pleasure of directing, and in a recent article from his pen carved out from that unpublished monograph ("Derivational transparency and sound change; the two-pronged growth of.-IDU in Hispano-Romance", Romance Philology, XXXI:4, 605-617). (k) By way of exception from the general norm that inordinately technical papers, of possible appeal solely to specialists, should be barred from this miscellany, the unswervingly narrow-gauged piece "The five sources of epenthetic /j/ in Western Hispano-Romance; a study in multiple causation" (1969), was, after some soulsearching, included for the sake of the broader implications of the chosen problem, hinted at in the subtitle. This paper, riveting attention neither to Spanish nor to Portuguese, but ֊ predominantly — to "intermediate" Asturo-Leonese material, addresses itself to one of the core issues in all manner of historico-comparative in quiries, a bent of curiosity transcending the boundaries of Romance: Why does a given change occur in one area but not, despite the apparent similarity, even nearidentity, of conditioning factors, in another? The reason for the baffling disparity, with special reference to epenthetic / j / , seems to lie in the varying conjuncture of several operative forces, some of them clearly visible, others minor and, to the hurried and inexperienced eye, almost imperceptible, that in the aggregate may suffice to produce that elusive change. Among other targets, the paper aims to show how splendid data collections, interspersed with explicative comments, such as those that Ramón Menéndez Pidal has handed down to his successors, can be put to

14


excellent use by a somewhat differently oriented generation of linguistic analysts. (1) "On hierarchizing the components of multiple causation" (1977) goes just one step, but a decisive step, farther along the same promising road. If it is true, as a chorus of affirmative responses to earlier questions appears to substantiate, that appeal must, again and again, be made to the assumption of multiple causation, how can the causes so implicated be organized into a single hierarchic formula? How does one handle principal, in relation to concomitant, factors? Here practically everything remains to be thought through anew and to be formulated with a preci sion unanticipated by our predecessors. By way of mere "appetizers", to whet the initiative of future researchers, as it were, two problems are sketched out: the history of the Spanish preposition hasta 'until' (Old Sp. ata, fata, fasta; Ptg. até), a severely isolated Arabism in this word class; and the Romance adverbial element -ment(e), -mient(re), initially a compositional ingredient, later metamorphosed into a derivational suffix. Whether I have succeeded or failed in my attempt to suggest an acceptable hierarchy, certain specifics in the long-winded history of hasta were later elucidated, in quick succession, through two far more circumstantial investigations - unfortunately, overlapping in part: "Español antiguo des(de), fa(s)ta, fazia y fascas", Homenaje a Julio Caro Baroja (Madrid, 1978), pp. 711-733; and "Problems in the diachronic differentiation of near-homophones", Language, LV:1 (1979), 1-36, see esp. 16-31. (m) A study, on prima facie evidence, of an episode in the history of linguistics, the piece entitled "Conflicting prosodic inferences from Ascoli's and Darmesteter's 'laws'?" (1975), concomitantly also involves a deep bow in the direction of modern accentology. Strictly speaking, this is a distressingly late clarification of an un acknowledged misunderstanding in the dialogue between two intellectual giants, a century or so ago, which should have materialized but never did. Ascoli's "law" bears on the sloughing-off of an unstressed initial syllable containing just one front vowel, typically e-, the alternative being the expansion of that e- into something more resistant to attrition (say, en-). In testing the agency of this "law" in various Romance languages, the great Italian scholar either failed to notice, or, if he became at all aware of the resultant situation, refused to take into account, the all-important fact that Northern French, en bloc, did not participate in this development. Darmesteter, on the other hand, in observing, on the occasion of his celebrated 1875 paper, that in Old French any word-initial syllable was countertonic, regard less of its distance from the main stress, overlooked the equally conspicuous fact that this state of affairs was peculiar just to the chosen language and by no means extended to its congeners. If one now remembers that the langue d'oïl was the most heavily Germanicized among all Romance tongues and that the stress, in Germanic, ordinarily hits a word's opening syllable, then the apparent contradiction can be easily smoothed over: The "Lex Ascoli", at the critical juncture, simply did not work in a Romance language thoroughly alienated from that family's major rhythmic tradition. (n) "Etiological studies in Romance diachronic phonology", which dates from approximately the same period (actual publication date: 1974), is a complexly structured article bearing, cross-linguistically, on several branches of Romance, with the principal stress falling, for once, on Italian. It comes close to being a min-

INTRODUCTION

15

iaturized book, and gains from being read in conjunction with Edward F. Tuttle's important elaboration, published almost immediately thereafter: "Sedano, senero, prezzemolo and the intertonic vowels in Tuscan", Romance Philology, XXVII:4 (May 1974), 451465. My own article, I repeat, attacks a wide range of classic issues in Romance diachronic phonology, especially the idiosyncratic development of intervocalic vowels in Tuscan (for which multiple causation, with full attention to the influx of Hellenisms and Gallicisms, to morphological interferences, to strictly phonetic circumstances, and to structural features, is argued) and an elusive pattern of consonant dissimilation, again in Italian, namely r. . .r yielding either to r. . .d or to d. . .r. Sundry Hispanic problems also receive their share of attention, including the tendential reduction of word-initial, unaccented /kwa/ to /ka/ and of /je/ to /e/, certain striking sound developments in composition and in suffixal derivation, and the particularly challenging, at first glance inexplicable occasional transmutation of V. Lat. /dj/ into Old Sp. /ts/ spelled ç, as in baço, hoçar, raça, also in vergüença. The central point of the argument, as it was developed in the early 'seventies, was the theretofore overlooked (or underestimated) circumstance that in practically all instances alternative solutions, involving the more readily understandable shift /dj/ > / j / , were also on record: bayo, rayo, vergüeña, an ingredient of the total picture which invites a new and bolder interpretation of the ç, in a framework approaching the sociolinguistic model. Details apart, the recurrent idea that ties together the fairly numerous and outwardly heterogeneous elements of this study is, as the title announces, the belief that sound changes, whether or not purpose fully initiated by the given speech community — a problem to be separately investi gated by teleologists —, at the very least invite a close analysis of their demonstrable causation. (o) Of the three studies devoted well-nigh exclusively to suffixal, or assumed suffixal, derivation, the first is marginal to the overall purpose of this collection, inasmuch as "One characteristic derivational suffix of literary Italian: -(t)aggine", composed for a special occasion (1976), hardly carries any major message for the generalist. Its minor aim is to follow in its course, across several centuries, a semilearned suffix of diaphanous Latin provenience, an element which developed splen didly in a single Romance dialect (and, through that dialect's cultural prestige, before long spilled over into a major literary language), but failed to get off the ground in the cognate languages and did not lend itself to secondary diffusion, through borrowings, either. Because there exist comparable situations elsewhere (e.g., -ez in Spanish, -ice in Portuguese, -ise in French, etc.), the Italian-oriented paper offered here, if it meets with approval, may serve the purpose of exemplifying a transferable technique, but stops short of spelling out any underlying theory. (p) The succinct, similarly unpretentious note, "The double affixation in Old French gens-es-or, bel-ez֊or, Old Provençal bel-az-or", which actually appeared one year after the official publication date (1973) of the issue of the Swedish periodical that hosted it, concerns itself with the phenomenon of suffix chains, and with the related problem of non-final suffixes, for which the newly-minted term "interfixes" was elsewhere proposed and championed at some length. So much for the note's possible appeal to the generalist, who may be in search of a new, finely-graded inventory of categories of affixes, with appropriate exemplification. The specialist

16


may find in this piece some shreds of information on the next-to-last convulsions, on Gaulish soil, of characteristic Latin "irregular" comparatives in -(I)ŌRE; on their use geared more often than not to suppletion; and on the corresponding adverbial formations. (q) "The rise of the nominal augments in Romance" (1972) is an attempt to come to grips with an exceptionally elusive morphological element in Hispano-Romance. The difficulties this element poses are of two kinds: first, there is no consensus among scholars on how to categorize it; and second, no truly convincing hypothesis has yet been advanced concerning its provenience and apparent limitation to just one subfamily of Romance dialects. The paper aims at grappling with both issues (which, for all we know, may interlock), one of which pertains to general and the other to particular linguistics, in diachronic projection. The starting point for research along both axes is Menéndez Pidal's landmark article of 1905, "Sufijos átonos", whose tenets and terminological choices, perhaps felicitous for that distant period, have been repeated by epigones over and over again. The Madrid scholar himself returned to the fray on two occasions — in 1926, then again in 1953 — adding a profusion, conceivably an excess, of fresh factual details, but hardly improving, in the process, the over-all picture. Thus, the challenge for myself was to achieve some sort of break-through where the best-informed expert, after a half-century of gropings, had run into a blind alley. The tag "suffix" (i.e., 'derivational suffix') was in itself suspect, first, because, except for a small group of non-final suffixes (alias "interfixes" or "empty morphs"), such an element, by definition, is expected to convey a semantic message — as, e.g., the -lo of obs. murciégalo 'bat' in lieu of older murciego, lit. 'blind mouse', defini tely does not — and second, because at the Romance stage, though not necessarily in the parent language, such suffixes were consistently stressed, so that "unstressed (derivational) suffix", in Romance context, becomes something of a paradox. The alternative label "augment", here advocated, minimally has the advantage of not blurring the dividing-line between separate categories. Romance augments, inci dentally, come closest to Germanic "excretions" (as in G. Ax-է 'ax', Habich-t 'hawk'): In both situations speakers seem to delight in certain word-final segments, -lo in Spanish and -xt, -cht in German. The genetic facet of the question is far more difficult. If in 1905 Menéndez Pidal was non-committal on that score, by 1953 his thinking was irreversibly tilted in the direction of substratal ("palaeo-Mediterranean") extraction, and he adduced alleged parallels from other, extinct circum-Mediterranean languages ֊ an extremely hazar dous game, given the number of unknowns in each such context. As the subtitle of my own article discloses ("Graeco-Latin and Tuscan clues to the prehistory of Hispano-Romance"), my preference has led me to a hard kernel of Hellenisms in Romance characterized by a "dactylically" structured word-final segment: -'--. The vowel in the penult was often, but not mandatorily, a - depending to some extent either on the vocalic configuration of the model word or, as in Tuscan, on an interplay with the following consonant. The sweeping generalization of the central vowel was a characteristic symptom of the broad leveling process that presided over the genesis of Spanish in numerous respects. (r) "Identification of origin and justification of spread in etymological analysis"

INTRODUCTION

17

(1969) is conceived as a sharp reminder of a fact only inexplicitly acknowledged, namely that a scholar, in proposing a new etymological solution, owes to his readers two separate, if often interwoven, statements: one on the rise within, or the pene tration from without, of the etymon invoked; the other - equally important — on the viability and ultimate triumph of that lexical unit, within a certain sphere. After all, thousands upon thousands of word blends, or compounds, or derivatives, etc. spring into ephemeral existence, day by day, in every characteristic speech com munity, only to be contained in their exiguous socio-spatial habitat. The diffusion, i.e., success, of such a form may depend not only on its inherent qualities (pleasing syllabico-accentual structure, flattering or amusing semantic associations, and the like), but also on certain ailments (e.g., harmful homonymy) afflicting the standard forms prevalent until then and subject to instantaneous replacement. Illustrations are provided chiefly from Hispano-Romance sources. (s) A study which, prima facie, is little more than an elaborate inquiry into a trinuclear lexical family, "Primary, secondary, and tertiary etymologies: The three lexical kernels of Hispanic saña, ensañar, sañudo" (1974), turns out to be actually, in the first place, a contribution to etymological theory and technique, questioning as it does the widespread and, in many but not all instances, perfectly justified belief that a single, neatly-structured word family, with a clearly recognizable center ("headword"), is best investigated by following the development of that word across the ages. In a not insignificant minority of cases an entirely different con stellation of circumstances appears to have prevailed. To account for Sp. saña 'fury, rage', flanked by the adj. sañudo 'furious, enraged', and the reflexive verb ensañar 'to fly into a rage, become angry' (all three with Portuguese counterparts using -nh- for --), scholars have appealed to three indepen dent lexical kernels of the parent language: (a) the element SĀN-, which can be safely extracted ("peeled off', in laboratory jargon) from SĀNUS 'healthy, sound', ĪNSĀNUS 'of unsound mind, mad, raving', INSĀNIA 'madness', SĀNĀRE 'to heal', and INSANIRE 'to be mad, rage, rave'; (b) SANNA 'mocking grimace', flanked by SANNIÕ -ÖNIS 'buffoon' (cf. also OSp. sosañar < SUBSANNĀRE); and (c) SANIĒS 'corrupted blood' which, predictably, boasted the variant SĂNIA. While the tradition until now has been for every explorer to opt for one of these three seemingly discrete possibilities and to reject, not infrequently with a certain stridency, any alternative, the message of this paper is, on the contrary, to plead the assumption of a certain confluence, at the time level either of Latin or of Pro to-Spanish, using the relations of -N-, -NN-, and N + J as one's guideposts and throwing in, for good measure, the commonsensical evidence of semantic development. Thus, sañudo — judging from the grammatical and semantic ambits of the nouns the suffix -udo is indepen dently known to have joined — points to SANIĒS/IA as its starting-point, and its synonym sañoso beautifully fits SANIŌSUS. With the steady growth of INSĀNUS and ĪNSĀNIA, a new verb, *ĪNSĀNIĀRE, may easily have come into existence, overlaying the older INSĀNĪRE. The contact between *(ĪN)SANIĀRE and SĂNIA may very well be traceable to Late Antiquity, since it presupposes solely the removal of one barrier through neutralization of the prosodic contrast between the a's of SĀ- and SĂ-. If SANNA was allowed, on Spanish soil, to survive into the Middle Ages, it would have emerged as *saña in the vernacular and, with its meaning of

18


'mocking grimace', might have smoothly been attracted into the orbit of the sanwords (meanwhile firmly coalesced into a single family). If this reasoning is correct, we have before us, in all three instances, partially correct and mutually complemen tary and corroborative solutions of the one vexing problem on hand. This article overlaps with a number of other exploratory papers. As regards the occasional diagnostic value of a derivational suffix glued to an old and important member of a word-family under scrutiny, one may be excused for citing the analysis proposed for Fr. feu 'late, deceased' < OFr. feü, faü, traceable to a base in -UCU rather than -ǕTU (see "Ancien français faü, feü, malostru", Mélanges Paul Imbs, 1973). With respect to a single word's, or a single word-family's, roots in two dif ferent sectors of the given lexicon, viewed at an older stage, attention can be drawn to a paper jointly written by members of a Berkeley group of etymologists, includ ing myself as its senior leader: "AEQUUS versus (IN)GENUUS; etymological studies on Old Spanish (y)engo 'free', (y)e(n)guedad 'freedom' " {Romance Philology, 1978). On the negative side of the ledger, this paper, painfully enough, highlights one deeply deplorable lacuna, namely the continued absence of any in-depth in quiry into the coexistence of, and eventual conflicts between, the two Latin pre fixes IN-, the one illative, the other reversive. (t) "Etymology and modern linguistics" (1975), unlike the remainder of the studies here collected, was, from the start, designed for oral delivery and, as a result of such planning, aims at an incisive presentation, not to say dramatization, of an issue usually hushed up: the divorce between front-line, and even mainstream, linguistics, on the one hand, and, on the other, the etymologically-oriented rearguard. This distribution of roles may accurately reflect the present-day capricious state of affairs, but its wisdom is scarcely borne out by any stretch of non-partisan thinking. The paper starts out with an almost random survey of reactions, by influential shapers of "public opinion" in linguistics, from nineteenth-century pioneers to the late Uriel Weinreich, to the burning issue of (general) linguistics vs. etymology. The gradually declining interest of many leading figures in the filigree work pro duced by several etymological workshops is noted, but by no means endorsed. The discussion then shifts to the status of etymology within the ensemble of linguistic disciplines, with some attention to its fluctuating relevance against the background of "strong" vs. "weak" phonetic change. The concluding pages concern themselves with two legitimate categories of etymological puzzles that deserve to be called relevant or "interesting", in the sense that mathematicians are wont to confer on the latter qualifier. Once more one detects numerous overlaps with earlier or contemporaneous studies, including one written in French for a comparably wide audience: "Deux catégories d'étymologies intéressantes", Revue de linguistique romane, 1975. Re garding the measurement of "strength" in a sound shift, a field still wide open for debate (but merely a side-issue in this particular context), see my short initial state ment (1962) in the Mélanges W. A. van Groot {Lingua, Vol. XI). For the circum stantial analysis of Sp. terco 'stubborn', whose results are here made the corner stone of the basic categorization, turn to a paper as far removed in time as Vol. LXIV of the Publications of the Modern Language Association, 1949. As a whole, this 1975 piece carries on the discussion initiated, perhaps under a luckier star,

INTRODUCTION

19

almost twenty years ago, in a key issue of the then flourishing journal Word, Vol. XVIII. (u) Probably no single paper in this miscellany is more programmatic as concerns the prevailing bifocal approach explicated and championed above, than the lengthy item "The interlocking of etymology and historical grammar", with the soberingly narrow subtitle: "Exemplified with the analysis of Spanish desleír" (1979). The entire article, undeniably, qualifies as an exceedingly detailed, not to say almost too circumstantial, account of a single etymological controversy, concerning the ulti mate source of the Spanish verb desleír 'to dilute, diffuse', as spanning a period of three centuries. There exist, all in all, eight to eleven major conjectures, depending on one's criteria of selection and extending from the guesswork of 17th-century trail-blazers (such as S. de Covarrubias and F. del Rosal) to certain pronouncements by undeniably sophisticated analysts of our own age. Upon closer inspection, it turns out that three seasoned 20th-century practitioners of Romance linguistics (Menéndez Pidal, Carda de Diego, and Corominas) went wrong in their etymologi cal diagnosis because they failed to heed various finer points, easily overlooked, of historical grammar. The one founding father of Spanish philology who, in retro spect, seems to have had the right hunch happens to be Rufino José Cuervo (1893), who opted for, possibly, the least immediately-persuasive ancestry, namely Lat. DĒLĒRE 'to destroy'. It can be demonstrated, through patient sifting of the evi dence, that every single intermediate step which must be posited before one dares to accept Cuervo's bold conjecture harmonizes splendidly with the trend of events painstakingly reconstructed by historical grammarians. Many of the difficult points with which the problem seems to bristle concern morphology (conjugation classes, prefix change) rather than phonology. (v) "Contacts between BLASPHĒMĀRE and AESTIMĀRE" (1976) is a note attempting to explain some of the deposits long deemed "inexplicable" of BLAS PHĒMĀRE 'to curse' (e.g., Sp. lastimar 'to wound') on the assumption that its semantic opposite, AESTIMĀRE 'to esteem', acting as the positive pole, in learned transmission significantly influenced its form. If this surmise is correct, then the ideas rather sketchily presented in "Lexical polarization in Romance" (see Language, 1951) would seem to be vindicated. The entire discussion of the phenomenon invites thorough re-working, not least in light of this latest proposal. In an excursus an attempt is made to claim Sp. Ptg. tomar 'to take' for (AES)TUMĀRE, a wide spread variant of AESTIMĀRE, conceivably with some help from severely isolated AUTUMĀRE 'to assert, aver' and with surgical removal of AES 'copper (money)'.

VII. It now remains for me to thank the many persons who, in one way or another, have made it possible for this complexly structured book to come into existence. Obviously, I remain beholden to those managing editors of learnèd journals and executive secretaries (or members of steering committees) of scholarly meetings who, in the first place, extended their hospitality or sponsorship to the papers here assembled. I refrain from supplying any lengthy "complete roster" of names, as-

20


suming one can at all be reconstructed at this distance from the events: If I may be selective in my statements of appreciation, I would like to single out for mention, among those friends and colleagues unfortunately no longer alive, my Scandinavian peers Bengt Hasselrot in Uppsala and Knud Togeby in Copenhagen. I am further indebted to those business managers of publishing houses which held the copy rights to these papers for freely authorizing me to make use of this material for the purpose of the present miscellany. The idea of assembling a not inconsiderable number of articles marked by a certain easily recognizable leitmotiv came to me, it seems, approximately two years ago, and I ventilated it first upon making the personal acquaintance of John Benjamins in Hamburg, early in September 1977, on the occasion of the Third International Congress of Historical Linguistics. John and Claire Benjamins have ever since gone out of their way to translate a dream into a realistically planned project, and they have shown touching patience in the face of certain enforced procrastinations on my part. John W. M. Verhaar has displayed initiative and com mendable understanding, in collaboration with Werner Abraham, by allowing this particular collection of articles to be absorbed into the prestigious "Studies in Language Companion Series", after having previously invited me to contribute a paper to the opening volume of the newly-founded journal, Studies in Language. Much of the basic research, of the necessary revisions and elaborations, of the - in part, unavoidably dreary and technical — correspondence, and of other routine operations could not have been so promptly and accurately carried out without the constantly cheerful support that I have been privileged to receive from my talented research assistants. In this context the names of Prudence Ashley, Anne Donaker, Iisa Moore, and Candida Silva come to mind at once. All in all, this has been a very gratifying experience, and one from which I have learned a good deal.

Berkeley, March 30, 1979

A. GENETIC LINGUISTICS

LINGUISTICS AS A GENETIC SCIENCE*

I am designedly using the term 'genetic' rather than 'historical' in this context because, after having been for thirty years a devotee of the discipline conven tionally labeled 'historical linguistics', I have at present grave misgivings about the unqualified suitability of the label and, far more important, I have become alert to the implications of its possible inappropriateness. To begin with, the label 'history', like many terms of ancient scholarship, is fraught with imprecision. It refers to an analytical discipline concerned with the study of the past, but also with a segment of that past regardless of any analysis, as long as there exist any written records. Only thus can we explain the fact that prehistory designates, at least in normal usage, another, earlier segment along the time axis, whereas such terms as prelinguistics (which may or may not have been coined) and prephilology (which has actually been toyed with on one occasion) could meaningfully refer only to a prescientific stage of a discipline; that is to say, to cognition itself, not to an object of cognition. My second source of doubt is the awareness that any serious work of history involves a number of commitments on the part of the practitioner not only as to the purpose, scope, and techniques of his research, but also as to the forces credited with causing change. This hard core of curiosity about causation in general, with special reference to evolution, must, of course, be shared not only by straight historians, i.e. students of broadly significant events in the succession of human societies, but also by scholars combining a special topical expertise with a flair for sequential reconstruction—say, art historians or historically oriented analysts of jurisprudence. To my surprise, I discovered before long that, despite the familiarity of many old-style linguists with an astonishing profusion of historical details, very few among them operated explicitly with any clearly delineated philosophy of history. True, Ēmile Littre, the late-19th-century author of a massive dictionary of Modern French, had, at the outset, a generous share of close contacts with Auguste Comte's positivism; also, Otto Jespersen's disputed optimistic belief in the steady improvement of language as a tool— witness his book Efficiency in linguistic change—implied a sharply silhouetted world view, as one would expect of the rationalistic author of a publication en titled The philosophy of grammar. But, generally speaking, among the older linguists addicted to the historical approach, particularly those immersed in etymology, it is easier to find individuals with a strong side-interest in various facets of material civilization, in navigation and caravan routes, and even in botany (cf. Vittorio Bertoldi), ichthyology (witness Paul Barbier), and orni thology (I refer you to Richard Riegler) than persons with any pronounced * Substantially unaltered text (except for the added excursus on primary vs. secondary determinants) of the Presidential Address delivered to the Linguistic Society of America, in Chicago, on 28 December 1965.

24

GENETIC LINGUISTICS

flair for, let alone systematic training in, etiology—the name under which the theory of causation is known in philosophical circles. The third doubt that has assailed me relates to a curious paradox which bids us revert to the aforementioned distinction between history and prehistory. Students of evolutionary linguistics are in the habit of complaining almost ritualistically about lacunae in their information, but, strangely, such problems as suffer from few gaps offer them no real challenge and are thus unlikely to attract many first-rate minds. There exists almost an excess of data on the transi tion of written Latin from the Golden Age to the Silver Age, and the sociocultural background, for once, is amenable to minute inspection; but such a topic, for all its philological legitimacy, falls flat among red-blooded linguists, as would the study—equally defensible yet equally unexciting—of, say, English terms of sports and athletics recently absorbed by miscellaneous modern languages. I t is prehistory, with its conjectures so rich in hazards, that titillates the daring linguist's imagination; and this proclivity again makes the term 'historical linguistics' something of a misnomer. A fourth idiosyncrasy of evolutionary research in linguistics must now claim our attention. Peculiar to the data-gathering process in historiography is the gradual concentration on details and a concomitant attitude of tactical meticulosity which measurably slows down strategic progress and therefore makes genuine breakthroughs increasingly rare. Through a cruel irony, it is the honest, patient, scrupulous, erudite worker who is most inhibited by his knowledge of residual gaps in the information garnered and by his excruciating awareness of any severe limitations of his personal competence. Leaving aside such talented amateurs as H. G. Wells and cynics whom I prefer to leave unmentioned, the rare professional historian attempting a bold synthesis based on a new theory of cycles—I have in mind men as gifted and hard-working as Spengler and Toynbee —must reckon with an investment of thirty-odd years of sustained effort and unstinted labor, i.e. practically a lifetime spent on a single project. The rank and file of workers infected with historical curiosity, however, shy more and more away from such audaciously architectured edifices and prefer to earn polite recognition if not enthusiastic acclaim through rigorous inquiries into paro chially delimited problems. Thus there unfolds before our eyes a world of shrink ing horizons, in which monumental structures are allowed to crumble, and only compact, tightly built cabins seem safe enough to survive. I submit that this kind of attitude of apprehension, carried to an extreme, can have a stifling effect on evolutionary linguistics and that, to avert lasting harm, we must counter balance the ever-present temptation of dêtaillisme by carving out, from the over flowing mass of raw data and of minute accurate findings, such salient discoveries as lend themselves to wider generalization and higher abstraction, thus constitut ing a welcome feedback into the main stream of general linguistics. It is on this battleground, throughout the last half-century, that descriptivists of many hues have had all the advantages, and evolutionists, in the guise of historians, all the drawbacks. Historical linguistics, to be sure, has its continued justification for specific painstaking probings into unique societal settings—a slant of study to whose charm, for well-known sentimental reasons, most 19th-century scholars

LINGUISTICS AS A GENETIC SCIENCE

25

succumbed and not a few of their 20th-century successors continue to respond, and which has, I am confident, a future. But to ensure that future, which pre supposes increased maneuverability, we must learn to isolate a core of theory and exemplification—if necessary, at the cost of studied restraint and even of painful surgery—and establish a discipline concerned with the universals, as against the particulars, of change. Whether we decide to call it glottodynamics, or evolutionary linguistics, or genetic linguistics, is a matter of expediency and esthetic preference. I realize that in life sciences, genetics, concerned with neatly foreseeable hereditary features, and evolution, focusing on mutations far more difficult of prediction, are mutually opposed—in some quarters, almost to the point of irreconcilability; but this contrast seems not to apply to the growth of language. There need occur no divorce from conventional historical linguistics— in some instances, only the friendliest of separations, with visitation hours, will suffice; in others, still better, the alternate performance, by the same versatile worker, of both assignments will maintain a most desirable liaison, as when Emile Benveniste has been cultivating the two perspectives, before our eyes, in an elegant zigzag movement. One supervenient reason for favoring 'genetic' over 'historical' is that the latter label applies, by definition, to the past, whereas the former, in Janus-like fashion, points to events both behind us and ahead of us. This statement implies no simplistic belief in any easy predictability of approaching trends of develop ment bearing on language. Unlike his naturalistic counterpart, the linguistic geneticist will perhaps be well-advised to curb his optimism, confining himself to the hope that from the observation of the past and of the present he can try to extrapolate one ingredient or streak of the foreseeable evolution, an ingredient whose effects can be reinforced, weakened, neutralized, or in extreme cases even reversed by other elements either utterly imponderable at this state of knowledge or, at best, tentatively assessable from the vantage points of disciplines other than the study of language. Even if that much ground has been gained, the genetically oriented linguist, unlike the straight historian and quite unlike the orthodox descriptivist among his confreres, may well cease to protest, in an almost morbid recoil from personal involvement in reality, that he is eager to observe, to record, and to analyze, but not, under any circumstances, to in fluence events. Perhaps we can transcend this long-dominant attitude of impas sibility bordering on passivity by arguing that at the moment of actual dissection the analyst should in fact declare himself totally uncommitted but that, at the immediately following stage of evaluation, he is perfectly free not only to draw forceful inferences as to certain factors emerging as potentially influential, but also, if circumstances so warrant, even to throw his weight behind shifts and movements deemed desirable—if not qua linguist stricto sensu, at least qua linguistically enlightened and chastened layman. The substance of what I am propounding (as against certain emphases placed conceivably for the first time) is by no means new, and the very word 'genetic' was sporadically applied to linguistic science by some mid-19th-century workers —witness the Vocabolario genetico-etimologico delta lingua italiana by G. B. Bolza, traceable to the year 1852. More important, the implications of this

26

GENETIC LINGUISTICS

approach were felt on both sides of the Atlantic throughout the decades separating, on the one hand, the concluding phase of W. von Humboldt's activities and their immediate posthumous reverberations and, on the other, the temporary consolidation of neogrammatical theories and practices around the mid-'eighties. A. Schleicher's organic view of the development of language has been attrib uted, in conventional retrospect, to the vogue of Darwinism. In more careful re appraisals, it has recently been observed that crude or refined evolutionism, by the mid-19th֊century, was somehow in the air, nourishing initially both Darwin's biological inquiries and Schleicher's parallel thinking along a less sensational line. In the final elaboration of his technique and terminology, Schleicher could, of course, have freely profited from the powerful impact of such universally in fluential works as The origin of species by natural selection (1859); he did not live to see the publication of The descent of man (1871). As happens not infrequently when scholars toy with the wisdom of a fresh start or grope for new shapes of lingering ideas, those of us eager to see diachronic linguistics rehabilitated must, among other things, strive to reassess some earlier judgments which, by dint of repetition, have become rigidified but not necessarily unassailable. If, in casting a glance at 19th-century linguistics, we allow our selves to be guided exclusively by Holger Pedersen's well-known book, readily available in a tasteful English translation, we may easily obtain a skewed, grossly distorted picture. Pedersen, a seasoned Celticist and an expert Indo-Europeanist, was swayed in his retrospective value-judgments by what—on the scale of the early 'twenties—seemed most beneficial for certain types of Indo-Europeanist reconstruction; though he exceeded the Indo-European domain in sheer scope of coverage, he fell short of transcending it in methodological preference. Through a queer and inexcusable caprice, he also abandoned a solid Danish tradition stretching from Vilhelm Thomsen to Viggo Brøndal and beyond in completely slighting Romance linguistics. By dividing all of 19th-century scholarship into two parts and by assigning to the first the experimental writings of a Bopp, a Rask, and a Schleicher, while reserving for the latter the technically far superior, but in other respects distinctly narrower analyses of the leading neogrammarians, he conjured up an image of progress which, I think, can and should be chal lenged unless one implicitly agrees to equate certain favored chapters of I E historical grammar with linguistics as a whole. For while there indisputably occurred a change around 1870, that change carried with it not only an undeniable increase in precision and refinement, but also numerous retrenchments and a withering of curiosity along several once promising lines. With regard to liber alism, I am tempted to compare the third quarter of the 19th century in Europe and, thanks to William Dwight Whitney, in New England to the intellectual climate in this country throughout the 'twenties and especially the 'thirties, when linguistic research, as symbolized by Edward Sapir's many-splendored and imaginative if disciplined gropings, was carried out in freedom from the op pressive constraint so unpleasantly characteristic of the 'forties and the 'fifties. I t was only in the last quarter of the 19th century, which I venture to regard as a period not only of triumphant advances but also of tacit retreats, that the concept of historical linguistics, with its full scale of implications and connota-


27

tions, began to crowd out the concept of genetic linguistics. Perhaps the turning point, on the semi-popular level, was Hermann Paul's long-celebrated and tonesetting Prinzipien der Sprachgeschichte (1880; enlarged 2d ed., 1886). What sets off, by no means disadvantageously, the pioneer geneticists or evolutionists from the bulk of 19th-century language historians is a set of qualities such as these: first, greater concern with, and keener understanding of, universals of linguistic change, sometimes achieved—let us candidly admit—at the cost of premature generalization; second, sustained curiosity about fluctuations of meaning, with no concurrent neglect of changes of form; third, superior openmindedness toward non-Indo-European languages; fourth, full recognition of the importance of etymology, conceived not only as a purveyor of dependable equa tions for historical grammar, but as a discipline worth pursuing for its own sake; fifth, unabashed delight in modern dialect research and, coincident with this predisposition, growing finesse in the requisite open-air operations and in subse quent analyses of data so obtained. Let us exemplify some of these propensities and, in the process, observe how they mesh and intertwine. For general linguistics the contemporaries of Whitney, moving in the groove traced out by Humboldt, used the term 'philosophy of language'. In calibrating the output of these trail-blazing scholars, one notices a balance between datasaturated monographic researches and methodologically slanted early syntheses, including the two major book ventures by Whitney himself, of the years 1867 and 1875, respectively. The accumulation of austerely monographic probings, unrelieved by experiments in theory and speculation, becomes, in contrast, characteristic of the concluding decades of the century. It may be argued, as I realize only too well, that in certain contexts premature syntheses are harm ful; but that argument yields in persuasive power to the contention that the blunting and stunting of the explorer's imaginative drive may, in the last analy sis, represent the greater evil. As a temporary moratorium, the exclusion of so exciting a topic as the origin of speech from the deliberations of a prestigious linguistic circle may be understood and condoned; but, in sober fact, the shelving of a problem of this magnitude and intensity of appeal marks not a victory, but the masked admission of a stunning defeat. As regards heightened concern with meaning, Michel Bréal's Essai de séman tique inevitably comes to mind, a book whose publication date—1897ât first glance removes it from the early genetically biased period of linguistic science. But if one recalls that the author was born in 1832 and had published, by the early 'sixties, four monographs bearing on Zoroastrian religion, comparative mythology, and Persian anthroponymy, it is plausible to assume that the book's period of gestation pertains to a distinctly earlier, pre-neogrammatical period. Rather characteristically, Bréal's first translator and annotator, J. P. Postgate, confesses at the very outset of his lengthy Preface, dated 1900, that as a young man in 1877 he had selected the science of meaning for the subject of a 'Fellow ship dissertation' at Trinity College, transforming that essay almost twenty years later, but before Bréal entered the arena of semantics, into an inaugural address at University College, London. The radiance surrounding the activities of a startlingly precocious Ferdinand de Saussure during his Parisian period

28

GENETIC LINGUISTICS

must not blind us to the fact that there were other remarkable French linguists at work in the capital—scholars unjustly eclipsed, whose belated rehabilitation must be based on the rôle they played as precursors of Bréak I have especially in mind two experts whose names I rarely hear in linguistic quarters. One, already alluded to, was Littré, a companion-in-arms, you will recall, of the positivist Auguste Comte; his early writings include a tract on the electrodynamic theories of Ampère; closer to home, he authored a treatise entitled Comment les mots changent de sens, issued posthumously in 1888. The other was the talented Arsène D armesteter, who in his highly readable Sorbonne lectures at the start of an all too short career, collected two years earlier than the date of the Littré pamphlet and published under the title La vie des mots étudiée dans leurs significations, included the most elaborate version of an ambitious 'théorie du développement des sens par rayonnement et enchaînement', clearly a kind of embryonic semantic wave theory. For the mid-19th-century's impressive ability to straddle IE and non-IE languages—that is to say, for the enthusiastic espousal of the platform of the trail-blazing typologist Wilhelm von Humboldt rather than the program of the early comparatiste Bopp, Grimm, and Rask—I wish to adduce, as a clinching if slightly unexpected argument, the evidence of Romance scholarship. Between the founding father of Romance linguistics, Friedrich Diez—a counterpart, if you wish, of Jakob Grimm—and the vigorous systematizer of that discipline at the threshold of the 20th century, Meyer-Lübke—whose work you are free to view as a pendant of Brugmann's—lies a hazily silhouetted territory dominated by two major figures, Graziadio Isaia Ascoli and Hugo Schuchardt. Ascoli's lasting merits lie in the domain of fine-grained Romance dialect studies, especially the monumental mosaic of Raeto-Romance which he pieced together in his Saggi ladini (1873) and his later characterization of another sharply pro filed dialect cluster which he considered a separate Romance language—the torso of a less laborious monograph embodied in his Schizzi franco-provenzali (1878). But the theoretical conclusions which he drew from the minute inspection of medieval and modern dialects—namely the substratum theory (or, as some Italians call it, 'dottrina delle reazioni allogene') with special reference to Celtic, subjacent in certain territories to Latin and Romance—these important in ferences, transcending his immediate purposes, he would hardly have reached without that independent acute curiosity about Celtic which prompted him, in 1878-79, to examine the Irish glosses of the Codex Ambrosianus. Celtic, again, was to him merely a major prong of his earlier broader advance in the direction of Indo-European, a preoccupation which yielded his well-known—and, I understand, not undisputed—inquiry into the velar series of the parent language, and which also cast off studies in Persian and in Gypsy; while the Indo-European family in all its richness acquired a sharpened contour within his private cosmos through confrontation with Semitic, a step made possible because Ascoli's early contact with Hebrew had awakened in him to desire to become an Orientalist, bidding him to acquire the necessary tools. This is to say nothing of his incidental excursions into Turkish, Dravidian, and Chinese. Similarly, Schuchardt, after taking as his starting point Vulgar Latin, and after


29

presenting the wave theory in a nutshell, with verbal and pictorial illustrations, nine years ahead of Johannes Schmidt's programmatic treatise, branched out from these modest beginnings on a scale unprecedented in the annals of Romance philology, not only appropriating, one by one, such adjacent domains as Slavic and Celtic, but also developing an astounding competence—reminiscent of Trubetzkoy's and Sapir's elasticity—in Basque and in paleo-Iberian; in Hun garian; in Georgian; in miscellaneous Creole languages—amalgams involving English, Spanish, and Portuguese, on the one hand, and West African, Indic, and Malayo-Polynesian tongues on the other (witness his nine sketches published in quick succession from 1882 to 1891); as a separate inquiry, in the speech of Saramakka Negroes transplanted to the Dutch colony of Surinam in the North east corner of South America; in Berber languages; and even venturing, ahead of Otto Jespersen, into the realm of artificial languages such as Volapiik (1888). On a less ambitious scale and without Schuchardt's unique intuitive flair, a scholar almost forty years his senior—in fact, distinctly older than Ascoli— also reconciled his active layman's curiosity about English and Romance dialects with heightened concern about Basque, not entirely dilettantic: I refer here to Prince Louis-Lucien Bonaparte (1813-91), a nephew of the Emperor, who, despite his modest rank as a scientist, deserves mention hic et nunc be it only because his incomparable collection of early dialect monographs—a veritable treasure trove—was acquired quite some time ago by Chicago's Newberry Library, which is within a stone's throw from the place where we are assembled. Over against this slant of performance, Meyer-Lübke, typifying to some extent neogrammatical preferences and standards of craftsmanship at the close of the century, displayed commendable industry, competence, and honesty in turning out, again and again, either historical grammars or articles conducive to such grammars ; in so doing he seldom abandoned the sheltered Romance hunting ground and practically never mustered sufficient audacity to strike out beyond Indo-European. My fourth point, the uninhibited enjoyment of etymology for its own sake rather than merely as an ancillary operation, stands somewhat apart. Undeniably the merit and sheer necessity of sound etymological analysis were quite unapologetically recognized by the pioneers. The emerging custom called either for the same scholar to produce, as mutually complementary repositories of knowl edge, a grammar and a dictionary, in launching a two-pronged attack on a language or a family of languages, as practiced by Grimm and Diez ; or for one expert's grammatical edifice to flank another's string of etymological vignettes, witness the experiments of opp alongside those of Pott. I can think of no plea in defense of etymology more eloquent than Whitney's in the concluding chapter of his book The life and growth of language; let me quote from this passage just a few lines: The whole process of linguistic research begins in and depends upon etymology, the tracingout of the histories of individual words and elements. From words the investigation rises higher, to classes, to parts of speech, to whole languages. On accuracy in etymological proc esses, then, depends the success of the whole; and the perfecting of the methods of ety mologizing is what especially distinguishes the new linguistic science from the old.

30

GENETIC LINGUISTICS

So far Whitney; his forceful commitment in this matter clashes pathetically with Saussure's and Bloomfield's baneful inarticulateness. It is true that the summit of etymological escalation was reached only several decades later, at the turn of the century, after the first flowering of dialect geography and the crystallization of the Wörter-und-Sachen approach spearheaded by Meringer and Schuchardt. But it is equally true that the then novel techniques which gave such powerful impetus to 20th-century etymology were nourished by the spadework of freewheeling mid-19th-century evolutionists (or geneticists) far more than they were by the gradual retrenchments of neogrammarians, for all the superiority of the latter group's organization. If one were to press into a single formula the chief services that a grand strategy of lexicology, including its etymological kernel, renders to linguistics, beyond its humble illustrative and exemplifying function, I am tempted to propose that it be recognized, first, as the main artery of communication with the ceaselessly changing hard facts of the outside world and, second, as a safety valve for the study of individual, especially of abnormal, growth—for matters, that is, which stringent grammarians eager to concentrate on patterns tend, for the sake of economy or tightness, to sweep under the rug. Yet even the most austere, grammatically oriented linguist is unlikely to deny that erratic configurations sporadically scattered over the surface may conceal invaluable clues to older and even pristine patterns, partially overlaid. Thus any prolonged withdrawal from etymology, under whatever pretext, threatens to produce the blocking of vital supply lines and, in the end, the total inanition of historical linguistics. The fifth and last salient trait of the trail-blazing researches, before the various lines of curiosity hardened into a fixed hierarchy, was the workers' active concern with dialect speech. Of course the records prepared were clumsy, the transcrip tion used inadequate, the underlying theoretical assumptions—if any—often naïve; but openmindedness was clearly in evidence. The later, far more elaborate dialect monographs and the currently accepted canons of fieldwork have their ultimate roots, I believe, in the mentality or, if you wish, attitudes of the scholars toiling approximately a century ago. It is not my intention here, let me emphasize, to extoll the somewhat dimmed merits of one generation of linguists and of their lineal descendants, pushed into the background, while belittling (or attempting to cut down to size) the accom plishments of a more aggressive subsequent generation. The real point at issue is quite different and has distinctly greater bearing on some of today's events, both exciting and disquieting. When speaking of progress in linguistics, we can use this term with unalloyed pride only where one technique supersedes another without leaving any residue. To provide a handy frame of reference: the jetplane passenger service from San Francisco to Chicago marks an improvement over the older propeller-plane service because it indisputably accomplishes the same clearly circumscribed task, transporting an individual and his luggage from one city to another, under measurably more favorable conditions, as regards time, boarding procedures, seating, comfort, protection from noise, proc essing of checked suitcases, and en route entertainment—visual and gustatory. In other words, there remains in commercial aviation virtually no residue of


31

pleasurable or useful experiences lost in the course of planned technological replacement. The situation is less clear-cut where ferryboats have been super seded by automobile traffic over bridges, because in that context the strictly utilitarian and the esthetic or sentimental benefits seem to be at loggerheads. In the annals of linguistics there may have prevailed some smooth transitions, without concomitant losses. But even the improvement which for years passed off as the smoothest of all—the substitution of phonemic for phonetic descriptive models—left a heavy residue; and the two shifts from early evolutionism, characteristic of the 1860's, to the strict historicism of the neogrammarians and later to ironclad structuralism entailed, counter to widespread belief, spectacular retreats as well as advances. In mapping out the future of our discipline, it is entirely legitimate to inventory the losses suffered at earlier junctures and easily overlooked in the flush of triumph, so as to reabsorb into the mainstream of linguistics any valuable ingredient that may have gone astray. Before veering away toward the present and the future, let us briefly examine some terminological innovations launched by the early evolutionists and their direct successors. The term SUBSTRATUM—in French, alternately substratum and substrat before the fairly recent standardization of the latter—was reportedly coined in the mid-18th century, to denote a philosophical concept, and it emerged again, after a certain lapse or a period of intermittent use, apparently—in 1876, on the pages of the Parisian Revue des deux mondes, then a tone-setting journal read all over Europe. The application of the term to linguistic layering dates in France from 1882 and, in all likelihood, presupposes intervenient specialization in the direction of geology; I gather that Italy, through Ascoli as a spokesman for its semi-autonomous tradition of glottologia, for once preceded France in the coinage of an unequivocal designation of the subjacent speech layer. 'Substra tum', by virtue of its prefix, invited an unambiguously polarized name for the reverse situation, a 'crust' overlaying the principal deposit, and W. von Wartburg, in the early thirties, provided an answer to this need by advocating the label SUPERSTRATUM, which would apply, for instance, to Merovingian Frankish superimposed on Gallo-Romance, to Maghrebi Arabic partially superjacent vis-à-vis Hispano-Romance, to an unidentified variety of South Slavic—possibly akin to Macedonian—stretching over Daco-Romance and permeating it with its ingredients, and to Mexican Spanish forming a roof over Maya in the Yucatan Peninsula. 'Substratum' and 'superstratum' admirably fulfilled an actual de mand of stratigraphic linguistics—a discipline we must now carefully distinguish from stratificational analysis—and have lately enjoyed unqualified recognition on both sides of the Atlantic. I am uncertain as to which 20th-century expert, from the ranks of Romanists, presumably, in fact launched ADSTRATUM, but I find the word far richer in noteworthy potentialities than the inconclusive uses so far essayed would lead one to aver. If we are to believe J. Marouzeau's ter minological glossary (Lexique ... 3; Paris, 1951), 'superstratum' and 'adstratum' have of late been employed interchangeably, a situation which, once proven true, would call for the elimination of the less widely adopted member of this set of doublets. Salvatore Battaglia, in the opening volume (1961) of his Grande dizionario, maintains that 'adstratum' involves a state of balance or equality

32

GENETIC LINGUISTICS

between two layers, the implication being, it seems, that substratum, in contrast, suggests not only anteriority visually dramatized by subjacency, but also hierarchical inferiority; the 'adstratum', to quote Battaglia, 'indica il contatto fra due lingue di cultura che si influenzano reciprocamente senza però che si abbia la prevalenza netta dell'una sull'altra'. Would this apply to a social cleav age like the one between aristocratic Anglo-Norman and the native rustic speech that one is led to interpose between Old English and Middle English? Or should we look to countries like Belgium and Switzerland for exemplification? I, for one, am tempted to reserve 'adstratum' for a rarer, yet more neatly delimitable situation, as when nuclei of Greek and Osco-Umbrian speakers spread, carrying their languages with them, over the western Mediterranean under the aegis of the rapidly expanding, culturally liberal Roman Republic. By the same token, 'adstratum', adjusted to the context of modern immigration, could aptly label certain admixtures of Italian and Galician-Portuguese speech to the uninter ruptedly predominant layer of Spanish in the La Plata zone (Buenos Aires and Montevideo); also, under comparable conditions, it might point to recently identified elements of Italian, Yiddish, and Caribbean Spanish in the racy speech of New York City and other metropolitan centers of this country abounding in newcomers from abroad. Of course, the coinage of neologisms in -stratum can become as much an idle parlor game as was, for two long decades, the deriva tion—half-serious, half-jocular—of qualifiers in emic ('sememic', 'tonemic', 'stylemic', etc.) and ֊fix (witness 'interfix', 'coaflix', 'superfix', 'transfix'); small wonder, then, that Carlo Battisti recently overreached himself by experimenting with the ugly hybrid parasnato in Chapter IV of his Sostrati e parastrati nel l'Italia preistorica. On the other hand, I heartily applaud Italian linguists for freely experimenting with the ending -oid: SUFFIXOID is crisper than 'quasisuffix', LIGUROID is most helpful to convey the idea that a language spoken out side Liguria was genetically or typologically akin to Ligurian proper. To revert to geology, a useful borrowing from its lexical inventory has been un filone 'streak, vein (in the earth)' for all kinds of minor but far-flung admixtures, a metaphor favored by some Italian scholars, e.g. Vittorio Bertoldi, and occasionally by their Spanish counterparts (una veta) ; sediment and deposit have been welcomed in many quarters ; so have been such key words as residual and vestigial. Another developmental science with which linguistic evolutionists established an early rapport was paleontology. The term 'linguistic paleontology' itself enjoyed but a short vogue; I can trace it to the three editions of Meyer-Lübke's classic Einführung in das Studium der romanischen Sprachwissenschaft (the first, of 1900; the third, of 1920) ; did the author borrow it from Ascoli? The neologism failed to strike root in linguistic quarters; but a few satellite formations, includ ing 'petrifact', 'fossilization', and 'ossification', have acquired citizenship rights, whereas 'skeletal outline', propelled by some descriptivists, pertains to a dif ferent metaphoric strain. Medicine, though not a genetic science, is so closely interwoven with biology as to have benefited from this tide; while the old-fashioned arbitrary concept of 'decay, corruption', as laid down by dogmatic preceptivists, was gradually yielding ground to more sober and more scientifically definable notions, part of


33

this archaic imagery was salvaged and allowed to fuse with the new terminology spreading from medical schools. Gilliéron's 'pathology' and 'therapeutics' have left a profound, indelible impact on genetically oriented lexicologists; to this day they are certainly more defensible, if one agrees to make allowances for teleo logical concepts, than the same scholar's later, inexcusably flamboyant catch words such as 'bankruptcy' (1919), 'hypnotism', 'phantasmagoria' (1921), and even 'thaumaturgy' (1923). But though Gilliéron, not least through his over statements and his flair for theatricals, drove a wedge for this biological approach and its terminological trappings, it was a scholar less given to sensationalism, namely Littré, who as early as 1880 included in a by and large forgotten miscel lany of articles one piece entitled Tathologie verbale, ou lésions de certains mots dans le cours de l'usage'. Darmesteter's oft-cited La vie des mots provided the bridge between the writings of the two men: Birth, life, death, evolution, trans formation, struggle, health, creativity, stagnation, attrition remain the slogans from cover to cover, and, in perceptive anticipation of Gilliéron's probings into the outlandish and the erratic, one stumbles here over a reference to 'teratology', i.e. the study of monstrosities. Long before Darmesteter, Whitney, endowed with almost uncanny foresight, warned of the potentially perilous inroads of physical sciences and of psychology on the domain of linguistics; but, as is shown by the very title of his Outline', The life and growth of language, he readily came to terms with the life sciences. But enough of the past. This retrospect, which is not meant to sound a nos talgic note, will have served its limited purpose if it has aroused your curiosity (as it has my own) as to what the most imaginative geologists, paleontologists, and paleobotanists may have done in recent decades and, in fact, may be ac complishing at this very moment that should be worthy of at least some lin guists' sustained or passing attention. To cite just one concrete instance of a blind alley into which linguists, by dint of gratuitously isolating themselves from other evolutionists, have run in the last quarter-century: The age-and-area hypothesis, i.e. the tentative reconstruction of a sequence of events from the present terri torial distribution of a pair of erstwhile rivals, has been inconclusively eyed by a handful of linguists with an occasional furtive side-glance at sociology. But, surely, at least some geologists and paleontologists must have reached note worthy conclusions as to the cogency of such inferences from space to time in their own disciplines. I should very much like to see the problems of spatiotemporal linguistics (as I think this branch deserves to be called) reopened for vigorous discussion, on condition that the participants agree first to take cogni zance of any parallel progress recently achieved by standard-setting naturalists. Finally, if I may venture a very ingenuous remark on botany, I have followed with a layman's admiration the skill with which naturalists manage to narrate the biography of a tree from a closely reasoned analysis of its rings. Thus, if the innermost rings are relatively broad and evenly spaced, the analyst infers that at first the tree grew with no disturbance. Subsequent rings of the same tree may show, through the varying impairment of their earlier breadth and evenness, that something pushed against it, making it lean (the rings are then wider on the opposite side) ; that the crowns and root systems of the tree's neighbors deprived

34

GENETIC LINGUISTICS

it of much of the requisite water and sunshine; that, after the harvesting of surrounding trees, there was once again ample nourishment and adequate exposure to sunlight; that a fire swept through the forest, perhaps only scarring the tree; that a prolonged dry spell once again slowed the tree's growth; that some insect (like the larva of the sawfly), by eating leaves and leafbuds, may have adversely affected the health of our much-troubled tree. This mosaic-like re construction, under optimal conditions, of the probable sequence of events can never be duplicated by linguists, simply because languages do not cast off, at regular seasonal intervals, crusts or barks hardening into rings or shells; but a good deal can be observed and assimilated, I surmise, both from the heartening breakthroughs and the inevitable pitfalls of techniques adopted by favored neighboring disciplines. There exist several other universals in linguistic reconstruction which cry out for separate treatment and may even demand the immediate remedy of existing practices. In the last thirty years not a few previously established sound shifts which long appeared unintegrated, dangling loosely in the air, so to speak, have acquired a new and more interesting perspective through the efforts of diachronic phonemicists. Scholars so oriented have often succeeded in placing isolated events in a meaningful chain or in establishing helpful cross-connections. In each instance the earlier description or attempt at explanation turned out to be not necessarily inaccurate and worthy of rejection, but rather unpalatably meager, if viewed in critical retrospect. In numerous cases the modern analyst's ability to operate with entire systems rather than with individual sounds has indis putably had an eye-opening effect. The advance thus achieved has been from a survey of the surface to an inspection of the depth, if I may resort to two ultrafashionable words ordinarily used in a different context. Yet even the most com mendable idea or correlated technique has a limit; any indiscriminate applica tion of the new insight beyond the saturation point threatens to produce rapidly diminishing returns. While diachronic phonemics retains its usefulness as a handy means of bracketing and underpinning an otherwise crumbling congeries of individual sound changes, it provides no answer to several legitimate questions of considerable weight, such as this one : Under what ensemble of circumstances can an aberrant morphological condition set in motion a drift of events even tually conducive to a regular sound change? We are so accustomed to seeing phonology take precedence over morphology, starting with the typical organiza tion of introductory textbooks and monographic grammars, that any such morphological adjustment (e.g. leveling, diversification, wholesale replacement of suffixation by prefixation, etc.) as marks a speech community's reaction to a preceding sound change accountable in strictly phonological terms strikes us as natural, while the presumably rarer reverse concatenation of cause and effect seems counter to expectation. Undeniably, one runs in one's readings across parenthetic remarks to the effect that certain minor sound changes may have been stimulated by analogical regroupings in the inflectional, affixal, or composi tional scaffold of the language under study. But a systematic inquiry into this category of possible, if infrequent, change—an investigation so ambitiously con-


35

ducted as to be cross-cultural, cross-spatial, and cross-temporal—remains, un less I am grossly misinformed, a mere desideratum. I could easily here enumerate other problems of comparable magnitude which clamor for a resolute frontal attack rather than for evasive, inarticulate allu sions. In order not to tax your patience, however, I shall concentrate, for the remainder of my Address, on just two, admittedly many-faceted issues: that of multiple (or complex) vs. single (or simple) causation, and that of primary vs. secondary determinants. This thread of discussion will lead us straight into the center of etiology and may, at the same time, add a much-needed touch of re deeming concreteness to this evening's deliberations. There has been considerable progress in the recognition of different categories of change, especially within phonology: Thus, it would hardly occur to any Ro mance expert at present to champion the application of, say, 'Verner's Law'— which defines the impact of the word-stress on the outcome of certain consonants in paleo-Indo-European—to a sliver of material that does not fall under that 'law's' jurisdiction. This century's scholars, unlike some of even the keenest among their predecessors, have learned to distinguish quite sharply between universals and particulars. However, no major efforts have yet been made to explore the compatibilities or plausible successions of different determinants of change. Without any reliable guide to a searching analysis of the simultaneous impact of these affinities, complementarities, and reconcilabilities (or, conversely, to the patterns of their mutual exclusiveness), no serious study of complex causa tion can be undertaken. By the same token, the typical sequences of causes and effects in temporal projection and, within this schema, the hierarchy of major (or primary) and minor (or secondary) determinants deserves increased atten tion. To drive home my point, let me first succinctly restate, under a new angle, four problems of Romance linguistics, two of which have given rise to memora ble discussions, while one—the first—has never been posed, let alone examined.1 The kindred problem of primary vs. secondary determinants will, conversely, be thrown open in abstract terms, involving at most an imaginary situation in a nonexistent language. PROBLEM N O . 1. There exists a sizable number of modern Spanish adjectives (most of them disyllabic) with a masculine singular ending in unstressed ֊io, e.g. agrio, amplio, lacio, limpio, lucio, medio. Is it warranted to posit a productive derivational model? To answer this question, we must first break down the twenty-odd formations collected into their historical sources. The diachronicist will distinguish between two major subgroups, of which the first includes adjec tives, both vernacular and learned, whose segment ֊io could have been foreseen from elementary knowledge of phonology, while the second encompasses the more difficult and thought-provoking cases. The first subgroup falls into three minor subdivisions : (a) vernacular descend1

The following section of the Address embodies, in drastically compressed and partially rephrased fashion, the results of a monographic investigation ('Multiple versus simple causation in linguistic change') which is to appear in the new testimonial volume for Roman Jakobson planned by Mouton & Co. (The Hague).

36

GENETIC LINGUISTICS

ants of Latin -IDUS adjectives, e.g. FLACCIDU 'flabby' > lacio, LiMPlDU 'trans lucent' > limpio; (b) one vernacular descendant from a Latin -EUS adjective, involving a radical-final bilabial consonant: RUBEU 'red(dish)' > rubio 'blond'; (c) learned descendants—some of them very old—of Latin -ius adjectives, regardless of the radical-final consonant, e.g. MEDIU 'half' > medio, TERTIU 'third' > tercio, also NESCIU 'ignorant' > necio 'silly' and NIMIU 'excessive' > nimio 'prolix, stingy'. In the other subgroup, further categorization is also possible, but within every narrower category the marked tendency of each word history to yield a highly individual vignette is in evidence throughout. In at least three cases the common denominator for the anomaly is the use of -io for expected *-o. The three adjectives at issue are soberbio 'proud', from SUPERBU; agrio 'sour', from ĀCRU beside ĀCRE; and amplio 'wide', from AMPLU. In the vernacular layer of Old Spanish, SUPERBU should have cast off *sobiervo, while SUPERBIA 'pride' should have emerged as sobervia, and in fact did so. May one argue that the 'ideal' form *sobiervo 'proud', placed alongside the adjectival abstract sobervia, would have produced a morphophonemically unparalleled pair, an impending situation that has led to leveling in favor of the substantival stem variant? With agrio, which has superseded the perfectly regular OSp. agro, there is room for interpretive disagreement : Some scholars note the pressure of agriar 'to sour' (despite the late appearance of that verb) and adduce as parallel the equally startling transformation of AMĀRU 'bitter' into Sp. amargo, doubt less through contamination with amargar 'to embitter' < AMĀRICĀRE; others seek the source of associative interference in the widespread fluctuation between, say, ve-, vi-dro 'glass' < VITRU and vidrio 'id.' < VITREU, lit. 'glassy'. To account for amplio, wé may recall that AMPLIUS frequently furnished an equivalent of ULTRĀ, PRAETEREĀ, PLUS and, as a result of this extended use, gave rise, within the bounds of Latin, to the verb (EX)AMPLIĀRE rather than *AMPLĀRE. In another triad of formations, even more intricately structured, there has oc curred an accent shift, from io to '-io. The three adjectives in question are sandio 'foolish', zafio 'uncouth', and reacio 'obstinate'. In the old texts, the combined evidence of rhymes and meter shows that the word for 'foolish', etymologically opaque, was stressed sandio՛, cf. the continued comportment of its congener Ptg. sandeu. The Spanish word later fell into desuetude; when, after centuries of dormancy, it was revived, probably as a consequence of newly awakened interest in Cervantes' Don Quixote, readers identified it as sándio, conceivably on the analogy of its near-synonym nécio. Zafio is in all likelihood an Arabism, though Orientalists are in disagreement as to its specific model. The latest conjecture classes it as a cross of SAFÎH 'stupid, shameless' and ŞÂFI 'pure'. Be that as it may, the earliest witness to its transfusion into Spanish, the Granadino convert Pedro de Alcalá, ca. 1500, placed a stress mark unequivocally on the i. Reacio 1 stubborn', recorded in older texts as re-fazio, -hazío, is unmistakably an offshoot of OSp. refazer 'to do over and over again', from remodeled REFICERE, and in volves the same derivational suffix -îvus as does OFr. restif՛, cf. Eng. restive and mod. Fr. rétif.


37

At the periphery of the lexicon one detects, particularly in dialects, a few iso lated formations whose testimony carries much weight. Extr. ludio 'leavened' matches standard leb- > leu-do, from LEVITU, and may involve the avoidance through metathesis of a doomed falling diphthong in some such immediate proto type as *liudo. The other item, apparently branching off directly from pender 'to hang', is pen- or pin-dio 'bent, leaning', peculiar to the north of the Peninsula. If agrio, amplio, ludio, pindio, reacio, sandio, soberbio, and zafio exhibit a certain vitality of this type within Spanish, the scheme thus authenticated would raise the number of unstressed derivational suffixes once deemed unproductive in Romance. We have thus far isolated the following causes, ascertainable or highly prob able, behind the crystallization of ֊io adjectives: perpetuation of prototypes in -IDUS (limpio); limited survival, in the vernacular stratum, of bases in -EUS (rubio); adoption of learned formations now early (necio), now late (nimio); influence of the co-existent abstract (soberbio) ; reverberations of the autonomous use of the Latin comparative (amplio, perhaps with the collateral support of the verb ampliar) ; analogical extension of wavering between an old substantive and an erstwhile adjective tending to evict it (ve- vi-dro ~ vidrio —-> agro ~ agrio); accent shift in a word reintroduced into restricted use (sandio) or in one favored uninterruptedly (reacio) ; restructuring of a lexical item borrowed from an exotic language (zafio); and free-wheeling use of -io (dial, pendio), sometimes coinci dent with metathesis (Extr. ludio). Each individual explanation, in some in stances adduced to account for a single adjective, seems defensible; is any one powerful enough to have allowed an observer to predict, with assurance, this particular course of events? Such a tangled situation provokes a number of interlocking questions. If half way satisfactory conventional explanations are readily available for all or most of the items, why hypothesize a new derivational model such as radical-stressed adjectival -io? If, despite our misgivings, we accept this model as a contributing force, do the other explanations fall by the wayside, or can two (or more) causes be declared mutually complementary and co-efficient? Does the assumption of multiple causation clash with the principle of maximum economy as a yardstick of cogent scientific analysis, to the extent that such parsimoniousness is recon cilable with the complexity of the facts analyzed? In seeking to provide answers, one must discriminate between mere descrip tion and causal explanation. When we state that liudo was locally transformed into ludio and list this process under the rubric 'metathesis', we simply attach a handy tag to a shift which, within that dialect, happens to be practically un predictable. Our second step is to remind ourselves that, even where probabilistic state ments can be ventured, such assessments as involve a low degree of predict ability prod researchers to seek for additional determinants. Assuming that a unique socio-historical situation involves a protracted state of bilingualism, thus favoring the surmise of substratum influence, surely the staunchest supporter of such hypotheses will grant that only selected features of sound structure, lexicon,

38

GENETIC LINGUISTICS

etc. are bound to filter through. Clearly certain asymmetries, lacunae, or ambigui ties—in short, flaws in the architecture of the receiving language—may control the extent of assimilation of the ingredients adopted from the donor language. Our third thought turns toward quantification. If, to simplify matters, we argue that the crystallization of all -io items except those extracted from Latin through normal channels involves the convergence of two forces—Force A, the agency of an apparently productive derivational model not yet fully identified, and Force B, some collateral pressure familiar from earlier inquiries—we observe that, in the slice of material examined, Force  varies from case to case: Bi (accent shift), B 2 (adaptation of a loan), B 3 (contamination by a related verb), etc., while Force A, pervasively at work, ties together all these motley minor alterations. The positing of a heretofore unsuspected force—in this context, of an expanding derivational type—gains in plausibility if that force is each time paired off, in a series of complex mutually related processes, with a companion force of distinctly narrower scope. One final consideration: Can the agency of the new factor be observed, at least occasionally, in isolation? For agrio, amplio, reacio, etc., at least one other driving force could be tentatively identified, with varying degrees of accuracy. The tidy isolation of even one clearcut instance would immeasurably bolster up any hypothesis reckoning with a new separate ingredient in multiple or complex causation. PROBLEM N O . 2. This is a classic illustration of irreconcilable disagreement between two scholars of comparable caliber; it revolves around the alternation of [δ] and [r] in Hispano-Romance, [δ] being locally an allophonic member of / d / . Of the two disputants, E. G. Wahlgren (1930) brought to the arena a superior knowledge of cognate languages, his critic T. Navarro (1931) a more intimate glimpse of fluid dialect usage. Wahlgren granted that in certain Romance languages the transformation of d into r could represent a regular sound shift, a kind of 'rhotacism'. Elsewhere (e.g. in Spanish), he argued, the process was set in motion by a subtle interplay of (a) lexical or affixal analogies and (b) saltatory (sporadic) sound changes; but he apparently did not allow for any intermingling or alliance of these two major groups of causal ingredients. Thus, in dissecting Spanish words (predominantly of dialectal stock), he implicated lexical contamination, as with badajo 'clapper (of a bell)', changed into barajo allegedly through contact with baraja 'pack (of playing cards)'; the intrusion of a derivational suffix, as in Ast. antroiru 'carni val' instead of expected antr oidu, through pressure of -oiru, or in dial. párparo 'eyelid' < párpado, standard lámpara 'lamp' < LAMPADA, through association with '-aro, '-ara՝, the interference of a co-existent infinitive, as in seguirilla be side older seguidilla 'form of stanza' and mentira 'lie, falsehood' beside Cat. mentida, through contamination with seguir 'to follow' and mentir 'to be false' respectively; and miscellaneous disturbances subsumable under assimilation (Cloro- < Clodomiro), dissimilation (paeres < parees < paredes 'walls'), and metathesis (poderón < paredón 'thick wall'). In a small residue of more compli cated relationships, Wahlgren recognized a joint effect of some of the agencies


39

just mentioned, while steering clear of any assumption of regular phonological change; this more elastic attitude applied to regionalisms like arbolera 'grove', buirador 'brassworker', pelarela 'loss of hair', porvaera 'cloud of dusť, and vēdera 'sidewalk'. The vulnerable spot in Wahlgren's position is the extraordinary range of diverse phenomena, many of them of an optional rather than obligatory inci dence, to which he resorts in an effort to elude any appeal to regular sound shift; one notes, as a result, the strikingly thin representation of each phenomenon in the inventory of determinants. Why not first deduct a few ancient cases (like lámpara), which must obviously be fitted into a different context; then posit for the 20th century a state of widespread fluctuation [֊δ֊]/[-Γ-], still bidirectional, but unequivocally pointing toward r as the ultimate victor, a situation recalling a sound shift in statu nascendi, , i.e. one which may or may not eventuate, de pending on the final outcome of the interplay of contending forces? If we agree to view these disturbances as mere symptoms or harbingers of a pending shift, does it not stand to reason that this preliminary scouting and groping—as speakers unconsciously feel their way toward a possible breakthrough which, once it is a fait accompli, linguists will dub a sound-shift—should be directed toward points of least resistance? These are just such points, in the edifice of language, as might cause the existing structure to crack through pressure in dependently applied by miscellaneous processes, such as metathesis, dissimila tion, lexical or affixal analogy, etc. If these heterogeneous established causes ef fect, again and again, the same shift, one has every reason to suspect that there has been operative, initially in alliance with and through them, some new and sharply focused force pressing in a single direction. PROBLEM N O . 3. The explicit marking of a given morphological category in excess of the traditional norm has been labeled hypercharacterization. Thus, Latin possessed the category of gender and frequently expressed relevant con trasts with maximal crispness and neatness: FĪLI-US, -A; BON-US, -A, -UM. In other contexts the gender was signaled less vigorously, cf. TBIST-IS, both masc, and ferm., and VÊTUS, a single form shared by all three genders. From the parallel changes of GRANDIS to Fr. grand, -e, of VIRIDIS to Fr. vert, ֊e, and of a pair un characteristically marked by its endings, such as SOCER 'father-in-law' vs. socRUS 'mother-in-law', to clearcut It. suocer-o, -a and Sp. suegr-o, ֊a, we can infer a protracted trend toward hypercharacterization of gender. In a few instances this trend seems to be at work alone, as when OSp. cuchar (fem.) 'spoon' yielded to mod. cuchara. In most cases there occurred a splicing of this tendency with others, liable to separate isolation. Thus, if VETUS -ERIS, ailing from inadequate marking of gender (hypocharacterization), was replaced in some varieties of provincial Latin by VETULUS (cf. It. vecchio, Fr. vieil), in others by *VETERU distilled from VETERĀNU (OPtg. vedro, OSp. viedro), in yet others by a cross of VETULU and *VETERU (as is true, on phonological evidence, of Sp. viejo), then affixation and inflection—for one detail, the co-existence of PAUPER -ERIS and MISER -Ī 'poor, wretched'—may also have shared heavily in the process; cf. Sp. pobre, alegre alongside It. povero, allegro.

40

GENETIC LINGUISTICS

In the domain of zoönymy, the marking of the animal's sex through gender was at the outset limited to the barest minimum, except with certain domestic animals where the male and the female of the species provide radically different services (say, BŌS VS. VACCA). At the start, LEŌ 'lion' and LUPUS 'wolf' were consequently ambigeneric, MASCULUS and FĒMINA being added at intervals for incidental emphasis on the sex, much as are bull (or buck) and cow, cock and hen, he- and she- in English or, for that matter, ֊bock, -kuh, and -kalb in German. Only later did speakers coin LUPA for 'she-wolf', fig. 'harlot', and LEA for 'li oness'. Significantly, wherever lexical innovations struck root in Romance zoönymy, regardless of the specific justification of each replacement, one recurrent fringe benefit was the sharper delineation of the gender (and sex). Thus, for ambi generic CAN-ĒS or -is 'dog', Spanish displays an exotic substitute, perro, flanked by perra 'bitch'; near-by Portuguese proved hospitable to cão < CANE, closely linked with cadela 'bitch', from CAT-ULA, -ELLA, orig. 'female puppy'. French offers a third solution, placing chienne alongside chien, much as, in a French man's world-view, a lionne shares the lion's lair. The lexical and grammatical vicissitudes of the 'cat' were slightly different: The original designation FĒLĒS was feminine, like UOLPËS 'fox' and MĒLĒS 'marten, badger', and the reference was initially quite vague as regards both sex and species ('wild cat', 'weasel'?). The introduction of CATTUS (GATTUS), an invader not attested before the 4th cen tury, in all likelihood signaled three events: the adoption of a new label, the importation of a hitherto unknown species, and the acceptance of a novel social institution (domestication of the mouse-chasing pet) ; but—more vital to us—it coincidentally provided a welcome means whereby speakers could instantly tell a male from a female. The descendants of LEPUS 'hare' invite similar analysis, and one could almost indefinitely expand the roster of illustrative examples. Every single elaborative substitute may be amply justified, I repeat, in terms of such trivial phenomena as lexical borrowing, orchestrational attachment of augmentative or diminutive suffixes, analogical innovation, etc. But if the student of causation succeeds in isolating one consistently emergent by-product of all these multifarious processes—namely the more explicit marking of gender and sex—he is entitled to vindicate for the agency effecting the change the status of an autonomous codeterminant, because this force brackets untold developments doomed otherwise to appear disparate, meagerly exemplified, and almost random. PROBLEM N O . 4. Consonant dissimilation at a distance has been a notorious crux of explicative linguistics. The event that best qualifies as the starting point for the peak of the debate is the publication of M. Grammont's doctoral thesis (1895). That monograph attempted to elevate the chosen class of dissimilation to the rank of a set of stringently formulated sound laws, cross-temporal and supposedly binding on all speech communities. Grammont's book, and a major complementary article from his pen, provoked a flurry of formal reviews and lively prises de position ranging from glowing or lukewarm approval to flat rejection and involving such experts as Brugmann, Meillet, Niedermann, and Schuchardt, to cite just a few names. Gauchat, likening the saltatory change at


41

issue to mutations in botany, warned against indiscriminate use of the facile label ('moyen commode d'explication'). Fresh vigor was injected into the discussion by Rebecca Posner's Oxford dissertation (Consonantal dissimilation in the Romance languages, 1961), which had the merit of eliciting a bold counter proposal from K. Togeby only two years later. Like Grammont, Posner believes in the survival of the fittest as the overriding principle; but she skillfully nuances her predecessor's teachings by heeding phonemics, observing the incidence of sounds, spelling out explicit conditions for regressive effects, separating the central from the peripheral sections of the lexicon, and distinguishing, especially as regards the sequence of two r's, between attitudes of individual languages. In his sparklingly phrased counter-theory ('Qu'est-ce que la dissimilation?', Rom. Philol. 17.642-67, 1964), Togeby admits that consonant dissimilation oc casionally operates with the precision of a genuine 'sound law', e.g. where the succession of labiovelare is the stumbling block, as in QUINQUE. For the most part, however, it makes its appearance sporadically, a situation which forces the analyst to use the term, at best, as a descriptive tag ('étiquette sans valeur') and, in fact, urges its elimination from any causal argument (650). The core of Togeby's own reflections resides in his repeated insistence on the dispensability of any assumption of a consonant's dissimilatory change or fall if some other explana tion, equally plausible, can be proffered. A generous string of such persuasive rival explanations is next displayed in searching detail, the possibilities varying from a medieval scribe's lapsus calami and an early etymologist's faulty base to lexical borrowing, hyperurbanism, onomatopoeia and expressivity, crosses within the same word family, blends with some other word family, folk ety mology, suffix or prefix change, agglutination and deglutination of the article. What adds a touch of drama to these comments, aside from the author's polemic mastery and technical versatility, is to see a scholar reared in the tradition of radical structuralism and, to this day, steeped in it, side for once with such champions of the opposite extreme as Schuchardt and Gilliéron. Despite my admiration for Togeby's spirited advocacy, let me provisionally state this in partial criticism of his contention: From the fact that a phenomenon, while definitely identifiable, is seldom found in isolation, I incline to infer that this isolable phenomenon usually operates in conjunction with others. Once the common occurrence of multiple causation is granted axiomatically, there will be no need to proceed to radical extirpation of any suspected redundancy. To cite concrete instances, Togeby is at his most seductive when he corrects the geographic trajectories of certain words traced hazily by Posner, as when he argues that it is not Fr. caramel which must be confronted with CALAMELLU, but the French word's immediate prototype, Sp. caramel 'lollypop'. In other cases, possibly more numerous, Togeby's alternative hypotheses lead to richer orchestration of the earlier analyses rather than to their complete abandonment. Thus, in almost-learned Sp. plegaria 'prayer' < PRECĀRIA the dissimilation in voked by Grammont seems to be the decisive force behind the withdrawal of the first r in favor of l but plegar 'to fold' < PLICĀRE may have been, on the seman tic side, a contributing factor (by suggesting genuflection). Furthermore, the switch in Spanish from an older, rustic form pregar 'to fold' to a socially more

42

GENETIC LINGUISTICS

elegant variant slegar created an ideal setting for the simultaneous replacement of pregaria by plegaria. Our attention, then, is drawn to an ensemble of circum stances, by no means mutually exclusive: (1) a latent dissimilatory trend, (2) a stimulating semantic or imagerial bridge, (3) false regression, i.e. a typical socio-linguistic reaction. Each single ingredient isolable from this amalgam can be charged with the full burden of responsibility for the crystallization of plegaria, but it is the joint momentum that most satisfactorily explains the outcome. There is, needless to say, no reason to assume that situations of this kind are idiosyncratic to Romance. To supply a concluding illustration from Germanic, let me refer you to one set of idiomatic formulas in which that language family abounds: the irreversible binomials. An inquiry into these inlayings, which lend such verve and raciness to English and German, has disclosed that, in spon taneous and stylized discourse alike, several discrete forces are at work designing the shape of the binomials (A + B). Chief among these forces, in modern IndoEuropean, are (a) one rhythmic pattern favoring the attachment of a longer  to a shorter A; (b) one semantic pattern giving sequential precedence to the positive over the negative quality—'positive' and 'negative' being relativistically defined as a matter of consensus within the frame of the given culture. There are on record certain binomials in which A precedes  by virtue of form alone (greater brevity)—Eng. each and every, root and branch, part and parcel, Ger. auf Biegen und Brechen—or of meaning alone ('positive', 'concrete', 'palpable' feature first)—bigger and better, hither and thither, with equality or near-equality prevailing in regard to the companion trait. Admittedly, in a few isolated cases a binomial has crystallized even though its outward configuration or its inner structure was so unpromising that this deficiency could have thwarted its genesis. But in the overwhelming majority of the cases it was, of all circumstances, a lucky coincidence of rhythmically appealing form and semantically attractive message that, in the first place, accounts for the birth and rapid propagation of the formula: all or nothing, assets and liabilities, rise and decline. To summarize: The prevalent notion of unicausality needs thorough revision much as did until recently the long-accepted belief in dominant monolingualism (or, for that matter, the earlier typological classification of languages on the basis of a single structural characteristic, however salient). By starting out with the expectation of plausible pluricausality, we shall do fuller justice to the intricacies of reality. The problem of concurrent causes must be very sharply divorced from the equally intricate and far more sorely neglected issue of primary vs. secondary causation. Let us assume that—in an undisclosed locale—through an interplay of substratum influence (which operated through bilingualism) and structural streamlining (which removed the rough edges), a given phoneme of the parent language, say -g-, had dropped out mtervocalically in the north while surviving intact in the south. Between the neatly silhouetted dialects of the extreme north and those, equally trim, of the extreme south there stretches a major transitional zone, which exhibits an, at first, perplexingly erratic distribution of -g- under the stated condition. Not only do the subdialects of this zone increasingly favor the


43

retention of the controversial occlusive as they approach or adjoin the 'solid south', but there emerge, again and again, in the same localities, words behaving in a mutually contradictory manner, even though one would expect their com portment to be alike. At closer inspection it turns out that the results are not en tirely motley, inasmuch as numerous minor patterns become visible amid the chaos. Thus, it can happen within the confines of this area that words containing three syllables (or more) lose their -g-, while those of fewer syllables retain the consonant. Again it is conceivable that the place of word stress acts as a dif ferentiator: If -g- precedes the accented vowel, its chances of survival may be quite different from what they would be under reverse conditions. Some adjacent sound may seem to produce the split: Though in the general phonological struc ture of the given language it makes no difference whether a consonant is flanked by / a / or / o / , the fate of ֊g- in the ill-delimited dialect zone under examina tion may hinge on whether it forms part of the sequence / g a / or / g o / . Outside the realm of strictly phonological conditioning, it is not impossible that the loss of -g-, all other circumstances being equal, occurs with impressive consistency in adjectives, but neither in nouns nor in verbs; or that, within the verbal para digm, -g- disappears only from certain moods, tenses, numbers, or persons. Beyond the domain of grammatical categories there extends the vast, hazily circumscribed zone of meaning: From this zone, too, speakers seem, occasionally, to have taken the cue in accepting or discarding certain forms involving the locally unstable -g-. I t is surely no matter of mere coincidence if in a close-knit series of meliorative or propitiatory words, the -g- has invariably asserted itself, while in a comparable series of pejorative words (all other conditions, once more, being equal), the -g- was doomed to extinction. Clearly, this short list of readily identifiable determinants of a bifurcation does not exhaust the possibilities : In languages where vowel quantity is distinctive, long vowels before, or after, or both before and after, the -g- may have produced a special effect not observable otherwise. Such languages as grant structural significance to lexical tone may, in turn, impose a unique set of tonal conditions on an endangered consonant if that consonant is to emerge unscathed. Not only can examples of separate condition ing factors be almost indefinitely multiplied, with the understanding that only a selection of such factors is likely to be operative in any given language during a certain evolutionary phase, but characteristic combinations of these factors, productive of mutual reinforcement or reciprocal blocking, still further increase the range of possibilities. In the older historical grammars the dividing line between all these kaleido scopic varieties of MINOR causation and the necessarily few forces of MAJOR causation, such as external pressure and internal tightening, was not always drawn with all desirable precision. At this point, an attempt at clearcut hierarchization of these two categories must be made. If dialects sharply profiled in general, or at least conspicuously tidy in their treatment of certain sound changes, exhibit a neat cleavage, as when, in our imaginary country, the extreme north abandons the ֊g- which the extreme south faithfully preserves, then the chances are that a contrast of this magnitude, easily involving hundreds of examples, has been produced by the agency of a PRIMARY cause. But if, in a transitional zone

44

GENETIC LINGUISTICS

rich in blurred contours, minute coincidences in features of word stress, phonotactics, grammatical category, semantics, etc. must be painstakingly ferreted out to justify the self-contradictory comportment of exiguous clusters of lexical items (not infrequently as few as two), then it would be quite unrealistic to con tend that these weakly represented determining conditions, which shall be known as SECONDARY factors of causation, are on a par with, say, substratum influence or structural adjustment. The underlying historical process is likely to have taken, typically, this course. For social reasons not directly related to the fabric of language, the cohe sion of speakers in the north and in the south appears to have been such as to have produced two powerful nuclei of speech, each almost homogeneous within its own boundaries, though not identical with the other, witness—possibly among other conflicting peculiarities—the contrastive treatment of ancestral -g-. In the center of the territory—perhaps a kind of no man's land for many critically im portant years—there arose no comparably leveled major dialect. Rather, as a result of successive infiltrations, annexations, partitions, and other disturbing changes of authority, there eventually crystallized a maze of subtly diversified subdialects, which only at a distinctly later date, through a delayed change in fortune for the raiders and settlers concerned, coalesced into a superficially uni fied dialect abounding in deep-seated discrepancies. Inside such a hastily as sembled structure there must initially have co-existed countless doublets (e.g. péga ~ péa, tigúr ~ tiúr, etc.). In the course of the ensuing coalescence, the critical delay of the rapprochement and the continued lack of organic unity pre vented, in each hamlet, the complete subordination of local autonomy to a terri torial norm, i.e. either the sweeping abandonment of the ֊g- or its consistent restoration. On the other hand, superimposed over-all tightening was sufficiently strong to squeeze out any pointless duplication of forms, making it impossible for péga and péa, tigúr and tiúr to survive side by side for an indefinite length of time. As a result, there began to form small kernels of resistance to the spreading ero sion of ֊g-, the leader words of the opposition being conceivably, at the outset, such lexical items as enjoyed greater currency in territories closer to the con servative south. On the analogy of these key-words, in whose ranks the process of selection was controlled partly by straight linguistic, partly by extraneous fac tors, speakers selected variants either with or without ֊g- of such other words as shared with the key words some characteristic feature of form (phonic, gram matical) or meaning. Practically any isolable detail of prosodic, phonemic, in flectional, derivational, compositional, or semantic structure could qualify for this role of secondary determinant, and it would obviously be quite misleading to regard the limited and delayed agency of such a random detail as a direct force of causation remotely comparable in its impact to potent substratum influence via bilingualism or to structural adjustment in the direction of economy and sym metry. The situation is not radically different where, through a concatenation of his torical antecedents and relative laxity of structure at the critical moment, a speech community has at its free disposal a number of competing derivational suffixes serving to produce abstracts from nouns, adjectives, and verbs, i.e. the


45

equivalents of Eng. -dom, -hood;-th,-ness, -ity; -ing, -al, -ment. Ordinarily, the suffixes in question are traceable to different sources or 'strains'; in English, e.g., one can readily set off those of Germanic, Anglo-Norman, and learned (Greek or Latin) provenience, and at stages of relative stability of usage it would not occur to speakers to detach such suffixes from the primitives to which they have tradi tionally been welded and to experiment with their free interchange. Thus, in contemporary English, it is inconceivable that anybody should toy, except in jest, with the substitution of -ity for -ness in coolness, deafness, eagerness, or, contrariwise, should try to foist ֊ness on the kernels of facility, stupidity (even if facileness and stupidness eke out a marginal existence in the most otiosely in clusive of our dictionaries), though there admittedly exists a thin fringe of free or almost free variation (briefness beside brevity, scrupulousness alongside scrupu losity). Now assume that in some unidentified language a group of such func tionally related suffixes have been yanked loose from their moorings to specific derivatives and are thus freely available to enter into new relationships with an unlimited number of primitives. Make the additional independent assumption that for some undisclosed reason—narrowly linguistic or broadly socio-cultural— the tone-setting members of that speech-community are not, at this crucial juncture, economy-minded, i.e. are disinclined to seize this golden opportunity to effect the drastic reduction of the various 'morphs' of a single 'morpheme' to just one 'morph' (the equivalent of -hood, -th, -al, say). In the reshuffling of lexical and suffixal units that this mood of permissiveness is bound to entail, certain nuclei will predictably crystallize around chosen key words. If, for instance, the local counterparts of -al and -ment are each with striking frequency found at tached to a conspicuous primitive, then these particular derivatives may well act as magnets in the ensuing reapportionment, attracting to the separate folds of -al and -ment, respectively, additional primitives sharing with the key words some such isolable feature as the number and configuration of syllables, the stress pattern, the grammatical form class, the semantic spectrum, etc.—regardless of etymological circumstances, of which the present generation of naïve, aggressive speakers is, of course, utterly unaware. In a broad perspective it would be quite inaccurate to attribute to these secondary determinants, for all their importance, any major power of causation; they merely represent so many thin threads tying, in moments of cataclysmic flux, countless lexical items set loose by the flood to a few islets of relative security, i.e. of resistance to the unleashed forces of change. Primary factors of causation thus include the actual disruptive forces in linguistic growth; in contrast, secondary determinants, being analogical by nature, account for partial stabilization. These very sketchy—and contrastively styled—illustrations of problems in the causation of language change do not begin to exhaust the range of legitimate in quiries into this domain, which is at the very heart of genetic linguistics. It re mains to be seen whether the principles of linguistic etiology lend themselves to encoding in a rigid sequence of statements or formulas, or whether, on balance, a looser, more discursive presentation, of the type favored in historical research, is to be preferred as more germane to the analysis of events, in recognition of their infinite complexity.

. HISTORY OF LINGUISTICS

HISTORY AND HISTORIES OF LINGUISTICS

UNDENIABLY, a flurry of book-length publications bearing on the history of linguistics has of late hit the market. The specific reason, or reasons, for this unforeseen surge — as regards both the sudden need to impart such knowledge and the equally sudden demand for it in such generous quantities — is in itself a tantalizing problem, which future historians of linguistics, in focusing their attention on the mid-'sixties, will some day have to ventilate. This boom is something of an innovation and, strangely enough, almost coincides with the centennial of the possibly most distinguished and ambitious venture of its kind, Th. Benfey's bulky classic — limited, it is true, to the German scene (Geschichte der Sprachwissenschaft und orientalischen Philologie ...). From 1870 to I960 fairly little along this line of intro- or retro-spection was accomplished, except in introductory chapters to all manner of Grundrisse and to textbooks;1 concei vably the most valuable and indisputably the most honest studies in this domain were those explicitly confined to a given period, such as H. Steinthal's Geschichte der Sprachwissenschaft bei den Griechen und Römern ... (1863; 18902). One way of measuring the temporary decline of historical self-concern among linguists during that long stretch is to compare, point by point, L. Bloomfield's rather circumstantial treatment of these problems in his Introduction to the Study of Language (1914) with his distinctly crisper and a shade impatient account in his far more influential and original book Language (1933). Generally speaking, first-rate scholars, with rare exceptions, had readily understandable reasons for steering clear of book-length probings so slanted ; one such exception was W. Streitberg, who piloted a five-volume Geschichte der indogermanischen 1 One of the most sorely neglected sketches, even though it is replete with factual information, is G. Gröber's triptych, introductory to his Grundriss der romanischen Philologie, I (Strassburg, 1888): "Geschichte der rom. Philol." (3-139); "Aufgaben und Gliederung der rom. Philol." (140-153); and (in collaboration with A. Tobler and with an expert in paleography) "Anleitung zur philologischen Forschung" (155-280); in the revised 2d ed. (1904-06) there appeared gener ously expanded versions of these three sections (1-185, 186-202, and 203-368). To the extent that historical linguistics could be lifted from its philological matrix and its ties to medieval literature could at all be severed through skillful surgery, W. Meyer-Lübke's Einführung (1900, 19092, 19203) may have partially replaced Gröber's third section (but note that the original edi tion of the Einführung preceded the revised "Anleitung" by a margin of four years). For a full appreciation of Gröber's somewhat hidden talents and continued relevance see his celebrated disciple E. R. Curtius' commemorative essay, included in the latter's Gesammelte Aufsätze zur romanischen Philologie (Bern and München, 1960), pp. 428-455; cf. here p. 544, below. H. M. Hoenigswald has further drawn my attention to B. Delbrück's Einführung in das Sprachstudium . . . (Լ., 1880; 18842; 18933), a title later changed to Einleitung in das Studium der indo germanischen Sprachen (19044; 19085), all five editions with a revealing subtitle: Ein Beitrag zur Geschichte und Methodik der vergleichenden Sprachforschung; Eng. tr., 1882.

See the Bibliographic Guide (p. 82) for notes on the books discussed in this article. Part IV of the original text, contributed by Margaret Langdon, has here been omitted.

50

HISTORY OF LINGUISTICS

Sprachwissenschaft... (1916-36), leaving it, unfortunately, a torso. The present review article aims at a preliminary sifting of representative writings of recent vintage and at the delineation of problems — acknowledged or unacknowledged — which they raise in the minds of the readers. Parts I-III, contributed by the senior member of our team, examine certain prolegomena to this genre of inquiry ; then categorize the sources of information available to the chronicler or the full-fledged historian (with special attention to life-sized likenesses, such as those recently anthologized in T. A. Sebeok's Portraits of Linguists) ; and, against this backdrop, finally survey some major recent publications, particularly those authored by European experts. Part IV, for which the junior member of our team has assumed sole responsibility, concen trates on just three such books — one from Belgium (Leroy), one from Sweden (Malmberg), one from Yugoslavia (lvié). The critic selects for her analysis a single facet of the problem: the picture these three authors, from their separate European watchtowers, have drawn of the North American landscape. Each member of our team has acted independently and should receive plaudits or blame solely for his or her own share in the common venture. I Whatever the reason for the sharp increase in the current output, it is wholesome to remind oneself that the history of linguistics, like all specialized varieties of history, is — by virtue of its mixed status — an exceptionally exacting discipline, presupposing dual expertise. Much as a first-rate historian of che mistry or, for that matter, of music should fulfill two requirements : demon strate the specialized knowledge of a scientific or artistic domain and, at the same time, prove his ability to grapple with a temporally slanted projection of human activity, with full attention to the key problem of causation, so a historian of linguistics, before trying his hand at a challenging assignment, ought to know a good deal about intellectual history (embedded within the matrix of general history) and about the unique, more technical aspects of linguistics. This remark may sound trivial, but the point needs to be re stated. Let me grant from the outset that a scholar eager to report on "recent trends" in any organized human endeavor need not show his credentials in thorough historical training. Such an eyewitness report, often propagandistic, likely to be frothy and bubbly rather than plodding, has its limited usefulness if it is timely and witty. At its weakest, it is journalistic, reportorial in tone and implications ; at its strongest, it may be likened to a distinguished piece of literary criticism, presupposing, above all, taste, alertness, and sensitivity on the part of the ar biter, and a bare minimum of erudition. The greater the time-depth at issue, the more legitimate the appeal to the straight historical approach, involving as its prime desiderata a sense of perspective and an attitude of objectivity. These qualities cannot be acquired through quick immersion, unless we are striving after "instant history" ; considerable patience — and genuine eagerness to do archival work and to engage in interviewing — are needed to amass the neces-


51

sary raw data. When it comes to sifting them, experience and detachment be come additional qualifications. Perhaps the four most desirable conditions for developing satisfactory working habits as a historian of linguistics are to have personally witnessed the rise and decline of one or more fashions ; to have lived in several countries long enough to have absorbed their disparate intellectual climates, from grammar school to university seminar ; to have cultivated, with a certain alacrity, more than one major genre of linguistic investigation; and to have focused attention, at least during one's years of apprenticeship, on a period definitely closed, with whose chief protagonists the writer has not been so closely involved, in terms of personal relations, as to have developed any bias, be it animus or subservience. In reality, most linguists — including those writ ing historical accounts — are youthful, idealistic, enthusiastic and, by the same token, often both uncritical in their endorsements and unfair in their con demnations and, above all, in their — more fashionable and more deadly — tacit omissions. Such older linguists as may turn to reminiscing or to digging into the antecedents of their discipline are plagued by a mood which is the reverse of the same medal : Feeling pushed aside by the pitiless tide of history, they will tend to indulge in the idealization of a past time segment, viewed in nostalgic retrospect as a Golden Age. The nouvelle vague status of linguistics at this juncture hardly favors any forceful, yet balanced, historical appraisals com parable in sustained quality to the finest, maturest research carried out by, say, seasoned art historians. Any analysis of the history of linguistics that aims to transcend a mere chronicling of recent events must, first of all, raise this question : Do the various schools of thought which, in the name of linguistics, have struggled so hard for supremacy each really offer an alternative solution for essentially the same set of problems? Have they carved comparable edifices of structured knowledge from a congeries of random ideas and loose bits of factual information ? Or have they bandied about the same label, applying it over the years to fairly dif ferent provinces of cognition, overlapping only to a slight extent or, worse, not at all ? If the latter situation has obtained — and the last few decades have witnessed events where this almost seemed to have been the case — then one can no longer, without disclaimers, speak of the organic growth of one welldefined discipline bent on steady self-improvement, but only of an erratic sequence of thrusts into terra incognita — some strong, others feeble — all of them involving the application of an obviously appealing, intriguing label — namely the WORD linguistics — to all sorts of scholarly endeavors loosely connected with inquiries into human speech. It is this peculiar state of affairs which, again and again, emboldens adherents of a given, often very respectable, school of thought so to lose their sense of perspective and proportion as to maintain, in a rather humorless vein, that just because certain genres of serious and honest research bearing on language happen to fall outside their private definitions of linguistics "at its best", these genres do not deserve to rank as fully accredited among "real" linguists. One hears more and more frequently remarks to the effect that important subdisciplines, placed at the periphery or even at the very core of linguistics

52


(metrics, poetics, phonetics, semantics, etymology, onomastics), simply fail to qualify for inclusion, that their representatives and mouthpieces do not "belong". Fortunately, every ten years or so, this discriminatory practice, as if in response to some hidden rotating mechanism, hits, with unabated fury, yet another group of unsuspecting victims. A noted student of physical sciences, upon observing at close range some dialogues de sourds carried on at a linguistic conference, once remarked goodhumoredly that the participants reminded him of a corps of astronomers, each training his powerful telescope on the same immense dark celestial body. Of the total surface of that body only relatively small, bewilderingly diversified por tions were caught by each instrument's lens. Small wonder that every astrono mer present distrusted his fellow-scientists' reports on what they had seen; worst of all, alarmingly large patches of the dark surface had completely eluded the battery of lenses. By the same token, it becomes increasingly difficult to claim any measurable progress — except along certain predetermined lines — where the prime condi tion for gauging advances, namely neat comparability of old and new, is almost entirely lacking. All one can safely contend is that one way of attacking, on the level of academic discourse, certain problems pertaining to language is of vastly greater appeal to a given generation of students than is another. Occasionally spokesmen for a new theory or technique will be able and eager to demonstrate that they are in a position to settle questions which an earlier generation was utterly unprepared to raise, let alone to solve. Such welcome demonstrations verge on a proof of genuine advance; unfortunately those announcing such victories seldom bother to mention retreats from excellence which may with equal justification be charged to the very same revolutionary movements. Given these two forms of bias, it is hardly surprising that the picture of an inspired quest for ever clearer understanding should so often have been blurred and stained by a wide range of all too human stirrings. Such familiar "im ponderable" ingredients of intellectual struggle and strife as persuasion, seduc tion, the diagnostically correct anticipation, by an experienced puppeteer, of a younger generation's secret cravings, the skillful appeal to pride and other sentiments (plus, in extreme cases, the exploitation of prejudice), in short, manipulation, have all, for better or worse, played their part in the vicissitudes of linguistic fads and fashions. The fact that the record of linguistics shows, in lieu of a neatly drawn line, a strangely curved trajectory reminds one, incidentally, of the biography of languages themselves. For the sake of convenience and tidiness — here invoked in reference to a single language's development symbolized by the tags X, Y, and Z for consecutive periods — some prefer to think of Stages Y and Z as, simply, later phases of Stage X. Ever since a historic pronouncement erroneously attributed to Saussure, spokesmen for this opinion have had a pretext and an authority for deeming it downright inelegant to appeal to extra-linguistic factors. In reality, this sacrifice at the altar of elegance involves, if pushed to its ultimate consequences, a pathetic self-delusion. Undeniably, the geographic, ethnic, and social matrices of X, Y, and Z may, in characteristic instances, be so


53

different as to bar any straight continuity, hence, indirectly, to discourage a point-by-point comparison. Our written records are, almost by definition, sorely deficient — above all, irremediably fragmentary — and over any ex tended period of time the country in which the language under scrutiny has been primarily used may very well have changed its total size, its outer con figuration, and, most important, the inner apportionment and contour of its cultural zones and administrative units. Then again, the scanty records available may involve totally irreconcilable genres, some gratifyingly close to the spoken language presumed to underlie the records (farce and comedy ; epigram, diary, and epistle), others highly stylized (ode and tragedy, historiography, legislative and notarial documents). Finally, those originally charged with the transmission of records may have been members of widely discrepant social groups (princes, aristocrats, scholars, clerks, monks, teachers, soldiers, actors or minstrels, courtesans, vagabonds, wall-scribblers). The primary sources we actually handle in our archives form, as is well known, a kaleidoscopic assemblage of fortuitously preserved bits, scraps, shreds, and remnants. To transform this medley into an ordered whole is the challenging task of diachronic linguistics allied with philology — a task which, through the confusing heterogeneity of data, thus runs parallel to the equally taxing assignment given a historian of linguistics. (Predictably, scholars gravitating toward the descriptive approach will, on the average, make far less satisfactory historians of linguistics than will those steeped in reconstructive operations.) It may be argued that most, if not all, disciplines are slightly "fuzzy" or "jagged" around their edges; that the transfer of responsibility for certain strings of problems from one discipline to another has become an almost daily occurrence ; and that the steady rise (and, often, rapid decline) of transitional disciplines profiting from their strategic position astride two or more wellestablished domains — say, astrophysics, biochemistry, anthropological lingui stics, sanitary engineering — bespeaks the fluidity of all border-lines provision ally drawn inside the edifice of knowledge. All this may be true (and, to some of us, heart-warming) ; but linguistics undeniably represents an extreme case of interlocking, being well-nigh unique in this respect, as in so many others. A tremendous chunk of the legitimate corpus of organized linguistics can be effortlessly accommodated within the adjacent provinces of anthropology (properly understood), sociology, psychology, and, of course, philology at its richest. It is this circumstance, not to say inherent quality, which best explains the — oft-bemoaned, but seldom accounted-for — astonishingly late date of the self-identification of linguistics as an autonomous discipline. These links to parallel, especially neighboring, repositories of knowledge can not with impunity be rashly destroyed. Thus, an impatient L. Bloomfield, after espousing a new commitment to "pure linguistics", clearly over-reacted to an obsolescent, fundamentally classicist cultural pattern in trying so to delimit the newly emancipated discipline as to divorce it radically from old-style philology. On several memorable occasions, while at the peak of his prestige, he indulged a personal whim by placing linguistics, in his favorite scheme of scholarly pur suits, at the same distance from philology as from, to cite sciences at random,

54


astronomy and botany. 2 But — whimsies, vagaries, and underlying frustrations apart — the fact remains that philology and linguistics, as currently defined, are intimately related. Indeed, they are so inextricably enmeshed that for cer tain significant operations equal expertise in both fields is strictly mandatory. Generally speaking, SCHOLARLY DISCIPLINES ARE NOT EQUIDISTANT FROM ONE

The enthusiastic, uncritical adoption—by a whole generation of excited and, in part, embittered followers — of certain polemic overtones in some of Bloomfield's strident or sarcastically low-keyed pronouncements for a while threatened to wipe out historical linguistics from the American scene. As with all evolutionary modes of inquiry, there inevitably arises, for the historian of linguistics aspiring to be more than a mere annalist, the question : Why is one particular school of thought eventually supplanted by another? Is it (as historical sketches will ordinarily state) because the loser in the struggle for hegemony actually offers fewer stimuli and outlets or performs less satis factorily (e.g., leaves certain legitimate questions unanswered)? Or is it because — given the acceleration of wear and tear in the total pattern of our restless culture and, in particular, the intrinsic mobility of its avant-garde fringe into which linguistics has gradually been shoved—any kind of sharply focused intellectual enthusiasm is bound to evaporate within a matter of decades unless fresh vigor is skillfully instilled into it — in which case the craving for novelty, the hankering for change actually precedes "revolutionary" upheavals and literally invites the infusion of resoundingly new ideas? Can one reduce the immensely complicated situation to clashes between magnetic personalities contending for power and prestige — 20th-century condottieri, as it were, who, among other and nobler qualities of leadership demanded of them, cannot help acting as shrewd impresarios, effective propagandists, colorful embodiments, and astute, tireless talent scouts ? Is our boat being rocked so violently because contagious shifts occur in other, perhaps more powerful sectors of social, intellectual, and artistic life and, given the interplay of attractions, unavoidably suggest tempting alternatives to, and alluring departures from, accepted prac tices to linguists trained, to the point of boredom, in striving after certain goals ? Is it, finally, the rise and fall of influential national cultures long known as patrons of linguistic research that lift certain favored varieties of active linguistic curiosity to unforeseeable heights, only to doom them later, with a change in political fortunes, to their slow decay, if not drag them to their ultimate ruin? Should one, by any chance, reckon with the mutual concomitancy of all, or at least some, of these — and possibly other — isolable forces and, if so, into what pattern can one fit their interplay? Above all, does one dimly recognize in all these bewilderingly multifarious events certain recurrent features worthy to be labeled universals ? And can one expect to organize such ANOTHER.

2 Rather characteristically, on the opening page of Vol. I (1925) of Language one reads, in Bloomfield's programmatic article "Why a Linguistic Society?": "The layman—natural scientist, philologian, or man in the street — does not know that there is a science of language". Bloomfield's book Language (1933) contains derogatory or condescending statements on philo logists (e.g., 22) and abounds in remarks of this type: " I t is important to distinguish between philology and linguistics .... since the two studies have little in common" (512).


55

universals cogently into smoothly contoured spirals or cycles ? The answers to these programmatic questions would require a book rather than the meager space of a few pages. If such a book could at all be written, it would be radically different from those currently flooding the market, surpas sing them in both candor and sophistication. I cannot begin to discuss the intricate issues in any systematic way, though I may have unwittingly in sinuated certain tentative answers. However, let me try to illustrate with a few eloquent examples the conflation of several currents at many crucial junctures. The Neogrammatical movement certainly should be judged on its own doctrinaire merits and on the basis of the performance of its participants ; but, having been firmly moored, especially at its start, to the tone-setting German school of Indo-European studies, it does not lend itself to discussion, so far as its incipient dynamism is concerned, without reference to the euphoria that gripped German university teachers in the late 'sixties and early 'seventies of the past century — the moment of triumph in the history of their country. By the same token, while Saussure's posthumous Cours (1916) invites and richly deserves assessment on the strength of its splendors and accomplishments, its instantaneous impact, especially outside Central Europe, cannot be properly measured without some allowance for the feelings of that time : The acceptance of the leadership of a French-Swiss genius connoted for many Westerners then opposed to Germany a strongly desired, rationalized escape from the world of Brugmann, Leskien, Osthoff, and Paul. In Germany herself something similar was happening, in the ranks of Romanists : While K. Vossler's earliest polemic writings of the years 1904-05 (Positivismus und Idealismus in der Sprachwissen schaft; Sprache als Schöpfung und Entwicklung) had but a faint immediate echo, the yearning for a radical change, for rejuvenation after the loss of a war and the collapse of a régime, allowed Vossler's ideas to come to the fore in the period 1921-27, temporarily eclipsing other, more solid and austere, traditions of Romance scholarship. As regards the trend toward cyclic configuration of research in language, one can, schematically, diagram a certain crumbling of the initial impetus, a gradual loss of vigor extending over a span of time which may correspond to three or four generations. Let us assume that the perspective favored by a given school is the spatio-temporal; on this understanding one witnesses, as a rule, two parallel processes unfolding, as it were, on two planes. On the analytical level one observes a quest for enhanced precision and for ever closer approximation to completeness, a search conducive to slower work, to more circumspect conjec tures, and to the choice of an increasingly fine network and a more and more elaborate battery of filters — all of this, inevitably, to the detriment of daring ideas, broad vistas, sweeping generalizations, exciting hypotheses, and, in general, a zest for unhampered experimentation. On the level of the emotional subsoil — which, if nourishing, can superbly feed a worthy intellectual impulse — one observes the gradual evaporation of the original sentiment that spurred the pioneers to stake their reputations on risky, exacting ventures. To put it differently : A group of independently-minded enthusiasts launches a number of hazily defined hypotheses, involving wide expanses of territory, long stretches

56


of time, dimly recognizable alliances of tribes, powerful jolts not immediately traceable to an epicenter and their apparently sporadic reverberations. This audacious "first draft" is jotted down and before long published—to spark discussion — at the risk of numerous errors (lacunae, oversights, hasty in ferences, etc.), since grand strategy calls for speedy operations. The next generation, less pressed for time and sentimentally not nearly so aglow, identi fies the gaps and tries to fill them, weeds out the errors one by one, and modifies in important details the theory bequeathed, in fluid state, by the trail-blazers. The cool-headed generation concluding the cycle — comprising the détaillistes — is engaged in leisurely, small-scale mopping-up operations, ferreting out and correcting the residue of inaccuracies, but no longer bothering to reshape the general outline of the theory. A slightly different variety of the gradual blunting of an originally razorsharp cutting edge is offered by research in dialect geography, ordinarily clustering around a linguistic atlas : Here the step-by-step exhaustion of fresh methodological insights is the main cause of debilitation. The first small batch of a total of, say, one thousand maps yields radically new ideas concerning — to cite topics at random — areal patterns, folk etymology, abandonment of eroded forms, compromises between metropolitan and regional preferences, dialect boundaries, recoil from taboo (from polysemy, from homonymy), substitutions of more effective, suggestive synonyms, concessions to humor and playfulness, and the like. The more maps one subjects to scrutiny, the more severely limited one's chance of scoring a new major methodological breakthrough. Like every other investigative technique, the cartographic projection of linguistic facts obeys its own laws of diminishing returns. Small wonder that dialect geography has tumbled from the position it occupied at the turn of the century. The frustration ultimately produced among late-comers marks their incipient readiness to switch to a less labored approach, beckoning from other quarters. As concerns the depletion of the emotional subsoil — a very serious hazard —, recall that early European comparativism of the type associated with the advent of Franz Bopp throve, for all its purely intellectual accomplishments, on such moods and delights as the fascination generated by the thrilling rediscovery of ancient India, as the sustained magic spell cast by classical Antiquity (among an élite steeped in the knowledge of Latin and Greek), and as the all-pervasive romantic Schwärmerei favoring persistent curiosity about, even tacit selfidentification with, the spokesmen's own ancestral — i.e., paleo-Germanic — culture. Modern Indo-Europeanists the world over have, of course, inherited the scholarly kernel of this legacy, but today's society-at-large is no longer marked by the same emotional undercurrents, least of all in countries and continents — like North America — physically remote from the cradle of Old World civilization and dominated by forward-looking rather than nostalgically retrospective strategists. Hence the difficulty of recruiting strongly motivated devotees and, even more, of sustaining the enthusiasm of those converted — against the grim background of widespread indifference. The model of a phasic development here outlined represents, at best, a PLAUSIBLE or TYPICAL sequence of events ; it should certainly not be construed


57

in deterministic terms as something obligatory and ineluctable. With a modicum of mental elasticity and of deftly directed will-power one can, in fact, guard against anticipated decay. A truly perceptive influential teacher, sensitive to the ever-threatening change of tides, will curb his desire to promote most energeti cally such students as seem to follow all too faithfully his own line of research and will instead lend his maximum encouragement to followers a trifle less loyal and less sheltered, but endowed, by way of compensation, with versatility, resilience, fierce independence, and enviable dynamism. Such a policy can later keep the teacher's legacy, or at least its finest ingredients, in a state of continued fluidity, thus preventing it from being swept away by any powerful reaction. (In fact, a minimum of inoculation may be wholesome ; the far-sighted teacher best-known as an etymologist, stylistician, or cultural historian will be well advised to expose his most enterprising students to the serum of abstract linguistic theory, and vice versa.) On the other hand, a resourceful younger scholar, unguided by any senior expert but feeling greater affinity for the preferences of an earlier generation rather than those of his own, will successfully avoid the hazard of epigonic sterility and will salvage what he most highly treasures by deliberately transcending his intellectual heroes' time-bound scope and techniques. Only thus, by dodging onslaughts and constant maneuvering, does one stand a chance of achieving CHANGE WITHIN THE PATTERN OF CONTINUITY.

The scheme here proposed is, one hopes, no less realistic than the one widely associated with Thomas S. Kuhn's sensationally successful book The Structure of Scientific Revolutions (1962). Admittedly, various characteristic features of Kuhn's paradigm, in particular the image of a young scientist — surrounded by a phalanx of youthful supporters — shattering the doctrine of an entrenched academic régime or establishment, harmonize with the mood of the late twen tieth century and, above all, immensely flatter the younger generation.3 Few slogans have, of late, been more effective than "change of guard", "Young Turks", and the like; 3a certainly the norm has been at all times for the new generation to rebel against, and try to supplant, the old. But since human society, especially its top échelon of intellectuals, is more complexly organized than a beehive, struggles between schools of thought exceed in degree of intri cacy (and, one hopes, in level of ethics) the seasonal confrontations between an older and a younger queen bee, each surrounded by her own swarm. Even Yu. Tynjanov's rival formula of Archaizers vs. Innovators, i.e., of conservatives locked in rivalry with experimenters within a given generation, fails to do full justice to numerous situations. Specifically, one must distinguish in the annals 3 Kuhn's provocative book furnished the leitmotiv for the Anthropological Research Symposium — sponsored by the Wenner-Gren Foundation and piloted by Dell mes — on "Revolution vs. Continuity in the Study of Language" (Burg Wartenstein, Aug. 15-25, 1964); the papers presented on that occasion were later dispersed. The extent to which Kuhn's ideas haunt the imagination of younger linguists can be seen from the literary critic R. Sklar's article — based on interviews? —: "Chomsky's Revolution in Linguistics", The Nation, Sept. 9-, 1968, pp. 213217. 3a For an Anglist and literary historian's humorous view of Young Turks see R. H. Robbins, "Old Guard and Young Turks: the Problems of Reviewing", ASNS, CCV (1968), 177-188.

58


of linguistics between the actual occurrence of a breakthrough and its wide spread recognition. An observer as experienced as A. L. Kroeber reckoned with a typical time lag of thirty years between the two phases of any such process ; but the accelerating speed of communications, the proliferation of symposia and congresses, the mushrooming of visiting professorships, and other parallel manifestations of a "one-world climate" may have substantially reduced this temporal buffer. Still, it is wholesome to remind oneself that the early phonemic theory of the pioneering Kazan' School is one thing, and the sweeping "victories" of the phonemicists in the second quarter of this century quite another. Aside from such instances of DELAYED recognition (involving, in extreme cases, POSTHUMOUS acclaim), there are also on record examples of REVIVALS, in cluding several returns, under varying sets of circumstances, to W. von Humboldt (Sapir/Whorf, Chomsky), and a few exceptional occurrences of REHABILITATION.

The problem of (instantaneous, gradual but slow, blocked) recognition is itself arresting for the cultural historian and forms part of the broader issue : diffusion of scientific thought. Geographic distance, counter to expectation, turns out to be a minor circumstance (a data-oriented German Switzerland's massive resistance to Saussurean abstractionism is a perfect illustration of this point), though changes in residence and affiliation may facilitate contacts and hasten the spread of ideas, key-words, and techniques (witness the influence exerted by Russian emigres in Central and Western Europe throughout the 'twenties, and the impact of European newcomers on the American scene after 1935). There are some noteworthy instances of conspicuously late appointment to a teaching position (when rewarded with a chair at Liège, Jean Haust began to assert his influence in Walloon Belgium as a man approaching his sixties) ; of a scholar's slow growth toward his pinnacle of originality ; of the gradual or sudden removal of powerful obstacles. Hugo Schuchardt, for one, published his famous pamphlet against the Neogrammarians in the mid-'eighties, but his impact made itself felt only after the advent to power of dialect geographers some twenty years later.

II Sebeok's anthology of obituaries (Portraits of Linguists) being a prime example of a source book, there may be some point in prefacing to its assessment a brief typological inventory of the varieties of primary material available to the historian of linguistics (and of any comparable discipline). These may tentatively be broken down thus, with a minimum of illustrations accompanying each category : (a) Autobiographic sketches and self-appraisals, some found in a scholar's private files by his executors, others published in his lifetime (e.g., Karl Pietsch's self-portrait in Һ, X X V I I [1929-30], 388-394, ushered in by G. T. Northup's appreciation). A special genre is


59

the "curriculum vitae" attached to doctoral dissertations in Central European countries and ordinarily relevant only to the extent t h a t it sheds a few shafts of light on the candi date's teachers and on the atmosphere prevailing in certain lecture courses and seminars. (b) A scholar's memoirs. Not many linguists have shown the inclination (or possessed the talent) to write book-sized recollections, but a few have interspersed retrospective analyses of important episodes with personal remembrances of things past, e.g., A. L. Kroeber in his evaluation of H. W. Henshaw's specific share in Major J . W. Powell's celebrated classifica tion (1891[-92]) of this country's Indian languages. 4 (c) Authors' prefaces, introductions, and, more sparingly, epilogues to their own writings, insofar as they contain statements highlighting the — sometimes dramatic — history of the given research project, its relation to earlier spade-work and, conceivably, to rival under takings as well as the writer's prognosis as to the presumable future course of events. Particularly valuable, in this context, are prefaces to radically revised editions, because they may accurately record (or, at least, imply) changes in the surrounding intellectual climate. (d) Exchanges of letters. Generally speaking, concern with Gelehrtenkorrespondenz, so characteristic of the 19th century, has waned in our own time ; I cannot recall any recent counterpart, at least not on the Romance side, of the slim volume of F . Diez's Freundes briefe — despite the fact t h a t some of our near-contemporaries have poured into their missives an impressive corpus of priceless information and original ideas (here the names of G. Guillaume and J . J u d come to mind). 5 A veritable treasure-trove of older epistolary reactions, on the European scene, is hidden away in the Ascoli Archive (Rome). 6 (e) The composition of all manner of Festschriften (Melanges, Miscellanee, etc.) geared to special events in a distinguished scholar's private life or academic career, or perhaps issued on the occasion of his death (or even of an anniversary of his death, as in the case of P . Henríquez Ureña). With due allowance for all kinds of minor distortions and interferences, one can thus piece together a fairly reliable picture of Sapir's "inner circle" from a study of the six-column Tabula preceding, in 1941, the Essays in Memory of Edward Sapir (Language, Culture, and Personality), as well as from the dissection of the volume itself. (f ) Prefaces, usually from the pen of a colleague or some favorite student, to collections of, or selections from, the writings — or, at least, the shorter writings (Kleinere Schriften, Scripta Minora, Reliques) — of a revered master often contain nuggets of noteworthy information. Such invitational prefaces or Editor's Introductions may be serious and substantial, measured by the yardstick of their scholarly content (witness Leo Spitzer's to his Hugo Schuchardt Brevier, 1922, 1928 2 ; and David G. Mandelbaum's to the Selected Writings of Edward Sapir in Language, Culture, Personality, 1949), or they may be chatty and anecdotal (as is true of A. Torres-Rioseco's to the J. E. Gillet Memorial Volume of HR, X X V I I I [1959] or of A. Zamora Vicente's to the three-volume set of Studia Philologica dedicated to Dámaso Alonso [1960-63]). 7 4 See AnL, I I (1960) : 4, 1-5, for an elaboration on W. C. Sturte vant's earlier note in IJAL, XXV (1959), 196-199. D. H. Hymes, in turn, provided a kind of post-script to Kroeber's remi niscences; see AnL, I I I (1961): 6, 15f. 5 Selected letters may also form part of full-scale biographies, entering into portfolios of "intimate", "authentic" illustrations. One thinks of Margaret Mead's luxuriously printed An Anthropologist at Work; Writings of Ruth Benedict (Boston, 1959), containing annotated specimens of R.B.'s correspondence with Sapir and Boas. 6 D. Gǎzdaru, who had temporary access to this material a quarter-century ago, has since published several chunks of it in miscellaneous journals and Festschriften. For a bird's-eye view of his many-pronged activities see the Prefatory Note to his book Controversias y documentos lingüísticos (La Plata, 1967) and my rev. in RPh, XXI, 360f. The latest major event along this line is the appearance of the Epistolario de R. J. Cuervo y H. Schuchardt, ed. D. Bross (Bogotá, 1968). 7 A border-line case is the publication, especially at a time of emergency, of a scholar's biblio graphy as a separate slender pamphlet. Such an honor was accorded Amado Alonso at the very moment when he departed for Harvard (1946), leaving behind his flock of Argentine disciples; on balance, María Rosa Lida's "semblanza" heralding that modest venture has left a far more powerful impact than Alfonso Reyes' lax Foreword to the 1953 A.A. Memorial Volume of the NRFH.

60


(g) Congratulatory or commemorative articles, which typically herald Testimonials, Memorials, Étrennes, Coronae Querceae, and the like, but may also appear on such occasions as a change of helmsman in the piloting of a learned journal or of a monograph series (one example: H. Lausberg's vignette of G. Rohlfs as editor, ASNS, CXCI [1954-55], 1-2). (h) Certain classes of book reviews, when the occasion calls for more than a sober ap praisal of the volume at issue and when critic and editor respond to the challenge of un usual circumstances. Unique opportunities for boldly drawing such balance sheets arise in the following situations: [α] When the book under scrutiny was the author's last venture, particularly if it also marks the peak of his efforts or, alternatively, the close of an epoch ; [β] when a major posthumous book is involved; [γ] when a book is reissued long after its original publication date, thus inviting a "second look" in a radically new general climate of taste and opinion, with full attention to some earlier responses to it, or when a long-awaited translation appears on the market ; [δ ] when a (predominantly, posthumous) collection of scattered articles encourages the reviewer to see a scholar's total œuvre in a broader perspec tive. 8 In all these cases we observe the crystallization of an intermediate genre between an ambitiously planned book review and a straight necrology. (i) Provocative summings-up, in each instance, by one tone-setting participant of a given symposium, esp. of an ad hoc conference (a "milestone") devoted to the elucidation of a single major issue. 9 (j) Such incidental shreds of information as can be gleaned from digests of lectures, minutes of discussions (BSLP), announcements of meetings — which may range from regional convocations to international congresses —, bulletins and catalogues of universi ties, research centers, and institutes, the old-style "Programmschriften" issued by the better German gymnasia, open letters of polemists, candid underground journals (a few issues of one such "maquisard" periodical appeared in the Netherlands' academic jungle in the early 'fifties), programmatic statements scattered over tracts, opening pronouncements made on the occasion of newly started journals and series, incisive editorial comments disguised as page-fillers, formal promises — not always redeemed — of "forthcoming contributions", book-dealers' descriptions of specialized collections on sale, traces of feverish rewriting left on heavily marked-up galleys (some of them still in the printer's hands), university and municipal librarians' records of the circulation of key-books, regis trar's records of the attendance in courses (number and names of persons enrolled), 10 and similar seemingly worthless bric-à-brac which in the hands of a deft, inspired historian can easily become an inexhaustible gold-mine.

Against this background of the available latitude of sources, S.'s self-limita tions to portraits of dead linguists represents a deliberate retrenchment ; to say the least, a few figures of commanding stature now past their prime and deftly portrayed in their lifetime might have marginally qualified for inclusion had the criterion of actual demise been enforced less rigidly. True, not every single 8 For all four (or five) possibilities here envisaged numerous illustrations can be offered. Class (b) may be exemplified with a review of my own (BPh, IX, 237-252), centering about A. Alonso's De la pronunciación medieval a la moderna en español but aiming at a word-portrait of the colorful author. For Class (d) it seems proper to adduce two reactions to Mandelbaum's Sapir anthology ("editio maior"), each based on a broad spectrum of personal recollections in addition to a wide range of readings : Z. S. Harris' in Lg, XXVII (1951), 288-333, and II. Hoijer's in RPh, V (195152), 311-315. 9 Several such position papers have, in recent years, been contributed by R. Jakobson. One also thinks of A. L. Kroeber's "Concluding Review" (Chap, xx, pp. 357-376) rounding out An Appraisal of Anthropology Today, ed. Sol Tax (Chicago, 1953). 10 I certainly do not mean to be facetious. It is well known how the minute examination of successive printer's proofs has furthered Balzac studies. Such students of modern literature as elect to concentrate on readers' responses have learned the art of analyzing circulation cards pasted into books loaned to readers by influential and representative public libraries. É. Benveniste has recently persuaded a younger worker to analyze in microscopic detail the attendance in F. de Saussure's classes conducted at the Parisian Ecole pratique des Hautes Études.


61

piece in this gallery of ninety likenesses is a formal obituary culled from a learned journal or yearbook, especially not in the opening volume. One finds, interspersed among narrowly defined necrologies, commemorative essays by non-contemporaries (e.g., F. Edgerton on Sir William Jones, I, 1-18), relevant "flashback" chapters lifted from prestigious book-length ventures (thus, H. von Steinthal's analysis of W. von Humboldt was entombed in the portraitist's book Ursprung der Sprache, published fifty-three years after his subject's death), and lengthy entries reproduced from biographically arranged reference works. But a certain less than attractive air of necrophilia undeniably pervades the entire expanse of almost twelve hundred pages. If the two volumes encompass a total of ninety pen portraits (a final choice made, one reads in the Introduction, from a preliminary collection twice as long), the number of coryphaei so honored is measurably smaller, because not a few leading figures are contrastively represented twice, for the sake of a "stereo phonic" effect. In fact, the somewhat elusive, not to say enigmatic, F. de Saussure has posed as a model to as many as three painters (II, 87-110: R. Gauthiot, A. Meillet, and W. Streitberg) ; so has, as we shall see in a moment, Sir William Jones — though, one gathers, for quite different reasons. The idea of multiple profiling, from different angles, is in itself commendable, on account of the pleasingly pungent ingredient of dramatization it is apt to add. (This procedure can, incidentally, be traced to Leo Spitzer, who, I distinctly recall, enlisted the help of a much younger Helene Adolf in silhouetting Elise Richter against two slightly different Viennese backgrounds [RPh, I, 329-341] — two canvases, as it were, juxtaposed on the same easel.) It is, one readily grants, exciting to find a noted British student of Indies such as Sir William evoked, first, by a modern New World Sanskritist acting, as did Edgerton (see above), as a spokesman for Western culture ; then by S. K. Chatterji, a mouthpiece of un disputed authenticity for present-day India (I, 18-36) ; then again, with a new abrupt change of perspective, by a second Western specialist, this time an authority on Iranian linguistics, G. H. Cannon (36-57) ; or, for that matter, to hold J. Vendryes' magisterial, quaintly detailed evocation, from patently intimate first-hand knowledge, of his teacher — despite but a minor difference in age — and companion-in-arms A. Meillet (II, 201-240) against the terse, but equally incisive assessment from the pen of a favorite Scandinavian follower, Alf Sommerfelt (241-249). But the satisfaction thus experienced is tempered when one realizes that space limitations — due to steep production costs — have set a high price-tag, entailing regrettable omissions, on this indulgence in an otherwise legitimate extra luxury. One laments the exclusion of the aforemen tioned two pieces on Elise Richter, which would have shed a welcome light on Old Vienna, shored up the project on the Romance side (where it is most vulnerable), and gallantly added to the gallery the profile of at least one woman. A truly irreparable loss has been the neglect of E. R. Curtius' unsurpassed essays on F. Diez and, I repeat, on that great systematizer of Romance studies G. Gröber (see the critic's Gesammelte Aufsätze..., pp. 412-427 and 428-455): As a stylistically spell-binding essayist and a master-practitioner of Einfühlung into successive intellectual climates of the recent past the elegant, empathetic

62


Curtius, needless to say, towers above every other contributor to the two volumes. Would not S., all told, have struck a better balance by publishing, on each leader of linguists deemed important enough for such preferential composite treatment, one full-length article plus generous excerpts from complementary or contrastable rival portrayals ? The temporal sweep of S.'s exhibit of paintings is impressive, but the selection of Sir William Jones as the opening figure, heavily underlined by, we recall, the set of three full-length portraits (I, 1-57), is a bit puzzling. Is it a tribute to his colorfulness and to the touch of exoticism in his career ? Can one interpret it as an obeisance meant to enchant or to placate English-speaking readers? Or does it represent a bow in the direction of comparativism ? But then, Jones was not the first to have observed the kinship of Sanskrit and certain European languages (cf. Edgerton's appositely qualifying remarks on G. L. Cceurdoux and N. Fréret [1688-1749], I, 16-18), nor was he a comparativist in any strictly technical sense of the word. A brief prophetic statement of his, however felicitous, hardly justifies the measure of attention lavished on him in this particular context — significant as his share indisputably was in the long growth of Indic PHILOLOGY. To lend a less invidious slant to my stricture, I wonder why such giants as Descartes, Leibniz, and Vico have been swept under the rug. Since all three have definitely been in the focus of modern discussion, it is difficult to believe that there exist no such suitable pen portraits of them as render, at least, partial justice to those phases of their activities and of their thinking which should rightfully catch and hold a linguist's attention. Any pos sible counter-argument to the effect that linguistics, as we currently understand the scope of this discipline, failed to occupy the center of those trail-blazers' life work runs afoul of our awareness that neither was Sir William Jones a linguistic analyst, though he was indeed a magnificent connoisseur of divers languages, dead and living. With all due respect for the prelude to the 19th century and for the historicist vogue it generated, I venture to think (and on this score I happen to be in tune with my time) that the opening selections might well have afforded a few glimpses of genuinely pre-1800 linguistics, whetting the readers' appetite with, say, specimens of rudimentary phonetics as developed by the 16th- and 17th-century teachers of the deaf-mute.11 Since S. has included neither a section of notes nor any running commentary, 11 Why not entice the readers — instead of leaving them completely in the dark on the Spanish tradition — with such choice morsels as T. Navarro Tomás' unjustly forgotten twin articles, "Doctrina fonética de Juan Pablo Bonet (1620)", RFE, VII (1920), 150-177, and, especially, "Manuel Ramírez de Carrión y el arte de enseñar a hablar a los mudos", ibid., XI (1924), 225-266 ? Pablo Bonet is the author of a treatise on the method of instructing deaf-mutes : Reducción de las letras y arte para enseñar a hablar los mudos [letra here means 'sound' and arte 'grammar, technique, handbook']. Ramirez de Carrión, the deftest practitioner of the newly devised method, was less of a writer, though he reserved some space (fols. 127-129) for the matter of deafness in his didacti cally biased Maravillas de naturaleza (Montilla, 1629). The forerunner of these therapists and, reputedly, the actual founder of the Spanish school of remedial pedagogy was Fray Pedro Ponce (1526-84). Vol. III (1921) — unfortunately, inaccessible to me — of the extinct Barcelona journal La Paraula was dedicated to this Renaissance flurry of phonetic curiosity, devoting one major article (23-47) to Pablo Bonet's biography. Elizabethan England could probably match these accomplishments by the linguistic implications of its spade-work in early, crude varieties of shorthand.


63

one can only guess at the criteria that may, in each instance, have guided him in pairing off two contrastive portrayals. In some cases one of the analyst-bio graphers was a compatriot of the man under discussion, while the other was a relative outsider, though not necessarily a total stranger; this dichotomy ac counts, we remember, for the gap separating Vendryes' from Sommerfelťs equally valid (if not equally readable) illuminations of Meillet, and may further apply to M. B. Emeneau's assessment of F. Boas, as distinct from R. Jakobson's (II, 122-139), but certainly not to N. Haislund's and L. Hjelmslev's parallel, uniformly Scandinavian appraisals of O. Jespersen (II, 148-173), nor indeed to B. Bloch's and . . Sturtevant's mutually complementary views, ineradicably American, of L. Bloomfield (II, 508-521), a duality perhaps best explained by a neat trans-generational contrast within the matrix of the same national culture. In other sections of the book, the lenses of the two cameras trained on the same person correspond to the chosen critics' radically different tempera ments, "styles", or ideologies rather than to conflicting social or educational backgrounds; no one would have expected M. B. Emeneau and E. Adelaide Hahn, despite their erstwhile proximity to Yale, to proffer nearly identical portraitures of Sturtevant (II, 365-384). Not infrequently, a combination of national roots and scholarly personality produces a particularly sharp dis sonance : Thus L. Hjelmslev's forceful, nay, revolutionary reexamination of his compatriot Rasmus Rask (I, 179-195), based on the younger Dane's familiarity with the pioneer's less well-known writings, drafts of treatises left unfinished, also letters preserved in archives, and clearly presided over by the writer's commitments to glossematically colored general linguistics so strongly over shadows a well-intentioned Kemp Malone's competent, but woefully conven tional approach to Rask as one of the founding fathers of comparativism (I,195-199) that it makes the latter, cruelly enough, appear anticlimactic, not to say verging on anachronism. In some cases it might have been both fair and advantageous to oppose a frankly eulogistic to a candidly critical, if not down right morose, account of a leading personality's life and œuvre. Conceivably this remark applies to the somewhat skewed vignette of L. Spitzer (II, 522-526) ; but at Spitzer's death Romance linguistics was in such low estate that no firstrate vindications of his possible merits in that domain appeared anywhere in sight, though R. Wellek did prepare a moving and erudite balance sheet of that controversial polygraph's accomplishments as literary critic (CL, XII [1960], 310-330, followed by a four-page bibliography, cf. XIII, 378f.). These preliminary considerations take us to the broader, and also distinctly more de batable, problem of the general relations between biographer and biographee, far beyond the side-issue of infrequent multiple treatment. The necrologies themselves, sometimes, contain discreetly concealed hints at personal ties, as when a self-effacing Vendryes depicts, with a few delicately applied brush strokes, the Latin Quarter apartment in which a still young Meillet used to receive a sprinkling of visitors, including the first generation of his favorite students, or as when Jakobson, by unmistakable implication, pinpoints the receipt of a lengthy missive of his own addressed to Trubetzkoj as the precise turning point in his distinguished friend's development away from bare description or mere old-style history toward sophisticated diachronic structuralism. I n most instances, however, the degree of

64


personal contact or involvement (if any) between the two protagonists of each encounter as well as the exact tenor, nature, and "temperature" of their relations remain, understandably enough, undisclosed. Few fledgling readers of E. Gamillscheg's studiedly restrained, even serene, evocation of K. Vossler (De mortuis ...?) will suspect the degree of vituperative bitterness with which these two antagonists for decades publicly taunted and ruthlessly ridiculed each other from lectern and writing-desk alike. Nor will it dawn on the unforewarned peruser of S.'s anthology t h a t the author of the thoroughly sympathetic biographic sketch of K. Jaberg (II, 454-458) was never privileged to meet in real life the Swiss savant he so visibly admired. In the overwhelming majority of the tributes, however, the tone fluctuates between laudatory and reverential — as befits hero worship — because the wri ters, being former fellow-students, direct disciples, junior colleagues, even successors to the chairs of the great scholars memorialized, or members of the same teams, coteries, circles, academies, and "movements", or perhaps self-proclaimed intellectual heirs to their legacies, tend to present, often no doubt consciously so, subtly IDEALIZED portraits. This attitude is humanly understandable, even touching, and the degree of canonization becomes simply a matter of tact and taste and literary savoir faire. Whoever remembered Benveniste's warmly admiring review (BSLP, X L V I I I , 21-23) of Vendryes' 1952 collection of scattered and outof-reach articles could not but expect, from the pen of the younger scholar, a charmingly human appreciation in retrospect, on the occasion of his aged teacher's death, like the one reprinted here (II, 385-393). Yet, despite these bonds of personal attachment, the devoted necrologist so finely nuances his words as to make unmistakably clear his disinclination to place the dearly loved Vendryes on the same pedestal of solid achievement as the more dynamic and far more original Saussure and Meillet. Not every portraitist uses the resources of his palette quite so skillfully, though few contributors engage in unctuous praise or ful some posthumous flattery ; on this score I have found Benfey's patriotically tinted tribute to J . Grimm (I, 120-169) quite unenjoyable. In any event, the future historian, in using this fine source book, will, again and again, have to discount the ingredient of conventional homage. He will also have to remind himself of the fact t h a t the very few orthodox necro logists aiming at a balanced, austerely critical appraisal, i.e., viewing their task as essentially paralleling the assignment of a book reviewer pledged to objectivity, have proceeded at the severe risk of jeopardizing their own reputation for decorum. Informative remarks on real-life relations — if any — and intellectual cross-connections between the two parties in each confrontation; accurate references to other significant retrospective essays bearing on the same key-personalities and, occasionally, pithy, helpful epitomes of such rival essays; lastly, subjective, necessarily impressionistic, statements, from the editor's comfortable swivel-chair, on the pictorial achievements and shortcomings of the corps of portraitists retained — all these provide the minimum of guidance t h a t some readers, either because they are utterly confused or because they are inordinately demand ing, might have expected to find in a separate section of notes. One regrets to report t h a t no such section — comparable to the sparkling comments lavished by a not yet embittered Leo Spitzer in his truly remarkable anthology Meisterwerke der romanischen Sprachwissen schaft (2 vols. ; München, 1929-30) — has here been organized. 12 Perhaps the predominantly hostile reaction of critics to M. Joos' unceremonious demeanor as arbitrating editor of Readings in Linguistics (I) has served as a deterrent, bridling S.'s volubility. He does speak up, however, with complete freedom from restraint in his Introduction, chock-full of tokens of amenity and urbanity (I, vii-xiv). There, aside from doling out credits to co-workers and consultants, he admits t h a t the extra-rapid rate of attrition in linguistic fashions was the chief stimulus in goading him to examine our discipline's distant and recent past (tantalizingly enough, he does not further pursue this point) ; he disarmingly confesses that one determining consideration in the making of all final selections was his private taste — hence, among other idiosyncrasies, the discernible emphasis on Finno-Ugrian studies ; and he observes that certain categories of obituaries (e.g., those channeled through daily news12 The closest approximation on the American scene is, I believe, Dell Hymcs' approach in Language in Culture and Society (N.Y., etc., 1964).


65

papers) have proved less rewarding t h a n others, written — one gathers — sub specie aeternitatis՛, but what about K. Jaberg's splendid Jakob Jud necrology, which appeared in the tone-setting Neue Zürcher Zeitung ? The only other editorial features detectable are, in each instance, the scrupulous identification of the original locus of publication (along with courteous acknowledgements to author and publisher) and a single elaborate Index Nominum at the very end of Vol. I I (587-605, in double-column). 13 Steering clear of any educational stance, S. effectively displays his wares, so to speak, without trying to instill in his readers any idea of how to put to the best possible use the valuable merchandise he has assembled. By keeping aloof and refusing to show condescen sion, he thus leaves the burden of such suggestions to the sophisticated readers themselves, as well as to the critics. There are a few stray possibilities t h a t have occurred to one peruser. I t may be worth-while to analyze, in con figurational terms, the life-spans of leading students of language. There are those among them who lived — and enjoyed to the hilt — long lives ; where does the peak of their activities typically lie, in their 'twenties (as is rumored to be true, on the average, of inspired mathematicians) or in their 'fifties (as seems to apply to many ponderous historians) ? Perhaps the trajectories of linguistic theorists and of language historians clash in this respect. Among those scholars blessed with longevity — the late R. Menéndez Pidal, slighted in the anthology under review despite the excellence of his Orígenes, almost reached his one hundredth birthday — some remained buoyant and active to the very end (H. Schuchardt, say) ; others, including W. Meyer-Lübke, lost much of their initial élan; still others, quite exceptional, exhibit two summits, as it were, separated by a deep valley of visibly low productivity, though their thinking and their academic teaching may, of course, have constantly kept their minds in a state of unabated alertness, even of ecstasy (in this context, one is reminded of Saussure's "curve"). "Trubetzkoy war mit 15 Jahren ein reifer Forscher", Jakobson remarks (II, 529) ; true — but, clearly, he had not yet reached the pinnacle of his originality. Saussure's celebrated predoctoral Mémoire, on the other hand, stamps him as a precocious Originalgenie. Schuchardt, though steadily growing in diversity of interests, scope of commitments, and novelty of views — he was the experi menter par excellence in the Romance fold — never again achieved the power of synthesis he was privileged to show in his Vokalismus des Vulgärlateins, a doctoral dissertation immediately expanded into a three-volume magnum opus (1866-68), which thus became the fountain-head of all his subsequent inquiries, stretching over sixty more years. If the cases of gradual tapering-off of energy are numerous and those of loss of initial impetus, due to sickness, personal frustration, or academic disappointments, are by no means infrequent (here the name of Rasmus Rask comes to mind), long spells of total indifference to lingui stics, whatever the provocation, are quite exceptional, at least in this category of all-time greats; yet the concluding thirty years of S. Gyarmathi's long life (1751-1830), after the publication of his Affinitas (1799), were marked by an abrupt retreat from excellence, des pite the appearance of his Hungarian etymological dictionary at the mid-point of this dreary span (Vocabularium, Vienna, 1816). A. L. Kroeber's case history—pieced together with conspicuous meticulosity and verve by his devoted junior friend Dell Hymes (II, 400-436, esp. 403, 414, 425, 433f.) — is even more striking and particularly complex because there 13 This list records, in a single alphabetic roster, the names of biographers, those of biographees, and those — most numerous of all — of persons incidentally mentioned. One curious spacesaving device : Arabic numerals used for page numbers refer to Vol. I, italics to Vol. II. The value of this inventory, prepared by Cologne's Ursula Stephany, might have been enhanced if boldface (roman and italic) had served to designate the pages actually devoted to the necrologic essay in honor of the given individual. Thus, the names of F. Bopp, K. Brugmaan, and B. Delbrück emerge in strikingly numerous contexts, but each of these scholars can claim only one profile concentrating on his total œuvre. Between thorough treatment and casual mention one discerns, in fact, an intermediate category — the fairly detailed examination of an accessory personage, as when, in M. Zsirai's evocation of S. Gyarmathi (I, 58—70), P. Sajnovics also receives a liberal share of attention (59f.), as do F. F. Fortunatov and A. A. Šaxmatov — neither accorded the honor of separate treatment — i n Jakobson's many-splendored essay on Trubetzkoj. Perhaps a special font should have been reserved in the Index for such ''(semi)finalists" or "runners-up" who but narrowly missed attaining the highest distinction.

66


never occurred any slack or slump in the rhythm of his general activity, known for its sus tained effervescence, though a long stretch of relative indifference to language neatly separates his early period of Indianistic research, immediately after his transfer to Berkeley, from his return to linguistic inquisitiveness, in the wake of the glottochronological vogue. (Kroeber, in the course of a conversation shortly before his death, spontaneously confessed to me that, in his student days at Columbia, it was passionate concern with linguistics which initially sparked his abiding curiosity about anthropology, thus launching him on a scintillating career, cf. Hymes, 403f.) Edward Sapir's well-known cooling, after 1933, to ward certain genres of linguistic work which had previously for many years roused his mind to a feverish pitch of activity is ably restated by . Ғ. Voegelin (II, 489-492).

The obstacles that beset Kroeber's and Sapir's separate, if intersecting, paths dramatize another set of problems which S.'s anthology encourages one to confront : the complications arising from most linguists' split commitments to several disciplines that compete for their time and loyalty. (To simplify matters, one can disregard other conflicts and cleavages as between pure scholarship, on the one hand, and theological, lay educational, or administrative duties, on the other.) Typological break-down along this line bids fair to reveal other con formations, some of them foreseeably startling: Students of language who originally were Orientalists, classicists, teachers of modern languages, archeologists, anthropologists, logicians form different crews or clans, as it were, use diversified jargons, obey contrastable traffic rules for Academia, and believe in irreconcilably discrepant scales of values. Typically, each group so delimited also inherits a wealth of prejudices from its earlier affiliation ; it is thus, to cite just one example, impossible to trace the tragic decline of Romance linguistics in this country without taking into account the barely mentioned fact that the founders of the LSA were, for the most part, professors of Greek, Latin, and German, who had small reason to rejoice at the prospect of any flowering of Romance studies ; a stab in the back was, of course, concurrently dealt by those literary scholars in the Romance departments — they formed a solid majority — who saw little point in encouraging an emphasis rightly or wrongly associated with the predominance of German culture. To revert to the main thread of our discussion, S.'s material, again and again, incontrovertibly shows the longprevalent harmonious blend of philology and linguistics ; so viewed, the rebel lion of "pure linguists" against philologists thirty years ago (see fn. 2, above), amounts to hardly more than an adolescent caprice. One more side-light worth noting : An astonishing number of trail-blazing theorists and reconstructionists of language were, simultaneously and often from childhood, excellent, versatile ''practical linguists". Such naïve linguistic experience (sometimes nourished by a bilingual parental home, or by some such supervenient event as intercultural marriage, extended leisurely travel abroad, or enforced emigration to another country) thus turns out to provide a superb preparation for all manner of analytic dissection in the domain of language. Despite certain qualifications that a Romance scholar is apt to make in sum marizing his impressions of S.'s book,14 this elegantly printed, luxuriously bound 14 Thus, as regards the small reserve of available portraitists, it is strange that S. should have overlooked the special aptitude of the late B. Terracini, one of whose books — of all possible


67

set of two volumes is certainly a welcome arrival. It richly deserves to figure prominently on every list of assigned or recommended readings in advanced courses in the history of linguistics ; in fact, it should grace the shelves of every discriminating book collector — though, as an objet d'art, it would have gained immensely from the inclusion of several plates, a few of which might well have been polychromatic (after all, Sir Joshua Reynolds DID portray Sir William Jones, using a genuine rather than a metaphoric brush to suggest his subject's countenance ; see I, 16). I would even have recommended the book for the much humbler purpose of helping autodidacts among graduate students of linguistics to improve their skill in reading semitechnical literature in foreign languages were it not for the deplorable fact that the articles written in French are marred by quite a few, and those appearing in German by far too many, misleading typographic errors and infelicities.15 titles — emblazons on its front cover the words Perfiles de lingüistas (Tucumán, Arg., 1946), and one of whose better-known articles, bearing on his flamboyant compatriot M. Bartoli, enters into the series "'Ritratti critici di contemporanei". In addition to numerous shorter obituaries, a genre he cultivated assiduously, Terracini wrote rather ambitious biographic sketches of such Italian linguists as G. I. Ascoli, 8. Debenedetti (strictly speaking, a philologist, who has decisively influenced, among today's notables. G. Contini and C. Segre), . Giacomini, Pier Gabriele Goidànich (a particularly weighty article traceable to the year 1953), E. G. Parodi, and C. Salvioni, aside from his essay on J. Gilliéron; for full details see pp. xi-xx of his miscellany Pagine e appunti di linguistica storica (Firenze. 1957). But it is, of course. Terracini's paper "W.D Whitney  la lingüística general" in RFH, V (1943), 105-147, that, if only for the sake of its cultural piquancy, should not have eluded S.'s vigilance. Some of the most vibrant pages on Gilliéron have, incidentally, been written by John Orr ; why not introduce, for a change, a Tasmanian-Scotchman's view of a Swiss dialectologist transplanted to Paris ? As concerns the men of distinction potentially qualifying for portrayal, Spanish America, inexplicably, has been completely neglected. The Venezuelan-Chilean A. Bello, personifying as he did a temporal overlap (a first-rate 18th-century mind concealed in the body of a major 19thcentury figure), and the Colombian R. J. Cuervo domiciled in Paris do boast sufficient stature to have been included, and there exists, moreover, on both figures a profusion of readily accessible literature. As for the Spaniard-Argentinian A. Alonso who, in 1952, died under tragic circum stances "exiled" at Harvard, perhaps an imaginative combination of Ճ. Rosenblat's ten-page memoiristic essay ("A.A.", 1952), now included in his omnibus volume La primera, visiόn de America (1965), with excerpts from a distinctly younger D. Catalan's La escuela lingüística española (Madrid, 1955) would have provided the necessary polarization. These three leaders should not have been omitted. 15 Aside from abounding in crude misprints (seebelow),the texts — especially those offered in foreign languages — suffer from all manner of minor inadequacies, all of which, in the aggregate, are irritating and some of which might have been anticipated and therefore eliminated. The neophyte will find no key to the abbreviations of book titles and names of journals, not even those pertaining to Hungarian (I, 59-62). The editor has failed to observe such a minimum of typo graphic uniformity as the use of italics for book titles (I, 119), and in H. von Steinthal's paper, enlivened by numerous quotations from Humboldt, ss and β seem to be distributed quite chaoti cally. German expository prose has changed considerably over the last century; perhaps there was no harm in preserving, for the sake of their quaint archaic flavor, such obsolete forms as äussres, letztrem, sichres, tiefrer, but surely it would have been advantageous to modernize the spelling oí wargenommen (I, 104) and gewart (1,117), also, of ohngefähr 'ungefähr', where more than mere orthography is at issue (I, 212) ; alternatively, S. might have appended to his book a short glossary of obsolete German words. Factual contradictions should somewhere have been noted and, if at all possible, clarified ; thus, in their rival evocations of F. Bopp, R. Martineau asserts that his former Berlin teacher, as a student in 1815, attended Chézy's Sanskrit course at the Collège de France (I, 202), while just a few pages later A. Leskien flatly denies this teacher-pupil association (p. 209: "Jedenfalls hat B. ... nicht den Unterricht Chézy's genossen"). The para graphing was done with glaring carelessness in the older German originals, and some URGENTLY needed changes might very well have been introduced to help the modern reader (e.g., I, 208, line 6 ; also pp. 214 and 219). Perhaps an index of key-words would have been a boon (note the use of "innere Struktur" by F. Schlegel, in anticipation of Humboldt's far more famous "innere Sprachform", and of "Organismus der Sprache" by Bopp — not, as one would have expected, by Schleicher : I, 212). Finally, a short supplement might profitably have been reserved for a chrono-

68


III Against this backdrop let me examine in somewhat more searching detail four books contributed over the last few years by a quartet of Europeans : a worldly Belgian Indo-Europeanist, a cosmopolitan lady from Yugoslavia, a widely traveled Swede, and an Italian stationed in southern England. M. Leroy's book, which has become overnight something of a bestseller in the French academic milieu and is now available in English as well, deserves sus tained attention less on account of its intrinsic merit, which is limited, than by virtue of its symptomatic value. Sparklingly phrased, but only in part bewitchingly argued and, by and large, deficiently architectured, it exemplifies a no doubt sincere, but somewhat lame attempt of a classicist turned comparativist to come to terms with a movement of such overflowing vitality as is modern linguistics. This new discipline is, in its scope, general and in its texture pre dominantly structural and thus, on both counts, conflicts with the highly particular and (as a result of irretrievable losses) woefully fragmentary material traditionally assigned to philological linguists. Supervenient complications arise because L.'s survey purports to embrace periods which to qualified presentday observers appear etched out in reassuringly neat outline — e.g., the one hundred years stretching from Bopp's and Rask's pioneering experiments to Saussure's and Sapir's crowning achievements — alongside more recent decades which as yet afford no easy orientation even to a seasoned analyst and which, I regret to report, have left the author baffled and confused. Also, his enviable familiarity with some cultural settings (Paris and Geneva, say) and his com mensurate aloofness from other locales, at present not one whit less important, 16 logical t a b l e of crucially i m p o r t a n t publication d a t e s . T h e reader would t h u s , a t a glance, see w h a t t h e members of certain " p l é i a d e s " — say, Grimm, B o p p , a n d H u m b o l d t , or Meillet, Gramm o n t , Vendryes, a n d Bally — were each engaged in a t a given m o m e n t . I t would, of course, be p e d a n t i c for a reviewer t o supply a complete list of even t h e more serious misprints t h a t h a v e been allowed to creep in ; I therefore offer j u s t a few specimens, relating t o p e r h a p s as little as one t e n t h of t h e entire material, a n d present in each instance w h a t I t a k e t o be t h e correct, r a t h e r t h a n t h e incriminated, reading : Vol. I, Introduction: praebet (vii); to restore the dead to life and to grant them (xi); Očerki (xiv). Steinthal on Humboldt : nicht anders denn (103) ; hervorgehoben (119). Hjelmslev on Rask : a gravité (184), telles que les langues (190). Malone on Rask: . Nyrop (195). Leskien on Bopp: die gebildetsten (207), mit der indischen Literatur ... verschiedener (208), genannt (210), aus Partizipien die schon (211), verglei-chende (215), ein, wie sich immer zeigt, ... (218), Kirchhofe (220). Von der Gabelentz on Pott: seiner (252), dieses (257). Delbrück on Böhtlingk: Mitgliedern (263). — Vol. II, Vendryes on Meillet : professionnel (208), intérêt (210), F. de Saussure (215), Johannes Schmidt ... Benveniste, une seconde édition (221), dans la langue ...indo-européennes (224) ... fantaisie (225), rassurer (236), directement ... indo-européennes (238), intérêt (239). Gamillscheg on Vossler: denen (336). Benve niste on Vendryes : secrétariat général (391). Jakobson on Trubetzkoj : spannenden (531), sprachgeschichtliche (535), slavistischen (536), Gesetze ... unerwartete (539), Psych. (540), phil.֊hist. (541), etc. I reckon with t h e possibility t h a t a few of these Schönheitsfehler are traceable t o t h e original source t a p p e d ; b u t does n o t t h e m a i n a d v a n t a g e of resetting, as against p h o t o g r a p h i n g , consist precisely in t h e o p p o r t u n i t y it affords of eradicating r a t h e r t h a n p e r p e t u a t i n g and, worse, in creasing this undesirable residue of mechanical slips ? 16 Aside from a pointedly reserved a t t i t u d e t o w a r d G e r m a n a n d a t h i n l y m a s k e d a n t a g o n i s m t o w a r d American scholarship (of which more anon), L.'s book a b o u n d s in other manifestations of s y m p a t h i e s a n d a n t i p a t h i e s . I t a l i a n learnèd concern w i t h language, from D a n t e , Calepino, Sassetti, a n d Vico all t h e w a y to Bartoli a n d Devoto, receives considerable if desultory a t t e n t i o n , while Spanish research — i.e., studies conducted b y or bearing on hispanophones — is consistently ignored. R e s u l t : T h e transition from A n t i q u i t y t o Middle Ages (Isidore of Seville), t h e flowering of two Hispano-Oriental traditions of n o r m a t i v e g r a m m a r , N e b r i x a ' s pioneering rôle, a h e a d of other E u r o p e a n s , in vernacular probings, t h e discovery a n d early exploration of " e x o t i c " (New World) tongues, t h e full s t a t u r e of H u m b o l d t as a s t u d e n t a n d a n a l y s t of I b e r i a n a n d B a s q u e , Cuervo's visionary idea of t h e overseas g r o w t h of a E u r o p e a n language, Menéndez P i d a l ' s u n i q u e l y successful a t t e m p t t o reconcile dialect geography w i t h philologically u n d e r p i n n e d historical


69

bave caused a certain unevenness in the placing of emphasis, the distribution of shares of authorial attention, and the degrees of intensity of emotional response. Finally, without disputing L.'s right to a reasonable measure of selectivity, particularly in a work of baute vulgarisation, I cannot help feeling that he has made entirely too liberal allowances for personal likes and dislikes, along both doctrinaire and national lines. In certain sections one further detects disquieting traces of haste and laxity, as regards topical coverage and critical verdicts, glossed over, I repeat, by an invariably polished, not seldom scintillating style 17 and by an almost impeccable typographic presentation.18 A perceptible lack of firmness in the theoretical foundations of L.'s book greatly debilitates its entire framework and dissolves the long concluding p a r t (75-173) into such a loose string of witty reactions t h a t the venture's prime raison ďetre seems to be L.'s intention to absorb into one book a profusion of elegantly worded earlier articles on linguistic themes (ix), keyed to current lines of lay curiosity. Only the Introduction (1-14) and Section I (15-60), which jointly take the reader from Pānini past Jones to the threshold of Saussure's Cours, follow the conventional, roughly chronological chain of events and, as a result, fall into some sort of recognizable and defensible pattern. So far as a relative outsider can judge, the Introduction is an attractive piece of writing, free from imbalance and partisanship. The chapter on the Greeks, with its contrastive grammar (1926), to say nothing of R. Lenz's and A. Alonso's intriguingly conflicting views of sub stratum influence in colonized territories, are all swept under the rug. Portugal, despite the excit ing linguistic implications of its contacts with Africa and Asia, shares the fate of Spain. Eastern European scholarship elicits a few remarks,and Hrozný, if not Mathesius (who laid the foundation for the Prague School), is incidentally mentioned (55) ; but the Russian antecedents of phonemics are sketched in far too hazily (80f.), Kuryłowicz is referred to only once, en passant (142), the entire cultural context which fostered an unmistakably original flowering of linguistics in that part of the world — or in neighboring Finland and Hungary, for that matter — is left out of the account, and only the characteristic Marr aberration is accorded a couple of solid pages (145f.) — in tribute to its picturesqueness and absurdity ? 17 It is sheer delight to evoke the gourmet dinner that L. serves his spoiled readers. The narra tive or expository prose is spiced with an exquisite assortment of anecdotes (21, 33, 80, 90, etc. ; the author turns out to be a tastefully restrained raconteur), salted with predictably entertaining examples (57), and sparingly seasoned with choice references to charismatic literary figures (23). Subtly interspersed remarks on suspenseful elements of human drama (book publications critically delayed, discoveries long left unheeded, credit for accomplishments assigned to the wrong persons) provide another bonbonnière of tidbits and morsels, whetting and promptly satisfying the appe tite. Any threat of pedantry and monotonous consistency is kept under tight control : If certain key books are identified with a prolixity of bibliographic detail, others, for no apparent reason, are barely hinted at, and biographic vignettes are doled out or withheld with the same charming nonchalance. Thus, from the ten pages devoted to Schleicher (22-32) the reader will learn neither the topics, let alone accurate titles, of that scholar's major writings, nor the dates of his birth and death — in his case particularly important for those tempted to place him beside Darwin. The reader visualized by L. is sophisticated so far as his humanistic schooling is concerned, to the extent of being receptive to Greek words and titles printed in the original script (passim) ; he need not be told the meaning of technical terms like scholiast (4) and visibly requires no intro duction to Aulus Gellus (9), Scaliger (11), or Hegel (27). Curiously enough, that same reader must be forewarned not to confuse pristine language with the babbling of infants or with modern speech used in primitive cultures (34f.) ; not to share Rousseau's, or Renan's, illusions (35) ; not to fall into the trap of racists (36-38) ; and to be diffident of such labels as "difficult" language and "ugly" sound (57). The presentation is enjoyably lucid, though it will take the uninitiated a certain effort to keep apart (48-51) Bréal's trail-blazing article (1883) "Les lois intellectuelles du langage" and his celebrated book (1897) Essai de sémantique. Exotic languages, cautiously cited, are seen through the prism of other scholars' inquiries (28f. : Chinese; 3 1 : Bantu; 84: Lower Cherkessan; etc.). The network of cross-references is superb. Judiciously enough, L. avoids any technical terms that might prove "rébarbatifs", the mention of "muscles thyro-aryténoïdiens" for 'vocal chords' (83) being an isolated exception. The guidance of the reader along the crazily winding path is, on the whole, firm and careful, though "expressivity" and "functionalism" may cause raised eyebrows at first mention (79) and though at least one gratuitously narrow-gauged excursus into paleo-I.-E. (44) is likely to prove unassimilable for the tyro. 18 A few self-explanatory slips are found on pp. 79 (fn. 3), 83, 105, 119.

70


treatment of Plato and Aristotle (4-6), benefits from L.'s classical expertise; later one sees, in quick succession, a parade of engaging cameos of A. Calepino, . Gesner, J . Megiser, G. Postel, J . J . Scaliger, and G. Vico (9-13). Hindu grammarians, medieval theorists, Dante, Leibniz, Arnauld (for whom L. cares less than he does for Vaugelas), and the encyclopedists receive a smaller share of attention than they deserve ; the Tower of Babel myth is mentioned cursorily (8) without reference to a recent exhaustive monograph devoted to its reper cussions, and the Bible figures as a linguistic stimulant solely apropos of translations into certain vernaculars, without full appreciation of its rôle in the history of traditional exegesis, textual criticism, and vocalic-musical notation (to say nothing of the silence surrounding Hebrew and Arabic grammarians). P a r t I is subtitled "La formation de la méthode linguistique" (15-60) ; it encompasses the 19th century plus t h a t slice of the 20th which preceded the First World War and, conse quently, the appearance of Saussure's Cours. At every step, by virtue of its scope, it invites comparison — more than do other sections — with rival treatments. L.'s account of this period is structured thus : After a quick glance at the West's early sporadic exposure to Sanskrit (Sassetti, Cceurdoux, Jones) and after incidental mention of P. Schlegel's pioneer ing use of the key term "vergleichende Grammatik", he establishes a small nucleus of founders (18-22) — with separate niches reserved for F . Bopp, R. Rask, J. Grimm (whom he discusses far more cursorily than did J . Waterman), and F . P o t t (whom he treats less cava lierly than he does fellow-etymologists) —, then devotes a consolidated ten-page essay to A. Schleicher, with an occasional side-glance at M. Müller. At this point the temporal progression yields the right of way to a flashback (32-35) highlighting various theories (from Herodotus' to Renan's) on the birth of language. The main thread of the narrative is again broken by the necessity to do justice to Humboldt's œuvre (35-39) which, in its technique and philosophic implications, runs parallel to the endeavors of 19th-c. IndoEuropeanists rather than constituting an isolable link in their continuous chain or than intersecting that chain. The inherent difficulty is aggravated through L.'s failure to buttress Humboldt's unique position by either placing his works in the matrix of contemporary German culture or at least connecting them with Central Europe's long-influential psy chologizing school, down to W u n d t ; Whitney alone, despite the geographic distance, is attached to him. The Neogrammarians now occupy the center of the stage (40ff.) ; Brugmann, Delbrück, Osthoff (but neither Leskien nor the somewhat younger Streitberg), on the one hand, and Meillet beside Verner, on the other, elicit remarks of varying length. Unfortunately, no clear-cut, if miniaturized, pictures of the German, French, Scandinavian, East European, or North American prongs of the movement emerge at any point. Meillet's unifying and synthesizing rôle dwindles through an almost reckless scattering of references to him from cover to cover ; the "Société de Linguistique de Paris" — his brainchild and the prototype of the various 20th-c. "circles" — acquires no sharp contour, and the ties binding the master to the far-off Slavic phonologists and such near-by disciples as Benveniste and Martinet remain unidentified. I n the context of Neogrammarian stirrings the young Saus sure, on the strength of his Mémoire, receives from L. his first accolade (42-45) ; but Schuchardt as the brilliant, undaunted critic and opponent of the school almost vanishes from sight. The "fin de siècle", L. argues (47), witnessed the fermentation of new ideas, which before long were to cause an upheaval. Among these cross-currents he singles out : (a) pho netics (but why omit Viëtor and Lenz, and cite instead M. Grammont and A. Grégoire, whose rise to fame falls into a distinctly later period?) ; (b) M. Bréal's rather than H. Paul's semantics ; (c) J. Schmidt's wave theory ; (d) J. Gilliéron's dialectology. The inclusion of the fifth subsection, gauchely subtitled "Les langues centum et šatem" (54f.), makes sense only if the emphasis is placed on the refusal of two newly discovered L-  . languages — Hittite and Tocharian — to fall into place geographically. The concluding vignette (56-60) of O. Jespersen as the author of Progress in Language (1894) stands apart and serves no visible purpose at this juncture. L. has reserved a separate, conspicuously slim section of his book for Saussure (61-74), as he has for no other individual. Few readers will dispute Saussure's intrinsic radiancy and powerful influence ; even so, the wisdom of any kind of academic hero worship is debatable.


71

One concomitant t h a t strengthens one's resistance to L.'s schema is the fact t h a t the section deals not even with the favored scholar's total œuvre, but with the Cours alone — a writing of slightly apocryphal status to boot. The Mémoire of 1878, we recall, receives L.'s rhapsodic attention at an earlier point; the recent exegesis of the Cours (78-80), particularly as practiced by R. Godei and R. Engler (the "third generation" of Genevans), falls into its slot within Section I I I ("Twentieth-Century Linguistics"). W h a t remains here as the irreducible kernel of the all too slender central pillar is the gently critical analysis of the Cours itself, or rather of its best-remembered leitmotive (such as the arbitrariness of the sign, the linearity of the signifier, the polarity of "langue" and "parole", the antinomy syn chronism vs. diachronism), a situation which practically reduces this chapter to a belated critique. True, L. enlivens his account b y occasionally exhuming a Saussurian metaphor long buried in the original lecture-notes, say, the "carré linguistique" (72) ; he even guides the reader to H. Erei's recent elaboration on t h a t concept, namely the "carré sémantique" (73), while the comments hidden away in A. Alonso's translation remain unsalvaged. But though L. does not hold back, here or elsewhere, with a few mild reservations concerning the validity of his spiritual guide's doctrine, he neglects to satisfy his readers' legitimate curiosity as to just what the Swiss scholar, placed on such a high pedestal, was doing between 1878 and ca. 1905. There is no hint either of the enthroned authority's intermediate period of writing or of his extensive readings, established or presumptive. Characteristically, though the sociologist É. Durkheim is mentioned repeatedly at strategic stations (90, 119f.), the neophyte has no way of inferring his immense influence on Saussure's thinking — except obliquely, through the descriptive title of W. Doroszewski's 1933 article cited in a footnote (120). The lengthy concluding part (75-173), on 20th-c. linguistics, is so loosely organized t h a t one is, I repeat, hard put to avoid regarding it as a bagful of aperçus and essays almost haphazardly thrown together; this shortcoming may be due to L.'s inability to weld his earlier exploratory articles into a single coherently architectured whole. One readily acknowledges the propriety of starting with the Geneva School (78-80), branching out into Prague-style phonology (80-89), then expatiating on structuralism (90-106, with separate bows in the directions of Brøndal, Hjelmslev, Harris, Chomsky, L. Bloomfield, in this idiosyncratic sequence). But once this advanced command post has been reached, it is slightly anticlimactic to be invited to revert to such Saussurian problems as "l'arbitraire du signe, langue et parole, synchronie et diachronie", even if at this stage the distant reverbera tions (Benveniste, Coşeriu, Wartburg), rather than the mentor's original doctrine, are a t issue (106-114). Also it is a matter of mounting surprise to see attention next deflected toward the Sociological School (115-123), with J.-J. Rousseau, É. Durkheim, M. Cohen, G. Dumézil (a spokesman for comparative mythology), and A. Sommerfelt, to name b u t a few in the motley crowd, hurriedly parading before the reader, privileged to catch, with a measure of luck, the names of Bloomfield and Sapir, but doomed to remain blissfully un aware of socio- and ethno-linguistics in their new garb. The farther we wend our way, the more insistently the clock is being turned back, because now an incommensurately long chapter (123-138) introduces us to Croce, Vossler, Bertoni, Bartoli as so many champions of "thèses individualistes" (here "neolinguistics" is showered with sympathetic attention). Veering abruptly from an ideological orientation to the array of specific facts, L. next con fronts his reader ("La dialectologie indo-européenne", 138-143), with such diverse per sonalities as, on the one hand, Bartoli, Bertoldi, Bonfante, Devoto, Pisani, and Terracini, and on the other, Meillet, Benveniste, and Kurylowicz, all of whom he patently likes, reserving two deftly aimed slaps for Brugmann and Sturtevant (142), whom he clearly detests. W h a t holds together these meandering pages is some such theme as "New Ideas on Indo-European" rather than any sustained concern with dialectology. If national teams and "schools" have so far surreptitiously dominated the scene, their hegemony becomes overt in the discussion of "Mechanism and Mentalism", i.e., North America [over a quartercentury] vs. Europe (143-145), and reaches its peak in the section devoted to "Soviet Linguistics" (145-148; the inclusion of M. Cohen should һve counseled L. to adopt some such alternative, less strident, title as "Marrian and Marxist Linguistics"). The Russian

72


landscape evoked is far removed from Moscow's present-day involvements. The last major subsection — and easily the one best harmonized and most pleasingly phrased — bears on linguistic typology (149-163), i.e., on the typology of languages, not of genres of linguistic inquiry, which, though concealed from L.'s readers, has meanwhile, on its own, ripened into a legitimate avenue of approach. Here the views of the principal mouthpieces : F . N. Einck, J. Lohmann, E. Lewy, E. Sapir, É. Benveniste, receive, as they should, benevolent attention, while the nuanced opinions of Schuchardt, Meillet, and Jakobson provide delicate orchestra tion, and so much as any hint of bias or irritation is absent or repressed. The remarks on stylistics, so meager as to occupy a single page (163f.), are somehow left dangling, as were before the woefully threadbare comments on the psychological slant (114f.) and on phonetics (148f.)—presumably the three most vulnerable spots in the entire survey, unless one wishes to add to the number of near-failures a few noncommittal musings on the statistical method, toward the very end (172f.). The subchapter on semantics (164-172), which acquaints the reader with Kronasser, Ullmann, Trier, Guiraud, and Matoré, to the exclusion of the latest vicissitudes in the English-speaking countries, abounds — as one would expect — in warm overtones of approval, but, wedged in as it is between typology and statistics, hardly lends itself to any meaningful integration.

From beginning to end, L.'s book is permeated by an undercurrent of fervid attachment to the ideal of seeing linguistics admitted to full membership in the alliance of "sciences humaines" (or "sciences de l'homme"). Regrettably, this key concept is at no juncture neatly defined, still less delimited against its traditional counterpart, the "sciences sociales" ; neither is it anywhere cali brated with sufficient precision to invite comparison with Germany's nebulous Geisteswissenschaften or the English-speaking domain's excruciatingly hazy humanities, to say nothing of liberal arts — by no means synonymous tags and catch words. True, at intervals L. discloses certain characteristics of this kind of learning19 (which in English usage happens to be excluded from association

19 L. sets the tone and the standard in this preliminary remark: "Dans les sciences humaines, auxquelles appartient la linguistique, nous sommes sur le terrain du continu et toute compartimentation stricte est vaine" [except for narrowly didactic purposes] (viii, cf. 78). He is patently on the defensive thirty pages later : C'est le moment d'affirmer une fois de plus que la linguistique est science de l'homme et non de la nature et que les sciences humaines ne se laissent pas ramener à ces schemes rigoureusement compartimentés, à ces tableaux à accolades... (28) Could it not be argued that, just as a tree, though plainly visible above ground, has its subter ranean roots or just as only a small portion of an iceberg emerges above the surface of the sea, so linguistics, according to the particular subdiscipline envisaged, could with equal justification be assigned to humanities or to sciences and that, in its total expanse, it may very well pertain to both realms? L.'s polemic continues under two guises : Through sharp repudiation of predecessors and fellow scholars who, in his estimate, have adopted a platform perilously close to that of straight scientists (Schleicher: 28, the Neogrammarians: passim, and many moderns; see esp. 97-99), and through subtle endorsement — via inexplicit self-identification with the experts summoned to the witness stand, almost in the celebrated tradition of the "style indirect libre" — of periodic rebellions against excessive rigor or dogmatism. The working of this latter technique can be best observed under M. Bréal (50) and especially under M. Bartoli, whom L., as if to com plicate matters, elects to see through the unavoidably distorting prism of a follower (130f.). Further revealing of L.'s inner sanctum are his statements on the anti-aprioristic character of the historico-comparative method (14), his distrust of typological classification (29-31), his inability to tear himself loose from the obsession with scientific checking: "hypothèses pas toutes invrai semblables, mais indémontrables" (22), "ensemble de conjectures invérifiables" (34), "affirma tions incontrôlables" (35). More promising are L.'s casual remarks on an ingeniously conducted interplay of "idées directrices" and "observation des faits" (106) — ideas which might have yielded the cornerstone for the entire edifice, even though they hardly ulumine the alleged dif ference between "sciences de l'homme" and science in general. Why must the selection and accu rate assessment of criteria (93, 119f.) rank as a distinctive feature of humanities?


73

with any league of sciences); but a thoroughly reliable frame of reference remains consistently unavailable. This flaw is probably the worst strain on L.'s grand strategy, as distinct from his many tactical errors. His heart — the heart of a classicist steeped in fine literature — beats strongest when it comes to championing the embattled position of Croce, whom he hails as the advocate of the supremacy of mind over matter, of the gifted individual over shapeless society, of the esthetic-expressive over the jejunely communicative function of speech (123-126). This enthusiasm is palpably rooted in a philosophic pre disposition, not in a linguist's genuine laboratory experience; once aroused, it prompts L. to accord an astonishingly high rating (127f.) to Vossler (despite the fatal damage done to that Romanist's prestige by the ill-advised post humous publication of certain drafts, a matter L. leaves unmentioned) ; to lump together a fiercely independent critic-philologist  Spitzer (trained by MeyerLübke, not to forget Becker, and inspired by Schuchardt) with Vossler's authentic students and orthodox followers (128, 164) ; to take a charitable view of that once sober philologist G. Bertoni's gratuitous feuillet ones que vagaries of linguistic theory (129) ; and to compound the difficulty by soldering too tightly Bartoli's "neolinguistica" — an ephemeral fad if there ever was one — to Crocean idealism, instead of finking it — as simultaneous, parallel manifesta tions of the same gropings — to the hard-boiled sociologists' Age and Area Hypothesis (129-132). L. seems to be untroubled by the fact that these "thèses individualistes", for all their conceivably flattering appeal to the philosophers in us, have led linguists astray as regards zestful, vigorous, and imaginative research, except for one narrow by-path : stylistics. Paradoxically, if one dis regards L.'s professions of faith, he steers, as a practicing linguist, a course much closer to the position of Saussure, whose superb edifice, we recall, he rightly ad mires. For a while he even cheerfully joins Meillet and that scholar's associates in applauding the "Sociological School", which operates, by definition, with society-centered cultures rather than with the comportment of lone individuals (115-122). In the end he grows aware of the need for making a choice between loyalty to humanities and fealty to social sciences; but, for all the dramatic flavor of their delayed confrontation (131-138, the book's climax), he offers no escape from this dilemma other than a reference (132) to Devoto's and Nencioni's no longer new concept of INSTITUTION "qui rend compte du double aspect, social et individuel, des faits de langage". Could not a spell of immersion in Sapir's writings have supplied him with the needed pass-word? L. does not mince matters in condemning, or brushing aside as irreconcilable with his cherished "sciences humaines", such brands of linguistics as have operated with biological models, mechanistic and behavioral presuppositions, and patterns of retrenchment and schematization admissible in logico-mathematical disciplines. Striking this antagonistic pose is indisputably his privilege. What he and other polemists — including many who come from the opposite camp — never make clear is this : Just how much of a distinguished worker's effort must be written off as a waste if he is suspected, accused, or convicted of having been a heresiarch? Assuming L. Bloomfield's theoretical basis and personal style irritate a critic, should this disagreement, in the critic's opinion, deprive such monographs

74


as Eastern Ojibwa (1957) and The Menomini Language (1962) of intrinsic merit on all levels ? If not, would it be safe to argue that the landscape of hard lingui stic facts and connections remains fundamentally unchanged with the advent to influence of new theories and terminologies, or at least that its contour changes no more radically than does the delineation of any real-life landscape under varying conditions of lighting — natural and artificial ? Milka Ivić's book, whose Serbo-Croatian original appeared in 1963 (thus making it strictly contemporary with L.'s venture), suffers from a less than distinguished, in part downright pedestrian, translation. 20 Its chief merits are the elaborate apparatus of up-to-date bibliographic hints and the author's honest, by and large successful effort to strike a balance between the accom plishments, old and new, of East European, West European, and North Ameri can scholarship, no easy undertaking amid today's chaos. M.I.'s tact and kind ness transpire in the restraint with which she treats such authors, books, and movements as are clearly not to her liking. On the negative side of the ledger, at least for readers spoiled by L.'s light-winged approach, are the frankly didactic slant of her book — a veritable honeycomb of 480 narrow-meshed paragraphs — and a slightly candid pervasive optimism, driving her to espouse the cause of modernity, without awareness of certain losses which this choice,  any other, inescapably entails (see esp. 69-77), and bidding her applaud with equal fervor not a few mutually incompatible trends, provided they have been made safe by some such sanitary label as ''structural", "progressive", or "modern". The architecture of M.I.'s book, elegant in its basic design if not in details, has vastly simplified her task of doing justice to various currents and cross-currents of 20th-c. research, though even her elastic approach fails to overcome the intrinsic difficulty of disentangling the knot of (a) topical emphases (say, dialectology or stylistics) ; (b) methodological pre ferences (e.g., structuralism, distributionalism) ; and (c) national traditions ordinarily congealed, not to say rationalized, into schools of thought. As a result, her pattern of slicing the material, though superior to L.'s, is also less than satisfactory. Two extra-long chapters, the 14th and the 15th, each requiring two tiers of subdivisions, jointly occupy almost as much space as sixteen other chapters lumped together. The 14th, devoted to the non structural strain (78-112), encompasses linguistic geography, the "French School" (with such further ramifications as psychological, sociological, and stylistic inquiries), esthetic idealism — which, of course, might have been alternatively ticketed "Italo-German School" — including Neolinguistics (despite the fact t h a t this doctrine, in the last analysis, is much closer to the glottogeographic than to the glottoesthetic streak and should have been correspondingly classed), the Progressive Slavist Schools (a fanciful common denominator for the Kazan' Group, Fortunatov's Moscow School, and Yugoslavia's chameleonic A. Belić), the Marrist interlude, and, in their wake, unaccountably, experimental phonetics. The outsized Chap, x v (113-185), designed to cover cis- and trans-Atlantic structuralism, starts with Saussure (see below) and the Geneva School, reaches its summit in the pages devoted to "The Phonological Epoch" from Trubetzkoj's forerunners to Jakobson and M a r t i n e t , then, incongruously, sets up as a unit "The Schools of [North] American Lingui20 A few random examples: "German" has been substituted for "Germanic" in a particularly unsuitable context (41.9); "edition", in lieu of "editorship", borders on the unintelligible (43.1). One section is marred by a combination of misprints (114 : r. inadmissible), slips (121 : r. Travaux du Cercle...), and improper idiom (ibid.: for "actually" r. "at present, currently"). Certain statements require editing to become understandable, e.g. : "Individual facts should not be taken in isolation, but always [as parts ofJ a whole" (114).


75

sties" (152-172) — granting in its scaffold a status of autonomy to psycho-, but neither to ethno- nor to socio-linguistics —, and leads the exhausted reader to Brøndal's and Hjelm slev's Danish insularity. Judiciously set off from the treatment of structuralism are three shorter chapters pointing, one gathers, toward the future: "Logical Symbolism in Lingui stics" (186-202), "Linguistic Syntax and the Generative Approach" (203-211), and, distinctly more elaborate, "Mathematical Linguistics", including separate sections on statistical linguistics, information theory, and machine translation (212-242). The chapters show a consistently high standard of bibliographic underpinning, b u t vary greatly in degree of sophistication.

Noteworthy is the treatment which M.I. has reserved for Saussure. A fairly short section of Chap, xv offers a single, consolidated analysis of Saussure's Mémoire and Cours (122-124), with practically no reference to the "opera minora" separating these two peaks of his production, but this time with heavy stress on his indebtedness not only to Durkheim, but also — a claim made at the cost of some exaggeration? — t o the Polish pioneers of the "Kazan' School", J. Baudouin de Courtenay and Mikołaj Kruszewski (97-100). The discussion of the Cours is succinct and to the point, allowing for the composite manuscript's distortion at the hands of Saussure's disciples, centering around the famous, immensely suggestive simile of the chess-game, and parsimoniously summarizing all of the book's fundamental ideas and classic antinomies. The analysis spills over into the following short chapter ("The Geneva School", 129-131), less convincing because here Bally's separate rôles as stylistician and general gram marian are not individuated with sufficient rigor ; also, the fact that a few of Saussure's seminal ideas, still in their state of incubation, percolated through Bally's and Sechehaye's early writings (1905-09, 1908), years before the delayed publication of the Cours, is nowhere clearly stated. Regrettably, the introduc tory remarks on structuralism which usher in the portrayal of Saussure are replete with inaccuracies regarding the past, with misleading clues to the present, and with presumably erroneous prognoses.21 M.I.'s book is not devoid of originality : She allows for a greater independence of Max Müller (46) than does L., who sees in him merely a deft popularizer; neither chronicler deems Fisk at all worthy of mention, clashing on this score with Waterman. M.I. surpasses L. b y a wide margin in her lucid, unpretentious treatment of H . Steinthal's psychologism (52f.) — with further threads leading from here to W. Wundt, G. von der Gabelentz, and A. Marty (53-57) —, as she does in her dissection of Neogrammarians (58-64) and in her brief evocation of Schuchardt (65f.). 21 Thus, the assertion that "structuralism in linguistics means, first of all, a new approach to facts already known" (113) hardly applies to the study of exotic languages and otherwise bares the movement's most vulnerable flank. The wholesale exclusion of mentalistic criteria from all manner of structural analysis is no longer tenable (114). It is incorrect to affirm, in a flush of un warranted optimism, that the initial diversity of attitudes in structural linguistics has been gradually overcome (ibid.). Bloomfield publicly admitted his early debt to Saussure (114f.). Yale University has nothing to do with the FOUNDING of structuralism, though it has played a major rôle in its soaring to prominence, then in its approaching decline ; significantly, Sapir and Bloomfield both taught at Chicago during the "movement's" critical formative years (116). Not all influential descriptive linguists accept the principle of binarity (118). The exultantly drawn landscape of a paradise on earth achieved with the advent to power of a certain type of struc turalism will, at best, elicit faint smiles (119). In §245 H. Hendriksen seems to receive greater credit than J. Kurylowicz ; §§264f. lay at the doors of Bally two mutually exclusive pronounce ments.

76


From its straight linear progression one would gather that Bertil Mahnberg's book exhibits the most tightly fitted structure of the four, but such is not neces sarily the case. The opening chapters, on the historical and comparative method (5-33) and on Saussure and the roots of structuralism (34-53), are distinguished and commendably even.22 Understandably, M.'s sustained concern with Scandinavian scholarship is in evidence throughout ; not only do the researches of Rask (7), Verner (9), Pedersen (11), Collinder (17), Brøndal, and Sommerfelt (19), to cite a few familiar names, receive loving attention, b u t some less well known writings, such as Axel Kock's by-and-large forgotten treatise on linguistic change (14), figure prominently in the discussion. The author's moderation shows to good advantage in those passages where the thread of his narrative is inextricably enmeshed with theoretical considerations and assessments; witness the sober position he takes toward such giddying issues as regularity of sound change (14), critique of the Neogrammatical doctrine (15), Jespersen's evolutionary optimism instilled into his concepts of progress and efficiency (16), internal reconstruction (17f.), assumption of the agency of suband super-strata (18-20). I find M. refreshingly flexible and persuasive in drawing a line between linguistics and philology (1-3 ; note the pithy remark : "The same researcher finds himself now a philologist, now a linguist"), and in discriminating between the strictly comparative and the historical approach (7) ; he is endowed with a special knack for con trolled simplification (8 : Germanic sound shift) and is as free from irksome political bias or prejudice as one can aspire to be. I n his Preface he insists t h a t the authors and titles cited involve a selection on the strength of typicality rather than of superior achievement and candidly admits a possible margin of arbitrariness. One stumbles over few downright infelicities. Not everybody, especially not among the youngest enthusiasts, will accept 1800 as the cut-off point for the genesis of linguistics in the "modern sense" (5). Disappointingly weak, for a change, is the discussion of the reputed centum/satem bifurcation (10f.), inasmuch as M. forgets to prepare the reader for the injection of the territorially perplexing Hittite and T o c h a r i a n evidence, this despite the fact t h a t in the preceding lavish inventory of L -  . languages he has reserved niches even for Kurdish, Kazjubian, Polabian, and Farnese. M.'s bifocal chapter on "Ferdinand de Saussure and the Geneva School ; Structuralism" — the second division of his book — is particularly solid (a model of its kind) not least because the author has treated its two nuclei on earlier occasions 23 and is thus treading thoroughly familiar ground. A cosmopolitan, he is discreet enough to list among the precursors of structuralism not only two neglected Swedes (A. Noreen and C. Svedelius), but German, Polish, American, Swiss, and Dutch trail-blazers as well. All of Saussure's work is jointly examined, and sensationalism is scrupulously avoided in reference to R. Godel's recent dis coveries (37), which were not meant to involve an exposé. Particularly unorthodox is the idea of examining the Mémoire after acquainting the reader with some basic tenets of the Cours (40f.) ; one benefit of this counterchronologic advance is the proof that, with respect to the delimitation of the phoneme, Saussure may have been more daring in 1879 than in 1908. The pioneer's own ideas on the separation of the two time axes, and the objections of his critics, are presented with pleasing impartiality (37-39). The "langue" : "parole" dichotomy receives its quota of attention ; Saussure's own concept of the "signe linguistique", with its margin of inconsistency, elicits shrewd observations, particularly incisive and astute anent the controversial "arbitrariness of.the sign" (42f. ; stimulated by D. Wester22 The book has been translated — on the whole, with the requisite taste and competence — from the 2d ed. (1962) of the Swedish original, with stray additions on such topics as glottochronology and generative grammar. In matters of doctrine M. is a middle-of-the-road man, and his avowed modernism is tempered by his earlier acquaintance with a conservative shade of Romance philology — an acquaintance which ended on a strident note (cf. G. Tilander's and Karin Ringenson's scathing reviews of his thesis). Despite subsequent reorientations and voltefaces his interest in Romance material has remained unabated (9, llf., 15f., etc.). 23 Cf. his book Structural Linguistics and Human Communication (1963) and the article "Fer dinand de Saussure et la phonétique moderne", CFS, XII (1954), 9-28.


77

mann, M. reckons with a heavy dosage of onomatopoeic formations in pristine languages). 24 One also runs across provocative remarks on such Saussurean key notions as "rapports syntagmatiques ou associatifs" [i.e., paradigmatic] and on "différence" vs. "opposition" (45). M. correctly traces the semiotic-sociological nucleus of Saussure's thinking (and some ingredients of Vendryes' as well) to the impact of L. Lévy-Bruhl and of Durkheim and assigns places at the periphery of the Geneva School to Meillet, Weisgerber, Glinz, Gardiner, Firth, Tesnière, Guillaume, and yet other minor luminaries (46-51). He joins L. Heilmann and C. Lévy-Strauss in embedding linguistic structuralism in broader currents of modern thought (54).

Of the remaining nine chapters some are excellent, others adequate, still others barely acceptable (the short piece on "Neolinguistics, the Vossler School, the Spanish School" unmistakably marks the nadir point, 69-73) ; but even if one disregards such mistrials due to the absence of so much as minimal rapport between the judge and those arraigned, it still remains true that themes ap propriately selected and treated are but seldom adroitly interwoven. Thus, the assured handling of Saussure's theoretical edifice clamors for smooth connection with another finely chiseled chapter, "Phonology and the Prague School" (7497) ; instead, the reader, barely familiarized with the Cours, is shoved into a sidealley—to use the author's, not the reviewer's, implied yardstick—namely "Dialect Study and Linguistic Geography" (54-68),25 and is then left adrift in the rough sea of Neolinguistics, where the author's piloting skill is at its weakest, before he regains access to traffic along the main avenue ("Phonology ..."). The treatment of structural phonology presupposes at every step elementary knowledge of old-style phonetics, and since M.'s New Trends, unlike (say) Bloch and Trager's Outline (1943), represents no introductory textbook, the omission of articulatory phonetics seems unobjectionable. Equally com mendable is M.'s decision to offer his readers a rather full picture of modern, esp. instrumental, phonetics, a domain in which he enjoys an enviable reputation of expertise (98-122 ; yet cf. Ilse Lehiste, UAL, XXXI [1965], 107f.). But wedged in as they are between Trubetzkojan structuralism and separate reports on Semantics (123-139) and Glossematics (140-157), the twenty-five chock-full 24 M. sides with those who contend "that there is an outer compulsion (social pressure) to associate a certain signifier with a certain signified, but no inner (naturally inherent) compul sion" (43). His final verdict declares the sign a "combination of a segment of the amorphous mass of sound with a segment of the amorphous mass of thought and concept" (44). 25 The hermetic isolation of the two chapters "Dialect Study and Linguistic Geography" and "Modern Phonetics" in M.'s book is best illustrated with the total lack of cross-references to the other chapters, most of which are, quite properly, interconnected with a network of such refer ences. Yet a modicum of good will, imagination, and flair for dramatizing events would have enabled the author to contrast the theoretical residue of linguistic geography and the structuralist doctrine as two assaults, from different vantage points, on the citadel of Neogrammarian as sumptions and practices. Aside from the attractive maps, the only redeeming feature of M.'s flat treatment of linguistic geography is the interweaving of Scandinavian evidence (Natan Lindqvist, Gösta Sjöstedt) with the familiar findings of Germaniste (from Wrede to Frings), Erench scholars (Gilliéron, Dauzat), and Hispanists (Menéndez Pidal); nearly all of this could have been written, far more zestfully, forty years ago. The author's studied restraint shows in his refusal to give credit to masterly studies by such fellow Swedes as K. Ringenson and B. Hasselrot who, in the company of other scholars, instead of turning apostates, have brilliantly bridged the threaten ing gap between dialectology and traditional philology. The chapter leads to a blind alley (66) and leaves the reader with bleak prospects. Appended to it is a colorless six-line statement on the study of place-names ; the reader will not realize that toponymy is flanked by several other ono mastic disciplines, including anthroponymy, which, piquantly enough, has been assiduously cultivated in — of all countries — Sweden (K. Michaëlsson, Å. Bergh, Olof Brattö).

78


pages on physiological and acoustic problems, copiously illustrated with draw ings and photographs (Figs. 8-30), represent an obnoxious incrustation, worse, a "Fremdkörper", and thus almost achieve the opposite of what the author, judiciously enough, hoped to accomplish : to narrow the Prague School's exces sive division of phonetics and phonology (97). Once "national styles" of research have entered the picture, "Modern American Linguistics", almost inevitably, receives special treatment (158-185) — a privilege denied, for no apparent reason, to Swiss, British, and Soviet counterparts. The concluding chapters have been reserved, as stands to reason, for "Statistical and Mathematical Methods" (186-203) and for "Psychological and Philosophical Contributions to the Study of Language" (204-220) ; given the pronounced modernity of semantics in its most recent twists and zigzags, it would probably have been wisest to assign it a place here, in closer proximity to the latest tendencies. From the total picture drawn by M. it certainly does not follow that one trend has been evolving from the other, still less that several initially autonomous directions in linguistics have begun to coalesce. Eather one infers that there have been several sharp advances or "stabs" along different lines, with magnetic personalities, tighten ing bonds of national culture, and true scientific insights all playing their sepa rate, loosely coordinated parts in these throes, gropings, and agonizing re appraisals. Few articles, in the recent annals of Italian linguistics, have riveted so much attention to their authors, and provoked such lively discussion as the opening installment of Giulio . Lepschy's "Aspetti teorici di alcune correnti della glottologia contemporanea", ASNP, XXX (I960), 187-267. Several factors com bined to allow a mere beginner to score such a conspicuous success. For one thing, L. was a self-made man ; though, in the introductory footnote, he acknowledged his debt to several friends and teachers, especially those — like T. Bolelli — he encountered at Pisa, it was the freedom they granted him ("la cui guida libe rale") that he prized even more than their technical advice ("il cui prezioso consiglio"). And the leeway that he so thoroughly enjoyed provided a chance to free himself from the strait-jacket of the local academic tradition and to examine critically what had been fairly recently accomplished, in general linguistics, OUTSIDE Italy. The monograph, in other words, encompassed an analytical survey of such movements as those of Saussure and his disciples, the Schools of Prague and of Copenhagen, and the early state of American structuralism ; the following installment, according to the prehminary announcement (187), was to contain, tantalizingly enough, a report (a) on post-Trubetzkojan, esp. "func tional", linguistics; (b) on post-Bloomfieldian, esp. structural, linguistics; (c) and on certain then very recent trends, esp. the transition from quantitative to transformational-generative grammar (the sequel appeared with considerable delay, in 1965). For an Italian scholar, even if one takes into account L.'s transfer to the London scene and to a teaching post at Reading, L. was impres sively well informed, quite up-to-date, and decidedly precocious.26 26 In a way, the article reminds one of I. Iordan's survey, in the Streitberg FS, of (then) fashion able trends in inter-bella Romance linguistics, an article which later grew into an oft-revised and lavishly expanded book. One may also liken L.'s paper, in its rôle and its impact, to the ensemble


79

In a way, L.'s new book, La linguistica strutturale ( = LS, 1966), may rank as an attempt to carve out, from the earlier monograph, a nearly complete pano ramic view of modernistic research in and about language. I use the modifier "modernistic" advisedly, because Chap, VIII, on T.G. (175-188; now called "trasformazionale" and no longer, as five years before, "trasformativa"), clearly deals, as everyone in this country knows, with a radical departure from, indeed a superbly managed revolt against, orthodox structural linguistics.27 The "Aspetti" were not L.'s first publication; they were closely preceded by a handful of middle-sized-to-long book reviews, searching in content and incisively phrased, among which his temperately negative reaction (in ASNP, X X I X , 289-298) to Carla Schick's ill-fated synthesis II linguaggio (Turin, 1960) is particularly noteworthy, because on t h a t occasion he first intimated how he himself would go about writing a book of such scope. Between his "Aspetti" and his LS Լ. published some forty articles, notes, surveys, and book reviews, ranging from such definitional inquiries as his studies of the key terms "arbitraire du signe" {ASNP, X X X I , 65-102) and "structure" (ibid., 173-197) via mis cellaneous original investigations all the way down to exceptionally scrupulous, b u t ineluctably unexciting, annual digests of research in Italian linguistics (YWMLS, XXVII [1965], 281-288 ; X X V I I I , 305-315). I confess t h a t I find L. far more engaging and stimulat ing as a critic than as an independent researcher ; 28 he is, in fact, a master reviewer, especi ally in the field of general linguistics, which obviously arrests his attention far more than do the minute details of Romance scholarship. I n fact, it is sometimes difficult to draw a sharp line between L.'s performance as an appraiser of other experts' prospecting and his personal research. Thus, his article "Problems of Semantics" (Linguistics = Li.s, X V [1965], 40-65) starts out with a balanced discussion of four recent book ventures (A. Schaff's Introduction, P. Guiraud's La sémantique, and S. Ullmann's Semantics: an Introduction besides the latter's Language and Style: Collected Papers), but before long broadens out into L.'s own explora tions of semantic space — if I may borrow this felicitous phrase from U. Weinreich, to whose approach and stance L.'s show unmistakable resemblances. As an assessor, L. some times succeeds in writing a weighty review, brimful of ideas, of a fairly vacuous book, rather than wasting his time — and his readers' patience — tearing down an obvious failure (cf. his unequivocal reaction [Li.s, I X , 88-90] to J. Tognelli's Introduzione all "Ars puntandi") ; he also knows Һ to expatiate boldly beyond an esteemed author's self-imposed latitude (cf. his imaginative preview of potentialities in grappling with M. L. Alinei's Dizionario inverso; see Li.s, I I I [1964], 116-123). Finally, he is very adept at composite

of two slim book-length projects by E. Alarcos Llorach on the Spanish scene (1950, 1951). For once Spain preceded Italy by a full decade in its attempt at reorientation. 27 Between the two major ventures L. busied himself on two occasions with T.G. : "La gram matica trasformazionale" (SSL, IV [1964], 87-114); and "Trasformazioni e semantica" (LeS. I [l966],23-25.)Interesting, for an early reaction of his to Chomsky's recourse to "intuition", as against the confirmed behaviorists' reliance on "observation", is L.'s response to the challenge of three Texas Conferences on Problems of Linguistic Analysis of English (1956-58; publ. in 1962) ; see Li.s, X X I I I (1966), 104-113. 28 The range of L.'s independent researches is rather neatly circumscribed by the following titles: "Fonematica veneziana", ID, XXV (1962), 1-22; "Morfologia veneziana", ibid., XXVI, 129-144; cf. the English-language counterpart, "The Segmental Phonemes of Venetian and Their Classification", Word, X I X (1963), 53-66; "Note sulla fonematica dialettale", ID, XXVII (1964), 53-67 ; an edition of the paleo-Milanese grammar Prissian (1606), flanked by a discussion of selected points of older Lombard phonology, ibid., XXVIII, 143-180; and a comment on the theory that the Italian language may have existed as early as the 5th  A.D. (IMU, VIII [1963], 295-308). One major study stands topically apart, being devoted to Latin, but it shares with the others the same markedly critical and retrospective, rather than brashly forward-looking, bias : "II problema dell'accento latino. Rassegna critica di studi sull'accento latino e sullo studio dell'accento", ASNP, XXXI (1962), 199-246.

80


and at contrastive treatment. 2 9 Particularly praiseworthy is the fact t h a t L. refuses to become a victim of infatuations and revulsions, and, occasionally, revises his earlier verdicts. Inescapably, the second judgment — as the charm of novelty wears off— tends, as a rule, to be more severe, some times distinctly sterner, than the first.30 Since LS is, admittedly, a distillate of the "Aspetti", its chronological sweep is similar — except for the Preface (7-10) and a substantial Introduction (17-41) it does not take into account the pre-Saussurian period. The author has, however, meticulously up-dated his heavy bibliographic apparatus (which includes numerous references to research published in Russian) for the years 1920-60 ; and his revised survey also encompasses a listing of the investigations conducted over an additional six-year period. This expansion has had several consequences, not all of them favorable. For one thing, any historian surveying a develop ment to the very threshold of the present time is bound to see the earlier periods in much sharper perspective ; as it comes to his own day, there emerge countless trees, b u t seldom a neatly profiled forest. This liability applied even to a scholar as experienced as P . Henriquez Ureña in his Literary Currents (1945), a classic embodying a cycle of Norton Lectures de livered at Harvard ; it is pathetically observable in L.'s latest venture. W h a t exacerbates the difficulty is t h a t "modern linguistics" itself, in the early and mid-'sixties, passed through a strange and violent reversal of its immediately preceding phase. The result is t h a t the most advanced scholars, as L. presents them to the uninitiated reader, seem to be working a t cross-purposes as regards not mere details or trappings, but the very core and direction of their discipline. The total impression will, therefore, inevitably be one of añ exciting and restless community of intellectuals split into many tiny groups led by resourceful and un daunted condottieri. I think t h a t L.'s book would have been more tightly organized, more clearly arranged, and far more cogently reasoned had he had the courage to confine it to the tidily silhouetted period which came to a — less than glorious — close ca. 1955, with a brief epilogue forewarning the reader t h a t the latest movements showed such wild swinging of the pendulum as not yet to allow major events to fall into place. Such a radical decision would have implied a painful self-inflicted deprivation; while sharpening L.'s analysis, it might have impaired his design to arouse his compatriots' concern with the dernier cri. One cannot effectively reconcile the intrinsically opposed rôles of observer and participant (or, worse, propagandist).

I have insisted on L.'s prowess as a zestful reviewer because that performance, and the special aptitude underlying it, have to an appreciable extent colored his book-length project. L. is, undeniably, a tireless reader and an assiduous annotator and collator of his readings — two qualities that have immensely enriched the purely informational or referential component of his LS. Ironically, they have also had certain drawbacks, two of which invite a brief discussion. 29 The former approach is best illustrated by his delicately nuanced joint discussion (Li.s, XVI [1965], 92-100) of three Italian books traceable to the same background (A. Pagliaro and W. Belardi's Linee di storia linguistica; . Taglia vini's Introduzione alla glottologia5 including its lengthy outline of a history of linguistics ; and the same scholar's painstakingly detailed Storia di parole pagane e cristiane — a modera counterpart, let me add on my own, of H. Rheinfelder's classic Kultsprache und Profansprache). For a highly entertaining contrastive treatment see L.'s comments (Li.s, XXIV [1966], 123-125) on P. Hartmann's Theorie der Grammatik (spiced with excerpts illustrative of the author's obtrusively turgid style) and, by way of counterview, on A. Martinet's feathery and, above all, lucid Elements of General Linguistics, tr. Ē. Parker. 30 In this respect, it is reassuring to compare his, all told, favorable response to R. A. Hall's Linguistics and Your Language in ASNP, XXIX (1960), 299-303 with his studiedly cold reaction to the same scholar's incomprehensible Idealism in Romance Linguistics in Li.s, VI (1964), 99105 — L.'s critique is perhaps the most restrained and judicious statement available in English on the impact of croceanesimo on linguistics. Equally remarkable is L.'s dampened enthusiasm (Li.s, V [1964], 79-92) in coming to grips with Martinet's Functional View of Language (1962) after the more ecstatic reaction, in SSL, II (1962), 123-137, to the French scholar's Éléments.


81

First, the genre of the BOOK itself, as a tool for disseminating knowledge and for indoctrination, has not at all times exerted the same degree of influence on our discipline. Some personally reserved, not to say retiring, linguists have, let us grant, left their main impact through books ; a splendid example would be Saussure's Cours, if only we were sure (as we no longer are) that it represents a replica of an authentic piece of writing (or lecturing). Perhaps the last truly influential "monolithic" books were Bloomfield's Language (1933) and Trubetzkoj's Grundzüge (1938). What has, by and large, replaced them are such agents of transmission as classroom theatricals, seminar sessions, phonograph records, leadership in a circle, the stewardship of a symposium (annual meeting, summer institute, international congress), the issuance of a Festschrift, work on an editorial board of a journal or of a monograph series, and many other tonesetting modes and vehicles of self-expression. One learns little about these things by sitting at one's desk and perusing, page by page, the file of some prestigious quarterly. Even the most comprehensive periodicals in our domain have never REFLECTED, but only REFRACTED, reality. What is needed, as a corrective, is a quota of field-work in halls of graduate studies ; in this respect, L. has done far too little, and the lapses into naïveté caused by this omission are legion. The other consequence of L.'s excessive reliance on the reading of specialized literature is that he underrates the influence, on the recent growth of linguistics, of world events — which, of course, receive no discussion in technical textbooks and learned journals, but which directly shape the lives of "charismatic" leaders and of their docile retinues and thus often obliquely impinge on their tastes and doctrines. Some of these quakes L. mentions incidentally; but parenthetic references are quite inadequate. The Third Reich and the Second World War lost by Germany greatly accelerated the decay, and finally the col lapse, of "traditional linguistics" and caused a craving for radical change; the arrival of political émigrés on the American scene decisively affected the course of linguistics, as did de-Stalinization in the Soviet Union, the profound restructuring of American society in the last quarter-century, and many other shifts of comparable magnitude. While ivory-tower-minded linguists are free to seclude themselves from their fellow-men, for the sake of the purity and auto nomy of their intellectual pursuits, one cannot meaningfully discuss the recent history of linguistics without constant sobering references to outside tremors. 31 The consequence of all this is that L. offers his readers a string of superbly executed vignettes and cameos, centering about certain favorite books, muchadmired œuvres, and overpowering personalities, with helpful summaries, the 31 L. is not so dry as completely to dehumanize his discussion, aiming as he does at the neo phytes and the uncommitted, but the anecdotal elements with which he playfully intersperses his account have not been properly absorbed and assimilated. What, for instance, is the point of his random remark (97) that numerous American linguists are of Jewish extraction? One expects either a fuller, more explicit statement to this effect, or complete silence. On the other hand, the major rôle played by Protestant missionaries and Bible translators — a milieu that has patently produced leading figures of the caliber of Nida and Pike and has characteristically determined some of their own, and their countless students', activities qua linguists — would have deserved thorough treatment.

82


whole presented, as one would expect from a stickler for typographic stan dards, 32 in flawlessly neat form. This service to the Italian public (which he, by implication, regards as being thus far ill-informed or, worse, misguided by academic dinosaurs) he performs very creditably indeed. Viewed in a more cos mopolitan perspective, however, his book—despite its avowed cosmopoli tanism — falls flat in important respects. Above all, he leaves completely un explained the central — and by far most dramatic — jolt of the entire period on which he has undertaken to train his lens : the rapid collapse, especially but not exclusively in the United States, of the type of structural linguistics that had flourished down to the mid-'fifties. According to L., the sailing went so smoothly for structuralists around 1955 — triumph after triumph, conquest after conquest — that it is bafflhig that a whole generation of younger workers should have so resolutely veered away from ideals — gaining acceptance — which they were supposed to cherish. The turn-about was complete and ruthless in some of its human manifestations. The insiders know, of course, if not the "whole story", at least the partial answer, and realize that certain symposia convoked to elucidate the Sapir-Whorf hypothesis; the rise to autonomy of ethno-, socio-, and psycho-Hnguistics ; the activities of the Columbia University group, including its youthful leader Uriel Weinreich (whom L., inexplicably, leaves unmentioned) ; the glottochronological experiment — noteworthy, even if, in the end, it misfired — were so many omens or symptoms of the gradual recession of 1950-style austere — almost puritanical — structuralism from its high-water mark. The full story of the mentalistic counter-revolution — not only of its blueprint, but also of its almost instantaneous success — must now be told, along with an account of the revival of thoroughly refurbished IndoEuropean and gratifyingly rejuvenated Romance studies in linguistics. All this L. has tossed out, apparently because it did not fit his "paradigm" or his extrapolations.

BIBLIOGRAPHIC

GUIDE

IVIĆ, MILKA, Trends in Linguistics. " J a n u a Linguarurn'', Series minor, X L I I (The H a g u e : Mouton, 1965). P p . 260. (Orig. Pravci  linguistici, Ljubljana, 1963.) LEPSCHY, GIULIO ., La linguistica strutturale. Piccola Biblioteca Einaudi, L X X I X (Torino : Giulio Einaudi, editore, 1966). P p . 234. LEROY, MAURICE, Les grands courants de la linguistique moderne. Univ. Libre de Bruxelles, Travaux de la Fac. de Philos, et Lettres, X X I V (Bruxelles: Presses Univ. de Bruxelles, and Paris: Presses Univ. de France, 1963). Pp. x, 198. Id.. Main Trends in Modem Linguistics, tr. Glanville Price (Oxford : Blackwell, and Berkeley & Los Angeles: Univ. of California Press, 1967). P p . xii, 155. MALMBERG, BERTIL, New Trends 32 Not unlike É. Benveniste, L. is, quite properly, a severe disciplinarian in his book reviews in assessing typographic accuracy and related matters; for illustration see Li.s, VI (1964), 99 104* XVI (1965), 64; XXIII (1966), 113; XXVII (1966), 106f. His own record of neatness has been most impressive, even in such a tricky operation as the transliteration of Russian names and titles. Equally commendable has been L.'s effort to write a major portion of his papers in forceful and sophisticated English — no mean accomplishment for a scholar of his background. If his earlier experiments show occasional slips, such as the confusion of substitute and replace (Li.s, Y [1964], 82) or the awkward string of adverbs: "we can usually fairly accurately gather" (ibid., IX [1964]. 90), his current performance has become unimpeachable.


83

in Linguistics: an Orientation. Bibliotheca Linguistica, I (Stockholm-Lund, 1964). P p . [viii], 266 ; 44 drawings in text. (Orig. Nya vägar inom språkforskningen; en orientering i modem lingvistik, Stockholm, 1959). ROBINS, R. H., A Short History of Linguistics (Bloomington and London, 1967). P p . vi, 248. SEBEOK, THOMAS Α., ed. Portraits of Linguists. A Biblio graphic Source Book for the History of Western Linguistics, 1746-1963. 2 vols. Indiana Univ. Studies in the History and Theory of Linguistics (Bloomington and London, 1966). Pp. xvi, 580 ; vii, 605. Related publications by some of these same authors include: Malmberg, Structural Linguistics and Human Communication; an Introduction into the Mechanism of Language and the Methodology of Linguistics (Berlin, 1963, and New York, 1963) ; and Robins, Ancient and Mediæval Grammatical Theory in Europe (London, 1961). Some similarly slanted publications were examined by me in RPh, X V I I , 823-828 ("Bibliographic Notes: History of Linguistics") : J. T. Waterman, Perspectives in Lingui stics; an Account of the Background of Modern Linguistics (Chicago, 1963) ; L. Kukenheim, Esquisse historique de la linguistique française et de ses rapports avec la linguistique générale (Leiden, 1962) ; also I. lordan and W. Bahner, Geschichte und Methoden der romanischen Sprachwissenschaft (Berlin, 1962), with a flashback to Iordan's 1924 article : "Der heutige Stand der romanischen Sprachwissenschaft". John H. R. Polt has assessed for our readers (XIII, 95f.) another monograph of Banner's: Beitrag zum Sprachbewusstsein in der spani schen Literatur des 16. und 17. Jahrhunderts, a revised translation of which is La lingüisticaespañola del Siglo de Oro (Madrid, 1966). Other fairly recent publications, a few of them obliquely touched upon here, include : A. Graur and L. Wald, Scurtă istoria a lingvistici (Bucureşti, 1961) and C. Tagliavini, Storia della linguistica (Bologna, 1963; = Introduzione alla glottologia, Bologna, 1963 5 , pp. 19-380), plus a variety of historically oriented readers, esp. H . Arens, Sprach wissenschaft (Freiburg-München, 1955) ; T. Bolelli, Per una storia della ricerca linguistica; testi e note introduttive (Napoli, 1965) ; W. P . Lehmann, A Reader in Nineteenth-Century Historical Indo-European Linguistics (Bloomington and London, 1967; cf. RPh, X X I , 359f.) ; G. Mounin, Histoire de la linguistique des origines au XXe siècle (Paris, 1967) ; and V. A. Zvegincev, Istorija jazykoznanija ... (Moskva, 1960). Of the two Danish pioneering ventures, V. Thomsen's shorter but better balanced Sprogvidenskabens historie: en kortfattet fremstilling (København, 1902; pp. 87) has become best known through H. Pollak's tr. : Geschichte der Sprachwissenschaft bis zum Ausgang des 19. Jh.; kurzgefasste Darstellung der Hauptpunkte (Halle, 1927); the Spanish version (Bar celona, 1945) by J . de Echave-Sustaeta contains a Preface and an Epilogue. H. Pedersen, though by far the younger man (he wras born in 1867, exactly a quarter-century after Thomsen), wrote his first pamphlet earlier: Sprogvidenskaben (Flensborg, 1899; pp. 64); returned to the fray in a presidential address of 1916 (Et blik paa sprogvidenskabens historie med særligt hensyn til det historiske studium av sprogets lyd, København), and presented his definitive account in 1924 (Sprogvidenskaben i det nittende aarhundrede; metoder og resultater (København; that same year there appeared a Swedish translation in Stockholm). Outside Scandinavia Pedersen's book became familiar through J . W. Spargo's translation (Cam bridge, Mass., 1931) : Linguistic Science in the Nineteenth Century; Methods and Results), of which a paperback edition (Bloomington, 1962) bears the fancy title The Discovery of Language. On B. Delbrück's Einführung see p. 49, above. At proof I can announce the appearace of a four-hundred-page book by t h a t talented Italian medievalist A. Vàrvaro: Storia, problemi e metodi della linguistica romanza (Napoli, 1968), whose originality consists in deliberately merging the history of a discipline with the survey of problems and methods germane to it.

. CLUES AS TO DATING

RANGE OF VARIATION AS A CLUE TO DATING

The Method at Issue FOR ALL scholarly disciplines concerned in one way or another with time, the matter of dating is crucially important. Alternative or mutually complemen tary techniques of chronometry, whether absolute or relative, are binding on the geologist, the paleontologist, and the paleobotanist no less than they are on the archeologist, the art historian (including the connoisseur of verbal arts), and the diachronically oriented linguist. Indeed, if within the complexly structured domain of "philology" the last-mentioned expert may afford to be less excited by problems of authorship than has been traditionally the literary historian, both in equal measure engage in the neatest possible reconstruction of sequences of events. Granted this intensity of their commitment to the time axis, it is small wonder that both theorists and practitioners of temporally slanted disciplines should be bent on devising ever new techniques of direct and indirect dating. Alongside the time-honored comparative method in linguistics, operating, for an assumed succession of splits, with the test of common innovation, imaginative scholars have of late placed the subtler mosaic of internal recon struction, which unavoidably bears on temporal concatenations of processes. The recurrent correlation, in certain controllable instances, of a particular segment of time and a specific territory has emboldened some linguists to draw temporal (stratigraphic) inferences from characteristic patterns of areal distribution even in default of any independent control. The assumption of an average rate of attrition, hypothesized in certain key sections of a typical I am greatly indebted to a number of friends and colleagues who have carefully read a pre-print version of this paper, giving me the benefit of their advice and, in many instances, of their specialized knowledge (none of them, needless to say, is responsible for the article as a whole or for any particular s t a t e m e n t ) : Samuel G. Armistead, Robert Austerlitz, Emmanuele Baumgartner, Madison S. Beeler, Curtis Blaylock, Margaret S. Breslin, Catherine A. Callaghan, Diego C a t a l á n Menéndez-Pidal, Jerry R. Craddock, Peter F . Dembowski, Søren Egerod, Percival . F a y , Erica . García, J. Gulsoy, Dell H. Hymes, Henry R. and Renée T . K a h a n e , Ilse Lehiste, Winfred P . Lehmann, Paul M . Lloyd, Raimundo Lida, Thomas Montgomery, Rebecca Posner, Ruggero Stefanini, Knud Togeby, Karl D. Uitti, and M a r t h a G a r r e t t Worthington.

88

CLUES AS TO DATING

lexical stock, has prompted venturesome anthropologists to extrapolate from patterns of verbal diversity (interpreted as results of gradual estrangement and differentiation) the approximate starting points for characteristic twoor three-way cleavages or even for the total collapse of major language families.1 In analogous fashion, literary scholars have assembled their own batteries of test-tubes. Narrowly defined grammatical preferences, phonic and accentual idiosyncrasies, metric and strophic peculiarities (in some instances) form one ensemble of tell-tale features — traits not always perceived, and, a fortiori, not always deliberately modulated and self-imposed, even by acknowledged masters of verbal artistry. Over against these data one is free to place another set of revealing details — period pieces, as it were, which in the hands of sensitive analysts provide criteria for establishing a score of a quo and ad quem dates: the author's (or his protagonists') references—straight or oblique—to contemporary events (wars, reigns, sensational crimes, cata clysms, discoveries, inventions) or to conspicuous personalities; specification of garments, finery, jewelry, furniture, objets d'art, gastronomic tidbits, and other items neatly datable because subject to sharply contoured vogues and caprices of fashion. Add to this volume of topical information the conclusions that one can draw from the implements and paraphernalia of script and print, including shape of letters, spelling habits, quality of parchment (or paper) and binding, watermarks, illuminations, etc.2 1 For an authoritative pronouncement "On the Methods of Internal Reconstruction" t u r n to J. Kurylowicz's paper so titled, in Proceedings of the Ninth International Congress of Linguists..., ed. H. G. L u n t (The Hague, 1964), pp. 9-36; this technical term — trace able, one learns, to the late 'thirties — serves to denote " t h e diac hronic conclusions t h a t may be drawn from a synchronic analysis of linguistic d a t a without or before having recourse to comparison, linguistic geography and «areal linguistics», and glottochro nology". M. Birnbaum deals with this m a t t e r in his article, ' O n the Reconstruction and Predictability of Linguistic Models...", Sc.-SI., X I I I (1967), 105-114 and promises to return to the fray in Vol. X I (1968) of IJSLP. T h e best-known guides to glottochronology and its ramifications are the two articles by Sarah C. Gudschinsky, " T h e ABC's of Lexicostatistics (Glottochronology)", Word, X I I (1956), 175-210, and by D . H. Hymes, "Lexicostatistics So F a r " , CA, I (1960), 3-44, the Supplement to w h i c h — e m b o d y i n g reactions of fellow-scholars — contains, interestingly, a mise au point by Miss Gud schinsky (pp. 39f.). For a later and still essentially valid notion of Hymes' view see his editorial remarks in the reader Language in Culture and Society (New York, e t c . , 1964), pp. 567-573. One European reaction is E . Coseriu's "Critique de la glottochronologie appliquée aux langues romanes", Actes du Xe Congrès Int. de Ling, et Philol. Romanes, I (Paris, 1965), 87-95. On W. Meyer-Lübke's and E . Richter's experiments in relative chronology — in the wake of . Bremer's pioneering paper — see my article "A T e n t a tive Typology of Romance Historical G r a m m a r s " , Lingua, I X (I960), 321-416, esp. 353-356; reprinted, with slight changes, in the miscellany Essays on Linguistic Themes (Oxford, 1968), pp. 71-164. 2 The sources of information on techniques of dating in literary scholarship are in exhaustible, but I know of no single treatise geared to this sole aim and cutting boldly through ages, territories, and national cultures — a book, in short, comparable to F . E . Zeuner's Dating the Past (1950). We do have a storehouse of excellent monographs. Acclaimed as a milestone in methodology and partially confirmed by subsequent dis coveries has been S. G. Morley and C. Bruerton's frankly experimental Chronology of Lope de Vega's " Co medias"... (New York, 1940); for reactions, see B. M. Woodbridge's bibliography in RPh., VI, 221 ( # 4 8 ) . A good example of closely reasoned argument has been the controversy surrounding the genesis of the Old French epic, with special refer-


89

The development of formal criteria and of rules of thumb for dating has been progressing at a brisk pace in natural and in social sciences. Suffice it to recall the benefit anthropologists and prehistorians have of late been deriving from the radioactive carbon (C14) method applied to fossils, in conjunction with such earth-science approaches as the stratigraphic-geomorphic method, the association with glacial deposits, observations of changes in the shore-line level, the computation of rates of dune migration and travertine deposition (growth of stalagmites and stalactites), and the like. Geochronologists harness, in their schemes, data abstracted from the astronomical elements of the inclination of the earth's axis, the eccentricity of the earth's orbit, and the longitude of the perihelion, or they establish a graph determining the fluctua tions of solar radiation received by the earth's atmosphere. 3 Rather than providing a thumbnail sketch of all currently accepted or tolerated methods of dating, the present article aims at drawing attention to yet another, apparently hitherto neglected, technique of arriving at a rough estimate of time lapse. This technique is meant to be applicable only under certain conditions; it seems to be perfectly compatible with the most solid among the older and better established approaches, and its rôle may thus be described as supporting, corroborative, or qualifying.4 Oscillation in language growth — dramatized by the most authoritative research in dialect geography — is such a prominent condition of change that even a scholar as prone, by temperament, to schematization as was L. Bloomfield could not refrain from granting it a generous measure of recognition in his influential book Language (see Chap. 22: "Fluctuation in the Frequency of Forms"); yet even he did not propose on that occasion to use the range of ence to the newly discovered " N o t a Emilianense" and to the incubation of the Roland; for one investigator's perspective see R. N . Walpole, RPh., I X (1955-56), 370-381 and X, 1-18 (review of D . Alonso, " L a primitiva épica francesa...", RFE, X X X V I I [1953], 1-94), and, for a counter-view, G. Menéndez Pidal, "Sobre el escritorio emilianense en los siglos X a X I " , BRAH, C X L I I I (1958), 10-19. A classic instance of wide discrepancy in dating has been the Song of the Host of Igor', which one school of thought regards as authentically archaic, and another as merely archaizing — a particularly ticklish di lemma. Another famous chronological crux has been Petronius' Satyricon; cf. RPh., I X , 353-359. 3 1 owe most of my second-hand information to two surveys which, significantly, figure as the opening pieces in the trail-blazing miscellany Anthropology Today: An Ency clopedic Inventory, ed. A. L. Kroeber (Chicago, 1953): R. F . Heizer, "Long-Range D a t i n g in Archeology" (pp. 3-42), and K. P . Oakley, " D a t i n g Fossil H u m a n R e m a i n s " (pp. 43-56). See also R. L. Carlson, " R e c e n t Developments in Radiocarbon D a t i n g : Their Implications for Geochronology and Archeology", CA, V I I I : 4 (Oct. 1967), 349-352. M a n y related topics, judging from the prospectus, were broached at the UCLA Confer ence (Oct. 26-28, 1967) "Applications of Science to Medieval Archeology", whose pro ceedings are to appear as a separate volume. 4 I n focusing attention on the conclusions t h a t an experienced etymologist can some times draw from the absence of variants in the dossier of the word or word-family under investigation, I became for the first time aware of the independent possibility t h a t the record of fluctuation might conceivably yield clues to approximate dating as well. See my contribution ("Etymology and General Linguistics") to a special, expanded issue of Word, X V I I I (1962), 198-218, at p. 215, and the revised version in Essays on Linguistic Themes, p p . 193f.

90

CLUES AS TO DATING

variation, be it only experimentally, as an autonomous means of, or clue to, dating. 5 Yet granted that the impact of certain linguistic upheavals — not unlike some rock hitting the smooth surface of water — produces clearly discernible undulatory movements tapering off into gentle ripples, can one not visualize a technique enabling the analyst to infer from the duration and liveliness of these ripples the — ideally — exact moment of the impact as well as the most conspicuous accompanying circumstances? If that much is allowed, should one not start out from situations in which impact and ripples are independently observable, then, after establishing a network of plausible relationships, apply the experience thus gained under optimal circumstances to less transparent situations? As when the ripples are again amenable to inspection and measure ment, but the original impact is irremediably shielded from direct observation? Under certain circumstances, I repeat, one is tempted to surmise that an abundant and, above all, erratic corpus of variants is apt to yield valuable clues to chronology—very broadly understood. What are, typically, such sets of circumstances, and which variants bid fair to offer the most eloquent testimony? There exist perhaps in language history three major causes for the rise of a plethora of variants. One is dialect mixture: Two systems, tightly organized and at the outset self-sufficient, become subject to contact and interpénétra tion (as a result of, say, geographic contiguity or proximity stimulating cultural osmosis) ; as a consequence, there emerge countless doublets — which at a later juncture may, but need not, be eliminated or semantically differ entiated. Another cause of internal proliferation is analogical innovation: Alongside the established form, particularly one less than forceful in conveying the grammatical shell of the intended message, speakers place a rival form endowed with greater strength of suggestion. ("Weakness", in this context, may be due to ambiguity or to uncharacteristic contour; those speakers of Italian who have substituted cantavo Ί was singing' for -ava < -ĀBAM achieved at one blow a modicum of contrast of the 1st and the 3d pers. sg. and pleasing hypercharacterization of the former.) Innovations of this type do not neces sarily triumph, because the liabilities they turn out to entail may, on balance, outweigh their initial assets, or because an influential conservative ingredient of the speech community may reject them as conflicting with accepted usage and may in the end succeed in stifling them. The third kind of variants are apt to erupt as an uncoordinated set of reactions to a new situation. Suppose a powerful wave of apheresis, syncope, or apocope of unstressed vowels hits a language; or a phonological tendency suddenly imperils an inflectional trend. The speech community is bound to be profoundly shaken — on various levels 5 One can now profitably consult the extract Language History from "Language" by Leonard Bloomfield, ed. H. Hoijer (New York, etc., 1965), with a sprinkling of crossreferences (p. 500) to textbooks of later vintage by C. F . Hockett (1958), W. P . Lehmann (1962), and R. A. Hall, Jr. (1964).


91

of awareness — by such a cataclysmic event, and small groups of its members, as if running for cover from a disaster, will each in certain instances be un consciously experimenting with adjustments, expansions of neuro-muscular skills, interplays of analogy and other ' 'therapeutic' ' devices and emergency solutions. At this point, the possibility of drawing cogent inferences as to dating from the sheer multitude and disarray of remedial actions launched becomes very seductive. Speech communities are diversely organized in social terms, some being known for speedily leveling out unprofitable multiformity, while others, more loosely structured as regards communication lanes or more apathetic in their reactions to emergent infelicity, seem tolerant of the protracted coexistence of practically undifferentiated doublets and triplets. It is, patently, to the former type of community, known from other contexts for its efficient handling of lexical and grammatical "deadwood", that our considerations apply most appositely. If such a community has, in general, established an unbroken record of fairly quick and thorough disposal of superfluities, then it is permis sible to infer, from the quite exceptional copiousness of variants ("ripples") in, say, a given segment of time, that the initial impact which set the surface in motion must have occurred shortly before that time, rather than in the distant past, because only thus does the failure of the expected leveling action to make its effects felt receive a plausible explanation. The purpose of this article is neither to offer a general, close-knit theory of the chronological implications of the scope of variation, nor to parade before the reader a long procession of heterogeneous illustrative examples, each but briefly highlighted. By way of "first approximation", I have chosen a radically different approach: Only two problems will be subjected to fine-meshed scrutiny, which in their cases was on any account long overdue. The limited results of these separate inquiries should stand on their own feet and perform useful services regardless of the acceptance or rejection of the general assump tions here experimented with; but it is within the framework of these assump tions and as an optimistic prelude to more ambitious studies so slanted that the narrowly circumscribed monographic investigations here undertaken can, I hope, acquire their full value. The first problem chosen is the strong preterite of TRAHERE 'to draw, drag' in Hispano-Romance. By way of preliminaries let me remark that Spanish, not unlike other leading European languages, occasionally shows two rival forms, regionally or socially stratified, in its verbal paradigms. Thus, from andar 'to walk' the standard preterite 'I walked' is anduve, but the less erratic counterpart ande is still widespread in dialect speech, just as English tolerates dreamt and leapt beside dreamed and leaped or as German admits wog beside wägte 'I weighed', or, with stylistic differentiation, wob beside webte 'I weaved', or, with semantic differentiation, schuf 'I created' alongside schaffte 'I per formed, accomplished', both correlated with schaffen. The unique and astonish-

92

CLUES AS TO DATING

ing peculiarity of Spanish traer 'to bring' and of its Portuguese equivalent trazer, anciently also trager, is that not two, not three, not four, but approxi mately ten major morphological variants are on record. Of these the two far and away best-known are both characterized by some such consonant phoneme as / s / , / s / , or / x / , synchronically unpredictable from the infinitive, but historically either rooted in the / k s / cluster of the Latin prototype TRAXI Ί drew, dragged' or attributable to a transparent analogical deflection. One variant (and its satellites) is dominated by a nuclear back vowel—witness Ptg. trouxe, OSp. troxe, mod. dial. Sp. truje; the other by a central vowel: Stand. Sp. traje. On closer inspection the specialist discovers a much greater wealth of variants, at present either obsolete or relegated to obscure dialects. The most striking items include OPtg. trouve and trousse, both of which share with OLeon. trogue the nuclear back-vowel or a related diphthong. Another cluster of variants comprises the widely scattered, thinly spread items marked by some such nuclear diphthong as ai or ei, of which the monophthong e is the obvious reduction. Most of the relevant forms are medieval and marginally represented; I can adduce, from philologically filtered sources, OPtg. treisse alongside OSp. trexe and ONav. traísse which, respectively, reflect Western, Central, and Eastern usage; to these three one may cautiously attach the rare modern var. trije, identified in Viscaya, i.e., in close proximity to Basque territory. The picture is rounded out by two ancient Eastern forms tied together by an a monophthong, a very aberrant trasqui used in the early 13th century by the oldest Spanish poet known by name, Gonzalo de Berceo, who hailed from La Rioja, an area describable as an apple of discord between Navarre and Old Castile; and, structurally at the opposite end of possibilities, trai, found sporadically in Old Aragonese and classifiable as a completely regularized form, not unlike E. dreamed and leaped in their relation to dreamt and leapt.6 It is difficult to escape the impression that the process on which our lens has been trained involves a sort of "rash", all the more so as the cognate languages, by way of perfect contrast, show an almost entirely uneventful development of the sigmatic TRAXÏ paradigm. An eruption, if I may embroider on the metaphor, calls for the service of a diagnostician, whose 6 I find myself in sharp disagreement with R. de Dardel, Le parfait fort en roman commun (SPRF, L X I I ; Geneva, 1958), p. 53, who — superimposing the familiar Italian conjugational p a t t e r n on the congener languages, as it were — tries to segregate the· variants according to whether the given form is radical-stressed or suffix-stressed. I n his eagerness to press this issue, which is at the heart of his entire theory, he neglects to distinguish between simple and compound verbs, which have not at all times behaved alike, at least not in Hispano-Romance, where retraer may tend to part company with traer, see fnn. 44f., below. For a critical appraisal of Dardel's controversial book see R. L. Politzer, MLN, L X X V (1960), 717-719, and G. Meilor, MLR, LV (1960), 440f., who both persuasively attack his central thesis; J. Bourciez, RLaR, L X X I V (1960-61), 133f., who concentrates his attention on details; H. Rheinfelder, ASNS, C X C V I I (1960-61), 357f., who chides the author for rigidity of analysis, for staking out excessive claims for "Common R o m a n c e " , and for disregard of readily available sources; and, above all, R. Posner, RPh., X V I I (1963-64), 419-431, in her masterly review article "Phonology and Analogy in the Formation of the Romance Perfect".


93

prime assignment is the prompt identification of an irritant or of an ensemble of irritants. Disappointingly, older scholars have channeled all their efforts into diligently collecting, sifting, localizing, and dating the scattered variants — a useful preliminary operation, but clearly no end in itself — though even within these strict limits none seems to have endeavored to prepare a full inventory.7 The trail-blazers were, on the whole, quite successful in isolating those models on which each individual form may plausibly have been pat terned (say, trasqui after nasqui 'I was born' and visqui 'I lived', trouve after houve I had', troque in the wake of -uī preterites, and tro(u)xe through a blend of -uī and -χι or of their direct descendants), but they stopped short of raising the fundamental question : How can one account for the — at first glance — bizarre-looking concentration of variants so numerous and so exuberantly diversified in just this southwestern corner of the entire far-flung ROMANIA? 8 Also, as a rule pioneers failed to pinpoint the phonologically "regular" out come of TRAXĪ (OSp. trexe or traxe?), a maneuver needed to gauge the more accurately all deviations from the norm. Most disturbingly, they were quite inexplicit about the likeliest chronological dimension of the epidemic of variants : Should their respective prototypes be traced to the level of provincial Latin, or does their incubation fall within the confines of the most critically important period of the Middle Ages (10th-13th centuries)? Coming back to the kernel of our own problems: Though phonemes may from the outset be suspected of having played a major part in this entire process, by providing at least the initial spark, most phases of the ensuing chain reaction distinctly give the impression of representing a complicated interplay of analogical imitations and adjustments. The ingredients of the situation were in part similar, in part radically different in the case of the Old Spanish syncopated future, which has here been selected as the second specific assignment and testing ground. The resem blances consist in that the object examined contains specimens of the verbal paradigm particularly rich in radical-final variants, which must consequently again be classified as morphological (specifically, inflectional) in nature; and in that the ignition was once more — and this time quite patently and in controvertible—provided by sound-change, namely by the loss in proto7 Particularly irksome is the tendency of Lusists to disregard the Spanish material and vice versa, although the two sets of d a t a are equally valuable and mutually illumi nating. Thus, A. Magne, in his Glossary to A Demanda do Santo Graal, I I I (Rio, 1944), s.v. trager (pp. 397-399), has painstakingly and tidily assembled almost all pertinent data, so far as Old Portuguese is concerned, without at all extending his alacrity in the direction of Old Spanish. 8 The sheer number of variants in itself commands attention, just as the sudden crowd ing of hypothetical forms has an intrinsic diagnostic value; for one cautious hint see my article " T h e Hypothetical Base in Romance E t y m o l o g y " , Word, VI (1950), 42-69, esp. Section 35, devoted to multi-pronged development: Parallel series of hypothetical forms strengthen rather t h a n weaken the assumption; the impression they produce is t h a t speakers, groping for the right derivative still absent from the s t a n d a r d language, were making impatient a t t e m p t s in various directions. I n default of any authentic s t a n d a r d word no leveling force was available to efface the initial wealth of tentative coinages.

94

CLUES AS TO DATING

Spanish, shortly before the crystallization of the literary language at its most archaic, of all intertonic vowels inherited from the ancestral language with the sole exception of a: hence com(e)rá 'he will eať, mor(i)rá 'he will die' against morará 'he will dwell'.9 Again one witnesses an inordinate accumulation of variants, but on this occasion the flux has been provoked less by "wild leaps" of morpho-semantic analogy (which we saw so clearly at work in the case of trouve, trogne, etc.) than by a succession of minor jolts and tremors indicative of purely phonetic adjustments, some analogical in essence, many of them gradual. The roster of forms recorded in medieval texts and, on a minor scale, of those extracted from modern dialectal specimens is most impressive and includes such categories of transmutation as (a) straight loss of the vowel, i.e., vocalic syncope (podrá); (b) loss of the vowel with further reverberations, e.g., insertion of a homorganic buffer consonant (tembrá, pondrá), metathesis (terná), partial adjustment of the first consonant within the newly formed cluster (istrá from exir, tandrá from tañer, codrá from coger), emergence of a new phoneme / R / from the welding of the first consonant to the second (morrà); () syllabic syncope (dirá, fará from dezir and fazer, respectively); (d) successful resistance to syncope in certain small nuclei—in part ver nacular (sofrirá), in part learned (reprehenderá) — which, in due time, became outposts for a powerful countermovement. The novelty of the treatment here offered consists (a) in a scheme of syncopated futures that takes bet ter into account the dynamics of the zigzagging process than did other wise equally meritorious schemes previously advocated; (b) in a perceptibly heavier emphasis on the hard core of resistance to syncope, discernible even at the height of its vogue; and, to revert to our principal theme, (c) in the tentative assessment of the chronological evidence that can, through a blend of rigor and imagination, be extracted from the profusion of variants. In a nutshell, what ties together the two otherwise disparate inquiries is the idea t h a t T H E RECENCY O F A PROCESS CAN B E I N F E R R E D FROM STRIKING CON

GLOMERATIONS OF VARIANTS-PROVIDED THESE CLUSTERS ARE EMBEDDED IN A MATRIX CHARACTERIZED BY THE CEASELESS AGENCY OF ECONOMY AND OF LEVELING FORCES.

Problem A. Latin

TRAXÏ

in Romance: its Descendants

and Substitutes

I. Preliminaries. By way of anticipation it may be stated that on its way from Latin via Old Spanish to Spanish the language under study gradually shed most of its "irregular" (i.e., partially "strong") preterites. The tighten ing of resources occurred on two levels, as it were: Speakers eliminated entire classes (e.g., the reduplicative class; last remnant: OSp. estido ← STETIT ← *STESTIT) and sharply reduced, one by one, the numbers of the representatives of such classes as were marginally tolerated (sigmatic: dixo ← DIXIT; long9 In dealing with the syncopated future, I usually cite, for a variety of practical considerations, the 3d pers. sg.


95

vowel: fe-,fi-zo ← FĒCIT). It is, therefore, realistic to assume that, amid these mounting losses the accidental rise of a new conjugational pattern, such as trexo ← TRAXIT vs. inf. traer/ pres. ind. trago, conflicted violently with the broad trend and must consequently have been acutely unwelcome to the speech community. This is the first element of the situation, the irritant. As regards the second element, the comparative speed and efficiency with which Spanish, in general, keeps the proliferation of variants under control, eloquent examples are plentiful and can be culled from practically any facet of the language viewed in diachronic projection. While Portuguese to this day wavers between oi and ou, proto-Spanish quickly resolved the incipient fluctuation uo ~ ua ~ ue; it refused to shelter side by side the derivational suffixes -ero and -duero, while Italian is still reconciled to the protracted coexistence of such pairs (semantically undifferentiated) as -zero and -iere, or -aio and -aro; its economy further shows in the compression of four conjugation classes into three (counter to the preference of Catalan and of its extraPeninsular cognates); in the reduction of models of weak preterites from three to two (counter to the looser budget of Portuguese); in the simplified handling of possessive adjectives (mi casa, mi padre — counter to the consensus of Portuguese and Italian) ; in the relative scarsity of wavering in the ranks of preterites and past participles (cf. visto 'seen' vs. It. visto/veduto); in the streamlined mechanism of auxiliaries used to derive past tenses (counter to French and Italian); and in thousands of lexical situations (such as the char acteristically distributed reflexes of PRETIUM 'price': Sp. preci-, except for archaic prez, vs. It. prezz-/preg- vs. Ptg. preci-/preç-/prez-). With rare excep tions, the more complicated structure currently rejected was, at some earlier stage, actually peculiar to Spanish as well, so that we can confidently vindicate observable simplification, not just continued echoes of initial simplicity. There may be considerable wisdom in examining, before the detailed analysis of the descendants of and replacements for TRAXI, certain circum stances surrounding that central problem as well as a few side issues which might otherwise clutter up the discussion at a point where brisk progress will be most advantageous. 1. HISTORY OF THE PROBLEM. The proliferation of preterite forms in the paradigm of a

few Old Spanish verbs — especially estar 'to stay, stand, be' < STĀRE 'to stand (still), remain standing' and traer 'to bring' < TRAHERE — has been known for quite some time, mainly as a matter of curiosity or as evidence of loose grammatical organization.10 10 Thus, F. Hanssen, Gramática histórica de la lengua castellana (Halle, 1913), §255, cites OSp. est-ido, -odo, -udo, -ovo, and -uvo as an extreme case of "confusion" and prolif eration. To these offshoots must be added Ptg. esteve and, above all, the most conserva tive reflex, OLeon. estiedo < STËTIT (cf. Alexandre, ed. Willis, O, 574b, matching estido in MS P), a variant which did not elude the vigilance of E. Gessner, Das Leonesische; ein Beitrag zur Kenntnis des Altspanischen (Berlin, 1867), p. 23 — which, in fact, already caught the attention of A. Fuchs, Über die sogenannten unregelmäßigen Zeitwörter in den romanischen Sprachen (Berlin, 1840), p. 69. The reason for the amplitude of the waver ing—left unstated by Hanssen—was the severe morphological isolation of esti{e)do

96

CLUES AS TO DATING

T h e fact t h a t some of the most bizarre variants (from the modern angle) happen to occur in 13th-c. texts which aroused the interest of pioneer philologists has understanda bly increased the corpus of speculation. 1 1 While J. D . M . Ford, in his doctoral disserta tion (1897), could afford to show indifference to the problem, 12 a more mature F . Hanssen, t h a t same year, amassed valuable information from medieval and modern dialectal sources — not a few of t h e m difficult of access —, and, in addition, critically assessed scholarly pronouncements ranging over a quarter-century. 1 3 Also in 1897, a distinctly less talented and less scrupulous worker, A. Gassner, assembled numerous data, all of them in need of further filtering, bearing on the inflection of the Old Spanish verb, with considerable attention to traer.14 I n the following seventy years the situation has not changed substantially: Scattered pronouncements on individual forms — some original < STĚTIT, orig. *STE-ST-IT, as the lone and mutilated remnant of the Latin reduplicative device; cf. A. E r n o u t and A. Meillet, D É L L 4 (Paris, 1960-61), p . 651a. OAst. estit, in lieu of estido, in the hybrid Fuero de Aviles (§72) m a y involve a compromise with OProv. estet; see R. Lapesa, Asturiano y provenzal en el "Fuero de Aviles" (Acta Salmant., F. y L., 11:4 [1948]), p. 80. 11 As a result, in large part, of T. A. Sanchez's alertness and sharpness of focus (Colec ción de poesías castellanas anteriores al siglo XV, I I [Madrid, 1780], 554; I I I [1782], xxxvi and 437f.), we find traxe, truxe, tro{u)xe, and trasqui all five mentioned b y F . Diez, Grammatik der romanischen Sprachen, I I (Bonn, 1838), 154, and subsequent revisions; two years later, a very youthful Fuchs, op. cit., p. 35, displayed his familiarity with obsol. troxe and mod. traje (which he somewhat ambiguously tagged " m i t dem Lateini schen TRAXÎ mehr übereinstimmend" —■ does this remark on similarity presuppose retardation or regression?). Still laboring within t h e tradition of Diez, . von Reinhardstoettner, Grammatik der portugiesischen Sprache auf Grundlage des Lateinischen und der romanischen Sprachvergleichung (Strassburg-London, 1878), p. 240, recalled OPtg. trouv- forms, characteristic — so he asserted — of municipal ordinances, as alter natives to tro(u)x- forms and credited them to the influence of jazer and prazer. For s t a t e m e n t s of varying degrees of accuracy and usefulness see P . Förster, Spanische Sprachlehre (Berlin, 1880), p. 380; A. Keller, Altspanisches Lesebuch mit Grammatik und Glossar (Leipzig, 1890), p . 152; id., Historische Formenlehre der spanischen Sprache (Murrhardt [Wtb.], 1894), pp. 61f. (with an astonishingly neat delineation and adequate documentation of all major branches); E . Gorra, Lingua e letteratura spagnuola delle origini (Milan, 1898), pp. 57 ("...trasque 'trassi', di cui la forma forte normale general mente non occorre") and 149. Noteworthy is the silence of A. Morel-Fatio, "Recherches sur le texte et les sources du Libro de Alexandre", Rom., IV (1875), 7-90, esp. 34f., whose lexicological prowess is, conversely, so much in evidence throughout this article embody ing his dissertation. I regret not to have at my disposal F . A. Coelho, Teoria da conjuga ção em latim e português; estudo de gramática comparativa (Lisbon, 1870). 12 " T h e Old Spanish Sibilants", [Harvard] Studies and Notes in Philology and Litera ture, V I I (1900), 119. 13 "Über die altspanischen P r ä t e r i t a vom Typus ove, pude", Verh. des deutschen wissensch. Vereins in Santiago, I I I (Valparaiso, 1897-98), 1-7 ("Ansichten einiger Gelehr ten"). Hanssen's selective survey spans the period from W. Förster's masterly s t u d y of Romance metaphony (and H. Schuchardt's prompt reaction to it) to Meyer-Lübke's comparative venture and, anticlimactically, to E . Gorra's modest historical grammar. Neither W. Förster's article — left unfinished —, "Beiträge zur romanischen Lautlehre: Umlaut (eigentlich Vokalsteigerung) im Romanischen", ZRPh, I I I (1879), 481-517, nor the comments of his three celebrated critics: G. Paris, Rom., I X (1880), 330-332, J. Cornu, "Mélanges espagnols: I. Remarques sur les voyelles toniques", ibid., X I I I (1884), 285-297 (esp. 295ff.), and H. Schuchardt, ZRPh, IV (1880), 113-123, are directly concerned with the Romance vicissitudes of TRAXI, b u t they jointly form the backbone for all phonemic and morphophonemic inquiries into Romance metaphony. (An earlier article on metaphony by the pioneer N . Delius in JRESL, I [1859] was still held in high regard b y Meyer-Lübke, "Zum schwachen Perfektum", I X [1885], 223-267, esp. 259.) 14 Das altspanische Verbum (Halle, 1897), pp. 170 and 176f. By splitting his inventory into two p a r t s the author lost sight of the entire spectrum of variants and remained blind to its implications. I n his devastating review of this book (LGRPh, X V I I I [1897], cols. 202-206: "...Materialsammlung...vorausgesetzt, daß der Leser die gute Münze von der falschen zu unterscheiden weiß") J. Cornu corrected Gassner's misstatements on Berceo's maltraxo and Ruiz's tráyo, see fnn. 45, 72, and 75, below; cf. also fn. 99.


97

and authoritative, others trivial — have been steadily piling up, with little prospect of synthesis and no a t t e m p t to apply new linguistic insights. 1 5 I t has meanwhile become clear to all participants in the debate t h a t the point at issue is not the bare structure of t h e competing sets of six preterite forms, b u t also the configuration of three allied tenses: troxieron (say) is inseparable, in the problems it poses, from trox-ieran, -ieren, and -ies(s)en.

2. T H E RECORD OF COGNATE LANGUAGES. What most dramatically sets apart the polymorphy of Spanish and Portuguese representatives of TRAXI is the recurrent isomorphism, to the point of monotony, of the descendants of that same verb form and its closest allies in the cognate languages. The consensus extends to Catalan, in the immediate geographic vicinity of Spanish : Here the medieval forms were consistently trac, traguist, etc., and if the modern lan guage, chiefly in Valencian territory, shows instead tragué 'he took out, removed', the underlying shift to the weak preterite (with maintenance of the velar consonant at the morphemic boundary) is an innovatory feature shared by all preterites originally falling under the rubrics of -sī and -UĪ archetypes, and is thus by no means peculiar to this one verb; cf. Mod. Cat. degué = Sp. debió.16

Outside the Peninsula the initial preservation of TRAXI, with due allowance; for predictable regional sound changes, repeats itself again and again and is readily understandable against the background of the general spread of sigmatic preterites.17 All authorities on Old Provençal adduce trais alone as the local echo of TRAXI,18 and the situation in Old French, a language which has been so thoroughly investigated, shows no major departure from this basic pattern of uniformity;19 the subsequent semantic shrinkage of the verb in French, where its scope has been ultimately reduced to 'milking', and the 16 E v e n a modern scholar of A. Alonso's s t a t u r e and sophistication did not dare go beyond the fairly trivial s t a t e m e n t t h a t dial. truje and traje were a pair of morphological rather t h a n phonological variants. See his (and A. Rosenblat's) elaboration on A. M . Espinosa, Estudios sobre el español de Nuevo Méjico, in BDHA, I (1930), 80, as well as his t r e n c h a n t and, in other respects, eye-opening critique of A. Post's studies in Arizona Spanish (RFE, X X I I [1935], 67-72, at p . 70). 16 See A. M . Badía M a r g a r i t , Gramática histórica catalana (Barcelona, 1951), p . 317 (where the phrasing * T R A C U I T > tragué is awkward, not to say misleading); and F . de . Moll, Gramática histórica catalana (Madrid, 1952), p p . 232, 234, 235f. The infinitive traure is another t r a i t conferring on the Catalan t r e a t m e n t a measure of autonomy. 17 C. H . Grandgent was forced to raise his estimates of new -s(s)ī and -xī formations from ca. twenty to over t h i r t y in fairly quick succession; contrast his two succinct pres entations of this problem in An Outline of the Phonology and Morphology of Old Provençal (Boston, etc., 1905), §178:3, and An Introduction to Vulgar Latin (Boston, 1907), §429. T h e steep increase in V.Lat. -uī and -sī may be due, as Dardel made p a t e n t in 1958, to partial homonymy between certain key-tenses as already vestigially found at the classi cal level (incidit 'falls' ~ 'fell') and later augmented through the disintegration of the reduplicative mechanism. 18 C. Appel, Provenzalische Chrestomathie 5 (Leipzig, 1920), "Abriss der Formenlehre", xxxib-xxxiia : 2 traissist, 3 (a)trais, 4 traisęm, 6 traissçron; . Schultz-Gora, Altproven֊ zalisches Elementarbuch 4 (Heidelberg, 1924), pp. 99 and 104, who comments on the oc casional extension of the pris : pres model to mis : mes, b u t does not implicate trais in this wavering. 19 See K. N y r o p , GHLF, I I 2 (Copenhagen and Oslo, 1924), §§180f. (omitted from I n d e x ) ; W. Meyer-Liibke, HGFS, I 2 _ 3 (Heidelberg, 1913), §336.

98

CLUES AS TO DATING

grammatical reverberations of that shrivelling involve secondary processes of scant concern to the comparativist.20 Outside the Iberian peninsula and Gaul, it should suffice to appeal to three languages for further corroboration. In Sardinian (Logudorese) the traditional preterites, including that in s, survived unhindered until the 16th century; though M. L. Wagner adduces in this context aiunsi, arreposit, benedissi, battussi, etc., without specifically mentioning any direct reflex of TRAXI, the survival of TRAHERE in this island (REW3, §8841) argues for including its sigmatic preterite in the same stock. The wholesale replacement of this group after 1700 — comparable to the fate that befell its counterpart in Catalan — in favor of a new, analogical -esi, -isi past tense ordinarily grafted onto the present-tense stem (presi—>prendesi) or, alternatively farther south (Campidanese, Nuorese, Bittese), the extension of the "passato prossimo" at the cost of the pristine strong preterites represent irrelevant secondary complications.21 In literary Italian as well as in the dialects of the North, of the South, and of rural Tuscany, "strong" sigmatic preterites are firmly entrenched; while the position of some is not undisputed (cf. apersi ~ aprii Ί opened', [Pis.-Roman] morsi ~ morii, offersi ~ offrii 'I offered', persi ~ perdei 'Ί losť, resi ~ rendei Ί gave back', scersi ~ scernei 'Ί discerned, made out', volsi ~ volbi 'Ί turned'), precisely trassi may claim, it would seem, a kind of monopoly.22 This is the more noteworthy as the development of -KS-, which has long been a phononological crux among Italianists, 23 is characterized by a bifurcation within the vernacular layer: COXA 'thigh' > coscia, EXIT 'he leaves, walks out' > esce beside SAXU 'rock' > sasso; but, as Grandgent aptly observed forty years ago,24 the radical-stressed preterites, by virtue of their solid assignments to / s / rather than / š / , are here the one firm spot amid a maelstrom of insecurity. On the state of affairs in Rumanian, old and modern, we are fortunate enough to have several mutually complementary sources of information, including A. Lombard's comprehensive monograph.25 Here again, as in most situations so far surveyed, one recognizes a dichotomy between older usage 20 Cf. the classic analysis by J. Gilliéron and M . Roques, Études de géographie lin guistique d'après l'Atlas linguistique de la France (Paris, 1912), Ch. n i : "Traire, MULGĒRE et MOLERE", pp. 10-18 and m a p in portfolio. For a bird's-eye view of diversified semantic specialization of TRAHERE in other Romance territories consult Meyer-Lübke, R E W 3 (Heidelberg, 1930-35), §8841. 21 See Wagner, "Flessione nominale e verbale del sardo antico e m o d e r n o " , ID, X V (1939), 15f., 19-21, and La lingua sarda: storia, spirito e f orma (Bern [1951]), p p . 336f. 22 G. Rohlfs, HGIS (Bern, 1949-54), 11:1, §581. 23 This point has been very carefully studied, for once within the format of dialect geography rather t h a n of s t r u c t u r e , by R. A. Hall, Jr., " L a t i n -KS- in Italian and its D i a l e c t s " , Lg., X V I I I (1942), 117-124, including several maps. The -ss- development was initially characteristic of central and southern I t a l y (Tuscany included) as well as of N E I t a l y , with subsequent simplification to -s-; -KS- > [is], [š] originally characterized NW I t a l y . 24 From Latin to Italian... (Cambridge [Mass.], 1927), §132. 25 Le verbe roumain: étude morphologique (Lund, 1954-55); see p p . 971f., 976, 1073-75. As m a n y as 31 recorded sigmatic preterites have been preserved in Dacia, and one ob serves numerous accretions.

99


(sometimes fossilized or obliquely observable in conservative dialects) and more modern preferences; but these differences relate rather to segmental phonemes and stress patterns of personal suffixes26 than to the basic sigmatic category, which is neatly exemplified by trase < TRAXIT, surrounded by adáose < ADAUXIT 'he increased', aduse < ADDUXIT 'he brought', ajunse < ADIUNXIT 'he joined', derease < DĒRĒXIT 'he directed', (în)cinse < ( I N ) CINXIT 'he draped, wreathed, surrounded', înţelese < INTELLEXIT 'he grasped, understood', merse (obs. meár-, -se) < MERSIT 'he dipped, plunged', and the like. Dialects include such departures from the main course of events as Transylv. (1st sg.) tras, also, outside Daco-Rumanian, Megl. tras and Arom. trapşu,27 all of which conveniently fit the mould of TRAKSI beside, at most, *TRAPSĪ — a trivial variation, if measured by the yardstick of Balkan Romance. To sum up this excursus, nowhere except in the Spanish-Portuguese domain does the record of regional forms, extinct or living, offer the kind of jigsaw puzzle we described at the outset — a profusion of coexisting, overlapping, or mutually adjoining forms whose genesis, rise to prominence, and, in many instances, rapid eclipse inevitably teases the imagination of any reconstructionist. 3. H I S P A N O - R O M A N C E TRANSMUTATIONS OF THE SOUND SEQUENCE ֊ax-. Any

attempt

to

discriminate between the one normal and the many erratic outcomes of TRAXĪ clearly presupposes, on the p a r t of the analyst, the ability and readiness to commit himself as regards the local products of the segment -AX-, in general. Unfortunately, the picture t h a t unfolds as one examines these products is motley in the extreme, while any search ing t r e a t m e n t of the problem and its ramifications is still unavailable. Suffice it to re capitulate t h a t at the s t a r t -AKS- in all likelihood cast off / a j s / and t h a t the all-impor t a n t / j / element was thenceforth able to influence, according to prevalent circumstances, either t h e preceding vowel (in the direction / a j / > / e j / > / e / ) or the following conso n a n t (in t h e direction / j s / > / j š / > / š / , spelled x) or both, almost as the / j / acted in 26 I n Lombard's scheme (p. 971), the L a t i n sigmatic preterite (UNXI 'I oiled, greased') is exemplified by this Old R u m a n i a n paradigm: (sg.) ú n ş u , unséşi, únse; (pl.) únse-mu, -tu, -rǎ (with the final vowel in *- Şi < -sī analogically replaced b - as the more char acteristic marker of the 1st sg.). The subsequent development witnessed the intrusion of -éi upon the 1st sg. (a gambit which entailed a one-step move in the direction of the " w e a k " preterite) and the leveling of the plural in favor of the 3d pers.: únser-ăm, -ati, -era. I n Transylvania Forms 4 and 5 underwent an analogical accent shift: e.g., tråsér-am, -ǎti on the analogy of furár-ǎm 'we stole, robbed', -ăţi; dormir-ăm 'we slept', -ǎţi; tacúr-am 'we kept silent', -ăţi (see Lombard, p . 976, with a reference to earlier t r e a t ments b y É . Bourciez and H. T i k t i n ) . All of this re-structuring is in no way peculiar to the one verb here at issue. 27 On Transylv. tras see the field-work report b y D . Sandru, " E n q u e t e linguistique... V. Vallée de l'Almăj ( B ă n a t ) " , BL, V (1937), 142; on Megl. tras, T . Capidan, "Megleno românii, I : Istoria şi graiul lor", Acad. Rom., Studii şi cercetări, V I I (1925), 166; from Arom. trapşu A. Graur seems to have extrapolated the matching ODac.-Rum. *trapşi in Cum vorb'im, 111:2 (1951), 3f. (a s t a t e m e n t unfortunately known to me only indirectly, through Lombard, p . 1074). On the broader problem involved see A. Rosetti and A. Graur, " L e traitement du groupe latin cs dans la flexion du parfait r o u m a i n " , ap pended to "Sur le traitement des groupes lat. c et cs en roumain", BL, I I I (1935), 74-84; re printed in A. Rosetti, Mélanges de linguistique et de philologie [1921-45], Soc. roum. de ling., 11:5 (Copenhagen and Bucharest, 1947), pp. 277-288.

100

CLUES AS TO DATING

the bilinear development (assibilation of the dental consonant and crystallization of *oi

>

ue)

of

VERĒCUNDIA

'shame'

>

OSp. vergüe-ña,

-nça

and

of

NASTŮRCIU

'water-cress' > OSp. mas-, raes -titerco.,28 From this point on the multiplication of solu tions — not a few of them ephemeral — depends on the degree of complexity of each case history. Word-finally 29 AXE 'axis, axle' > */ajs/ yielded either ax, as in N a v a r r o Aragonese, through the closest possible alliance, then amalgam, of / j / and / s / with complete indifference toward / a / , or eixo (as in Portuguese), ex(e) (as in Old Spanish), through the bidirectional agency of / j / . In t h e more intricate case of -ILLA, *-ELLA 'jaw', one expects a greater abundance of finely nuanced products branching out from [majsjeλa], and, in fact, mais-, mas(s)-, mayx-, and max- were pitted in competition with -. T h e almost completely depalatalized form mas(s)- m a y involve a kind of dissimila tion vis-à-vis t h e [λ] or t h e [j] of -iella. Here again, the a- vars. are distinctly charac teristic of Old Navarro-Aragonese (Berceo's Riojan included). 3 0 T h e dendronym TAXUS 28 Despite its heavy documentation m y article on t h e Hispanic progeny of VERĒCUNDIA in SP, X L I (1944), 501-520, has, with t h e passage of time, accumulated so m a n y leaks as to be no longer seaworthy. Mestuerço was t h e form favored b y J u a n Ruiz. T h e contra diction between CUNEU 'wedge' → cuña, on the one hand, and -ÕNEU > -ueño (cf. F A V ÕNIUS 'gentle west wind, Zephyr', and Sp. ris-ueno 'smiling', also Hisp.-Ar. halag-üeño 'flattering'), on the other, remains unexplained, despite Menéndez Pidal's a t t e m p t at a powerful synthesis in his Manual de gramática histórica 6 (Madrid, 1941), §8 bis. On CORIU 'leather' > cuero (via coiro, as preserved in Portuguese) and the like, see some hints in m y forthcoming paper " L e nivellement morphologique comme point de départ d'une «loi phonétique»", to appear in t h e Mélanges Jean Frappier. 29 T h e ֊ of P t g . eixo, OArag. axo {Alexandre, ed. Willis, P , 856c) has been secondarily added to buttress t h e use of this word as a masculine. Speakers were long undecided on its gender, as is shown most eloquently b y t h e contrast between t h e two versions of t h e Alexandre in 986d ( P : "Que el ax de la rrueda jasie t r a s t o r n a d a " ; O: "Que el ex de la Rueda azie t r a s t o r n a d o " ) ; t h e masculine was favored by Berceo: " M a s volvióse la rueda, fue el ax t r a s t o r n a d o " (San Laurencio, 24c). T h e modern form eje is closest t o OLeon. , unapocopated in Alexandre, 856c (O): " E l exe de fin argent que cantasse mejor". On the two major processes involved in this proliferation of variants see R. Lapesa, " L a apócope de la vocal en castellano antiguo: intento de explicación histó rica", EDMP, I I (1951), 185-226; and my own article "Diachronie Hypercharacterization in R o m a n c e " , ArL, I X (1957), 79-113, and X (1958), 1-36, esp. 22, with numerous additional illustrations. 30 Here is a small array of examples. T h e forerunner of the modern form is mexilla, found in M S Ρ of Alexandre (632α, 648α — t h e corresponding q u a t r a i n is absent from O), where it m a y constitute a Castilianizing feature. T h e same version contains t h e (doubtless more authentically Aragonese) form maxilla: "Los cabellos en t u e r t o , la maxilla delgada" (34b); "sedie mano a maxilla Venus, duelo faziendo" (630α1), whereas the older M S O, known for its Western flavor, offers the more archaic forms mayxiella (34b) — v e r y striking i n d e e d — a n d maxiella (630d). T h e -x- is, then, endemic to t h e poem, though t h e range of scribal diversity through complex transmission has left an ample margin for two sets of v a r i a n t s : (a) may- ~ ma- ~ me- and (b) -iella ~ -illa, a point swept under t h e rug b y J. Keller, Contribución al vocabulario del "Poema de Alixandre" (Madrid, 1932), p . 128. T h e corpus of Berceo MSS contains both -χ- forms, e.g. maxiella (San Millan, 2096, 229α) and -s(s)- forms, e.g. mastella (Milagros, I, ed. Soialinde, 5086) and massiella (Duelo, 346, and San Millan, 372c), with complete consistency, judging from V. R. B . Oelschläger's compilation (1940), about t h e vowels (diphthong included) flanking the sibilant and with t h e limited possibility of interpreting -s(s)-, like the more common ONav. -iss-, as a local graphy for / š / ; cf. such graphies as aduso, dey sos, Leso, cueysa beside ysieron in t h e Old Navarrese Libro de las generaciones. T h e eventual choice of mexi(e)lla m a y have been dictated b y the need to steer clear of colli sions with t h e products of MISELLU 'wretched' (OSp. mesiello, as in Berceo, Milagros, 471d; note Solalinde's reference to Buenos proverbios, ed. H . K n u s t , p . 56) and of -ULA, *-ELLA 'stain, mesh (of a n e t ) , blemish' > maziella, later manzilla, cf. m y two analyses in " T h r e e Hispanic Word S t u d i e s " , UCPL, 1:7 (1947), 227-243, 269-282, and " A L a t i n Hebrew Blend: Spanish desmazalado", HR, X V (1947), 272-301. I wonder whether t h e course of mexi(e)lla was to any extent co-determined b y the fate of its near-synonym que(i)x-, quix- 'jaw, chin' < CAPSU beside more frequent CAPSA 'holder, box, case', see Lg., X X I (1945), 142-183. On the semantic ingredient of this side-issue see H . (R.)


101

'yew' and SAXUM 'rock, boulder' tidily exemplify the prevalence of -eix- [ejš] in the West, the former also of ֊ex- > -ej- [ex] in the Center; 3 1 so do, on a conceivably even bolder scale, LAXĀRE 'to extend, relax, slacken' and its family, including the adj. LAXUS 'roomy, loose, relaxed', which was transmuted into the adv. lexos 'far (away)', a successful rival of lueñe < LONGĒ; 3 2 it seems otiose to appeal, with Corominas, to the compar, adv. LAXIUS, since the ֊s can be explained, in less cumbersome fashion, as the Romance ad verbial suffix. Conversely, the behavior of FRAXINUS 'ash tree', with respect to the seg ment at issue, 33 calls to mind the comportment of AXIS and MAXILLA, SO does the split of MATAXA lit. 'raw silk, rope' into (a) OSp. madexa 'skein, hank, mass of hair' (Ptg. madeixa, etc.) and (b) U p p . Arag. madaxa (Ansó); indeed, their joint evidence strongly tempts one to posit *TRAXĪNĀRE — an assumed provincial counterpart of *TRAGĪNĀRE (from TRA-HERE, *-GERE) > Fr. traîner — as the base of an Old Spanish verb which haunts Aragonese texts as trasnar (beside postv. traxna) and Leonese texts as tresnar (beside postv. tresna);34 cf. I t . trascinare 'to drag (along), pull, trail', an old crux of K a h a n e , "Die Bezeichnungen der Kinnbacke im Galloromanischen", BBRPh, II:2 (1938), and "Designations of the Cheek in the Italian Dialects", Lg., X V I I (1941), 212-222. (All references to San Millán have been checked against G. Koberstein's edi tion of 1964.) T h u s , one finds it hard to agree with Hanssen's verdict — for once, not dialectally nuanced: "Unregelmäßig sind maxiella und ax bei Berceo" (Spanische Grammatik auf historischer Grundlage [Halle, 1910], §9.4). On the other hand, if ORioj. ma(s)siella actually contained an / s / or a / z / rather t h a n a disguised / š / , then its contrast to ax/ex might be a t t r i b u t e d to the different position of the segment within the syllable, cf. the norm laid down by G. Millardet in RLaR, L V I I (1914), 124, and Menéndez Pidal's qualifi cation in Manual 6 p . 45, fn. 1. 31 The p h y t o n y m P t g . teixo, Sp. tejo must be distinguished from the zoönym Leon. teijo 'badger' (El Bierzo), an obvious descendant of TAXU which is recorded in glosses as a variant of TAXO (see fn. 37, below). P t g . seixo 'pebble' < SAXU — involving a seman tic deterioration which interested K. Jaberg (Aspects géographiques du langage [Paris, 1936], pp. 64-68) — seems to extend only short prongs from the coast into the interior of t h e Peninsula, and G. Rohlfs' a t t e m p t to extract Arag. saso from SAXU has elicited few plaudits from J. Corominas, see DCE, IV [1957], 412a, with  dual reference to an earlier note ("Los nombres de la lagartija...", RFH, V [1944], 8f.) and to the e n t r y Sas in the forthcoming Onomasticon. On traces of SAXU in Sanabria, Salamanca, Astorga, and L a M o n t a n a as well as in some strongholds of Judeo-Spanish see M . L. Wagner, RFE, X (1923), 240, and X X I V (1950), 9-106, s.v. seso. 32 If one disregards the vexing encroachment of d֊ on I- which Hispano-Romance shares with Sardinian, it is fair to speak of a clear picture emerging from the record: ֊e(i)x- indisputably prevails in I-, d֊e(i)xar as well as in archaic delexar (cf. DÏLAXARE in Tironian N o t e s ) . Unfortunately from the analyst's angle, the area of lexos seems not to encompass the Atlantic Coast, where longe has at all times been unassailably entrenched. 33 One can theoretically explain Sp. fresno on two assumptions: E i t h e r syncope oc curred before [js] had yielded [š] and the progress toward that goal was interrupted, or t h e stage -xn- was actually reached and a kind of depalatalization ensued, in a recoil from the uncommon / š n / cluster, cf., for a parallel, the fut. istré 'Ishall leave' beside the inf. exir < EXÎRE. The record clearly favors this alternative, cf. the archaic forms fréxeno (A.D. 932), frexno (A.D. 1084), freisno (A.D. 1188), fresno (A.D. 1210) borrowed b y Corominas from Menéndez Pidal and Oelschläger. The a- vars. are, as expected, traceable to Aragonese, both medieval (fraxno: A.D. 1404; see M . Serrano y Sanz, " I n v e n t a r i o s . . . " , BRAE, I X [1922], 118) and modern, concealed in mountainous territory (fraxino: see A. K u h n , " D e r hocharagonesische D i a l e k t " , RLiR, X I [1935], 53). Interesting against this background is Moz. frahsino; see M. Asín Palacios, Glosario de voces romances (Ma drid and G r a n a d a , 1943), No. 251. 34 The full monographic t r e a t m e n t which this difficult word invites cannot be provided in this context. Of the two major redactions of the Alexandre (ed. Willis, 2254bc), the Western uses tresnar (v.), tresna (.), while the Eastern distinguishes between trasnar (v.) and traxna (.). A. Morel-Fatio's brief comment (ed. El Libro de Alixandre; manuscrit espagnol 488 de la . N. de P. [GRL, X; Dresden, 1906 j , p . xxvi, is unhelpful, as is J. Keller's belated echo of it. Corominas, DCE, IV, 569a-5706, furnishes a richly docu mented account, b u t neither its nuclear s t a t e m e n t ("probablemente tomado del fr.

102

CLUES AS TO DATING

Romance etymology. 3 5 The relatively best preservation of a in this particular phonic environment is exhibited by words either non-Latin or superficially Latinized, cf. OSp. axuar, OArag. axovar 'trousseau, house furnishings' as against P t g . enxoval < Ar. šuwâr36 as well as the parallel offshoots TAXU, TAXÕNE, and *TAXŪCU 'badger', a word of Ger manic provenience absorbed into L a t e Latin. 3 7

The conclusions it seems permissible to draw from these scraps of informa tion are these: No two words among those inspected show, in cross-dialectal perspective, exactly the same distribution of reflexes, but on the whole the West favors en bloc -eix- and certain strongholds of conservatism in the Navarro-Aragonese zone ֊ax-, while Castile ordinarily goes along with the West — on a sweeping scale in the case of semantically broad-gauged words (lexos, le- and de-xar) — being in league with the East only in the rare in stances of more concrete technical items either distinctly non-Latin or, at least, not strictly Latin. To rephrase this summation in more abstract terms, we are here confronted with a state of affairs casting the heartland of the traîner ' a r r a s t r a r ' , y éste del lat. vg. * T R A G Ī N Ā R E " ) , nor the chief explicative remark: Me parece muy probable que estemos ante un galicismo cinegético tomado del fr. traîner; aunque éste, que primero fue traîner, no tuvo nunca s etimológica, la prolongación de la vocal causada por la contracción de las dos vocales daría la impresión de que el vocablo perteneciera a la categoría de voces como frêne — fresno, âne = asno, y así lo españolizarían, convirtiéndolo en tresnar

carry conviction. This explanation, for which the author, with characteristic fickleness, is willing to s u b s t i t u t e , a few lines farther on, the alternative hypothesis of a vernacular development of *TRAXĪNĀRE, is u t t e r l y questionable, because the distribution OLeon. -esn- ~ OArag. -asn֊/-axn- unmistakably bespeaks two parallel s brains of native de velopment. 35 T h e aberrancy of *TRAXĪNĀRE, which involves the well-nigh unique addition of -ĪNĀRE to a preterite stem, is best explained by the fact t h a t , of all sections of the E m pire, t h e L a t i n i t y of I t a l y and of the Spanish provinces had given rise to *TRAGICĀRE 'to drag along' > Sp. P t g . C a t . tragar 'to swallow' (→ Logud, tragare), I t . straccare 'to fatigue, wear o u t ' ; cf. REW3, §8835. The hazardous proximity of this verb — its sway did not extend to Gaul where, consequently, *TRAGĪNĀRE has survived unscathed — , could very well have stimulated local speakers to circumvent a tradition and deflect *TRAGĪNĀRE to *TRAXĪNĀRE, a deviation hardly more startling t h a n the coinage of asir 'to grasp' from asa 'handle', since the right to the more readily postulated asar's "space" had already been locally prempted by the frequentative of ARDĒRE 'to b u r n ' (cf. Sp. asar 'to roast, b u r n u p ' ) . 36 N o t e w o r t h y archaic v a r i a n t s include axuuar, i.e., axuvar, in the Cantar de Mio Cid, w . 1650, 2571 (see Menéndez Pidal's ed., p p . 490f., with a reference to Arag. and Segorbe ajobar՛, axuvar has struck root in Judeo-Spanish) and exovar, in a dated text, A.D. 1139 (see Oelschläger). On t h e Arabic ancestry consult A. Steiger, Contribución a la fonética del hispano-árabe y de los arabismos en el ibero-románico y el siciliano (Madrid, 1932), p. 296, and E. K. Neuvonen, Los arabismos del español enei siglo XIII (Helsinki, 1941), pp. 94f. T h e distribution of the forms P t g . Teijo : Sp. Tajo, from the Arabicized hydronym TAGUS, bears remote resemblance to the situation here studied. 37 Here, in a nutshell, is the inventory of t h e most characteristic forms: OSp. texón, Arag. (Eche) tax-, taj-ón beside P t g . teixugo, OSp. tas-, tax-, tess-ugo, mod. dial, taj-ugo (Soria), tasugo (Burgos, Avila, Soria), tajugo (Soria), Arag. (Aragüés, Valley of Echo) tajubo, all of which involve suffix change and the last-mentioned also "acoustic equiv alence"; once we disregard the Avila record as u n t r u s t w o r t h y , the E a s t e r n character of the taj- zone, including its outposts in E a s t Burgos and N o r t h and E a s t Soria, will s t a n d out quite sharply (epistolary comment b y D . Catalán). F o r clues to further raw d a t a and literature see Corominas, DCE, IV, 4116֊412a, who refers the reader to G. Rohlfs' and J. Brüch's discussion of the suffix -ugo — see t h e l a t t e r ' s note in ZRPh, L V I I (1937), 69-79, omitting the s t a t e m e n t s independently made in Lg., X X I (1945), 153, fn. 109, and esp. in my "Studies in Spanish and Portuguese Animal N a m e s " , HR, X X I V (1956), 115-143, 207-231, at 132.


103

Peninsula in the rôle of a transitional area,38 quite unlike the situations so deftly highlighted by Menéndez Pidal in his Orígenes del español, where novelty-ridden Castile was pictured as driving a fan of deepening wedges into a ring of tradition-bound Hispano-Romance dialects. As far as TRAXI is concerned, the lesson these facts teach us is unequivocal: The "regular" outcome should have been treix-i, -e in the West, trexe in the Center, and traxe solely at isolated spots of the Navarro-Aragonese domain. Outside that territory, traxe represents a deviation from the norm, though, to be sure, one marked by excessive retention rather than by accelerated innovation. 4. OTHER PECULIARITIES OF THE HISPANO-ROMANCE PARADIGM OF trahere.

It is, of course, hazardous to isolate the preterite (and its appendages) from the total paradigm of the verb under scrutiny. In a way, TRAHERE, with its weak central pillar doomed to early disappearance,39 before long became an "filing verb"; note the total dislodgment from Romance of its companion-inailment VEHERE 'to carry, convey, transport' ( = G. 'fahren'). Rum. trage and OPtg. trager, whose joint evidence leans heavily indeed on the strength of their polar geographic spread ("marginal, lateral, or peripheral areas"), militate in favor of a prototype *TRAGERE, which is also reconcilable with Sp. traer, though that verb, assayed in isolation, could equally well be said to reflect TRA(H)ERE. T h e elaboration *TRAGĪNĀRE, which is generally believed to underlie Fr. traîner (OFr. trainer), presupposes a type *TRAGĪNA, conceivably adverbial like the more familiar *AGĪNA inferred from Sp. aina, OPtg. ag-, az-inha 'hurriedly' (cf. obs. I t . agina ' h a s t e ' ) . *TRAGĪNĀRE independently points in the direction of *TRAGERE, and the assumed paral lelism of *AGÏNA and *TRAGĪNA, by way of fringe benefit, identifies AGERE 'to drive, lead' as the verb which, before its own obliteration, may plausibly have stimulated the trans m u t a t i o n TRAHERE → *TRAGERE, much as AGŌ, AĞAM are often credited with having paved the way for OSp. fago 'I do, m a k e ' , faga 'let me do' (as against the rectilinear descent of P t g . faço < FACIÓ, etc.; dig-o, -a may also have intervened, as it surely did in the case of dial. I t . stag-o, -a and vag-o, -a). B u t granted the validity, as a general principle, of multiple causation in language growth, 40 the allowance made for the rôle of AGERE need not prevent us from joining C. Michalis de Vasconcelos, J. Leite de Vasconcelos, and others 4 1 in arguing t h a t it was the configuration of the preterite (TRAXI, i.e., TRAKSI) 38 As regards the p a t t e r n of regional distribution, the reflexes of -AX֊ thus fall into t h e same pigeonhole as those of -D-; they also call to mind certain relationships of -ҫ- and ֊z- (OPtg. coraçon, peçonha vs. OArag. corazón, pozón — provided this z s t a n d s for [dz]), the distribution of -er and ֊ir in verbs traceable to -ERE, -ERE, -ĪRE, and the gradient of syncope in its bearing on the future and the conditional tenses. 39 N o t for nothing did E . Richter s t a r t the procession of Romance sound changes with the loss of ; see her Beiträge zur Geschichte der Romanismen, I : Chronologische Phonetik des Französischen..., Suppl. 82 to Z RPh (Halle, 1934). 40 See m y contribution ("Multiple Versus Simple Causation in Linguistic Change") to the new testimonial To Honor R. Jakobson (The Hague, 1967), II, 1228-47, and, in partial overlap with t h a t paper, my LSA Presidential Address (Chicago, 1965) "Lin guistics as a Genetic Science", Lg., X L I I I (1967), 223-245, esp. 235-242. 41 Cf. the spectrum of finely graded conjectures in Ford, Old Spanish Readings Selected on the Basis of Critically Edited Texts* (Boston, 1911; rev. ed., 1939), p. 301a;  Michalis de Vasconcelos, "Glossário do Cancioneiro da Ajuda" [completed ca. 1905], RL, X X (1920), 90b; J. Leite de Vasconcelos, Textos arcaicos3 (Lisbon, 1922), ρ. 194α, in partial criticism of a juvenile note of his own (RL, I I [1890-92], 269-271, 349).

104

CLUES AS TO DATING

which, after the subtraction of the sigmatic aspectual or tense marker, left a residual velar element. This ingredient, through a subsequent interplay of analogies (REGÕ 'I rule' ~ RĒXĪ beside DICO 'I say' ~ D Ī X Ī ) , could have enabled speakers to extract such new infinitives as *TRAGERE and even *TRACERE, two snugly fitting prototypes for the dichotomy OPtg. trager alongside (eventually victorious) trazer. Again, the history of our verb's present tense abounds in all maimer of vicissitudes, as a result of the in terference by antihiatic devices and by the velar augment, so t h a t "etymologically correct" tra{g)-o, -es..., was, before long, flanked by smoother trayo, traes..., and ultimately evicted by traigo, traes..., a side step alien to Portuguese after t h a t language generalized the trazer branch. Finally, the future tense in the West involves syllabic syncope, inti mately associating trazer with (a) fazer 'to make' and (b) dizer 'to say' — a pair of far more potent magnets by virtue of their close mutual resemblance than have become their inflectionally dissimilar equivalents hacer and decir in Spanish — and the old imperative likewise shows vestiges of this alliance. 42

In short, TRAHERE has challenging problems in store for the language historian sensitive to form quite apart from the perplexing development of its preterite. But do these supervenient complications have a direct bearing on the kernel of our inquiry? The answer is: Only on a very modest scale. The restructuring of TRAHERE as *TRA-GERE (var. -CERE) doubtless falls into a conspicuously early period, implicating as it does Gallo-Romance (*TRAGĪNĀRE) and Rumanian (trage) on a par with Hispano-Romance. The course taken by future and imperative relates to a smaller geographic area, being of concern, strictly, to Lusists alone. The joint impact of intercalary / j / and / g / (trayo, traigo) admittedly struck traer in the late Middle Ages and thus seems chronologically a shade more relevant to the study of the preterite; but, significantly, traer has in this respect cast its lot with a group of verbs displaying enucleated radicals (caer 'to fair, huir 'to flee', roer 'to gnaw', etc.), with most of which it has nothing in common on the level of the preterite. The isolation of our central problem from surrounding issues, then, is analytically defensible and operationally advisable. 5. T H E COMPOUNDS OF traer. Traer is ringed at present by a large number of compounds, which share its inflectional peculiarities; many are crude Latinisms,43 of fairly recent vintage, and even the few that have been transmitted through vernacular conduits and thus, for a while, enjoyed relative freedom (such as retraer) have, with the passage of time, been semantically readjusted 42 The presently standardized future trarei (which in medieval texts rivaled tra-gerei, -zerei) is just as dependent on farei as are, in the present, 1st ind. traço, 1st and 3d subj. traça on faço < FACIÕ and faça < FACIA-M, - T . OPtg. treides (2d pl. ind.) and trei (sg. imper.), treide (pl. imper.) figure among the last local remnants of the authentic -ERE conjugation, along with certain debris assignable to FACERE and DĪCERE; see A. Magne, ed. Ճ Demanda..., I I I (1944), 398, in his digest of . Michaelis' pioneering remarks to this effect (RL, I I I , 188). 43 J. Wiggers, Grammatik der spanischen Sprache* (Leipzig, 1884), pp. 164f., lists compounds involving abs֊, a֊, con-, de-, dis-, ex-, re-, retro-, and sus-, all of transparent meaning.


105

44

to the Latin model. We shall not, in the following, focus attention on these verbs, because there is no evidence that they have, through any kind of "feed-back", influenced the course of traer itself so far as the preterite and allied tenses are concerned. Suffice it to remark that the best-studied of these verbs, retraer, has demonstrably participated in most of the experimental solutions which speakers have imposed on traer.45 On the other hand, since compounds in Hispano-Romance are known to part company with simple verbs on many occasions (witness the eloquent contrast between SEDERE 'to sit' > Sp. Ptg. se(e)r and POSSIDĒRE 'to own' > Sp. poseer, Ptg. possuir46), one need not be surprised to find (weak) retrai used by speakers and writers who would have winced at abandoning (strong) traxe or tro-, tru-xe in favor of traí.47 6. PERSONAL ENDINGS. Since it has been one of the collateral contentions of this paper t h a t the choice of the person (or of the specific personal suffix) has no measurable bearing on the selection of the stem variant which is here alone at issue, we can briefly record the fact t h a t trasqui in Old Riojano shares Berceo's preference for ֊i over -e in many verb forms, as well as in demonstratives, plus a sprinkling of nouns and adverbs. 4 8 Again, on the dialect level traer, in the 3d pl. of the preterite, occupies a patch of middle ground between -eron, as in dijeron < dix(i)eron, and -ieron, with analogical restoration of i, as in crujieron 'they crackled, creaked, chattered' and tej-ieron 'they wove'; in this respect it is reminiscent of hinchfijeron (see Hanssen, Span. Gramm., §26.2). Especially 44 See E. Cotarelo, ¡'Semántica española: retraer", BRAE, I I I (1916), 685-705. T h e typical medieval meanings were 'to recount (lit. 'bring back to m i n d ' ) , insult, grumble, reproach, upbraid'. The point of r e t u r n to the semantic core of the Latin prototype ('to withdraw') m a y be El Corbacho (mid-15th-c). The s u b s t a n t i a t e d infinitive retraire 'refrán' in Don J u a n Manuel is, of course, a Gallicism. Cf. M. R. Lida de Malkiel, " T r e s notas sobre Don J u a n M a n u e l " , RPh., IV, 154-194, at 187. 45 T h u s Cotarelo cites retraxieron from the 1460 version of the Libro de los siete sabios de Roma՛, retroxol from the Alexandre,  (the passage corresponds to q u a t r . 10646 in Willis' ed.); and retruxera from Gonzalo Perez's translation (1550) of Ulixea, Bk. ix, fol. 170v°. For one example of maltroxo see T. Montgomery, "On the Verb System of Biblia Escurialense 6", HR, X X X V (1967), 129-140, at p. 132, fn. 4 — an example t h e more noteworthy as maltraer otherwise tended to assert its full measure of independence (p. ptc. maltrecho 'mistreated, mangled, maimed', in sharp contrast to traído, from which trecho 'stretch' had branched off as a fossilized noun). 46 A parallel example is the gap separating proveer < PRŌVIDERE 'to see in the distance, foresee' from ve(e)r < VIDĒRE 'to see'. Add the well-known contrasts romper : interrum pir, verter : convertir, meter : remitir, in general more characteristic of Spanish t h a n of Portuguese; however, of all families t h a t of TRAHERE happens to show the reverse distri bution, since Sp. traer and atraer 'to a t t r a c t ' are neatly paired off, whereas P t g . trazer and atrair fail to match. 47 Let me adduce, from Cotarelo's copious inventory, these three eloquent examples: "Dixioron todos a Nestor las fes por synåles / que los que retrayesen que fuessen des leales" (Alexandre, 745cd []; here checked against Willis' ed. and emended); " e t otro sí, deven catar logar, de guisa que lo que retrayeren que lo digan a..." (Partidas, II.ix.30) ; "el dulce reposo buscaba de grado, / e yo retraýme facia mi m a n i d a " (Marqués de San tillana, Obras [Madrid, 1852], p. 117). 48 G. Tilander, " L a terminación -i por -e en los poemas de Gonzalo de Berceo", RFE, X X I V (1937), 1-10, at p . 7, argues t h a t the -i was inherited in the 1st sg. from the " w e a k " preterite, then extended to the 2d sg. (traxisti). For a thorough reexamination of this intricate issue see J. Gulsoy's forthcoming major article " T h e -i Words in the Poems of Gonzalo de Berceo" (to appear in RPh.).

106

CLUES AS TO DATING

in the New World one finds traces of -eron and -ieron alike soldered, significantly enough, to both traj- and truj- 49 . 7. A HOMONYMIC CONFLICT. I t is widely known t h a t traer < TRAHERE and trair 'to be t r a y ' < TRĀDERE (cf. the l a t t e r ' s remnants traidor 'traitor' and traición 'treason, treach ery') were near-homonyms in Old Spanish, a situation sufficiently irksome to have caused the extinction of trair, replaced by traicionar.50 In Portuguese the infinitives traz-/trag-er and tra-ir/obs. ֊er were adequately differentiated, b u t the sequence trei- was shared by certain archaic present-tense and imperative forms of traz-, trag-er and by members of the trair family (e.g., treiçõ 'treason'), 5 1 justifying the coinage, if not the generalization, of atraiçoar 'to b e t r a y ' . On the level of the preterite s t r a y vestiges of an overlap have been observed, with troxo meaning either 'brought' or 'betrayed'; 5 2 this is no more sur prising t h a n the ambiguity of crovo ('he grew; he believed'). Since, in this contest of strength, it was the word for 'betraying' which suffered irreparable reverses, there is little point in laying at the door of homonymy any episode in the local history of TRAXI. 8. RELATION TO OTHER V E R B S . J u s t as one cannot defensibly separate one tense of a given verb from its entire paradigm, so it would be unwarranted to sever the link of, say, traer with other verbs with which it shares or, for a while, shared salient features of form and meaning. Barring the aforementioned case of quasi-homonymy conducive to repulsion (collision of TRAHERE and TRADERE), these resemblances usually entail mutual attraction. T h u s , in territories where TRAHERE (*TRAGERE) yielded traer and CADERE 'to fall', caer, it is a fair guess t h a t the two verbs rhyming in the infinitive and in numerous tenses (present, imperfect, etc.) could also influence each other in the few divergent forms, including their preterites. 5 3 I n Portuguese, trazer was unlikely to coexist with, on the one hand, dizer 'to say' and, on the other, jazer 'to lie' and prazer 'to please' without concomitantly allowing the preterite disse < DÎXÎ to influence trouxe (orig. / t r o w š e / ) , to the point of transforming it into / t r o w s e / , and without whetting the speakers' appetite for trougue as an experimental counterpart of jougue < IACUĪ and prougue < PLACUĪ. B u t it cannot be asserted t h a t such affinities and proximities truly activated speakers groping for a new and more satisfactory preterite of tra-er, -zer, -ger; they merely channeled t h e preëxistent gropings into certain neatly bounded grooves. T h e deep roots of the dissatisfaction must lie concealed within the paradigm of the stricken verb. We stand a chance of discovering t h e m by meticulously examining the record of the various preterites toyed with by the visibly restless speech community. 49 T h u s , traj-ieran is found in southern Arizona (Yuma), where it may have been recently superimposed on truj-ieran, cf. A. C. Post, "Southern Arizona Spanish Phon ology", Univ. of Arizona Bull., V:l (1934; = H u m a n . Bull., I ) , pp. 21, 49. Though E . C. Hills was inexplicit in his slim original survey (1906) of New Mexican usage, his exegete P. Henríquez Ureña extrapolated from dij-eron ~ ֊ieron also tra֊ and tru-jieron, see BDHA, IV (1938), 38f. Trujieron — in the company of 3d sg. trujió — has been identified in Old Mexico, witness A. M . Carreno's epistle (1916) to J. M . Dihigo, capitalized upon by Henríquez Ureña, op. cit., p. 279. Truj(i)eron, truj(i)ese were observed at the thresh old of this century by J. Calcano, El castellano en Venezuela (Caracas, 1897), §188. Cf. fn. 105, below. 50 Cf. Լ. Spitzer, " C a s d'homonymie gênante en espagnol", RFE, X V I (1929), 173f., and my comment in "A Cluster of Four Homophones in Ibero-Romance", HR, X X I (1953), 20-36, 120-134, at 20f. 51 For examples of treyçõ (fol. 199v°) and treições (fol. 202v°) see H. H. Carter, Paléo graphie Edition and Study of the Language of a Portion of Codex Alcobacensis 200 (Phila delphia, 1938), Glossary, with references to J. Leite de Vasconcelos' ed. of Fabulário português, Vocabulary, s.v. treyçom, and to Henry R. Lang's ed. of Das Liederbuch des Königs Denis, Glossary, s.v. treïçom. 52 See T. Montgomery's remark in HR, X X X V (1967), 132, fn. 4. 53 In Portuguese, caer, no doubt under the pressure of sair 'to go, come, get out' < SALIRE 'to jump, bound, hop', had become cair and could thus exert no influence on either trazer or trager.


107

II. The Normal Outcome of Latin traxi. 1. A N E W STRATEGY. More than half a century ago, the earliest student of paleo-Hispanic dialectology, F. Hanssen, made a timid attempt to classify by areas the most conspicuous preterites of traer.54 In assigning a territory to each pattern, he asked himself neither whether the Spanish record can with impunity be separated from Galician-Portuguese material, nor whether the border-lines of the individual forms were sharply or hazily drawn. We shall adopt an altogether different procedure, starting from the remnants of the product presumed normal (OSp. trexe and its counterparts), examining next the form that testifies to retardation through inflectional resistance to sound change (traxe), and advancing thence toward the many analogical innovations. The latter may, for the sake of convenience, be so assembled as to highlight the contrast between the radical central-vowel group (trasque, etc.) and its radical back-vowel opposite (trouxe, etc.) —though alternative arrangements could be defended with almost equal justification. 2. T H E RECORD OF THE " E / I PRETERITE". An important hint regarding Ast. trix-i, -eron (in the company of dixi Ί said', fixi Ί did, made', quixi Ί wanted', and in rivalry with Ծօ-, tru-xi) was dropped by Å. W. Munthe, who also pointed out the affinity of these forms to Sant. tre- ~ tri-jo.bb Both trixi and trexo occur in a comparatively old anthology of modern Asturian poems.56 In his splendid 1897 monograph (p. 23), Hanssen added this shred of evidence to A. de Rato y Hevia's testimony 57 (while remaining oblivious of P. de Mugica's independent findings58) ; later, in the synthesis embodied in the two consecutive versions of his grammar, he deftly conjoined the dialectological data and such precious slivers of philological information as ONav. tresso and Alfonsine trexo,59 to conclude with the lapidary statement: ''TRAXI mußte lautgesetzlich trexe werden" (Span. Gramm., §31.18). Hanssen's dis54 In his Gramática histórica, §250, he credited trogo to León, troxo to Toledo, i.e., New Castile, and traxo to Old Castile; he listed trasco separately (§254), without bothering to localize it. 55 Anteckningar om folkmålet i en trakt af vestra Astuvien (Uppsala diss., 1887), p . 49. 56 Colección de poesías en dialecto asturiano (Oviedo, 1839), by an anonymous com piler, pp. 92 and 239. 57 Vocabulario de palabras y frases bables (Madrid, 1891-92), p . 119a, s.v. tría 'traía' (inf. trer). 58 Dialectos castellanos: montañés, vizcaíno, aragonés (Berlin, 1892), p. 9, with a refer ence to Pereda's Escenas cántabras (i.e., montañesas). I find no corroboration either in Pereda's own glossary appended to his novel Sotileza (1885), or in E . de Huidobro's pamphlet Palabras, giros y bellezas del lenguaje popular de la Montaña elevado por Pereda... (Santander, 1907), or, for t h a t m a t t e r , in J. González Campuzano's " A p u n t e s " revised by Huidobro (BBMP, I I [1920]); b u t a helpful thread leads to G. A. García-Lomas y García-Lomas, Estudio del dialecto popular montañés (San Sebastián, 1922), p. 24; id., El lenguaje popular de las montañas de Santander (Santander, 1949), p. lxiv, where (a) trij-e, -o, (b) trej-o, and () traj-ierin are p u t on record without any a t t e m p t at narrower localization. I t is not devoid of interest t h a t trej- is more characteristic of the 3d sg., esp. in light of C. Blaylock's conjecture; see the following page. 59 Fuero general de Navarra, eds. P . Harregui and S. Lapuerta (Pamplona, 1869), p. 57 (here indirectly cited) ; Primera Crónica General, ed. R. Menéndez Pidal (NBAE, V; Madrid, 1906), p. 55b35: " e t trexo consigo tres legiones".

108

CLUES AS TO DATING

covery did not pass entirely unnoticed but aroused, on balance, little curios ity; 60 as a result of this apathy, when J. J. Nunes drew attention to OPtg. (fut. subj.) treixer61 and when J. Leite de Vasconcelos, with superior precision,62 identified tre(i)x-/treis[s]- forms in a municipal 13th-c. text from Évora, where they alternated with tross- forms (closely related to familiar Western troux-) — a dependable sign that tre(i)x- was in full retreat at the given moment —, it apparently occurred to no one to consolidate the profusely scattered Old Portuguese, Old Spanish, Old Navarrese, modern Asturian, and modern Santanderino scraps of evidence into a single mosaic endowed with a meaning ful message.63 (The relation of trix- to trex- is not entirely transparent. One way of ac counting for it is to surmise, with Curtis Blaylock, that Alfonsine trexo is a Western form which cut loose from the metaphonic paradigm trixe, trex-iste, -o, etc.; in this eventuality the vowel change in TRAXÏ > trixe would represent the dual, two-stage effect of / k s / on / a / , and of word-final / l · / on the resultant nuclear / e / . This original state of affairs, Diego Catalán independently reminds me, seems to have crystallized in Lena: trix֊e, -iste, ֊emos, -esteis, -eron as against trexo, perhaps also in Alto Aller: trixi ~ trexo beside analog ically deflected trixi. Elsewhere the trend has been for one of the two stem variants to become generalized; hence trex-i, -o in Cabranes and trix-e, -iste, tr-ixo [beside ֊uxo], trix-emos, ֊estes, ֊enon [beside ֊iren] in the "bable" of Cabo Peñas, as recently investigated by  .  Diez Castañón. To the scant me dieval record one may add the fut. subj. trexiere in the Fuero de Brihuega, ed. J. Catalina García, p. 147. Cf. also fn. 68, below.) 3. A N ALLIANCE OF PERIPHERAL ZONES? Comparing these shreds of knowl edge, one inevitably notes the relegation of the "e/i preterite' ' to marginal 60 T h u s , A. Zauner, Altspanisches Elementarbuch 2 (Heidelberg, 1921), §118, recorded, without documentation, trexe as a var. of traxe. 61 Compendio de gramática histórica portuguesa: Fonética — morfologia (Lisbon, 1919), p. 343, fn. 1. 62 Textos arcaicosz, p. 37 — strictly speaking, in the wake of J. Cornu, "Die portu giesische Sprache", in G. Gröber's Grundriss2, I (Strassburg, 1904-06), 1076. The same 13th-c. text uses in close succession trexerem (lines 7, 9), treis[s]erem (line 11), and tros[s]essem (line 10). T h e editor operates with tre(i)xe < TRAXÏ as the starting point, without making it entirely clear whether he ascribes to metaphony the fronting of the central vowel (a gratuitous hypothesis, in my view, as against the wisdom of declaring e > i metaphonic). The symmetric reduction of ei to e and of ou to  represents a local twopronged evolutionary feature. The wavering between the e and the  preterite may point to a " g a p " between the original scrivener's speech habits and those of a later copyist, or to an unresolved conflict within a single idiolect torn by a dilemma through exposure to outside pressure; cf. my critique (Lg., X X X I [1955], 261-291) of M. Gorosch's grossly inadequate analysis of the similarly split language of El fuero de Teruel. 63 Let a single example suffice. One is admittedly safe in p u t t i n g to use bibliographic d a t a extracted from Ä. Rosenblat's tidy but superficial " N o t a s de morfología dia lectal", §219; see P a r t I I of his elaborate revision of A. M . Espinosa's Estudios sobre el español de Nuevo Méjico, BDHA, I I (1946), 270-273. Yet in P a r t I (1930) of the a d a p t a tion, for which A. Alonso and Rosenblat were jointly responsible, an editorial question mark (p. 80, fn. 1) discloses the fact t h a t this team of scholars either doubted the au thenticity of tre-, tri֊jo or were unable to account for them.


109

areas, as often as not in the hilly or mountainous terrain, off the main roads of traffic and diffusion if not always beyond easy accessibility, whether in the West (Alentejo), or in the North (Asturias, Montana de Santander), or in the East (Navarre). Where the sound sequence -AX- followed its "natural", unobstructed course, tending to produce -e(i)x- except at isolated points, the pattern of rapid disintegration or gradual recession has irreversibly asserted itself from the Middle Ages. Just what sealed the doom of the "regular'' forms, those organically transmitted? The only plausible reason for their ultimate rejection is morphological. While there was plainly nothing objectionable about trexo /trešo/ or, at a later stage, trejo /trexo/ on the phonological side, the grammatical mechanism pres. A/pret. E was unparalleled in Spanish and at best residual (fez < FĒCIT) — hence hardly capable of extension — in Portuguese after the large-scale injection into the paradigm of the initially rare metaphonic fiz-stem (OSp. fiz, fiz-ieste...; Ptg. fiz, fiz-este...). III. Paradigmatic Resistance to Sound Change: Old Spanish traxe. 1. T H E STATUS OF THIS VARIANT. If trexe is ruled to be the normal continuator of TRAXĪ, OSp. traxe plainly cannot be charged with performing that same rôle. Surprisingly, authorities in unison decline to specify the character of traxe — in part because at first glance it happens to LOOK normal, in part because, being the immediate ancestor of the modern standard form, it may have bored scholars in search of piquancy and exoticism. Traxe is best defined as an erratic form whose deviation from the norm is due to paradigmatic resistance to phonetic change: An older state has been here preserved in defiance of a sound shift which threatened to produce an isolated and, in the given context, undesirable conjugational model (A : E). Essentially the same phenomenon has been observed at close distance in the case of the old preterites vide Ί saw', vido 'he, she saw', which for centuries displayed an unshakably steady primary d while well-nigh all phonological analogues exhibited pro tracted vacillation between d and zero: OSp. cru(d)o 'raw', (d)esnu(d)o 'bare', ni(d)o 'nesť, su(d)o Ί am perspiring', etc.64 The -d- was so effectively protected in this niche, one argument runs, because, at the critical stage, a simple radical-final consonant, through a whim of circumstances, had become the expected marker of the strong preterite in Old Spanish; phonologically, the d of vid-, -o received a generous measure of support (a) within Castilian, from crudo, nido, sudo, etc. — initially, in all likelihood, a minority group, as against prevalent crúo, rúo, suo — and (b) from neighboring Navarro-Arago nese usage, where primary d, to begin with, was at no time endangered except through Castilian encroachment. By the same token, one is free to affirm that ֊ax- in traxe was kept or restored through the instrumentality of such 64 See "Paradigmatic Resistance to Sound Change: The Old Spanish Preterite Forms vide, vido Against the Background of the Recession of Primary -D-", Lg., XXXVI (1960), 281-346. Cf. the appraisal by K. D. Uitti, "Problems in Hispanic and Romance Lin guistics", HR, XXXIV (1966), pp. 242-255, at 247f.

110

CLUES AS TO DATING

residual Castilian doublets as axuvar/exoval or tax-/tex-ugo and, again, under the pressure of adjoining Navarro-Aragonese usage, where, on the evidence of such medieval forms as ax and max-iella, the normal outcome of -AX- was -ax-, in preference to -e(i)x-; see Section 1:3, above. Incidentally, the con comitant pressure of Aragonese here theoretically extrapolated would account for the fact that Portuguese fails to share in the restoration of trax- (a peculiar ity shrewdly noted by Dardel, but left unexplained) : The geographic distance from Aragon was here simply too forbidding for diffusion to assert itself. 2. T H E RECORD OF THIS VARIANT. In standard Spanish, tiraje has emerged triumphant from its long-drawn-out competition with numerous rivals, of which truje — to this day entrenched in untold dialects — proved to be by far the most formidable. Despite Nebrixa's support, 65 traxe did not score any uncontested victory in preclassical or Golden Age literature; 66 neither is it, at present, necessarily the commonest rustic form on either side of the Atlantic, though it is everywhere on the march, propelled by powerful socio-educational advantages rather than by any intrinsic, i.e., purely linguistic, superi ority.67 Specifically, trax- can be traced to certain varieties of contemporary Asturo-Leonese, as regards both literary usage and lay utterances elicited or overheard by field-workers.68 On the medieval level, trax- can be illustrated with Castilian,69 Castilianized "Western", 70 and Navarro-Aragonese71 texts — 65 See this Andalusian humanist's Gramática de la lengua castellana (Salamanca, 1492; ed. I . González-Llubera [1926]), Bk. v, Ch. 6, where  preterites are otherwise favored: cupe, plugue, pude puse, supe, tuve, uve; cf. Henríquez Ureña's comment in BDHA, IV (1938), 279. F o r t y years later, J u a n de Valdês, born in Cuenca (New Castile), withheld his personal endorsement, b u t granted: "Muchos cortesanos, cavalleros y señores dizen y escriven traxo". 66 See G. Ciro է, ' 'Quelques remarques sur les archaïsmes de M a r i a n a et la langue des prosateurs de son temps (Con ; ugaison)", RF, X X I I I (1907 = Mélanges Camille Chabaneau), 883-904, esp. 893-895. The author affords a welcome insight not only into the preferences of spokesmen for fine literature (including historiography: Mariana, Zurita, Gomara, Garibay, Moneada, etc.), b u t also into the didactic lines toed by tone-setting grammarians, language pedagogues, and lexicographers (C. Oudin, A. de Salazar, H . de Techeda, and others). 67 The recency of traje on the local scene, as a substitute for deeply rooted truje, has been observed b y A. C. Post, loc. cit. (see fn. 49, above). 68 Hanssen's 1897 monograph (p. 23) refers the reader to poems by Teodoro Cuesta. Menéndez Pidal, " E l dialecto leonés", §18.8, in RABMz, X I V (1906), 305, singles out, in a different context, Leon. (Carreño, Gozón) traxenon 'trajeron'. 69 Poema de Fernán González, ed. M a r d e n , 66c: traxyere; 287c: traxyst(e). Rimado de palacio, ed. Kuersteiner, traxo: N , 1406d ( = E , 1322d). Hanssen's 1897 monograph, pp. 13 and 15, supplies documentation from Archbishop Rodrigo's Estoria de los godos (Documentos inéditos, L X X X V I I I , 71) and from documents issued in King Alfonso the Learned's chancery {Memorial histórico, I -  ) , Nos. 5 and 104 (the spelling with obvi ously constitutes an indefensible modernization). 70 Dança general de la muerte, ed. . Appel (1902) : " C a traxo al mundo un solo b o c a d o " (vd) ; " e s t a mi dança traxe de p r e s e n t e " (ixo). Readings adopted by Ford, OSR, p p . 67f. 71 The characteristically E a s t e r n 14th-c. spelling traisso (with iss meanwhile pre sumably advanced to / š / , cf. Menéndez Pidal, Orígenes, §6:5) is authenticated b y J. F . Brutails, ed. Documents des archives de la Chambre des comptes de Navarre, 1196-1384 (Paris, 1890), N o . 98 (Pamplona, A.D. 1364). To the texts appearing in an eastern garb one m a y add the Libre dels tres reys d'Orient, which M. Alvar has now rebaptized Libro de la infancia y muerte de Jesús (Madrid, 1965): " E l que alguna cosa traxiesse / non ha haber que le valiesse" (vv. 102f. ) ; copied by the same 14th-c. scribe to whom we owe Apolonio and Maria Egipciaca.

RANGE OF VARIATION AS ACLUE TO DATING

111

including several MSS of Berceo's poems, all of which show, albeit in different degrees, gradual Castilianization of a Riojan (Navarrese) nucleus.72 IV. Innovative Formations. All types other than trexe (treisse) and traxe (traísse)—whether "strong", "weak", or "mixed" — have this in common that they presuppose a sharp break with the Latin and paleo-Romance tradi tion, faithfully preserved elsewhere throughout the rest of ROMANIA except where whole categories of preterites were shunted off or remodeled. The wisdom of separating this bulky residue into a smaller central-vowel series and a discernibly larger back-vowel series is rooted in geographic considera tions. The α-radicals are characteristic of the Navarro-Aragonese zone (which, we recall, likewise offered a fertile subsoil for traxe) and thus adjoin the terri tories of Catalan and other circum-Tyrrhenian languages and, beyond them, the remainder of ROMANIA. Conversely, the ou, , and  radicals are peculiar to the Atlantic Coast and adjacent territories and show Hispano-Romance, for once, with its back turned to the Mediterranean. 1. T H E CENTRAL-VOWEL SERIES. This weakly developed series comprises only two types, one "strong": trasque, the other "weak": traí. (a) T H E trasque T Y P E . Here is a rare variant alternating in a few rival manuscripts with traxe, significantly to the exclusion of trexe, troxe, and trogue. In particular, it is characteristic of two Berceo texts reflecting directly or ob liquely the 13th-c. "manuscrito en cuarto" known for its relative faithfulness to the original: the Milagros, as transmitted in MS I (but not A), and Santo Do mingo, as accessible through MS H (but not E) or the Vergara text (1736); it also occurs in rhyme in both versions of the Alexandre — a matter of potenti ally pivotal import for the disputed genealogy of these branches —, and in the Liber Regum, traceable, significantly enough, to Navarra del Ebro, ca. 1200 : Omne desanetaa vida que irasco grand cordura {Milagros, I, 49c; this quatrain is absent from A); dixo Peidro: "En vida trasqui grand avaricia" (ibid., I, 250a; A: traxi); el cuerpo, el que irasco esta alma consigo (ibid., 276c; A: iraxo) ; erramos duramiente, grand locura trasquiemos (ibid., I, 392b; A: traxiemos); que trasquiessen el ninno del mont a los poblados (ibid., I, 576c; A: traxiessen); mandó el sancto Padre que irasquiessen del ujno (Santo Domingo, V, 307a; J. D. Fitz-Gerald emends to traxiessen)՛, los pannos que trasquieron, esos mismos levaron (ibid., 484c, in Fitz-Gerald's reconstruction; MS E: traxieron; H: trasquieran); mas gualardon malo priso del lazerio que trasco (Alexandre, 1374d, O; P : mas mal gualardon priso...); et estidieron en Egipto cxx annos, tro que los en trasco Moysén (Liber Regum, .17).

On account of their outlandishness these forms promptly aroused the cu72 1 can cite traxo from San Millan, 435c (confirmed by G. Koberstein in his critical ed. [1964] ; misspelled trajo by F. Janer) ; maltraxo from Santo Domingo, 6866 (the quatrain is absent from MS H; Gassner's remark [§419] to the effect that TRACTĀEE underlies this form is an irrelevancy suggested by Fr. maltraiter, see fn. 14, 7above); traxiestes from Santo Domingo, 280a (H: trasistes); traxieron from Loores, 108a , and from Santa Oria, 99d (MSS A, J) — a passage not affected by M. R. Lida de Malkiel's emendations (RPh., X, 19-33).

112

CLUES AS TO DATING

riosity of medievalists. In 1897, Gassner (§420) spoke of "wahrscheinlich halbgelehrte, starke Bildungen". Baist, who as early as 1888 had classed trasque with nasque and visque, went one step further in 1906, tracing all three erratic preterites to the ecclesiastic conduit ("aus der Kirchensprache stam men"). Shortly afterward Menéndez Pidal skillfully grouped visque and nasque, in this order, with Old French and Old Provençal counterparts and amply documented them;73 he assumed learned metathesis of -KS- and the accentuation vesquit in clerical Gallo-Latin; most important, he demonstrated the growing vitality of this limited pattern by proving through internal analysis that the copyist of the Cantar exceeded the poet in its use; but he stopped short of wondering whether visque and nasque might not be interpreted as Gallicisms introduced in the wake of the Cluny Reform, as were indis putably fray, frey 'friar' for earlier native frade < FRATER and as was monge 'monk', for indigenous mόnago < MONACU; nor did he evince any surprise at the consistent avoidance of the preterite of traer by the author of the epic he knew so intimately. Dardel, as recently as 1958, saw in nasque a pan-Romanic formation of old standing, declaring visque and trasque local extensions of this prototype. But, granted that the semantic link between nasque 'he was born' and visque 'he lived' is obvious, it must be objected that trasque shows not the slightest affinity to them along the referential axis; also, while it is true that it occurs in hagiographie texts, it lacks by itself any ecclesiastic or so much as religious overtone. The fog surrounding trasque, trasco can be dispelled once we recall that Berceo's native district, the Riojano, formed part of Old Navarrese, a variety of Hispano-Romance where the "regular" development TRAXIT > tresso left isolated traces and where, independently, groups of speakers also experimented with traisso = tra (undeniably, a symptom of confusion); and once we further remember that one major branch of the Alexandre, which some experts view as closest to the original, was couched in Old Aragonese, hence akin to texts composed in Riojano. The somewhat forced association of trasque with nasque (here the rhyme, at least, supplies a redeeming link) and, a fortiori, with visque loses much of its oddity if we interpret trasque as another ephemeral attempt by a small speech community — an attempt abetted by the cultural prestige surrounding presumably imported nasque and visque — to elude, within the frame of traer, the morphologically unattractive á : β alternation in the relation of present tense to preterite — unattractive because it repre sented an inopportune enrichment of resources amid a trend toward leveling and frugality. Like traxe and, as we shall soon hear, troxe, the form trasque makes sense if analyzed as another tentative escape from an irritating con straint. The metathesis /trakse/ > /traske/ in the pronunciation of Church 73 Ed. Cantar de Mio Cid; texto, gramática y vocabulario (Madrid, 1908-11), pp. 279f.; note one minor correction in the rev. ed. (1944-46), p. 280, starred fn.


113

Latin may have been a concomitant, supplying the model but hardly explain ing its eager acceptance. (Diego Catalán obligingly draws my attention to two other stray vestiges of the spread of the /sk/ preterite : fusco in lieu of fuxo or fuyó 'he fled' appears in the Liber Regum, see BRAE, VI [1919], 195 and 200; and remasco, in preference to remaso 'he remained' troxo; A. Alonso and Ā. R o senblat, BDHA, I (1930), 80; Huber, Altportugiesisches Elementarbuch, §402; Williams, From Latin to Portuguese, §§92.9 and 200.6; G. B . Pellegrini, Grammatica storica spagnola (Bari [1950]), p p . 192 and 194. 97 N o t e t h a t trogo and troxo were found in neighboring territories — e.g., within the Old Leonese zone, the former in the City of León, the l a t t e r in Sahagún — and t h a t , judg ing from the jumble of manuscript readings, there m u s t have been a liberal margin of actual overlap; cf. fm. 115, below.


119

secondarily been transferred to the system of TRAHERE. 98 If we disregard this scintillating but unproven and semantically vulnerable "Einfall", there remains the question whether it is defensible to project the prototype of tro(u)xe, etc. onto the level of Latin — however far one goes in stretching the limits of that label. Why should *TRAXUĪ have arisen — and thrived — in certain sections of Roman Spain, but have emerged, still less struck root, neither in Gaul nor in Italy? The old-time experts did not bother to raise, let alone to answer this — to us — rather obvious, not to say decisive, question. Now, if we argue that tro(u)xe sprouted in early Romance rather than in Late Antiquity, through a "switch'' from tre(i)xe preëminently in areas favoring the inf. trazer (i.e., in territories offering the immediate model OPtg. jazer ~ jougue, prazer ~ prougue, also saber ~ soube, etc.; cf. OSp. yazer ~ yogue, plazer ~ plogue, also saber ~ sope, etc.), then a reason for such a "switch" becomes at long last apparent: It occurred as an alternative to the threatening change to phonologically regular tre(i)xe (or trixe), a shift which would have almost uniquely pitted preterital -(i)- (or -i-) against infinitival and presenttense -a-. What we are witnessing, then, is an attempt to escape — through the safety valve of analogy—from morphological isolation looming as a punishment for blind obedience to sound laws. Examples of the older texts' preference for the trox- type and its closest variants are so numerous as to defy any compact inventorying. One could break down the corpus by using as the prime classifier either the area (assigning to each zone the relevant subtypes) or the variant itself (specifying under each item the regions over which it temporarily spread). For operational reasons we shall here elect the latter procedure, (α) Trox- shared with trogu- a position of preminence in Leonese, probably surpassing it by a slight mar gin; but, unlike its rival, it also proved its appeal to Castillans. On the side of its Western supremacy (which extended a timid prong into Old Galician-Portuguese99) one may cite its occurrence, throughout the preterite and allied tenses, in the Leonese MS  of Alexandre (where it alternates with trogu-, while the Aragonese MS  adopts the "weak" paradigm; but trasco straddles both major versions); in a 13th-c. charter traceable to Sahagún and in other Western documents; in the Fuero de Ledesma, in the Poema de 98 Dardel fails to take cognizance of my article "Antiguo español y gallegoportugués trocir 'to pass' ", NRFH, Χ (1956), 385-395. If I were to rewrite that note, I would pay heightened attention to the interaction of OSp. trocir and OPtg. deçer, OSp. deçir 'to descend' < DISCĒDERE 'to go away, depart', a factor which might account for the trouble some appearance of ֊ҫ- in lieu of expected -z the single most serious obstacle to the acceptance of the etymon TRĀDǕCERE. Corominas' Hispano-Gothic alternative base *TRUCCIRE, from *THRUKKJAN (1957), an ad hoc concoction in which Dardel seems to take stock, has a disquieting admixture of free-wheeling hypothesis and had best be discarded. The idea of a given word's transfer from one lexical family to another is, in itself, unex ceptionable; for a parallel see my note "La etimología de cansino", NRFH, II (1948), 186-194, and the recapitulation of the analysis in "Etymology and the Structure of Word Families", Word, X (1954), 265-274. 99 The entry troxe 'trató' in the Marqués de Valmar's Glossary (p. 7866) to the Acad emy edition of the Alfonsine Cantigas is due to some misunderstanding. The item, ex tracted from No. 64.13 ("Mais a dona troxe peor ca un can" ; I.j .1 : "Mais la dona a trouxe peyor que a un can") should have been consolidated with trager (mal), see p. 783a. The normal form of the preterite in this milieu is, incidentally, troux-(e) (No. 58.9).

120

CLUES AS TO DATING

Alfonso XI;10Q in the comparatively late Salamancan MS S of Juan Ruiz (while the older MS G wavers between trax- and trox-) ; and, a century before Columbus, in Seville. Char acteristic of the Central Peninsular use of trox֊ are — among other specimens — miscel laneous samples of Alfonsine prose, representative versions of Don Juan Manuel's writings, and, half a century later, MS A of Pero López de Ayala's Rimado de palacio 101 (β) The more advanced variant truj- soared to prominence in the late 15th c., 102 in the company of sope → supe, tove → tuve, etc., 103 and remained abundantly represented well into the 17th c , without any noticeable taint of rusticity. 1 0 4 On the present-day dialect level it is solidly entrenched even in the E a s t (Murcia, Lower Aragon), where it con s t i t u t e s , in all likelihood, a fairly recent deposit, to say nothing of its continued hold on such Western or Western-style dialects as modern Asturian, Leonese (Lumbrales, Robleda de Sierra de G a t a ) , Castilian proper (Cuellar in Segovia), and, above all, overseas Span100 To s t a r t out with the poetic texts, cf. Alexandre, ed. Willis, 450c (O): troxieron ( = P : trayeron); 822c (O): troxiera ( = P : llevava); 971c (O) : troxier ( = P : trayere); 1170b (): troxo ( = Ρ , 1170: rrogó); 1503e (O): troxo ( = Ρ : traxo); as against 1449c (  ) : troguyemos ( = Ρ : trayemos); also, 1004b (  ) : retroxol ( = gap in Ρ) — Hanssen, in 1897, here recognized a Leonese strain; Ruiz, 23c (S): troxo (G: traxo); 117c: troxiese; 223d: troxo; 773b: troxieron; 779a: t[r]oxo; 783b: troxistes; 900c: troxieron ( = G ) ; Poema de Al fonso XI, 947a: troxiesen; 1800a: troxieron; Rimado de palacio, ed. Kuersteiner, A, 509d: troxe; 636a: troxiste. Note J. Corominas' embarrassingly hesitant statement concerning the prevalent usage in mid-14th-c. New Castile, in his critical edition of the Libro de buen amor (Madrid, 1967), p. 80b:

En el pretérito del verbo traer aparece en G por lo menos cinco veces la forma trax, traxo y sólo una vez troxieron՛, en S hay 6 de trox- por 2 de truxo y 2 de trax (T sólo da un troxeron). Una forma tan moderna como truxo no per teneció desde luego al Arcipreste, y la oposición entre S y el ms. de Castilla G revela que la forma en a predominaba en aquel tiempo en el ambiente del Arcipreste, y por lo tanto, le doy la preferencia, salvo allí donde no figura en ningún ms., pues no es improbable que troxo no fuese ajeno al uso castellano del siglo XIV.

Interestingly, it is in addressing a mountain girl t h a t the Libro de buen amor's protagonist uses trax (G, S, 1039c). As for prose passages, cf. the above-cited Sahagún charter edited b y E. Staaff; Fuero de Ledesma, §395: "...o carne de puerco troxier p o r s u r o s t r o " in the Castro-Onis collection (1916) ; independently, OLeon. troxier is adduced in H a n s sen's 1897 monograph (p. 20) from T. Muñoz y Romero's collection. The Sevillian text, dated 1403 and edited by M. Gaibrois de Ballesteros, pairs off veno and tróxole; see M. R. Lida de Malkiel and Renée Kahane, " D o ñ a Angelina de Grecia", NRFH, X I V (1960), 89. 101 Documents from Alfonso X ' s chancery (Memorial hist., I—II) were shown by H a n s sen to contain troxier e(n) alongside more advanced frux- forms, undoubtedly due to " r e mozamiento", and a sprinkling of trax- forms, the whole further complicated through such tampering as the intermittent modernizing substitution of j for x (see 1897 monograph, p. 15); the same scholar cites troxo from the Estoria de los godos (ibid., p . 13). Trox- is typical of Alfonsine historiography, though it never entirely crowded out rivals, includ ing, we recall, archaic trex-; for one example see Primera Crónica General, ed. Menéndez Pidal, p . 10b: "de las diez naves que el troxiera dexara la u n a . . . " . As early as 1897 Ford drew attention to trox-, -iesse in the 15th-c. MS of D o n J u a n Manuel's Libro del cavallero e del escudero (ed. S. Gräfenberg, RF, V I I , 462.20 and 481.6). 102 Truxeron in J a n e r ' s ed. of Fernán González, 584d — uncritically echoed by A. Gassner in 1897 — is now known to involve a misreading of traxeron, which the rhyme p a t t e r n , in any event, forced .  M a r d e n (1904) to replace b y llevaron. 103 Less t h a n illuminating, for our purposes at least, is R. K. Spaulding's note " O n the Introduction of the Preterites in  {Hubo and its Congeners)", HR, I (1933), 161-167; the s t r a y remarks on plugue have some bearing on trugue and, indirectly, on truxe. 104 Here are a few milestones: "Así que me truxiste alivio para el padecer" (Diego de San Pedro, Cárcel de amor [Seville, 1492; B H , X V : Barcelona-Madrid, 1904], p . 70 — be side estuvo, pluguiera, sostuvo, tuv-iera and -iste, e t c . ) ; trux֊o, -eron in J u a n de Valdés' translation of St. Paul's Epistles to the Romans and to the Corinthians, corroborating his s t a n d in the Diálogo (cf. E . Cotarelo, "¿Quién fue el autor del Diálogo de la lengua?", BRAE, V I I [1920], 288); trux-eren, -o (El laberinto amoroso [Barcelona, 1618], ed. . Vollmöller, RF, VI, 3.26 and 46.12,14, as noted by Ford in 1897) ; "el vicioso proceder / de las mocedades mías / trujo el castigo y los días / de mi t o r m e n t o . . . " (Lope, El castigo sin venganza, I I I , 10; for a literary and stylistic appreciation of this passage see M . R. Lida de Malkiel, La originalidad artística de "La Celestina" [Buenos Aires, 1962], p . 339).


121

ish. 105 The — hardly frequent — infiltration of trux- into late Old Portuguese texts represents a neatly detachable Castilianism. 106 (7) Troux- was the radical underlying the standard forms of the Old Portuguese can cioneiros, whose authors, for esthetic reasons t h a t elude us, eschewed troug(u)֊ and trouv-.107 So far as spelling is concerned, troux- to this day dominates the Portuguese scene, except t h a t the current pronunciation of  as / s / , patently erratic, betrays an irresistible outside pressure; influence of dissel (δ) OPtg. troixe, an uncommon variant despite t h e widespread alternation of ou and oi in other sections of the lexicon (clous ~ dois 'two', ouço ~ oiço 'I hear', ouro ~ oiro 'gold', etc.), was peculiar to the E storia troyãa equally known for t h e preference it accorded oyve, poide, soive over corresponding -ou- forms, etymologic ally vindicated. 1 0 9 I n some Portuguese dialects troixe is still in use, flanked by truixe.110

Interestingly, this type — specifically its variant trouxe modified by the impact of disse — won out in Portuguese (where it faced the competition of treixe, trougue, and trouve), but not in Spanish where, at least in literary usage, truje was blocked and eventually overrun by traje. (We recall that the cradle of the type most plausibly lay in the West, where trazer temptingly rhymes with jazer and prazer.) In dialects, both Peninsular and ultramarine, the rivalry between the a and the  variant to this day continues unabated, except that, by virtue of the elevation of traj- to the preferential rank of a standard form, the contrast traje : truje has ceased, especially in the New World, to 105 Some shafts of light are thrown on the situation in the SE by A. Sevilla, Vocabula rio murciano (Murcia, 1919), p . 197: trujo (beside 1st sg. pres. ind. trayo, impf, traiba), and b y J. García Soriano, Vocabulario del dialecto murciano (Madrid, 1932), p p . xcviii and 128b: (rust.) truj-e, -iste, -iera (on the latter ending see fn. 49, above). Interestingly, anduve has failed to overlay andé in this area. I n a territory reminiscent of Segorbe (as known through C. Torres Fornés' pioneering s t u d y of 1903), where Valencian, Aragonese, and Castilian a b u t ca. 50 km. to t h e N W of Valencia, trujo has been found and perhaps hastily identified as a "Castilian a r c h a i s m " : see V. Llatas, El habla del Villar del Arzo bispo y su comarca (Valencia, 1959), I, 71f., accessible to me through J. Gulsoy's digest in RPh., X I V (1960-61), 350f. For rural Castile I can adduce the testimony of A. de la Torre, "Έ1 habla de Cuellar (Segovia)", BRAE, X X X I (1951), 141: hubon 'hubieron', trujon 'trajeron' ; and Menéndez Pidal's careful localization of vin-ioren, truj-ioren ("El dialecto leonés", §18.8) ; Hanssen's 1897 monograph (p. 23) supplies miscellaneous Asturian data, including truxi from M u n t h e , truxo from the 1839 anthology, and truxeron from R a t o y Hevia. 106 One finds some d a t a on trux-o —■ with an -0 t h a t betrays its Leonese or Castilian background — as well as on troux-isti, -estes, and -e in Huber, Altportugiesisches Elemen tarbuch, §402:3. 107 Cf. the above-quoted t r e a t m e n t s b y C. Michaelis de Vasconcelos and Huber. Trouxalso prevailed in Old Galician poetry and prose alike, witness the Alfonsine Cantigas and the Miragres, ed. J. L. Pensado (see p . 357, s.v. trager). 108 E . . Williams, loc. cit., mentions dixe as an OPtg. variant of disse, wondering whether they could have been initially polarized like pus and pôs, fiz and fêz, etc. M y own feeling is t h a t in DĪKSĪ > /dijsi/ the / j / element could have been easily absorbed b y the preceding homorganic vowel (cf. FRĪCTTJ 'fried' > Sp. frito rather t h a n *fricho), and t h a t this circumstance, more t h a n any other factor, accounts for the contrast between the older trouxe, with / š / , and disse. Through later attraction trouxe succumbed to the spell of -sse. On the prevalence of fiz- and pud- see m y article on "Diachronic Hypercharacterization" (Part I ) , ArL, I X (1957), 79-113, at p.107, with a reference to H . Meier, Ensaios de filologia românica (Lisbon, 1948), p p . 31-54. 109 See Huber, Elementarbuch, §402:3. 110 Leite de Vasconcelos, Esquisse, p . 141: trou-, troi-, and trui֊xe exist side by side in the South (Beira).

122

CLUES AS TO DATING

reflect a clash between regional dialects and, as a rule, embodies a social confrontation of imposed literate with spontaneous rustic usage.111 (e) MIXED TYPES. T O round out this picture, it is fitting to mention a dialectal "mixed" form, involving a "weak" preterite grafted on the truj- stem, e.g., And. truji, as observed by H. Schuchardt in his pioneering essay,112 and Arag. brujió beside trujo, both recorded by P. de Mugica, Dialectos..., p. 79; in 1914, García de Diego specifically attested this paradigm's logically inferrable counterpart, substandard trajió. As another bundle of side-lines one may identify the sporadic extensions of truj- to sections of the paradigm other than the traditionally interconnected tenses in -ra, ֊re, and -se, whether additional tenses, including the imperfect (Extr. trujian), or non-finite members of the system, such as the gerund: Trujiendo has left traces in Santanderino (witness García-Lomas), in Eastern Leonese (at Cespedosa de Tormes), in Salamancan (as observed more than a half-century ago by J. de Lamano y Beneite), and again in Extremeño. An early model was Juan Ruiz's toviendo for 'teniendo' (S, 1390α).113 V. Patterns of Wavering. The roster of forms here paraded is incomplete (it omits, e.g., presumed scribal slips, such as trasxiera, a hapax legomenon114). But one circumstance must not remain unmentioned even in a cursory sur vey: the fact that the same medieval text, or league of closely allied texts, with conspicuous frequency exhibits wavering between rival variants — say, trox- and trogu-,115 or trax- and trox-,116 or trox- and trux- (truj-)117 or trasqu- and trax-118 or trouv- and troux-119 111 For a preliminary array of raw data see BDHA, II (1946), 270-273. 112 "Die Cantes flamencos", ZRPh, V (1881), 249-322, at 321, with a reference to E. Lafuente y Alcántara, Cancionero popular (Madrid, 1865), II, 371.7. 113 Sal. trujiendo alongside fuendo, hubiendo, hiciendo, supiendo has rigidified into an "invariable" pattern, on the authority of Lamano y Beneite (p. 61). García Soriano does not expressly implicate traer, but his mention (p. xlviii) of dijiendo, entretuviendo, hu biendo, and — again — fuendo (alongside endo, indo 'yendo') would make the local flower ing of trujiendo a safe guess. Fuendo figures in an even earlier dialect sketch: S. Alonso Garrote, El dialecto vulgar leonés hablado en Maragaterîa y tierra de Astorga (Astorga, 1909), p. 182 (2d ed., Madrid, 1947, p. 239), and it is this frame of reference which M. L. Wagner used in ID, XV (1939), 19, on the occasion of his Sardinian explorations. 114 Thus, Hanssen's 1897 monograph contains a clue (p. 13) to unparalleled trasxiera — a lapsus calami? — extracted from the Estoria de los godos, p. 9. Note, however, the occasional use of the digraphs sz, sç for / š / ; see Menéndez Pidal, Orígenes, §6:6. 116 Thus, in the fragmentary MS E (= Esc. II-M-18) of the Fuero de Zamora, §11, one finds in close succession: "Et se uno dixier al otro : «Mays omnes trouxieste que non ouieste a traer...», jure por su cabeça que mays omnes non trogo" {Fueros leoneses, eds. Castro and116Onís, pp. 19f.). Cf. fn. 123, below. As Hanssen (p. 17) and Ford simultaneously observed seventy years ago, Don Juan Manuel's Libro de la caza, ed. G. Baist (Halle, 1880), contains such forms, mutually ir reconcilable at first glance, as (a) trax-iere (11.2), -o (40.6) and (b) trox-iere (25.13, ¿7.17, 32.18, 67.8). Leomarte's Sumas de historia troyana, ed. A. Rey (Madrid, 1932), includes, the Glossary affirms (p. 4156), traxieron but troxistes; unfortunately, both page references supplied by the editor are grossly inaccurate. The 15th-c. MS of the Lapidario made available by K. Vollmöller (Ein spanisches Steinbuch [Heilbronn, 1880]; here cited via Ford's dissertation) is hospitable to both trax֊o (23.26, 28.5), -iere (28.15, 29.24, 30.21, 24) and trox-iere (28.2). The 14th-c. version of the "Visión de Filiberto", ed. . de Toledo (ZRPh, II [1878], 40-60), tolerates, as Ford observed, an isolated instance of trax-iste (55.38) pitted against trox-iste (55.39), -iese (53.32), and -iesen (53.28). "La estoria del Rey Anemur e de Josaphat e de Ballaam", ed. F. Lauchert (from a 15th-c. MS), RF, VII


123

The keenest theorist among this century's dialect geographers, K. Jaberg, has categorized areas typologically, starting from their configuration on a linguistic map, as hazily or sharply contoured, discrete or overlapping, etc.120 Though our own material here is too fragmentary, and our dossier on the medieval writers concerned and on their copyists (most of them, in any event, anonymous) too lacunary, for any formal cartographic approach, our mental map of the Iberian peninsula arising from the analysis of the data so far digested unmistakably points in the direction of overlapping zones. This peculiar configuration of territories, in turn, supports our initial assump tion that clusters of speakers were each groping for a suitable solution — al most any solution — that might help them overcome a morphological distress or emergency. I T FURTHER BUTTRESSES THE CONCOMITANT HYPOTHESIS, RICHER IN IMPLICATIONS, THAT THE PROCESS OBSERVED WAS STILL IN A STATE OF FLUX; IN OTHER WORDS, THAT THE VERY SCOPE OF VARIATION YIELDS AN OBLIQUE CLUE TO RECENCY.

VI. Avoidance of the Preterite of traer. One ingredient of the situation hitherto, it would seem, completely overlooked is the fact that speakers be wildered by a multiplicity of variants vying for their attention might have been searching — to be sure, subliminally rather than consciously — for some convenient way to circumvent their impasse. We know from Gilliéron's masterly dialect studies that puzzled patois speakers, under comparably dis turbing circumstances, have grasped for unequivocally correct Parisian words, to the lasting detriment of regionalisms.121 Some Spanish "cultismos" (e.g., dulce 'sweeť) might, in the last analysis, owe their introduction to a longdrawn-out, inconclusive rivalry between equally appealing vernacular coun terparts (doz, duz, doce, duce, etc.). In our own case, some speakers of Old (1893), 331-402, is sharply split in its preferences between (a) trax-ieren (383.28), -iese (364.35, 374.44), ֊ (377.11) and (b) trox-ieron (373.23, 393.36), -iste (380.23). I n the speech of the prelate, the Rimado de palacio, 509d, shows a cleavage between MS A ("que troxe d'Aviñón") and M S E ("que traxe de Aviñón"). To this day traj֊ and truj- also compete, we recall, on the dialect level almost everywhere, e.g. in Salamanca (Lamano y Beneite, El dialecto vulgar..., p . 62). 117 See fn. 123, below, for one example from the Primera Crónica General. 118 Recall the neatly silhouetted discrepancy between MSS I and A of Berceo's Mila gros. 119 The Demanda do Santo Graal tolerates, side by side, trouv-e (fol. 9c), -er (fol. 169α) and troux-e (fol. 10c), -erom (fols. 12b, 14c); see I I I , 399. Add the d a t a included in fn. 85, above. I owe to John M . M c C a r t n e y ' s courtesy the timely reminder t h a t dialectal Brazil ian Portuguese still uses tr(0)uv-and tr(o)ux- (with χ = / š / ) in competition, the upshot of the local development being t h a t trouve and trouxe correspond to the 3d sg. only, while truv- and trux- are peculiar to all other persons, the 1st sg. included. See M . Marroquim, A lingua do Nordeste (Alagoas e Pernambuco) (São Paulo, 1945), p . 128, and, less reliably, A. Nascentes, Ο linguajar carioca (Rio de Janeiro, 1953), p. 106. Elsewhere (p. 104), the l a t t e r philologist also opposes dial. (1) sube to (3) soube. The models were, of course, fiz : fèz, pus : pôs, pude : pôde. I n the extreme South (Rio Grande do Sul), M c C a r t n e y ' s own field-work has netted the form (1) trusse. Clearly then, both / s / and / š / "realiza t i o n s " of the spelling trouxe are observable in far-flung sections of modern rural Brazil. 120 See his Aspects géographiques du langage (Paris, 1936), p p . 43f. 121 Gilliéron and Roques, Études de géographie linguistique..., p . 13; also p. 122 and M a p 12, showing patches of coq~ squeezed in between areas of G C A L L U S , P U L L U S , and aisan.

124

CLUES AS TO DATING

Spanish who may have been in the grip of an acute embarrassment amid the welter of overlapping, interlocking possibilities open to them: traxe, trasque, trai, etc., could very well have preferred jettisoning altogether the capricious preterite of traer and using instead — in this particular tense and in its ap pendages alone on such a sweeping scale — the more easily manageable forms of a synonym (say, aduzir). This is exactly what seems to have occurred in the Cantar de Mio Cid, from whose lexicon traer, spectacularly enough in view of the dependence of an epic on narrative mechanism, is absent in, of all tenses, the critical preterite and its satellites.122 It is known from the meticu lous inspection of variant readings in Old Spanish texts that aduzir and traer were, in fact, to a large extent freely interchangeable.123 VII. Conclusion. In drawing the balance-sheet, one may state that the phonologically expected descendant of TRAXÏ, namely Ptg. tre{i)xe, treisse and Sp. trexe, tresse, is only vestigially observable in the Middle Ages and is found in modern dialects even more sparingly (in part disguised as trije, clearly under the impact of dije, though the original stimulus for e > i in the 1st sg. may have been metaphonic). Morphological pressure—the speakers' reluctance to sanction as a new grammatical device the alternation a (pres. and infin.) : (i) (pret.) under a regime of grammatical austerity — accounts for the emergence of numerous substitute solutions, which amount to so many eva sions; none of these solutions, consistently enough, involves experimentation with an entirely new pattern. The departure from the foreseeable norm may be in the direction of conservatism (faithfulness to ֊ax-, as in traxe, counter to what one is led to expect from one's familiarity with the reflexes of AXE, MATAXA, MAXILLA, and LAXARE) or in the direction of bold analogical games, conducive either to ou/o/u forms, chiefly in the West: trouxe (trousse), trogue, trouve, or to a forms, "strong" or "weak", "vernacular" or "semilearned", chiefly in the East: trasque, trai. In the end, the heartland of the Peninsula became the battleground between the two most successful contenders: traxe and truxe՛, the conflict has also been carried into Spain's erstwhile overseas possessions where, even more than in metropolitan territory, it has of late as sumed the character of a confrontation of socio-educational rather than strictly regional traditions. On the dialect level, Portuguese still tolerates several older contenders for supremacy (in fact, it has enriched their roster); in the standard language, the spelling eventually adopted, with  (elsewhere indica tive of / s / ) , and the pronunciation now consistently favored except in certain 122 See Menéndez Pidal's monumental edition (1908-11 and 1944-46), p . 430 (s.v. aduzir 'traer una cosa, una persona', including aduxier, v. 181), and p. 870 (s.v. traer). For independent observations and inferences on this point see T . Montgomery, " N a r r a tive Tense Preferences in the Cantar de Mío Cid", RPh., X X I , 253-274, at pp. 255 and 260. 123 T h u s §11 of the Fuero de Zamora reads in MS Q: ' Έ se uno dixier al otro: «Mays omnes aduxieste que non ouiste aduzir a este plazio», jurele por su cabesça que mays omes non troxo que los que deuiera atraer" (see the Castro-Onís collection, p p . 19f.; for a different version of this passage, cf. fn. 115, above). I n his 1897 monograph (p. 16) Hanssen cites telling passages from the Primera Crónica General, as available at t h a t j u n c t u r e in Menéndez Pidal's La leyenda de los Infantes de Lara (Madrid, 1896): troxiste (p. 233), troximos (vars. traximos; E : aduxiemos—p. 234); troxieron (vars. aduxieron, tra- and tru-xieron —■ p. 237). For one additional example see fn. 78, above.


125

islets of rural speech, namely with / s / , seem to be at loggerheads. May one argue that the merger of trouxe and trousse, plus the pressure relentlessly exerted by disse} could have cumulatively generated sufficient force to explain the shift / s / > / s / , left unreflected in official orthography? Such, in rough outline, are the facts as they primarily concern the Romance scholar. These bare data, for all their intrinsic value to the specialist, conceal a potentially far more memorable lesson for the general linguist. In reverting to our initial question: WHAT COULD POSSIBLY HAVE SET OFF THE RASH OF HlSPANO-RoMANCE VARIANTS IN A VERB WHOSE PRETERITE IN COGNATE LANGUAGES SHOWS A QUIET, UNEVENTFUL DEVELOPMENT?, w e Can n o w s t a t e

with greater assurance that nothing—neither the evidence of the parent language, nor the comparative analysis of congeners, still less the structural dissection of the Hispanic corpus—justifies the projection of a multitude of rival bases onto the plateau of provincial Latin. The speech community's ceaseless, almost frantic search for replacements cannot be fortuitous; it is understandable solely on the assumption of a kind of emergency or crisis, which might be labeled, in Gilliéronesque terms, "détresse (gêne) morpho logique" The tension arose when the multi-pronged development of the Latin sequence -ax- / a k s / threatened (and in part managed) to create a unique and, in the given context, obviously unwanted alternation tra(z)-, trağ-/tre(i)x~. In many parts of the Peninsula speakers spontaneously rebelled against this supervenient complication (which ran afoul of a general trend toward inflectional leveling) ; they fell back — let me repeat — on analogy in their unconscious search for avenues of escape, or retained the original vowel of -ax֊ beyond the predictable spatio-temporal norm, or else shunned alto gether this verb in certain vulnerable tenses. Now the crest, i.e., the critical stage in the diffraction of /ajs/ < -AX- into / a š / , /eš/, etc. falls, as Menéndez Pidal and others have independently established in sifting the record, into the period 900-1300 A.D. rather than within the bounds of regionally tinted Latin. The impressive range of variants, especially in medieval texts (and, at later stages, only in outspokenly conservative dialects), harmonizes with this dating because—the combined agencies of rivalry, selection, and attrition being what they are — the coexistence of so many competing variants as early as, say, 400 A.D. would incontrovertibly have given way to sharper tightening eight centuries later, when the curtain shielding the vernaculars went up at long last. One senses, then, in this concrete ensemble of circumstances at least, a direct correlation between the profusion of variants and the compara tive recency of the process that must have set them in motion. In analogous instances, where there is separately available no immediate clue to the prime movers of erratic inflectional proliferation, it will conceivably be henceforth permissible to advance one step further and to extrapolate, even without the privilege of philological underpinning, a recent approximate date from a contextually striking range of variation.

D. CONVERGENCE AND DIVERGENCE

FACTORS IN THE UNITY OF ROMANIA

A bold, imaginative synthesis of Romance cultures has, for generations, been the dream of advanced humanistic scholarship. One such inspired dreamer, at the turn of the century, was that unforgettable Strassburg professor of Romance philology, Gustav Grober, to whom we owe the founding of the most influential learned journal in our field, as well as the launching of the most authoritative encyclopaedia, if that is an adequate translation of Grundriss. There is hardly any need to stress the fact that Grober was an accomplished practitioner of both historico-comparative lin guistics and history of literature. Less well known is the anecdote that in the original design of the Grundriss Romance art history and related disciplines were also to be represented, to round out the many-splendored panorama of Romance culture. Hierarchically, in the structure of his journal and of his encyclopaedia alike, Grober granted pride of place to linguistics. Since prejudice or emotional preference is inconceivable in his particular case, we may well ask ourselves — to set the discussion in motion — what prompted Grober to adopt this scale of values. Romance linguistics has a priority over Romance literature in two vital respects. Spatially, the linguistic experts among us can study the remnants and the indirect reflexes of Latin as spoken in Celtic Britain, in pre-Arabic North Africa, along the eastern coast of the Adriatic, as well as in the Rhine and Danube valleys. Cologne, Munich, Vienna, Dubrovnik-Ragusa, and Tunis are to the Romance linguist cities with a Latin past. Few literary scholars would care to join us in the quest for the retrieval of this kind of Latinity. — Temporally, we are concerned with an impressive segment of over two thousand years, from Ennius and Plautus all the way to the late twentieth century. The oldest vernacular texts go back to the 9th century; the oldest samples of vernacular poetry — the Cantilene de Sainte Eulalie in northern France, the Boethius in Provence, the Mozarabic jarchas in southern Spain — are traceable to the 10th. This makes for a difference of twelve hundred years between the respective trajectories of languages and literatures. True, one can argue - persua sively so, where, for instance, hagiographic themes are involved - that the artistic exploitation of the given topic begins in Church Greek, thence passes into medieval Latin, perhaps in the 7th or 8th century, and reaches its conclusion in the vernaculars, somewhere between the 11th and the 13th centuries, an approach which tends to close the temporal gap by a considerable margin. Nevertheless, the fabric of the cognate languages remains the centerpiece of Romance culture, particularly if one elects to stress its unity rather than its diversity. It is a truism that, in the past, historically-oriented linguists have preferred to concentrate their attention on divergences rather than on convergences. The classic technique in grammatical and etymological studies has been to work toward the reconstruction of a complex process of fragmentation or differentiation; I refer

130

CONVERGENCE AND DIVERGENCE

you to widely used publications thus skewed by Meyer-Lübke and Wartburg, to cite just two names almost at random. No comparable corpus of investigations is readily available to satisfy the layman's curiosity about processes of convergence, except perhaps for the microscopic inspection of itineraries of certain loan-words — all told, a peripheral phenomenon. Yet the alternative agencies of convergence and divergence would alone explain the eventual preservation of a unity in the Romance domain. There has been no congealment, but a lively sequence of hardening and loosening processes. Now that we have warmed up to the subject, the moment has arrived to raise the central issue: How many discrete factors in the unity of ROMANIA may the linguist, qua linguist, be in a position to set off; how can these factors be best defined or characterized; how have they interacted? Only after this issue has been exhausted is it advisable to muse, by way of afterthought, whether nonlinguistic factors of comparable sharpness of profile are apt to be discovered. Let me announce from the outset that, in addition to the basic factor — namely inheritance, or common descent from an ancestral language, or genetic proximity — which is so obvious, flowing as it does from the annals of recorded history, as not to require any explicit justification, one can distinguish between at least four discrete contributing factors. However, the basic factor, for all its transparency, itself invites a short preliminary comment. Assuming, for the sake of simplicity, that it is permissible to visualize a gradual crumbling, or a sudden break-up, of the parent language, and assuming in addition that a feature x which we associate with the ancestral tongue appears in none of the daughter languages, does it follow from this state of affairs that, at the moment of the break-up, that particular trait, once sharply silhouetted, had already complete ly disappeared from the spoken (rustic or urban) variety of Latin? Earlier generations of scholars thought so; if the old passive {amor, amaris, amdtur. . .) or the classical future (habebo, habebis,. . . ; agam, ages. ..), except for a few scattered vestiges, was consistently absent from medieval and modern languages and dialects, then — so the argument ran for decades - the evaporation of these forms must have preceded the fragmentation of Latin. Today's analysts have become more cautious; our genera tion of workers is satisfied with inferring from such testimony as the silence of Romance dialects that, at the critical juncture, the constructions under scrutiny must everywhere have been declining and yielding ground to one of several available substitutes. So inheritance, counter to first impressions, can mean several things. Where preservation is involved — e.g., the strikingly even representation of CAPRA 'goat' or of DICERE 'to say' from the Atlantic to the Black Sea - the process is simply one of inertia; but where common losses are at issue, the bequest of the concluding phase of the parent language may very well have been a mere disinclina tion to favor the form, provided that reluctance was growing rather than declining and provided further that homogeneous or heterogeneous replacements were in sight. Once this qualification has been introduced, a big slice of the conspicuous features which even a sensitive layman - because these features are observable at the surface - would readily credit to Romance as a whole, can be safely ascribed to joint descent from a common stock. Some of these characteristics are phonic (e.g., the limited


131

role of pitch, the lack of aspiration in obstruents, the absence of glottalized conso nants); others are morphological (the presence of compound tenses, the "analytical" grading of adjectives and adverbs, the limitation of gender to masculine and feminine, the prevalence in derivation of suffixes over prefixes, the near-atrophy of composi tion); yet others syntactic (the predominance of the sequences substantive-adjective and subject-verb-object). Of the four contributing factors now on my agenda, I have placed one ahead of the rest on account of its resemblant affinity with genetic kinship. It involves, let me remark parenthetically, a situation little understood before the advent of twentieth-century linguistics; Meillet is said to have adumbrated it, Sapir dramatized it in the celebrated Chap. 7, titled "Drift," of his inspiring book Language (1921), and Joseph Greenberg, in the mid-'fifties, gave it a very sharp formulation in his masterly Essays in Linguistics. The factor presupposes the agency of polygenesis, as against monogenesis. Suppose a given feature is encountered in all or most "daughter languages": It need not have existed, certainly not in neatly-crystallized form, in the parent language, onto whose level pioneer comparatists would doubtless have tended to project it, but may very well have arisen independently in the derivative languages on account of their similar structures and related lexical inventories. Take the definite article, highly characteristic of all Romance dialects, yet absent from classical Latin. It is gratuitous to assume that after its genesis in colloquial Late Latin, perhaps through that language's connubium with Greek, all Romance dialects went, first, through a Phase A, involving post-position, as in Rumanian (lupu-l 'the wolf,' soare-le 'the sun,' etc.), then, during Phase B, switched, in the Center and the West, to pre-position (il lupo, le loup, ellobo), the way Ernst Gamillscheg envisaged the development in an ill-fated Academy lecture four decades ago. It is by far safer and smoother to hypothesize a certain fermentation, in the various provincial varieties of Imperial Latin, a playful, tentative experimentation aimed toward introducing some sort of definite article, then the selection, in the divers provincial societies or speech communities, either of ILLE or, especially around the Tyrrhenian, of IPSE, i.e., of available identifying or demonstrative elements, for this new purpose, with additional local circumstances determining the exact phrasal position of this novel grammatical tool. Assuming this argument is tenable, the post-position in Rumanian of ILLE (when attached to a noun), far from reflect ing Common Vulgar Latin usage, emerges as a feature conditioned or superimposed on a fluid colloquial Latin by a paleo-Balkanic preference - taking the Balkan languages not as a family, but as an alliance, a Sprachbund. The three remaining factors are distinctly easier to grasp and seem to be of greater immediate relevance to the general culture historian uncommitted to any favorite linguistic theory. For one thing, Late Latin at no time ceased to be in active use all over the West throughout the crucial medieval period. None of the so-called barbaric kingdoms pursued, as a goal of its Kulturpolitik, the extirpation of Latin or even the lowering of the rank it occupied. Even in southern Spain controlled by the Muslims (though not, of course, in North Africa beyond the sixth century) a cultivated variety of Latin, stylistically chastened, successfully held its own side by side with spontaneous Romance speech. This uninterrupted survival, in Catholic Europe, of Latin as the

132


long-obligatory language of liturgy and as the preferred language of hagiography, theology, historiography, legislation, and incipient science produced everywhere a unique climate of bilingualism and cultural osmosis, making Latin something of a lengua general. In Romance territory the situation was almost unique (the closest typological approximations that come to mind are the roles played by Church Slavic vis-a-vis certain Eastern and Southern Slavic vernaculars and by Sanskrit vis-a-vis modern Indo-Aryan), because Quranic Arabic, in its relation to Persian and to Anatolian Turkish, and classical Hebrew, in its relation to Judeo-Romance and to Judeo-German (i.e., essentially, Yiddish), also Quechua in its relation to certain Amerindian languages, though hierarchically comparable, could not exert any similar linguistic influence beyond incidental lexical borrowing, given the enor mous distance between, say, Arabic, a Semitic tongue, and Persian, an Indo-European tongue. The communication between Medieval Latin and the Romance vernaculars was, significantly, bidirectional. The vernaculars absorbed, for instance, at successive dates a generous portion of characteristic ecclesiastic words, recognizable as late intruders by their failure to have participated in striking sound shifts, and of whole formulas as well; you will find a classic treatment of this problem in Hans Rheinfelder's unsurpassed monograph, Kultsprache und Profansprache in den romanischen Ländern, marked by heavy emphasis on Italian. But Low Latin and, later, medieval Latin also assimilated a by no means inconsiderable number of vernacular lexical units, including some of Celtic, Germanic, and even Oriental (particularly Arabic) provenience; moreover, it qualified as a secondary channel of transmission. Thus, if one discovers a Gothic word in Old French (although no Goths are known to have settled in that territory) or, conversely, a Frankish word in Old Galician-Portuguese, spoken in a corner of the Peninsula shielded from any Frankish invasion en masse, then one of several explanations between which the language historian has to make a choice is that the erstwhile foreignism was, at first, sucked in by the local variety of medieval Latin and, at a later date, exported, at first within that medium, to neighboring countries as part and parcel of a pan-European heritage. The cultural factor which we have just outlined may serve as a justification for drawing the dividing line between Western Romance and Eastern Romance so as to exclude Rumanian from the majority group, because Dacia came early under the influence of Byzantium and was, for over a millennium, cut off from contact with, say, France, Spain, and Portugal, though probably not with the Adriatic coast of Italy. This is the way Amado Alonso, in Argentina, and Maurer, in Brazil, have interpreted 'Western Romance' - very much in contrast to scholars (W. von Wartburg, among others, the believer in the La Spezia-Rimini line) who have been guided by phonological criteria and who, obviously, to this day represent the Establishment in our discipline. For another thing, both in the Middle Ages - starting with the time of the Cluny Reform and of the Crusades - and in subsequent periods, down to our own century, one particularly prestigious Romance vernacular would strongly influence its counter parts.The oldest case and the example par excellence on account of its neatness is the radiation of the Old Provencal troubadouresque vocabulary, along with the diffusion of certain concepts and attitudes associated with Southern France's courtly culture, if not necessarily the mythical Minnehöfe. Initially concurrent with, and


133

in part indistinguishable from, the raying-out of Old Proven9al influence, both narrowly linguistic and broadly cultural, was the spread of Old French chivalric manners, matters, and words all over Europe, England and Holland included, but with particularly strong reverberations in Romance territory. It might be boring if I were tQ attempt to verbalize here, through sheer enumerations, what can be more effectively presented in the form of a chart: the record of Italian influence on France, Catalonia, Spain, and Portugal; of the Galician-Portuguese impact on Castilian; of the Spanish ascendancy, as reflected in the Hispanisms absorbed by French (shortly before the Age of Louis the Fourteenth) and by Italian, etc. Suffice it to state that the lexical (and in small part, also phraseological, syntactic, and stylistic) preeminence of France was sharply felt by the cognate cultures (and, of course, by others as well) at different points: in the 13th to 15th centuries, after a long interruption in the 17th and 18th, then, after a short break, once more in the 19th and 20th century. Needless to say, this hegemony was not restricted to lan guage. Many smaller influxes, too numerous to inventory here, have also served as side-connections in numerous sectors of the Romance edifice: the Tuscan influence on Corsican, Sardinian, and Friulan; the Venetian pressure on Dalmatian; the Castilian encroachment on Valencian, which forms part of the Catalan domain, and the like. The last powerful factor that has underpinned the unity of Romance cultures was the joint, zestful participation of most of them in Humanism and Renaissance, a process that - among many other, undoubtedly more significant and entertaining facets — involved self-enrichment through deliberate return to Greek and Latin sources, starting with practically all aspects of the written language. The generous, nearly-simultaneous tapping of an identical source obviously increased the resem blance between the daughter languages, via secondary rapprochement. Illustrations from the lexicon are so copious and, in general, so trivial as to be gratuitous in this context. More subtle and difficult to detect is the new crop of syntactic parallels; note, for instance, that in pretentious Renaissance style there appears the same proclivity toward the subjunctive, especially after como, come, comme, erroneously equated with Lat. CUM; the same overabundance of participle absolute constructions and of imitations of the classical accusative with infinitive, whether we read Italian, French, Spanish, or Portuguese prose or poetry. Even on the level of phonology one discovers some elusive traces of reinforced affinity. To cite just one example chosen almost at random: Refined speakers of Spanish borrowed from Greek and Latin numerous words beginning in Φ- and F-, respectively, and thus put an end to the threatening estrangement of their language from its neighbors, a process which started with the transmutation of Lat. FĪLIU 'son' into /fizo/, later /hizo/, /i§o/. Spanish fervor, a typical Renaissance word, is very close to Fr. ferveur and to It. fervore, whereas the much older hervir 'to seethe, boil' is at a considerable distance from It. fervere. Admittedly, the parallelism of infiltrations was incomplete; one is left wondering why It. lepido and odierno lack counterparts in French and Spanish; or why English borrowed from Latin obstreperous and from Greek deleterious, whose equivalents are used far more sparingly, if at all, in Romance. But while there remain, annoyingly enough, certain unknowns in our equations, the direc tion of the central thrust of the Renaissance is unmistakably clear.

134


Between the five factors we have so far succeeded in isolating there exist un doubtedly loose connections and even powerful conjunctions. Let me adduce a single category of cases where dual credit is appropriate. Suppose French 19th-century scholarship introduced, as a scientific label, some Latinism (including the possibility of an artificial neo-Latinism) or Hellenism, and suppose further that this tag was later adopted by several, or all, other cognate languages. In one sense, it is legitimate to invoke the agency of Factor 4, because a Gallicism is involved; in another sense, it is equally defensible to hold Factor 5 responsible for the innovation, inasmuch as the appeal to classical nomenclature is in a direct line of descent from humanistic practice. Having marshaled the linguistic evidence for five essentially diverse factors pre siding over the unity of ROMANIA, we are entitled to ask ourselves, in a brief after thought, whether other factors, of a radically different character, e.g. esthetic, religious, folkloric, literary, ethnic, could not be adduced with equal justification? I venture to think that some of them could and indeed should; I would not be surprised if their discussion were to prove more stimulating than the necessarily austere linguistic analysis you have just been exposed to; I doubt, however, that these supervenient factors can be isolated with equal neatness. Until now we have been making the silent assumption that the individual Romance languages are equals, and it certainly would be an impardonable error to deny that they are peers and to attempt to rank them, qua languages, on the basis of the literary accomplishments of the national cultures they represent. It is perfectly legitimate, on the other hand, to assess them on the basis of their typicality. In this respect, the major surprise is that French, for a variety of reasons that invite technical formulations transcending the scope of this symposium, has, from the Middle Ages on, been the least characteristic of all Romance tongues, far less typical, indeed, than a cousin as humble as Sardinian. The only comparable paradox is that Castilian, upon close inspection, has turned out to represent the least typical dialect of HispanoRomance; this startling discovery (1926) must be credited almost in its entirety to Ramon Menendez Pidal, who dramatized it very effectively. Another afterthought is that it is no longer permissible to operate with the concept of strictly Romance languages — for the simple reason that they do not exist. Once, however, we allow for the fact that the lexicon, if not necessarily the grammar, of certain indisputably Romance languages has been strongly Slavicized, Germanicized, or Arabicized, we may go one step further and agree to recognize, as the generation of Grober and Meyer-Lubke did not, semi-Romance, or partially Romance, or else thinly Romanicized languages, as is true to a large extent of English and to a slight extent of German, to say nothing of Albanian and Basque. After all, these two Germanic languages were exposed to the radiation of medieval Latin and Old French, no less than were, say, Spanish and Italian. In this manner, the once rigid mould of ROMANIA can be broken and the spread of certain fashions, preferences, and canons of taste which rightly preoccupy the literary scholar need not be severely isolated from the findings of orthodox linguists.

E. DIFFUSION

REVIEW ARTICLE Lexical borrowing in the Romance languages: a critical study of Italianisms in French and Gallicisms in Italian from 1100 to 1900. By T. E. HOPE. 2 vols. Pp. xiv, 782. New York: New York University Press, 1971.

Hope's exceptionally serious book, which may well embody the fruit of ten years or more of sustained research, has so far produced a weak impact. Perhaps a study devoted, in the main, to French and Italian should for practical purposes— i.e. out of consideration for the vast majority of its prospective readers—have been submitted in one of these two world languages; thirty-odd years earlier another Englishman, William D. Elcock, achieved a major success with his monograph on Pyrenean dialects (1938), a probe phrased in splendid French. A weak alterna tive might have been to offer such a study in German as the traditional medium of comparative Romance linguistics. Weightier reasons for the relative indifference of critics have been the two facts that Elcock's light-winged book, at the time of its appearance, seemed to stand very close to the front-line of research, whereas H's far more cumbersome investigation, except for the concluding pages, does not give the impression of echoing any part of the spectrum of tastes of the early seventies, however liberal one's interpretation of that spectrum; and, further, that the heavy architecture of the two volumes obstructs, rather than facilitates, the view. The major originality of H's book lies in the fact that it focuses attention on bidirectional borrowing; the earlier monographs on lexical borrowing involving Romance language-partners have consistently studied unidirectional infiltration—e.g. B. E. Vidos' and Bartina H. Wind's books on Italianisms in French (1939 and 1928, respectively), Marius Valkhoff's inquiry into Batavisms in French (1931), Juan H. Terlingen's Utrecht dissertation on Italianisms in Spanish (1943), A. Steiger's (1932) and E. Neuvonen's (1941, 1951) investigations of Arabisms in Hispano-Romance.1 Given this major advantage of H over practically all his predecessors, one is surprised that he has not capitalized on it, starting with the choice of a felicitous title. In fact, the book, in all other respects so thoroughly honest, has been badly mistitled (the inaccur acy is in part redressed by the subtitle, absent from the jacket): It does not deal with lexical borrowings between members of the far-flung Romance language family as a whole (and their neighbors) except, occasionally, by implication; and the reader would look in vain for any information on Gallicisms or Neohellenisms in Rumanian, Lusisms in Castilian, Provencalisms in (Old) French, and the like. The focus is exclusively on the lexical relations between two stand ard languages, French and Italian, with some attention marginally paid to dialects, especially as regards the Italian side. This is shown very clearly by the elaborate Index verborum (753— 76), unfortunately limited to Part 2 of the venture (see below). Such limitation, like any other self-confinement, can, I suppose, be defended with such arguments as achievement of superior thoroughness, finer-meshed sifting of the raw data, and more leisurely consultation of secondary literature. My own feeling is that such advantages are outweighed by heavy losses of perspective. Nothing is apt to throw into bolder relief the vicissitudes of French words in Italian than their radically different adoption by Spanish. Thus, while speakers of Italian were split in their acceptance of the French suffix -ier (used in agentives, names of containers, certain adjectives)— 1 To the familiar sources available in print, one may add Daniel J. Pratola's Berkeley disserta tion (1952) on Portuguese words of Italian origin, which runs to a total of over 450 typescript pages and includes a solid bibliography (427-43).

138

DIFFUSION

with the result that, even centuries later, Tuscan wavers between minimally adjusted -iere and maximally adjusted -iero—the speakers of Spanish swiftly and resolutely abandoned -er (except in an exiguous residue of Catalanisms, such as mercader 'merchant', and of course family names traceable to Catalonia, e.g. Ferrer, lit. 'ironsmith') and switched to ligero 'light-footed, quick', from OFr. legier < LEVIARIU. 2

The structure of the handsomely printed and bound and (on the whole) superbly proof-read book is in itself a clue to some of its assets and liabilities. The division into two volumes—admittedly, a convenience and a source of esthetic pleasure— is purely incidental; in fact, the dividing line, awkwardly enough, cuts through Part 2. I submit that, with slightly better strategic planning and at small (if any) extra cost, distinctly better results might have been achieved. The bulky book falls into five parts, if one disregards the introduction (ix-xiv), which really is a preface and should have been consolidated with a foreword (vii) containing just a few acknowledgments. Part 1 is titled 'Methods and principles' (1-23); Part 2, which patently represents the core of the venture and straddles the two volumes, is 'The historical aspect' (25-573); Part 3, 'The formal aspect' (575-634); Part 4, 'The semantic aspect' (635-77); and Part 5, 'Romance loan-words and the neological process' (679-742). Among the appended sections not yet mentioned, let me single out a list of 'Abbreviations relating to authors and works' (743-52) and a welcome, if exceedingly motley, Index rerum (777-82)—one finds in this catch-all, side by side, key concepts of linguistics, specific suffixes, names of classics of belles-lettres (Ariosto, Boccaccio, Brantdme), and names of cities and provinces (e.g. Venice and Piedmont). All this is commendable, but the most important and urgently-needed bit of information is shielded from quick inspection. A linguistically oriented reader would like to find out from the start whether H has availed himself of innovative works like Haugen 1950 and Weinreich 1953 on the broad subject of 'languages in contact'—and, if so, how often and in precisely which contacts. The answer happens to be yes; but this vital information has been withheld by a bashful H from his readers; to ascertain the most important of all sources, they are apparently expected to wade through 800 pages.3 2

These problems are discussed in Anita Katz Levy's brilliant predoctoral study (196567), which became available in the mid-sixties but must have escaped H's attention; cf. the following footnote. 3 As a matter of fact, H has made good use of at least some characteristic writings of both Haugen (577-8, 623, 638-9, 645, 651, 676) and Weinreich (579, 638-9, 650, 681). I became sufficiently titillated by the question here raised to have prepared a private index of H's second ary sources for Parts 3 and 4, which happen to aggregate a one-hundred-page segment (577-677). Influential practitioners of general linguistics and of IE studies do come up, at intervals, for incidental mention: M. Breal (669), L. Deroy (577, 579, 638, 650), G. Devoto (661), J. R. Firth (674), A. L, Kroeber (611, via U. Weinreich), H. Kronasser (638, 650), J. Lyons (674), J. Marouzeau (614, 651-2, 655), A. Martinet (675, 682), A. Meillet (665-6), E. Nida (669), V. Pisani (577), I. A. Richards (646), F. de Saussure (578), H. Sperber (668), H. Vogt (650), and E. Windisch (579). H has also paid a modicum of attention to Anglists, Germanists, and representatives of similar groups: W. Betz (638), W. E. Collinson (654), P. Schach (613), Aasta Stene (662). His best friends in these quarters were apparently Deroy and Marouzeau; one is surprised by the absence from the roster of E. Benveniste (who assessed Deroy's book in a severe review, 1959) and, especially, of L. Bloomfield (whose long-authoritative book Language contains three chapters on borrowing, and operates with the controversial concept of 'intimate

LEXICAL BORROWING IN THE ROMANCE LANGUAGES

139

It may meanwhile have become clear that Parts 1, 3, 4, and 5 (all four relatively short), which jointly represent the theoretical framework, or at least an approximation thereto, belong together, not least in tone and style of presentation, and should preferably have been assembled in Vol. 1, which would then have had an unequivocal appeal to general linguists, students of diffusion and bilingualism, sociolinguists, experts in diachronic analysis, and devotees of comparative Romance linguistics. Vol. 2 might then have been reserved for the heavilydocumented Part 2, qualifying for consultation by philologists, textual critics, Renaissance scholars, specialists in the growth of material civilization, and professional lexicographers (etymologists included)—in short, a contingent of humanists and historians. Whether we like this situation or not, these two academic crowds no longer mix with any degree of frequency; slanting each volume, realistically, in the direction of each coterie's predilection would have vastly increased the usefulness of the monograph, making it truly 'double-barreled'. (The only benefit that can be cited in defense of the lay-out favored by H is his ability to refer back, in Parts 3-5, to case histories discussed at some length in the preceding historical section; however, the advantage is slight, applying as it does to the rare, perhaps non-existent reader who intends to plough, page by page, through the entire ponderous book.) One can learn a good many other things from even a superficial examination of the book's structure. The most elaborate section, Part 2, has been so organized that the five major sub divisions are chronological, stretching from the Middle Ages to the 19th century; our own century, despite its obvious liveliness and the countless examples of Italo-French collaboration and even teamwork, has been excluded from microscopic inspection, though one discovers stray references to recent events (580 fn. 1). Each of the five temporal subdivisions has been organized in identical fashion, falling into four shorter segments: (a) Italianisms in French, (b) Italian influences on French vocabulary, (c) Gallicisms in Italian, (d) French influences on Italian vocabulary. Such a pattern is defensible, making for easy comparability; but it also entails a certain monotony, and is better suited to a reference work than to an exploratory monograph. At the end of the historical survey, looking back at over five hundred crowded pages, the reader searches in vain for a paragraph summing up and characterizing crosstemporally the entire curve of Italo-French relationships—or offering, by way of brief counterview, the landscape of these relationships arranged on the basis of an alternative principle, e.g. by contrasting one direction with its opposite, or by pitting the -isms against the influences. In contrast, the structure of the four theoretical, or rather methodological, chapters displays

borrowing'). Most of the authorities invoked are, understandably, Romance scholars of varying caliber, especially those active thirty to sixty years ago: P. Barbier (577), V. Bertoldi (620), R. R. Bezzola (594, 624, 626), F. Brunot (602, 607), A. Carnoy (669, via S. Ullmann), A. Dauzat (621), E. Gamillscheg (654) etc.—but with a sprinkling of younger workers: R. P. de Gorog (639), K. J. Hollyman (647), M. F. M. Meiklejohn (588). One is left wondering about the implication of this state of affairs: Have loanword studies declined in quality, in quantity, or in both? or did H collect most of his bibliographic data, except those for Part 5, fifteen to twenty years ago? The latter possibility is not unlikely, since, in discussing a famous crux of French suffixal derivation—namely the source of -ade, the exotic counterpart of -ee—H deems it necessary (602) to cite A. H. Luther's worthless and wholly antiquated Columbia dissertation (1912) and C. S. R. Collin's more meritorious, but also obsolete writings (1906-18), while omitting some obvious modern, and, in part, stimulating collateral sources of information: e.g., J. M. Piel's revision (1966) of the classic Franzosische Wortbildungslehre by W. MeyerLübke (in that book, see 197-8 for reviser's relevant addenda to original §116) and E. S. Georges' posthumous article on past-participial nouns (1968), to say nothing of Georges' and Annegret Alsdorf-Bollee's subsequently issued book-length treatments of the same subject (both 1970), which may indeed have appeared too late to be taken into consideration. I have noted through out how little attention H has paid even to extensive and searching book reviews; thus the Piel edition of Meyer-Lubke has elicited critical reactions possibly more substantial and incisive than Piel's addenda—reactions which surely might have enriched H's treatment not only of -ade, but of Fr. -age > It. -aggio as well.

140

DIFFUSION

great variety and unpredictability. As one plunges into Part 3, 'The formal aspect', H's 'Pre liminaries' (577-9) give one a few interesting shreds of information on such familiar matters as loan-words vs. loan-blends and loan-shifts—or, more original, on certain differences between the context of bilingualism in an immigrant milieu and in the past relation between Frenchmen and Italians. But one receives inadequate guidance as to what is to follow immediately: the specific interference patterns and problems (palatalized vowels, 579-81; / + consonant, 581-3; s + consonant, 583-6; groupsp1/,bl,fl, kl, 586-8; stress, 588-90; nasal vowels, 590-93; the representation of certain French diphthongs, both rising and falling, 593-5; change in gender, 595-7; prefixes, 597-9; final consonants and suffix analogy, 598-609), the formal facets of the process of borrowing, and many other highly relevant matters. If the reader, in cutting a path through the historical part, was overprepared for things to come, almost to the point of surfeit, his power to anticipate forthcoming discussions suddenly becomes eroded. H is not unaware of the difference in slant between the philological and the linguistic sections of his magnum opus; and his introduction, in which he traces the development of the project, is a human document of sorts. Apparently his original plan called for the study of cultural relations between the two countries in light of lexical borrowings (ix-x), a strategy which would have made him sail in the wake of Frazer Mackenzie's standard-setting treatise Les relations de I'Angleterre et de la France d'apres le vocabulaire (1939).4 A crisis ensued, when H dis covered that such an approach would only make his inquiry ancillary to historical research—and, worse, might fall short of enriching our historical insights (xii).5 Coincident with this despondency was the suspicion that a group of influential frontline linguists, on both sides of the Atlantic, had meanwhile adopted a negative attitude to lexico-etymological studies in general, and had even questioned the wisdom of operating with borrowings in synchronic analysis; here H apparently refers to the age of militant descriptivism (say, the mid-forties in America). But H recovered his zest and the momentum of research—so he reports—by switching to linguistic methodology and theory, vindicating the status of loan-word study in terms of its benefits to general linguistics; hence Parts 3-5. This report of a crisis and subsequent conversion, which best explains the unusual delay in the completion of the book,6 should some day be re-examined, in balanced retrospect, along with other testimonies and indirect proofs of a widespread collapse of traditional Romance linguistics, a fiasco which seems to have occurred in the fifties.7 I find H's admission of temporary 4 The phraseological model, in German, for this kind of attack on a moot problem of re lationship is: ... im Spiegel ... 'reflected by, mirrored in'. K. Vossler's controversial book, Frankreichs Kultur im Spiegel seiner Sprache (1913, later re-issued under the less challenging title Frankreichs Kultur und Sprache, 1921) does not stand alone in this category. 5 This particular apprehension seems to me unfounded; certain etymologies open up vistas to which other historical sources provide no avenues of approach. On this point see my recent paper, 'Etymology and modern linguistics' (Malkiel 1975). 6 My suspicion that a draft of this book goes back to the fifties, based on internal evidence, is borne out by repeated citations (639, 643) of an essay on 'The analysis of semantic borrowing' from H's pen, traceable to a testimonial volume published in 1960. 7 It is interesting to observe how J. G. C. Herculano de Carvalho, who veered away from sparkling studies in the 'Worter und Sachen' key (cf. my review, Lg. 33.54-76, 1957) to a fairly pale kind of structuralism, correctly diagnosed certain weaknesses of the earliest Establishment (e.g. in his trenchant 1967 review of W. Giese's Los pueblos romdnicos y su cultura popular), but himself never attained his previous buoyancy. I have touched upon this situation in the opening paragraphs (plus fn. 1) of my forthcoming review of B. Malmberg's miscellany, Linguistique generate et romane.


141

defeat humanly moving, as a period piece; but his attempt at extricating himself from the predica ment is less than successful. Even in America, the recoil from or aversion to etymology, among front-line theorists and innovators, was at no time universal; suffice it to refer to the concluding page in the excellent Chap. 3 of Greenberg 1957. The gradual comeback of lexicography, under the aegis of linguistics, started shortly thereafter with the prestigious 'Problems in lexico graphy' Conference (cf. Householder & Saporta 1962). Weinreich, a leading participant on that memorable occasion (I can testify to that fact as an eye-witness), was not only an imaginative theorist but also, unlike most of his peers, an inspired practitioner—to be specific, the author of a major bilingual dictionary and of a sophisticated, lexically-oriented dialect atlas; both projects were then at an advanced stage of preparation (only the dictionary has since appeared, posthumously), and both were equipped with an implied diachronic dimension. For H to start a program of re-orientation with a search for loose metaphors hardly strikes me as a particularly bold or promising move; 8 what he could have achieved instead is to cleanse his own prose of an annoying residue of Victorian and Edwardian English reminiscent of Skeat—a phrasing far too genteel for the taste of most Anglophone linguists, who, on the other hand, seem not to mind newly-minted metaphors, however strident or grotesque.9 Finally, the real difficulty that H appears to have encountered, and with which he has not quite come to grips, was not the clash between a well-defined older style of research and another neatly-delineated newer one, but rather a conflict between a basically homogeneous traditionalism (despite varying nuances of presentation, the older workers, in the main, used the same scholarly discourse) and a hetero geneous, internally-split modernism, in which—to cite just two examples—semantics and lexi cology are ruthlessly extirpated by some analysts, while they are placed on a pedestal by others. It is the faction-ridden, self-contradictory character of modernism that threatens to confuse conservatives like H, who in their formative years were attuned to orthodoxy. Finally and most significantly, Romance linguistics can to a large extent be rejuvenated from within; among the hundreds of borrowings inventoried and briefly studied, there were surely some that H could have effectively placed in the center of etymological excursuses—thereby invigorating a genre which, since the days of Hugo Schuchardt, has proved conspicuously germane to Romance scholarship.10

In harmony with the preponderant taste of this journal's readership, I shall henceforth stress the broad-gauged Parts 1 and 3-5. However, despite its seductive title ('Methods and principles') Part 1 is, to a considerable extent, a time-honored but less-than-exciting survey of earlier research and available sources, of H's actual program of intensive reading and excerpting, of his techniques (mostly rules-of-thumb on back-dating, detecting ghost-words, etc.) for checking the accur8

' One might begin in a small way by re-examining minor conventions of practice which tend through long acceptance to become immune from criticism' (xii). It hardly need be demon strated that such display of timidity—unfortunately, by no means atypical of H's book—is unlikely to be efficacious. 9 H's introductory pages contain such obsolete expressions as 'neo-Latin languages', 'Romance idioms' [= R. languages], 'comparative philology', 'acceptation' and 'significa tion' in lieu of 'meaning'; the same proclivity continues throughout. Side by side with these archaisms H tosses in, anachronistically, ultrafashionable expressions like 'metalanguage'. His use of philosophical terms seems imprecise to me; he speaks of 'deductions' (ix) where 'inductions' or 'inferences' would have been more accurate. Again, I am baffled by his talk (xiii) about 'historical or positivistic points of view' and about 'an approach, atomistic yet organic and teleological'. 10 Of course, I am not advocating lavish digressions on genetically obscure Romance words just because they happen to have been transmitted, perhaps millennia after their coinage, from one language to another. Truly relevant situations are those where the possibility or plausibility of borrowing is pitted against that of native descent (or spontaneous creation).

142

DIFFUSION

acy and completeness of various lexicographic compilations, and the like.11 'Methods and principles', so interpreted, applies not to the broader linguistic problem on hand, but to the sifting and cataloguing of rules or norms followed in preparing those word lists which constitute the bulk of Part 2. 12 Where possibilities of more theoretical discussions were latent, H has not always grasped such a chance. Take the moot issue of periodization (8-11). H marshals arguments for offering his material sliced up in centennial segments, except for the medieval period presented as a whole (in part, as a result of meager or lacunary evidence), and he rejects on good grounds such alternatives as fifty-year or ten-year segments. But while it is permissible to start from a highly artificial edifice of regularly recurrent calendar units—as a provisional construct, so to speak— the goal of a truly linguistic operation in diachrony is to arrive at a succession of such periods of varying length as are defined by internal criteria (be they phonological, morphological, or semantic). The unsurpassed model in Romance scholarship is the section on phonology in Meyer-Liibke's classic HGFS (1908). Unfortunately, H's 'Methods and principles' offers no clues to this definitive, self-correcting approach to periodization.

H comes closest to tackling grammar in certain sections of Part 3. One such section, on ' Change of gender' (595-7), commands heightened interest, not least on account of the many challenging questions raised (but in part left unanswered) by other researchers, in regard to oscillations of gender in Romance. 13 With the wealth of material at his fingertips, H makes a noteworthy discovery: Numerous Italian masculines, upon penetrating into French, become feminines. Examples in point are biscotto 'biscuit' → biscotte; bro(c)catello 'brocade, peacock-blue and white marble' → brocatelle (via broca-tel, -del, still masc.); crinolino' hoop-petticoat' 11

Given the extraordinary experience which H was privileged to acquire over the years, his value judgments certainly carry weight; I am thinking, e.g., of his unequivocal comments on the etymological component, not yet tested, of the relatively new Grande dizionario della lingua italiana (3-4). His praise of B. E. Vidos' remarkable thoroughness (5) will cause little surprise; and I am pleased to see H confirm (7) the reliability of the Migliorini-Duro Prontuario (1950) —Migliorini's flair for Gallicisms may be due to the fact that he started his career as an expert in French. But despite H's meticulousness in certain respects, especially as a passionate excerpter, I detect in this introductory part not a few inexplicable lacunas. He seems to have made ex tensive use of the original edition of E. Gamillscheg's etymological dictionary of French (EWFS), traceable to 1926-28, without paying heed to the thoroughly revised 2d ed. (1966-69). Again, H's tribute to Vidos is not balanced, as it should be, by equal attention to the remarkable Mediterranean studies, stretching over a period of almost forty years of inspired labor, by H. and R. Kahane. 12 The reader will be amused and enlightened by H's cautions regarding certain quirks and biases of widely-quoted authorities (6-7). Of particular relevance is his warning on the tendency of older authors of bilingual dictionaries to invent artificial equivalents (7)—though such phantom words, through a further caprice of history, may later actually be minted, in which case the spurious entry becomes a premature bit of documentation. Let me add on my own that certain successful writers can be safely suspected, if not convicted, of casting around for neologisms in the muddy waters of unsifted dictionaries—a situation which would close the circle of the comedy of errors. The artificial coinages, incidentally, are apt to creep in most frequently where cognate languages like French and Italian are involved; most of them are suffixal derivatives arrived at through simple equations predicated on the (false) idea of sym metry, such as -eur = -atore, -ment = -mento, -et = -etto. 13 In this context H cites the English version of my own paper on diachronic hypercharacterization (1957-58) and embroiders on it with great sophistication. He could have profited even more from familiarity with the investigations conducted by Garcia, Hasselrot, Kahane, and various Berkeley-trained younger workers; see below.


143

→crinoline; fresco 'type of painting' →fresque; ombrello '(leafy) shade, parasol, etc' → ombrelle (still masc. in Montaigne); ritornello 'refrain, repetition, ritornel' → CI. Fr. (Moliere) ritournelle; schiopetto 'very short carbine'→ escopette (via chopet); sigaretto ' small cigar' → si-, ci-garette. H offers a not implausible ex planation : Speakers and writers involved in the transmission of these words were eager for the characteristic last consonants—whether /k/, or /l/, or /n/, or /t/—to be preserved, and the neat word-final articulation of these phonemes in French suggested the choice of the feminine rather than the masculine gender (here one would have liked to see heavier stress placed on such adjectives asfou,folle 'crazy', fin, fine 'thin', muet, muette 'dumb', which could have played a mediating role). Nevertheless, H finally (597) admits his inability to explain away sojfitto '(false) ceiling' → sojfite, which has remained masculine to the present day. Actually, there is nothing mysterious or exceptional about sojfite, if only one agrees to modify H's pattern of analysis. In the aforementioned cases (biscotte, brocatelle etc.), it was, I feel, principally morphological pressure, namely the ready availability of the derivational suffixes -elle, -ette, -ine, -otte, which clinched the decision on the gender.14 Since French, unlike Spanish (-ita), has no such productive suffix as *-ite, the phonological pressure brought to bear on soffite proved of itself inadequate to produce the expected change (though in the case of stallo 'stall, seat of honor, throne' → stalle, the switch did occur after a period of protracted wavering; for Cotgrave, stalle was still masc.). Only where suffixal pressure was inoperative, then, can one expect to find an area of indeterminacy; here all sorts of contingencies and minor ingredients may be predicted to play a decisive role, and long-drawn-out vacillation is the rule rather than the exception (cf. the case histories of boussole 'compass', caprice 'whim, freak', cartouche 'escutcheon, cartridge', as pieced together by H himself). As regards the speakers' decision in the case of redoute 'redoubt'—which, though traceable to It. ridotto, became femin ine (300, 596)—I happen not to share H's perplexity: Here the hazard of collision with the family of douter 'to doubt' < DUBITARE, especially with the postverbal (masc.) doute which replaced DUBIUM (cf. the spelling of Eng. redoubt), could plausibly have provided the stimulus for differentiation, in conjunction with la route 'road' etc. One is reminded of Erica C. Garcia's successful demonstration (1970) that gender switch in Spanish derivation frequently runs parallel to semantic estrangement. The most intricate among the situations reconstructed by H involve a process of lexical backlash: Certain Italianisms, after infiltrating the French lexicon and undergoing the change to the feminine in their new habitat, re-entered the stream of Italian (or, at least, were allowed to influence its course), giving rise secondarily to Italian feminine doublets. H affirms this zig zagging development for sigaretta (though not for ombrella, counter to earlier opinion), and he toys with the idea of extending it to saponetta 'gentle, fragrant kind of soap', in lieu of earlier -etto, from sapone (masc). At this point, however, his conjecture runs afoul of the hypoth esis of the diminutive (or hypocoristic, or potentially meliorative) masculine vs. the augmenta14

To this argument one is tempted to add a consideration advanced by Karen H. Kvavik (1975) in regard to Spanish (but presumably applicable to French and Italian as well): While a slight majority of primitives in a typical Romance language are masculine, the overwhelming majority of suffixal derivatives are feminine. The Italian-French gender switch is conducive to smoother integration with the general design of the language.

144

DIFFUSION

tive feminine in Romance—a heatedly-discussed key problem of derivation; cf. the connective earlier studies by B. Hasselrot (1945, 1951) vs. those by H. and R. Kahane (e.g. 1948). Part 3 contains, we recall, two rather detailed surveys of the vicissitudes of the derivational suffixes -ade and -age, each one exemplifying diffusion in a different geographic direction. This trespassing by a lexicographer on the territory of morphology is all to the good; but would it not have been even more meritorious for H to have offered his readers a synopsis of the deriva tional elements that, in the process of lexical borrowing, oozed from French to Italian and vice versa? In addition to straight suffixes, he might have mentioned characteristic final wordsegments (suffixoids = the 'primary suffixes' of L. Bloomfield 1933 and the Sanskrit grammar ians)—which, though not segmentable, have interacted with genuine grammatical morphemes and have thus become pertinent to diachronic morphology. In search of information along this promising but neglected line, I have culled, from a single word-list of the many compiled by H ('The sixteenth century: Italianisms in French', 148-227), examples furnishing evidence of strong Italian influence on the configuration of French words. We find a palpable increase of lexical items in -ache (bardache 'mignon'—and verbs, e.g. amouracher ' t o infatuate'); -asse (bancasse '[naut.] locker'); -aille {antiquaille 'antiques, relics'); and -an (artisan 'craftsman', balzan 'horse with white stockings') alongside -in (algousin 'petty officer in charge of the galley slaves', assassin 'murderer', baldaquin 'canopy', bambin 'baby'). Many ways were open to H for meaningful and economic grouping of such data, which he had laboriously assembled over fifteen years: He could have experimentally isolated the suffixes and suffixoids in a, e, i etc., to test the recurrence and (possibly) the semantic load of characteristic stressed vowels, oral and nasal (cf. /a/ and /i/, above), which preside over grammatical morphemes and approximations thereto—even the agency of vocalic gamuts. In this context, I suspect that the influx of Italianisms vastly increased the representation of French stressed a and especially of o, meagerly scattered over the native stratum after the changes a>e and o>ou>eu in free syllables, and thus enhanced their hierarchical status: cf. babiole 'bauble', baioque 'money of little value' etc., with a dimly recognizable phonostylistic value. Alternatively, the consonantal pillars of suffixes and structurally kindred segments might have been isolated for separate analysis, in an effort to test their salient features; impressionistically, I note the heavy representation of -ss- /s/ and -ch- /c/ ( > /§/). Then the length (mono-, bi-, triphonemic) and the canonic configuration (VC, VCC etc.) of the suffixal segments would have yielded a third, equally legitimate and significant criterion of classification. Because Tuscan, if not Gallo-Italian, was phonologically far less eroded than French, certain suffixes with which it endowed the borrowing language had distinctly more ' b o d y ' to them than their native transAlpine counterparts (cf. -esque, as in arabesque, barbaresque 'barbaric, barbarous, of the Barbary Coast', vs. indigenous -ois, as in Francois 'Francis' beside francais 'French'). A fourth, slightly riskier classification suggested itself: To what extent could an alert average bilingual of that time be credited with the ability to 'equate' the Italian suffix and its French congener? Surely, there could have been no difficulty about -elle (as in animelle(s) 'sweetbreads, offal'); -aille {antiquaille 'antiques'); -ment (appartement 'apartment'); -ette (baguette 'rod, stick'); -ance (bastance 'sufficiency'); -(t)ure (battiture 'scales or sparks which fly from metal being forged'), and the like; but what about -esque and its relations to autochthonous -oisl Fifth, the numerical relation between genuine suffixes and pseudo-suffixes might also have invited a separate probing, with several side-issues: e.g. It. bozz-ello 'stone moulding and projection' is an authentic derivative from bozza; while bosel, bossel, bozel in Renaissance French is a cluster of completely isolated forms lacking a primitive—a situation which reflects on the gram matical status of -el. Apart from such general breakdowns, numerous fascinating individual studies—with a builtin linguistic rather than merely lexicographic slant—were within H's easy reach. A few hints must suffice. Indigenous Fr. -on < -ONE is either neutral as to size or diminutive (anon is a 'donkey's foal', fig. 'silly child', 'little fool'); conversely, It. -one (like its Hispanic counter parts) is either neutral or, more frequently, augmentative (ballone is 'big ball', donnone is 'fat, majestic woman'). The importation of It. -one words thus threatened to produce a serious conflict. A word like balcone 'balcony', immune from segmentation, was hardly conducive to misunderstandings; but ballone, which the French adopted as ballon, definitely was fraught


145

with ambiguity. One possible outcome of such clashes is dramatized by Fr. caraf(f)e 'waterflask, decanter for wine' vs. carafon 'small decanter'—two neatly defined words familiar to habitues of French restaurants the world over: carafe, H confirms (149, 174), is a transparent Italianism (rooted in the last analysis in an Arabic word); but carafon, which he leaves unmentioned, gives the impression of being a French innovation grafted onto the Italianism, though it might conceivably turn out to be a radically re-interpreted ('disguised') borrowing. On the other hand, Sp. garrafon, which for chronological reasons is more likely to be an Italian ism than a direct borrowing from Arabic, means 'large carafe'. Equally exciting might have been an analysis of the Italian pressure exerted on two homonymous native French suffixes: -ece < -ITIA, used to derive adjectival abstracts (cf. altesse 'highness, loftiness') and -esse < -ISSA, serving to designate female agentives. Through an unfortunate coincidence, two eyeopening papers bearing on these twin suffixes, by Margaret S. Breslin (1969) and Kathleen Connors (1971), respectively, appeared almost simultaneously with H's book, so no crossfertilization occurred. 15

Part 4, devoted to the semantic facet of lexical borrowing, contains four loosely strung essays: on semantic loans (which H also calls semantic extensions, 63751), pejoratives and pejoration (651-61), change of meaning (661-8), and synonymy beside near-synonymy (668-77)—though H has scrutinized some semantic problems, a bit incongruously, in the much longer and far more complexly architectured preceding sections, in discussing effects of loan translation, loan-creation or false loans, and paronymic attraction. The general impression made by these forty pages is one of marked maturity of judgment and unusual sophistication. In this province, H has clearly done a good deal of original thinking for over ten years, rather than mere diligent data-collecting. His independence in matters termino logical (particularly confusing) and analytical is apparent throughout—though he has not neglected a heavy share of homework, namely assigned readings, either. The influence of his teacher S. Ullmann, which H freely acknowledges from the start (vii), is apt to have been stronger here than elsewhere, as shown by the pervasive predilection for the concept of polysemy, a notion which U. Weinreich—another source of inspiration (642)—of course also favored. Several successive layers of Haugen's thinking, as disclosed by his publications, have been tapped (638-9), in addition to a meaningful epistolary exchange between the two men (639, fn. 2).16 Certain ideas of Hans Vogt's (1953) on semantico-stylistic, phonemic, and gram matical 'convergence phenomena', assimilated by H (650), remind one, not sur prisingly, of Meillet (for many years Vogt was one of the French comparatist's most loyal students)—but one finds counterparts, let me again add, in Sapir's classic treatment of drift (1921) and in J. H. Greenberg's 1957 elaborations on Sapir. Some of H's readings turn out to be particularly fruitful. Thus, a re-examination 15

For the same unfortunate reason, I could not avail myself of the valuable material amassed by H in preparing my own contribution to the Delattre Memorial, submitted several years before the actual publication date (1972). 16 One point of doctrine on which Hope disagrees with Haugen—probably without knowing it—is that Hope equates 'phonetic' with 'formal', using these terms in contradistinction to 'semantic' (617, 644), whereas Haugen operates (in my view correctly) with a tripartite struc ture: phonological, grammatical, and semantic (638). My alternative interpretation of gender switch as a concomitant to borrowing, above, shows why I favor Haugen's codification over Hope's. The fact that, of various monographic and critical writings by Haugen (dated 1950, 1952, 1953, 1954, 1959 etc.), Hope adopts the terminology and underlying approach charac teristic of the first (loan homonyms and loan synonyms as the basic variants of loan shifts) is another indication that the early fifties were the actual period of incubation of Hope's book

146

DIFFUSION

of Valkhoff's list of Batavisms in French (1931) discloses that, in this group of borrowings, the percentage of semantic loans is extremely low (from 1% to 2%), no doubt because of the formal incomparability of the two languages and, more interestingly, because 'the Dutch loans are ill-matched semantically, in the sense that they often refer to objects or circumstances which could not readily be de scribed by a polysemic extension or figurative transfer of existing French words' (649). The real element of surprise, however, is the fact that the percentage of semantic loans is distinctly higher in Italian than in French, 11.1% vs. 6.25% (639-40). H aptly describes this disproportion: 'Homophony is more likely to arise between established Italian words and Gallicisms than in the case of loans in the opposite direction; consequently the probability that a new meaning will become current without the acquisition of a new form is greater in Italian than in French' (640). H wittily explains this paradox by the high frequency of learned borrowing in French, the idea being that Latinisms provide 'intermediate patterns ... which often happen to be closely in accord with those of Italian'—a good observation felicitously phrased, except that one misses here a broader hint of the pervasive camouflaging of Latinisms in Tuscan, a feature that sharply distinguishes that language from, say, Spanish (cf. It. interrompere, integrated with rompere, vs. Sp. interrumpir, isolated from romper; and untold similar instances). Among the highlights of Part 4 let me mention also H's succinct discussion of those rare cases —real tidbits of lexicography (641)—where semantic borrowing has occurred in relation not to an established native word (ordinarily, 'vernacular' or 'popular') but to a loanword culled earlier from the same source language, e.g. the culinary term canape (19th century) grafted onto canape 'sofa' (17th century), and similar niceties. In discussing loan homonyms, H remarks: 'Usually a broad difference of meaning on the synchronic plane corresponds to a greater divergence historically' (642). He goes on to observe that, of two loan homonyms, one or both are usually drawn from non-Romance languages, and offers a few telling examples; e.g., Fr. civette 'civet' (animal and perfume) is in the last analysis an Arabism, which civette 'chive' is not (its source, which H inadvertently conceals, is Lat. CAEPA 'onion'). This is a new and potentially important criterion in gauging the plausibility of etymological conjectures. Another original flash of thought enlivens H's discussion of loan translation, lato sensu (643-4). With out entirely endorsing Kristian Sandfeld's pioneering essay (1912), which lumped together formal caique (Fr. entre-vue → It. intervista) and semantic caique (Fr. canard 'lying newspaper report', lit. 'duck' → Ger. Ente, to which one may add Russ. utka), H nevertheless insists on the wisdom of granting separate status to the semantic caique (rather than subsuming it, as most moderns do, under loan synonyms and loan homonyms), because of the element of deliberateness, of 'a conscious desire to exploit the parallelism'. This analysis implies, of course, a general philosophy of language distinguishing between higher and lower degrees of intention on the part of, at least, an elite of speakers within a community. In loan-polysemy (= synony my), we further learn, the borrower need not be aware of both the formal and the semantic state of affairs in the source language; it suffices if he develops an awareness of the latent affinity between the intruder word and its native counterpart in the target language (644). H resigns himself to the thought that, in reference to older periods, the degree of awareness and intentionality cannot be flawlessly reconstructed from a semantic loan's formal aspect.

Part 5 stands completely apart, 17 constituting in essence a paper in general linguistics; it bears on loan-words viewed as one of several categories of neologisms, with Romance examples and (in particular) lexical exchanges between French and Italian intervening as an occasional frame of reference (note the excellent vignettes 17 Unlike the preceding parts, this one is marred by a long series of typographic slips, many but not all of them self-explanatory (682, 683, 685, 686, 687, 688, 690, 691, 693 [two], etc.)


147

on Fr. manquer, imported as a substitute for indigenous [de]faillir, and on Fr. contraste, introduced to replace a cluster of less serviceable native words, 704-6). This part also belongs to a discernibly more recent layer: H's scattered preliminary studies which have fed this section go back only to the years 1962-65; 18 and the semantic literature examined in a critical vein includes writings not only by J. R. Firth (1935) but also by J. Lyons (1963), which in turn means that American generative semantics comes up for peripheral mention. Perhaps the latest theoret ical treatise taken into account is A. J. Greimas' Semantique structurale (1966): If my surmise is accurate, this would imply that H's book was concluded and thence forth left unrevised five years before it went to press—a hazardous lag in linguistic theory these days. H's terminology and phraseology also become livelier and less archaic here: The pages teem with references to empirical groupings, total reality, methodological constructs, semantic solidarity, discrete taxonomy, models and metaphors, finite conclusions, semantic universals, and the dynamism behind neologisms—while old-fashioned 'cultural loans' emerge, rejuvenated, as 'lexical interferences' (681).19 The two key problems discussed in this part, with a minimum of dependence on documentation, are: 'firstly, WHY ARE WORDS BORROWED? And, secondly, HOW CAN THE PROCESS OF LEXICAL INTERFERENCE BEST BE ENVISAGED AND MOST EFFICA CIOUSLY DESCRIBED?' (681). The questions are relevant and stimulating; so the

reader's only regret is that H has discussed issues of comparable breadth—e.g. 'what a loanword is, how it may be recognised, the sort of cultural and historical information it can provide', and has criticized 'traditional methods of loanword analysis' in such a helter-skelter way, and with distinctly less neat isolation from the overflow of data (the Index rerum, incidentally, offers no clue to the locus where a loanword is defined). As one scans page after page of H's discussion, one recognizes that he could be best described as a 'neo-traditionalist'.20 In the preliminaries, while overwhelming the reader with stray refer ences to Wittgenstein (691) and Chomsky (683) and invoking 'Gestalt' (687), he quickly parts 18 The carelessness that attaches to the proofreading of this part is, unfortunately, also re flected in the corresponding entries of the bibliography. Even among H's own three preliminary papers (1962-3, 1965a, b), two are listed with incomplete data; the third, presumably in a show of modesty by H, has been altogether skipped. I have tried to supply the full information in my references, below. Again, H's habit of citing or quoting the latest editions of secondary sources is not commendable where a course of events or a chain of pronouncements is to be reconstructed. Thus the long passage from W. von Wartburg's Problematik und Methodik, reproduced (685, fn. 1) from the revised ed. (1963) of a French translation, is traceable in the last analysis to a celebrated Festschrift article of the late thirties—a circumstance which accounts best for its affinity to Bally's hypothesis invoked immediately thereafter. 19 This terminological euphoria has a few harmful side-effects: e.g., H uses 'cultural' in two utterly irreconcilable senses, practically side by side (684). When he refers to borrowings as 'cultural symbols', it is the social anthropologist's culture that is unequivocally involved; yet the phrase 'cultural languages', a transparent (and perhaps gratuitous) imitation of Ger. Kultursprachen, presupposes the humanist's (and the layman's) traditional view of culture ('languages endowed with a legacy of sustained intellectual and artistic refinement'). 20 Though the label was coined by Menendez Pidal in reference to medievalist studies in literature, Rebecca Posner has subsequently extended it to a climate in linguistics that seeks a compromise between the old and the new, and strives for continuity rather than rebellion against classicism (1970:433-51).

148

DIFFUSION

company with the brand of experimental semantics that deals primarily with logico-grammatical structures, and retreats to the more comfortable—because more conventional—exploration of moves 'from lexical towards cultural situations as well as in the opposite direction' (683-4). He is clearly very much at home in the linguistic cosmos of Orr, Bally, Wartburg, Guiraud, and Ullmann—but somewhat afraid of Fries and even of Martinet, whom no one else, to my knowledge, has so far declared a straight descriptivist (682). Later, in the analysis of H's first key problem, the situation repeats itself. He starts out well, elegantly and persuasively contrasting two older, but still widely-discussed semantic models: on the one hand, Jost Trier's 'Feldtheorie' (1931) and some of its immediate reverberations (Ipsen, Porzig—but not, surprisingly, Weisgerber 21 ), which operate with an internal segmentation of the lexicon (688-95); on the other, the attempt by Georges Matore (1953), with his small band of followers, to see in the fluctuations of the lexicon almost exclusively the impact of the vicissitudes of social history, i.e., of the outside world (695-701 ). 22 H correctly recognizes the strength and the weaknesses of each method; e.g., the school of Matore is unable to grapple with 'non-historical' words, purely relational or broadly human (700). So far, so good; but instead of capitalizing on his show of strength and proposing a schema superior to Trier's and Matore's, H suddenly decides to retreat (701) to a much older point of view, namely to 'the time-honoured conception of LEXICAL DEFECTIVITY or INADEQUACY remedied by the creation of neologisms'—even though formulating this problem more forcefully than was customary half a century ago. This dis appointingly prudent strategy suggests a streak of conservatism or a lack of boldness.

A critic can almost indefinitely pursue the discussion of a monograph which, had it appeared in print easier to the eye, could well have run to over one thousand pages. As regards sheer massiveness and the total investment of time, energy, and devotion to a less than overwhelmingly exciting topic, I cannot readily adduce a single work of British scholarship in the Romance domain that might dispute this book's pre-eminence. But if one's yardstick is competence in carpentry, balance between theoretical insightfulness and organization of data, or similar criteria, a distinctly less favorable verdict will have to be turned in. Hidden away in an in conspicuous footnote is the following stunningly frank confession: 'Within these terms of reference [a 1955 paper published in Lg.], of course, the present work does not qualify to be called a loanword study. But we must not think there is 21 Through another irritating coincidence, the same year that saw the publication of H's monograph also witnessed the appearance of H. Geckeler's equally ambitious book Zur Wortfelddiskussion (inspired by the teaching of E. Coseriu), which goes in incomparably greater detail over much the same ground. Having recently reviewed Geckeler's book rather circumstantially (1974), I refrain from going into minutiae. But even some of the sidelines pur sued by the two scholars are similar and invite comparison—e.g. their curiosity, in the context of Trier's 'Feldtheorie', about Hjelmslevian glossematics, particularly as seen through the prism of B. Siertsema's well-known Amsterdam dissertation (characteristically, H used the first edition, of 1955, and Geckeler the second, of 1965). As if to complicate matters, the Russian linguist Scur was meanwhile busy, in the very early seventies, preparing his elaborate reaction (1971) to Trier's theory, with the result that we now have a late flowering of critical responses, outside Germany, to the 'Feldtheorie' and its progeny (sometimes subsumed under the label of a neo-Humboldtian trend). 22 Since Matore's own research ran within the groove of French monolingualism, H tested its presuppositions and applicability by consulting P. J. Wexler's studies on French railway terminology borrowed from English (1955, 1964). Ironically, there had meanwhile appeared in Vienna—again probably too late (1969) to be taken into account—Herbert Peter's companion study on Italian railway terminology in the 19th century; the novelty of the situation that crystallized in Italy was that three languages rather than just one—namely English, French, and German—provided ready-made models for the nascent Italian nomenclature. For many search ing critical observations, see R. Stefanini's long-delayed appreciation (1974) of Peter's thesis.


149

something paradoxical in these asymmetrical viewpoints. It is all part of the "historico-cultural" versus "linguistic" dichotomy that runs through contempor ary linguistics, producing gaps which I for one would dearly like to see bridged' (651). It can be said that, whereas earlier loanword studies (e.g. Bezzola 1925, DeForest 1916, and Terlingen 1943) were prepared in a climate of intellectual stability, H's book, conversely, testifies to a period of crisis and transition; academ ically, the person who prepared the first rough draft of Part 2 was simply not the one who, twelve or fifteen years later, was toiling on the completion of Part 5. The fact that a book of this magnitude was completed at all certainly counts as a credit to Hope's perseverance. REFERENCES ALSDORF-BOLLEE, ANNEGRET. 1970. Die lateinischen Verbalabstrakta der U-Deklination und ihre Umbildungen im Romanischen. (Romanistische Versuche und Vorarbeiten, 34.) Bonn. BENVENISTE, EMILE. 1959. Review of L'emprunt linguistique, by L. Deroy. Bulletin de la Societe de Linguistique de Paris 54:2.47-8. BEZZOLA, RETO R. 1925. Abbozzo di una storia dei gallicismi italiani nei primi secoli (750-1300): saggio storico-linguistico. Heidelberg: Winter. BLOOMFIELD, LEONARD. 1933. Language. New York: Holt. BRESLIN, MARGARET SINCLAIR. 1969. The Old French abstract suffix -ise: studies in its

rise, internal diffusion, external spread, and retrenchment. RPh. 22.408-20. CARVALHO, JOSE G. C. HERCULANO DE. 1953. Coisas e palavras: alguns problemas

etnograficos e linguisticos relacionados com os primitivos sistemas de debulha na Peninsula Iberica. Coimbra. -. 1967. Review of Los pueblos romanicos y su cultura popular: guia etnograficofolklorica, by W. Giese. RPh. 21.223-6. CONNORS, KATHLEEN. 1971. Studies in feminine agentives in selected European lan guages. RPh. 24.573-98. DEFOREST, JOHN B. 1916. Old French borrowed words in the Old Spanish of the twelfth and thirteenth centuries ... Romanic Review 7.369-413. ELCOCK, WILLIAM D. 1938. De quelques affmites phonetiques entre l'aragonais et le bearnais. (Toulouse dissertation.) Paris: Droz. FIRTH, J. R. 1935. Technique of semantics. Transactions of the Philological Society, 36-72. GAMILLSCHEG, ERNST. 1966-69. Etymologisches Worterbuch der franzosischen Sprache. 2d ed. Heidelberg: Winter. [Orig. ed.: 1926-28.] GARCIA, ERICA C. 1970. Gender switch in Spanish derivation (with special reference to -a → -ero, -o —> -era, -a → -in, -on). RPh. 24.39-54. GECKELER, HORST. 1971. Zur Wortfelddiskussion. Untersuchungen zur Gliederung des Wortfeldes 'alt—jung—neu' im heutigen Franzosisch. (Internationale Bibliothek fur allgemeine Linguistik, 7.) Miinchen: Fink. GEORGES, EMANUEL S. 1968. Past-participial nouns: their development from Latin to Romance. RPh. 21.368-91. . 1970. Studies in Romance nouns extracted from past participles, ed. by Jerry R. Craddock & Yakov Malkiel. (UCPL 63.) Berkeley & Los Angeles. GREENBERG, JOSEPH H. 1957. Essays in linguistics. Chicago: University of Chicago Press. GREIMAS, A. J. 1966. Semantique structurale: recherche de methode. Paris: Larousse. HASSELROT, BENGT. 1945. Du changement de genre comme moyen d'indiquer une relation de grandeur dans les langues romanes. Uppsala Universitets Arsskrift, 107-25. . 1957. Etudes sur la formation diminutive dans les langues romanes. Uppsala.

150

DIFFUSION

HAUGEN, EINAR. 1950. The analysis of linguistic borrowing. Lg. 26.210-31. HOPE, T. E. 1960. The analysis of semantic borrowing. Essays presented to C. M. Girdlestone, 125-41. Newcastle upon Tyne: King's College. ——. 1962-63. Loan words as cultural and lexical symbols. Archivum Linguisticum 14.111-21, 15.29-42. . 1965a. The process of neologism reconsidered, with reference to lexical borrowings in Romance. Transactions of the Philological Society 1964:46-84. . 1965b. L'interpretation des mots d'emprunt et la structure lexicale. Actes du 10e Congres International de Linguistique et Philologie Romanes, 149-55. Paris. HOUSEHOLDER, FRED W., and SOL SAPORTA (eds.). 1962. Problems in lexicography:

report of the conference on lexicography, Indiana University, Nov. 11-12, 1960. (UAL 28:2, part 4.) Bloomington: Indiana University. KAHANE, HENRY and RENEE. 1948. The augmentative feminine in the Romance lan

guages. RPh. 2.135-75. . 1973. [Analytical bibliography of their writings, by Angelina Pietrangeli.] Issues in linguistics, ed. by B. Kachru et al., 1-31. Urbana: University of Illinois Press. KVAVIK, KAREN H. 1975. Patterns of derivational affixation in a Romance dialect. RPh. 29.57-66. LEVY, ANITA KATZ.

1965-67. Contrastive development in Hispano-Romance of

borrowed Gallo-Romance suffixes. RPh. 18.399-429, 20.296-320. LYONS, JOHN. 1963. Structural semantics. (Publications of the Philological Society, 20.) Oxford: Blackwell. MACKENZIE, FRASER. 1939. Les relations de l'Angleterre et de la France d'apres le vocabulaire. 2 vols. Paris: Droz. MALKIEL, YAKOV. 1957. Review of Coisas e palavras, by Herculano de Carvalho. Lg. 33.54-76. [Included in Essays on linguistic themes, by Y. Malkiel, 281-310 (Oxford: Blackwell, 1968), under the title 'Words, objects, images: shapes, makes, and names of the flail in Portugal'.] . 1957-58. Diachronic hypercharacterization in Romance. Archivum Linguisticum 9.79-113, 10.1-36. [Included with addenda in Linguistica generate—filologia romanza—etimologia, by Y. Malkiel (Firenze: Sansoni, 1970).] . 1972. The pan-European suffix -esco, -esque in stratigraphic projection. Papers in linguistics and phonetics to the memory of Pierre Delattre, ed. by Albert Valdman, 357-87. The Hague: Mouton. . 1974. Review of Zur Wortfelddiskussion, by H. Geckeler. Foundations of Language 12.271-86. . 1975a. Etymology and modern linguistics. Lingua 36.101-20. . 1975b. Review of Linguistique generate et romane, by Bertil Malmberg. To appear in General Linguistics 15:4. MATORE, GEORGES. 1953. La methode en lexicologie: domaine francais. Paris: Didier. MEYER-LUBKE, WILHELM. 1908. Historische Grammatik der franzosischen Sprache, 1: Laut- und Flexionslehre. Heidelberg: Winter. [2d and 3 eds., 1913.] MIGLIORINI, BRUNO, and ALDO DURO. 1950. Prontuario etimologico della lingua italiana.

Torino: Paravia. NEUVONEN, EERO K. 1941. Los arabismos del espanol en el siglo XIII. (Helsinki thesis.) . 1951. Los arabismos en las Cantigas de Santa Maria. Boletim de Filologia (Lisboa) 12.291-352. PETER, HERBERT. 1969. Entstehung und Ausbildung der italienischen Eisenbahnterminologie. (Wiener romanistische Arbeiten, 8.) Wien & Stuttgart: Braumuller. PIEL, J. M. 1966. Rev. of Historische Grammatik der franzosischen Sprache, 2: Wortbildungslehre, by W. Meyer-Lubke. Heidelberg: Winter. POSNER, REBECCA. 1970. Thirty years on. Supplement to An introduction to Romance linguistics, its schools and scholars, by I. Iordan & J. Orr, 2d ed. London: Methuen. PRATOLA, DANIEL J. 1952. Portuguese words of Italian origin. (Berkeley dissertation.)


151

SANDFELD, KRISTIAN. 1912. Notes sur les caiques linguistiques. Festschrift Vilhelm Thomsen, 166-73. Leipzig: Harrassowitz. SAPIR, EDWARD. 1921. Language. New York: Harcourt Brace. SCUR, G. S. 1971. Ob osnovnyx teorijax polja v jazykoznanii ['On fundamental field theories in linguistics']. Voprosy Filologii, ed. by N. Je. Uljanov, 81-126. Omsk: Gosudarstvennyj Pedagogiceskij Institut. SIERTSEMA, B. 1955. A study of glossematics: a critical survey of its fundamental concepts. The Hague: Mouton. STEFANINI, RUGGERO. 1974. Review of Peter 1969. RPh. 27.527-31. STEIGER, ARNALD. 1932. Contribucion a la fonetica del hispano-arabe y de los arabismos en el ibero-romanico y el siciliano. (Revista de Filologia Espanola, Suppl. 17.) Madrid. TERLINGEN, J. H. 1943. Los italianismos en espanol desde la formation del idioma hasta principios del siglo XVII. (Utrecht dissertation.) Amsterdam: N. V. NoordHollandsche Uitgevers Maatschappij. TRIER, JOST. 1931. Der deutsche Wortschatz im Sinnbezirk des Verstandes: die Geschichte eines sprachlichen Feldes, 1. Heidelberg: Winter. ULLMANN, STEPHEN. 1953. Descriptive semantics and linguistic typology. Word 9.225-40. . 1959. The principles of semantics. (Glasgow University publications, 84.) Glasgow & Oxford. VALKHOFF, MARIUS. 1931. Etude sur les mots francais d'origine neerlandaise. Amersfoort: Valkhoff & Cie. VIDOS, B. E. 1939. Storia delle parole marinaresche italiane passate in francese: contributo storico-linguistico all'espansione della lingua nautica italiana. (Biblioteca delL Archivum Romanicum, 2:24.) Firenze: Olschki. VOGT, HANS. 1953. Language contacts. Word 10.365-74. VOSSLER, KARL. 1913. Frankreichs Kultur im Spiegel seiner Sprache. Heidelberg: Winter. [Translated by A. Juilland: Langue et culture de la France (Paris: Payot, 1953).] WARTBURG, WALTHER VON. 1937. Betrachtungen liber die Gliederung des Wortschatzes und die Gestaltung des Worterbuchs. Zeitschrift fur Romanische Philologie 57.296-312. . 1943. Einfiihrung in Problematik und Methodik der Sprachwissenschaft. Halle: Niemeyer. [2nd ed., with Stephen Ullmann, 1962. 3rd ed., 1970.] 1946. Problemes et methodes de la linguistique [translation of the preceding item]. Paris: Presses Universitaires de France. [2nd ed., with Stephen Ullmann, 1963.] WEINREICH, URIEL. 1953. Languages in contact: findings and problems. (Publ. of the Linguistic Circle of N.Y., 1.) New York. WEXLER, PETER J. 1955. La formation du vocabulaire des chemins de fer en France (1778-1842). Geneve: Droz. [Rev. by Percival B. Fay, RPh. 10.362-7 (1957).] . 1964. Frein: the naming of a vehicle-brake. Cahiers de Lexicologie 5:2.69-83. WIND, BARTINA H. 1928. Les mots italiens introduits en francais au XYP siecle. (Amsterdam thesis.) Deventer: A. E. Kluwer.

F. LANGUAGE VS. THE REAL WORLD

GENDER, SEX, AND SIZE, AS REFLECTED IN THE ROMANCE LANGUAGES

1. Preliminaries Let me attempt to circumscribe from the start the intended scope of this paper. It is to contain no announcement of any sensational discovery. The threads that lead from the examination of grammatical gender to the study of naive statements about size, or of suggestions of size, with some background attention paid to sex, allow us to piece together the fascinating story of a protracted scholarly dialogue. The discussion started in low gear almost a century ago; it reached a much higher plateau of intensity from the early 'twenties to. the late 'forties, and some of the participants of that debate are still with us, while others passed away not so long ago. New voices, including young ones, have been heard over the last few years, though no new chorus has yet emerged. My paper, then, concerns a classic issue in Romance linguistics, one which abounds in implications for general linguistics — an issue distinctly better defined and understood at present than it was half a century ago, but by no means exhaustively solved in all its complexity. I shall exercise my right to offer, here and there, bits of criticism and certain slight qualifications, but I shall not presume to settle all such ramifications of the chosen problem as have been left dangling. The relation of sex to gender — in those languages which accept gender as a separate grammatical category - varies from culture to culture. In this respect, the configuration of English, in which, poetry and allegory apart, he stands for 'male', she for 'female', and it for 'inanimate', is fairly exceptional. Some major languages of the world, German included, do not even possess separate terms to distinguish be tween 'gender' and 'sex', between 'masculine' and 'male', or between 'feminine' and 'female'; the same lexical items, Geschlecht, männlich, and weiblich, each perform the same dual function. Russian separates 'gender' from 'sex' (rod vs. pol), but not, e.g. 'masculine' from 'male'. The Indo-European languages are, thus, characterized by a partial overlap of 'gender' and 'sex'. This overlap is particularly easy to observe in the older (or modern but tradition ally-patterned) Indo-European languages. In Classical Latin, for instance, a grammati cally masculine word may designate a male animate being (say, puer 'boy') or some inanimate object (say, lapis 'stone'); and the same holds for the feminine (witness puella 'girl' vs. mensa 'table'); neuters do not refer to animate beings in Latin (typical examples being helium 'war' and mare 'sea'); but in Russian they may occasionally do so (e.g. ditja 'baby, child'), although in all other relevant typological respects Russian coincides with Latin. The gender in practically all Indo-European languages can often be recognized by sharply characterized forms of qualifiers that accompany the nouns at issue, including articles if these are available, or by pronouns substituted

156

LANGUAGE VS. THE REAL WORLD

for these nouns; in certain languages the ending of the given word supplies reliable or at least probabilistic concomitant information. Thus, in Latin, despite a few tricky cases, such as socrus for 'mother-in-law' and nurus for 'daughter-in-law' (or, for that matter, domus for 'house'), the ending -us is normally a safe guide to the masculine gender, as in hortus 'garden' and exercitus 'army'; also, -a in the singular, except for tiny, neatly-delimited groups of exceptions, is practically a guarantee of feminine gender, though the state of affairs is entirely different in the plural, where -a is associated with the neuter, far beyond the confines of Latin. Other endings are inconclusive as to gender. In Spanish and Tuscan you find to this day a very similar situation; it is, incontrovertibly, el tio 'uncle' ~ la tia 'aunt', lo zio ~ la zia, but what about arte and fin(e)? Small wonder that here predictability is at its lowest and the decisions of the speakers sometimes appear not only arbitrary but hesitant, within the same language — as between Sp. el arte gotico, but las bellas artes or between It. la fine 'end' and il fine 'purpose' — but even more so if one compares one language with others: Sp. la sangre 'blood' vs. Ptg. o sangue, It. il sangue. In French, on the other hand, the shape of a noun, in general, and its final segment, in particular, cast only faint light, or no light at all, on the gender. We happen to know that oncle is masculine and tante is feminine, but it is not the res pective sequences of sounds that offer us these clues. Many more such elementary facts could be recapitulated here, but my assignment at this point calls for a switch of attention to the linguistic correlates of size. In the parent language, as well as in its congeners, special modifying suffixes were available for suggesting a half-hypocoristic (or -ironic), half-diminutive message; thus, regulus (from rex) could mean 'small king', 'charming little king', 'petty king', and the like. The daughter-languages developed a whole network of hypocoristic-miniaturizing and derogatory-augmentative suffixes; their genesis and diffusion is an intricate problem, into which we need not delve. Can size be linguistically suggested in yet other ways? Otto Jespersen, a firm believer in the effects of sound symbolism, once remarked in a celebrated short note that the vowel /i/, cross-culturally, was apt to evoke diminutiveness, an idea to which Joseph H. Greenberg, not so long ago, reverted, by no means disapprovingly, in his stimulating Essays in Linguistics. Meillet was willing to endow, at least in an individual language such as Latin, the a of the adjectival root-morpheme with a suggestiveness of comparable latitude. At this juncture there arises the question whether the inventory of size-suggestive devices, in a family or rather subfamily of languages such as Romance, could also have included the mechanism of gender-switch. For instance, could the existing label for a 'large container', particularly if that label was equipped with an ending sharply characterized as to gender (say, -a in Spanish or Italian), give rise to the name of a tag for a 'small container' through substitution of -o for -al Posed in these terms, the issue is difficult enough, involving as it does a grammatical plus a referential in gredient and appealing as it must to the straight linguist and to the student of material civilization steeped in Worter-und-Sachen analysis. The problem is bound to be further complicated if one adds the element of sex, as observed especially in humans and in larger, familiar animals. Is there, in the problem at hand, a component that, for lack of a more appropriate term, one might be tempted to call, with Damourette and Pichon, 'sexuisemblance'?

GENDER, SEX, AND SIZE...

157

It turns out that the entire multi-pronged issue is by no means new and that we have now sufficiently prepared ourselves not only for attacking it frontally, but also for critically inspecting and shifting the record of earlier tentative formulations. The one shred of information still misssing is that in the course of the transition from Latin to Romance the neuter disappeared in the class of nouns (though not of pronouns), merging frequently with the masculine in the singular and/or with the feminine in the plural — producing collectives, as in It. il dito 'the finger' beside le dita 'the five fingers of one hand' (or 'the five toes of one foot'), or casting off 'ambigenerics', i.e. words masculine in the singular and feminine in the plural, as in Rumanian. 2. The crystallization of the problem If you search for the earliest manifestation of originality in Romance linguistics, as displayed in the discovery of one particularly rewarding genre of research, you will perhaps agree with me that the ferreting-out and classification of doublets (doppioni, Scheideworter) was one such experiment or, if you prefer, fad. There was an ingredient of parlor game in this type of inquiry which, launched almost simultane ously by a minor figure, Auguste Brachet, and by the great comparatist Michel Breal around 1868, spread like wildfire, catching the imagination not only of Romance linguists, such as Ugo Canello, Carolina Michaelis, and Francisco Adolfo Coelho, but of outsiders as well, witness the British etymologist W. W. Skeat. Ordinarily, doublets (and, in exceptional cases, triplets) were coexistent reflexes - typically, with semantic differentiation - of the same ancestral bases, involving different channels of transmission (say, learned vs. vernacular or borrowed vs. indigenous), as in Sp. macula 'blemish' vs. mancha 'spot' or in Fr. rigide 'strict' vs. raide 'stiff, tight, taut'. But once curiosity about such pairs had been sensitized, it was possible to transfer it to similar configurations of semantic units. o

Thus, when the Swedish pioneer dialectologist A. W. Munthe, working in Western Asturias in the mid-'eighties, recorded in his field notes bintanu 'small window' alongside bintana 'window', the local counterpart of Sp. ventana, the brilliant Portuguese linguist Goncalvez Viana, reviewing Munthe's Upsala dissertation in the opening volume of Revista Lusitana a few years later, was led to expatiate on the entry, distinguishing between two kinds of such semantically nuanced and formally polarized pairs in standard Portuguese: either the feminine was preexistent and the masculine, secondarily arrived at, involved diminution, as in caldeira 'caldron, kettle', canela 'shin bone', cortiga 'cork bark', vs. caldeiro 'small kettle', canelo 'horse shoe', cortigo 'beehive'; or the masculine was preexistent and the feminine, representing the innovation, was augmentative, as in sapato 'shoe' and pogo 'well', vs. sapata 'low shoe, brakeshoe' and poga 'shallow pool, pond, puddle'. This laconic comment, running to a few lines, contained the seeds of one of the most celebrated issues in the annals of Romance linguistics.1 1) While Portuguese dialects later moved out of the center of discussion, O. de Pratt, in 1911, offered a very similar shred of information: Minhoto janêlo, in less careful speech jinêlo 'peephole, small window', beside janela 'window' [from IANUA 'threshold'] (Viano-do Castelo

158


Meanwhile, the year 1883 had witnessed the appearance on the scene of a young scholar who was soon to influence events in the Romance field, and of a book of major, if indirect, relevancy to the issue under scrutiny. In his doctoral disserta tion Wilhelm Meyer - the future Meyer-Liibke - concerned himself with the vicis situdes of the Latin neuter in Romance. Such facets of the problem as the long coexistence of, say, folium, sg., leaf and folia, pl., 'leaves, foliage' (cf. Fr. feuil ~ feuille, as in OFr. chevrefoil 'honeysuckle' beside feuille) or brac(c)hium 'arm' and brac(c)hia 'arms' (cf. Fr. le bras, the anatomical term, as against la brace 'fathom'); or as the relation, in Tuscan, of the singulars dito 'finger', osso 'bone', uovo 'egg' to the peculiar collectives in le dita, le ossa, le uova\ or as the crystallization in Rumanian of the two ambigeneric nominal paradigms into which eventually even words of Hungar ian and Anatolian Turkish ancestry were pressed - oras m. 'town' /orase f. 'towns', dulap m. 'closet' / dulapuri f. 'closets' - these and related issues, bearing on collectivity and augmentation, were to acquire considerable prominence in the discussion of lexical pairs differentiated as to size. On the Indo-European side, Meyer-Lubke's study likewise fell into place very smoothly, because here scholars of the rank of Johannes Schmidt also, independently, busied themselves, in the late nineteenth century, with the formation of the neuter plurals. In his Italienische Grammatik of the year 1890, Meyer-Liibke himself reaped benefit from his earlier probings, first, by citing several pairs of nouns where the masculine, in -o, designated a smaller object than the corresponding feminine, in -a: ber(r)etto 'cap, headgear', buco 'hole', bugnolo 'straw-plaited basket', canestro 'wicker basket, fruit basket, washing basket', cavicchio 'bolt, peg, wooden pin, rung of a ladder', cerchio 'circle, ring', fiasco 'glass bottle (with straw covering)', fungo 'fungus, mushroom', orezzo (poet.) 'light breeze, cool shade' (291f.), and, second, by observing, in another, not closely related passage (303), that feminine primitives are accompanied by genuine diminutives in -ina (e.g. casa 'house' - casina 'small house', carrozza 'car' — carrozzina 'small car') where the corresponding -ino derivative normally reserved for this purpose has accidentally become semantically deflected (casino, carrozzino). Meyer-Liibke treated both points briefly and superficially,2 and, alas, was to pay scant attention to the problem and its ramifications in his subsequent writings.3 Perhaps as a result of this relative indifference on the part of the tone-setting specialist, the discussion, for thirty long years, was to pick up momentum very and adjoining areas), likening it to such regionalisms as cancelo beside cancela 'low, wooden gate'; sete-estrelo 'the Pleiades' beside estrela 'star'; and panelo alongside panela 'pot, pan, kettle' (see 158b-159a). 2) Thus, he did not ask himself the key question whether only primitives ending in -o or -a can produce relevant counterparts, and, unlike Goncalvez Viana just two years earlier, he did not bother to distinguish between masculines giving rise to feminines and vice versa. 3) It is particularly surprising that as late as 1921 - in Part 2 of his innovative Historische Grammatik der franzosischen Sprache, where he grouped derivational suffixes in semantic or functional clusters - Meyer-Liibke should have left the augmentative feminine unmentioned as a rival device in discussing -as f-asse), -dtre, and -aud (§§172-5), an omission not remedied in J. M. Piel's disappointing supplement to the 1966 ed. of his teacher's handbook. MeyerLubke's dictionary contained, of course, a number of relevant entries (e.g. CANISTRUM/ KANASTRO, CISTUS, SACCUS), but the material offered in them was not so organized as to cast into bold relief the processes here studied.


159

slowly. To be sure, individual observations, on a small scale, were made here and there, e.g. by Karl Jaberg in his first introduction (1908) to dialect geography, when he discussed the maps chaudron and chaudiere of the Atlas linguistique de la France. In reference to Occitan usage (which marks the masculine with zero and the feminine with -o), Jaberg confirmed that pairol ordinarily denotes the 'small kettle' and pairolo the 'large kettle'. Pairol, by way of exception, may denote the large kettle if preceded by the qualifier grd, unless an entirely different word is pressed into service for the 'small kettle'. Gillieron himself, the mastermind of the glottogeographic movement, focused attention in a celebrated excursus, ten years later, on a separate, but related problem: In reference to very young girls, a masculine form of one of the available diminutive-hypocoristic suffixes is attached to their names, transmuting Louise into Louison, Marie into Marion, Suzette into Suzon, this despite the fact that the less disquieting use of -onne, or -ette, or -ine would, so it seems at first glance, have been admissible. This stimulating side issue has produced a sizable corpus of literature of its own, into which it is tempting to plunge - but we must curb the temptation; suffice it to state that through Gillieron's sparkling essay gender switch and derivational suffixation were for the first time closely associated. The time was now ripe for a livelier course of events, and a dialogue soon ensued. The year Gillieron's study, appended to his monograph on the Gallo-Romance names of the bee, appeared in print, a Catalan polymath or poligrafo, M. de Montoliu, published in a Barcelona journal - to be specific, in the Bulletin of the Catalan Institute — a short list of regionalisms recorded in the rural environs of Tarragona. Two consecutive entries, given no particular prominence, were llangost 'locust' and llangosta 'somewhat larger locust'. These entries captured the attention of a young and, at that juncture, still little-known Swiss Romanist, W. von Wartburg, who, conceivably under the continued impact of his teacher Gillieron's sensational monograph on entomonymy, asked himself, in a tightly-packaged note which ap peared three years later, whether the root of the suggestion of larger size by the feminine variants was traceable to the names of living beings, e.g. insects, and could thus, in the last analysis, go back to man's — specifically, an untutored speaker's — observations of the facts of nature. At this point, in 1921, the so far unsuspected possibility of a biological element or kernel began to loom. After briefly considering the wisdom of this explanation, which he called 'object ive' and which we would tend to associate with conditions of 'real life', Wartburg repudiated it, in part because biological facts failed to support it — male locusts, at least those stationed in Spain, happen to exceed their females in size — in part, more significantly, because the majority of lexical pairs related to inanimate referents, e.g. Cat.anell 'ring' vs. anella 'large ring'. 'Une difference naturelle ne peut done plus en etre la cause', argued Wartburg, casting around for other examples of names of tools and containers, such as Sp. bieldo ~ bielda 'rake, pitchfork', panero ~ panera 'breadbasket', 'baker's basket', saco ~ saca 'sack, bag', with the feminine member of the pair consistenty labeling the bigger object (cf. also Fr. rdteau 'rake' vs. Gasc. rastero 'big rake').4 Wartburg selected as the starting point for his — strictly 4) Typically Gillieronesque was Wartburg's quip that plato 'plate, dish' failed to generate

160


grammatical - explanation the collectively nuanced plurals of the Latin neuters — prdtum 'meadow' vs. prāta 'meadows', as reflected in Fr. pre vs. prée, arguing that the mass-noun meaning could gradually have changed into an augmentative meaning: from 'several adjoining small meadows' there was but a step to 'single large meadow': 'On s'est e'videmment figure le tout comme compose de plusieurs morceaux. Puis, peu a peu, l'idee de la pluralite s'est effacee et I'on a concul'e'tendue des pres comme un tout'. After choosing saco ~ saca as the presumable leader words among the designations of tools and containers and recognizing in Sp. dedo 'finger, toe' beside Ast. deda 'big toe' an advanced stage of the development, Wartburg rightly observed that this pattern, from the start, was far more weakly developed in French than in Occitan and the languages of the Iberian Peninsula, but made a serious blunder by flatly declaring that the model was absent from Italian and, presumably, from Rumanian - a surprising remark indeed after Meyer-Liibke's well-documented state ments to the contrary, at least as regards Tuscan. On later occasions a more experienced and better-read Wartburg would revert to this topic, but his subsequent pronounce ments no longer formed links in the principal chain of events.5 Wartburg's 1921 note - provocative and, in part, clearly misleading as it was did not at once stir up any controversy; thus field workers lucky enough to discover further vestiges of such lexical pairs failed to refer to it - it was not ignored but simply overlooked.6 But the related issue of the paradoxical development of the suffix -one continued to worry scholars throughout the 'twenties. In addition to the *plata 'large dish' because the right to the occupancy of that latter word's niche had been pre empted by plata 'silver'. In sober fact, had *plata 'large dish' sprung into existence at an early date, it might have blocked its homonym's entry and strengthened the precarious position of ARGENTUM to the south of the Pyrenees. 5) Wartburg was, of course, again and again reminded of his early note by hundreds of articles, stretching over a half-century, in his monumental Franzosisches etymologisches Worterbuch. He refrained from including the note, in its original or in any revised form, in his selection misleadingly subtitled collection - of articles, Von Sprache und Mensch; gesammelte Aufsdtze (Bern: Francke, 1956). Wartburg's later 'Stellungnahme' to this problem - a summary and an elaboration at the same time - is, I regret to report, neither fair nor effective; see Einfuhrung in Problematik und Methodik der Sprachwissenschaft 68 and Problemes et methodes. . . 66-7 (tr. P. Maillard); even in the 3d German edition (1970), brought up to date by P. Ineichen and enriched by a temporary program of collaboration with S. Ullmann, the results are unsatisfactory. Wartburg cites his 1921 note without admitting its basic error as regards Italian; there is a reference to the shorter and less significant of the two papers by Hasselrot, while the crucial studies by R. Toole and H. and R. Kahane - closer to Wartburg's own position - are swept under the rug; the author cites enthusiastically an opinion by J. Grimm and declares himself unconvinced by Brugmann's strictures, without either stating these objections or, at least, identifying the place where a curious reader can find them. Above all, it remains unclear under what conditions purely formal or semantic circumstances determine the association of gender with size (and strength?). 6) Thus, F. Belloni-Filippi, 'La lavorazione di un frantoio idraulico in territorio di Buti (Pisa)', after recording a few conspicuous cases - canala 'un grande canale di acqua', coltella 'piu grande e robusta del coltello', barroccina 'piu ampia e commoda del barroccino' (233-4) mused: 'Curiosa questa tendenza di mettere al femminile gli accrescitivi'. He was apparently unaware of holding in his hand two trump cards: the refutation of Wartburg's arbitrary exclu sion of Italy, and the proof that a masculine in -e could also be flanked by an augmentative feminine in -a.

GENDER, SEX, AND SIZE. . .

161

slightly piquant problem of the French hypocoristics Louison, Madelon, Marion, etc. there was that other puzzle familiar to every traveler touring Europe: Whereas in France carafon is a small and carafea.standard-sized decanter, in Spain garrafon denotes a large and garrafa a standard-sized carafe. How can one reconcile these diametrically opposed evolutionary lines? It. caraffa, as an old liquid measure, stands alone; were it accompanied by an appropriate satellite formation, Italian usage would, on circumstantial evidence, harmonize with Spanish preferences, leaving French alone. But aligning majorities is not yet offering a satisfactory ex planation. In 1921 E. Gamillscheg credited the semantic peculiarity of Fr. -on to Germanic influence, which indeed was stronger in Northern Gaul than in most other provinces. Leo Spitzer, in a paper attached to Gamillscheg's, opted for a witty, psychologizing interpretation of the facts. Eight years later, Gamillscheg reversed himself: he was now prepared to distinguish, in the parent language, between an individualizing suffix -o, -onis, attached — among other uses — to certain animal names, and a miniaturizing suffix -id, -ionis, appearing in the names of small, flitting living beings (cf. papiliō 'butterfly' and, judging from Fr. poisson, *piscio 'fish', originally 'small fish, fishlet', which had been allowed to spread at the expense of Class, piscis preserved elsewhere - It. pesce, Ptg. peixe, OSp. pece, etc.). The two suffixes merged, and the second, being more copiously represented in Northern France, left its semantic stamp on Fr. -on, making carafon in the end a small rather than a large decanter. 3. The peak of the discussion In addition to engaging in research, Gamillscheg, after his transfer to the University of Berlin in the mid-'twenties, doubtless devoted lectures to these topics and to others germane to them. Against this background and within such an intellectual climate, it becomes understandable why one of his best students of those vintage years, Renee Toole, should, in her doctoral dissertation completed by 1931 and published in 1934, as a slim but meaty book on a point of French dialect geography, have prepared an excursus of over fifteen pages on the topic under discussion. Her analysis, distinctly more original and sophisticated also on the side of documenta tion, and better buttressed than anything previously offered, produced few im mediate results, for a variety of reasons,7 even though it was to trigger certain long-range effects. From the start, after briefly surveying the few available statements on the chosen category of word pairs, the author demolished Wartburg's hasty denial of the exist ence of this pattern in Italian (324). She then offered well-arrayed selective lists culled from Italian (34-7), Occitan (37-9), Hispano-Romance (40), and French 7) For one thing, the monograph series in which this exciting dissertation appeared was hardly known for its high quality; see my necrological essay on E. Gamillscheg; for another, the title of the dissertation led readers to expects little more than a routine interpretation of one map (818: MARMITE) in Gillieron's atlas, with side glances at a few other collaterally useful maps (715: JATTE; 1065: POT; 1509: COCOTTE). Clearly, to spark discussion, Gamillscheg should have persuaded and helped his star pupil to separate the pan-Romanic excursus from the bulk of the booklet and to channel it through some widely-read journal.

162


(40-1) sources; each section was so organized that primitives came first and derivatives involving so-called modifying (in most cases, diminutive) suffixes followed. For example, under the rubric of Italian, one finds, at first, an inventory of pairs structured like capanno 'small cabin' ~ capanna 'cabin', cucchiaio 'spoon' ~ cucchiaia 'big spoon' and, later, others architectured like chies-in-o 'tiny church' vs. chies-in-a 'small church' (primitive: chiesa), colt-ell-o 'knife' vs. colt-ett-a 'big knife' 8 The advantage of this approach, within the frame ofRenee Toole's thinking, will become apparent at once; one of its liabilities, for the diachronically-oriented linguists, is the author's failure to distinguish between original feminines that gave rise to masculines, such as capanna, and the reverse process. Rene'e Toole's analysis of the data assembled (41-3) contained several daring ideas. She felt that the switch from a collective to an augmentative meaning was, generally speaking, rare (witness the record of the minutely-examined suffix -dta), but that two special circumstances expedited the transition in this particulair in stance: semantically, the prevailing tridimensionality of the object referred to; and formally, the fact that the overwhelming majority of the cases involved nouns already equipped with a diminutive suffix, so that the typical feminine member of a word pair designates the slightly larger of two varieties of a miniaturized image: '. . . so dass also die feminine Form die Bezeichnung eines ausdriicklich kleinen Raumes in semantischer Hinsicht vergrossert'. Straight augmentative suffixes (say, -one or -accio) could not very well have been pressed into service as an alternative here, because they would or might have been neutralized by the preceding diminutive suffixes. Let me immediately add, by way of criticism, that Renee Toole's illustra tions have almost all been extracted from dictionaries - both monolingual, like Tommaseo-Bellini, and bilingual, like Rigutini-Bulle; she did not properly weight the factor of actual incidence, paying equal attention in her statistics to very com mon and exceptionally rare words, not a few of which may even have been manu factured by overambitious lexicographers. Another original idea of Rene'e Toole's flowed from her preceding qualification of the switch from collective to augmentative. The number and diversity of the factors involved allowed, in a third evolutionary stage, never before isolated, a new function to develop: the feminines in -a (or > -e) become paler, retaining a merely modifying, limiting, nuancing rather than discernibly augmentative force, as was previously hinted at by F. Hanssen: 'Das Nebeneinanderbestehen der Endungen -o und -a wurde dann als Mittel zur Schaffung von Bedeutungsdifferenzen gebraucht' (129). There was henceforth less emphasis on the names of containers as primitives, i.e. a lighter stress on tridimensionality. After offering another assortment of illustra tive examples from Italian, Old Provencal, Occitan, Spanish, and French (43-5), once more organized according to her favorite principle, Renee Toole comes up with a big surprise: At the extreme fringe of the nuancing function (Stage III) one 8) Forty years ago scholars did not operate with the concept of suffixoid, so it is hardly . surprising that Renee Toole made no effort to build a bridge from cucchiaia, which to the average speaker has no identifiable primitive, to the more transparently carpentered derivatives in -aia, which she lists separately under -ARIUS. It is puzzling, however, why coltell-o, -a, equally "monolithic" - by her standard - for the uninitiated, should appear under -ELLUS rather than among unsegmented primitives.

GENDER, SEX, AND S I Z E . . .

163

finds isolated cases of the feminine partner, in such word pairs, denoting the smaller object. Such stray examples of the downright diminutive function of the feminine are found in Italian, where tina designates a smaller 'vat' or 'tub' than does tino\ the same is true of bacchetta 'rod, staff, navicella 'small craft', and rapa 'turnip' vis-a-vis bacchetto, navicello, and rapo, respectively. The chances for the feminine in -a to end up as a diminutive are relatively best where the primitive has been previously expanded by the augmentative suffix -on-: thus, if one starts out from camera 'room' and casa 'house', a camer-on-a and a cas-on-a turn out to be somewhat smaller than a camer-on-e and a cas-on-e. The symmetry to the earlier case of chiesin-a ~ chies-in-o is transparent. One recognizes something approaching the configura tion of a quadripartite cycle: I. Collective; II. Weakly-Augmentative; III. Nuancing; IV. Weakly-Diminutive. A brilliant schema indeed, argued with passionate persuasi veness and in scintillating German by a very young woman, barely twenty-four years old, and one born in a different country and into a different culture.9 Through a bizarre twist of circumstances, Mauricio Schneider's paper — dedicated to Amado Alonso — appeared on this side of the Atlantic the same year as Renee Toole's excursus did in Europe. The two investigations, whose authors were unaware of each other's involvement, shared one more peculiarity: They aroused little im mediate attention, and no third party rushed to the foreground of events to offer any bold synthesis. The title of Schneider's monograph-length article was descriptive: 'El colectivo en latin y las formas en -a con valor aumentativo en espanol'. Perhaps the claims Schneider staked out were, on balance, too modest. In addition to canvass ing the feminine augmentive (62-86) he also neatly set off its occasional counterpart, the masculine diminutive (86-8), as when standard Spanish tolerates the innovation ruedo 'small sphere' alongside traditional rueda 'wheel', from ROTA, or as when the Santanderino dialect in northern Spain accepts anguilo 'name of a fish smaller than the eel' beside anguila 'eel', from ANGUIL(L)A. There is a more satisfactory balance between the Latin prelude and the Romance drama in Schneider's article than in any other study that has come to my attention, partly so because his main emphasis is on the early evolutionary stage, marked by hypertrophy of mass-noun function. Thus, he devotes a whole chapter - rather elaborate at that — to the socalled 'heterogeneous' nouns in Latin (we might be tempted to call them 'ambigeneric'); i.e. nouns with a recorded gender switch from singular to plural, which to him marked the starting point of the entire development, e.g. acinus 'berry', with tendential semantic differentiation; also armentum 'herd', caementum 'rough stone from the quarry', nervus 'sinew', whose plurals display all sorts of irregularities in the suggested direction. Finally, Schneider scrupulously combed the latest Spanish Academy Dictionary then available as well as numerous Old World and New World dialect dictionaries of Spanish in search of fresh data, coming up in the end with a real corpus, however limited, rather than with randomly selected illustrations. Unlike Renee Toole, he attributed no particular 9) As her philosophical guide to lexico-derivational problems of language (history), Renee Toole took the Polish scholar Jan Rozwadowski's once influential book Wortbildung und Wortbedeutung, on which Wartburg and other experts leaned heavily as late as the mid-'forties.

164


importance to any division of the Romance words affected into primitives and diminutives. The impetus of Bengt Hasselrot's eventual involvement in this problem is due to the fact that his main purpose, patently, was not to criticize or qualify opinions voiced by any predecessors, but to offer a sliver of original thinking. In the early 'forties, the Swedish scholar was slowly, but firmly and methodically, at work on his magnum opus, the rise and spread of certain Romance diminutive suffixes, chiefly -ittus and variations thereupon. This project was to keep him in a state of alertness for over thirty years; he clearly was in no hurry. The extraordinary thorough ness of his data-gathering is evident throughout the first of his two monographs entitled Du changement de genre comme moyen d'indiquer une relation de grandeur. In that slender 1944 tract, he added to the inventory of relevant languages Sardinian and at least two islets of Rhae to -Romance; drew fine lines between the separate usages of Gascon, Provencal proper, and Franco-Provencal rather than lumping together shreds of evidence under some such indiscriminate label as Occitan; used dialect monographs and glossaries, guides to hydronymy, phonetically-transcribed field notes (in part unpublished), and maps of dialect atlases, in addition to more conventional sources of information; and ferreted out numerous long-hidden hints in a wide spectrum of scholarly treatises.10 On the technical side, his treatment showed unrivaled mastery in "pinpointing", and maturity of judgment as well. Hasselrot's starting point coincided approximately with Rene'e Toole's midpoint and altogether transcended Schneider's horizons. Taking for granted the expectation that the suffixally-derived diminutive in Romance will normally tend to adopt the gender of the primitive, he isolated three groups of exceptions,11 then focused atten tion (108) on the one group involving pairs of diminutives clashing in overtly marked gender and referring to objects of varying size — with the feminine variant ordinarily matching the larger of two paired-off objects. However, in the word-list that he actually supplied (108-12), he no longer distinguished carefully between primitives and suffixal derivatives, placing Sp. charco - chorea 'puddle' beside Cat. ganivet ganiveta 'penknife', and he did not use as a classificatory criterion the etymological priority of the masculine over the feminine partner, or vice versa, in such binomial relationships. In the analytical section of his monograph, Hasselrot initially accepted the standard explanation, surmising that the feminine augmentative in Romance rested on a col lective function, and that neuters like prātu 'meadow' ~ prāta 'meadows', '*one large meadow' paved the way for masculines like hortu(s) ~ V.Lat. horta 'garden(s)', muru(s) ~ V.Lat. mūra 'wall(s)'. He sensed that Classical Greek and certain Semitic languages also illustrated the semantic switch from mass noun to augmentative, 10) On gender switch in suffixal diminutives the earliest — to this day important — pronoun cements by Indo-Europeanists seem to be K. Brugmann's 1906 note, 'Der Genus der Diminutivbildungen', and its critique twenty years later by H. Jakobsohn. 11) The two groups less worthy of sustained attention involve either those primitives which, unlike their satellites, have undergone a fairly late change in gender (thus, Fr. amourette echoes older [fern.] amour) or those suffixal derivatives that have undergone the influence of their near-synonyms, as when Fr. dial, baquette 'small vat' (Ardennes) echoes the ending (and, ines capably, the gender) of tinette 'small tub'.


165

for feminines. A few pages farther down, however, Hasselrot — and herein lies his real originality — rebelled against the feminine augmentative, calling its use abnormal, or, as we would say, counter to expectation, counterintuitive. Using as his guides Jakob Wackernagel's celebrated lectures on comparative syntax and several bold syntheses made over a quarter-century span by the noted Africanist Carl Meinhof, Hasselrot now protested that in most healthy cultures the larger size associatively goes hand in hand with the evocation of male sex and, through the widespread overlap of sex and gender, with the masculine gender. Listen to his own words: Le féminin augmentatif, bien qu'assez repandu comme nous venons de le voir, constitue tout de meme, reflexion faite, une anomalie. Dans plusieurs langues, c'est au contraire le masculin qui peut assumer une fonction augmentative et, puisque rhomme, normalement, est plus grand et plus fort que la femme et que le meme rapport existe entre le male et la femelle dans la plupart des especes, cela parait plus normal, plus conforme a la na ture des choses.

At this point Hasselrot, in the wake of Hermann Jakobsohn (1926), cited Franz Boas's sketch of Chinook, an Amerindian language which makes large animals mas culine and smaller animals feminine, as well as African languages classing designations of females and of smallish males with those of inanimate objects. He also joined Grimm, Wackernagel, and Wartburg who, unlike Brugmann, all three operated with the assumption of a far-going "sexualisation des objets du monde inanime", as seen in the names of countless artisans' tools, in the technical parlance of sailors very explicit in this regard, etc., though, let me add on my own, it is not so much differ ences of sheer size as those of configuration which appear to have been decisive in matters of nomenclature.12 The conclusion that Hasselrot drew from the findings of yet other scholars — botanists, students of ancient medical and veterinary science, and toponymists included — was that the situation was more complex than had been anticipated by the pioneers. There has existed in Romance, on the one hand, an augmentative feminine, which throve on an internal, grammatical process — the disappearance of the neuter and the various consequences which that event entailed; and, on the other hand, there has, independently, developed an augmentative masculine, rooted in ever-present, ever-active personification and sexualization, to which, upon occa sion, there has further been added, as a third ingredient, the interplay of the designa tion of inanimate objects with suffixes ordinarily used to tag cubs and whelps. In Spain, a small river will be known as riato, from rio, and one tributary of the Duero will be called Durato, involving -ato, the label for young animals. In France, the streams flowing through the villages Auzon,Leuc, and Hon will be known as l'Auzonnet, le Lauquet, le Hogneau, but the streamlets emptying into those streams will be I'Auzonnette, la Lauquette, and la Honnelle. 'Nous avons affaire ici a une combinaison des phenomenes de personnifications et de sexualisations', argued Hasselrot, 12) An even greater — almost wild — allowance for the physiological and psychological implica tions of sex in grammar was made in many of his writings by Leo Spitzer; perhaps more explicitly than elsewhere in his study (1921) of epicene nouns in Romance. For a more recent and, all told, far more judicious assessment of a closely-related (in part interlocking) problem see Anita Katz Levy's 1973 paper on plural form vs. singular meaning.

166


I'affluent de I'Auzonnet, p.ex., etait considere, pour ainsi dire, comme sa femme et sa femelle'. The conflict between the two patterns can be resolved in many striking ways. Ordinarily each speech community decides whether the masculine or the feminine is principally associated with aggrandizement. Because the augmentative feminine is solidly entrenched in standard Spanish and Portuguese, these languages show only minimal traces of the augmentative masculine; but individual Hispano-Romance dialects may steer a different course. In Spanish, jarra 'earthen jar'refers to something bigger than jarro', but in Western Asturian, xarro denotes a bigger container than xarra, and Upper Aragonese offers seven instances of the augmentative masculine. In Franco-Provencal, the spread of the augmentative masculine encountered weak resistance, because in Gallo-Romance the augmentative feminine happened to be underdeveloped. In conclusion, Hasselrot admitted that, despite these subtle regional variations, a measure of ineradicable ambiguity had crept in. Once the natural, biological asso ciation had become grammaticized and lexicalized, it was in the way of the pratum ~ prdta scheme, which, from the start, had been a strictly grammatical device. Since neither mechanism succeeded in completely dislodging its rival, they neutralized each other, and their initially-promising advance ground to a standstill. They lost out to the more explicit - and, let me add on my own, more expressive - augmenta tive and diminutive suffixes: 'L'opposition masculin-feminin ne peut done, en fin de compte, remplir convenablement la fonction d'indiquer sans ambigu'ite une rela tion de grandeur'. In the penumbra of the gradual withdrawal or shrinkage of this device, there crystallized certain hazily-delineated relationships between doublets, in which the -o partner and the -a partner (to cite the Italian and Spanish carriers of the contrast) were semantically differentiated along an axis other than that of size. Thus, It. berretto and calzetto refer to garments worn by men, while berretta and calzetta perform the opposite service. This zigzag takes us back, as Hasselrot could not fail to observe, to Renee Toole's 'nuancing function'; but what to her had been a fairly early link in the chain of events impressed Hasselrot as being, most plausibly, the concluding link. During the war years and immediately thereafter, no sustained dialogue ensued; for a short while it seemed to matter little whether a paper by one investigator preceded or followed upon a paper by another. No major close-knit theory nearly so innovative as Toole's and Hasselrot's was presented by any party; but minor gaps in documentation were randomly filled and an occasional trial balloon was allowed to float. The corpus of Catalan examples was slightly expanded in 0. Bernhard's otherwise undistinguished Zurich dissertation. G. Gougenheim presented a short list of modern French formations in -ette, some of them of very recent vintage and frankly experimental, which seem to lack a masculine counterpart; though the paper was written for and channeled through an American journal, the author was unaware of the impact of mock-French American on genuine French usage in the climate of fraternization peculiar to 1944-45, though a hybrid like gangsterette should have alerted him to his possibility; after all, in this country a blonde and a divorcee even now normally refer to a woman, and petite has been borrowed, while petit has not. 13 The well-known Latinist Marouzeau, on one of his periodic forays into con-

GENDER, SEX, AND SIZE.. .

167

temporary French, raised a number of questions without providing explicit answers, though there was no dearth of hints in his comments. He felt that, in dealing with the standard language, psychologizing explanations with sexual overtones were inadequate to account for change in grammatical gender: Fr. la mer 'the sea', he argued, underwent the shift from neuter to feminine under the pressure of its se mantic opposite la terre 'the earth' (and in substandard speech Vair 'air' followed suit),14 without the slightest implied reference to a woman's charm, tempestuous beauty, or fickleness. In examining specimens of racy speech, however - on the level of 'francais argotique' - Marouzeau was less certain of his point: in numerous stereotyped phrases involving faire and etre 4- noun; in characteristic sequences such as se la fouler, se la mettre, on se la casse 'on fuit'; in newly-coined facetious postverbals, say, la creve (from crever 'to burst') and I'epate (from epater 'to baffle, dazzle'); in derogatory references to individuals, especially to males: canaille, crapule, fripouille, gouape, nouille, and in certain other sharply delimited groups the feminine was used almost exclusively, which could hardly be due to coincidence. Had Marou zeau been writing thirty years later, he might have allowed for a streak of "male chauvinism" in French society, at least at certain levels or in certain contexts. In any event, Marouzeau's inconclusive note added a sociolinguistic or stylistic dimen sion to the problem. Finally, the discussion of the ticklish French hypocoristics Louison, Madelon continued unabated; the controversy, started by Gillieron in 1918, now aroused the attention of Dauzat and, predictably, of Spitzer, who offered heavily speculative and psychologizing interpretations.15 When the dust settled on the war and on its immediate repercussions, the debate was resumed on an ambitious scale. The next item in my chronologically-arrayed bibliography, a forty-page article, straddling the years 1948 and 1949, by Henry 13) Appended to Gougenheim's note is a brief and light-weight editorial comment - presumab ly, on circumstantial evidence, from the pen of Leo Spitzer. The two-way link between Am.-E. and Fr. -ette was better grasped by B. Hasselrot in his second attack (1950) on the problem. For the most successful attempt to place the agentives in -ette within an appropriate cultural framework see Kathleen Connors' 'Studies in feminine agentives. . .'. 14) Marouzeau, a professional Latinist but a mere amateur in the Romance domain, was apparently oblivious of the use of la mar (obsolete, poetic, figurative) in Spanish where, in the absence of rhyme, no appeal can be made to any irresistible pressure of tierra. The influence of TERRA could have been, at most, a concomitant, as I pointed out shortly thereafter in my paper on lexical polarization. 15) To reconstruct in detail Spitzer's zigzagging line of thinking one would have to examine a raft of his studies, spanning a full quarter-century, from 'Entwicklung des Gegensinns. . .' (1921) via 'Feminization. . .' (1941) all the way to his lengthy essay on hypocoristics. The pervasive factor of his reasoning was the assumed primacy of sensual associations over gram matical mechanisms, a hierarchy which in 1950 shocked even the outspokenly "realistic" Hassel rot. Specifically, in his 1946 article Spitzer expatiated on Gillieron's excursus and attempted to show that Margot, Ninon suggested 'des garcons manques'. Had he had access, during the war, to A. Dauzat's earlier note in FM 9.161-70, he might have thought differently: Dauzat stated (convincingly, I think) that Marion, (Mar)goton, Margot pertain to amorous parlance, in which the interversion of gender forms part of the general cajolery and persiflage (I can cite on my own the colloquial use of Sp. jfeo/, /negro! in the sense of 'handsome!'). Dauzat adduced as a parallel grotesquerie the emotional stress in epOUvantable; note also the many corrections in his review {FM 15.75-6) of Spitzer's article.

168


and Renee Kahane, titled The augmentative feminine in the Romance languages', was, and has remained, the single most elaborate piece of research devoted to this problem and in certain crucial respects, even in today's perspective, marks the summit of the entire controversy. It also carried with it a few mild surprises, insofar as, first, Mrs. Kahane turned out to be the very same person as Miss Toole; as, second, here was for the first time a contribution written and published in English, albeit by two scholars steeped in European learning; and as, third, the article was editorial ly processed by a newly-founded American journal.16 The thoroughness and maturity of the Kahane paper, on the separate levels of documentation and of analysis, makes it indeed a finely-chiseled masterpiece. The fact that after such a long search they found no vestiges of the phenomenon under study in Balkan-Romance, including Rumanian, has become an important negative characteristic of word-formation in that bundle of dialects. Their survey of earlier, necessarily sketchier treatments of the process is, on the whole, fair in verdicts and restrained in tone. The one link in the chain a reader misses is an at tempt at detached self-criticism: What did the Kahanes themselves, from their vantage of 1948, find worthy of acceptance, of modification, or of excision in Renee Toole's 1931-34 excursus? Despite the tribute they paid to Hasselrot, the Kahanes, as they were approaching the mid-century point, basically adopted and fortified Renee Toole's earlier position. The presentation of the material remained, in essence, identical, though numerous data, this time, were culled from dialect glossaries and dialect maps, an. opportune deviation from earlier practice in the dual direction of authenticity and specificity. The concept of "nuancing function" originally toyed with was now replaced by that of "discriminative function or meaning", for which a separate, generouslyexpanded dossier was now prepared, which in turn invited a good deal of accurate geographic pinpointing. The rare instances of the diminutive feminine were once more, with increased authority, recognized as constituting just an unusual shade of the discriminative service performed by -a. Finally, on the side of I'historique du probleme, all interpretations so far advanced were divided into two major classes: the strictly grammatical ones, championed by Wartburg, Toole, Bernhard, and Schneider, and the staunchly psychological ones, spearheaded by Damourette and Pichon (after an earlier advocacy by the seldom-remembered pioneer R. de la Grasserie in 1904), with Spitzer and Hasselrot emerging as less than entirely successful would-be mediators. The school of psychology involved was, let me add at once, 16) I hope I am committing no breach of confidentiality by testifying to the fact that the authors, though initially eager to see their vigorously expanding paper appear in the opening issue of the journal, later curbed their impatience in exemplary fashion and waited until all the evidence, down to the minutest detail, had been scrupulously assembled and sifted. This thoroughness at the concluding stage of the operation must be superadded to the eighteen years or so of previous thinking and data-gathering on the part of Renee Toole Kahane. The echoes to the Kahane paper, as listed in A. Pietrangeli's retrospective bibliography, were favorable but, on the whole, faint. M. Garcia Blanco provided a detailed summary in Spanish; M. Paiva Boleo dramatized the chief paradox and cited the authors' own conclusion; B. Pottier stressed the difference between the rich yield of the device in Spanish and Italian and its relative meagerness in French; G. Rohlfs was satisfied with praising the authors' industriousness (the accolade he had given fifteen years earlier to Renee Toole's thesis excursus was warmer and less hasty).

GENDER, SEX, AND S I Z E . . .

169

conjectural and speculative. The Kahanes' personal sympathies lay indisputably with the camp of grammarians. They neatly defined, without resorting to polemics, their limited disagreement with Hasselrot, whom they — rightly or wrongly - viewed as influenced by Kuhn: The strongest objection to Kuhn's and Hasselrot's interpretation seems to lie in chrono logical considerations: the diminutive feminine appears so much later than the augmenta tive feminine that it is doubtful whether in its incipient period natural philosophy still influenced Romance gender, (p. 169)

With Henry and Renee Kahane's balanced analysis, proffered in measured terms, the discussion, I repeat, reached its all-time peak. Unfortunately, Hasselrot yielded to the temptation of once again (indeed, more than once) entering the fray. His chief revision, summation, and rebuttal are contained in a 1950 article published in the Swiss journal Vox Romanica. What mars this second attack on the problem — which, had it been written in English, might well have borne some such title as "Gender Switch Revisited" - is that most of the space is given over to the circum stantial refutation of an extremely silly remark dropped and, unfortunately, later reiterated by Dauzat. The net result of the controversy is to confirm that Occitan preponderantly coincides, as was to be expected, with Spanish and Italian in favoring the augmentative feminine over its masculine counterpart. In certain zones (e.g. Savoy and French Switzerland) one encounters a good deal of confusion as between one village and another. On the positive side, one notes Hasselrot's success in filling one minor gap in his earlier documentation: Central Rhae to-Romance (specifically, the Fassa Valley) was now firmly credited to the territory of the augmentative feminine. In 1950, Hasselrot voiced the belief that Northern France initially likewise belonged to that area, but that the derivational model was subsequently eroded to the north of the Loire through the vigorous spread of the augmentative masculine, so that a sort of neutralization and ultimate decay ensued: 'Le masculin augmentatif et le feminin augmentatif s'equilibraient a peu pres et se nuisaient ainsi mutuellement au point de s'entretuer' (p. 141). As regards the kernel of the disagreement, Hasselrot, in 1950, was unwilling to yield ground on the issue of the evocation of biological facts: 'Le masculin augmentatif, he averred, 'doit son origine a un rapprochement entre genre et sexe, a une comparaison toute naturelle et spontanee entre le regne inanime et la relation de grandeur et de force qui existe entre male et femelle dans la plupart des especes animales qui interessent rhomme et surtout entre rhomme et la femme eux-memes'. With respect to the feminine augmentative, he accepted the standard hypothesis of a collectively-colored ancient plural neuter as the bridge to the Latin ending -a, but, apropos names of containers, he toyed with the possibility of animization, indeed, personification: 'Pour ce qui est des re cipients, il peut aussi reposer sur une comparaison entre le[s] corps des humains des deux sexes' (p. 143). To put it differently, Hasselrot was willing to operate with anatomical imagery; on the other hand, he resolutely rejected as unproven, indeed, undemonstrable, Spitzer's attempt to drag in popular notions, however suggestive, of a female's psychology. He next considered the relevant pages of Orion Bernhard's and rejected, perhaps not strongly enough, that scholar's bizarre phono-symbolic

170


explanation of augmentative -a.11 Only the last four pages dealt with the article of the Kahanes, which received unstinted praise on the side of documentation, but was mercilessly criticized on the level of causal and chronological interpretation. As regards particulars, Hasselrot's rebuttal corrects a few factual inaccuracies; then reasserts his fundamental belief in a steady interaction of grammatical pressures (today some of us conceivably would use instead the word "constraints") and ever-present animistic forces, which could perhaps have been unleashed in the inter mediate zone between the animate and the inanimate realms, e.g. in hydronymy and in vernacular phytonymy. Hasselrot appeals also to folklore and even to learned allegory (['opinion, reine du monde) in his search for collateral support. Reduced to a still sharper, more abstract formulation, and all details apart, Hasselrot argues that the Toole-Kahane hypothesis hinges, in the last analysis, on a single and unique state of affairs in Late Latin: prāta denotes not only something larger than pratum, but also, at least potentially, something different from prātum, whereas his own interpretation is essentially binary. A second ineradicable contrast between the two principle theories: Whereas the Toole-Kahane analysis attaches major importance to the fact that the given word either constitutes a primitive or is endowed with certain modifying suffixes, Hasselrot disregards this circumstance entirely, but feels, by way of compensation as it were, that the semantic cleavage in the development of a few such suffixes, particularly -one and -otto, now diminutive now augmentative (de pending on the dialect), constitutes a relevant parallel to such fluctuations as aug mentative feminine vs. diminutive feminine - in terms of language universals.18 Thus one finds a syntagmatic, pitted against a paradigmatic, appeal to derivational suffixes within the frame of our problem. Hasselrot's last-mentioned remark was no doubt designedly programmatic, be cause he was at that point actively, almost feverishly, at work on his magnum opus on Romance diminutive suffixes - perhaps the last, enduring classic of old-style Romance scholarship — a book in which he reverted incidentally to the problem here at issue, as he did in a number of shorter pieces, simultaneous or subsequent.19 17) Hasselrot struck a cavalier attitude toward Bernhard's bold appeal to phonosymbolism, but failed to point out just what made that appeal so grotesquely inopportune. One may argue about the propriety of so interpreting a stressed vowel, including a (in It. -actio, Sp. -azo, Fr. -as, say), but the application of this principle to unstressed final vowels receives no support from any side. Another fantastic idea of Bernhard's was to justify the link between the names of domestic animals and the device of the augmentative feminine with the argument that the females of those species, though actually of smaller size, often happen to be more useful to man, hence more valuable, and thus loom bigger. 18) Hasselrot also feels that the referential contrast evoked by gender and by the light endings -o vs. -a (or their local counterparts) need not be slighter than the contrast suggested by the use of the heavier standard derivational suffixes, such as -ino. 19) In his contribution to the Rohlfs Testimonial Volume, Hasselrot, aside from surveying, briefly or in searching detail, earlier studies (by Tobler, Gillieron, Dauzat, Spitzer, Jaberg, Rohlfs, Herbillon, Bratto, Michaelsson, Jakobsohn, and others), proffered one original idea: The flowering of the declensional type Hue(s) (subj.) ~ Huon (obi.), patterned on a Germanic model, in Northern France more than in any other Romance country, unavoidably restricted in that territory the range of -on as a hypocoristic suffix to the feminine sector of the onomasticon. Whereas Marion was apt to express 'cute little Mary', Huon could not perform a parallel service with regard to 'Hugo'. Interestingly, Hasselrot mentioned (204) the insertion of an 'antesuffixe' (i.e. interfix)


171

But there were no longer to come any dramatic confrontations with the TooleKahane hypothesis, while the Kahanes themselves, of late, have apparently regarded their mission as completed, except for details that may have emerged at the surface in their far-flung etymologicon. The involvement of derivational suffixes, whether in paradigmatic or in syntagmatic perspective, makes it incumbent upon me to mention very briefly several brilliant studies which may someday be called upon to play a major role if and when the discussion of our major issue reopens. Erica Garcia, using in part statistics, in part introspection as a native speaker of American Spanish, has associated gender switch in suffixal formations with the establishment of a wider semantic gap between primitive and derivative: hierarchically, papeleta is more autonomous vis-a-vis papel than is papelito, and the relation of ventana 'window', ventanilla 'small window', and ventanillo 'peephole' can be similarly defined. Suzanne Fleischman has thrown a bridge from the denominal augmentatives in -on to the action nouns in -on suggestive of sudden, violent movements, such as tiron; similar problems have attracted the attention of a noted senior scholar, Marcel Bataillon, a rare visitor to the precicnt of linguistics. Karen H. Kvavik has observed the fact that, in the class of Modern Spanish nouns, most primitives are masculine, while most suffixal derivatives are feminine, and has drawn exciting tentative conclusions from her observation.20 Sufficient ammunition has once more become available for the battle to be resumed; let me qualify: a strictly cognitive battle.

4. A tentative conclusion Let us attempt to draw a provisional balance sheet. Have the eighty years or so of intermittent discussion produced any tangible results? I believe that the answer should, at least in part, be positive. In all likelihood, all the major causative factors and all or, at least, most of the relevant perspectives in which the growth is amenable to observation have been identified; but the all-important concatenations of these factors and of the events produced by them remain, in most instances, a matter of continued dispute. Some of these concatenations, the simplest to describe, involve cogent chronological sequences: By what margin (if any) did the augmentative masculine precede or trail behind the augmentative feminine? Others refer to like -es(s)-, ~ign-, or -ill- as the one device that could be seized upon to protect the hypocoristic value of -on under such circumstances. Ch. 8 of Hasselrot's Etudes sur la formation diminutive. . . is also concerned with the various functions of Fr. -on. In his last major book (1972), confined to present-day French {Etude sur la vitalite de la formation diminutive francaise au XXe siecle), Hasselrot reverted, again and again, to such favorite themes as 'disaccord en genre entre derive et mot base', 'genre indice de grandeur' (see 16f., 60, 64, 79, 101), expatiating on earlier analyses of his own. 20) Kvavik's review article, 'Patterns of derivational affixation in a Romance dialect', aside from its intrinsic value, commands interest because it digests several other writings of the author, in press (one article) or not scheduled for publication (her Stanford dissertation). In addition, it summarizes and criticizes the ideas and data included in the book at issue, which happens to be a monograph of my own: Patterns of derivational affixation in the Cabraniego dialect ofEast-Central Asturian (1970).

172


compatibility: To what extent do aggrandizement and miniaturization exclude each other? Despite the successive appeals to statistics, phonosymbolism, animization, sexuality, and what not — only substratum influences have not yet been invoked — we still have no truly satisfactory answers to such basic facets of the problem as the absence of the augmentative feminine from Rumanian; its rarity in Northern French, as against its superabundance in Occitan and Catalan, to say nothing of Italian and Spanish; the syntagmatic or the paradigmatic role of competing augmentative and diminutive suffixes; and the wisdom of accepting or rejecting a binary schema: grammar vs. personification, which of course would imply the interplay of linguistic and extra-linguistic forces. Should the dichotomy 'marked vs. unmarked', whose use in other contexts has, upon occasion, produced splendid results, be experimen tally tried out in our impasse? And is the word-final contrast 'central vowel vs. back vowel', characteristic of the state of affairs in Spanish and Italian, to be placed on exactly the same level as the contrast 'zero vs. any vowel' which grosso modo describes the situation in Gallo-Romance, Catalan, and Modern Rumanian? The inventory of side issues left unsolved or, perhaps worse, incompletely solved or, still worse, misleadingly formulated, could be generously expanded. The balance sheet here drawn, let me conclude, is not atypical of Romance spatio-temporal linguistics, to the extent that it is explicative, i.e., concerned with causation. Many problems have been raised; some have, in the process, been made to appear excessively complicated through the discovery of all sorts of collateral issues and interferential factors; relatively few have in fact been definitively solved. This situation is, I imagine, apt to produce two radically different reactions on the part of younger researchers. Either they will decide to disregard and scuttle quite recklessly the corpus of earlier studies, so as to be unimpeded in their attempts to strike out anew, preferably through use of a fresh approach and a novel termino logy, at the heavy price of cutting the thread of continuity. Or they must agree to add to their bold and ever-changing explorations, as a separate dimension, the sophisticated probing of earlier hypotheses, be it only as an exercise in the history of science or as a token of self-discipline. If this alternative is chosen — as, I think, it should be —, then few problems will command higher interest and priority than the record of studies in gender, sex, and size as reflected — or refracted — in Romance.

REFERENCES A. General (in alphabetic sequence) Damourette, Jacques, and Edouard Pichon. 1930-49. Des mots a la pensée: essai de grammaire de la langue francaise. 7 vols. Paris: Collection des linguistes contemporains. Greenberg, Joseph H. 1957. Essays in linguistics. Chicago: The University of Chicago Press.


173

Jespersen, Otto. 1922. Symbolic value of the vowel i. Philologica 1.1-19. Prague & London. (Reprinted in Meisterwerke der romanischen Sprachwissenschaft, ed. by Leo Spitzer, 1.53-68. Miinchen: Hueber.) Meillet, Antoine. 1928. Esquisse d'une histoire de la langue latine. Paris: Hachette. Rozwadowski, Jan. 1904. Wortbildung und Wortbedeutung: eine Untersuchung ihrer Grundgesetze. Heidelberg: C. Winter.

B. Special (in chronological sequence) 1868. Brachet, Auguste. Dictionnaire des doublets, ou doubles formes de la langue francaise. Paris: A. Franck. 1871. ; Supplement [au Dictionnaire des doublets]. Paris: A. Franck. 1873. Coelho, Francisco Adolfo. Formes divergentes de mots portugais. Romania 2.281-94. 1883. Meyer[-Liibke ] , Wilhelm. Die Schicksale des lateinischen Neutrums im Ro manischen. Halle: M. Niemeyer. 1887-89. Viana, [A. dos R.] Gongalvez. Review of A. W. Munthe, Anteckningar om folkmalet i en trakt af vestra Asturien (diss. Uppsala). Revista Lusitana 1.27985, at 283. 1889. Schmidt, Johannes, Die Pluralbildungen der indogermanischen Neutra. Weimar: Bohlau. 1890. Meyer-Lübke, Wilhelm. Italienische Grammatik. Leipzig: 0. R. Reisland. 1906. Brugmann, Karl. Der Genus der Diminutivbildungen. Indogermanische Forschungen 19.215-6. (Cf. Grundriss der vergleichenden Grammatik. . . 22:1. 609.) 1908. Jaberg, Karl. Sprachgeographie; Beitrag zum Verstandnis des 'Atlas linguistique de la France'. Aarau: H. R. Sauerlander. 1910. Hanssen, Friedrich. Spanische Grammatik auf historischer Grundlage. Halle: M. Niemeyer. 1911. Pratt, O. de. Linguagem minhota. Revista Lusitana 14.145-68. 19[ll-]20. Meyer-Liibke, Wilhelm. Romanisches etymologisches Worterbuch. Heidel berg: C. Winter. 1918. Gillieron, Jules. Gene'alogie des mots qui designent I'abeille, d'apres l'Atlas linguistique de la France. Paris: Bibliotheque de 1'École Pratique des Hautes fitudes. (Excursus 13 [Suffixes masculins dans les pre'noms feminins] was included in Meisterwerke der romanischen Sprachwissenschaft [2 vols.], ed. by L. Spitzer, 1.279-89. Miinchen: Max Hueber.) . Montoliu, M. de. Petit vocabulari del camp de Tarragona. Butlleti de dialectologia catalana 6.38-51. 1920-24. Wackernagel, Jacob. Vorlesungen über Syntax, mit besonderer Beriicksichtigung von Griechisch, Lateinisch und Deutsch. 2 vols. Basel: Kommissionsverlag von E. Birkhauser. 1921. Wartburg, Walther von. Substantifs feminins avec valeur augmentative. Butlleti de dialectologia catalana 9.51-5.

174


. Meyer-Lubke, Wilhelm. Historische Grammatik der französischen Sprache, 2: Wortbildungslehre. Heidelberg: C. Winter. (2d ed., revised by J. M. Piel, 1966.) . Gamillscheg, Ernst. Grundziige der galloromanischen Wortbildung, in: Gamillscheg & Spitzer, Beitrage zur romanischen Wortbildungslehre. Bibliotheca dell'Archivum Romanicum 2:2. Geneve: L. S. Olschki. . Spitzer, Leo. Uber Ausbildung von Gegensinn in der Wortbildung, in: Gamillscheg & Spitzer, Beitrage zur romanischen Wordbildungslehre. 1926. Jakobsohn, Hermann. Review of J. Wackernagel, Vorlesungen uber Syntax. . . Gnomon 2.369-85 (see esp. 375f.). 1929. Gamillscheg, Ernst. Zur Frage der Auswahl bei der suffixalen Ableitung. Behrens-Festschrift: Dietrich Behrens zum siebzigsten Geburtstag dargebracht. Zeitschrift fur franzosische Sprache und Literatur, Suppl. 13. Jena & Leipzig: W. Gronau, W. Agricola. (Reprinted in E. G., Ausgewahlte Aufsatze [1]: Festschrift zu seinem fiinfzigsten Geburtstage, ZFSL, Suppl. 15 [1937],) . Belloni-Filippi, Ferdinando. La lavorazione di un frantoio idraulico in territorio di Buti (Pisa). L'ltalia dialettale 5.226-38, at 233f. 1934. Schneider, Mauricio. El colectivo en latin y las formas en -a con valor aumentativo en espanol. Boletin de la Academia Argentina de Letras 2.25-91. . Toole. Renee. Wortgeschichtliche Studien: toupin und bronze. Berliner Beitrage zur romanischen Philologie 3:4 (dir. E. Gamillscheg). Jena & Leipzig: W. Gronau. 1941. Spitzer, Leo. Feminization del neutro. Revista de Filologia Hispanica 3.33971. . Dauzat, Albert. Les interversions de, genre a valeur affective; les hypocoristiques du type Marion, Margot: leur explication. Le Francais Moderne 9.161-70. (Reprinted in Melanges de linguistique francaise 52-61 [Paris: D'Artrey, 1946]. 1943. Bernhard, Orion. La formation de nombres por sufijos en Catalan. Zurich diss. Zurich-Altstetten: Buchdruckerei H. Schraner. . Wartburg, Walther von. Einfiihrung in Problematik und Methodik der Sprachwissenschaft. Halle: M. Niemeyer. (3ded. [revised by G. Ineichen], Tubingen: M. Niemeyer, 1970.) 1944. Hasselrot, Bengt. Du changement de genre comme moyen d'indiquer une relation de grandeur dans les langues romanes [1]. Sprakvetenskapliga Sallskapets i Uppsala Forhandlingar 1943-1945, pp. 107-25. 1946. Gougenheim, Georges. Les feminins diminutifs en français moderne. Modern Language Notes 61.416-9. . Marouzeau, Jules. Un aspect du fe'minin français. Le Français Moderne 14.241-5. . Spitzer, Leo. Suffixes masculins dans les prenoms fe'minins en français. The Romanic Review 37.127-49. 1948-49. Kahane, Henry R., and Renee T. The augmentative feminine in the Rom ance languages. Romance Philology 2:2-3.135-75. 1948. Dauzat, Albert. Review of Hasselrot, Du changement de genre... [1]. Le Français Moderne 16.76. 1950. Hasselrot, Bengt. Du changement de genre comme moyen d'indiquer une relation de grandeur dans les langues romanes [2]. Vox Romanica 11.135-47.


175

1951. Malkiel, Yakov. Lexical polarization in Romance. Language 27.485-518. 1957. Hasselrot, Bengt. Etudes sur la formation diminutive dans les langues romanes. Uppsala Universitets Arsskrift 1957:11. 1958. Hasselrot, Bengt. Le type Marion, Louison - une monstruosite' lexicale? Romanica: Festschrift fur Gerhard Rohlfs, 200-7. Halle: M. Niemeyer. 1965. Craddock, Jerry R. A critique of recent studies in Romance diminutives. Romance Philology 19.286-325. (Includes rev. of Hasselrot, Études sur la forma tion diminutive. . , and of all critical reactions to that monograph [287-310].) 1970. Garcia, Erica C. Gender switch in Spanish derivation (with special reference to -a→-ero, -o→-era, -a→-in, -on). Romance Philology 24:1.39-54. . Malkiel, Yakov. Patterns of derivational affixation in the Cabraniego dialect of East-Central Asturian. University of California Publications in Linguistics 64. Berkeley, Los Angeles, London: University of California Press. 1971. Connors, Kathleen. Studies in feminine agentives in selected European lan guages. Romance Philology 24:4.573-98. 1972. Hasselrot, Bengt. Etude sur la vitalite de la formation diminutive francaise au XXe siecle. Studia Romanica Upsaliensia o. 1973. Fleischman, Suzanne. Collision of homophonous suffixes entailing transfer of semantic content: The Luso-Hispanic action nouns in -on and -dela/-dilla. Romance Philology 26:4.635-63. Levy, Anita Katz. Plural form versus singular meaning in Hispano-Romance nouns. Romance Philology 27:1.13-25. . Malkiel, Yakov. One short-lived genre of glottohistorical research. Romance Philology 26:4.749-51. . Malkiel, Yakov. Ernst Gamillscheg (1887-1971) and the Berlin School of Romance linguistics (1925-1945). Romance Philology 27:2.172-89. Pietrangeli, Angelina. The writings of Henry and Renee Kahane: an ana lytical bibliography, § §2 and 87. Issues in linguistics: papers in honor of Henry and Renee Kahane. Urbana: University of Illinois Press. 1974. Bataillon, Marcel. Bofetones mecanicos. Estudios filologicos y linguisticos: Homenaje a Angel Rosenblat 97-109. Caracas: Instituto Pedagogico. 1976. Kvavik, Karen H. Patterns of derivation in a Romance dialect. Romance Philology 29:1.57-66. 1979. Malkiel, Yakov. The analysis of lexical doublets: The Romanists' earliest contribution to general linguistics. Studia Gratulatoria, Robert A. Hall, Jr. 191196. Madrid: Playor.

G. THE SOCIAL COMPONENT OF CHANGE

THE SOCIAL MATRIX OF PALAEO-ROMANCE POSTVERBAL NOUNS

THE SPECIFIC mechanisms that could have led to the crystallization of the Romance postverbals (or, as some scholars prefer to dub them, deverbals) were satisfactorily established as early as the turn of the century, and the genetic explanations offered by such trail-blazers as Grober, Meyer-Lubke, Paris, and Tobler — to say nothing of Diez — to this day command respect. Nevertheless, a total revision of the older thinking is long overdue — on account of numerous minute facts that have since come to light, and for many additional reasons. For one thing, the rapid internal diffusion of the "clipped" verbal abstracts, against the background of the utter structural atypicality of the model, remains un explained; it certainly is not enough to appeal to ''regression", by citing a few alleged parallels, which upon closer inspection turn out to be of dubious rele vancy. The pioneers were apparently unaware of the wisdom of offering separate explanations for the RISE and for the SPREAD of any innovative feature. For another thing, the paradoxically incongruous roles assumed by postverbals in Latin and in the Romance daughter languages — especially with respect to the contingents of the verbs themselves — raise a problem of the first magnitude, which has, again and again, been virtually swept under the rug. The Romance postverbals cry out for an up-to-date book-length treatment, which obviously cannot be undertaken here. The single problem selected for fairly leisurely inspection is the SOCIAL MATRIX, the milieu which, once it has been tentatively identified on a conventional philological basis, bids fair to help us to answer a few of the more provocative linguistic questions. We shall, first, cursorily examine the highlights of earlier discussions, some of them admittedly disappointing. In the process, the question is apt to acquire a sharper profile. Our next task will be to propose a new hypothesis, concerning the specific social mooring of the slowly-emerging pattern. Finally, we shall turn our attention, for one quick look, from the large contingent of verbal abstracts to the small, noto riously controversial group of postverbal agentives. I. HIGHLIGHTS OF EARLIER ANALYSES

During the boldly exploratory, "romantic" phase of historical linguistics, the discussion of Romance postverbal nouns started on a lively note. As early as

180

THE SOCIAL COMPONENT OF CHANGE

1838, Diez not only delimited their domain with remarkable felicity of formula tion ("Ihre eigentliche Quelle ist die 1. Konjugation, ihre Form die des Präsens, ihre Bedeutung die des Infinitivs, also abstrakt"), but also neatly set off this particular field from the contiguous fields of nominalized participles and infini tives — an impressive accomplishment.1 As an imaginative if cautious scholar, Diez was perfectly aware of the resemblance (minus the element of apophony) of this rhizotonic derivational model to the one familiar from German, but pre ferred to think in terms of merely typological similarity: "...an diesen Substan tiven, welche sich von Seiten ihrer Einfachheit den deutschen durch Laut und Ablaut gebildeten (Band, Binde, Bund) nähern". Characteristically, he did not allow himself to be swept off his feet by this near-parallelism, and made a point of also citing counterparts from two classical languages, Latin (see below) and Greek. And when a far more enthusiastic — and, occasionally, less judicious — contemporary, namely Jakob Grimm, began to toy with the possibility of a genetic influence by Germanic on Romance, Diez, in revising his own grammar, tersely observed: Grimm, I I I , 785 fragt ob in solchen romanischen Wörtern wie fr. vol von voler, etc., nicht Einfluß des Deutschen, vgl. Flug von fliegen, etc., anzunehmen sei. Man darf mit Nein antworten, da mit dieser Art von Wortschöpfung schon das Latein vorangegangen war. 2

Interestingly, neither pioneer reckoned with the probability of a blend or second ary reinforcement arising from a Romano-Germanic cultural symbiosis. Diez was doubtless aware of one paradox, even though he chose not to empha size it: If the Romance postverbals belong, for the most part, to the -ar(e) class, such postverbals as he could, for the sake of comparison and reconstruction, adduce from Latin sources pertained, almost ironically, to the -ĔRE, the -ĒRE, and, if only at rare intervals, the -ĪRE class. Moreover, on the semantic side they did not, for the most part, involve abstracts, but referred to occupations, parts of plants, living beings, tools, etc. — in short, to the entire referential spectrum of nouns: 3 1

Grammatik der romanischenSprachen, II (Bonn, 1838), 232-235. The definitive 3d ed. goes back to 1870-72 and is available in several printings and in a French translation; cf. GBS5, II, 6l4nl. Perhaps advisedly — and, ifso, on good grounds — Diez refrained from mentioning coLLĒGA 'partner (in an act ofinvestiture in power)', which, as its nuclear vowel shows, cannot be traced directly to LĚGO -ĔRE 'to gather, read', coLLiGō -ĔRE 'to assemble, draw together'. It invites joint study with COLLEGIUM 'association, partnership' (both words pertained originally to the religious sphere). One faces here a grammatical dilemma: Either one espouses, with Bréal and Bailly, the hypothesis that COLLEGIUM was extracted from LEx 'law' on the model of cONFĪsriUM 'common boundary, neighborhood' from FĪNIS 'border, end' and ofcoNSORTiUM 'com munity of goods, fellowship' from soRS -Tis 'lot, share', in which eventuality coLLĒGA must be declared a derivative one more step removed from LEx, arrived at as was coNVĪVA 'table compan ion' from coNvīviuM 'living together, social feast, banquet, "party'" (see theseauthors' Did. étymol. lat.). Or, and this is the alternative to which A. Ernout seems to lean (DÉLL±, p. 354a), one starts out from LĒGŌ -ĀRE 'to dispatch, send with a charge', arguing that coLLĒGA was ex tracted immediately from the verb as were ADVENA from ADVENĪRE and mcoLA from iNCOLERE (see below), cutting a groove for COLLEGIUM. The only drawback here, Ernout confesses, is the unavaila bility Of *COLLĒGĀRE. 3 Diez's illustrations are here reproduced in full, with glosses and elaborations culled from Lewis & Short, A Latin Dictionary (Oxford UP), after certain finishing touches were added to the spell2

THE SOCIAL MATRIX

181

-ĚRE: ACUERE 'to sharpen, wheť→ACirs 'needle, pin'; COQUERE 'to cook, bake, boil, roast, parch' → COQOUS 'cook'; IN-coLERE 'to dwell in, inhabit, cultivate' → iN-coLA 'resi dent'; CURRERE 'to run, move quickly, hasten' → CURRUS 'chariot, car';FALLERE 'to deceive, trick, betray' → (hapax legomenon) FALLA 'artifice, trick' ( = EALLĀciA); MERGERE 'to dip in, immerse, plunge into, sink' → MERGiJS (Varro) 'diver' [water fowl], (Columella, Palladius) 'layer of a vine'; PREMERE 'to press into or upon, force in'->PULLi-PREMA 'paederast'; RUMPERE 'to break, burst, tear, rend' → (Varro) RUMPUS 'vine branch, runner'; SCRĪBERE 'to write, draw' → SCRĪBA 'public writer, clerk, secretary, scribe'; TRAHERE 'to draw, drag (along), haul' ֊^ (Columella, Vulgate) TRAHA 'drag, sledge' = Vergil's TRAHEA; TRŪDERE 'to thrust, push, shove (forward), press on, impel' → (Vergil, Isidorus) TRÜDis 'pointed pole'; VEHERE 'to bear, carry, convey', 'to be borne, ride, sail' → VEHES 'carriage loaded with..., (cart-)load'; -ĒRE: ARCĒRE 'to shut up, enclose, keep in order, hold off, keep at a distance' → ARCA 'place for keeping anything; chest, box'; CALLĒRE 'to be thick-skinned, hardened, unfeeling, skillful, experienced' → CALL-us ~ - 'thick skin, hard flesh, callousness' ; cĒNSĒRE 'to tax, assess, rate, estimate' → cĒNSFS 'registry and rating of citizens or property'; SEDĒRE 'to sit' → PRO-SEDA (Plautus) 'common prostitute' [ = 'she who sits or lounges in public']; -ĪRE: AD-VENĪRE 'to come to a place, reach, arrive'→ AD-VENA 'foreigner, stranger'.

It seems to have slipped Diez's mind that two important Latin derivatives, at least, were traceable to -ARE verbs, namely LUCTA 'wrestling match', from LUCT-ĀRE/-ĀRĪ 'to wrestle, struggle', and PUGNA 'fight fist to fist', from PUGNĀRE 'to combat, contend' — a fatal omission, as this paper will attempt to demon strate. But the implication of some profound and unexplained discrepancy be tween the states of affairs in Latin and in Romance is all the more dramatic. There is evidence that Diez, for years, was deeply troubled by certain disquiet ing aspects of Romance postverbal derivation — even while writing the Intro duction to his etymological dictionary in the early 'fifties, he was haunted by such thoughts 4 — a restlessness which redounds to his honor. In the writings ing. Throughout the editions of his grammar the author kept one Greek example: αρχζιν 'to rule, be leader of' → άρχός 'chief, commander', but in the end substituted reívew 'to stretch, strain' → τόνο? 'rope, cord, board' for another, less representative pair. He refrained from commenting on the contrast in vowel quantity between TRŪDERE and TRŮDis. 4 Etymologisches Wörterbuch der romanischenSprachen (Bonn, 1853), "Vorrede", pp. xxiii f.; see my earlier paper here, "Friedrich Diez and the Birth Pangs of Romance Linguistics". Despite a number of immediately recognizable blunders, Diez's mid-century statement is worth summariz ing. Starting out from — parsimoniously represented — prototypes (scRĪBA, coQuus, nux, Ё), Diez distinguished three evolutionary lines within Romance: (a) Deverbal names ofpersons, in -o (or neutral -e, in French): It.furbo 'cunning man' (beside forbire 'to furbish, polish'), It. dial, lecco ['glutton', 'sponger'?] (beside leccare 'to lick') —these two he traced to Gmc. models; It. allievo 'pupil, nursling', Fr. élève ("mehr sächlich als persönlich"), Sp. trasgo 'imp, goblin' (beside trasegar 'to upset'; Diez admitted the inadequate fit), plus — Diez's trump card ("unläugbar") —FR.j 'judge' &\ongsidejuger 'to judge'; (b) Masculines in -a, produced through personification of an object, cf. — outside the field of deverbals — boia 'fetter' > 'executioner' (here Diez apparently telescoped Lat. BOIAE [pl.] 'de tachable collar' and It. boia 'hangman'). Other examples, in decreasing order ofplausibility: Sp. boga (m., f.) 'rower' beside boga 'vogue, rowing, stroke' and esp. Ptg. voga 'oar'; It. spizzeca 'tongs' ( = 'Knicke, Kneipzange') besides spizzicare 'to pinch, nibble' (the noun is ofdubious authenticity and, upon Diez's admission, of questionable meaning to boot); It.farfulla 'stammerer' besideFarfullare and Ptg. beberrica 'drinker, alcoholic', in the shade oïbeberricar; (c) Masculines in -a, produced through personification ofabstracts: Diez here cited, in addition to It. ascolta, scorta and Prov. uca, also Prov. bada ('Obacht' > 'Wächter'), crida ('Schrei' > 'Schreier'), and It. gonfia ('Aufblasung' > 'Glasmacher'), which wereto elude the attention of

182


of some among Diez's contemporaries one discovers all sorts of useful shreds of information on lexical bric-a-brac — but no truly important alternatives.5 A new chapter in studies of Romance postverbals began with the publication, in 1894, of W. Meyer-Lübke's Romanische Formenlehre, in which the author reserved ample space (§§397-402) for a discussion of all major facets of the prob lem. Meyer-Lübke must have recognized at a glance the built-in paradox in Diez's presentation of the antecedents, namely the pioneer's failure to adduce any pertinent examples from the ranks of the -ĀRE class, the one that mattered most at the post-Latin stage; but he did not bother to engage in polemics. Instead, he calmly proffered a new hypothesis, which became at once the stan dard doctrine. Meyer-Lübke — on that occasion and later6 — reckoned with two starting points. For one thing, numerous -ĀRE verbs were produced, within Latin, from a variety of nominal primitives, thus: LUTUM 'mud, mire' → LUTARE 'to bedaub, besmear' (with mud or clay); PLANTA 'sprout, shoot, twig, sprig, graft', 'sole of the foot' → PLANTARE 'to set, (trans)plant', 'to fix in place, form'; SERRA 'saw' → (Late Lat.) SERRARE 'to saw (up), saw into pieces'; SONUS -Ī (rarely -us) 'noise, sound' → SONARE 'to make a noise, (re)sound'.7 Given the frequency of this relationship, argued Meyer-Lübke, untutored speakers could easily be se duced into minting words in reverse direction, i.e., into starting out from an -ARE verb and experimenting with radical-stressed derivatives in -u or -A. For another thing, as the number of post-participial iterative and intensive verbs E. L. Adams and other successors; he allowed for the semantic opacity of the relation of[dial. ?] It. trecca 'greengrocer' to treccare 'to cheat, deceive'. In 1900, A. Tobler, in his review of G. Lené's monograph (see below), offered additionalillustrations of the shift — within the boundaries of French — "nomen actionis" > ''nomen agentis'' : ótage 'hostage', OFr. message 'mess-age, -enger', témoin 'witness', OFr. prison 'prison(er)'. Today it is known that Fr. élève is a mid-17th-c. adaptation of It. allievo (see NDÉH), which Migliorini and Duro liken, in turn, to ALUMNUS viewed as an outgrowth ofALERE 'to nourish, raise'. Fr. juge is, of course, a direct deposit of IŪDICE, influenced by jugier < IŪDICARE; iŪDiCE also survived in OSp. júez; later, under the impact of /we/, it was transmuted into juéz. 5 For one example see H. Rönsch, Itala und Vulgata; das Sprachidiom der urchristlichen Itala und der katholischen Vulgata2 (Marburg, 1875; the orig. ed. appeared in 1869), pp. 83-88. The author offers valuable clues to actual loci containing traces of FALLA, LUCTA, PROBA, and COMPUTUS and documents such compounds — to which Diez was also alert — as AQUI-, CON-, DĒ-, LŪci-, REFUGA; BUB-SEQUA; HERĒDi-, HONÕRi-PETA; but he fails to discriminate between abstracts, agen tives, and the remainder of characteristic words in -A and -us. 6 Compare the almost identical analysis in GRS, II : Formenlehre, §397, where the postverbals ranked as one variety ofvocalic derivation, and in HGFS, I I : Wortbildungslehre (Heidelberg, 1921), §§108-113, wherethey again occupiedthe niche of"Verbalabstrakta", except that the later word ing took cognizance ofthe key rôle ofthe molecule PUGNUS ~ PUGNARE ~ PUGNA (see below) and also pressedinto service such pairs as MOLA 'millstone' beside MOLĚRE 'to grind' and PLUViA 'rain' alongside PLUĔRE 'to rain'. The author's relative indifference to postverbals manifested itselfin their exclusion from the most urgent tasks of glotto-paleontology which he outlined in his tonesettingEinführungin das Studiumder romanischenSprachwissenschaft (Heidelberg, 19092), §§178190; cf. 19203, §§196-209. 7 The number of examples can, of course, be almost indefinitely extended; cf. MACULA 'spot, stain, mesh' → MACULARE 'to spot, stain'; MEDicus 'doctor, surgeon' → MEDic-ARE/-ARi 'to cure, heal'; MENDicus 'beggar' → MENDÏC-ARE/-ARÏ 'to beg, go begging'; siGNUM 'sign' → siGNÃRE 'to mark, stamp'; soMNIUM 'dream' → SOMNIÃRE 'to dream of, daydream about', etc. On DAMN-/ -ÄRE and LUCR-UM/-ÃRE see below.

THE SOCIAL RI

183

steadily increased, and as their distinctive semantic or functional feature faded more and more, it became tempting for speakers to associate not only CANERE 'to sing', but, secondarily, also its rival and eventual substitute CANTĀRE (orig. 'to sing vigorously or habitually') with CANTU, i.e., with the past participle of CANERE which, through a propitious coincidence, could also be mistaken for the verbal abstract CANTUS (orig. gen. -ūs) 'song, singing'.8 Further illustrations of this model, in Meyer-Lübke's thinking, included CURSĀRE 'to turn hither and thither' ~ cursus 'running, course, way, march' ; DUCTĀRE 'to lead, draw, con duct', 'to allure, deceive' ~ DUCTUS 'line, row, form, shape, structure'; iACTĀRE 'to throw, toss about, shake, drive about' ~ iACTUS 'throw(ing), cast(ing)'; SAL(i)TARE 'to dance, perform a pantomime' ~ SALTUS 'leap(ing), spring, bound'; *ūsĀRE (beside well-attested ūsiTĀRE) 'to use often, make habitual use of, employ' ~ ūsus 'use, employment', 9 against the background of cuRRERE, DŪCERE, IACERE, SALĪRE, and ŪTĪ. The gradual coalescence of these two develop ments was apt to produce a well-rounded new derivational pattern — namely a masculine and a parallel feminine series of radical-stressed verbal abstracts. Because the masculines could draw on both sources and the feminines on just one, the masculine series — as was to be expected — was initially somewhat stronger than its feminine counterpart. 10 Does this classic explanation, which the author and his immediate followers as well as his intellectual heirs were to repeat on many occasions, really hold water ? I am afraid that, despite its obvious sophistication, Meyer-Lübke's edifice of hypotheses is subject to grave doubts on several scores. First, Latin denominai verbs in -ĀRE have been in existence since before the start of the literary period, apparently without producing a ripple on the surface of the lexicon; thus, LUCRĀRĪ 'to gain, win, get' from LUCRUM 'profit' and DAMNĀRE 'to levy a fine, find guilty, do damage' from DAMNUM 'loss, damage, 8 On the positive side of the ledger must be placed the author's shrewd observation that the crucial step was from (a) -Tus, -sus formations to (b) -u used regardless ofthe preceding consonant. A vital prerequisite for the entire development here sketched out was the merger, in Vulgar Latin, ofthe 2d ("o") declension, to which the past participle was linked, with the 4th ("u") declension, to which the verbal abstracts belonged. On these processes see, in addition to the older writings by Carl S. R. Collin, the recent investigations by E. S. Georges: "Post-Participial Nouns; Their Development from Latin to Romance", RPh, XXI:4 (1968), 368-391; Id., Studies in Romance Nouns Extractedfrom Past Participles, edd. J. R. Craddock and Y. Malkiel, UCPL, LXIII (1970); and Annegret Alsdorf-Bollée, Die lateinischen Verbalabstrakta der U- Deklination und ihre Umbil dungen im Romanischen, RVV, XXXIV (1970), as well as a welter of critical reactions to these writings: by H. J. Wolfin RF, LXXXIII (1971), 333-339; by A. K. Levy in RPh, XXVI : 2 (1972), 4 1 2 - 1 9 , etc. 9 I t is not immediately understandable why Meyer-Lübke dragged into the discussion so un characteristic an example asCUBITR'to lie down often' beside CUBITUS (-U) 'elbow, ell'. 10 The conclusion ofthis paragraph is particularly disappointing: The author, rather brusquely, declares that all Romance items under scrutiny fall into three categories: (a) products ofLatin words in -us (-) -Ï, e.g., It. assaggio, esordio, prezzo [from , 0j,1]; (b) products ofancestral formations in -us -us, e.g., It. abuso, accento, affetto [from ABŪsus, ACCENTUS, AFFECTUs]; (c) Romance innovations which alone are to be known as postverbals, e.g., It. accordo, perdono. This argument sweeps under the rug such preexistent model words as LUCTA, MOLA, PUGNA, to which Latinists freely apply the label "postverbals" (invented by Bréal), and gratuit ously disregards Diez's by no means incompetent earlier treatment.

184


harm' belong to the Plautine vocabulary. Just why, one is entitled to ask, should derivation-in-reverse have suddenly been activated eight or nine centuries later? Second, while it is true that the triangular relationship CANERE: CANTu: CANTĀRE increased its range and popularity in colloquial Late Latin, it certainly is not the central link (i.e., the noun) that ended up as the chief winner. One telling example is furnished by the family of UTI. The preëxistence of the p. ptc. ūsus encouraged speakers to coin the new verb *USĀRE, in preference to well-established USITĀRE and, of course, to the primitiveUTI(doubly undesirable as a deponential verb and as one of the 3d conjugation). New nouns based on proportional relationships were coined very sparingly, in view of the preference accorded to -ĀTA and similar derivational devices. Third, Meyer-Liibke's schema operates with purely formal variations and encounters, disregarding such factors as felicity, which produces great appeal and is conducive to imitation. Moreover, one badly misses any reference to a particular social context or milieu where the new model might have been warmly welcomed, and one learns nothing about any stylistic advantage that might have accrued to the early crop ofpostverbals from their two most characteristic, inalienable features: brevity and stress on the radical. Fourth, Meyer-Liibke's analysis is geared to the expectation that, especially at the formative stage, the masculine variant far outdistanced its feminine counterpart, since it drew on two separate sources. However, microscopic inspec tion of the earliest records, to which we shall revert, unmistakably shows the prevalence of the feminine, in the oldest layer: witness LUCTA, PUGNA; MINA(E) beside MINĀCIA(E), FALLA beside FALLĀCIA, etc., over against fairly isolated COMPUTUS, also DAMNUM beside LUCRUM (for details see below). Because the feminines, at the Late Latin (i.e., the embryonic) stage, so clearly outnumber the masculines in texts, Meyer-Liibke's subtle conjecture emerges discredited from the philological test-tube. Fifth, and easily most important: The general tendency in Late Latin was, for better or for worse, (a) toward lengthening rather than clipping ofderivatives, and (b) toward the use of heavily-stressed affixes. Suffice it to document this pervasive two-pronged evolution with such almost trivial examples as (a) the elaboration of -ĀMEN, -ĪMEN, -ÚMEN in the broad direction of -ĀMENTUM, etc., and (b) the replacement of dimin. -ULA, -ULA by -ELLU, -ELLA, as well as the gradual triumph of Greek-style -IA over Latin-style -IA.11 Admittedly, there are 11 There are on record many more instances ofsuch lengthening; e.g., the widespread substitu tion of -ĀCEU, -ĪCEU for -ĀCE, -ĪCE; the gradual retreat of-ÍA (after its triumph over -iA) before -ARÍA; the emergence ofnew models for the designation offruit-trees (in -RIU, -, as shown by Fr. poirier, Ptg. pereira, or in -ĀLE, as displayed by Sp. peral 'pear-tree', also moral 'mulberry tree'), as against the older type exemplified by It. pero, cf. Sp. castaño 'chestnut tree'. In each case a unique alliance offorces may have been at work; but two pervasive factors are clearly recogniz able — the speakers' propensity toward lengthening, rather than shortening, the inherited words, and the ease with which they succumbed to the fascination ofaccented derivational suffixes. Seen in this perspective, the postverbals stand out as entirely exceptional.

THE SOCIAL MATRIX

185

on record isolated instances of back-formation, such as are apt to arise in any context of intense flux and consequent insecurity; on another occasion, MeyerLübke himself skillfully arrayed a small collection of words in -ICU, -ICA, from diminutives in -ICULU, -ICULA, testifying to a ' 'gewisse Neigung zur Rückbil dung": rustic MANICUM 'handle of a plough' ( = MANUBRIUM), from MANICULA; AVICA 'bird, fowl, goose', through compression and, to speak with J. Gilliéron, "dédiminutivisation" of AVICULA 'birdie' (from AVIS); *CORBICA 'basket', from CORBICULA, itself rooted in CORBIS 'wicker basket'; *RUDICUM 'ladle', in lieu of RUDICULA, from RUDIS 'stick, rod'; *VĀS(I)CA in preference to VĀSICULA, from 12 VAS -is or VASÚM -Ī 'vessel, dish, utensil'. But even though a certain pattern begins to acquire its contour in these examples, reconstructed from scattered vestiges (Fr. manche /Sp. mango, It. oca ¡ Fr. oie, Bol. K, Surs. rudi, It. vasca), one visualizes nothing remotely comparable in sheer sweep to the Ro mance postverbals. These represent, then, the effects of a certain backlash or countercurrent, which may either have served to smooth away several defi ciencies (harmful side-effects) of the mainstream of events, or else could, at the outset, have materialized in a relatively secluded, sheltered, autonomous milieu, before a more favorable concatenation of circumstances eventually dragged it into the center of changes. In any event, the clash between this sort of clipping and the general proclivity toward wasteful suffixal embroidery in Late Latin clamors for some kind of tentative justification, which Meyer-Liibke pointedly abstained from supplying. One dimension of the rise of postverbals in Vulgar Latin — a facet of rela12 On this regressive model, see Meyer-Lübke's Einführung2 §182 ( = Einführung3 §200), with clues to further literature. Categorially G. Paris may have been justified in endorsing as 'ingenious' Lené's attempt to lump together the postverbal model of false regression with such isolated in novative instances as Fr. nerférer after nerférure (or nerf-férure) (vet.) 'Overreach'; arc-bouter 'to strengthen by a flying buttress' after arc-boutant 'buttress, support, pillar'; and, above all, médecin 'healer, doctor' (for older mire < MEDICU) after médecine 'medical science, medicament': "forma tions que l'on peut regarder à l'origine, ainsi que les nôtres, comme pathologiques, c'est-à-dire, comme reposant sur une erreur" (Rom., XXIX, 441). In fact, Paris even toyed with the idea of including in this group such adjectives newly extracted from nouns as châtain, -e 'chestnut brown, light auburn', violet, -tte 'violet-colored'. But, in so doing, he completely disregarded the differ ences in date, social setting, and level of speech, to say nothing of his neglect of the preëxistent segments (a) -in in nouns (e.g., cousin), and (b) -ain beside -et in both nouns and adjectives. More promising — again categorially or typologically — is the comparison with modern trends of clipping, as observed, lexically, by H. Kjellman, Mots abrégés et tendances d'abréviation en français, UUÅ, 1920; and, syntactically, by E. Bichter, Die Entwicklung des neuesten Französisch (Bielefeld & Leipzig, 1933). Kjellman not only lists, but documents and authoritatively assigns to a special variety oîfrançais argotique (favored by students, military, lower-class groups, railway personnel, Stock Exchange brokers, etc.), such late-19th- or early-20th-c. coinages, typical of highly colloquial Parisian speech, as absorb 'absorption', achar 'acharnement', admiss 'admission', alloc 'allocation', applic 'application', bifur 'bifurcation', combine 'combinaison', commiss 'com mission', consom(m) 'consommation', distrib(e) 'distribution', liquide 'liquidation', etc. (pp. 31-53). Kjellman recognizes in this process the apocope ofthe standard suffixes -ation, -aison, -ment, etc., an analysis which is indeed defensible. But, as the case of absorb, in lieu of *absorp, irrefutably demonstrates, the verb absorber also exerted a measure of influence, and to that extent these semifacetious neologisms remind us ofpostverbal abstracts ofVulgar Latin stock. The slangy character ofthis spate ofexperimental^n-de-siècZe Parisianisms offers a perfect parallel to the athletic-mili tary-custodial social dialect of Latin here posited as by far the most plausible matrix for the incu bation of the postverbals almost two millennia ago.

186


tively subordinate importance to us — visibly intrigued the late-19th-century comparatist: He was eager to determine the representation, in the nouns, of VLat. [i], contending that the masculine series favored straight -U, whereas the feminine counterparts usually ended in I + A, so that speakers in search of appropriate derivatives from, say, DOLĒRE would waver between the two types DOLU and *DOLIA; similarly *FALLU contrasted with *FALLIA and *voLU with *VOLIA. The author offered one of his ingenious explanations for this discrepancy, seeking the clue to I before A in fine details ofthe verbal paradigm. 13 The point is that his scheme simply does not work, given the fact that IACĒRE produced *IACIU, not *IACU, a complication which, a quarter-century or so later, prompted the author to modify his argument radically and to explain *IACIUM as an echo of GAUDIUM 'joy' and of STUDIUM 'zeal' in their parallel relations to the verbs IACĒRE, GAUDĒRE, and STUDĒRE.14 A simpler approach would be to argue the sporadic influence ofthe rapidly spreading adjectival abstracts in -iA, cf. FORTIA 'strength' which — though formally a plural of the subst. adj. (n.) FORTE — bespeaks the pressure of INERTIA 'laziness, rudeness', soLLERTIA 'skill, shrewd ness', and the numerous words in -ANTIA/-ENTIA.15 There occurred no extension, on a comparable scale, of the scope of-iUM.16 So far as Romance offshoots are concerned, the only reverberation of this minor tremor has been a small-scale variation, word-finally, between /1/ and /λ/; also, in one isolated case, between /n/and/N/." The acid test of Meyer-Liibke's skill in reconstruction was that section of his comparative grammar in which he set offthe older, common stage ofthe develop ment from its later ramifications in the individual daughter languages ("einzelsprachlich"). The author was visibly in a hurry to acquit himself of this self13 Whether it is, in the last analysis, correct or erroneous, the argument (Romanische Formen lehre §398) bears pondering. Meyer-Liibke felt that — on the level of unadulterated form if not of meaning — speakers closely associated derivational -a with inflexional -a (3d ps. sg. pres. ind. and subj.), with the result that the /ja/ of DOLEAT 'it may hurt, he may feel pain', FACiAT 'he may do', etc. were reflected in the suffix -ia observable in It. doglia, voglia, etc. From here a path leads to the thinking of Tollemache; see below. 14 See HGFS, II, §108. The author subtly distinguishes between *IACIUM, observable through the prism ofFr. jas 'erster Teich der Seesalzwerke', and Fr. deuil, which involves a belated modi fication of OFr. duel. On this point one may profitably re-read the analysis proposed by K. Togeby, "Qu'est-ce que la dissimilation?", RPh, XVII, 642-667, at 649 (reason for contrast between [a] écueil, orgueil, ac- and re-cueil, cercueil, plus chevreuil and écureil, and [b] aïeul, épagneul, glaïeul, ligneul, tilleul). Togeby neglected to include duel > deuil in his specimen. 15 To these may be added VERĒCUNDIA 'bashfulness, shame', which — judging from OSp. vergüença (beside vergüeña), OPtg. vergonça (beside vergonha) — was undoubtedly pronounced -UNTIA, at least in certain quarters, very much in contrast to the provincial variant that underlies Fr. dévergondage 'barefaced shamelessness'. On FORTiA, see the recent article by Ilse Schön, "Romanisch FORTiA 'Kraft, Macht'—ein Gräzismus?", RF, LXXXV (1973), 255-274. 16 On the contrary, ARBITRIU 'decision, judgment, control, authority' was distorted to albedrio in Spanish (cf. poderío 'power', señorío 'dominion, sway, rule'), while in Tuscan IMPERIU 'command, Empire', emerged as impero. 17 Meyer-Liibke's set of examples: OFr. vaille from valoir, veuille from vouloir, meschaille from meschahir, and even hogne 'mockery' from honir is helpful (HGFS, II, §108), and his proposal to trace them all to faille from falir (the predecessor of mod. falloir ~ faillir) is very much to the point; but the bridge he rather desperately tries to cast from *FALLiA to GAUDIA, the pl. of GAUDIUM 'joy', seems far-fetched in retrospect.

THE SOCIAL MATRIX

187

imposed task and thus had recourse, inexplicitly, to a somewhat rudimentary technique: Where several Romance languages seemed to point toward a common prototype, he posited that hypothetical ancestor, reconstructing, e.g., [*]CERNA on the strength of It. cerna 'selection, sorting; raw recruit', Sp. cierna 'anther' [of flower of wheat, the vine, etc.], and Ptg. cerna, of dubious authenticity (againstthe back drop of CERNERE 'to separate, sift, discern'). Other such experiments at reconstruction involved DOLU 'pain', supported by Rum. dor, It. duolo, OFr. duel [see above], [0]Sp. dudo, Ptg. dó; [*]FALLIA 'flaw, failure, fault', under which one can subsume Olt. faglia, OFr. faille, OProv. & Ptg. falha; [*]IACIU, underlying Neap, yattse, OLomb. giaçço, OProv. ja/z; [*]TIMA 'fear', retrievable through Rum. teamă, It. & Surs. tema ( = Engad. temma), beside OFr. críeme (cf. mod. craindre, crainte); and[*]vOLIA 'will', as the source of Rum. voie 'permission', It. voglia 'wish, fancy, longing, craving' ; to all of which the author, by way of an afterthought, added [*]GRUNDIU 'grunt' as the allegedly sole representative of an -ĪRE verb, pieced together from the interplay, in the hands of an analyst, of (a) GRUNNĪRE/arch. GRUNDĪRE 'to grunt' plus (b) I t . grugno 'snout', fig. 'face, mug' (far il grugno 'to pout, sulk, pull a long face'), Fr. groin 'snout' (cf. grognon 'grumbler'), OProv. gronh.

However, using a kind of undeclared ''internal reconstruction", Meyer-Lübke also operated with bases inferred from a single reputed offspring (e.g., [*]DOLIA, [*]FALLU, and [*]VOLU, as presupposed by It. doglia 'pain, ache', It. fallo 'failing, slight fault, slip', and OFr. vuel, respectively), whenever such weakly-supported bases could be interpreted as mere variants of strongly-endorsed counterparts; i.e., without stating his presupposition in so many words, the author apparently felt that DOLU lent a modicum of credibility to *DOLIA, etc. From today's vantage point Meyer-Lübke's modus operandi appears crude, in general, and vulnerable in many particulars. Even if one disregards the author's anti-philological bias and excessive faith in the comparative method of re construction, which apparently led him to equate solidly documented bases (thus, DOLUS is amply recorded in epigraphy) with passing fancies of conjectura lists, and even if one makes allowances for an excusable margin of plain error (as, for instance, the attribution of Rum. voie to the Romance rather than the Slavic stock),19 there still remains, as a basic flaw, the total neglect of the mere possibilities of (a) diffusion through borrowing (i.e., imitation) and (b) coinci dence through drift, as re-defined not so long ago by J. H. Greenberg, in the wake of E. Sapir's and A. Meillet's earlier thinking. 20 Small wonder that absolutely no suggestive cultural (semantic, stylistic) pattern emerges from the random collocation of CERNA, DOLU/*DOLIA, *FALLU/*FALLIA, *GRUNDIU, *IACIU, *TIMA, 18 Ptg. (obs.) cerna, a word of marginal importance, is listed by some lexicographers as a var. of cenra, which is, in turn, equated with seara 'grain field, harvest'. 19 Cf. Russ. volja 'will'. Interestingly, S. Puşcariu, Etymologisches Wörterbuch der rumänischen Sprache, I: Lateinisches Element... (Heidelberg, 1905), omitted voie from consideration. As Α. Cioranescu, Diccionario etimológico rumano (La Laguna & Madrid, 1966), ρ. 902α, has clearly shown, the Slavic ancestry of voie — which in Old Rumanian meant 'free will', 'favor, liberty' (cf. the obsolete adj. volnik 'free') — was established as early as the trail-blazing writings of F. Miklosich (1860) and A. de Cihac (1879), and was reiterated by B. Conev (1921) and H. Tiktin (1925), all of which makes Meyer-Lübke's slip, in 1894, rather embarrassing. 20 See Chaps,  ("Genetic Relationship Among Languages") and iv ("The Problem of Lin guistic Subgroupings") of his excellent Essays in Linguistics (Chicago, 1957), esp. p. 46.

188


*V0LU/*VOLIA as, allegedly, the thin subjacent layer of the grandiose edifice of Romance postverbals. The section under scrutiny ends up with a half-page devoted to the exempli fication of regional developments, based on provincial Latin usage: in Rumanian, Western Raeto-Romance (Surselvan), Italian (with some attention to northern and southern dialects), Old and Modern French (once more with a quick glance at dialectal usage), Spanish (without proper attention to the deep divide between -o and -e or zero in the masculine series); one example from Catalan (Mallorcan) is thrown in for good measure — while Occitan and Sardinian, for no good reason, remain completely unrepresented. The basic design of this concluding segment is, all told, satisfactory, except that the randomness of the illustrations carries with it serious disadvantages. Combat, débat, and soutien, for instance, are all three listed as peculiar to French, while socorro figures among the formations confined to Spanish. In reality, combate, debate, and sostén have long enjoyed wide currency in Spanish, while Italian boasts sostegno, moreover, French flaunts its own secours (→ E. succor), of striking resemblance to It. soccorso. To what extent does inter-Romance borrowing account for the distribution of items in -bat(e) and -t(i)enl-tegnoi And how far can one defensibly go in completely separating postverbal from postparticipial derivations when they are formally as resemblant as (a) Sp. socorro vs. (b) It. soccorso, Fr. secours and, in addition, practically identical on the semantic side? Compare also Ptg. venda 'sale' to Sp. venta, Fr. vente, beside It. vendita. Meyer-Lübke's 1894 treatment thus leaves not a few legitimate questions un answered, and fails to dissipate many doubts. Nevertheless, it marks a point of distinct progress in certain respects — be it only by bringing together various doubtful cases. This particular service G. Gröber, in his slightly earlier "Vulgär lateinische Substrate romanischer Wörter", had unfortunately shied away from providing in any systematic way, 21 though, on careful reading, one is apt to learn a good deal from some of his entries. Thus, even though he happens to regard ASSALTUS 'attack' (I, 244) mainly as a trace of an unusually contoured past participle (which it indeed is, as a foil to ASSALĪTUS and against the back ground of ASSALIRE), the relation of Fr. assaut (orig. assalt) to assaillir is not at all the same as the relation of Sp. asalto to asaltar, while It. assalto, through its dual connection with the near-synonyms assalire and assaltare 'to attack', mediates between its sister languages (cf. also E. to assail, in the closest vicinity 21 The various consecutive installments, plus a Supplement, appeared in ALLG, I-VI (1884-89). On numerous occasions, in listing side by side a noun and a corresponding verb, Gröber refrained from discussing the underlying patterns ofhierarchy; see I, 548 (' ~ COAG'LĀRE); II, 438 (GYRU^GYRME);rV,515'(LUCRU ~ LŮCRARE); IV, 452 (puLSu ~ puLSĀRE); V, 459 (SARc'LU ~ SARc'LĀRE). He was, on balance, more concerned with post-participial than with post-verbal nouns; rather characteristically, he discussed in detail the relation of the twin participles FALSUS and FALL(i)TUS to FALLERE 'to deceive' (II, 282f.), recording Fr. faute 'mistake', Cat. Sp. It. falta 'dearth, lack', etc.; but he stopped short ofincluding, in his canvassing, the relation ofthese off shoots of FALLiTA to those of *FALKEA, etc. Note also his comment on MORDĒRE 'to bite' ~ MORSUS 'bit' (IV, 120).

THE SOCIAL MATRIX

189

ofTassault). Gröber, without expressly using the label "postverbal", unmistak ably allowed for a formation of this type to have branched off from *STAGNICĀRE (V, 479) and candidly admitted his inability to define with any degree of cogency the relation of *RETINA 'rein' (cf. Fr. rêne, Sp. rienda, Ptg. rédea, etc.) to RETIÑERE 'to hold back'. 22 We shall revert to the issues thus raised, at a later juncture. Despite its share of serious shortcomings — immediately clear to any ex perienced late-20th-century observer — the research by Gröber and MeyerLübke might have generated many exciting discussions. Unfortunately, the seeds did not fall on particularly fertile soil. For one thing, the domain was partitioned among second- and third-string scholars, each concerned solely with the metamorphoses of a single language; thus the vital dimension of re construction-through-comparison came to be neglected. For another, the tri umph of the archeologically oriented Wörter-und֊ Sachen approach deflected at tention from something as ''pale" as verbal abstracts, which invited neither drawings nor photographs, still less maps. Moreover, the problem seemed to lack the welcome appeal of interdialectal or (potentially even more engaging) inter-Romance borrowing,23 to say nothing of the absence of any exotic flavor (Grimm's musing and Diez's parenthetic remark had meanwhile been forgotten). Finally, the prevailing atomization of research blunted any sustained curiosity, on the part of the explorers, about the discrepancy between these derivational models, which were characterized by energetic clipping, and all the others, marked by generous proliferation or, at least, elaboration of affixes. Hence an atmosphere of boredom has long prevailed in many quarters, as well as one of concentration on relatively unstimulating frills. The period of doldrums began with the by no means incompetent, but narrowgauged and uninspiring Uppsala dissertation of G. Lené, Les substantifs post verbaux dans la langue française (1899). For more than one reason, one wishes 22 "Die Herleitung aus dem lat. RETINERE ist ebensowenig wie aus den romanischen Vertretern dieses Zeitworts glaubhaft zu machen...rätselhaft". In sober fact, the single striking feature in the biography of RETINA is the eventual blurring of its compositional design; at a certain point, speakers ceased to analyze it into RE- and -TINA and thus to associate it with TENERE (hence Ptg. rédea, Sp. rienda, Fr. rêne, E. rein, etc.). The word's feminine gender is hardly astounding (Gröber himself cites It. voglia and Prov. falha, and we have already encountered other examples), while the stress pattern, though admittedly infrequent in verbs of the -ERE class, is not unparalleled; Gröber must have overlooked Plautine PROSEDA 'whore', already adduced by Diez. HABĒNA and TENACULUM 'strap, reins', which he cited, indeed represent noteworthy alternatives open to the Romans, but hardly invalidate the derivation of RETINA which he sought to question if not to impugn. As for Fr. soutien, Sp. retén, It. ritegno, they transparently pertain to a later stratum and testify to a process of recomposition. 23 An exception must be made in reference to the Hispanic masculine series in -e or zero, where vestiges of Gallo-Romance pressure were recognized at a fairly early date. Cf. the statements made by F. Hanssen, Gramática histórica de la lengua castellana (Halle, 1913), §276: "La mayoría de estos vocablos son de procedencia extranjera"; and, above all, R. Menéndez Pidal, Manual de gramática histórica española (Madrid, 19416), §83:5 : "Contribuyeron a la formación y aumento de estos postverbales los muchos de origen extranjero...". My own studies, conducted with closer attention to detail, have fully borne out these astute remarks; see "Fuentes indígenas y exóticas de los sustantivos y adjetivos verbales en -e", RLİR, X X I I I (1959), 80-111; XXIV (1960), 201-253 (to continue).

190


the research had been conducted by an Erik Staaff.24 Still, Lené's memoir, especially if used in conjunction with two searching book reviews which it pro voked — the one by Gaston Paris, the other by Adolf Tobler25 — lends adequate service for the study of the postverbals within the spatio-temporal limits of the French lexicon, thanks to the author's conscientious extracts and excerpts. It is the all-important prehistory — the Vulgar Latin segment, to put it bluntly — that refuses to lend itself to this one-sided, not to say parochial, approach. The other book-length treatment, from the pen of F. Tollemache, appeared after 1950, and its analytical meagerness is additional proof of the slowness of genuine progress along this particular axis. 26 Aside from an initial cursory glance 24 Staaff's name comes to mind for two different reasons — quite apart from the anecdotal circumstance that he happened to be Lené's near-contemporary at Uppsala (they were both students of P. A. Geijer's). First, in his own doctoral dissertation (Le suffixe -2RIUS dans les langues romanes, 1896) and in a later (1927), shorter monograph on the relation of Lat. MINUS- and Gmc. miss- to Fr. mé{s)- he brilliantly demonstrated the possibility of ultimate merger of similarlyshaped Romance and Germanic affixes, provided there prevailed a modicum of referential affinity. Second, at the concluding stage of his career, he supplied an unsurpassed study — truly a model of its kind — on the extraction of adjectives from verbs, a problem of distinctly more modest scope: "Sur la formation des adjectifs postverbaux en français", StN, IV (1931-32), 97-119. For the sake of accuracy let me report that Germanic figured less prominently in Staaff's thinking, in 1896 than in 1927, when he merely nuanced G. Lozinski's more radical thesis (1929). 25 See Bom., XXIX (1900), 440-445, and ASNS, CV (1900), 203-209, respectively. Each of the two reviews is, in its own way, a model of turn-of-the-century scholarship. At the time they were writing, both critics had reached the pinnacle of their experience and sophistication. Of these anthology pieces, Tobler's can best be described as a running commentary, in which he aimed at the elimination of all spurious items ("ghost-words") and questionable readings. Of broader rele vance are Tobler's remarks on the gender of certain words; on competing modes of lexical enrich ment mistaken for instances of regressive derivation (substantiated adjectives: grief, lait; nominalized phrases: à bandon, à compte, à pat); and on the wavering between -l and -/λ/, in offshoots of do(u)loir and vo(u)loir — as a consequence of mental association of the nouns with different members of the respective verbal paradigms. Tobler's introductory paragraph is more laudatory than his conclusion. Conversely, Paris is less preoccupied with philological details and more concerned with the basic grammatical design and with genetic issues. Interesting is the reve lation that some of A. Darmesteter's ideas appealed to by Lené go back, in the final analysis, to the teaching of G. Paris. 26 Federigo Tollemache, I deverbali italiani, Biblioteca di "Lingua Nostra", X (Firenze, 1954). Since this monograph series was directed by an editor as dynamic as Bruno Migliorini, it is a fair guess that some of the better features of the venture are traceable to that leading expert's advice. Franca Ageno's severe review of this booklet (AGI, XL [1955], 169-171) — t h e only one ever published on the authority of R. A. Hall's comprehensive bibliography — is worth examining in this context. The side-issue of a subtle thread linking the postverbals to the compounds of the type Fr. porte-plume 'penholder', Sp. matafuego 'fire-extinguisher', It. portabandiera 'standard-bearer' already haunted certain late-19th-c. writers (G. Lené, for one, dealt with this problem at length). The matter acquired fresh prominence through Tollemache, whose first venture had borne on — of all conceivable topics — Le parole composte nella lingua italiana (Roma, 1945). The connection alluded to is the relation of the given compounds and regressive derivatives (i.e., "postverbals") to specific members of the corresponding verbal paradigms. The thrust of Franca Ageno's effort is toward demolishing Tollemache's claim — made in 1945 in regard to the compounds, then re iterated in 1954 with respect to the postverbals — to the effect that certain privileged forms of the pres. ind. paradigm are exclusively involved. She is firmly convinced that, instead, the imperative (sg.) is at issue (on the assumption that some of the compounds were initially nicknames) and she marshals cogent phonological, semantic, and stylistic evidence to drive home her point, in close alliance with A. Darmesteter's classic treatise and with shorter pronouncements by A. Prati (1931), Լ. Spitzer (1952), and G. Bonfante (1954), plus the original version of B. Migliorini's memoir on Italian masculines in -a. Important, from our point of view, are her comments on two series of modern racy colloquialisms, namely (a) Vaccbmoda Obliging person, factotum', il buratta 'puppet', il girella 'turncoat, political weathercock', {Maestro) Impiccia 'bore, troublesome person', il procac cia 'paid messenger, errand boy', lo sbercia 'duffer, incompetent person' [lit. 'one who shoots wide of

THE SOCIAL MATRIX

191

at certain writings by Diez and by a young Meyer-Liibke, Tollemache, in the "historique du problème" he supplies, brashly disregards the entire corpus of technical literature on the postverbals in cognate languages and concentrates exclusively on Italian. 27 Carrying to its furthest consequences an idea tossed off by the puristic lexicographer Fanfanı, Tollemache argues that the Italian postverbals have at all times involved nominalizations of the 1st or the 3d pers. pres. ind.; this is how he analyzes contegno 'demeanor', ritegno 'reserve, restraint', sostegno 'support, prop, bracket', convégno 'meeting, appointment'; taccio 'rough estimate', puzzo 'stench', singhiozzo 'sob, hiccup'; vaglia 'merit', vòglia 'will'; and, with specific reference to an older conjugational paradigm, also allièvo 'pupil', rilièvo 'relief, prominence', sollièvo 'relief, comfort', diniègo 'denial', (poet.) pr(i)ego 'prayer, request'. Tollemache holds that his conjecture is valid not only for Italian, but for the congeners of his native tongue as well.28 Amid his enthusiasm, he forgets to explain why the shift in the conjugational system, contegno→contengo, convegno→convengo, rilievo → rilevo, etc. has so far failed to be reflected among the alleged nominal satellites of the present tense. And how can Sp. sostén be accounted for in terms of his proposal, over against the present-tense forms sostengo ¡sostiene ? Even though Tollemache's insistence on the close ties between the postverbals and certain members of the verbal paradigm may, despite obvious exaggerations and inaccuracies, have not been inopportune, he contributed disappointingly little to our grasp of the rise of the category. The embryonic stage remained in a vacuum; to be sure, there was an allusion (12) to a scattering of Latin prototypes — PUGNA beside PUGNARE and PERFUGA, TRANSFUGA 'deserter, turncoat' along side PER-, TRANS-FUGERE — but these forerunners were not firmly linked to the older, let alone the later Romance flowering of the model. Despite the lack of any real breakthrough, minor advances, by dint of great effort or through sheer luck, were here and there achieved. Menéndez Pidaľs the mark'], il tartaglia 'stutterer' (character of the Neapolitan theater) beside the verbs accomodare, tartagliare, etc., and (b) Ser contrapponi 'Mr. Objection', il dormi 'sleepy person', lo spiovi = lo spiovente 'watershed'; note the mock-respectful titles. Clearly, if all or some of these forms turn out to be substantivated imperatives, they no longer qualify as postverbals, despite the similarity of appearances; to that extent Franca Ageno continues — without citing him — in the tradition of A. Tobler (1900), who similarly trimmed Lené's Old French list (see fn. 25, above). My own feeling is that the Genoese critic is entirely convincing in her repudiation of Tollemache's idée fixe of the obligatory dependence of such nouns on the pres. ind. On the other hand, one is left wondering whether genuine imperatives were pervasively involved, or whether -a and ֊i merely served as "minimal suffixes" in verbal nouns for the 1st and 2d-4th conjugations, respectively, and could upon occasion have been secondarily endowed with imperatival overtones. 27 Even among Meyer-Lubke's writings, the one item that interested him most was the GermanSwiss scholar's least mature grammatical venture, namely his Italienische Grammatik (Leipzig, 1890). A strong influence on him was exerted by Fanfanı & Arlia's Lessico dell'infima e corrotta italianità (Milano, 18903) — although he refused to subscribe to the authors' belief that most postverbals were truncated suffixal derivatives ("raccorciamenti") — and various explorations by C. Merlo, especially those into dialect speech. 28 Tollemache so interpreted Fr. aveu, épreuve, espoir, manœuvre, relief and Sp. acierto, acuerdo, cierro, comienda, huelga, etc. This hypothesis still leaves unexplained the fact that there exist, in each language, two major series — each connected with a characteristic ending and assigned to either gender.

192


statement in the relevant section of the Cid grammar was not exempt from infelicities,29 but in the definitive revision of his Manual he correctly pointed out the extreme rarity of the patterns in classical Latin (citing LŮCTA and PUGNA as the starting points for the future development, but stopping short of drawing any conclusion from their astonishing formal and semantic resemblance).30 Menéndez Pidal also noted the sharp rise of the model in Late Latin ("posterior mente abundan"), without attempting to justify such late blossoming, and added inferred *DUBITA 'doubt', the replacement for DUBIUM, 31 to the recorded triad FALLA 'trick, deceit', PROBA 'test', and COMPUTUS, without paying attention to its split, on the local scene, into cuento 'story' vs. cuenta '(ac)count'. This picture clashed significantly with the one Meyer-Lübke had drawn in 1894. In his ambitious historical treatise on French word-formation, K. Nyrop as signed a generous niche to post- or, as G. Paris preferred to call them, de-verbals (§§540-553) ; 32 however, the opening paragraph, reserved for the elucidation of their origin, turned out to be disappointing through its admission of defeat. After tacitly adopting Meyer-Liibke's CANERE ~ CANTUS ~ CANTARE scheme, the author candidly declared: "Ce genre de derivation devient de plus en plus général dans la langue vulgaire, SANS QUE POURTANT IL SOIT POSSIBLE DE SUIVRE SON DÉVELOPPEMENT" [emphasis mine]. In passing, Nyrop projected onto the 29 Ed. Cantar de Mio Cid: texto, gramática y vocabulario (M., 1908-11), §69.1. Thus, one fails to understand why the author adjudicates acorro alongside duelo to the (Latin? Spanish?) third con jugation. In revising this book for inclusion in his Obras completas, i.e., in the early 'forties, Menén dez Pidal retracted his championship of (ar)rebata 'sudden attack', no doubt under the influence of J. Oliver Asin's 1928 monograph on the Hispano-Arabic background of this word-family; however, revata fits snugly into the picture on semantic grounds, as do alcança, alcanço, alcaz despite their Arabic veneer. That is to say, even though (ar)rebata is unquestionably descended from an Arabic model, the fact that this noun was accompanied by a verb in the borrowing process (cf. mod. arrebatar 'to snatch, carry away' beside unmistakably Hisp.-Lat. arremeter 'to spur [a horse], attack, assail, rush forth') made it ultimately possible for untutored speakers to associate it with the "military" kernel of the Romance postverbals. Moreover, the lively semantic development of rebata in Old Spanish ('cavalry attack' > 'panic, fear, fright, scare, shock' — a sensation actively caused at the outset, then passively experienced later on) confirms the absorption of this once technical term into the core vocabulary. Significantly, its cognates (a)rrobda and ronda 'sentry, (night-)guard', examined in lavish detail by Menéndez Pidal (ed. Cantar, pp. 475-478 and Supple ment, p. 1219), testify to the same kind of hybridism. Once more, words of indisputable Oriental provenience have been, grammatically, pressed into a model of Latin ancestry, with — conceiv ably — a light overlay of Germanism. Thus, guarda and arrobdafronda illustrate processes of sym biosis and neither pose a genetic dilemma nor invite a volteface of the type Menéndez Pidal essayed in revising his Cid edition. One might have expected firmer guidance to parias 'tribute', which stands alone in lacking in the chosen text a corresponding verb to lean upon; note the use of sg. paria in stereotyped phrases: entrar en paria, meter en paria. 30 Manual de gramática...6, §83:5. The author is here more explicit in labeling FALLA, PROBA,

and COMPUTUS as replacements for standard FALLACIA, PROBÃTIÕ, and COMPUTÃTIÕ, but stops

short of explaining the advantage that may have accrued from these substitutions in a certain milieu. He declares the agentive escucha ('scout, vedette, chaperon', lit. 'listener') a descendant of the verbal abstract. 31 Here the formal distance, in classical usage, between DUBIUM and DUBITARE may have been uncomfortably long, a disadvantage fostering the demand for some remedial action. Historically, the primitive was DUBĀRE, based in turn on a submerged primary adjective; the secondary adj. DUBIUS sprouted from that archaic verb, much as LŪDIUS 'stage-player, pantomimist', 'gladiator' did from LUDERE 'to play' and scius 'knowledgeable' from SCIRE 'to know'; DUBITARE, which in the end emerged as the victor, was initially a mere satellite, signaling the frequentative meaning. See Ernout & Meillet, DELL (P., 1959-604), p. 185a. 32 See GHLF, I I I : Formation des mots (Copenhague, etc., 1908).

THE SOCIAL MATRIX

193

level of Vulgar Latin COSTUS 'cost' and GUSTUS 'taste' (flanked by COMPUTUS) and *CAPTIA 'hunt(ing), chase' (alongside PROBA), as so many examples of the earliest innovations. 33 He did not elaborate on the striking mutual resemblance of the assumed leader words. Five years later E. L. Adams, in his expanded Harvard dissertation on affix ation in Old Provençal, offered little more than an orderly catalogue of classified illustrations, culled from acceptable sources and subsequently filtered.34 As re gards postverbals, he repeated the statements made previously by G. Lené and K. Nyrop, giving appropriate credit to these authors. One finds the expected remarks on the unpredictability of gender and ending and on the wealth of doublets; the one distinctive feature that Adams pointed out in medieval South ern French, as against its northern neighbor, was the unequivocal character of the ֊e ending. It unvaryingly involved the supporting vowel, placed after a heavy consonant cluster, in a masculine formation (among corresponding feminines the equation OProv. -a = OFr. -e prevailed). Immediately after World War I, Gamillscheg — who could fall back on Nyrop but, while temporarily stationed at Innsbruck, lacked access to Lené — expati ated on a semantic facet of this category of verbal abstracts in Gallo-Romance.35 His specific concern on that occasion was with the groove in which an abstract would move to achieve concretization; much depended, he argued, on the syntax of the underlying verb. If that verb happened to be transitive, the chances were that the postverbal would acquire the secondary meaning of a tool. Among such designations of "Werkzeuge", the author cited batte 'long wooden staff or beater' [ > E. bat], biffe 'sham, buffer', époussette 'dusting-brush, duster', étire 'stretching iron', fraise 'tool for enlarging a drill-hole', gratte 'scraper', perce 'borer', pince (pl.) 'tweezers, sugar-tongs', presse '(copying-, printing-) press', sonde 'soundingline, lead'. Gamillscheg viewed agentives so shaped as mere extensions of this function, involving a switch from (inert) instruments to humans engaged in an activity: OFr. avise 'sentry', cerche 'spy', crie 'herald, announcer', escoute 'ob server', in addition to mod. garde, guide [orig. guie], etc. As the lone designation of a locale, a half-convinced Gamillscheg adduced échauffe, as in the phrase mettre les peaux en échauffe 'to heap the hides' [a phase of the tanning process].36 33 This is hardly an adequate assortment. GUSTUS, on the authority of Latinists (cf. DELL, s.v.), actually preceded GUSTARE by a sizable margin. *CAPTIA 'hunt' is unlikely to have sprung into existence before CAPTIO ('deception, fallacy, harm'), a haplological distortion of CAPTĀTIO, fell into desuetude, unless *CAPTIA is a bold reinterpretation of CAPTIÕ, made possible by an allusion to the many tricks used in traditional hunting (by the pursuers and by the quarry alike). What is the philological foundation for the projection of *COSTUS onto the plateau of (Common) Vulgar Latin ? 34 Word-Formation in Provençal (New York, 1913), pp. 537-548. 35 "Grundzüge der galloromanischen Wortbildung" (1921), pp. 25f. 36 Gamillscheg seems to have overinterpreted échauffe when he translated it by 'Kübel', thereby gratuitously endangering his own favorite hypothesis to the effect that only nouns extracted from intransitive verbs were apt to refer to a locus. Elsewhere in his treatise (21f.) the author referred to PUGNĀCULUM 'bulwark, wall' beside RETINACULUM 'rein, bracket', TENACULUM 'rein' — all three words of major significance in today's context — and on that occasion freely confessed that one may be in doubt as to whether a tool or a locale was alluded to.

194


But could not a container so used, marginally, qualify as a tool (for the heating process) ? At first glance one senses that M. L. Wagner, a quarter-century ago, gave pride of place to postverbals in his guide to Sardinian word-formation.37 Upon closer inspection, however, one discovers that, while treating with enviable competence and loving care many finer points involved, he came close to skirting all the central issues. One learns from Wagner things that one has somehow anticipated seeing confirmed and that, as a consequence, are less than exciting: Postverbal nouns are frequent; they already abounded in older texts and typi cally accompany ֊are verbs; although fundamentally abstract, they often adopt concrete coloration; among those marked by abstract meaning there are numerous borrowings from Spanish and Italian (as a rule, it is the subjacent verb that has been borrowed) ; there exist two parallel series of approximately equal size, with masculines in -u pitted against feminines in -a; etc. 38 Wagner's comments on individual words, of course, retain to this day their undiminished importance and must be heeded by a new generation of reconstructionists, but the value of his book is seriously impaired by its many evasions; by the prevalent atomization through lexicalization of its grammatical message; and through constant diversion of attention to marginal issues. Astonishingly meager and, for the purposes of reconstruction, downright unenlightening is the information which Rohlfs, two years later, offered in a sort of hastily drafted supplement (§§117If.) to the section on suffixation in his

37

Historische Wortbildungslehre des Sardischen (RH, XXXIX; Bern, 1952). The author further set off several formations confined to poetic discourse (§4), then briefly reviewed a few items either of dubious derivational category  prima facie etymologically opaque, but apt to gain in translucency once they are analyzed as postverbals (§7). Apparently the single most conspicuous feature of Sardinian postverbals is their predominance in the ranks of verbs descending from -ICĀRE, -IGĀRE models or coined in their image (see conclusion of §3); here, for once, the masculine gender seems to be predictable. One is tempted to liken this state of affairs to the striking expansion of -eo in modern Spanish — again to the virtual exclusion of -ea; see the aforementioned discussion by Hanssen (GHLC, §276), the comment by Alemany Bolufer (see η 43α, below) and, in more lavish detail, my own analysis toward the close of the 1959 paper. Incidentally, this rule was not yet strictly enforced in older Spanish; a good counterexample is pelea 'fight, struggle, row, brawl' (its gender and ending perhaps co-determined by those of its principal synonyms, riña and rencilla, orig. renziella, from RINGI 'to show one's teeth, snarl, be snappy'). Pelea is based on pelear 'to fight, feud', lit. 'to tear one's opponent's hair' (from pelo < PILU). A similar relation between noun and verb obtains in Portuguese: pelejar → peleja. In general, the Ptg. -ejar verbs, regarded as native, lack corresponding abstracts; where they are adjectival (like branqu-ejar 'to whiten', es-brav-ejar 'to shout loudly', negr-ej-ar 'to be, appear, or become black', pret-ejar 'to blacken, fill the streets', ronqu-ejar 'to roar', rouqu-ejar 'to croak', tol-ejar 'to act foolishly', vermelh-ejar 'to redden'), the abstract, if at all needed, will be extracted directly from the primitive: ronqu-ice, tol-ice, a situation allowing speakers to short-circuit any experi ments with -ejo, let alone -eja. In the ranks of the -ear variant (borrowed from Spanish ?) one finds more easily a sprinkling of corresponding - nouns: passeio 'promenade' (from passear), mod. bloqueio 'blockade' (from bloquear), but appealed to far more sparingly than in Spanish; witness Ptg. toi-, tou-rada 'bull-fight', directly from touro 'bull', vs. Sp. tor-eo, via the detour of the verb torear (Sp. torada, a collective, means solely 'drove of bulls'). Despite the abovecited counter example (pelea, peleja), a slight tendency toward preferring masculines for descendants of -IDIARE verbs seems to be almost pan-Romanic; cf. Fr. octroi 'grant, concession'; the familiar Italian musical terms arpeggio, solfeggio; and the like. 38

THE SOCIAL MATRIX

195

Italian grammar. 39 One confidently expects to find, and indeed finds, numerous carefully distilled dialectal forms — these stray data are the extent of the use fulness of the two paragraphs. One reads again about the numerical prevalence of the derivatives from -are above all others, about the interplay of abstracts and concretes, etc. 40 One original idea for which the author deserves credit despite the peripheral relevance of the problem (§1173): It. doglia and voglia, along with faglia '(geol.) fault', as well as contegno and convegno, along with compagno 'fellow, companion' and invoglio 'packing, bundle, wrapping', are no longer treated as severely isolated problems, but are subsumed with many other peculiarities of comparable magnitude under "Palatalization of the Word-Final Segment". 41 On the explicative level Rohlfs, at the mid-20th-century point, offers distinctly less than did Diez and Meyer-Lubke, adducing as the presumable starting point a few randomly selected postnominal verbs (ANIMA ~ ANIMARE, FORMA ~ FORMARE, and the like); at the end, by way of afterthought, he brings in, as his sole contribution to the prehistory of the pattern, the example of PLĀNA 'plane' — scarcely an apposite illustration, since that label for a car penter's tool was, primarily, a substantiated adjective (PLĀNU, -A level, even') 42 and the relation to the v. PLANARE 'to polish, smooth over' was secondary. It is unfortunate that the one book by Meyer-Lübke (f1936) whose revision materialized after the Second World War — namely the Französische Wortbil dungslehre — should, far from benefiting by the elaboration, have turned out to be a shoddy product; J. M. Pieľs addenda to §§108-113, which are of special concern to us, are singularly undistinguished and have, as a result of their flatness, failed to goad the imagination of several fine reviewers, who channeled their energies into other directions.43 Thus, the 39 Historische Grammatik der italienischen Sprache und ihrer Mundarten, I I I : Syntax und Wort bildung (Bern, 1954). 40 The author rarely pauses to apply such tags as "semilearned" to certain specimens he exhibits; one might have expected that much in the case of bonifica, notìfica, qualifica, ratìfica, specifica (§1171). But then, should not these and similarly structured words have, generally speaking, Deen dissected under the rubric of composition ? 41 I take it that this is what Rohlfs actually meant when he wrote "Palatalisierung des Wortstammes". 42 This background of PLĀNA, ensured by Ernout and Meillet's listing (DELLı, p. 512Ò), also explains the rise of the dimin, PLANULA = 'eynoms (cf. REW 3 6567, 6580). 43 The original edition of Part 2 of HGFS produced before long two weighty critical reactions, one by W. von Wartburg, in ZRPh, XLII (1922), 504-508, who did not expatiate upon postverbals but dropped a potentially helpful remark (507) on the Germanic background of the OFr. verbal abstracts in -ine (gast-ine, later desert-ine, etc.); the other by Leo Spitzer, in AR, VII (1923), 194210, who — among many other comments — ascribed a certain phonosymbolic effect to a highlycompressed word such as rebond 'rebound', as against its staler counterpart rebondissement (205). In attempting to refute this opinion at a later date, Piel missed the entire point by resorting to the lame argument that rebond should be traced to bond rather than rebondir. While onomatopoeia may not have been the right label, the postverbals, undeniably, display a starkness and sponta neity which set them apart from heavier suffixal derivations. Pieľs addenda and corrections (173-213, at 196f.) appended to the 2d ed. (Heidelberg, 1966) of HGFS, II, cannot be safely used without collateral appeal to the devastating reviews that revision provoked, mainly those by M. Höfler, in ZRPh, LXXXIII (1967), 104-114, cf. his brief comment ondéchet ~ dèche (112); and by H. D. Bork, in RF, LXXX (1968), 421-439, at 433. Other reactions, only one notch less negative, came from G. Merk, in RLiR, X X X I (1967), 225-227; J. Klare, in

196


total impression one gains is one of a gradually ebbing tide of enthusiasm. 43a As if to compensate today's analyst for so many and such bitter frustrations, B. Migliorini in the mid-'thirties wrote (and, a quarter-century later, expanded) a memoir of which a substantial portion (§11) bears on one facet of the subject here under study. 44 In focusing attention on a sharply circumscribed group of Italian agentives, namely: (a)scolta 'sentry, night watchman', (obs.) g(u)aita, scaraguaita 'look-out post', (obs.) guarda > (mod.) guardia 'guard', guida 'guide', (slang) piant(ell)a 'policeman or soldier on point duty', ronda 'night patrol', (Sic.) sciurta 'night watchman', scolcajsculca 'spy', scorta 'escort, guide', sentinella 'sentry', staffetta 'courier, dispatch rider', vedetta, orig. veletta 'look out post',

Migliorini argued that all members of this genetically motley but grammatically close-knit series, whatever their individual backgrounds,45 were initially verbal BRPh, VII (1968), 173-175; and O. Jänicke, in VR, XXVIII (1969), 329-338, esp. 333 (on dégoût). To the general untidiness and the specific infelicities already exposed, let me add the observation that the particular passage in Meyer-Liibke which his direct pupil Piel despaired of understanding rather transparently meant that juge 'judge' and neige 'snow' — going back, as in the last analysis they do, to the primitives IŪDICE and NĪVIS — cannot possibly rank as straight postverbals though their forms indeed betray traces of secondary pressure by the correlated verbs jugier 'to judge' < IŪDICĀRE and neig(i)er 'to snow' < *NIVICĀRE 'to snow on and on' (in lieu Of NĪVERE).

43a The general slackening of progress in the 20th century, after an encouragingly brisk start (1880-1900), is best exemplified by J. Alemany Bolufer's uninspiring analysis, in his Tratado de la formación de palabras en la lengua castellana: la derivación y la composición (Madrid, 1920), §§1-3. The author classifies the -a, -e, -o endings peculiar to postverbals as "sufijos átonos", a term doubly unfortunate since Menéndez Pidal, as early as 1905, preempted that label for an entirely different morphemic or submorphemic category. (For a critique of Menéndez Pidal's theory see my paper "The Rise of the Nominal Augments in Romance", RPh, XXVI, 306-334.) Alemany Bolufer breaks down the -a and -o forms according to conjugation class and to meaning (agent: adivino 'soothsayer', tool: sonda 'plummet', time or season: siembra 'seedtime', beside the solid block of abstracts) and he correctly observes the sweeping spread of -eo. But he does not pause to ask him self whether in such words as fallo 'judgment, verdict' (from Osp .faūar 'to find') and recibo 'receipt' (both of which pertain to the legal domain) it is not preferable to recognize the fossilized use of the 1st pers. sg. ('I find or opine', 'I receive or have received'), etc. To be sure, livelier discussions of erratic modes of word formation ensued in books geared toward general or descriptive linguistics, such as F. Brunot's La pensée et la langue՛, . Bally's Linguistique générale et linguistique française; J. Dubois' Étude sur la dérivation suffixale en français moderne et contemporain՛, and É. Pichon's Les principes de la suffixation en français; Venrichissement lexical dans le français d'aujourd'hui. But these more innovative approaches, for all their sparkle, did not in and of themselves move the genetic problem here under scrutiny away from "dead center". 44 "I nomi maschili in -α", Studj romanzi, XXV (1934), 5-76; the revised text occupies pp. 5 3 108 of B. M.'s Saggi linguistici (Firenze, 1957). From M. L. Wagner's encomiastic appraisal (see VKR, VII [1934], 270-272) it is worth citing his endorsement (272) of Migliorini's derivation of sentinella 'sentry'. 45 The number of Germanic families is conspicuously high: guaita, guarda, guida, scolcajsculca, scorta, spia, staffetta; interestingly, even though (a)scolta goes back to A(U)SCULTĀRE, it gives the appearance — on any level — of being a Germanic word and may thus have interacted with those that are genuine ingredients ofthat stock. Ronda is Hispano-Arabic; see ?29, above. Sentinella is undoubtedly somehow connected with sentire 'to hear', but — on this issue I must swerve from the author's straight path — the interfix -in- points toward secondary association rather than direct descent; cf. the near-synonym vel- > ved-etta, under the impact of vedere 'to see'. It. guida dislodged native Fr. guie (reminiscent of Sp. Ptg. guía) in favor of guide, which invaded English. Could occasional il guardia, in reference to a single guard, involve imitation of Spanish usage, as regards gender ?

THE SOCIAL MATRIX

197

abstracts; 46 he might have added that here, for once, gender and ending are predictable, as they are also in the case of Sp. -eo and of its Sardinian counter part. In any event, Migliorini incontrovertibly identified the starting point of the entire development in classical Latin: (pl.) EXCUBIAE 'a watching, keeping watch' (lit. 'a lying out of the quarters, a lying-out on guard'), VIGILIAE (Plautus) 'nightly vigils', 'watchfulness'. Perhaps, heavier emphasis on PER- and TRANS FUGA might have been relevant. Another redeeming element in the 20th-century phase of scholarly gropings has been the steadily growing measure of attention which lexicologists, as against grammarians, have paid to early postverbals — especially those used during the transition from Latin to Romance. In fact, occasionally the same expert has performed more effectively in tracing the nascent postverbals through the lexicon than in categorizing them and accounting for their genesis in strictly grammatical context. We shall draw on these findings before long; suffice it to state here, by way of anticipation, that Meyer-Liibke, in the revised edition (1930-35) of his dictionary, markedly improved, in a few scattered verdicts, on his earlier pronouncements on postverbals (1894, 1921), which had been cast in a grammatical mould. Thus, he no longer operated with *CERNA and *TIMA, being instead satisfied with listing the clusters of Romance derivatives noncommittally under the respective verbal entries (§§1832, 8737); *DOLIA, *FALLU, *VOLU, and * VOLIA also ceased to qualify as archetypes projected onto the plane of Latin. DOLU retained its status (§2727) and was newly supported by a refer ence to Basque; *FALLIA was salvaged on the strength of epigraphic FALLA (§3168); *IACIUM 'lair' emerged as the real winner (§4566); GRŮNIUM, defined as an anatomical term ('Schweinsrüssel', i.e., 'pig's snout'), was spared (§3894), but deprived of any mandatory link with GRUNDĪRE/* GRUN JARE (§3893).47 On the other hand, COMPUTUS 'account' (§2109) flanked COMPUTARE 'to (re)count' (§2108). The total impression the reader gains is one of more sharply pronounced realism and enhanced fidelity to textual transmission. IL A NEW RECONSTRUCTION OF EVENTS

Latin, rather transparently, had two groups of, at best, loosely connected nouns that could be extracted from verbs without the mediation of suffixal machinery: 46 Migliorini's argument in favor of the filiation 'abstract' → 'agentive' is persuasive, being based on the wide use of certain stereotyped phrases (far la guardia, andar di rorida, esser di scorta,stare in or star di sentinella) and attuned to the intrinsic affinity of abstracts to collectives (in the case of guarda, I suppose), while in other instances "questi termini, dato che spesso la funzione è esercitata da singole persone, hanno anche valore singolativo". However, in his survey of earlier opinions he might have gone beyond contrasting two pronouncements by Meyer-Liibke (GRS vs. REW, the latter under the influence of the Bruckner monograph); after all, Diez (1853), Lené (1899), the latter's critics, and many others had participated in this debate. 47 GRUNIUM is recorded at a fairly late date, in an Oribasius translation. Hence the wisdom of M. Niedermann's analysis, in which GRUNITJM was interpreted as a postverbal of *GRTJNĪRE, a byform of GRUNNĪRE inferrable on the analogy of MAMM-A ~ -ILLA; GRÜNNÎRE could of course have lingered on in some other corner of the Empire, supported bv^ GRUNN-IÖ -ITTNT. This explana tion has earned its advocate the applause of Ernout-Meillet's DELLı, s.v.

198


(A) There was a sizable body of nouns in ֊A, normally but not consistently masculine, which referred to the occupation, or status, or temporary behavior of individuals, sometimes in a humorous tone. The verbs at issue were chiefly of t h e -ERE, -ERE, or -ĪRE class (e.g., SEDERE, COLERE, PETERE, FUGĚRE, SEQUI, VENIRE), to the virtual exclusion of -ĀRE verbs. The architectural design of these

words was such that they would contain either a prefix (AD-, IN-, PER-, PRO-, RE-, or a noun stem (AGRI-, BUB-, HĒRĒDI-, HONORI-, of verb stem, in which case they represented compounds. The extreme brevity of grammatical equipment at the end almost seemed to counterbalance the considerable length of these words — minimally trisyllabic — in other segments. A few of these forms made their appearance late and may involve pretentious exercises in artificial coinage; but some of the formations can be traced to archaic Latin, and have indeed figured prominently in discussions among Indo-Europeanists.48 Here are a few illustrations: TRĀNS-) plus the bare verb stem, LUCI-) plus the very same type

ACCOLA 'dweller near a place, neighbor', ADVENA 'foreigner', AGRICOLA 'farmer, ploughman', (Cicero) AGBiPETA 'field-oriented, land-hungry', AQTJIFUGA 'one fearful of water, one suffering from hydrophobia', BTTBSEQTJA 'cowherd, herdsman', CONFUGA 'refugee', DEFUGA 'runaway', (Petronius) HĒRĒDIFETA 'legacy-hunter', (gloss) HONÕRIPETA 'one striving for honors or striking ahead' ( = G. Streber), INCOLA 'resident, inhabitant', LUCIFUGA 'light-shunner, one who has turned night into day', PERFUGA 'deserter', PROFUGA 'fugitive, refugee, escapee', PROSEDA 'whore', PULLiPREMA 'paederast', REFUGA 'apostate', TRANSFUGA 'deserter'. RETINA would belong here on formal grounds for three separate reasons: It has the right ending, boasts a prefix, and leans against an appropriate verb. Semantically, however, designating as it does a tool, or a part of the harness ('halter'), it stands shockingly apart. In Romance, this entire group — which, we recall, already stirred the imagina tion of Diez as early as 1838 — was doomed (except, of course, for haphazard learned incrustations, cf. Sp. agrícola, tránsfuga, It. [stilted and obsolete] incola, etc.). The major exception has been RETINA, relatively well preserved (Fr. rêne, Sp. rienda)՝, but the price for its survival apparently was the severance of any ties to the — comparably transmitted — TENERE family.49 (B) Within the second group — equally well silhouetted in the language, but less neatly defined in scholarly literature and far less thoroughly discussed — two words, measured both by themselves and in terms of their collocations with in the respective families — stand out very sharply: PUGNA and LUCTA. Both have already been cited here and there as isolated remote models (e.g., in 48 Migliorini, "І nomi maschili..." (see Saggi..., p. 56), opportunely recalls that in compounds of the rAGRicoLAn type, some Latinists — Brugmann, Lindsay, Sommer — recognized ancient ab stracts (i.e., nomina actionis) subsequently blurred on the semantic side, whereas Saussure, spurred by his usual originality, pleaded in 1909 for an alternative analysis: He descried here relics of disyllabic roots ending in ә, which were eventually absorbed into the 1st (i.e., -a) declension. 49 Another highly atypical formation — SCRIBA 'clerk' — erratic on account of the unexpected absence of some prefix, also survived into Romance, but only after undergoing bizarre transmuta tions, through identification, first, with a Romano-Germanic type for agentives (-Ā, -ANIS) and, ultimately, with -ĀNU. The "golden rule" still holds: The danger of light-weight words is averted by equipping those bereft of a prefix with at least an appropriately substantial suffix.

THE SOCIAL MATRIX

199

Menéndez Pidaľs Manicai), but the full implication of the astounding parallelism in their formal and semantic make-up has not yet been properly emphasized. Note the identity in the number of syllables (two) and phonemes (five), and in the syllabico-accentual configuration as well: CÝC-CV; observe further the re current brevity of the stressed vowels and the shared "velarity" — a distinctive feature — in the opening segment of their consonant clusters. Add to these resemblances the similarity of meanings: PUGNA, the older of the two, was tantamount to 'fight fist to fist, man to man', hence 'battle, combat, action, engagement', and its -A clearly served the purpose of differentiating it from PUGNU 'fist', the head of the entire family. LUCTA, a Late Latin innovation, meant 'wrestling match', and its -A, scarcely motivated by any retreat from the hazard of polysemy or homonymy, sounds simply like an echo of the ending of PUGNA. Finally, PUGNA was hierarchically subordinated to a high-frequency active verb of the -ÄKE class, namely PUGNARE 'to fight, give battle' ; similarly, LUCTA was the satellite of a verb of the A- conjugation which, at the outset, functioned deponentially (LUCTĀRĪ 'to wrestle'), but, toward the close of Antiquity, was absorbed into the rapidly expanding active branch of the -ĀRE class. Once LUCTĀRĪ, by becoming -ĀRE, was grammatically joined to PUGNARE, there was one more cross-connection between them in addition to their age-old semantic link and their numerous formal coincidences. Shortly thereafter, LUCTA, predict ably, sprang into existence; and PUGNA beside LUCTA, given their above-described status as virtual twin-formations, in all likelihood became the trail-blazers for the crystallization of the feminine postverbal series.50 The victory of the PUGNA/LUCTA team (with the center of gravity moving more and more in the direction of LUCTA, away from lexically "doomed" PUGNA, which was meanwhile succumbing to the onslaught of BATT(U)ĀLIA) explains numerous significant details : the revolutionary preponderance of the -ĀRE class, once totally divorced from the postverbals; the sudden liberation from the older prefixai or compositional design; the continued importance of the -A series; and the neatly delineated grammatical rôle of verbal abstracts. But there is one circumstance that exceeds all others in the weight of its consequences: We can now for the first time lay our fingers on the aberrant prosodie design of these words. To be sure, Latin had a normally contoured derivative from one of the two leader verbs involved, namely LUCTĀMEN 'wrestling, struggling' (reminiscent 50 One cannot overestimate the contagiousness of a pair of near-synonyms which are also phonically highly resemblant. To be sure, there existed in Latin isolated postverbals of a very different shape and slant, e.g., TRUDIŠ 'pointed pole' and VEHES 'cart-load', caught in transparent relation to TRŪDERE and VEHERE, respectively. But this molecule hardly constituted a promising start; the u of the noun clashed with the ū of the verb, and the nom. sg. in -is or -ES carried with it no mes sage of a productive declensional paradigm. On the formal side, ACUS 'needle, pin', COQUTTS 'cook', MERGUS 'diver, layer of a vine', RUMPUS 'vine branch' were less objectionable, and some of these words actually lingered on in Romance, without, however, becoming rallying points for the launch ing of any new series. (Cf. It. aco 'needle', cuoco 'cook', etc.) TRAHA 'drag, sledge' constitutes a severely isolated tool-name, a state of affairs which gave rise to vars. already extant in Antiquity (TRAXEA), to say nothing of the vulnerability of the -AHA segment.

200


of CERTAMEN 'contest', EXAMEN 'test', etc.), and no preëxistent word would have blocked the rise of *PUGNÃMEN 'fist fight'. But in the highly peculiar social atmosphere of army quarters and training grounds — among legionaries, pugi lists, gladiators, acrobats, wrestlers, and gymnasts — verbal sparseness, immedi acy, and unceremoniousness have, the world over and from time immemorial, been special virtues, not only because every split second counts in serious con frontations, but also because the elegant use of elaborate lexico-grammatical structures seems out of place. We have now tentatively identified the social milieu, within the fabric of Roman Antiquity, in which the "clipped" style underlying the Romance postverbals could very plausibly have originated. Given many unknowns in our equations, it is impossible to avoid the appeal to educated guesses in further reconstruction. It is a fair guess that FUGA 'flight', also characterized by  and by disyllabicity, joined the PŰGNA/LŰCTA kernel; in fact, FUGA lent itself to mediating between the -ĀBE verbs and the alliance of verbs assigned to the other conjugations because, though it was extracted from FUGĔRE 'to flee', it was also apt to be associated with FUGĀRE 'to rout'. To EXCUBIAE and VIGILIAE, already adduced by Migliorini, one is tempted to add another fem, pl., namely ĪNSIDIAE 'ambush, trap', which once more widens the spectrum of conjugation classes pressed into service (cf. INSIDERE; Ernout and Meillet explain ĪNSIDIAE as 'fait de s'établir à un endroit pour y guetter l'ennemi'). The coexistence of (fem.) ĪNSIDIAE (a word preserved in Romance, hence endowed with marked vitality 51 ) and of the triad (n.) OBSIDIUM 'blockade', PRAESIDIUM 'guard, garrison, escort, convoy', SUBSIDIUM 'the troop stationed in reserve' 52 may have been instrumental in creating among the suspected group of speakers an awareness of two parallel series, one in -u, the other in -A, and may thus have paved the way for such rival experiments as It. aiuto 'help' vs. Sp. ayuda, Ptg. ajuda — which may, incidentally, very plausibly have also been military terms at the outset ('auxiliary troop, or outfit, or platoon'). In a succession of investigations, the early appearance of FALLA 'deception' has been stressed; 'stratagem, military trick, maneuver' can all be readily sub sumed under 'deceit', so that the word may easily have been arrived at within the same social matrix as those previously here exhibited. It is not improbable that FALLA joined FUGA, etc. via a detour. There existed in the lexicon a nominal primitive, MINAE, whose pristine meaning was something like 'projecting points of rocks or pinnacles of walls' (the DÉLL 4 offers the definition: 'saillie, avance d'un mur, d'un rocher, surplomb; chose suspendue sur...'), and which subse quently developed into the equivalent of 'threats'; from this noun MINĀRĪ, 51 On OPtg. ensejas, also the verb ensejar (presumably contaminated by the family of EXAGIUM) etc., see my paper, "Old Spanish assediar and its Variants", HR, XVII (1949), 183-232. 52 Interestingly, Meyer-Lübke toyed with the possibility of reconstructing *ABSEDIUM and *DĒSEDIUM, with *E replacing classical Ї through recomposition. Add EFFUGIUM 'flight', REFUGIUM 'refuge', and the like. While the -IUM and -IA words, undeniably, differ slightly from the straight postverbals on account of their / j / element and thus, at best, constitute a sideline paralleling the assumed mainstream of events, they may nevertheless have sharpened the speakers' feeling for the two series of abstracts polarized mainly by a contrast of gender.

THE SOCIAL MATRIX

201

eventually -ĀRE, 'to threaten' cut loose. The adjective MINĀX 'menacing' con currently paved the way for MINĀCIAE 'threats' (derived like DIVITIAE 'wealth'), and this suffixal derivative, in the end, crowded out the "simplex" MINAE (cf. E. menace). But while MINAE, MINĀRĪ (-ĀRE), and MINĀCIAE coexisted or over lapped, speakers using side by side FALLERE 'to deceive', EALLAX 'deceptive, deceitful', and FALLĀCIA 'trick, deceit' may very well have been induced, through the familiar agency of a proportional relation, to try out FALLA, given a certain semantic real-life affinity of 'threats' and 'tricks' — the common denominator is 'bluffing'. Finally, as long as there exists documentary evidence for the occur rence of PROBĀRE as the counterpart of 'to try, test' (frequently in the Vulgate) or 'to approve for military service', 'to recruit, enlist' (thus in the Epistles of Plinius Caecilius Secundus, who credits this use to Trajan), we are not too wide of the mark in arguing that PROBA may have been minted in the same army milieu, by hardened soldiers and commanders attuned to a "clipped" style — even in lexical derivation. PROBA, doubtless a colloquialism at first, is very meagerly attested; the fact that the historian Ammianus Marcellinus and the Codex Justinianeus used it — sparingly — for 'proof' does not exclude the pos sibility of its genesis in another environment, with some such meaning as 'screen ing, enlistment' or, alternatively, 'test' [of an enemy's real strength through a preliminary skirmish or diversionary maneuver]. Once the semantic gamut of PROBA had been extended to the juridical sphere (by the year 400 A.D. at the latest), there was no reason for hesitating to try out *DUBITA (or *DUBITU) 'doubt' — hence Ptg. dúvida, OSp. dubda > duda, Fr. doute, all of which offer the advantage of a smooth, tight fit with the corresponding verbs (duvidar, dubdar, douter).53 The trajectory of the masculine series is less easy to trace. One of the few available anchorages is the early appearance of DOLUS 'pain' in epigraphy, the general understanding being that DOLUS represented a substitute for DOLOR -ORIS, 54 within the DOLERE family, and that it succeeded in ousting its nearhomophone DOLUS 'trick, device, deceit', well-attested in fine literature, at all levels of discourse (cf. DOLŌSUS 'wily, cunning'). Since the morphological relation of DOLORE to DOLERE was by no means unfamiliar or unattractive to speakers of Late Latin and nascent Eomance (cf. CALŌRE 'warmth' > Fr. chaleur alongside CALÉRE 'to be warm or hot', 'to arouse strong feelings' > Fr. [obs.] chaloir 'to matter'), homophony alone must be held responsible for the replacement. To have produced this effect all by itself, it must have been acutely unsettling. 53 By preserving the organic offshoot of ancestral DUBIUM, Tuscan, in contrast, offers a transi tion from noun to verb (dubbio —> dubitare), fairly rough in two or three respects. 54 In French, douleur '(any kind of) pain' and deuil 'mourning', from older duel, were in the end semantically differentiated. OSp. duelo fell into desuetude and dolor, after a period of wavering, became consistently masculine. Portuguese, however, with characteristic laxity in matters of lexical economy, still tolerates, side by side, dó (m.) 'pain, compassion, sympathy, mourning' and dor (f.) 'pain', used not infrequently in the plural. On a variety of side issues: the rôle played by gen. pi. DÖLŌRUM in mediating between DOLOR and DOLUS, the emergence of the adv. DŌLŌSĒ (CIL 12.1939), and the relation to DOLŌ -ĀRE 'to cut' see DĔLL 4 p. 181.

202


Upon further reflection one discovers that military life, beside martial sports, provides one of the few contexts in which 'pain' (from a wound inflicted in battle, or from some other physical hardship) and 'stratagem, ruse' (as in a shrewdly executed maneuver, or in a feint) could seriously have been in each other's way. It is not easy to place COMPUTUS in a truly convincing perspective. Perhaps the "clipping" was favored by the fact that the word served as the equivalent of αριθμόν (interestingly, the illustration adduced by Eönsch from Irenaeus: ''secun dum Graecarum litterarum COMPUTUM" suggests the possibility of the writers' protracted awareness of this descent). COMPUTUS and (far more widely preferred) COMPUTĀTIO may have belonged to different stylistic registers. The selection of the masculine ending is less easily explained. Perhaps it is not irrelevant that, despite the use of the conservative spelling -PUT- (in the mathematical writings of Firmicus Maternus and in a few other sources), the actual pronunciation inferrable from the consensus of Romance representatives seems to have been *COMPTUS (witness Fr. conte, It. conto, Sp. cuento/cuenta). Given the préexistence in the Latin lexicon — as early as Lucretius — of COMPTUS 'band, tie' (from COMERÉ 'to bind together, comb, adorn, deck out'), one could toy with the con jecture of a folk-etymological association,55 making the new COMPTUS the suc cessor of the old. On second thought, a considerably less far-fetched explanation suggests it self, if one falls back on the strongest among Meyer-Liibke's arguments, applying that argument to two items whose importance he appears to have overlooked. From the dawn of the literary tradition, there coexisted in Latin LUCRUM -Ī 'gain, profit' and LUCRĂRI 'to gain', also DAMNUM -Ī 'fine, damage' and DAMNARE 'to levy a fine, inflict damage, find guilty'. The modern Indo-Europeanist is in a position to demonstrate that, hierarchically, in both families, the verbs sprouted from and initially depended on the respective nouns; but to the naïve speakers all that mattered in both contexts was the simultaneity of verb and noun plus the close tie harmoniously uniting them. The Romanist realizes how deeply these two families were anchored to folk speech in different parts of the Empire; 55 The secondary differentiation — in French through allographs and in Spanish (and Portuguese) through allomorphs — of conte/compte (also conter/compter) and cuento/cuenta {conto ¡conta), re spectively, involves the arresting problem of intolerable polysemy. The split cuento vs. cuenta raises several issues. Could cuento be an imitation of Fr. conte 'story', with the expected adjustment to local conditions except for the preservation of the trans-Pyrenean gender? In that eventuality, cuenta, reminiscent of ayuda, would alone constitute the indigenous form, clashing, with regard to gender and ending, with Tusc, conto 'account, story' and aiuto 'help'. If, alternatively, cuento descends from COMPUTU in a straight line, then cuenta would inevitably represent a local innova tion designed to remedy an excessive range of meanings. This semantic plethora of cont..., to which a-cont-ecer 'to happen' (from CONTINGERE) has also contributed, may, incidentally, explain the speakers' willingness to blur the morpheme boundary in PER-CONTĀRĪ 'to ask inquisitively, question strictly' (from CONTUS 'pilot's pole or pike' = κοντός) > Ptg. perguntar, Sp. preguntar 'to ask'. Note that 'count', as a title, stayed out of this tangle in Spanish (conde < COMITE, lit. '[king's] companion'), though it did not of course in French. The hypothesis of borrowing from GalloRomance, in the case of cuento, receives support from the typologically similar bifurcation of FRONTE > (l) fr () ente (f.) 'forehead' vs. 2. frente (m.) 'front' via Fr. front.

THE SOCIAL MATRIX

203

LUCRU and LUCRĀRE have fared exceedingly well in Rumanian (where a lucra means 'to work' and lucru has acquired an exceptionally wide scope: 'affair, busi ness, matter, thing') and in the Iberian peninsula as well (cf. Sp. lograr 'to get, obtain, attain, manage to produce, succeed'; logro 'attainment, profit, usury'; malograr 'to miss, waste, spoil'; -ado 'ill-fated, late [ = deceased]', etc.). This lexical family is less prominently represented in Italian: log(o)rare 'to wear off', and has left only light and scattered traces in Gallo-Romance : OProv. logre (plus the weak inference from Cymric, i.e., Welsh llygru as to LUCRU'S earlier lingering in northern Gaul); in fact, the thinness of Gallo-Romance evidence best explains, without excusing it, Meyer-Liibke's relative neglect of the entire family. It would be easy to show that DAMNUM, too, and, on a slightly more modest scale, DAMNĀRE likewise were deeply rooted in provincial Roman folk speech. Granted that much, one readily sees that the relation LUCRU ~ LUCRARE, DAMNU ~ DAMNARE, reversible in synchrony, could have stimulated speakers and even writers to experiment with COMP(U)TU as a satellite of COMP(U)TĀRE. Semantically, the fit is perfect: 'profit', 'loss', and 'count' clearly belong together and support one another. As all specialists agree, the Latin verbal nouns in -TUS, -SUS — which sounded and looked as if they were past-participial — bolstered the position of the mascu line postverbals. The anatomy of the transfer can be illustrated with one telling example. Ancestral CURSUS not only survives in Fr. cours (cf. its satellites concours, recours, secours, beside more learned discours) and It. corso, etc., but also underlies Sp. acosar 'to pursue relentlessly, harass' on the basis of the byform CU(R)SU. On the other hand, Old Spanish had acorro (beside acorrer 'to run up, hurry; help') and socorro (beside socorrer 'to succor'), and certain scholars support the derivation of Sp. corro 'group or circle of people, ring, ring-arounda-rosy' from correr 'to run' — despite the sparse documentation of corro in medieval sources and a disquieting semantic gap. 56 If this derivation is tenable 56 The analysis of corro as a back-formation from correr was already advocated by, inter alios, .  Rice in HR, I I I (1935), 162, with a reference to OSp. acorro and to mod. recorro; the note was reprinted, without elaboration, in the posthumous collection, arranged by U. T. Holmes, Jr., of Romance Etymologies and Other Studies (by Rice), UNCSRLL, VII (Chapel Hill, 1946), pp. 62f. Corominas, loc. cit., points out the semantic gap between corro and correr and the spotty medieval record of the former as the chief obstacles to Rice's conjecture, leaning toward belated derivation of corro from corral 'yard, stockyard' rather than toward a gambit in the opposite direction. How ever, strong counterarguments can be pressed against Corominas' plea. For one thing, the influx of Arabisms, reflecting the success of the Moorish cavalry's celebrated lightning attacks (cf. algara 'raid', etc.; note also the Arabicized veneer of OSp. alcançar 'to reach [in hot pursuit]', from encalcar 'to be at one's heel' < *INCALCIĀRE) could very well have greatly debilitated the position ornative corro *'running, raid'. For another, the expansion of the dynamic native suffix -eria led to the mintage of correria 'raid, foray, excursion', which is also apt to have eroded the original status of its arch-rival corro. If this analysis carries conviction, then rare late-medieval and mod. corro simply represents the meager residue of the original ambit and incidence of a word once firmly entrenched. In conjunction with the impressively strong record oí acorro and socorro (e.g., in Juan Ruiz), this evidence would support the conjecture here championed of a switch from CTJRSTJ to CTJRRU (through semantic reinterpretation of the latter) in a provincial milieu of late Antiquity, namely Hispània, Corro 'raid, foray' conceived as being, initially, a military term would harmonize beautifully with various words for 'skirmish, encounter', including Sp. encuentro, Fr. (r)encontre (—> E. en counter), It. riscontro beside incontro.

204


(as, I think, it is on balance), we would have before us a wholesale switching of -CURSU formations in the direction of -CURRU, once more within the social matrix of warfare and martial sports. Actually, the situation is even more arresting: Classical Latin already boasted an early postverbal CURRUS, with the semantic specialization 'chariot'. Perhaps such chariots, at a certain point, were discarded, and so a lacuna was produced, which invited the coinage of a masculine abstract. III.

POSTVERBAL AGENTIVES

The last major problem to be tackled here involves a certain subcategory of agentives, which formally resembles the postverbal abstracts and, in the opinion of most analysts, cannot be genetically separated from them. Diez confessed his serious worries about this subclass in 1853 and his candor, in retrospect, is more impressive than the self-confidence that prevailed among a few of his successors. Actually, one discovers before long the existence of three separate, if interlock ing, questions: (a) How does this particular class, represented by words like 'guard', 'sentry', 'spy', 'scout', functionally compare with the remainder of postverbal nouns ? ; (b) What is its numerical representation in the individual daugh ter languages, and what conclusions can one legitimately draw from any significant variation in this respect? ; (c) Is the fairly high percentage of relevant words traceable to Germanic (a situation well-established) a mere accident, or is it a circumstance inviting close attention in any scrupulously nuanced explica tive statement? G. Lené, at the threshold of this century, organized his corpus in such fashion that the prime division was into "nomina actionis" (50-136), "nomina agentis" (137-140) — the very same words now under scrutiny —-, and "nomina instru menti" (141-144).57 The disparity in sheer size, and also in architectural in tricacy, between these three chapters is obvious, and one readily understands how G. Paris, in his paternalistic rôle of senior critic (1900), could have been sufficiently shocked by this discrepancy to reproach Lené with having introduced an altogether gratuitous division. The reviewer preferred to see in the agentives and the instrumentals merely further semantic developments of the abstracts. It is less easy to empathize with Gamillscheg when he declared, twenty-one years later, that these agentives were simply labels for a special variety of tools. In the context of his long-term investigation he might well have observed that, 57 Only with respect to action nouns did Lené operate with further subdivisions, first according to gender (masculines [62-105] vs. feminines [105-136]) and then, as the final classificatory step within either group, according to conjugation class. Interestingly, the tool-names have, at all times, been overwhelmingly feminine, perhaps for the sake of neater contrast to the rival type porte-bouteilles 'wine-rack', cf. agrape 'sort of drill', biffe 'sham, duffer', bine 'hoe', brie 'tool of the vermicelli maker', broie 'brake' (for hemp), bute 'farrier's knife', (obs.) comporte 'wooden tub or vat used by vintagers', drousse 'wool carding-comb', épuise 'tool used in draining', estampe 'punch', etc. (diminutives, such as ¿poussette ' dust-cloth, dusting-brush', obey the same trend). Masculines have been few and far between, and have been drifting toward gender-switch: affile 'knotted cloth bag used in sharpening' (m., f.), cache 'guard for screening parts of a photograph in printing' (m., f.).

THE SOCIAL MATRIX

205

with the rarest of exceptions, words equipped with instrumental derivational suffixes (e.g., those ending in -MEN or -MENTUM or -(ĀT)ŌRIUM) do not serve to designate living beings, let alone humans. (Fr. jument 'mare' is one such excep tion, easily explained away.) While the movement 'abstract' → 'tool' has been both recurrent and gradual in Latin and Romance, the movement 'abstract' → 'agentive' in military context is almost unparalleled and seems to involve a leap; there certainly is no need to posit two successive gambits — first, from 'abstract' to 'instrument', and thence, to 'agentive'.58 As regards numerical representation, French (particularly in its medieval garb) seems to have a slight margin over Spanish and Italian. Characteristic formations include: (obs.) avise 'mounted sentinel' ( = mod. vedette); (obs.) cerche 'spy', lit. 'searcher'; (obs.) crie 'crier, herald'; écoute, őrig. esc- 'listener, watcher' (→ E. scout), as distinct from écoute 'secret listening-in, hiding-place for listening', which underlies the phrase être aux écoutes 'to eavesdrop'; (obs.) espie 'spy', replaced by espion, which has been divorced from the verb épier; garde 'keeper, guardian, watchman'; (obs.) gaite > (obs.) guette ~ guête 'lookout-man, signal-man', lit. 'one lying in wait' (only guete 'wait, watch' has survived); (obs.) guye 'guide', for a while in competition with guion (cf. espion, above, and Hue ~ Ruon; note also Sp. guia 'leader' ~ guión 'hyphen'), which gave way to guide under Italian influence (cf. English usage); (obs.) huche 'hawker', lit. 'one who cries by whistling'; (obs.) regarde 'keeper, warden'; (obs.) veille 'watcher, watchman', cf. Sp. véla 'wakefulness'. Note also aide 'military assistant, helper' and observe how English manages to differentiate, in spelling, aid from aide.

The pertinent Old Provençal formations assembled by E. L. Adams (who interspersed them within the ranks of abstracts) are relatively few: crida 'crier', espia 'spy', gacha 'sentinel', guida 'guide', and aca 'crier' (see 547?n1). Significantly, not a single formation of this kind was exclusively peculiar to Southern France. While the shift 'abstract' > 'agentive', within the given military-custodial se mantic sector, causes no insuperable difficulty, it is noteworthy that only the feminine variety of abstracts — i.e., those ending in -e in French and in -a else where — have qualified for this treatment. The subsequent adjustments in gender vary, as is well know, from one Romance language to another, with Sp. el espía, el guía and It. la spia, la guida representing the two extremes, while older French exemplifies the gradual transition (witness the residue of older usage in OFr. la crie 'crieur public', la guye 'guide'). The consistent restriction of the series of masculine agentives to original feminine abstracts is certainly a factor of prime importance, to which any explicative account must attempt to do justice. 58 For all its brilliance, É. Benveniste's Noms d'agent et noms d'action en indo-européen (P., 1949) is not particularly helpful in this context. Part I ("Noms d'agent") is based, fundamentally, on Indo-Iranian and Greek evidence; though the author, toward the end, aims at distilling a kind of language universal from the distinction he has established between "auteur" ("désigné à partir de l'acte qu'il a accompli") -τωρ and "agent" ("caractérise l'être comme voué à une fonction" -τήρ), one is left wondering how far this dichotomy can be stretched with impunity. Citing Fr. chauffeur beside brûleur and E. founder beside shaker, Benveniste feels that, in derivation, machines need not be strictly distinguished from human performers. This statement raises the issue whether the transfer of a suffix from the realm of humans to the domain of tools (i.e., the cases cited by Ben veniste) can be treated on a par with an alleged transfer in reverse direction (a description that would accurately fit Gamillscheg's conjecture).

206


The concluding facet of the problem is the wisdom of appealing to Germanic influence (even more: to Germanic provenience) in reference to this particular subclass. This neatly pinpointed claim is not to be confused with Grimm's sweep ing contention that all Eomance postverbals may be traceable to Germanic, or with Diez's more modest typological analogy. The most articulate serious spokesman for a limited Germanic thesis was Lené, who adduced several weighty arguments (137-139), including the fact that this was a small but sharply -silhouetted group ("on est en présence d'une formation différente et indépendante"), not merely a marginal extension of some larger cate gory; that its obligatory ending -e is a tell-tale feature; that not a single Latin masculine agential in-a[like NAUTA 'sailor', say] has been organically transmitted into Romance, while the percentage of Germanic words was strikingly high; finally, that several characteristic instances of dual borrowing from Germanic (v. garder, n. garde; v. gaitięr,n. gaite; v. espier, n. espie59) could have produced a novel derivational model. The first instance of a transfer to Latin material in volved escolte (which, Lené might have added, looks and sounds like a word of Germanic parentage, despite its descent from A(U)SCULTĀRE). The single supposi tion this conjecture required was, upon the admission of its proponent, the shift from action noun to agentive, either weithin Germanic or at the earliest Romance stage. Despite its persuasiveness, Lené's hypothesis was either expressly rejected (G. Paris) or tacitly overridden (Meyer-Lübke, Migliorini, and many others).60 At this juncture it seems apposite to turn to Palaeo-Germanists for further guidance. In a characteristic medieval Germanic language such as Old Frisian, whose system of abstracts has of late been expertly surveyed,61 one finds num erous verbal abstracts in -β, all of them feminine and radical-stressed, e.g. : 59 Lené here correctly recognized the phenomenon of dual borrowing; cf. the paper, by a group of Berkeley Hispanists, "El nùcleo del problema etimològico de pícaro ~ picardía; en torno al proceso del préstamo doble", Studia Hispanica...R. Lapesa, II (M., 1974), 307-342. 60 To the implied rejections must be added the silence of J. Bruch, Der Einfluß der germanischen Sprachen auf das Vulgärlatein (Heidelberg, 1913), who slavishly followed Meyer-Lübke in recogniz ing traces of -IA, -   , -ONE, -ARD, and -ÁLD in Gallo-Romance, but allowed for the invasion of -ING only elsewhere (§17) and emphatically brushed off the possibility of Germanic influence on inflection, despite the rise of the declensional types -oj-one, -a/-ane at the dawn of the Middle Ages and despite the local blocking of the disintegration of the case system through Romano-Germanic symbiosis; cf. my remarks on grading in the note "The Double Affixation in Old French gens-es-or, bel-ez-or, Old Provençal l-az-or", StN, XLV :2 (1973), 217-225. Contrast Brüch's rigidity with E. Staaff's flexibility in reckoning with cross-fertilization. More helpful is Brüch's analysis of WARDON as a leader word (§46) and his semantic formula ('auf der Wache stehend ausschauen' > [1] 'auf der Hut sein, aufbewahren', [2] 'schauen'). The hybrid character of RE + WARDÔN (cf. Fr. regarder 'to look', etc.) is grist for our mill, showing as it does the coalescence of Latin and Romance elements in the military-custodial sphere. 61 L.-E. Ahlsson, Die altfriesischen Abstraktbildungen (diss. Uppsala, 1960), pp. 1-9. The author provides references to many other Germanic languages, and his copious bibliography contains further clues. Ahlsson's chapter "Zum Übergang in konkrete und kollektive Bedeutung" (226-232) furnishes isolated examples of abstracts gravitating toward agentives, e.g., wakinge 'Nachtwache' (227f.); cf. *Mnenge 'Hausgenossenschaft' (228), ervemeîthe 'erbfähige Verwandtschaft' (229), soldie 'Söldnerheer' (231), frôdhêd 'Geschworene' (232). From the chapter on "Konkurrenz und Chronolo gie der Suffixe" (233-241) one gathers that -e was forced to recede before -ethe and -hed in adjectival abstracts and showed even less resistance in verbal abstracts — a state of affairs implying its strength at an earlier period, namely the time of repeated clashes between Latin and Germanic.

THE SOCIAL MATRIX

207

bere 'threat' (lit. 'threatening gesture'), bihelde, -ilde 'custody', dêpe 'baptism' (cf. G. Taufe), fêle 'feeling', halse 'amorous embrace', hêmsêke 'visitation' ( = G. Heimsuchung), hende 'ar rest', here 'hearing',  'complaint', lungenskrêde 'wound in the lung', rende 'tear, rend', weiwende 'blocking of the road, attack on a road', werde 'violation, wounding', werde ~ wirde 'formulation, testimony, proof', ondwarde 'answer' (cf. G. Antwort), were 'defense' (cf. G. Wehr), lond-, land-were 'defense of the land', liudwere 'people's defense', nêdwere 'emergency defense'(= G. Notwehr), hêrwere 'possession through lease', hof wer e 'ecclesiastic possession', râfwere 'possession through robbery'.

Other Palaeo-Germanic languages offer almost exact parallels; as a matter of fact, modern Standard German has not drifted far from this model. Particularly remarkable is the circumstance that, within the semantically motley group of formations surveyed, a significant percentage of items carried meanings which had already crossed our path in our earlier survey of Late Latin and Romance postverbals: 'threat', 'custody', 'attack on a road', 'defense'. If we now make a point of recalling to what extent Germanic warriors, at first individually recruited and, from the 4th century on, used in ever larger contin gents as allies and auxiliaries, formed the backbone of the Roman Empire's armed forces, it becomes plausible that we are here witnessing the conflation of two currents — a process conducive to a high tide. On the one hand, there existed a small contingent of native Latin postverbals; some of them pertaining to -ĀRE verbs; shorn of any prefix; feminine; abstract; and associated with military practices (PUGNA, LUCTA, MINA); plus the feminines in -IAE (EXCUBIAE, ĪNSIDIAE, VIGILIAE) — which surely lent themselves to de-pluralization, suggestive of small-scale military operations and sentry duty. On the other hand, one recog nizes a block of Germanic rhizotonic verbal abstracts exemplified by the material just cited from Old Frisian sources. Lat. MINA and Gmc. bere both meant 'threat, threatening gesture or mien'. In the bilingualism which must have been wide spread and deeply rooted in Roman military camps, compromise forms like *spëha and * warda were almost inevitable. Best of all, the interpretation here offered for the gitarda — gui(d)a — espia — escolta type may, mutatis mutandis, stand us in good stead in assessing the Romance postverbals as a whole. The military-custodial subclass is the one most sharply characterized (a rank which by no means implies that it was the most productive one 62), inasmuch as we here actually encounter a small block of solidly entrenched Germanic stems, a few of which — particularly the couple guardar — guarda — are likely to have infiltrated Latin long before the dis integration of the Empire. Within the larger and, all told, far more successful group of verbal abstracts, the preponderance of Latin word families was never 62 The tendential replacement in Spanish and Italian of guarda by guardia, less tightly welded to guardar(e), is one symptom of the subsequent decline of this type. In Chilean folktales, one finds Aguaitador as the name of a protagonist whose son is called Aguaitin Aguaitan, in preference to Palaeo-Romance guaita; see M. R. Lida, El cuento popular hispano-americano y la literatura (Buenos Aires, 1941), p. 9.

208


challenged, though Germanic speech habits may have intensified or accelerated an erratic or, at least, highly atypical trend within Latin folk speech. Nineteenth-century pioneers found it easy to produce scattered Latin models for Romance postverbals, but these assumed prototypes gave the impression of amorphous débris of archaic usage,63 and the configuration of such isolated words seemed sharply at variance, as regards all syllabico-accentual conditions and the morphological design, with the models that emerged as dominant in Roman folk speech. What appeared most urgently needed for any explanation to acquire plausibility was the discovery of a particular strain — a social dialect, a voca tional jargon, if you wish—within the ensemble known as "sermo plebeius" that would have justified the adoption and sudden rise of such a model. The key word turned out to be PUGNA, and once LUCTA joined it, the formal similarity of these words ensured their further success. Contact was established with FUGA and MINA, and the coinage of *FALLA followed suit. The social matrix within which this development occurred was, to all appearances, Roman military life and martial sports, including the circus and the gymnasium, where "clipped" terms, possibly used in laconic commands, certainly made better sense than derivatives heavily loaded with strings of suffixes. The genesis of the parallel masculine series — originally smaller — does not yet stand out with equal clar ity; the molecule comprising the primitives DAMNUM and LUCRUM, plus the re gression COMP(U)TUS, all three against the backdrop of -ĀRE verbs, provides the best starting-point. It is remotely conceivable that within the abstract-collective domain the coexistence, at least in certain provinces, of other masculine-feminine pairs (e.g., in names of containers) may have provided a minor stimulus, offering a model of symmetry. 64 In any event, the existence of a Romano-Germanic connubium in the military sphere is hardly needed to explain the origin of the postverbal abstracts, though it surely accounts for the impetus of their subse quent cross-lexical diffusion. University of California, Berkeley 63 Suffice it to re-read in this connection M. Bréal's note — surely not one of his best — "Noms postverbaux en latin", MSLP, IV (1881), 82f. The author mentions just a handful of variegated nouns: ARMA 'defensive weapons',LIBUM'(pan)cake offered to a deity', LŪCUS 'grove or thicket of trees which are sacred to a deity', lit. 'clearing' (beside COL-, SUB-LŪCĀRE), plus two participial adjectives (ADULTER, TRUNCUS) and a number of mythological names, some of which involve a prefix: Subigus, Ante- and Post-vorta, Per-tunda, Per-fica. 64 See my forthcoming paper "Gender, Sex, and Size, as Reflected in the Romance Languages", to appear in Studies in Romance Linguistics, ed. M. P. Hagiwara.

E X C U R S U S : OLD SPANISH ONTA

VS. ON(D)RA

—

ANOTHER CASE OF LEXICAL POLARIZATION?

The further growth of postverbals offers, literally, hundreds of challenging individual problems, which should prove of distinctly greater appeal at present than a half-century ago, at the height of the Wörter und Sachen fad. Here is a thumbnail sketch of just one such problem, which seems particularly recalcitrant.

THE SOCIAL MATRIX

209

There exists a rich and stimulating literature — both narrowly technical and broadly cultural — on the vicissitudes of Lat. HONOS, later HONOR -IS beside the verb HONORARE in Romance, including the Luso-Hispanic branch o f t h a t family; but certain questions seem to have consistently eluded the curiosity of highly-qualified explorers. I n most daughter lan guages, the noun and the verb have both survived, a but not independently — their farreaching interaction is one point inviting further research — and not necessarily on the same socio-educational level of discourse. Only in Hispano-Romance, however, has a postverbal noun arisen, by way of bold innovation, and has, in part or wholly, encroached upon the traditional domain of HONÖRE. Why do Spanish and Portuguese alone boast a newlydeveloped, regressively derived abstract, honra 'honor', whereas Fr. honneur (→ E. honor), It. onore, etc. quietly perpetuate the Latin prototype? How have honra (OSp. ondra, onra) and honor (OSp. onor) managed to coexist in Spanish, whereas in Portuguese honra has suc ceeded in crowding out its potential rival ? And why did speakers of Luso- and Hispano Romance, in the case of this culturally all-important word, opt for a feminine postverbal, to the strict exclusion of its readily imaginable masculine counterpart, *(h)onro Certain difficulties t h a t have beset the growth of this family are traceable to the Latin — indeed, to the proto-Latin — stage. The relationship of the subst, HONOS to the adj. HONESTUS is erratic; some scholars feel that there must once have existed a neuter var. *HONÖS -ERIS (cf. DÉÇUS -ORIS [.] beside DECOR -ORIS [m.] 'beauty, grace, charm, ornament'); if so, HONESTUS was transferred from doomed *HONOS to vigorously thriving HONOS and, eventu ally, during the early years of the Empire, to HONOR. 13 Then, the relation of HONESTUS to the correlated abstract HONESTAS involves a stumbling block; one is free to assume the agency of haplology in *HONESTITAS and to cite the parallel of TEMPESTAS 'time, season, stormy weather', in lieu of *TEMPESTITAS, alongside TEMPUS. c HONESTUS shows no signs of word-ofmouth transmission to the vernaculars; but, surprisingly, D(EH) ONESTARE 'to disgrace, dishonor', subject to violent distortions (reminiscent of the vicissitudes of PRAE-HEND ERE 'to grasp'), lingered on in medieval Spanish and Portuguese: cf. OSp. denostar, beside OPtg. doestar. Conceivably, association with MOLESTARE 'to annoy' and with MONÉRE/MONITÄRE 'to admonish' could have supported the verb at the critical evolutionary step. d HONÖRE survived into Old Spanish as onor, a word of fairly wide (though, as we shall see, uneven) currency and one t h a t looks as if it had been "regularly" transmitted. This impres sion is correct, so far as the consonantal scaffolding is concerned ; e on the vocalic side, how ever, onor could very easily have become *enor (cf. ROTUNDU 'round' > *rodondo > redondo and numerous developments), with the further risk of a "dangling" front-vowel being, in the end, completely sloughed off: ÉBRIACU 'drunk' > , in harmony with Ascoli's L a w / The fact that the initial segment of onor was preserved relatively intact is due to two cir cumstances: (a) continued association of t h a t noun with the powerful verb on(d)rar (plus certain offshoots of t h a t verb: ondrado, ondradamientre, ondrança); (b) the uninterrupted a Rumanian is one major exception: mod. onoare (pi. onoruri) and the corresponding verb a onora do not rank as indigenous formations. The Latin Index to S. Puşcariu's etymological dictionary contains not a trace of the ancestral HONÖRE family.  Most of the following remarks are based on Ernout and Meillet's magisterial treatment {DÉLL4, pp. 2986-299a). c Despite the minor anomaly, HONOS and HONESTUS were, of course, closely associated in classi cal Antiquity; for details see F. Klose, Die Bedeutung von "AOTIOS" und " honestus" (Breslau, 1933). d For three possible explanations of VLat. * ADMONESTARE 'to admonish', see REW3 §180. Amonestar was a fairly current verb in Old Spanish, equated with SUADERE by late-medieval glossators (see A. Castro, Glosarios latino-españoles de la Edad Media, M., 1936, s.v.); cf. Grail Fragments, 27Gr°; Barlán y Josafat, ed. Moldenhauer, 189r°, 196v°. e I am putting to good use — here and immediately afterward — the many excellent remarks made by F. Hanssen, GHLC (1913), and by Menéndez Pidal, MGHE6 (1941); for guidance see the respective indexes of words cited. f See HR, XIV (1946), 104-159, and RPh, XXVIII:4 (1975), 483-520.

210


cross-connection with HONŌRE, which was widely used in legal and liturgical Latin — a very relevant frame of reference if one recalls t h a t OSp. onor also served — among other functions — as a legal term (for details, see below). However, it cannot be argued t h a t onor was a straight Latinism; e.g., as regards gender, it was for centuries preponderantly feminine,^ and while the occasional switch in the direction masculine → feminine in such derivatives was nothing unusual in the older language, especially where it accompanied the wavering be tween -or and -ura,h the tendency, for a while, was discernibly stronger in the case of honor, despite the fact t h a t speakers refrained from experimenting with *honura. Gender-switch, consequently, must be explained in some different way; protracted coexistence of onra and onor, and a generous measure of semantic overlap between them qualify as good reasons. 1 If we now ask ourselves why HONÖRE has remained without representation in Portuguese, the answer is fairly simple: Given the local vulnerability of intervocalic -n-, the expected spontaneous outcome might have been something like *oor, *eor with further prospects of changing to, say, *or or ônhor; and while a few words have withstood such severe mutilation (or, alternatively, have had their -n- restored in the end, probably through welcome pressure from neighboring Leonese), the ready availability of onra, with which *oor, in most contexts, would have been practically interchangeable, made all such emergency solutions gratuitous. The speech community simply dropped the troublesome descendant of HONÖRE, rejoicing in the remedy offered to them by onra (later spelled honra). This clearing of the ground leaves us saddled with the responsibility for explaining the rise of on(d)ra, in preference, above all, to ôn(d)ro — though on(d)ra, in the end, also dis lodged on(d)rançaJ The one conceivable argument t h a t must be rejected at once is t h a t speakers, for reasons of euphony (variety, clarity, and the like) preferred the sequence —a to its rival —o: The préexistence of logro < LUCRU and lograr < LUCRARE; the mintage of corro, a-, so-corro from native correr, etc.; and the acceptance of torno beside tornar and of aforro (mod. ahorro) beside aforrar from the lexica of adjoining cultures all three at once prove this supposition wrong. k The clinching factor t h a t remains to be identified must represent a force independently known as having been active on the Peninsula, during the crucial period; and the constellation of favorable circumstances yet to be reconstructed s Cf. the explicit statement to this effect, and the abundant documentation, in Menéndez Pidal, ed. Cantar de Mio Cid, pp. 776-779; the individual shades of meaning: (a) 'gloria, esplendor', (b) 'demostración de respeto', () 'heredad, patrimonio', (d) 'usufructo de las rentas de alguna villa o castillo realengos' apparently exerted no controlling influence on the gender, except that the re current phrases honores et tierras, tierra e onor (amply illustrated by the commentator in support of the fourth meaning) could very well, in conjunction with onra, have deflected onor from its straight course. This early phase in the history of onor seems doubly remarkable when one remembers that in French — a language which has sanctioned the use of la couleur, la rougeur, etc. in the face of steadfastly masculine Latin usage (COLOR, RUBOR) — honneur, of all abstracts, has, in the long run, remained faithful to the ancestral preference. (Recall the long-drawn-out wavering, in this respect, oí amour.) h I am thinking of such doublets as dulç-or, -ura 'sweetness' (from DULCÖRE, but refashioned after dulce 'sweet'), amarg-or -ura 'bitterness', etc. Cf., in modern Portuguese, the feminines côr 'color' and dor 'pain'; and in modern Spanish, (m.) grosor 'bulk, thickness', but, occasionally, in poetry (f.) calor 'warmth' — an echo of obs. calura, which underlies caluroso 'warm', (fig.) 'affectionate' ? * Interestingly, for Juan Ruiz onor — a word he used sparingly — had already reverted to its original masculine status: "Tiróle su poderío Dios e todo su onor՝՝ (305d). j Concerning on(d)rança see Menéndez Pidal's Cid vocabulary; judging from my own excerpts (see UCPL, 1:4 [1945], 108, 135), this derivative was not otherwise used either in Old Spanish or Old Portuguese. The Latin model adduced or reconstructed by Menéndez Pidal is unfamiliar to me; Church Latin (Symmachus, Ambrose, etc.) distinctly favored the compound HONÖRIFICENTIA. k Some of these cases may involve dual borrowing; in a few instances the pair may represent not a primary verb and a noun extracted from it but, conversely, a primary noun and a verb added by way of superstructure. In any event, a team such as ôn{d)ro ~ on(d)rar would not, for phonic reasons or for lack of precedents, have run against the speakers' grain.

THE SOCIAL MATRIX

211

must be shown to have been peculiar to the given time and place far more than to any comparable spatio-temporal context. I t is the contention of this note t h a t the prime stimulus responsible for the coinage of on(d)ra came from its semantic opposite, onta 'shame, disgrace', with which it came to share a number of highly relevant features — prosodie, phonological, and grammatical: (a) the number of syllables, (b) the stress pattern, (c) the number of phonemes in such dialects as favored onra /onRa/ over ondra.1 (d) the distribution of vowels in both syllables, (e) the presence and position of one consonant phoneme, namely syllable-final n, with additional support drawn from the fact that onta's է and ondra's d were closely-related dental stops; (f ) the gender. In short, it is difficult to imagine closer formal proximity in the face of almost diametrical semantic opposition. Now, Hispano-Romance is well-known for its special predi lection for such piquant clashes: One finds here not only the expected, " t r i t e " examples of lexical polarization (e.g., 'right' vs. 'left', 'up' vs. 'down', 'mother-in-law' vs. 'daughter-inlaw', etc.), but certain peculiar instances of interlocking, less easy to detect, as when 'from' (desde) and '(up) to, until' (hasta, OSp. ata, fata, fasta) have, to some extent, conditioned each other's shape and trajectories or as when, at a certain point, 'first' (primero) signifi cantly influenced 'last' (OSp. postrimero, in lieu of expected ^-emero, from POSTRÈMUS; later reduced to postrero, in deference to the trisyllabic structure of primero).m I t is not inaccurate to say that French, in this respect, either has lagged behind Spanish or, alternatively, has, in course of time, veered away from this pattern: If it is true t h a t premier and dernier to this day rhyme, it is also a fact that dessus and dessous are less intimately linked to each other than were, in Golden Age Spanish, suso and yuso; again, belle-mère and belle-fille lack the thread of a shared stressed vowel, especially a heavy diphthong, of the kind that connects Sp. suegra and nuera, etc. One may now broach the question of the penetration of onta into Old Spanish and of its temporary entrenchment to the south of the Pyrenees. Onta represents a borrowing from OFr. honte /hônto / 'shame' (which, in turn, goes back to Gmc. haunipa); more accurately, it is just one reflex the speakers toyed with, the alternative being fonta, with ƒ- representing an attempt to render locally non-existent /h/ — a substitution independently familiar from the history of numerous Arabisms, including the name Maf ornada The biography oi f onta had its moments of irony, as when ƒ-, secondarily, was converted into h- shortly before reaching the stage of muteness, so that, for a while, Late OSp. (h)onta moved closer to OFr. honte than at the time of the actual contact and transfer. 0 The medieval variant of immediate concern to us, however, is the rival form onta. Both forms are copiously attested in literature 1 As Menéndez Pidal correctly emphasized, the solution -nr- of syncopated -N'R- involved more than mere preservation of the ancestral consonants; it also entailed the change /r/ > /R/. It is probably true that, in terms of sheer lengths of articulation, /R/, with its multiple vibrations, has been, approximately, the equal of the alternative ~dr- cluster. m For full details I refer to my forthcoming article, "Español antiguo des(de), fa(s)ta, fazia y fascas", to appear in the Madrid venture Homenaje a J. Caro Baroja; and to an earlier note, "Diachronie Lexical Polarization Once More: The Case of Spanish primero — Old Spanish postri mero — Classical Spanish postrero", BBL, XX :5 (1970: "Hommage à A. Rosetti"), 523-526; both pieces with clues to earlier writings. n The examples usually adduced are facha 'ax' (as distinct from facha 'ugly face', an Italianism of transparently later vintage), faraute 'herald' (cf. Confisión del amante, 68°, 171°), and j՝ardido 'bold'. Viewing the process under a stronger lens, one recognizes that OFr. hardi was normally rendered in the Juan Ruiz MSS by some such graphy as ard-it, -yt, -iz, while the corresponding feminine emerged as ardida and the adverb as ardidamente (cf. 52α, 64α, 455d, 482α*, 509c, 11196, as established by H. . Richardson); but MS G, known for its "folksy" tone, shows one lone trace of jurdida. It is not implausible that in some instances the loss of h- occurred "en route" to the South, as it were, somewhere in Provence; cf. arenque 'herring' (Ruiz, \\\2d), from Gmc. HARING via Occ. arene. However, in the language of the troubadours 'shame' was ordinarily expressed by vergonha, a word of diaphanous Latin descent. 0 This point was commented upon by Hanssen, GHLC, §13.

212


and, indeed, served as starting points for the verb a(f)ontar.p Another, more hidden mani festation of their powerful impact could very well have been — unless our conjecture has led us astray — the crystallization and subsequent rise to extraordinary prominence of on(d)ra,ą as onta's obvious semantic opposite; the contrast was further dramatized by the scale of values inherent in the given culture. Note that on(d)ra, in sharp contrast to onor, did not refer to any rights or possessions bestowed by royalty, but was confined to such semantic nuances as 'enjoyment of prestige, power, and splendor', Outward manifestation of this exalted status', 'esteem, respect, and admiration shown by others toward a person enjoying such a privilege', and the like; so seen, 'honor' was very close to 'fame' and 'glory'. The subsequent development of a conventional code of honor and its ultimate sublimation in the thinking of a few individuals (like Cervantes), who morally surpassed others of their age and environment, are topics which, for all their fascination, transcend our chronological limits by too wide a margin to warrant discussion. r The influx of Gallicisms (including Provençalisms), chiefly at the crest of the Cluny Re form, is a well-established fact, but their resultant inroads have not yet been exhaustively investigated. The better-known cases of interference involve such situations as the overlay of frai(re), frei(re) over native frade 'brother, friar' < FRATRE; and of monje over native monago 'monk' (cf. monaguillo beside monacillo 'altar boy, acolyte'); or the intricate rela tionship of precio 'price' and prez 'praiseworthy action' > 'honor, glory, worth'. s To this sort of subtle interweaving we can now safely add the drifting of the medieval Spanish speech

p Cf. F. éhonté 'shameless', scaffolded on honte, as distinct from honnir, which echoes directly the Germanic model. The prefix e- is reminiscent of the opening segment of dé-vergondage. զ This rise is observable through the prism of (h)onroso 'honorable' (behavior, position, etc.), an innovation unmatched by anything indulged in by the cognate languages. Also, honrado — developed from the verb but concomitantly supported by the noun — typically means 'honest' and has severely reduced the scope of the Latinism honesto, which has enjoyed unqualified success in French, Italian, and elsewhere; 'honesty' has traditionally been expressed by honradez rather than by honestidad, though the ratio of usages may be undergoing a reversal. Also, over against Fr. déshonneur (→ E. dishonor) Spanish has at all times favored des(h)onra (cf. Ruiz, 262a, 264c£, 265a, 679Ò, 681c, plus the introductory prose fragment, 6.16 in Ducamin's ed.) over deshonor, and, correspondingly, desonrar (Ruiz, 885a) over deshonorar, even though the Poema del Cid allowed for competition between desondra and (f.) desonor, to the point of tolerating the phrase α grant desonor (suggested by trans-Pyrenean models ?); but even there the sole verb available was desondrar. The extra-wide margin of on(d)ra over onor in this negative series constitutes added support for the contention that the pressure of onta was behind the rise and expansion of on(d)ra. See also fn. t, below. r The degeneration of honra, both social and lexical, has led to such facetious offshoots as the mock-diminutive honrilla '[exaggerated] concern for what people will say'; cf. the colorful phrase por la negra honrilla. The classic inquiry into the rehabilitation of this concept is Americo Castro's — apparently unfinished — article, "Algunas observaciones acerca del concepto de honor en los siglos XVI y XVII", RFE, III (1916), 1-50. On the idea and ideals of'fame' and 'glory' in older literature see the respective book-length monographs by María Rosa Lida de Malkiel and Fran çoise Joukovsky-Micha. s Cf., in Tuscan, prezzo vs. pregio. The situation is even more complicated in Portuguese, where (a) borrowed, (b) native vernacular, and (c) native learned forms can all be set off: (a) prezar, desprezo, desprezivel; (b) preco, menospreço; () apreciar; see my analysis in "Préstamos y cultis mos", RLiR, XXI (1957), 1-61. Similarly slanted explorations include: Anita Katz Levy, "Con trastive Developments] in Hispano-Romance of Borrowed Gallo-Romance Suffixes", RPh, XVIII, 399-429; XX, 296-320; and S. N. Dworkin, "Mester and menester; an Early Gallicism and a Cognate Provençalism as Rivals in Older Gallo-Romance", ibid., XXV, 373-389. While hypothe sizing Gallo-Romance pressure on selected derivational suffixes is nothing startlingly new (both Hanssen and Menéndez Pidal were, for instance, thoroughly aware of the heavy share of borrow ings in the genesis of the model of Hispanic postverbal nouns in -e), the assumption of such pressure on the verbal paradigm — as played with, not so long ago, by H. Lausberg in his account of the OSp. and OPtg. p. ptc. in -udo — is truly unprecedented.

THE SOCIAL MATRIX

213

community toward one type of postverbal, to the disadvantage of its equally available rival — through the instrumentality of lexical polarization stimulated by borrowings But, granted the validity of the hypothesis here advanced — to the effect that, on Spanish soil, imported onta significantly influenced the sprouting of a postverbal within the semantically-opposed family of HONORARE — why, one may ask, did not some kind of similar con tamination occur in France, where the reflexes of haunipa and of HONÖRE/HONÖRARE were likewise available and were, beyond dispute, endowed with identical semantic potentialities ? The answer is that, for two disparate reasons, no dramatic clash could possibly have occurred to the north of the Loire : First, OFr. honte, with its solid consonantal onset, and its incipient strong nasalization of syllable-final n, on the one hand, and the reflexes of HONÖRE, namely enor, onur, etc., u deprived of their ancestral - for over a millennium, characterized by the weakest conceivable nasalization (if any), and sporadically subject to vowel dissimilation — > e—, later e— (cf. ROTUNDU > OFr. reont > rönt, mod. rond), on the other, were by no means outwardly resemblant and thus fell short, on the all-important formal side, of triggering any associative process. Secondly, in Gallo-Latin the verb HONORARE — for some reason or, more plausibly, for some interplay of reasons — refused to undergo compression through syncope. v We thus find, from the start, enor er, its e controlled in part by the trend toward vowel dissimilation, in part by the interference of the prefix en-, typical of tri- and polysyllabic verbs. Given the moral or moralistic overtone of noun and verb, it is small wonder that, through coexistence of medieval Latin HONORE and HONORARE (both used in legal and ecclesiastical contexts), speakers and writers rejoiced in adopting culturally more prestigious Latinizing forms: honneur, honorer, etc. As a minimum concession, speakers of all shades preserved the endangered intertonic vowels. Inspecting the situation from this angle, one again reaches the conclusion t h a t there was not the slightest prospect of having honte pitted against enur, onur; in fact, no postverbal — whether masculine or feminine — from enor er, honorer would have been sharply enough silhouetted against enur, honneur, and so none came into existence. Italian, like Spanish, did borrow from Gallo-Romance onta 'shame, disgrace'; in fact, it went — as usual — beyond Hispano-Romance in its subservience to French models, opening its gates to ontoso 'shameful, disgraceful, ashamed' (from honteux) and using the phrase  Though I examine onta/fonta elsewhere in adequate detail, there may be some point in supply ing here, at least, skeletal documentation of their record. Onta has left traces in a number of "Western" texts, which, on internal evidence, may go back to (lost) Leonese versions: De una emperatriz... (eel. A. Mussafia), Chs. 7, 11 (twice), 27; Carlos Maynes (ed. A. Bonilla y San Martín). Chs. 4, 10, 17, 21, 23; Grail Fragments (ed. K. Pietsch), fol. 276r°. The first of these texts also contains ontar 'to (put to) shame' (Chs. 4, 33), which flanked aontar. The Poema de Mio Gid favored ondra (and offshoots) beside fonta. u The leading etymological dictionaries of French (EWFS, NDÉH, DÉLF, etc.) are, for once, agreed on all essentials. The fact that OFr. enur, etc. tended to be feminine and included, among its semantic components, the meaning 'fief, domain' is noteworthy and indeed calls to mind OSp. usage, as observable in the Cid. A modicum of influence may have occurred; but it was undoubtedly counterbalanced by local pressure of notarial Latin — witness the steadfastness with which the Peninsular scribes preserved the o- of the first syllables, curbing any temptation to imitate OProv. enor, anor; OFr. enur, etc. v One reason why speakers of many vernaculars — outside the Peninsula — have been so staunchly opposed to syncopating HONÖRÄRE was presumably the hazard of a very inopportune

collision of the families of HONÖRE and HORRORE. (The paradigms of TENERE and VENIRE disclose

in how many places N'R has yielded, or tended to yield, -rr-.) Actually, one finds, in Old Italian, sporadic traces of orrato 'honored' and even of orrevole for 'honorable'; cf. . Wiese, Altitalienisches Elementarbuch (Heidelberg, 1904), pp. 48 and 58. So onorare, strictly speaking, involves regression or restoration. Then, of course, the coexistence of verb and noun had its share of responsibility: Since HONORE was not amenable to syncope, it also shielded from erosion, in certain areas, the in tertonic vowel of HONORARE. Note that in Spanish a leveling occurred within the verbal paradigm; HONORO would, in isolation, have produced ¿.onoro, but on(d)ro was substituted instead in deference to the majority of the forms: on(d)ramos, etc.

H. LEXICAL INDEPENDENCE VS. GRAMMATICAL CONSTRAINT

EACH WORD HAS A HISTORY OF ITS OWN I The classic example, on the scene of nineteenth-century linguistics, of the almost quixotic rebellion of one man against a whole powerful and solidly entrenched school of thought is provided by H. Schuchardťs onslaught upon the Junggrammatiker ("Neogrammarians"). Because Schuchardt, before becoming a general linguist of note, had his early training in the Romance field, it is not impossible that the well-known idiosyncrasy of Romance linguistics, which continues to this day, was, to some extent, predetermined by his politely refractory attitude toward the dominant trend, although other forces were equally at work.1 What matters in the framework of this conference,* however, is the fact that the evolution of this partic ular phase of scientific thinking falls short of showing the ideal reconciliation between the rigidified old and the revolutionary new through the crystallization of some revised and more elastic orthodoxy, as it were. Schuchardťs originality shone through in his championship of two causes: macrocosmically, the wave theory (as against, or alongside, the projection of linguistic filiation in family-tree style);2 and microcosmically, the infinite variety and unique ness of word biographies or lexical vignettes ("Each word has a history of its own"). The connecting link between the two causes, the examining of facts at successive stages of balance and congealment, is the dynamic, rather than the static, view of change in language. For simplicity's sake, we shall here disregard the vicissitudes of the wave theory and concentrate on efforts to reconcile the conflicting postulates of historical grammar, with its inevitable emphasis on pattern, and of individual lexical growth. Though some excellent practitioners have brought about a fruitful rapprochement between Neogrammatical and Schuchardtian linguistics (especially noteworthy, in this respect, is R. Menéndez Pidal's trailblazing Orígenes del español, 1926; rev. ed., 1950), there has been no corresponding perfect merger on the theoretical level. The temporarily influential Swiss scholars (other than those of the Geneva Group, which was indifferent to etymology) have allowed the pendulum to swing very far, not to say perilously far, in the direction of individualism, remain*) This paper was originally presented to a Wenner-Gren Foundation for Anthropological Research symposium on "Revolution vs. Continuity in the Study of Language", Aug. 15-25, 1964. 1) H. Pedersen, The Discovery of Language; Linguistic Science in the Nineteenth Century, tr. J. W. Spargo (Bloomington, Ind., 1962; Engl, orig., 1931; Dan. orig., 1924), distorts the picture by portaying Schuchardt chiefly as a Basque scholar, pp. 124-127; for an attempt at idealization, see I. Iordan and J. Orr, An Introduction to Romance Linguistics, its Schools and Scholars (London, 1937; Ruman, orig., 1932). Cf. also my paper "Distinctive Traits of Romance Linguistics", to appear in Dell H. Hymes' Reader in Linguistic Anthropology. 2) On this point, especially on the priority of Schuchardt (1866) vis-à-vis J. Schmidt (1872), see my note "An Early Formulation of the Linguistic Wave Theory", Romance Philology, IX (1955-56),31.

218

LEXICAL INDEPENDENCE VS. GRAMMATICAL CONSTRAINT

ing aloof, en bloc, from efforts made elsewhere to refine the very concept of sound change. On this side of the Atlantic, L. Bloomfield was presumably aware of the gap between the implications of the conservative "comparative method" and of explosive "dialect geography", but did not, despite strenuous endeavors to this end, actually succeed in bridging it. II The dictum "Chaque mot a son histoire" has customarily been ascribed, by friend and foe alike, to Jules Gilliéron,3 that colorful French-Swiss scholar who, around the turn of the century, through his Atlas linguistique de la France launched the technique of massive cartographic projection of dialect geography and, shortly after, in a series of startling essays, examined the effects of homonymic clashes. This widespread belief in Gilliéron's authorship involves a dual oversimpüfication. On the one hand, Gilliéron, admittedly an indefatigable toiler and a man endowed with an unfailing flair for, shall we say, "detective" work in linguistic reconstruction but certainly no outstanding theorist, relied heavily on the truly original thinking of Schuchardt, to whom, characteristically, he dedicated — on the occasion of the revered master's seventieth birthday — the first collection of his pioneering essays.4 While the Swiss dialectologisťs own position vis-à-vis the Neogrammarians' platform had been one of dramatic volteface, if not of indecision,5 Schuchardt had, from the start and in quite unequivocal terms, declared himself in flagrant disagreement with that influential school of thought,6 a situation which immediately raises the key problem: Is the dictum under study at all compatible with the assumed "excep tionless" regularity of sound change, an axiom advocated by the Junggrammatiker? At the opposite end of the spectrum, we discover, counter to expectation, that a man of L. Bloomfield's stature — a scholar who, in sharp contrast to the consensus of European structuralists, almost passionately identified himself with the Neogrammarians' cause, program, and dogma — also unhesitatingly espoused the maxim emblazoned on our agenda. You will find it, itaHcized for emphasis, in his celebrated book Language (1933); to be specific, in Section 19.4 of the chapter on "Dialect Geography", with a reference to Karl Jaberg's eye-opening pamphlet Sprachgeo graphie (1908).7

3) Cf. Winfred P. Lehmann, Historical Linguistics: an Introduction (New York, 1962), p. 126, and, among the book's reviewers, W. Sidney Allen, in RPh., XVII (1963-64), 170-75, esp. 171. 4) Etudes de géographie linguistique (Paris, 1912), written in collaboration with M. Roques; for some essays included in that miscellany Gilliéron had enlisted the help of J. Mongin (d. 1910), another admirer of Schuchardt's (so the Preface reports). 5) In Lang., TV (1928), 288, fn. 6, Bloomfield asserted: "Schuchardt, through a historical accident, stood aside from the theoretical formulations of his contemporaries. Gilliéron, in spite of much vigorous and pleasing metaphor, insists [1918] upon l'implacable régime de la phonétique'." What about such Gilliéronian formulas as "la faillite de la phonétique", "mirages phonétiques"? See also Lang., VIII (1932), 231. 6) Über die Lautgesetze. Gegen die Junggrammatiker (Berlin, 1885). 7) See Jaberg, p. 6 ; Bloomfield, pp. 328 and 520.

EACH WORD HAS A HISTORY OF ITS OWN

219

At first glance there seems to be no serious incompatibility between individual word history and the postulate of regular sound change. No scholar, not even one of the most austere "regularist" persuasion, has denied the possibility of non-phono logical changes (say, those controlled by analogy) being superadded to the norrrţal "expected" transformations. It is further generally conceded that the agency of dialect mixture may add extra ingredients to the composite whole; the dialects involved may, in turn, be either regional or social (if the latter situation prevails, some scholars would favor the metaphor of different levels, channels, or conduits of transmission - vernacular, learned, partially learned). The existence and importance of semantic change has, in principle, never been disputed, though there has been no unanimity of opinion on the wisdom or even the possibility of dealing with this phenomenon in terms germane to linguistic analysis. Finally, depending on their general philosophies, some scholars do and others do not operate with "expressive" change; a few experts attach overriding importance to the avoidance of homonymy and polysemy, while others belittle this factor, etc. Perhaps the answer to our initial question of reconcilability between the two maxims lies in this: Assuming all changes, including those of the semantic order, could be calculated and recorded in percentages, it is clear that a segment of a given evolutionary line containing 80 % of readily predictable, "regular" changes (of the type called "phonetic") and 20% of all the others would tend to bear out the Neogrammatical line of reasoning. On the other hand, in an environment favoring extra-lively analogical changes (through morpho logical leveling and lexical contamination),commingling a variety of dialects, provoking social tensions and upheavals, etc., one can visualize a radically different distribu tion of transmutations, say, 10 % of the phonetic type and up to 90 % of all the other categories viewed jointly, categories which have in common a far less sharply profiled patterning than the sound correspondences. In such a climate, the classic "sound laws" would not necessarily be abolished (in fact, their residue might be doubly important for genetic reconstruction), but they would, psychologically, lose a good deal of their immediate appeal and certainly could no longer, without grave damage, dominate the scene of linguistic research. If, under such circumstances, the etymologist undertakes to play first violin, the entire character of the performance undergoes a profound change. Because etymology is inconceivable without an appeal to the interplay of several disciplines, its predominance spells the end of any isola tionist trend in linguistics. Not for nothing was it Schuchardt, that master etymol ogist par excellence, who was particularly active in launching the "Wörter-undSachen" movement. To the slogan "Every word has a history of its own" one thus feels constrained to add: "And one cannot observe it through the linguistic prism alone". The difficulty is immeasurably compounded if the areas occupied by such word variants as are differentiated solely by phonological discrepancies show unequal geographic extensions of identical sound shifts. The point is that a normal phonetic sound change — in marked contrast to that special family of transmutations some what indiscriminately termed "sporadic", "spontaneous", "saltatory" changes (metathesis, haplology, dissimilation, etc.) ֊- is, by definition, limited to a certain span of time and to a certain territory. Not a personal caprice of Gilliéron's, but

220


the stark fact that the maps of his atlas revealed major divergencies in the extension of conventional sound laws as exemplified by individual words led him and his adherents to question the validity of Neogrammatical doctrine. In this context it is rewarding to examine closely Chaps. 18 ("The Comparative Method") and 19 ("Dialect Geography") of a book reputedly as monolithic as Bloomfield's Language. As is well known, that scholar, emphatically and on many occasions, espoused the Neogrammarians' cause. Somewhat less familiar is the fact that Bloomfield, for many years, was almost equally fascinated by the precedentshattering discoveries of linguistic geographers.8 What concerns us most is that when, as a mature and experienced man, he was writing his second introduction to linguistics {Language, 1933), he was unable to reconcile the two points of view. The truth of the matter, of course, is that Language is far less monolithic than has been proclaimed, almost in unison, by a full chorus of the author's closest followers. A fact seldom remembered is that Bloomfield, in his years of apprenticeship, became an accomplished representative of Wortforschung at its most sophisticated and even artistic level.9 A decade or so later he read with heightened attention G. G. Kloeke's monographic study of the vowel phonemes of the local words for 'mouse' and 'house' in the Dutch-Flemish domain,10 accurately grasped its implica tions, and reviewed the book sympathetically and in considerable detail.11 Still later, he interested himself in Gillie'ron's researches and came up with a surprising ly serene and constructive appraisal of the underlying methods and of the results achieved.12 The chapter on dialect geography in Language embodies, then, the ripe fruits of years of concentrated thinking; it represents, indeed, an excellent synthesis, as forceful as the best provided by European scholarship and perhaps second in originality only to K. Jaberg's splendid College de France lectures (1933) assembled and published under the title Aspects géographiques du langage (1936). Against this background, it is all the more surprising that, precisely in elaborating on the adage or bon mot Every word has its own history, Bloomfield failed to draw a razor-sharp dividing line between two categories of apparent incongruences that dialect geography brought to light: (1) The failure of different isoglosses (phonological, morphological, syntactic) to coincide closely enough for sharply bounded dialect zones to emerge from the maps; (2) The failure of the same sound laws to be operative in the same territory, as between one word and another.

8) For Bloomfield's conceivably last tribute to the achievements of dialect geography see his review of F. Bodmer, The Loom of Language, in Amer. Speech, XIX (1944), 211-213. 9) "Physigunkus",Mod Ρhilol.,ΧV (1917-18), 577-602. 10) De Hollandsche expansie in de zestiende en zeventiende eeuw .. . (The Hague, 1927). 11) Lang., IV (1928). 284-288. 12) "On Recent Work in General Linguistics", Mod. Philol, XXV (1927-28), 211-230, esp. 227-229.


221

The first "irregularity" stabs at the heart of the ideal of neatly delineated dialects; the second shatters the concept of the sound law itself, at least as originally envisaged. In the concrete case of the wider territorial range — counter to sound law — of [mu:s] than of [hu:s], Bloomfield, in the wake of Kloeke, has very deftly sketched in the varying cultural background, arguing that a word such as 'house' is more prone than an "intimate" word like 'mouse' to succumb to the pressure of official and prestigious pronunciation, and convincingly demonstrates the superior force, in the given domain, of Hollandish over Hanseatic influence. In this context, be it noted in passing, he uses a "dynamic" and richly metaphoric terminology ("the innovator and aggressor" etc.) that might have delighted Gillie'ron; toward the end, he even toys, somewhat inexplicitly, with the Age-and-Area hypothesis which was to prove so controversial. But at the critical point he switches the discussion to the importance of relics turned up by dialectological field work (§ 19.5) without provid ing any guidance as to the precise techniques which may serve to identify classes of words "more confined to homely and familiar situations", and, above all, without admitting the extent to which discrepancies of the type [hu:s] / [mu:s] complicate phonological analysis and dent, or even breach, Neogrammatical doctrine. Later (§19.6) Bloomfield reverts to the dictum, slightly altering its actual wording ("Each word has its own history"), but the individual illustrations he borrows from Jaberg to substantiate his thesis reveal little more than the presence of some broad tendency or, if you wish, of some hazy scale of learnedness, one which forces the reader to allow for occasionally erratic distributions. And only after briefly introducing submerged areas and singling out toponyms as witnesses to phases otherwise overlaid (§19.7) does the author, in drawing his balance sheet (§19.8), admit the link of dialect geography to the wave theory (§ 18.12), "since the provincial types were examples of the differentiation of a speech-area without sudden cleavage", and expressly state his disappointment in dialect geography. Yet not even in this retro spect, for all its forcefulness, does one find a really tidy separation of such incongru ences as merely impede the delimitation of dialect zones and such as strike at the very root of the concept of a neatly bounded area for the agency of a single "sound law"; and there is not a hint here of the ineradicable contradiction between Bloomfield's own angry recoil from teleology13 and the avowedly teleological slant of all, or nearly all, significant lexical research in diachronic dialectology.

13) See rev. of Otto Jespersen, Language: its Nature, Development, and Origin, in Amer. Journal of Philology, XLIII (1922), 370 f; rev. of F. de Saussure, Cours de linguistique générale, 2d ed. (1922), in Mod. Language Journal, VIII (1924), 318; "On Recent Work in General Lin guistics", Mod. Philology, XXV (1927-28), 211-230, esp. 227-229; rev. of A. Stender-Petersen, Slavisch-germanische Lehnwortkunde, in Journal of English and Germanic Philology, XXVII (1928), 398; rev. of E. Hermann, Lautgesetz und Analogie, in Lang., VIII (1932), 224 and 231. In his book Language, Bloomfield - in the wake of Gilliéron - concerns himself with homo nymie clashes in, of all places, the chapter on the "Fluctuation in the Frequency of Forms" (§22.4).

222


III "Each word has its own history" may rank as a saw, or an aphorism, or an apophthegm, but, for all its pithiness, it certainly is no scientific formula. To be useful in analysis, it requires careful interpretation. "Word" in itself is a unit notori ously hard to define, and "history" means different things to different schools of thought; but, at this juncture, these two moot points need not disturb us. The root of the complication lies not in the semantic vagueness or ambiguity of any member of the proposition involved, but in a qualifier not formally expressed and which, to many users, would alone make the statement meaningful: " . . . its own independent history". The wisdom of operating with the dictum at issue hinges on one's interpretation of the implied adjective "independent". If a word, be it only in metaphoric use, has a life-span, a biography (we are reminded here of A. Darmesteter's once tone-setting booklet, La vie des mots), one is tempted to liken it to the life-span of a human being. Just how generous a share of true autonomy an individual may enjoy in real life depends, in part, on the observer's philosophy (a radical shade of determinism would obviously preclude any allowance for free will); in part, if the analyst's premises are not uncompromisingly deter ministic, on the given individual's ability to transcend the limitations imposed on him by his background and surroundings. A language itself is not, of course, en dowed with life, but its speakers are and the "independence" a word may enjoy hinges, in large part, on those speakers' ability to evade or escape "blind laws" of sound development — an ability varyingly assessed and in some extreme cases even expressly denied by experts. One alternative to this situation would be to exclude the users' voluntaristic involvement and to argue that the record of the word in question is unique by reason of a bizarre, unparalleled concurrence of circumstances. A word like Lat. aqua 'water' > Sp. agua boasts hardly any independent history of its own, because the few changes it underwent ([-kw- > -gw- > -γw-]; short stressed a transformed into quantitatively indeterminate a) pertain to the category of those predictably shared by many lexical items. Where phenomena such as folk etymology, hypercharacterization, false regression, blocking of change through learned in fluence, semantic shift, and avoidance of polysemy - to adduce but a few factors — come into play, predictability of results decreases sharply, the likelihood of the speakers' actual participation in the transmutational process increases, and the uniqueness of the resultant trajectory becomes unmistakable. There may be some wisdom in replacing the old dictum: "Each word has its own history" by this new one: "Many (or some, or just a few) words seem to have truly unique histories". The uniqueness, I repeat, may be traceable to accident or, more interestingly, to some imaginative speakers' partial involvement in the reshaping. Folk etymology, in particular, is the sworn enemy of regularity; it affects chiefly isolated words, lacking the support of powerful lexical families and fitting into no readily identifiable series. As happens in so many other provinces of linguistic investigation, the most profitable approach is to imagine some kind of scale of increasing complexity of lexical evolutions. At one extreme one might place developments as uneventful as the aforementioned shift Lat. aqua > Sp. agua (or > It. acqua, Rum. apă, but


223

certainly not Fr. eau, which fails to flow smoothly from eve, the commonest among many Old French forms); at the other, some of those instances of lexical "patholo gy", "therapeutics", and downright "thaumaturgy" which for many years enthralled Gilliéron and which to this day fascinate a few survivors — notably John Orr — from the original group of the Swiss pioneer's followers.14 A famous example, which Meyer-Lübke, in his REW3, tagged "doubtful" on account of several un documented intermediate stages, is the alleged metamorphosis of the Lat. ornithonym merulu aureolu, lit. 'golden blackbird', into OFr. *merle oriol; hence, through false segmentation and folk etymology, *mère loriol՛, still later, by interaction of grammar (gender) and lexical material, *père loriol՛, and, as a result of final elabo ration, compère loriot15 Here indeed, assuming that Gilliéron's reconstruction deserves credence, one witnesses a chain of events impossible of immediate predic tion, at this stage of our knowledge, and justifying the dictum: "Each word has its own history", taken not in its thinnest distillation, but with the full gamut of its implications of individual, pattern-bre aking growth. Most word histories fall between these two sharply silhouetted extremes, each lexical vignette illustrating, in varying proportions, the agency of sound laws (and of easily foreseeable morphological levelings provoked by them) and the interplay of the more loosely organized — if never entirely free-wheeling — forces in language. Human nature being what it is, some scholars will favor, for their inquiries, one or the other extreme group, or perhaps, better balanced intermediate cases. There might have emerged, in the case under study, a more intellectually slanted configuration of the three phases (a) original stabilization, (b) upheaval, (c) new balance, if we were living in a world more consistently rational, or if linguistic science, from the start, had been thoroughly mathematized. Neither condition has obtained in the situation at issue. For better or worse, in our discipline, small communities of scholars not infrequently accept a doctrine because they read with ease the language in which a potentially influential book is written; or because the given linguistic theory fits neatly into the current national pattern of cultural preferences; or because a certain saturation point has been reached on the local scene, creating a demand for something new that is also good or even for something new at any cost. Concomitant factors of this type rarely succeed in blocking com pletely the advance of linguistic science, but they may cause very painful delays and may thus distort the schema of brisk progress through appropriately speedy reactions to new ideas.

14) Among the recent publications see Essais d'étymologie et de philologiefrançaises(Paris, 1963) and Three Studies on Homonymics (Edinburgh, 1962), the latter extracted from the earlier miscellany, Words and Sounds in English and French (Oxford, 1953); and numerous book reviews scattered over the volumes of the Modern Language Review. (At proof, I must report John Orr's death in 1966.) 15) Généalogie des mots qui désignentI'abeüle(Paris, 1918), p. 229.

224


IV The intrinsically revolutionary character of Gilliéron's researches (as against the author's indisputable verve and originality) has been questioned not only in this country, by Bloomfield, but also in Europe, by scholars of conservative persua sion who, like Ernst Gamillscheg, inclined to stress the element of continuity far more emphatically than they did the ingredient of disruption.16 Actually the explo sive character of these inquiries is undeniable; yet it must be candidly admitted — and here Thomas S. Kuhn's stimulating recent book The Structure of Scientific Revolutions may indeed serve as an opening wedge — that, in itself, the theoretical gist of Gilliéron's critical remarks on the assumed regularity of sound change is too slight to rank him as a revolutionist of first magnitude. But if one takes into account such concomitants and accessories as the startling new techniques of field interviews and cartography, the unprecedented density of the network, the bold, flamboyant styling of model monographs, the provocatively aggressive manner of teaching and polemicizing, the courage — bordering on a cavaner attitude — with which he brushes aside, at least on an experimental basis, the most cherished collateral sources of information, a certain topical gaudiness of colors, and the brus que reversal of an outworn hierarchy through heightened attention to phenomena of traditionally minimized importance (say, folk etymology), then, and only then, the instantaneous crushing impact of Gilliéron's total œuvre becomes at once under standable . V It may be helpful to point out two metaphors frequently applied to the theory and the history of scientific inquiry. The first, familiar to every layman, at least in the English-speaking world, is that of "frontiers of knowledge", imaginary boundaries which the scientist, under any circumstances, and, ordinarily on a far more modest scale, the humanist also are expected to be steadily pushing forward. The image is here that of two adjoining domains, a "terra cognita" and a "terra incognita", as it were, of which the former, in any culture bent on uninterrupted acquisition of factual knowledge, must continuously expand and the latter corres pondingly shrink. By dint of repetition, the metaphor has become a diche; those who use it rarely visualize the specific mode in which the border-line or frontier separating the realms of knowledge and of ignorance can be most effectively ad vanced, either in strategic or in tactical terms. As a rule, no reference is made to any possible transitional zone between the presence and the absence of solid knowledge — a thin gray belt interposed and conceivably mediating between the vast, broaden ing white and the slowly receding black area (if we are to humor the advocates of the metaphor). Nor does the layman clearly visualize the ceaseless interplay of 16) Die Sprachgeographie und ihre Ergebnisse für die allgemeine Sprachwissenschaft (Biele feld and Leipzig, 1928), p. 11: "Wieder zeigt es sich, dass die Sprachgeographie nicht eine revolu tionäre Umbildung der in Jahrzehnten erarbeiteten wissenschaftlichen Ergebnisse bedeutet"; p. 72: "Die Sprachgeographie bedeutet also keine Revolution auf dem Gebiet der Sprach forschung, sie bedeutet nur eine Etappe im Fortschritt" (emphasis mine). The author identified himself with Menéndez Pidal's views (1926) of sound change in actual operation, tracing these views, in the last analysis, to Schuchardt (p. 76, n. 22).


225

advance and retreat, still less appreciate the need for periods of relative rest condu cive to gradual sifting and selective assimilation of slivers of knowledge after the exciting moments of swift and, occasionally, reckless irruption. Less frequently mentioned among non-specialists is the rival metaphor of the trajectory of knowledge, though one hears references to trail-blazing research or to pioneers spear-heading a certain movement. Yet the configuration of the ideal line, or lines, of cognitive progress must be continually haunting the minds of ambitious and responsible scholars. The number of geometric patterns readily available is in finite, but it may suffice to single out, at this preliminary phase, just the two extreme possibilities - one ultra-rigid, the other extra-flexible. On the one hand, we find the type of scholar for whom progress ֊ especially as regards refinement of methodology ֊ is symbolized by a straight line pointing, by definition, in a single desirable direction. Any deviation from this line will strike a scientist so predisposed as a regrettable departure from the path leading to truth absolute, as some kind of lapse, even apostasy, attributable to intellectual laxity or, worse, to moral frailty. Over against this orthodox purist stands the expert endowed with a more elastic mind. Without necessarily affirming the relativity or plurality of scientific truth, he will at least admit several itineraries — each involving a long, perhaps indeed bizarre, succession of Unes, both straight and curved — by which a significant advance can be achieved. In a way, this kind of researcher resembles a pool-player who, through a skillful and, above all, imaginative succession of shots, executed in a wide variety of directions and with numerous changes in his position, eventually succeeds in landing all the balls in the pockets. Flexibility, as this simile shows, must not be mistaken for hazy thinking or aimlessness; it requires, in fact, nerves of steel, but presupposes no fanaticism. If we apply to a domain as notoriously controversial as has of late been linguistics this tolerant view of progress shaped like a zigzagging or meandering Hne, our reac tions to research conducted at variance with our own favorite emphases or even assumptions will be conciliatory rather than irrevocably hostile. Supposing that we really prefer phonology as the discipline exemplifying, at its purest, neat lin guistic patterning and that etymology, with its wide margin of conjecture and its characteristic interplay of broadly historical and narrowly linguistic analysis, happens to be, to us, distinctly unattractive: We may nevertheless live unscathed through a period of sustained enthusiasm for etymological puzzles, knowing that such temporary deflection from the straight path can usefully replenish and, at the same time, filter a reservoir of lexical equations on which a later generation of austere phonologists can some day freely draw. By the same token, a fervent admirer of unadulterated Wortforschung may come to terms with a generation of fellow scholars oriented almost exclusively toward structuralism grammatically slanted.

226


VI One afterthought, of special relevance to the linguist: Not only does the actual progress of research fail to follow a straight line, but the development of language itself, i.e., the object of many linguists' prime curiosity, reveals, on microscopic inspection, a number of astonishingly sharp curves and breaks,17 an angularity which, as a rule, only in long-distance perspective yields to the soothing image of straight, beautifully drawn lines. The geometric design of continuity and revolu tion which we are detecting in an, all told, minor province of organized knowledge may thus, in a measure, be characteristic of cultural growth as a whole.

17) See my paper "Initial Points Versus Initial Segments of Linguistic Trajectories", read before the Ninth International Congress of Linguists (Cambridge, Mass., 1962) and printed in the Proceedings of the Ninth International Congress f Linguists (The Hague, 1964), pp. 402405.

I. MULTIPLE CAUSATION

MULTI-CONDITIONED SOUND CHANGE AND THE IMPACT OF MORPHOLOGY ON PHONOLOGY

Two ill-defined and controversial sound developments of late Old Spanish, namely the asymmetric diphthongizations ié > i and uè > e, become better understandable if one selects as starting points a set of morphological rather than phonological condi tions. Both verbal inflection and suffixal derivation can be invoked, including the rivalry of certain characteristic groups of preterits (-iemos, -ieste{s) beside -imos, -iste(s) etc.) and the competition of pairs or clusters of functionally more or less related suffixes : -ero beside -uero, -eño alongside -ueño—in addition to the pressure exerted, in the ranks of hypocoristics, by the close-knit series -/co, ֊ito and ~in(o) on isolated -iello. 0. The systematic exploration of the border zone between phonology and mor phology, viewed in diachronic perspective, does not date from yesterday. In the period between the two world wars it was, above all, the Swedish scholar Ernst G. Wahlgren (1879-1938)—not very much appreciated in his lifetime, and at present remembered only by a handful of specialists—who engaged in some hard thinking about certain shifts difficult to justify in the framework of traditional historical linguistics, which views ' regular ' phonological changes as primary phenomena and morphological changes, typically, as mere applications, adjustments, or reverbera tions, placing also a premium on etymologies which merely document or illustrate expected sound correspondences.1 To the extent that Wahlgren reserved a good deal of attention for individual word biographies, heeding associative interferences and 1 An apposite account of Wahlgren's intellectual growth remains to be written ; and of the few available short necrologies, even the relatively most satisfactory—Hasselrot 1938 and Melander 1937/38—are at best anecdotal. His activity as tireless researcher stretched, approxi mately, over a quarter century, from ca. 1910 until his virtual incapacitation in the mid-thirties. Outwardly a quiet man, known to his contemporaries for his meticulosity and modesty to the point of self-effacement, he was inwardly a rebel and a very forceful person, as his most pro vocative writings clearly show. Like most Romance linguists, he was more successful in de veloping a seasoned practitioner's knack and flair than in theorizing; in fact, the extra-heavy documentation that he amassed gave the (false) impression that his monographs were mere 'travaux de patience'. Less than successful in his career—he had to divide his time between marginal teaching at university level in Uppsala and a full-time job in a secondary school—he nevertheless forced himself to present his most ambitious studies over-documented, polished to a fine sheen, and extended to the entire gamut of Romance languages and beyond (he skill fully handled Germanic linguistics). Even his death had an ironic twist. He succumbed to a massive heart attack, no doubt from accumulated over-exertion, at the very moment when a major Scandinavian university was, at long last, almost ready to offer him a chair. Wahlgren's first major commitment to historical linguistics (his 1914 Uppsala dissertation, expanded into a weighty monograph six years later) involved a meticulous pan-Romanic probing of the agency of analogy, as exemplified by the reciprocal influences undergone by the 'weak' preterit and the past participle. The next major advance (1925), apropos so contro versial an issue as the so-called parasitic i of Old French, boiled down to an experiment in finely-nuanced distinction of the scribal signaling of genuine sound change, i.e. of the recording of actual sounds newly-introduced, as opposed to other uses of 'mute " as a purely graphemic device.

230

MULTIPLE CAUSATION

the like, he acted merely as a member of the combative generation of Romance scholars strongly influenced by the teachings of Schuchardt and Gilliéron.2 Con versely, in attempting to redesign the boundary between phonology and morphology and in constantly toying with the possibility of reversing their time-honored hierarchy, he struck out on his own and achieved genuine originality.3 Wahlgren began to tackle situations in which an inflectional or derivational change (triggered, e.g., by analogy brought to bear on a vulnerable part of a given grammatical structure) produced conditions either favoring or actually precipitating a sound change. Alternatively, he realized that a strictly morphological explanation could sometimes be persuasively substituted for a lame phonological hypothesis.4 Thirty years after Wahlgren's death, this kind of thinking began to be revived, 2 Thus, given the peripheral existence of a Romance sound change RS > s (MORSU 'bite' > OSp. muesso; VERSUS 'toward, into' > OProv. ves ~ vas), and consequently the ever-present possibility of false regression, Wahlgren 1928 wondered whether the toponym Marseille, from Gk. MASS-ALIA, -ILIA (an etymology at no time impugned), should be interpreted as a case of such regression, or alternatively, whether associative interference should be invoked. Sur prisingly, for the uninitiated, Wahlgren opted for the latter solution (influence of MARCELLUS and/or MARS). Having become sensitized by his 1925 paper to several conflicting functions of the letter i in records of medieval French, he attacked, five years later, two minor etymological problems in which the erratic occurrence ofiplayed a pivotal role (OFr. isnel, ignei ' quick, light, agile' and Fr. utague ~ itague ~ etague, a nautical term); and then devoted his last paper, published one year before his death, to the controversial i of the OFr. demonstratives icist, icil, thus isolating for evolutionary analysis a striking sound feature represented in a neatly con toured morphological class. 3 The most ambitious thrust, made by Wahlgren at the peak of his productivity and sophisti cation, is found in his many-faceted analysis (1930a) of the controversial d > r shift, which a respected fellow Swede, K. Michaëlsson, had previously (1924) posited and analysed—in refer ence to French alone—as a preeminently phonetic process. Wahlgren came up with the surprising discovery, supported by evidence drawn from all Romance languages, that (with the negligible exception of certain Italian dialects) the transmutation involved either lexico-etymological conflicts, or the interference of characteristic derivational suffixes—above all, the vernacular reflexes of -ARIUS. Critical reactions were mixed. To cite the two opposite ends of the scale, the author's next-door neighbor E. Staaff, himself an acknowledged authority in the domain of -Amus studies, pointed out (1930) certain flaws and exaggerations in Wahlgren's analysis, but fully recognized the author's boldness, originality, erudition, and seriousness of purpose; but T. Navarro, an inveterate phonetician, rejected (1931), so far as Spanish was concerned, Wahlgren's curtailment of the domain of phonology. A companion study of Wahlgren's, dealing brilliantly with the Romance names of cardinal points (1931b), offers a noteworthy elaboration. His purpose here is to drive home the illusoriness of the change sud > sur conceived as embodying a sound shift. Fr. surouest 'Southwest' is persuasively credited to what is at present called lexical polarization, echoing as it does nor(d)~ ouest, while Sp. sur, Ptg. sul are shown—on the basis of old logbooks etc.—as approximations to Eng. south (cf. L. Spitzer's quick and unqualified endorsement of this hypothesis, 1931). 4 After a considerable break, Wahlgren was slowly, methodically preparing for a return to his favorite terrain of research in conjugation, at the points of impact of phonology on mor phology—and, particularly, vice versa. Treading familiar ground (from his work on the afore mentioned dissertation) in the domain of Romance past tenses, he chose the format of a lengthy article, in 1933-34, to attack a crux of French historical grammar, namely the shift /we/ > /ε/ in the endings of the imperfect indicative and the conditional tenses (the paradigm is traceable to -E etc.).From the same period, apparently, is the opening fascicle (1931a)—the only one which was actually to appear—of the venture which Wahlgren presumably expected to become the magnum opus of his ripe age : a string of observations on the morphology and phonology (note this conspicuous sequence, doubtless rich in hierarchical implications) of the 'weak

MULTI-CONDITIONED SOUND CHANGE

231

and not only in Romance quarters—in part by the writer of these lines;5 in part, slightly later, by younger workers who either specifically referred to his gropings,6 preterit' in French with emphasis on dialectal evidence. The sequel did not come to fruition, probably as a result of Wahlgren's bouts with cardiac sickness ; but an impasse, or still other ingredients less tangible, may also be involved. A characteristic quality of Wahlgren's modus operandi, and one which eluded the attention of his critics and necrologists, was the close cohesion of his seemingly disparate studies. Even the deceptively isolated 1935 lecture on learned words in Italian (which appeared the following year, shortly before the author's death) hardly marks a new departure; as a matter of fact, it ties in rather closely with Un problème (1930a). 5 Most of the relevant research was conducted between themid-'sixtiesand the early 'seventies. On two occasions I drew attention to Wahlgren's ideas about d > r, and to an alternative view (1967a: 1239-41, 1967b:238-9). In the paper which represents the outgrowth of the April 1966 Texas Symposium in Historical Linguistics, the central problem subjected to a test was Lat. LG, NG, RG > OSp. (/)z, nz (beside n), and rz (1968:33-55), with the two monophthongizations, ie >  and ue > e, examined by way of a supplementary experiment or 'tentative parallel' (5562). The printed version differs but slightly from the preprint distributed at the Symposium; the Italian translation of Directions, prepared by R. Stefanini and now in press (Bologna: Mulino), contains a few addenda. The French-language article (1970a), written in 1966, restores the balance between the two problems, placing for the first time the monophthongization in the center of attention; the March 1970 post-script (734-5) refers the reader to the earliest critical reactions (R. Posner, K. Togeby, K. D. Uitti) and to a brief supporting note of my own. The paper dealing with /sk/ > /0k/ represents a third, entirely independent attack on the broad, underlying problem (1969a); all three studies were summarized, compared with each other, and modified, in retrospect, as regards a few trivial details, in Malkiel 1969b (p. 327). Work on /sk/ > /0k/ may have counseled careful reexamination of an eighty-year-old controversy re volving around the OSp. sibilants ç and z\ the solution offered (1971) introduced, as a new factor, the voicing, on a sweeping scale, of the consonantal pillar of all proto-Spanish deriva tional suffixes (in particular of -zar, -azo, -izo, -eza, and -azórì) except those of a hypocoristic or humorous or coarse nature—an assumption which virtually eliminated the troublesome residues of exceptions which had haunted earlier analyses. The programmatically-tilted French-language note (1973a) struck out in a new direction chiefly in attributing the rise of provincial triads in Portuguese {ęste-ęsta-isto, todo-toda-tudö) to the speakers' striving toward maximum dif ferentiation, i.e. hypercharacterization, of the three genders—a simple hypothesis which settles various difficulties with which it was practically impossible to cope in conventional phonological terms. A microscopic study, that same year (1973b), sought to explain the erratic leap from Lat. LACERTU 'lizard' to Sp. lagart-o, -ija (instead of *lazierto) through appeal to the wavering ERT ~ ART, characteristic of certain verbal paradigms, especially of past participles. An excursus attached to a more recent paper devoted to conjugational interfixation (1974:349-51) concerned itself with some further critical reactions to the Texas paper (A. M. Barrenechea, J. R. Craddock, M. B. Fontanella de Weinberg, G. Lepschy, . . Otero, O. Szemerényi) and filled a lacuna in the survey of older hypotheses (P. Fouché). An additional reason for the controversial develop ment of RG > rz etc., but hardly a substitute for the earlier analysis, was found when the general avoidance of the phoneme /ž/ by the Old Spanish speech community—as independently observable in the evolution of the suffixal triad -AGINE, -ÏGINE, -ŪGINE and in other contexts— was laid bare (Malkiel 1976b). 6 Rochet 1974, in a spin-off from his unpublished doctoral dissertation (Alberta, 1970), investigates specifically the merger of /ε/ and /ă/ in the majority of Old French dialects. As a matter of broad policy, he would like to have this morphological avenue of approach tested 'even when no puzzling or residual changes have to be accounted for' (44). His copious bibliog raphy contains references to pertinent writings, phrased in the sociolinguistic key, by Labov and by Ma & Heramsimchuk (43). Montgomery (1976:295), in examining the relation of stem-vowel to conjugation class in the transition of Latin to Old Spanish, goes so far as to declare: 'The sequence -i-í(C) of adquirir etc. had been absent from unadulterated Castilian until learned infinitives began to be coined.

232

MULTIPLE CAUSATION

or elected to advance independently.7 As a result, the problem of the fluctuating frontier between phonology and morphology, in diachrony, is gradually becoming, as it should at all times have been, one of the front lines of linguistic research. The aim of the present paper is to examine in adequate detail (though not exhaustively) another group of relevant laboratory cases, and in so doing to improve on one particular score a previous analysis offered tentatively in the mid-sixties. 1. Hispanists have traditionally focused their attention on diphthongization— principally on the change of stressed open vowels into rising diphthongs (as in BENE 'well' > bien; BONU 'good' > bueno), with less energy on the transmutation of falling into rising diphthongs (as in MEU 'my' > OSp. mió; CORIU 'leather' > /kojru/ > cuero).8 Little more than passing mention was, for decades, made of the (occasional) eventual monophthongization of the two most important among these newly-formed diphthongs, ie /je/ and ue /we/, observable in OSp. priessa 'haste, pressure' (a product of PRESSA, from PREMO -ERE) > mod. prisa, or in OSp. fruente 'forehead' (an outgrowth of FRONTE, from FRÖNS) > frente (f.) True, Menéndez Pidal (1926) announced a minor discovery, based on scrupulous sifting of dated and localized charters, in regard to the ie > i development. One must reckon, in his view, with a two-step development: first, in some instances, as early as the 11th century, dimin. -iello, -iella yielded -Ulo, -illa (sporadically); then, toward the end of the Middle Ages, a number of other words followed suit (e.g. aviespa 'wasp' > avispa, from VESPA), while the use of -ilio, -illa was generalized. The explanation which Menéndez Pidal offered lacked specificity (he saw in these events merely a confirmation of his broad view of the slowness and graduality of sound change), and the analysis at the outset triggered hardly any memorable discussion on the explicative level.9 It is still rare except in verb forms. In this case a morphological change acted as a moving force that affected the phonetic patterning of the entire language'—an unexceptionable statement, in my estimate. 7 O'Bryan 1974 uses Old Indic evidence to study interactions of morphological and phonolog ical processes. Theoretically, she takes her cue from writings of very recent vintage (1972-73) by P. Kiparsky, H. H. Hock, K. Dudas in collaboration with herself, and R. Wilbur. The key sentence in her point of departure is clearly stated : ' Morphological generalizations often go unnoticed because, as a result of the current phonological methodology, they appear upon first consideration to be correctly characterizable as phonological exceptions. Closer examination has revealed, however, that in many instances, simply marking them as exceptions to phonolog ical rules DESCRIBES a situation which could be EXPLAINED if looked at from a perspective in which morphological processes are considered' (49). Rochet and O'Bryan, though generally well-informed, are both unaware of Wahlgren's pioneering role along this road. On the side of Rumanian, Sala 1970 arbitrated the dispute concerning the relations of such aberrant forms as Rum. stea, dial, steaua 'star', to its prototype STÉLLA in a sense favorable to the general thesis here championed. Sala argued that steaua everywhere preceded stea and that the regional shift from the former to the latter form was triggered by a number of extra-frequent pronominal and adjectival models of the type ea 'she' ~ eale 'they' (f. pl.) (I owe knowledge of this highly relevant analysis to Merritt Ruhlen's forthcoming review of Sala's book in RPh.) 8 I have examined, cursorily, the change /oj/ > /we/ (1969c) and, in somewhat greater depth, the long-overlooked change /ew/ > /jo/, distressingly elusive on account of various overlays (1976a). 9 The deftest presentation of the problem is found not in the oft-revised and patched-up


233

At the mid-century, modernistically oriented linguists became aware that these changes had an importance out of all proportion to their, all told, modest lexical range. This was in part on account of certain methodological implications, in part as a consequence of their distinctly late date—a circumstance conducive to chrono logical accuracy; but the initial attempts at structural re-interpretation (e.g. by Alarcos Llorach), however well-meant, turned out less than successful.10 At the 1966 Texas Symposium, and in the context of subsequent publications, I suggested that Menéndez Pidal's Step 1 (-iello > -ilio etc.) be attributed to the pressure brought to bear on that diminutive suffix by an alliance of its rivals {-ico, -ino ~ -in, -ito, with only -uelo > -IOLU standing apart); and this portion of my argument was accepted without qualification even by the severest critic of the paper. 11 How ever, my more tentative remarks on (a) the slightly erratic further spread 'of ie > i (as in avispa, but to the exclusion of fiesta, siesta) and (b) the undeniable asymmetry of ié > i and uè > e (the upshot of the contradiction being that the historically later development—supposedly a mere echoing—is phonologically easier to understand than the earlier) failed to arouse a comparable degree of endorsement. An entirely independent study, which entailed the examination of the Old Spanish preterit (Malkiel 1976a), has meanwhile convinced me that, throughout the late sixties, I neglected to take into account one crucially important Manual de gramática, but in the two basic versions of Orígenes del español, see Menéndez Pidal 1926 (and 1950), §27. For the sake of completeness, let me summarize here the processes of monophthongization observable in pretonic or countertonic syllables (credit for the spadework must go to Zeitlin 1939:84-90). Here we find indeed, in ancient unretouched texts and modern dialectal records, plentiful examples of ie reduced to / (or e) and of ue reduced to  (apparently never to e): ¿quin sabe ? ' who knows ? ' (reminiscent of quiz-a, -as, orig. quiçab ' perhaps ') ; sintopié ' centipede ' (= stand, cientopies), barbi(e)speso (Cervantes) 'thick-bearded'; bie(n)fetria > bifetria beside benfetría > behetría 'confusion, pandemonium' (orig. BENE + FACTORÍA); gritada (Cane, de Baena) 'crackle, cracked surface' ( = stand, grietada); nus tramo (Torres Naharro) 'our master' (= stand, nuestro amo). Zeitlin so explains variants of weakly stressed words, e.g. cuerno ~ cum{o) 'how', luego ~ lugo 'at once', pues ~ pus 'then, since' and even—unconvincingly— toys with the idea of extending this interpretation to si(e)glo. Dial, and obsol. (El Corbacho) tútano beside equally old (E. de Villena) stand, tuétano 'marrow' might have been influenced by the verb destutanarse, since Jud.-Sp. totano points toward an ancestral o. The base may be onomatopoeic (Corominas 1957:624-5). 10 Having apparently read Menéndez Pidal's definitive analysis with insufficient care, one structuralist (Martinet 1952:184) jumped to the conclusion that everything would be simpler if i were traced to ie. The facts do not bear out this hypothesis. The idea that secondary i is best explained by starting from ie has been advocated with astonishing stubbornness and lack of realism by a Spanish follower of Martinet (Alarcos Llorach 1951:15-16, 1954:185-9 [§144]— and, with somewhat diminished confidence, 1961, §144). The idea of operating with a ie diphthong, championed by G. Baist and others, had been refuted almost a century ago by Meyer-Lübke (1885:254), except in territories where overwhelming evidence of its existence is available. Unfortunately, in the two most searching and balanced reviews of Alarcos Llorach's Fonología that have come to my attention, the respective critics (Jacobsen 1957 and Catalán 1964) were so preoccupied with the internal workings of phonology as to have overlooked morphological explicative alternatives. 11 See Craddock (1970:691). Among very recent assessments of varying breadth, not yet ad duced, let me cite Lehmann's (1973:229), essentially noncommittal, and Catalan's (1974:297-9), affirmative to the point of enthusiasm.

234

MULTIPLE CAUSATION

factor which, as a matter of fact, would have strengthened my general thesis, and in addition might have clarified several particulars. This factor—a conflict between two conjugational patterns in Old Spanish, e.g. (pret.) comiemos ~ comimos 'we ate'—has apparently eluded the attention of the other participants in the debate as well. The cases ordinarily cited to illustrate the propagation of ie > i are : aprisco (v.) ' I gather the sheep in the fold', (n.) 'sheep-fold', from *APPRESSICÄRE, a plausibly reconstructed intensive-iterative elaboration on PREMERE/PRESSUS ' t o press, squeeze, lie down o n ' (familiar to us frompriessa > prisa), avispa ' w a s p ' , mentioned above, which may owe its aberrant a֊ to association with abeja 'bee' < APICULA 'little or cute bee', from APIS. mirl-o, rarely -a 'blackbird' < MERUL-A, rarely -u (cf. Fr. merle, an ornithonym associated with a trail-blazing folk-etymological analysis by Gilliéron). níspero 'medlar-tree' and níspola 'medlar-fruit' < MESPIL-U beside -A (with dissimilation of consecutive syllable-initial bilabials). pingo ' I drip' < *PENDICO, a graphic elaboration on PENDERE ' t o hang'. prisa 'haste' (cited above) < OSp. pries(s)a. prisco 'kind of clingstone peach' < PERSICU, lit. 'Persian apple', with a characteristic metathesis (cf. OSp. Brima 'Bible' < Gr.-Lat. BIBLIA, as against crebar > quebrar ' t o burst, break', from CREPARE). remilgo (refl. v.) ' I am affectedly nice, squeamish', (n.) 'affected gravity, prudery', related to MEL ' h o n e y ' and involving the same verbal suffix -ICÄRE, extremely productive in Latin folk speech, as do aprisco and pingo.12 ristra 'string, row, file' < RESTE (f.) ' r o p e ' , with by no means unusual hypercharacterization or 'marking' of the less favored gender and—again, decidedly widespread—intrusive r after st (cf. estrella 'star' < STÉLLA). siglo 'world', later 'century' < SAECULU, absorbed presumably through the instrumentality of OFr. siegle, the predecessor of mod. siècle (cf. OPtg. segre, reminiscent of OGal. miragre, as against OSp. miraglo > mod. milagro 'miracle', another borrowing). visperaŲ) ' e v e ' < VESPERA (H)ÖRA 'evening'.

Where vernacular Portuguese counterparts are at all available, they display, as expected, primary /ε/ or secondary /e/ as counterparts of OSp. ie: merlo ~ melro, néspera, pêssego (with RS > ss, an ever-present possibility), pressa, restia, vespa, véspera. As regards the Old Spanish material, there is a point in distinguishing between nouns and verbs. In the nominal domain, which exhibits only a few traces of morphophonemic involvement, and on the side of derivation, the medieval and 12

The etymology of remilgar and remilgo strikes me as secure, given the extraordinary im pact of the semantic zone of 'sweet [like honey]', 'prim and finicky, affected, smirking', 'hyp ocritical' etc. on the entire lexicon—an impact which has its roots in the circum-Mediterranean importance of honey in the preparation of food. (For the latest crumbs of linguistic and broadly cultural information, see Seidenspinner de Núñez 1976.) As for the reconstructed -ICÄRE verbs, the varying date of the syncope of I, depending on the affinity of /k/ to the consonant preceding the doomed vowel, explains why pingar and remugar show a voiced velar, as against the voiceless velar of apriscar, reminiscent of rascar ' t o scratch' < *RÄSICÄRE (from RÄDERE ' t o scrape, scratch, shave', RÄSUS). I deliberately disregard here certain bizarre graphies of an archaic Toledan text, the Auto de los Reyes Magos (mid-12th-century?): bine 'well', quiro ' I want', quin ' w h o ' , timpo ' t i m e ' etc.; also seglo 'world', celo beside cilo 'sky, heaven'—an idiosyncrasy for which several ex planations, either exegetic-philological or sociolinguistic, have over the years been proposed (for the latest discussion, see Sola-Sole 1975:20-27).


235

Golden Age forms still display the diphthong : priessa, riestra, sieglo etc. Conversely, the three verbs so far securely identified no longer show any vestige of the expected morphophonemic alternation in their conjugational paradigm, and the leveling is further reflected in the corresponding postverbals. Thus, instead of inferrable *apriesco (> aprisco) ~ *aprescamos (a scheme suggested by aprieto 'I press [against the chest]' ~ apretamos 'we press', from Late Lat. APPECTORÄRE), one finds only i throughout the paradigm, whether the syllable is stressed or unstressed: aprisco ~ apriscamos. One is thus tempted to go beyond Menéndez Pidal's 1926 model and posit a triphasic development: (a) -iello > -ilio, a kind of prelude, confined to diminutives. (b) Limited diffusion of the i as a substitute for ie among nouns, in stressed syllables, under a set of phonetic conditions probably closest to contiguous /λ/— limited because predictability happens to be very weak, with tiesto 'broken piece of earthenware, flowerpot' (beside TESTA), fiesta 'holiday' < FESTA, and siesta 'afternoon rest' < SEXTA (HÖRA) seemingly contradicting riest(r)a. (c) Spread of the / to the pretonic syllable (from rhizotonic forms and postverbal nouns) in the paradigms of, minimally, three verbs—best understood as a recoil from an unwelcome e : i alternation, common to be sure in the -ir class (pedir ' to ask', pido 'I ask') but unparalleled in the -ar class. This categorization is not iron-clad, and one must reckon with transitions ; thus SELLA > siella 'saddle, stool' (later also 'seat, chair') contained a segment of the primitive (-iella) homophonous with the diminutive suffix, and could thus mediate between Group (a) and Group (b). More than any other facet of the elusive problem, it is the direction of the internal spread of ie > i, across the lexicon,13 that has baffled scholars. It seems that the concentration of certain consonant phonemes—either contiguous to the endangered ie or, even more astonishing, placed at a certain distance from it—stimulated monophthongization. After analysing /λ/ into its distinctive features, namely laterality and palatality, one is scarcely surprised to find that other phonemes containing one or two of these features were particularly conducive to the change ofie > i—initially, one is tempted to speculate, when following the diphthong, later in other positions as well. Since palatality is also present in /n/, /ž/, and /š/, one readily accepts OSp. Yéñego > Iñigo, Agierbe > Agirbe, and Guadiex > Guadix; and once the affinity of/λ/ to /1/ is granted on the basis of distinctive-feature analysis, i.e. of a universal, one is willing to accept the susceptibility of ie to the vicinity of other consonants with which /1/ is known to have formed a language-specific alliance in older Hispano-Romance, particularly /r/ and /R/. 14 As a rule, the co-occurrence with ie of /λ n ž š 1 r R/ is of itself not a sufficiently weighty circum13 'Internal spread', within a given lexical or grammatical system, is here contrasted with the 'external spread' of diffusionists and dialect geographers (cf. Wang 1969:9-25, Chen & Wang 1975:255-81). 14 While the specifics of the distribution of /r/ and /R/ may not have been the same in Old Spanish as in the modern regional and social dialects (the point has never been thoroughly investigated), the existence of two separate phonemes may safely be projected onto the level of medieval speech. (Let me also remind the non-specialist of the labio-dental character of v in Old Spanish ; v and b were not yet two rival spellings of the same sound.)

236

MULTIPLE CAUSATION

stance to have produced monophthongization; but the appearance of certain combinations of these phonemes—at rare intervals as initial clusters, somewhat more frequently as medial clusters, in most instances both preceding and following the diphthong amenable to change—has produced the effect at issue. Examples of such dyadic and triadic co-occurrences include the afore mentioned mi(e)rlo and Y(é)ñego, plus the toponym Xavierre > Exavirr, of Basque background. If developments have so far been falling into place, two details that have never before been satisfactorily explained must now claim our attention. First: no neighboring consonant has exerted on ie an influence remotely comparable to that of s. Now it is true that Castillan s displays a slightly more 'palatal' nuance than, say, French or Italian s; to the student of articulatory phonetics, it is apicoalveolar rather than predorsal (cf. Navarro Tomás 1944:106). But granted this well-established fact, and further granted that an occasional alliance of s [š] with χ [š] and g = j /ž/ in triggering the monophthongization of ie would not be astonishing, it remains a major surprise that s should have led all other Old Spanish consonants, during phase 2 of the development, in co-occurring with ie > i—this despite a complete lack of dynamic affinity between s and /, r, rr in all other pressures that were operative in the given language at the time implied. Here are the examples : ariesta > arista 'awn, beard (of grain)', from ARISTA, *-ESTA 'fish-bone'; aviespa > avispa 'wasp'; niéspera > nísp-ora, -ola 'medlar' ; priesco > prisco 'sort of peach'; priessa > prisa ' haste ' ; (em)priesta > prista ' he loans ' < PRAESTAT ; riestra > ristra 'string'; sieglo > siglo 'world'; siella > silla 'saddle'; viéspera > víspera 'eve'. To these must be added certain obscure toponyms in medieval records, e.g. Asieso ~ Asiso, Liestra ~ Listra, as well as the paradigms of a few important verbs, e.g. siego > sigo 'I follow', from SEQUOR, and siervo > sirvo 'I serve', from SERVIO, as against the consistent preservation of the corresponding noun siervo 'serf, servant', from SERVU—involving changes which, to be sure, can be explained away in entirely different terms, namely as a result of an interplay between diphthongization and metaphony,15 but which nonetheless, viewed in a germane perspective, are witnesses to the crucial transmutation sie > si in stressed syllables. Also to be noted are certain complicated episodes in the growth of the lexical family of quicio ' pivot hole' (for hinge pole), 'door jamb', assuming that the ç [Կ] was, for once, attracted by s.16 The cumulative impression of this evidence is that a separate reason must be 15

A paper of my own (1966) has had the good fortune of setting in motion the minds of several fellow scholars, including Harris 1969 and Craddock 1973, so that our knowledge of the impact of diphthongization on metaphony is once more in a state of flux. 16 It is actually resquicio 'crack, chink', fig. 'chance, opportunity, occasion', rather than quicio, an unusual back-formation, that best lends itself to genetic linkage to OSp. rescreçar— which apparently, in turn, presupposes *RE-(E)XCREPTIÄRE, in the last analysis an elaboration on CREPITARE/*CREPTÄRE (cf. Meyer-Lübke 1930-35, §2316). As one would expect of a family rich in morphological complications and phonosymbolic potentialities, the number of inter mediate stages is disquieting, and some of them are frankly puzzling; but the one decisive step established by Menéndez Pidal (1920:24)—OSp. rescrieço > mod. resquicio, a step com plicated by dissimilatory elimination of the second r and by the substitution of -io for -o (cf. agrio ' sour ' for older agro with some help from the verb agriar, or Eleuterio as the equivalent of ELEUTHERUS, with support from Desiderio, Eugenio, or escarnio 'taunt, scoff' from escarnir


237

found for such heavy representation of s as a phoneme ordinarily co-occurrent with ie > i.16a The less important side-issue previously neglected concerns the relative fre quency of the bilabials b m p and labio-dental ν in the particular environments favorable to the process of monophthongization. The examples, already mentioned in other contexts, are:avi(e)spa, mi(e)rlo, ni(é)spera, pri(e)sco, pri(es)sa,pri(e)sta, and vi{e)spera, plus Agi(e)rbe and Xavierre on the side of the onomasticon. The pre dominance of p is striking; the absence of/would be noteworthy were it not for the general weakness of that consonant in Hispano-Romance. It can perhaps be argued that p occurs not infrequently as part of a cluster of which another member (typically, an s or an r) is already familiar to us as a kind of catalyst in reducing ie to i, on the understanding that yet another consonant element endowed with similar force will be working in the same direction; thus sp + r cooperate in niéspera, viéspera;pr + s in priesco, priessa etc. (note that, without additional help, pr on its own falls short of transmuting ie into i, witness prieto 'darkish, black' from apretar 'to press, squeeze'—which, significantly, lacks any such by-form as *prito). The safest guess is that, from its frequent occurrence in such clusters, the;?—and, through it, eventually certain other bilabials and labio-dentals—derived a certain force, still ill-defined, to favor the monophthongization of ie. This analysis would leave such consonants as the dentals t d n and the velars /k/, g among those relatively least important in the context here studied, though their occasional occurrence in the words involved, presumably as neutral ingredients, is indisputable. In sieglo > sigh, I and s are likely to have favored the shift, without any counterpressure from g. In finiestra 'window' > mod. dial, hiniestra, the neutrality of the dentals n and է (witness siesta, tiesto—in niéspera an m was once involved) and the distance plus relative aloofness of ƒ (witness fiesta) apparently countervailed any positive strength vested in s + r (contrast hiniestra, with blocked monophthongization, and destra > ristra, with completed monophthongization). 2. It remains, then, to examine at somewhat closer range the one long-overlooked factor, already hinted at, which can be expected to account for the prominence of s in general, and of s followed by է in particular (by extension, also by/?), primarily where s (including st and sp) is preceded by the stressed vowel. This factor, strictly morphological, is the protracted rivalry between certain Old Spanish models of soetc.)—undoubtedly involves, at its very core, the monophthongization of ie to i, for once in the vicinity of a voiceless affricate, OSp. ç [bs]. (On several other ingredients of the problem, especially the role possibly played by desquiciar 'to unhinge, upset', see Corominas 1956:948a952a.) Apparently not immediately related to the problems of monophthongization here discussed is the co-existence, in the Galician-Asturo-Leonese domain, and originally elsewhere in the North, of the three pres. subj. forms of the verb se(e)r 'to be' : sega [seja], siega [sjéja], and sia. The first two are descendants of SEDEAT, from SEDÉRE 'to sit'; the third reflects VLat. *siat, a regional substitute for Class, SIT, flanking OLeon. dia 'he may give' and estia 'he may stand, be'. (See Lapesa 1948:77-8, with further literature, and Menéndez Pidal [1926] 1950:359; also, in the pan-Romanic perspective, H. Schmid 1949.) 16a At proof, let me mention one isolated instance of the parallel regional reduction of id to /, namely the top. Sitrama de Tera, orig. Siatrama; I discuss this point in my recent study of Diego, Diago, Diogo.

238

MULTIPLE CAUSATION

called weak (i.e. arrhizotonic) preterits of the -er and -ir conjugation classes. The number of variants securely identified is bewildering, and threatens to obfuscate the view; we shall therefore sift the available inventory, omitting forms apparently irrelevant—e.g. the dialectal rivalry between 3pl. -ieron and -ioron; or the interplay between apocope and restoration of the final vowel in the 2sg., which led to the co-existence of -ste ~ -sti ~ -st; or the vestigial survival in Old Leonese and Old Navarro-Aragonese of the separate 3sg. ending -eu in -er verbs, and many other frills. Reduced to a skeleton, the OSp. preterit of a verb like comer 'to eat' appears in two characteristic branches (here slightly idealized for the sake of clarity and contrast), one dominated by the monophthong i, the other by the diphthong ie: (a) -i, -iste, -ió, -imos, -istes, -iron. (b) -i -ieste, -i,ó -temos, -iestes, -ieron. In the dominant medieval varieties preserved in reliably edited texts, the forms of the lsg. and 3sg. straddle the two series, but the remaining forms are each restricted to just one series. Their actual distribution varied considerably from one verbal person to another, and the criteria of differentiation were not at all the same. Thus -iron was characteristic of the West (where it adjoined Gal.-Ptg. -irom and eked out a modest existence in steady competition with -toron), while -ieron prevailed in the Center. The vogue of -iemos, -testes was restricted, again in the Center, to the 13th and 14th centuries ; and it is clear that this fashion tied in with the strong representa tion of -ieron (which in the end outlived its erstwhile partners). There may very well have been a feeling in one tone-setting part of the speech community that ie some how characterized the plural as a whole—a conjecture whose likelihood is borne out by the fact that, even at the peak of the predominance of -iemos, -iestes, and -ieron, the 2sg. normally ended in -iste rather than -ieste. Since in the 'weak' preterit of the -ar class, at that same juncture, the prevalent scheme was -é, -est(e), -ó, ... (e.g. canté ' I sang', conteste, cantó, ...), it may be plausibly argued that a separate link connected the stressed monophthongs and diphthongs in the first and second persons of both numbers: amé 'I loved', -este, ... ; amamos, -astes, ... (from amar); bevi 'I drank', -iste, ...; beviemos, -iestes, ... (from bever).The distant origin of these series is lost in a complicated web of controversial issues, which need not be surveyed here—first, because the genetic problem is not directly pertinent to the tendential monophthongization of ie; and second, because a lengthy analysis of these century-old discussions, stretching from Schuchardt to Lausberg, has just been offered elsewhere.17 17 For full bibliographic details I can refer to the relevant sections (including Excursus A and Excursus B) of a monographic study completed only a few months ago (Malkiel 1976a). Briefly, the situation may be described thus: Menéndez Pidal, throughout his writings scattered over a half-century, and the narrowest group of his followers (e.g. Lapesa 1950), have aimed to explain the weak-preterit endings -iest(e), -iemos, -testes, -ieron without having recourse to the interference of the so-called DÊDÎ, STÊTÎ type (for the first time identified as a powerful agent by H. Schuchardt, as early as 1866 and 1868). The majority of other early-20th-century scholars involved (E. Staaff, J. Cornu, G. Baist, Α. Zauner, F. Hanssen, and several others) did make allowances for a measure of such interlocking—crediting, e.g., -ieron to dieron 'they gave' < DË(DÉ)RUNT; but they disagreed on numerous significant details. The discussion reached its peak in the decade 1904-14, then its intensity fell off sharply after World War I; but individual comparatists, including, of late, H. Lausberg, have kept it going.


239

The one facet of the problem which has immediate relevance here is the fairly rapid coalescence of the two once polarized branches—a process which, on the evidence of an abundant corpus, easily amenable to inspection, must have started shortly after 1400. The last remnants of bevieste, weak from the start, must then have succumbed before the onslaught; beviemos and beviestes, once strongly en trenched, also quickly yielded ground to bevimos and bevistes, while bevieron, conversely, dislodged beviron. Of these various conflicts, it is the sharp rivalry between -ieste and -iste in the singular and between -iestes and -istes in the plural— a struggle that in both instances led to the clear-cut victory of the monophthongal suffix—which could have influenced events on the phonological front as well. Speakers who, from one generation to another, heard everyone in their environment switch from -ieste(s) to -iste(s), and who eagerly joined the 'fad', could easily have been swayed to substitute, in the nominal domain, ristra for riestra—or at least to opt enthusiastically for ristra which may, until then, have led a marginal existence in the shadow of a more prestigious variant. The attraction of -ieste(s) > -iste(s) was apparently little more than a clinching factor, for it did not of itself suffice to monophthongize the ie oí fiesta, siesta, and tiesto, or to ram through prista 'he loans' at the expense ofpriesta;18 the fashion did not develop into a stampede. But if one adds to the strong weight of -i(e)ste and -i(e)stes also the slight weight of lpl. -i(e)mos, an ending which underwent the same significant reduction of ie to i (the only difference being that the s, for once, was not contiguous to the syllabic peak), one will grant that we have laid our finger on an independent factor explaining the marked affinity of s, as well as s + voiceless stop (i.e. st, with inherent potentialities of extension to sp), for the process under observation. A side-issue which can only be touched upon is: Why did the past subj. -iese etc. (originally -iesse) fail to develop so much as a moderately viable by-form in *-is(s)el The answer is clear: First, the shift ie > i depends on a plurality of favorable circumstances, not on a single factor, however advantageous ; second, -ies (s)e has at all times formed a close alliance with two other tenses, -iera and (lately almost extinct) -iere, which in turn have been geared to the 3pl. preterit in -ieron. The fact that -ieron won out over -iron suffices to show the strength of the diphthong in that particular position, whichever the ultimate reason for its victory (some analysts argue that there developed a strong need at least for near-symmetry, as between 3sg. -ió /jo/ and 3pl. -ieron /jeron/). Whatever argument we decide to appeal to, the speakers' choice of -ieron predetermined their selection of -iera etc.—a tense which has rapidly grown in strength to this day, and could thus block any leaning of individual speakers to experiment with -iesse > *-ise. One fringe benefit of our decision to heed the verbal paradigm is that we now understand, better than did the generation of Menéndez Pidal, the delayed spread of the monophthong from -ilio to the other cases. Apparently, it took the conver18 Prista—which occurs once in Elena y María, a short 13th-century dialog available in a scrupulous edition (1914) by Menéndez Pidal—lost out to priesta, because it would have set a precedent for a morphophonemic alternation in e:i obviously unacceptable for verbs of the -ar class. Note the contrast to the solution found for apriscar, pingar, and remugar, which consisted in leveling in favor of /, under a similar set of circumstances. But the development did not stop at that point ; see the following footnote.

240

MULTIPLE CAUSATION

gence of (a) internal phonological diffusion i < ie, and (b) the dramatically rapid prevalence of one paradigm of the preterit over another, around 1400, to produce certain limited results. It also becomes clear why the monophthongization of ie developed, erratically, in the direction of i rather than of e.19 3. After all these refinements have been introduced into our analysis, there re mains a residue of cases unsolved if not insoluble; it may be helpful to single out two categories for explicit mention. First, while measurable progress has been achieved in tracing the positive side of the evolution, i.e. the justification of actual change, the reverse side, namely the failure of certain words seemingly qualified to share in the development, can at this point be discussed only in conjectural terms ; to put it differently, predictability is still fairly low. Why has pierna 'leg, shank' fallen short of reaching the stage *pirna, even though ie was embedded in an environment favorable to change ? Was the force holding it back semantic (associa tion with pie, orig. piede [Berceo] 'foot') or phonetic (co-existence with [f.] tierna 'tender', in which the ie was more conservatively embedded) ?20 Why have miércoles 19 The less startling monophthongization ie > e was also in operation, though on a far more restricted scale. After certain consonants, especially ch, //, and ñ (anciently also χ /š/), the ie of certain suffixes, both inflectional (pret. -ieron, gerund -iendo) and derivational (-iego, -iento), was reduced to e; hence dix-eron (eventually dij-eron) 'they said' beside tem-ieron 'they feared'; bull-endo ' boiling, bubbling up ' beside gim-iendo ' sighing ' ; gall-ego ' Galician ' and manch-ego 'from La Mancha' beside and-ar֊iego ' r o a m i n g ' ; amarill-ento 'yellowish' beside gris-iento 'grayish'. More intricate cases include switches from radical-changing to entirely regular (i.e. leveled) verbs within the -ar class, as when older entriego ' I hand over', entregar ' t o hand over' and priesto ' I lend, loan', prestar ' t o lend' (alongside empriesto ' I borrow', emprestar ' t o borrow') have yielded ground to mod. entrego, entregar etc. (see Rosenblat 1946:280-82). In some instances the tendential avoidance of such heavy clusters as /ntrj/, /prj/ seems to have been the driving force—a cause which may also account for what on the surface looks like the interference of metathesis in OSp. criebo ' I burst, break', crebar (from CREPÖ ' I crackle, creak, rumble', CREPARE, cf. Fr. crever 'to burst') > quiebro, quebrar՝, but this explanation would not hold in the case of Class, aniego ' I drown' ( < ÊNECÖ ' I kill, exhaust, wear out') > mod. anego, or Class, tiemplo ' I temper, soften, ease' ( < TEMPERO ' I combine, blend, moderate') > mod. tempio. In the particularly disturbing development of PERTICA »pole, rod, staff' > OSp. piértega > mod. pértiga, there could have been some pressure from other members of the word family, differently stressed: pertigal 'id.', pertiguero 'verger', pertigueria 'office of the verger' (cf. Malkiel 1969d), while the change of the intertonic vowel calls to mind the vicissitudes of OSp. lagrema 'tear' (13th c.)> Class, and mod. lágrima. Finally, the generalization of weaklystressed es 'he/she/it is' < EST at the expense of heavily-stressed yes, in its final effect, contributed to the impact of ie > e. The two patterns of monophthongization, ie > i and ie > e, appear to have followed their separate courses with scant if any mutual interference, let alone conflict—despite their essential contemporaneousness. 20 In a truly innovative fashion, Montgomery, faced with the task of justifying the redistri bution of Lat. -ÉRE, -ERE, and -IRE verbs among the two Spanish verb classes available for this purpose (namely ֊er and -ir), not only takes into account the precise relation of stem vowel to thematic vowel (pursuing, in this respect, an attractive path opened up by Togeby 1972), but also strikes out on his own, studying systematically such configurations and sequences of sounds in slices of non-verbal material (e.g. in nouns) as may have provided models for the assignment of verbs to certain schemes. If one were to apply such an experimental approach to the problem on hand, one could, to say the least, wonder whether 2sg. pres. ind. ciernes (from cernir ' t o sift', whose [bs] and [n], aside from its membership in a verbal paradigm, were so many obstacles to the monophthongization of ie) could not, in a small way, have protected viernes 'Friday'


241

'Wednesday' (despite mierlo > mirlo) and viernes 'Friday' (despite viéspera > víspera) maintained their status quo, no doubt supporting one another in their opposition to change ? Could it have been because of the lag of closely-associated jueves 'Thursday', whose rising diphthong /we/ harmonized with the /je/ of miércoles and viernes? Over against these rather numerous cases of non-compliance with a strong trend, one is tempted to place a few equally baffling instances of over-fulfillment. Take the change of SĒRICU 'silken' into sirgo 'twisted silk, silk fabric'. Ancestral /e/ < Ē, I should of course normally have produced e, not i: DÏ(G)ITU 'finger' > dedo, PĒ(N)SU 'weight' > peso etc. In dealing with 'exceptions' like sirgo, scholars have alternately appealed to the agency of metaphony (a lame argument in the case of Castilian nouns, as against the Asturian), or to the influence of the heavy cluster /rg/— which may indeed have been a minor component in the aggregate of forces, especially if one starts from a pretonic syllable, as in sirguero or xirguero ' goldfinch, linnet' (later, through dissimilation of the r's, jilguero), the name of a bird known for its silken plumage. It is equally defensible, however, to argue that *serg-j*xerg֊ (i.e. the presumed original outcome of SĒRICU) were caught by the current that was transmuting sieglo into sigh etc., given such conspicuously favorable 'embedment' as /s/ or /š/ and /r/, plus neutral /g/, so that for once the shift e > і was allowed to flank the more common process ié > і. Such cases of sporadic overspilling of a sound change are at the root of numerous exceptions to regular correspondences, adding to the total impression of untidiness in a conventional historical grammar. The second—slowly shrinking—terra incognita concerns the tempo and relative chronology of the spread of і as a substitute for ie, even if one disregards the slight encroachment of metaphony on diphthongization in the verbal paradigm (sigo and sirvo displacing siego and siervo, at the threshold of the literary period). Here any attempt at analysis must await the meticulous preparation of twenty-odd miniature biographies. It is safe to predict minor surprises; thus sieglo, a relative newcomer if one endorses its classification as a medieval Gallicism, appears again and again as siglo (var. sygr) already in the poetry of Juan Ruiz (mid-14th century)—con ceivably with some assistance from above-cited sigo—despite the fact that segral 'secular' (for seglar < SAECULĀRE), which also entered Ruiz's lexicon, might have strengthened sieglo as a good morphophonemic partner (e:ie). On the other hand, the MSS of Ruiz's work show a trace of prisa beside favored priessa 'haste'. 21 from advancing to *virnes; whether lsg. and 3sg. pres. subj. pierda (from perder ' t o lose, squander, ruin') could not similarly have immobilized mierda 'excrement'; and whether (m.) tierno 'soft, tender'—its t and t inimical to the monophthongization of ie—could not have conservatively influenced infierno 'hell', despite the vicinity of to v. Despite their suggestiveness—or perhaps because of it—such arguments, at this stage of knowledge, are still very risky. 21 Judging from a comprehensive glossary (Richardson 1930), Ruiz consistently used -ilia, -ilio (e.g. nov-illa 'yearling cow', nov-illo 'young bull'), even where it was a mere sufiixoid (e.g. silla 'saddle' in four passages, and—analogically—also sillo alongside expected sello 'seal, sign, m a r k ' < SIGILLU). Arista, prisco, siglo (sygro), and víspera have all four reached the ter minal point of their respective evolutionary curves; priesa, undoubtedly the author's personal preference, still alternates with prisa (e.g., S, 479b, 480a, 550a, 971b: priessa; S, G, 512b: priessa; G, 984c: priessa; S, T, 1524b: priesa [in rhyme]; but S, 1691c: aprisa—clearly, a sign of scribal tampering). Elsewhere, the retention of ie was approved by posterity (avieso 'crooked',

242

MULTIPLE CAUSATION

4. One feature of the relation between the two monophthongizing processes, ie > i (beside ie > e) and the rarer ue > e, stands out very sharply at once : by the time the compression ie > i had already reached its maximum, the reduction of ue to e had barely started. To illustrate this crucial time lag, it suffices to scan the lexicon of Chancellor Pero López de Ayala's didactic poem Rimado de palacio (ca. 1400), observable through the prism of two satisfactory manuscript copies (N, E). The diphthong ue is still completely intact: afruenta 'dishonor, ignominious situa tion' (N, 87c, 470d), culuebra 'snake' (E, 703b), fruente 'brow' (N, 355b), corre sponding to mod. afrenta, culebra, frente. The preservation of ie—in contexts from which the rising diphthong was ultimately dislodged—is distinctly more sporadic. Thus, as regards the oft-cited diminutive suffix at issue, only -ilio is on record (cabd-, caud-illo 'head, leader, prince', chiquillo 'little, small'). Some of the other words pose exegetic problems because of conflicting evidence; thus E is conserva tive, with respect to a characteristic test word, in one passage (1806d, en el otro sieglo 'in the other world') and self-contradictory in another (572c, en los siglos de los sieglos 'for ever and ever'); it offers the reading ariesta 'straw' (1448d), but in rhyme with conquista, which at no earlier stage contained an ie: here the copyist appears to have heeded tradition more zealously than the relaxed poet-statesman. On the other hand, priesa 'urgent business' (N, 449b, 793b, 853c), a priesa 'in haste, speedily' (N, 513d; E, 1509b) point to continued use of ie, a preference deriving a certain morphophonemic justification (e : ie) from co-existence of the differentlystressed transparent cognate presura ' affliction, torment' (N, 795b, 846g; E, 1506c).22 Many (or, at least, certain) words indeed have histories of their own. The neatly measurable chronological lag of ue > e behind ie > i is only one of several factors tending to identify it as a sort of reverberation (a rather faint one, at that). The conspicuously small number of lexical items and proper names involved in ue > e is another relevant consideration: to afruenta, culebra, frente one can add little more than fl{u)eco 'fringe, flounce, raveled edge' (from FLOCCU); s(u)erba ' serviceberry, sorb ' (from the plural of SORBU) ; the obsolete and etymologically controversial combl(u)eça 'concubine'; and OSp. Burueva > mod. Bureba, the name of a section of Old Castile now entering into the province of Burgos. Some scholars lean toward adding to this inventory estantigua 'procession of hob goblins; bugbear, scarecrow' < OSp. huest(e) antigua, lit. 'old army', where the prieta 'brunette' etc.).Priz 'prayer' (242c) is an early Gallicism, due to reverberations of the Cluny Reform. One case difficult to diagnose is that of the postverbal for 'challenge to a single combat', a lexical item which occurs once (1203a), not in rhyme, and with all three manuscripts exhibiting divergent readings: S rrybto, G rrepto, Τ rriepto, from re{p)tar ' t o challenge' < REPUTARE. The regular form was, indisputably, rie(p)to; but in Classical and modern Spanish obsolescent reto 'challenge, dare, threat, (Amer.) insult' eventually won out. 22 I am basing my analysis on an unpublished Berkeley dissertation (Zeitlin 1931), whose author, in compiling the vocabulary, used not only A. Kuersteiner's by no means inadequate edition, but also photostatic copies of the two manuscripts. The material offers food for thought on many side issues, which limitations on space bid me to skip: the wavering -encía ~ -iencia {concencia over against mod. conciencia, diligiencia ~ deligencia), a process controlled by a subtle interplay of forces to the exclusion of the monophthongizing trend; e and i as tentative rival substitutes for ei in an exiguous group of imported words {deleit- ~ delet֊ ~ delit-acìón' delight, nleasnre) t.


243

diphthong figured in a counter-stressed syllable; but an early attempt to include cerdo 'hog' as an offshoot of SORDIDU must be written off as a failure. One also detects a certain parallelism between the impacts of ie > i and ue > e on the verbal paradigm, with the expected difference in intensity of spread. It will be remembered how the imminent possibility of the morphophonemic alternation e: i in -or verbs was averted, either through generalization of the i (apriscar, pingar, remugar) or through (temporary) return to e:ie (prestar). Something similar hap pened once or twice in the ranks of the verbs geared to a back-vowel in the radical. After afruenta 'offense', 'he/she offends' (inf. afrontar) began to gravitate toward afrenta, the likelihood of an unprecedented scheme afrentalafrontar loomed on the horizon; the complication was avoided through split of the original verb into (a) afrentar 'to affront, insult', flanked by the noun afrenta, and (b) afrontar 'to con front, bring to face', neither of them radical-changing. Similar sporadic attempts to allow lsg. pruebo (from probar 'to taste, test') to reach the more advanced form prebo, which has actually left traces in Asturias, Puerto Rico, and Mexico, were quickly rejected by the vast majority of Spanish speakers. Since, structurally, ie > e would provide a far closer parallel to ue /we/ > e than does ie > i—and given the fact, furthermore, that the two models of monophthongization ie > e and ie > i co-existed in Old Spanish, without being in one another's way—there inevitably arises the question: Is the traditional troublesome pairing of ie > i and ue > e defensible ? The query may be answered in the affirmative ; what holds ie > i and ue > e together is the pattern of conditioning by whole constellations of consonants, sometimes scattered over the entire word, rather than by a single immediately preceding consonant, as with ie > e : dixieron > dixeron. Among the consonants triggering the effect one encounters, as before, r and /; but the center of gravity has shifted to the bilabials and labiodentals—among which ƒ has now, at long last, acquired prominence : fr(u)ente, afr(u)enta—though b and v (less so m) are also important: Bur(u)eva etc. 5. It now remains for us to ventilate the expected question as to whether any set of lexico-morphological conditions could possibly have prepared the ground for the change ue > e, slightly recalcitrant to a smooth phonological explanation. One can, as a matter of fact, identify two such ensembles of circumstances ; unfortunately, they are somewhat less sharply contoured than was, e.g., the very neat case of the attraction of -iello by the triad of diminutive suffixes geared to I. There has, from Antiquity, existed a rather close relationship between the deriva tional suffixes -ÄRIUM and -ÖRIUM—a rivalry which, incidentally, has left a few traces in modern European languages (cf. Eng. inventory as against Fr. inventaire, Sp. inventario, Ger. Inventar՛, or Eng. sanitarium vs. Sp. sanatorio, Ger. Sanatorium, Russ. sanatorija). Among other functions which they shared, -ARIUM and -(ÄT)ÖRIUM could both be put to use to designate a place or a tool, except that the former pre vailed in nominal and the latter in verbal families—by no means a watertight division in Romance, where numerous families are bicephalous, i.e. grouped around one nominal and one verbal head. Galician-Portuguese to this day formally distin guishes between -eiro and -oiro, as does French between -ier and -oir, Tuscan between -aio (dial. It. -aro) and -oio, etc. In the earliest Spanish texts, many of them

244

MULTIPLE CAUSATION

Navarro-Aragonese rather than Castillan, -ero < -ARIU and -(ad)uero < -(ÄT)ÖRIU (via *-adoiro, cf. cuero 'leather' < *coirO < CORIU) 23 still co-existed, particularly in the language of Gonzalo de Berceo's devout poetry. Examples culled from a miscellany of 13th-century texts include: asmaduero 'worthy of esteem', cobdiciaduero 'desirable, provoking lust, whetting concupiscence', valeduero 'valid, bind ing'. In other contemporary or slightly later texts, or rival copies of the same texts, these and similar words end in -adero, -edero, -idero—which is of course the only family of composite suffixes familiar, ever since 1300, to speakers of Spanish {cerradero 'lock(ing)', duradero 'lasting', hacedero 'feasible'), in keen rivalry with -dizo (corredizo 'sliding', encontradizo 'bobbing up all the time', olvidadizo 'for getful, ungrateful').24 The original rapprochement between -ÄRIU(M) and -ÖRIU(M) was, I repeat, suffixal, hence based on morphological affinity; and in fact it trans cended the domain of Spanish. But around 1300 the adjectives in -duer began to merge with their counterparts in -ero, in favor of the latter ; specifically, to cite the name of a tool, the original descendant of TÖ(N)SÖRIA (pl.) 'scissors'—which, judging from Ptg. tesoira(s) ~ -oura(s), must have been *tisuera(s), *tixuera(s) in Proto-Spanish, with dissimilation of back-vowels—was allowed to become tixera(s) (witness mod. tijeras). Once this had occurred, there undoubtedly developed a feeling for a widespread change of ue to e, at least before r. The change at no time became pervasive; most words—e.g. fuero 'municipal ordinance' and the verb forms fuera, fuere (from seer 'to sit, be' and ir 'to go'), also the adv. fuera 'outside'—put up stiff resistance. But at least one noun was swept away by the current: STOREA 'mat(ting)', which of course contained no relevant suffix, but in the Late Latin pronunciation /storja/ resembled the words in -ÖRIA (f.), emerged in medieval Spanish from the start as estera, rather than *estuera. Can it be persuasively argued that shifts such as asmaduero > asmadero, valeduero > valedero, *tixuera(s) > tixeras, *estuerà > estera created a climate in which fruente > frente, Burueva > Bureba etc. become more readily understand able? The contiguity of r was, one. may assume, interpreted somewhat more liberally, with the treatment of (u)er being extended to r(u)e, as in fr(u)ente, or with a following r separated by a short segment (e.g. one interjacent [β]) being allowed to exert the same influence, as in cul(u)ebra. Given the extra-close ties between / and r, one understands, by stretching the imagination, the extension of the process to fl(u)eco and combl(u)eça.25 If one then wonders why flueco advanced to fleco, though clueca 'brooding' (speaking of a hen) failed to undergo the change to * cieca, and luego 'soon, at once' did not move over one step toward *lego, the obvious answer is that f through the proximity of its articulation to that of /w/, and independently, through its tiesto v and m as long-active conditioners of the ie > i 23

Note that Proto-Castilian /oj/ and /oj/ alike yielded ue. It is likely that the two falling diphthongs—to this day tidily kept apart in Portuguese—merged in Spanish before their transmutation into a rising diphthong, presumably through conflation with the far more fre quent ue < o. 24 For a very succinct, but no longer entirely satisfactory outline of the problem, see Hanssen (1913, §330). 25

Especially since comblueça may have been *combrueça (and before that *combuercaT), on the testimony of OPtg. comborça.


245

shift, could accomplish more, in conjunction with /, than / alone, as in luego, or than / supported by /k/, as in clueca—which was further hampered by the coexistence of the less phonosymbolic var. llueca. Moreover, the space of lego had already been pre-empted by lego ' lay ' < LAICU. One serious objection can be raised to the reasoning here advocated: If -duero changed to -dero shortly before 1300, while culebra, frente etc. are typically words of the concluding decades of the 15th century (at the earliest), is one not forced to infer from this chronology a gap of at least a century and a half—a lacuna that would exclude any direct influence from -d(ii)ero, est(u)era, tix(u)era etc. on Bureba, culebra, frente, and their companions ? The objection might be fatal, if the literary language known to us from philological evidence were the actual breeding place of sound changes. Experience teaches us, however, to expect humbler registers of speech—social and regional dialects alike—to be the far likelier matrix of such changes. Now in dialect speech, one finds all sorts of conflict; e.g., Galician -eiro < -ARIU and -oíro < -ÖRIU frequently alternate, witness tixeira ~ tisoira; in West Asturian, -oiro competes with -eiro, whereas in East Asturian, -oriu can be pitted against vernacular -eru and mock-learned -ariu etc. (cf. Malkiel 1969d: 262-4, 1970b: 43-4). From such a crazy-quilt of distributions one may safely extrapolate that, in peripheral rural dialects of Castilian, where oirjori had become uer, and eir had been reduced to er, the gradual change of ue to e (especially before r) continued long after 1300—a supposition which eliminates, with a vengeance, the aforehinted lacuna of a century and a half. 6. The second configuration of pre-existent morphological (once more, deriva tional) facts concerns the two adjectival suffixes -ueño and -eño. The former, neatly observable in halag-üeño 'flattering, attractive, charming' (from halagar 'to flatter', halago 'cajolery') and ris-ueño 'smiling' (from risa 'laughter', cf. sonrisa 'smile'), 26 failed to pick up momentum in Spanish, even though its Portuguese counterpart -onho (e.g. enfad-onho 'tiresome, tedious', trist-onho 'glum, unhappy, dejected') fared rather nicely.27 In fact, in rural speech Sp. -ueño suffered erosion; not unlike the Spanish descendant of -ŪGINE, it was fragmented and practically lost its identity, surfacing here and there disguised as -oño, -uño etc.—cf. terruño 'piece of ground', vidu{e)ño 'kind of grapevine', and a few other words, mostly obscure.28 On the 26

The third better-known adjective usually cited, ped-ig-üeño 'insistent, demanding, bother some', is best explained as patterned on halag-üeño resegmented as hal-agüeño, with substi tution of the thematic vowel i for a. The semantic link between 'adulation' and 'begging' in this particular culture has been independently adumbrated in a study of the vocalic gamut of limosna 'alms' and lisonja 'flattery' (Malkiel, MS). 27 The observations on -onho (apropos medronho 'fearful', also coll. dizonho 'saucy\palronho 'chatty') by Michaelis de Vasconcelos are still worth reading (see 1886, no. 24; 1887-89:301-3; 1893-94:132-3). 28 Some of the words cited—yet left untranslated and unlocalized—by Hanssen (§294), e.g. artuña ~ ortuña 'sheep struck by miscarriage' and redruña, are so marginal as to have been omitted by most standard dictionaries. The segment -uña has several other sources; in certain dialects it alternates with -una < -UNA; it may also conceal, in imitations of and elaborations on compounds, uña 'nail, claw' < UNGULA. The productivity of Ast. -oño is shown by seroño, as against OSp. seruendo 'maturing, ripening, late' < SÈRÖTINU (cf. Meyer-Lübke 1930-35, §7854.)

246

MULTIPLE CAUSATION

other hand, Lat. -INEU, practically doomed in Portuguese, developed splendidly in Spanish, on both sides of the Atlantic (limeno 'of, from Lima', madrileno 'of, from Madrid', porteño ' pertaining to the harbor [of Buenos Aires] ' etc. ; cf. also cañameño 'hempen', marfileño 'made of or resembling ivory', navideño [haplologically, in lieu of *-dadeño] 'of Christmas'). 29 It is plausible that, in its sweeping spread, -eño absorbed a few remnants of crumbling -ueño; if that happened (and the humility of most -ueño words makes the record of the suffix too spotty to invite adequate documentation), then another minor contributory factor, again morphological in essence, namely -ueño > -eño, has been identified as, at least, a strong possibility. 7. One expects to find, as the by-product of any extended and ' tortured ' sound change, a few instances of false regression (or hypercorrection). No disappointment is in store for us, provided we include in our purview the puzzling case (at first glance) of curtiembre ' tannery ' (from curtir ' to tan ') and urdiembre beside urdimbre 'warp, warping chain', fig. 'scheme, scheming' (from urdir 'to beam a warp', fig. 'to plot, scheme'); the underlying verbs may be, respectively, a local elaboration on RETERERE 'to grind again and again, to small bits', and ÖRDÏRÏ 'to start [weaving], undertake', a verb originally accompanied by such action nouns as EXand PRÏM-ÖRDIUM.

Early Romance inherited from Latin a triad of derivational suffixes used chiefly for action nouns (or, concretely, for the results of the actions involved) : -AMEN, -IMEN, and -UMĚN. Older Spanish, specifically, made wide use of -ambre (traceable to -AMINE or representing an extension of that suffix's original scope) and even more of -umbre (which can be similarly linked with -ŪMINE except that, at a certain point, it also contaminated -TŪDINE, a process which gave rise to such formations as much-ed-umbre 'crowd, abundance'). It is readily understandable that urdimbre came into existence in Old Spanish (alternatively, *ÖRDÏMEN may, with a minimum of risk, be projected onto the level of provincial Latin, where it could have ousted such, by then, awkward satellite formations as EX- and PRÏM-ÖRDIUM). But what about the var. urdiembre and its echo, particularly in New World Spanish, curtiembre, which seems to presuppose curtimbre as its starting point? There are no prototypes for -iembre available in the parent language; equally important, ie does not in the least fit into the vocalic gamut a, i, u, as an appealing variation upon i. The only explanation that comes to mind is that, at the height of the conflict between innovative ristra and older riestra etc., groups of stubbornly conservative speakers 'corrected' -imbre to -iembre. Note that the occurrence of m, b, and r in a single close-knit suffix, plus the preceding r, did give the over-all impression that ord-, urd-imbre was structurally close to mirlo, prisa, prisco, ristra etc. ; the same consider ation holds for curtiembre. 8. What conclusions are we to draw from the facts here assembled and the explan ations here proffered? If in 1966-68, in my first concentrated attack on the ie > i, ue > e 'riddle', I succeeded in isolating a single factor, involving analogical pressure by a close-knit group of derivational suffixes on one maverick suffix, 29

The currently best treatment of -eño, superseding my own and G. Bonfante's earlier notes (which served their purpose by eliminating Meyer-Lubke's infelicitous Sp. -eño < Lat. -IGNU), ÍS given bv Butler (1971 : 6 1 - 7 2 . See also the critical annrasial by asanoff (1975-561)


247

semantically germane, as a plausible co-conditioner of sound change (-iello > -ilio under the influence of -ico, -in ~ -ino, -ito), I have now been fortunate to detect as many as three more such ingredients, of varying depth of force and presumable impact on the erratic monophthongization : (a) the rivalry of two paradigms in the 'weak' preterits of OSp. -er and, on a more modest scale, -ir verbs; plus, slightly less sharply profiled, (b) the merger of -ero < -ARIU and -uero < -ÖRIU, in favor of the former; and (c) sporadic contacts, not yet very accurately describable, between -ueño < -ÖNEU (or -ONiu) and -eño < -INEU, with the former discernibly withering and the latter blossoming out. There was hardly any causal or typological connection among these four disparate factors at the initial stage; and the fact that, in each instance of rivalry, the monophthongal competitor won out may be due, at least in part, to sheer coincidence, since miscellaneous circumstances, difficult to sub sume under one common denominator, were clearly at work. Thus three hypocoristic suffixes geared to i outnumbered one involving ie; the /-preterit {-iste, -imos, -istes) found support, in terms of symmetry, in the powerfully represented apreterit (tom-aste, tom-amos, -astes), while the ie-preterit was fairly isolated; words in -ero, from the start, overwhelmingly outnumbered those in -uero. Without dismissing, then, the margin of pure coincidence, we may ask ourselves whether, nevertheless, some common causal element—a kind of coefficient—cannot be identified alongside the diversified ingredients. A positive answer can indeed be provided: Beyond a certain cut-off point, toward the close of the Middle Ages, the characteristic rising diphthongs ie and ue were past their crest, and there is inde pendent proof of their slow, gradual recession, e.g. in an equally widespread 'normal' sound change, such as ie > e (-iego beside -ego, -iento beside -ento, -ieron beside -eron etc.) Within the chosen language family, ie > e may defensibly be called 'normal', while ie > i and ue > e appear 'aberrant', on account of the divergence in their conditioning and predictability. A single phonotactic condition, clearly statable, controls ie > e, and the exceptions are few and easily explainable ; conversely, it takes a hazily defined constellation of circumstances (at least two, and preferably three to four) to produce the shifts ie > i and ue > e; also, the exceptions are relatively numerous and invite no simple justification. Categorically, it is indeed legitimate to invoke a multi-conditioned sound change, as different in its complexity from the remainder of phonological shifts as are (say) the nominal augments (Menéndez Pidal's 'sufijos átonos') from the bulk of Romance derivational suffixes. The chronological precedence of ié > i over uè > e is indisputable. For the oft-observed asymmetry (it is the first segment of ié and the second of ue that wins out) we offer the triple explanation that (a) compression of diphthongs was, generally, in the air after 1300; (b) in terms of broad structural outline, the diph thongs ie and ue (uo) condition each other as frequently as do the monophthongs i and u, or e and o, so that the compression of either diphthong serves as a stimulus for similar action on the other; (c) the urge toward compression used such channels as happened to be most easily available, as a result of the afore-cited morpho logical preludes. The impact—in diachronic perspective—of morphology on phonology, so brilliantly foreseen by Ernst G. Wahlgren, is by no means limited to unusual sound

248

MULTIPLE CAUSATION

changes, but is likely to have been at its strongest and is certainly best observable in that domain. REFERENCES ALARCOS LLORACH, EMILIO. 1951. Esbozo de una fonología diacrònica del español.

Estudios dedicados a Ramón Menéndez Pidal 2.9-39. Madrid: Consejo de In vestigaciones Científicas. . 1954. Fonología española. Rev. 2d ed. Madrid: Gredos. . 1961. Fonología española. Rev. 3d ed. Madrid: Gredos. BUTLER, JONATHAN LOWELL. 1971. Latin -ïnus, -ina, -ïnus, and -ïneus: from Proto-IndoEuropean to the Romance languages. (UCPL, 68.) Berkeley & Los Angeles: Uni versity of California Press. CATALÁN, DIEGO. 1964. Nuevos enfoques de la fonología española. [Review of Alarcos Llorach 1961.] Romance Philology 18.178-91. . 1974. Lingüística ibero-románica: crítica retrospectiva. Madrid: Gredos. CHEN, MATTHEW Y., and WILLIAM S-Y. WANG. 1975. Sound change: actuation

and

implementation. Lg. 51.255-81. COROMINAS, JOAN. 1956-57. Diccionario crítico-etimológico de la lengua castellana. Vols. 3-4. Madrid: Gredos. CRADDOCK, JERRY R. 1970. Review of Directions for historical linguistics, ed. by W. P. Lehmann & Y. Malkiel. Lg. 46.688-94. . 1973. Review of Spanish phonology, by J. W. Harris. Linguistics 109.83-109. HANSSEN, FEDERICO. 1913. Gramática histórica de la lengua castellana. Halle: M. Niemeyer. [Reprinted, Buenos Aires: El Ateneo, 1945.] HARRIS, J. W. 1969. Spanish phonology. (Research monograph 54.) Cambridge, MA: MIT Press. HASSELROT, BENGT. 1938. Ernst G. Wahlgren. Vox Romanica 3.352-3. JACOBSEN, WILLIAM H., JR. 1957. Review of Alarcos Llorach 1954. RPh. 10.258-66. JASANOFF, JAY H. 1975. Review of Butler 1971. RPh. 28.555-64. LAPESA, RAFAEL. 1948. Asturiano y provenzal en el Fuero de Aviles. (Acta Salmanticensia, Filosofía y Letras, 2:4.) Salamanca. ——. 1950. Historia de la lengua española. Rev. 2d ed. Madrid: Escelicer. [3d ed., 1955; 4th ed., 1959, etc.] LEHMANN, WINFRED P. 1973. Historical linguistics: an introduction. 2d ed. New York: Holt, Rinehart & Winston. MALKIEL, YAKOV. 1966. Diphthongization, monophthongization, metaphony: studies in their interaction in the paradigm of the Old Spanish -ir verbs. Lg. 42.430-72. . 1967a. Multiple versus simple causation in linguistic change. To honor Roman Jakobson: essays on the occasion of his seventieth birthday, 1228-46. The Hague: Mouton. . 1967b. Linguistics as a genetic science. Lg. 43.223-45. . 1968. The inflectional paradigm as an occasional determinant of sound change. Directions for historical linguistics, ed. by W. P. Lehmann & Yakov Malkiel, 20-64. Austin: University of Texas Press. . 1969a. Sound changes rooted in morphological conditions: the case of Old Spanish /sk/ changing to /0k/. RPh. 23.188-200. . 1969b. Morphological analogy as a stimulus for sound change. Lingua e Stile 4.305-27. . 1969c. The five sources of epenthetic / j / in Western Hispano-Romance: a study in multiple causation. Hispanic Review 37.239-75. . 1969d. Latin pedica, *pënsum, and pertica in Hispano-Romance. Études Romanes de Lund 18 ( = Mélanges Alf Lombard), 130-50. ——. 1970a. Le nivellement morphologique comme point de départ d'une 'loi phoné tique': la monophtongaison occasionnelle de ie et ue en ancien espagnol. Mélanges


249

de langue et de littérature offerts à Jean Frappier, 701-35. Genève: Droz. [Actually accepted in 1966.] . 1970b. Patterns of derivational affixation in the Cabraniego dialect of East-Central Asturian. (UCPL, 64.) Berkeley & Los Angeles: University of California Press. . 1971. Derivational transparency as an occasional co-determinant of sound change : a new causal ingredient in the distribution of -ç- and -z- in ancient HispanoRomance. RPh. 25.1-52. . 1973a. Deux frontières entre la phonologie et la morphologie en diachronie. Langages 8:4 (Le changement linguistique, éd. by S. Lecointre & J. Le Galliot), 79-87. . 1973b. Quelques avatars romans d'un zoonyme et d'un ornithonyme latins. Études de langue et de littérature du Moyen Age offertes à Félix Lecoy, 377-84. Paris: H. Champion. . 1974. New problems in Romance interfixation (I): the velar insert in the present tense (with an excursus on -zer/-zir verbs). RPh. 27.304-55. . 1976a. From falling to rising diphthongs: the case of Old Spanish io < *éu. RPh. 29.435-500. . 1976b. In search of penultimate causes of language change: studies in the avoid ance of /z/ in Proto-Spanish. Current studies in Romance linguistics : papers from the 1974 Texas Symposium, ed. by M. Lujan & F. Hensey, 27-36. . MS. Ancien espagnol losenja/lisonja 'flatterie' et (à)limós(i)na 'aumône': en marge du rayonnement transpyrénéen d'un provençalisme littéraire. To appear in Mélanges Jean Séguy. MARTINET, ANDRÉ. 1952. Review of Orígenes del español, by R. Menéndez Pidal, 3rd ed. Word 8.182-6. MELANDER, J. 1937/38. Nécrologe [Ernst G. Wahlgren]. Studia Neophilologica 10.193. MENÉNDEZ PIDAL, RAMÓN. 1904. Manual de gramática histórica española. [Rev. 6th ed.,

1941.] . (ed.) 1914. Elena y María. (Disputa del clérigo y el caballero.) Poesía leonesa inédita del siglo XIII. Revista de Filología Española 1.52-96. . 1920. Notas para el léxico románico. Revista de Filología Española 7.1-36. . 1926. Orígenes del español: estado lingüístico de la Península Ibérica hasta el siglo XI. Madrid: Espasa-Calpe. [Last revised ed. ( = Obras completas, 8), 1950.] MEYER-LÜBKE, WILHELM. 1885. Zur romanischen Laut- und Formenlehre, II: zum schwachen Perfektum. Zeitschrift für Romanische Philologie 9.233-67. . 1930-35. Romanisches etymologisches Wörterbuch. Rev. 3d ed. Heidelberg: Winter. MICHAELIS DE VASCONCELOS, CAROLINA. 1886. Studien zur romanischen Wortdeutung.

Miscellanea di filologia e linguistica in memoria di Napoleone Caix e Ugo Angelo Canello, 113-66. Firenze: Le Monnier. . 1887-89. Etimologías portuguesas. Revista Lusitana 1.298-305. . 1893-94. Fragmentos etimológicos. Revista Lusitana 3.129-90. MiCHAËLSSON, . 1924. Le passage J > r en français. Studier i Modern Språkvetenskap 9.259-98. MONTGOMERY, THOMAS. 1976. Complementarity of stem-vowels in the Spanish second and third conjugations. RPh. 29.281-96. NAVARRO TOMÁS, T. 1931. Review of Wahlgren 1930a. Revista de Filología Española 18.393-5. . 1944. Manual de pronunciación española. Rev. 4th ed. New York: Stechert. (The original ed. appeared in Madrid: Centro de Estudios Históricos, 1918.) O'BRYAN, MARGIE. 1974. The interaction of morphological and phonological processes in historic change. Linguistics 137.49-61. RICHARDSON, HENRY B. 1930. An etymological vocabulary to the Libro de buen amor ... New Haven: Yale University Press.

250

MULTIPLE CAUSATION

ROCHET, BERNARD. 1974. A morphologically-determined sound change in Old French. Linguistics 135.43-56. ROSENBLAT, ÁNGEL. 1946. Notas de morfología dialectal. Appended to translation, expansion, and annotation of Estudios sobre el español de Nuevo Méjico, II, by Aurelio M. Espinosa. (Biblioteca de dialectología hispanoamericana, II.) Buenos Aires: Instituto de Filología. SALA, MARIUS. 1970. Contribuţii la fonetica istorică a limbii române. Bucureşti: Editura Academiei. SCHMID, HEINRICH. 1949. Zur Formenlehre von dare und stare im Romanischen. (Romanica Helvetica, 31.) Bern: Francke. SEIDENSPINNER DE NUNEZ, DAYLE. The poet as badger: notes on Juan Ruiz's adapta tion of the Pamphilus. RPh. 30.123-34. SOLA-SOLÉ, J. M. 1975. El Auto de los Reyes Magos: ¿Impacto gascón o mozárabe? RPh. 29.20-7. SPITZER, LEO. 1931. Review of Wahlgren 1931b. Revista de Filología Española 18.185-7. STAAFF, ERIK. 1930. Review of Wahlgren 1930a. Studia Neophilologica 3.89-92. TOGEBY, KNUD. 1972. L'apophonie des verbes espagnols et portugais en ֊ir. RPh. 26.256-64. WAHLGREN, ERNST G. 1914. Étude sur les actions analogiques réciproques du parfait et du participe passé dans les langues romanes. Uppsala dissertation. . 1920. [Expanded version of preceding item.] Uppsala Universitets Årsskrift. . 1925. Sur la question de 17 dit parasite dans l'ancien français. Mélanges de philo logie offerts à J. Vising par ses élèves et ses amis Scandinaves ..., 290-335. Göteborg: N. J. Gumpert; Paris: É. Champion. . 1928. Le nom de la ville de Marseille. Studier i Modern Språkvetenskap 10.25-64. . 1930a. Un problème de phonétique romane: le développement d > r. (Skriften utgiven av K. Humanistiska Vetenskapssamfundet i Uppsala, 26:4.) Uppsala. . 1930b. Français utague, itague, étague. Studia Neophilologica 3.82-6. . 1930c. Ancien français isnel, ignei. Studia Neophilologica 3.116-29. . 1931a. Observations sur les verbes à parfaits faibles: étude de morphologie et de phonétique françaises, 1:1. Uppsala: [no publisher.] (On circumstantial evidence, this fascicle—apparently the only one that appeared—was privately printed between 1925 and 1928, and the cover in 1931.) . 1931b. Franc, surouest, suroît, esp. sur, port. sul. Studier i Modern Språkvetenskap 11.103-45. • . 1933-34. Contribution à l'histoire de l'évolution des terminaisons de l'imparfait de l'indicatif et du conditionnel en français. Studia Neophilologica 6.1-24. . 1936. Evoluzione semasiologica d'alcune parole dotte nell'italiano. Tr. Silvia Tomba. Uppsala & Stockholm: Almqvist & Wiksell. (Text of a lecture delivered in Swedish the year before.) . 1937. Sur l'origine de 17' initial des pronoms démonstratifs français du type icist, icil. Studier i Modem Språkvetenskap 13.105-43. WANG, WILLIAM S-Y. 1969. Competing changes as a cause of residue. Lg. 45.9-25. ZEITLIN, MARION Α. 1931. A vocabulary to the Rimado de palacio of Pero López de Ayala. Berkeley : University of California dissertation. . 1939. Unstressed ascending diphthongs in Spanish. Modern Language Forum 24.84-90.

MULTIPLE VERSUS SIMPLE CAUSATION IN LINGUISTIC CHANGE

There exists an extensive corpus of literature on the individual forces that are known to spring into action in the shaping of language. Depending on each research er's range of experience, catholicity of taste, and doctrinaire position regarding certain controversial matters such as the principle of teleology in evolution, he will incline to favor either external or internal factors, either the hypothesis of gradual improvement (under adverse circumstances, deterioration) of structure or the alterna tive hypothesis of mere regrouping of the structure's constituents, either an explana tion allowing for the intervention of speakers, at varying levels of consciousness, in the events affecting their speech or the rival explanation operating with unguided clashes of blind forces. There has been considerable progress in the recognition of different categories of change, especially within phonology: thus, it would hardly occur to any enlightened Romanist at present to champion the application of, say, 'Verner's Law' to a sliver of material that does not fall under that 'law's' jurisdiction, because this century's scholars, unlike their fumbling predecessors, have learned to distinguish quite sharply between universals and particulars. Yet the domain of linguistic etiology remains underdeveloped as long as no major efforts are made to explore the compatibilities of different determinants of change. Without a searching analysis of these affinities, complementarities, and reconcil abilities (or, conversely, the patterns of their mutual exclusiveness) no serious study of multiple or complex causation can be undertaken. Every historical grammar or etymological article is, of course, replete with incidental, non-committal remarks to the effect that concomitant circumstances may have accelerated the process at issue or have increased the likelihood of the event under study; yet a formal, systematic high-level inquiry into such concomitancies seems to be unavailable. The present article, as a result of its severely limited scope and its random illustra tions, does not aspire to filling this need: It remains purely exploratory from beginn ing to end and is meant merely to dramatize the need for a full-scale investigation so slanted and to anticipate a few of its possible conclusions. The paper falls into two parts: a semi-technical analysis of a derivational problem never before clearly outlined, I suspect, and the succinct restatement, under a new angle, of three earlier probings, two of which have given rise to memorable discussions.

I. SPANISH ADJECTIVES IN -IO A superficial inspection of the stock of modern Spanish adjectives suffices to show that a conspicuous proportion of them end in unstressed -io (m.sg.). Here is a random collection of examples:

252

MULTIPLE CAUSATION

agrio 'sour', amplio 'full, roomy', lacio 'faded, withered, languid'; 'straight (as hair)', limpio 'neat', lucio 'shiny', medio 'half', mustio 'sad, gloomy, withered', necio 'foolish, crazy', nimio 'excessive, stingy, small', reacio 'stubborn', recio 'strong, coarse, harsh, hard', rubio 'blond', sabio 'wise, learned', sandio 'silly', serio 'serious', soberbio 'proud, magnificent, fiery', sucio 'dirty', tibio 'lukewarm', turbio 'muddy, confused', zafio 'rough, uncouth'. As a group, these forms seem to have aroused scant attention, though on certain items there exists an abundant collection of writings. The diachronist will distin guish between two major subgroups, of which the first includes adjectives, both vernacular and learned, whose final segment -io could have been readily foreseen from elementary knowledge of phonology, while the second encompasses the more difficult cases. Of these, some involve -io instead of expected ~o; in others '-io re places -io. The dissection of any residual categories will be relegated to the very end of the survey. (1) Historically oriented linguists have long since recognized within this group typical descendants of Latin adjectives in -IDUS; in fact, if we choose to reverse the perspective at this point and to ask ourselves what have been the characteristic Spanish reflexes of ancestral -IDUS formations, we shall soon discern (A) a large group of adjectives in -io and (B) an exiguous parcel of nouns and adjectives (a) either in -do (cf. Fr. froid, raide, tiède) or (b) in -do alternating, under varying conditions, with -io.l Taking into account, to round out the picture, some dialectal and obsolete forma tions, one arrives at the following break-down: (A)

-lDU>-rø:

FLACCIDU 'flabby, weak' > Sp. lacio; LIMPIDU 'translucent' > Sp. limpio 'clean, neat'; LŪCIDU 'clear, bright' > Sp. lucio (OSp. luzio); NITIDU 'shining, glittering' > Leon, nidio ( = Ptg. nédio) 'sleek, glossy, chubby'; RANCIDU 'stinking' > Sp. rancio 'rank, stale'; RIGIDU 'stiff, unpolished' > Sp. recio (OSp. rezio); RÖSCIDU 'moistened, dewy' > dial, ro-, ru-cio; SAPID U 'tasty, flavorsome' > sabio 'wise' 1) To simplify matters, I am leaving out of account the learned and, for the most part, dis tinctly late adjectives in -ido, such as árido, cálido, candido, escuálido, espléndido, estólido, estúpido, fétido, flaccido, florido, frígido, insípido, límpido, lúcido, nítido, pálido, rápido, sólido, sórdido, túrbido, vivido (semilearned húmedo contrasted, at the medieval stage, with strictly bookish húmido). Note that Spanish uses these Latinisms far more sparingly than does English; thus, it lacks the exact counterparts of lurid, punid, and vapid, favors rabioso 'rabid' over rábido, and presses into service morboso where English seizes upon morbid; English, in turn, has no Latinism to match Sp. cálido, uses gravid more parsimoniously than Spanish does grávido, and marginally tolerates pallid. The modern languages have been reluctant to adopt certain adjectives in -IDUS; one such outcast has been CALLIDUS 'sly, cunning', perhaps on account of its proximity to CALIDUS 'warm, hot'. In modern languages GELIDUS 'frosty, ice-cold' has been adopted only in poetic discourse, ALGIDUS 'cold' practically on no level; LEPIDUS 'pleasant, charming, elegant' seems to be represented in Italian alone (lèpido 'facetious, witty'), on a very modest scale also in Portuguese. I further omit from consideration such lexical items as have reached Spanish by way of a detour, e.g., neto 'neat, net', traceable to NITIDUS via a French intermediary.

MULTIPLE VERSUS SIMPLE CAUSATION

253

(cf. resabio 'unpleasant aftertaste'); SŪCIDU 'juicy, full of sap' > Sp. sucio (OSp. suzio, var. sucio) 'dirty'; TEPIDU 'lukewarm' > Sp. tibio (OLeon. tebio);TURBIDU 'unquiet, troubled' > Sp. turbio (OSp. turvio).2 (B)

-IDU >-do:

(a) CALIDU 'warm, hot' > Sp. caído 'broth, sauce, gravy'; SOLIDU 'dense, firm' > Sp. sueldo 'salary, sold, ancient coin'; (b) FRIGIDU 'cold, cool' > frío (via *fri֊io) beside topon. Fontefrida՛, PŪTIDU 'rotten, nauseous' > pudio (cf. topon. Fuentpudia > Ampudia, RÏVU PŪTIDU > Repudio) beside OSp. pudto՝, RAPIDU 'swift > raudo 'rapid, impetuous' (of water), 'whistling' (of wind) ֊ but OSp, rabdo, occasionally rábido beside rabio in an archaic layer of Old Spanish. Though the fact of the bifurcation of -IDUS has long been accepted,3 the process has, to my knowledge, never been cogently explained. One finds hints of the phonetic affinity, hence mutual attraction, between (radiCal-fìnal) L and (suffixal) D, a circumstance which may indeed have provoked very early syncope (cf., within the bounds of Latin, VALDĚ 'very' < VALIDE), thus placing the -D- in a position 2) Some of these formations pose major phonological problems (such as the ֊z- of OSp. rezio), but nearly all are etymologically transparent. There arises one taxonomie question: In both editions of his comparative dictionary (1911-20 and 1930-35) W. Meyer-Lübke complicated a relatively simple situation by separating (a) Prov. moste 'wet', Cat. mustie ^withered, flabby, sad' (which he subsumed in § 5780 under *MUSTIDU, supposedly based on MUSTEUS 'must like, sappy, sticky; fresh, young' - an offshoot of MUSTUM 'must') from (b) Fr. moite 'wet', Oce. muide 'pale', which he classed as descendants of MUCIDU 'mouldy, musty' (from MUCËRE) X MUSTEU ( § 5711; Rum. muced, It. mùcido 'dank', Surs, miš were declared lineal descendants of MÜCIDU). But what is *MUSTIDU other than a blend of MUCIDU and MUSTEU? The two entries should clearly have been consolidated. E. musty exemplifies the same semantic affinity between 'must-like' and 'mouldy'; SŪCIDU 'juicy' > O S p . suzio 'dhty' offers a similarly curved semantic trajectory; E. moist perpetuates OFr. moiste which underlies mod. moite. On the lively sense development of Sp. mustio Corominas remarks: "... cuyo sentido ... evolucionó en castellano por la flojedad de las cosas mojadas". 3) In his less than entirely persuasive note "Modo de obrar el substrato lingüístico", RFE, XXXIV (1950), 1-8, esp. 5, R. Menéndez Pidal provided copious toponymie evidence of OSp. rabio < RAPIDU: aqua rapia doc. A.D. 1019; Rabias in Asturias, La Rabia in Santander, Ptg. Rabia beside Raiva, Guip. Fuenterrabía < *-râbia traceable to Fontem Rapidům as attested by Archbishop Rodrigo of Toledo (13th c ) . Menéndez Pidal attributed the retreat oí rabio before rabdo to the threat of a homonymie collision with rabia 'fury', rabiar 'to rage', from RABIES. He was less convincing in his insistence on the accuracy of the equations PALLIDU 'pale' > pardo 'dark-brown; drab, dull' and LIMP IDU >lindo. For a detailed refutation see J. Corominas, Diccionario crìtico etimològico, III (1956), lOlò- and 663Û-664Û; also "Supplement", IV (1957), 1039b and 1060Z?, with a reference to M. L. Wagner's parallel demurrer. On the evidence of COLLOCARE 'to place' > colgar 'to hang' the segment -LLID- could have yielded, at most, *-ld-. For the stress shift in Fuenterrabía it seems unnecessary to implicate the peculiar accentual conditions in Basque; as in rocío > rocío, the loss of its adjectival status may have propelled the word in the dkection of -ία, away from -ia. Rabdo calls to mind Rioj. jaudo, Jud.-Sp. -, tav-do beside aberrant Arag. jauto O S p . cabdal > m o d . caudal 'wealth, great volume', also caudillo 'chieftain', etc.

254

MULTIPLE CAUSATION

where it was not subject to erosion at the critical stage; the -LD- formations may have acted as a rallying-point for Group B. Through lexical polarization CAL(I)DUS could, in turn, have entailed the tendential contraction of FRÏGIDUS to FRÏGDUS (in actual pronunciation, perhaps /frijdu/)4 and have thus set in motion a new, short-lived trend. While this argument may hold water, another facet of the process is of conceiva bly greater relevancy to us: Even where the semantic development of the incontro vertible Հօ < -IDU formations has been strikingly lively (as is true of lacio and sucio), they have remained essentially adjectival. In contrast, the distinctly fewer -do items have evolved a predominantly substantival function. Note the nominalization of Sp. caldo, as against Rum cald, It. caldo, Fr. chaud, etc., all of which fundamentally still mean 'warm, hot', and the complete divorce of sueldo from the ranks of ad jectives. Fri(d)o, of course, straddles both form classes, but its exemplifying power is weak on account of the coalescence of its nuclear i with the opening segment of -io, a. feature which sets it accentually apart from all the words here surveyed. Ob solescent raudo seems to contradict the tendency; yet, not improbably, the differen tiation raudo (adj.) 'swift, tempestuous' -raudal(subst.) 'stream, torrent'; 'abundance' is merely secondary; cf. the dual grammatical function of its near-synonym manantial 'flow(ing)' alongside strictly substantival manantío. The failure of lindo 'beautiful', orig. 'courtly, well-mannered, well-dressed', to fit this description in an additional powerful argument - if any were needed — for assigning it to the progeny of LEGI TIMUS 'rightful' rather than of LIMPIDUS. The reverse of the medal displays an even smoother surface: The preponderantly adjectival character of Հօ is confirmed by every single example, making the equation SQUALIDU 'rough (from want of attention), dirty, waste' > Sp. escalio 'wasteland waiting to be cultivated' somewhat dubious despite its appealing semantics. Gal. murcio 'moisture exuded by slightly putrid pickled meat', a technical term classed as an outgrowth of MURCIDU 'lazy, indolent', is appropriately enough flanked in the West by an adjective of far wider scope and higher incidence, Ptg. murcho 'wilted, faded, wrinkled, droopy'.5 The clinching argument is provided by rocío 'dew' which, after its adventitious substantival ingredient had become more heavily weighted than the original adjectival kernel, actually underwent accentual restructur ing, with concomitant semantic support from regadío 'irrigable (land)', manantío,

4) For a broad view of the linguistic phenomena at issue see my articles "Lexical Polarization in Romance", Lang., XXVII (1951), 485-518, esp. 505-507 (on the syncope of the intertonic vowel in CALIDUS and FRÏGIDUS), and "Diachronic percharacterization in Romance", Arch. Ling., IX (1957), 79-113, esp. 103-106, and X (1958), 1-36. 5) Ptg. murcho, of course, involves Mozarabic transmission, all the more plausibly as barring the infiltration of an unusually early Italianism (a remote possibility) - Sp. marchito Vithered, languid' and marchitar 'to wilť are also best understood on the assumption of Moz arabic channeling of MARCËRE, *-ÏRE 'to droop, be fainť. For the latest pronouncement on this conduit see Corominas, Breve diccionario (Madrid, 1961), p. 373b. Meyer-Lübke operates with a long U in MŪRCIDUS; it seems wiser to accept A. Ernout and A. Meilleťs interpretation of the word as MURCIDUS {Diet. étymol4, s.v. MURCUS 'mutilated, cowardly, lazy') and to attribute the  of murcho to metaphony, cf. rubio and tibio.


255

etc., where -ίο echoes -ÏVU. In two additional, much smaller categories, of which one may be represented by a single piece, the emergence of -io presupposes no intrusion of any adventitious factor of causation: (2) Sp. rubio (OSp. ruvio) 'blond' < RUBEU 'red(dish)' - on all counts a vernacular word, with the characteristic preservation of /j/ after a labial (as against Fr. rouge 'red') and without any supervenient metathesis (as against Ptg. ruivo 'reddish-blond'). In fact, metathesis occasionally worked in the opposite direction: NERVU 'sinew, tendon, nerve' > OSp. niervo (matching Fr. nerf, Ptg. nervo, It. nèrvo, etc.) > Sp. nervio,1 with additional learned support from NERVIA, -ÖRUM.8 (3) Sp. necio 'silly' < NESCIU 'ignorant'. A learned, perhaps mock-learnèd. formation, beside even older, austerely learned medio 'half < MEDIU and tercio 'third' < TERTIU, as against thoroughly vernacular Sp. mastuerzo (OSp. mas-, mestuerço) 'common-cress, pepperwort' < NASTUR-TIU, -CIU 'kind of cress' or corzo (OSp. corco) 'roe deer' < *CURTIU.9 There are on record numerous more modern Latinisms: espurio, eximio, nimio, obvio, etc. The halo of connotations surrounding nimio ('prolix, stingy, worthless', etc.) testifies to an osmosis of erudite and racy formations in -io. (4) As we come to the next category (-io in lieu of expected *-o), it becomes necessary to resolve the joint discussion into a bundle of individual case histories. Special vignettes are called for to justify Sp. soberbio, agrio, and amplio in the face of Lat. SUPERBUS, ÄCRUS (beside more commonÄC-ER,-RIS,-RE), and AMPLUS, particularly in view of perfectly normal Ptg. soberbo, agro, and ampio. (a) Sp. soberbio < Lat. SUPERBU 'proud, haughty, magnificent' (alongside SUPERB-ΙΑ 'pride', a derivative comparable to FALLÄC-IA 'artifice', INERT-IA 'slothfulness', MODEST-ΙΑ 'temperance', MOLEST-ΙΑ 'annoyance', SOLLERT-IA 'cleverness'). Soberbia is residually available to speakers of Spanish for 'pride, arrogance, presumption, magnificence', though it yields the right of way to orgullo, 6) For the classic analysis of OSp. rucio, Gal. ruzo 'graying' (of hail), lit. 'sprinkled, spattered with patches of gray', a congener of Ptg. ródo 'dew' < RÖSCIDU (beside occasional rocío, as in Modern Spanish), see Menéndez Pidal, "Etimologías españolas", Rom., XXIX (1900), 334-379, esp. 369f. The older conjectures, such as RUSSEU 'reddish' (F. Diez, in the wake of R. Cabrera) and LUCIDU X RUSSU (Meyer-Lübke, 1890), have meanwhile all been discarded. 7) The treatment of /bj/, /vj/, whether primary or secondary, is highly characteristic of Spanish; contrast rubio with Fr. rouge and sabio with Fr. sage. Additional examples include LABIU 'Up' (CL- Lat. LABIA, -ÖRUM) > labio and PLUVIA 'rain' > lluvia (but FOVEA 'pit [as a trap], pitfall' ՝>hoya 'hole, pit'). Cf. Menéndez Pidal, Manual de gramática histórica^ (Madrid, 1941), § 8 bis, Section . 8) In linking Sp. nervio (so) dkectly with Late L. NERVIA, M. Niedermann, "Über einige Quellen unserer Kenntnis des späteren Vulgärlateinischen", NJKA, XXIX (1912), 313-342, esp. 325, has disregarded the rôle of OSp. niervo. 9) Within Hispano-Romance, the generous number of learned -io adjectives, even in semantic spheres where one would hardly search for any, is characteristic of Castilian; contrast Sp. tercio with Ptg. terço, Fr. tiers < TERTIU (a .discrepancy which reminds one of Sp. ֊encia as against Ptg. ֊ença < -ENTIA; see UCPL, 1:4 (1945) and cf. Sp. medio with dial, meo <MEDIU (see Me néndez Pidal, Orígenes del española, § 48:2, and the data assembled in UCPL, XI, 1954,114). Necio was a favorite with Juan Ruiz (16A, 56d, 114c, 159, 193a, 195c, etc.) who, incidentally, preferred the phytonym mestuerço (1544Ժ) to mastuerço.

256

MULTIPLE CAUSATION

ufanía (cf. OSp. loçanía), and yet other rivals.10 Does erratic Sp. soberbio (OSp. sobervio) show contamination of SUPERBUS by the correlated abstract in -IA, or does it mask an older vernacular *sobiervo, metathesized like niervo > nervio! In Italian nèrvo is normally developed, while superbo, by virtue of its -u-, has a learned garb. It is perhaps legitimate to argue that the "ideal" development *sobiervo 'proud' ~ sobervia 'pride' (the abstract gives the impression of a vernacular word, with / j / blocking the diphthongization of Ě, as in premia 'pressure' from OSp. prem-er, -ir)11 would have produced, on the morphophonemic level, an unparalleled pair of adjective and adjectival abstract, and that, to obviate this anomaly, a leveling process ensued, with generous help from OSp. sobervioso 'proud'. 12 (b-) The -i- segment of agrio and amplio has been attributed to contamination with agriar to sour' and ampliar 'to enlarge, broaden', respectively, and the paral lelism of amargo 'bitter', a cross of AMARU and AMĀRICĀRE (Itala) 'to embitter' > Sp. amargar seems, at first blush, to lend support to this hypothesis;13 cf. also AMÄRICÖSU 'full of bitterness' > Sp. amargoso, and note the contrast to Fr. amer (unaccompanied by any verb) and to It. amaro. Indisputably, the dominant OSp. form was agro < Late Lat. ÄCRU (Mulomedicina Chironis), beside Class. ACRE 'sour' (cf. It. acre and Sp. vin-agre, on the assumption that it is indigenous rather than borrowed from Provençal-Catalan); this medieval form, phonologically unimpeachable, neatly matches Ptg. agro.14 It seems best to separate the vicissitudes of amplio from those, no less tangled, of agrio. The comparative AMPLIUS is on record as having frequently furnished an equivalent of ULTRA, PRAETEREĀ, PLŪS and as having, in fact, given rise to the verb (EX)AMPLIARE; Italian, to this day, tolerates ampio 'roomy' < AMPLU (cf. 10) ö n OSp. locano 'arrogant' > 'handsome, satisfied, strong, luxuriant', (f.) 'noble, genteel, stately, portly' and the corresponding abstract loçanîa 'arrogance, beauty, gaiety' ( = OPtg. louçainha 'pride, elegance') see UCPL, 1:7 (1947), 248-257, 260-267, 284-288, with references to the near-synonyms orgull-o, -ία and ufanía. 11) For details see Thesaurus (BICQ, IX (1953-55), 1-138, fnn. 402f. 12) Sobervio is an adjective of old standing (Ruiz, 236a, 238Ò, 241d, 243d, 245ծ, etc.; Bañan e Josafâ, 138v°, 164r°), but sobervioso 'haughty' was also firmly entrenched in the me dieval lexicon (Ruiz, 1665z; Confisión del amante, 24r°, 50v°, 350r°, 356r°), which likewise had a niche for the verb soberviar 'to treat arrogantly' (Ruiz, 819c). The expansion of sobervicaused sobrevienta 'sudden assault, shock' (from sobrevenir) to cast off a variant sobervienta (Apolonio, Ruiz). 13) The modification of a primitive adjective by a verb derived from it, as in agriar ~ agrio, amargar ~ amargo, ampliar ~ amplio, Ptg. baix-ar 'to lower' ~ baix-o (as against Fr. baiss-er ~ bas Հ. Low L. BASSU) is accorded excessive prominence by Corominas, DCE, s.w. The phe nomenon undeniably exists - and has, I might add on my own, occasionally given rise to "Präfix verschleppung", as in Sp. endeble 'weak' beside It. debole < DEBILE (unless the synonym enfermo < INFIRMU deflected it from its normal course) and in Sp. desnu(d)o haked' beside Fr. nu, It. nudo, Ptg. nu (f. nua), from NŪDU contaminated by (D)ĒNŪDĀRE 'to strip bare', but it has seldom been the sole or prime mover. 14) Agro was the dominant form throughout the Middle Ages; cf. Confisión del amante, 278v°; the same text contains agramente, 170r°, and un agro dulce, 265r°and 364v° ( = m o d . agridulce 'bittersweet', lit. 'soursweet'). The adjective was flanked by agr-ura 'sourness' (Esc. Gl.) and agraz 'sour grape(s)' (Esc, Pal., Toi. GL; note A. Castro's comment, p. 155Ű, on the classification proposed by REW). Agrial '[s]cutel[l]a', an isolated entry in the Pal. Gl. ( § 460), pertains to the etymologically - and mythologically - controversial graal, grail family.


257

Sp. Ptg. ancho 'wide, broad') beside rarer amplio 'ample' (of clothes) < AMPLIU. Agriar appears too late in texts (ca. 1730) and has remained a verb of too low in cidence to warrant the assumption that it deflected OSp. agro from its natural course. If we agree to posit the (limited) productivity in Spanish of the adjectival di- or tri-syllabic -io model, agrio may pass off as a mere elaboration onagro, eventually adopted on the strength of its pleasing sound contour, and the crystallization of agriar would, in harmony with the record, represent the last link in the chain. Agrio may, in the process, have drawn much collateral support from the protracted co existence in Old Spanish of vidro 'glass' < UITRU (cf. It. vetro, Fr. verre, Ptg. vidro) and vidrio 'id.' < VITREU 'glassy' (cf. cirio 'candle' < CĒREU 'waxen'); above all, from the gradual withdrawal of the former before the latter.15 (5) The problem of the indebtedness of -io to accent-shift is again so multipronged and, at the same time, so closely enmeshed with other factors as to require, at the start, a number of separate analyses to foreclose the danger of any hasty generalization. The three cases at issue involve the vicissitudes of sandio, zafio, and reacio. (a) The continued use of Ptg. sand-eu, fem. -ia 'fooHsh' (rhyming with jud-eu, -ia 'Jew') and internal metric evidence — particularly rhymes — within older Spanish jointly militate in favor of stressing OSp. sandio on the i. After a period of dormancy the word, one of those quaint voces rústicas eventually consecrated by their emer gence in Don Quixote, has been revived and is today again widely understood among the cultured, if hardly readmitted to active use. The average speaker of Spanish tends to read it sandio (in accord with necio and in semantic contradistinction to sábio; one fringe benefit is the wedge thus driven between sandia 'silly girl' and sandia 'watermelon', of Arabic background). A modern commentator is free to strike a puristic pose and to combat this perfectly natural development, but to heap abuse, as does Corominas, on speakers favoring sandio and to liken the drift toward this richly suggestive variant to the groundless distortion — in the wake of its revival — of maguer 'though' into maguer, is to exhibit an utter lack of perceptiveness.16 (b) Sp. zafio 4incouth' (Nebrixa: çaflo) shares with sandio the distinction of etymological opaqueness; moreover, its medial -ƒ- stamps it as exotic. Arabic ancestry has been suspected for over a century; if Dozy's base GAFI, despite the endorsement it has received from Diez, Cuervo, and Meyer-Lübke, proves unsatis factory, some such elaboration on Eguílaz y Yanguas' rival conjecture as that recent -

15) The dialect map of the Peninsula (glass) shows areas of vedro zafío involves the hypercharacterization of a borrowing as an adjective, much as a change in the oppo site direction (say, rócio > rocío) serves to cut a word's adjectival ties and to trim it as a newfangled substantive. (c) The uninhibited use of adj. -io as a derivational tool is further dramatically illustrated by reacio 'obstinate', which, judging from its medieval prototype refazio, -hazio, clearly stressed on the /, may parallel Fr. restit and involve an -ĪVU > -ίο offshoot from refazer 'to do sth. over and over again'. The total obliteration of ֊h- liúdo (as in viudo 'widower') > ludio. But while metathesis may be said to have been at work, so far as the mere description or preliminary labehng of the process is concerned, incentives for the final shift, reached through slow experimentation no doubt, were, first, the chance it afforded speakers to rid their lexicon of a (to them) irksome diphthong and, second, the opportunity it gave them to expand simultaneously the immensely

17) The existence - possibly ephemeral - of a variant çaf- alongside çafî- follows from Ptg. sáfaro, an unmistakable congener ('inculto, vil, desprezível'). Ptg.sa-, çá-flo may well be a Castilianism, as is demonstrably tíbio; cf. Rom., LXXIV (1952), 145-176. 18) I have given op the alternative possibility of extracting refazio from OSp. (f)az 'face' < FACIE. Semantically there is no reference to 'grimacing', while phonologically the derivation from faz is blocked by the consistent use of -z- in preference to -ҫ-, counter to the record of posfaç-ar, later por- and pro-façar 'to mock', lit. 'to deride (behind one's face = behind one's back)'; see RPh, III (1949-50), 27-72. The inventory of medieval forms includes ra-, re-faziado {General Estória, I, 599U; II, \5\ab and 229b) and rehazlo, commented upon by Juan de Mena in his "Glosa a la Coronación'" (ed. 1548, fol. 6r-b);Iowe this datum to my late wife, María Rosa Lida de Malkiel. Refazío's Gallo-Romance counterpart, Fr. rétif 'restive', is best associated with RESTARE not in its general sense of 'staying, remaining' (E. Gamillscheg, EWFS, s.v.: "mit der Eigenschaft des Stehenbleibens"), but in its military use as synonym of RESISTERE ('to be refractory, rebellious'), see Ernout and Meillet, DÉLL4, s.v. STO (p. 653b). 19) For statistical data see T. Navarro, Estudios de fonología española (Syracuse, N. Y., 1946), 29. On the shift Eu- > U- see Α. Μ. Espinosa and A. Alonso, BDHA, I, 263.

MULTIPLE VERSUS SIMPLE AUSATION

259

appealing adjectival series in -io.20 The other formation, apparently branching off directly from a verb (pender 'to hang, be suspended') in the North of the Peninsula, is pen- or і-dio 'bent, leaning'.21 If agrio, amplio, ludio, pindio, reacio, sandio, soberbio, and zafío entitle one to assume a certain vitality of this model within Spanish, the schema thus authenti cated would raise the number of unstressed suffixes once deemed highly uncharacter istic of Romance. The rhizotonic set '-ago,'-alo, '-ano, '-aro, painstakingly and almost single-handedly reconstructed by Menéndez Pidal, and postverbal ֊e in abstracts (at present far superior in sheer productivity to its erstwhile rivals  and -a) would thus join hands with adj. -io, -ia in disproving or, at least, severely qualifying a rashly formulated accentual Umitation.22 The impression that the old nucleus of -io < -IDU adjectives had a lion's share in this process of consolidation and subsequent expansion is reinforced by the fact that in Portuguese, where -IDU did not nearly so often cast off -io as in the central peninsular dialects, the analogical extension of -io likewise failed to take place. The parallelism is too striking to be shrugged off as fortuitous. Thus one observes, on the one hand, limpo beside Sp. limpio, rijo beside Sp. recio, sujo (OPtg. also cujo) beside Sp. sucio, OPtg. tivo (flanking Gmc. morno) beside Sp. tibio ֊ a congener which eventually invaded the West and dislodged autochthonous tivo -, and turvo beside Sp. turbio; and, on the other, agro, amplo, and soberbo (alongside the ab stracts soberb-a and -ia, the latter slightly Latinizing). Within the same general trend of events, RUBEU very opportunely yielded ruivo 'reddish-blond' (cf. Rum. roib, applied to the coat of a horse) and sandeu remained accentually immobilized. If one cares to fall back on the pattern of thinking which linguistic geography has developed among its practitioners, one can argue that the inorganic atonic ՛iտօ far observed may have sprouted from the 'epenthetic -/-' peculiar, from the earliest records, to Asturo-Leonese dialect speech;23 some adjectives of the kind we are here 20) A. Alonso, "Cambios acentuales", Problemas de dialectología hispanoamericana (Buenos Aires, 1930; also included in BDHA, Vol. I), touches certain points concomitantly relevant to our inquiry: the unstressed suffixes (pp. 47f.), the spread of learned -ido to Cúpido, etc. (pp. 50f.). But, aside from an incidental remark on sandio (p. 52, fn. 1), he seems to have over looked the ramifications of the problem most pertinent to -io ~ ֊io. One side-issue generally neglected is the increase in authenticity that accrues to sandio from the abstract sandez 'folly' (for analysis, see CTL, III, The Hague, 1966, 342): Words in -io may or may not lose the -isegment when further suffixes are attached to them (contrast \a\necedad 'stupidity' and necear 'to talk nonsense' beside necio with [b]reciedumbre 'endurance', seriedad 'seriousness', variedad 'variety' beside recio, serio, and vano, in this order) while those in- invariably preserve that segment (except before i): vaciedad 'folly, nonsense' alongside vacío 'empty'. 21) Křepinský and Garcia de Diego, Inflexion, p. 22, fn. 1. G. A. García-Lomas y GarcíaLomas, Estudio del dialecto popular montañés (San Sebastián, 1922), 272, records the vars. pendio (Pereda, Peñas arriba, Châp. 3) and C.-Sant. péndiu. 22) For references to the relevant researches, chiefly by Menéndez Pidal and, lately, J. Hubschmid, see Hisp. Rev., XIX (1951), 238-263, 323-340; Spec, XXIX (1954), 588-594; and Lang., XXXVIII (1962), 149-185. 23) See R. Lapesa, Asturiano y provenzal en el Fuero de Aviles (Acta Salm., 11:4, 1948), p. 25, for a collection of pertinent verb forms. Menéndez Pidal, "El dialecto leonés", RABM^, XIV (1906), 152, supplies an illustrative list of substantives.

260

MULTIPLE CAUSATION

surveying (e.g. ludio and pindid) indeed give the impression of being mere regionalisms. But while constant fluctuation at dialectal border-lines in the scope granted to -/- may indisputably have furthered the spread of agrio, zafio, etc., such wavering does not of itself explain why -io should have become the distinctive mark of adjec tives rather than of some other form-class or of a random collection of lexical items.24 What matters is that we are here witnessing, excitingly close at hand, the birth of a genuine derivational model so far overlooked by analysts, of a device fit for subtle hypercharacterization. We have thus so far isolated the following causes, ascertainable or highly probable, behind the crystallization of Spanish -io adjectives: perpetuation of Latin proto types in -IDUS, authentic (limpio) or plausibly reconstructed (mustio), or else limited survival, in the vernacular layer, of bases in -EUS, -IUS (rubio), adoption of learned formations now early (necio), now late (serio), influence of the coexistent abstract (soberbio), reverberations of the autonomous use of the Latin comparative (amplio, perhaps via ampliar), extension of wavering between an old substantive and an erstwhile adjective tending to evict it (ve-, vi-dro ~ vidrio→agro ~ agrio), accent shift in a word reintroduced into restricted use (sandio) or in one favored uninterruptedly (reacio), restructuring of borrowings from exotic languages (zafío), infiltration of specimens of such dialect speech as abounds in epenthetic /j/ (pendio), free-wheeling use of -io coincident with the elimination of undesirable diphthongs through metathesis (Extr. ludio). Each individual explanation, in some instances adduced to justify the derivation of a single adjective, seems defensible; in all likeli hood none is powerful enough to have allowed an observer to predict, with assuran ce, this particular course of events. Such a tangled situation gives rise to a number of closely connected questions. If half-way satisfactory explanations culled from the existing inventory of processses are readily available for all or most of the formations, is it advisable to posit the productivity of a new derivational model such as radical-stressed adj. -io! In the event that we accept this model as a contributing force in the dynamics of the lan guage under scrutiny, do the other explanations fall by the wayside, or can two (or more) causes be recognized as mutually complementary and coefficient? Is it possible to establish certain criteria for such causal complementarity and for its opposite, mutual exclusiveness? Does not the assumption of multiple causation, especially if made on a liberal scale, clash with the principle of maximum economy as a yardstick of cogent scientific analysis, to the extent that such parsimoniousness is compatible with expository accuracy and with the complexity of the facts analyzed? In seeking to provide answers, one must draw a line between mere description and causal explanation. When we state that liudo was locally transformed into ludio and list this process under the rubric 'metathesis', we simply attach a conve nient tag to a shift which, within that dialect, happens to be unpredictable, even in 24) Unquestionably there remains in use a nucleus of very old adjectives in -ίο albedrίο 'frec will, caprice, precedent', echoing señorío 'dominion, sway, mastery, lordliness, gentry'.


261

probabilistic terms; not for nothing did Ascoli subsume metathesis under a bundle of seemingly capricious sound changes which he eloquently labeled 'general acci dents'. Our second step is to remind ourselves that, even where probabilistic statements can be ventured, such assessments as involve a low degree of predictability offer, in themselves, an invitation to researchers to seek for additional determinants. As suming that a given historical situation and its peculiar social setting (with special reference to a protracted state of bilingualism) favor the surmise of substratum influence, surely, even the most enthusiastic supporter of such hypotheses will grant that only selected features of sound structure, lexicon, etc. are bound to filter through. It is clearly plausible that certain asymmetries, lacunae, or ambiguities, in short, flaws in the architecture of the receiving language, control the extent of assimilation of the ingredients adopted from the donor language. This mutual dependence leads to a sobering realization: Though, for temperamental reasons, scholars may be divided into those preferring to operate with external pressures and those hypothesizing internal pressures to the limit of their imaginative power, the actual changes in language, in numerous instances, presuppose interplays of inner and outer forces. Our third thought may be directed toward quantification. If, for the sake of simplicity, we argue that the crystallization of all -io formations except those devel oped from Latin through normal channels involves the convergence of two forces: on the one hand, the agency of an apparently productive derivational model not yet fully identified (Force A) and, on the other, some collateral pressure familiar from earlier inquiries (Force B), we observe that, in the slice of material examined, Force  varies from case to case: Bı (accent shift), B2 (adaptation of a borrowing), B3 (contamination by a related verb), etc., while Force A remains pervasively at work and actually ties together all these minor heterogeneous modifications. The positing of a heretofore unsuspected force ֊ e.g., of an expanding derivational type ֊ gains in plausibility if that force is each time paired off, in a series of complex processes related to one another, with a companion force of distinctly smaller scope. Under these conditions, the newly introduced Force A brackets and underpins a structure that would otherwise suffer from amorphousness and tend to crumble. The final consideration in weighing the validity of a new factor of causation is whether its agency can be observed, at least occasionally, in isolation. We have seen that for agno, amplio, reacio, zafio, etc. at least one other driving force can be tentatively identified, with widely varying degrees of accuracy. The tidy isolation of even one clear-cut case would add immeasurably to the credibility of any broader hypothesis positing involvement of a new force as a separate ingredient in multiple or complex causation.

II. THREE SUPPORTING EXAMPLES Let me now weigh the advantages accruing from the assumption of complex causa tion to the solution of certain problems examined elsewhere in considerable detail, but not yet definitively settled.

262

MULTIPLE CAUSATION . Alternation of [δ] and [r] in Hispano-Romance

Of the three cursory discussions chosen for this supplementary purpose, the first, a classic case of irreconcilable disagreement between two equally distinguished scholars, revolves around the alternation of [δ] and [r] in Hispano-Romance, [δ] being locally an allophonic member of the phoneme /d/. Of the two disputants, the Romance comparatist E. G. Wahlgren brought to the arena a far better knowledge of cognate languages, his critic T. Navarro a more intimate glimpse of fluid Peninsular dialect usage - a discrepancy in training and exploratory flair which makes their clash of opinions, over thirty years ago, all the more dramatic.25 Wahlgren was by no means a scholar one-sidedly committed to a favorite idea (or to warring against strawmen and bêtes noires), and readily granted that the transformation of a d into an r could (a) in some languages (say, in Italian and in Provençal) represent regular sound shift, a kind of 'rhotacism', while (b) it was else where, e.g. in Spanish, set in motion by a subtle interplay of (a) lexical or affixal analogies and (β) saltatory (sporadic) sound changes; but he apparently did not care to reckon with any intermingling or alliance of these two major groups of ingre dients. Thus, in dealing with Spanish words (predominantly of dialectal stock), he appealed to lexical contamination, as with badajo 'clapper (of a bell)', changed into barajo allegedly through contact with baraja 'pack of playing cards', -ar 'to shuffle';26 to the intrusion of a derivational suffix as in (a) Ast. antroiru 'carnival' instead of expected antroidu < INTROITU, through pressure of -oiru < -ÕRIU, or in (b) dial, párparo 'eyelid' < párpado, stand, lámpara 'lamp' < LAMPADA, through association with unstressed '-aro, '-ara; to the interference of a coexistent infinitive, as in seguirilla beside older seguidilla 'form of stanza' (also 'special air and dance'), and mentira 'lie' beside Cat. mentida 'id.', through contamination with seguir 'to follow' and mentir 'to lie, be false', respectively; to association with a pre fix: dial, armiro < admiro and armitian < admitían;21 and to miscellaneous distur-

25) E. G. Wahlgren, Un problème de phonétique romane: le développement "d" > "r" ( = Skrifter utg. av K. Humanistiska vetenskapssamfundet i Uppsala, XXVI:4) (1930). T. Na varro's critique of this 250-page monograph appeared the following year in RFE, XVIII, 393-395. For a succinct reappraisal of the material see my article "Estudios de léxico pastoril: piara y manada", BH, LIII (1951), 41-80, esp. 61-66; for a subsequent elaboration on the special prob lem of mentira, which I regard as patterned on OSp. vera(s) 'truth' < VĒRA 'true things' and ju-, yu-ra 'oath' < IURA lit. 'rights', see "Ancient Hispanic verais) and mentirais): A Study in Lexical Polarization", RPh, VI (1952-53), 121-72. 26) Additional equations: ataure < ata-ul, -ule < ataúd 'coffin' X baúl 'trunk'; coriao Sp. reja 'grate'), one must not ascribe this complex reshuffling solely to a recoil from defect ive marking ofthegender: Affixation and inflection may also have shared heavily in the process. Clearly, in VETULUS the newly awakened fondness for diminutives was likewise operative, while in *VETERU one witnesses the concomitant intrusion of a rival declensional paradigm; compare PAUPER, -ERIS 'poor' with MĪSER, -Ī 'wretched' (or TENER, -Ī 'tender') and observe the contrast between Sp.Ptg.pobre (., f.) and It. pòver-o, -a ֊and the parallel contrast between Sp. Ptg. alegre (., f.) and It. allegr-o, -a, the common starting-point being AL-ACRE, *-ICRE 'excited, lively'. The analyst can fruitfully apply this general approach to one close-knit sector of the Latin lexicon viewed in its transition to Romance, namely the zoonyms.29 In this domain, the marking of the animal's sex through gender was initially kept at the barest minimum, except — understandably enough — in the case of certain domestic animals, where the male and the female of the species provide radically different services (BÕS vs. VACCA, ASIN-US and ASELL-US vs. -A, CAPER vs. CAPRA, etc.). At the outset, LEO 'lion' and LUPUS 'wolf were, consequently, ambigeneric, with MASCULUS and FEMINA added at rare intervals for incidental emphasis on the sex, much as are bull {buck) and cow, cock and hen, he- and she՜ in English, or, for that matter, -bock, -kuh, and -kalb in German. So deeply rooted was this usage that, when the Romans corned LUPA, it first denoted only the 'harlot'; chronologically, then, its figurative use took precedence over its basic one. Gradual ly, the scope of LUPA was widened to include the meaning 'she-wolf; also, LEA 'lioness' came into existence to contrast with LEO (which, by the same token, was referentially narrowed down). Before long there developed a vogue for consistently separate designations of at least those male and female animals whose behavior lent itself to ready observation. Now, it is noteworthy that wherever lexical innovations proved successful in Romance zoonymy, regardless of the specific justification for each replacement, 29) For details I refer to my above-mentioned article, "Diachronie Hypercharacterization in Romance", esp. to pp. 5-16, devoted almost exclusively to names of animals.


265

one recurrent fringe benefit was the sharper delineation of the gender (and sex). Thus, for ambigeneric CAN-ĒS or -IS 'dog' we find in Spanish an exotic substitute, perro, flanked by perra 'bitch'; in near-by Portuguese cão < CANE has survived in close association with cadela 'bitch' < CAT-ULA, -ELLA, orig. 'female puppy'; CATULUS, not necessarily cognate to CANIS, was eventually drawn into the latter's orbit. French offers a third solution of the problem, displaying chienne alongside chien, much as, in a Frenchman's world-view, the lionne shares the lair of the lion. The lexical and grammatical vicissitudes of the 'cat' were somewhat different: The original designation FĒLĒS was feminine, like VOLPĒS 'fox' and MĒLĒS 'marten or badger', and the reference was initially quite vague as regards both sex and species; conceivably hints of 'wild cat' and 'weasel' were both included. The introduction of CATTUS (GATTUS), a word first attested in the 4th century, in all likelihood marked three events: the adoption of a new label, the importation of a hitherto unknown species, and the acceptance of a new social institution (domestication of the mouse-chasing pet); but - even more important to us - it coincidentally provided a welcome means whereby speakers could instantly tell a male from a female, a discrimination for which there apparently arose a demand even outside Romance (cf. G. Kater ~ Katze, R. kot ~ koška). Lat. LEPUS, -ORIS 'hare' was ambigeneric, a state of affairs preserved in Italian; the congener languages have devised various ways and means for neater sexual differentiation, cf. Rum. iepure (m.) ~ iepurică (f.), Fr. lièvre (m.) ~ hase (f.) < Gmc, Sp. lebr-ón (m.) ~ liebre (f.). One can almost indefinitely increase the number of illustrative examples; every single one of these elaborative substitutes may be amply justified, I repeat, in terms of such phenomena as lexical borrowing, orchestrational attachment of augmentative or diminutive suffixes, analogical innovation, etc. But if the student of causation succeeds in isolating one consistently emergent by-product of all these multifarious processes — namely the more explicit marking of gender and sex — he is entitled to posit for the agency effecting the change the status of an autonomous co-determi nant, because this force ties together untold developments which would otherwise appear disparate, meagerly exemplified, and almost random. The pointed structuring of change need not to be confined to phonemics, where it was first traced out with such elegance. . Consonantal Dissimilation Had a more generous allotment of space been at my disposal, the pièce de résistance of the discussion might well have been consonant dissimilation at a distance, with especial reference to Romance, a notorious crux of explicative linguistics. If one disregards A. F. Pott's inconclusive pioneering essay (1833) geared to Indo-European, a probing which stamped the phenomenon as capricious and sporadic, the event that can best serve as the starting point for the entire debate is the publication of M. Grammont's doctoral thesis (1895). That monograph involved an attempt to elevate dissimilation so delimited to the rank of a sound law, or rather of a set of stringently formulated sound laws, cross-temporal and supposedly binding on all speech communities. Grammont's thesis, and a major complementary article (1907) from his pen, provoked a flurry of formal reviews and lively "prises de position"

266

MULTIPLE CAUSATION

at a juncture when Neogrammarian theory and practice were approaching their all-time crest. Among the participants,30 A. Meillet wholeheartedly approved the kernel of the book, skipping details; C. Salvioni evinced skepticism regarding Gram mont's rigid 'laws'; G. Paris spiced his plaudits with an ironic remark on the intricacy of these laws; M. Niedermann gained from the author's treatment the distinct impres sion that a common sound law was at issue; a realistic A. Thomas observed in action such lexical complications as seemed to make havoc of Grammont's severely phrased laws; K. Brugmann liberally assessed the scope of dissimilation and invoked psychol ogical motives (specifically, the "horror aequi") which obviate the need for any regularity of recurrence; in contrast, H. Schuchardt resolved the problematic changes arrayed by Grammont into a string of etymological biographies. E. Schopf, in his study of "Fernwirkungen", endorsed Grammont's doctrine in theory, without quite so zestfully applying it in practice. W. Meyer-Lübke sought to reconcile the acknowledgment of dissimilation as a major shaping force in evolution with de mands for individual attention to lexical details. L. Gauchat, likening the saltatory change in question to mutations familiar to the botanist, warned against indiscrimi nate use of the facile label ("moyen commode d'explication"). Fresh vigor was injected into the debate by Rebecca R. Posner's stimulating Oxford dissertation, Consonantal Dissimilation in the Romance Languages (1961), a book which had the merit of eliciting a bold counter-proposal from K. Togeby only two years later. Like Grammont, Posner believes in the survival of the fittest as the overriding dissimilatory principle; but she skillfully nuances her predecessor's teach ings by taking into account phonemics, paying attention to the frequency (incidence) of sounds, laying down more carefully distilled conditions for regressive effects, discriminating between the central and the peripheral parts of the lexicon (the latter include technical terms and the onomasticon), and distinguishing, especially as regards the sequence of two r's, between individual languages.31 Identifying himself, at least in part (pp. 647,652), with the thinking of such earli er Danish linguists as C. Møller, J. Byskov, and K. Sandfeld, and amplifying on classic pronouncements by Schuchardt, Gillie'ron, and Meyer-Lübke (p. 651), Togeby offers the readers of his ambitious critique of Posner's monograph a kind of sparklingly phrased counter-theory. The kernel of his proposal, brought to bear on Latin and Romance material, may be epitomized thus: (a) Consonant dissimilation operates either (a) with the precision of  genuine 'sound law' (e.g., k w - k w , preventively in the distribution of such suffixal doublets as Lat. -ĀLIS/-ĀRIS, -LUM/-RUM , Fr. -euilj-eul, -erer/-eler, pp. 648f.), or (β) sporad ically, a situation which forces the analyst to use the term, at most, as a descriptive 30) For a rather full and very perceptive bibliographic survey see K. Togeby, "Qu'est-ce que la dissimilation?", RPh, XVII (1963-64), 642-667; cf. the following comments and qualifica tions. One could add to Togeby's roster M. Grammont's third and final pronouncement in Traité de phonétique^ (Paris, 1950; orig. ed., 1933), "La dissimilation" (269-337). For dissimila tion in contact Grammont reserves the separate label "differentiation". 31) In a similar vein, the late Alf Sommerfelt, in a possibly unpublished lecture given a few years ago at Stanford's Center for Advanced Studies in Behavioral Sciences, used samples of Celtic and Norwegian material to demonstrate that the agency of 'saltatory shifts' is, in every language, intimately interwoven with the specifics of regular phonological change.


267

tag ("une étiquette sans valeur") and in fact invites its elimination from any causal argument (p. 650). (b) Inquiries into dissimilation revolve around three fundamental problems: (a) the cause, (β) the direction, (7) the outcome of each shift. On the first of these problems, we learn, there prevails a good deal of confusion; the second was attacked resolutely, but less than successfully, by Grammont, whose twenty 'laws', despite their sophistication, are vulnerable to criticism; on the third Posner has cast a ray of light. The hard core of Togeby's own reflections is to be found in his repeated insistence on the dispensability of any assumption of a consonant's dissimilatory change or disappearance if some other explanation, equally plausible, can be offered.32 An impressive string of such preferable rival explanations is next displayed in con siderable detail (pp. 653-664), the possibilities varying from a medieval scribe's lapsus calami and an early etymologist's faulty base to lexical borrowing, hyperurbanism, onomatopoeia and expressivity, crosses within the same word family, blends with some other word family, folk etymology, suffix or prefix change, ag glutination or deglutination of the article, and the . What adds a touch of drama to this discussion, aside from the author's general mastery of linguistic grand strategy and from his exemplary pan-Romanic versatility, is to see a scholar reared in the tradition of radical structuralism and, to this day, steeped in it, side, for once, with such champions of the opposite extreme as Schuchardt and Gilliéron. Despite my great admiration for Togeby's spirited advocacy of a deflated assess ment of consonant dissimilation, I cannot entirely subscribe to his views insofar as they concern the theory of causation. Preferring at this point to refrain from any formal commitment regarding a matter which has proved so vexatious to some of our discipHne's keenest minds,33 let me provisionally state this: From the fact that a phenomenon, while definitely identifiable, is seldom found in isolation I would infer that this isolable phenomenon usually operates in conjunction with others. Once the common occurrence of multiple causation is granted axiomatically, there will be no need to proceed to radical extirpation of any suspected redundancy. To cite concrete instances, Togeby is at his most persuasive where he corrects the geographic trajectories of certain words traced hazily by Posner, as when he argues (p. 654) that it is not Fr. caramel which must be confronted with  ALAM EL LU, but the French word's immediate prototype, Sp. caramel 'lollypop'. Similarly,

32) "Car on ne peut pas se contenter de la formule paradoxale de R. Posner qui veut parler à la fois de cause et de hasard. S'il faut chercher ailleurs la cause de ces changements, l'explica tion par la dissimilation devient superflue" (p. 651); "S'il faut chercher pour chaque mot une explication particulière du résultat de la 'dissimilation', cette dernière cause pourrait, à elle seule, expliquer la dissimilation, qui de nouveau devient, en tant que telle, superflue" (ibid.)՛, "Si le changement s'explique dans ces cas sans une action dissimilatrice, celle-ci devient illusoire" (p. 652); "Même les mots réfractaires à une autre explication ne peuvent prouver l'existence d'une influence vraiment dissimilatrice, car à côté d'eux on trouvera très souvent d'autres mots présentant le même changement de phonème sans qu'il y ait dans le voisinage un phonème identique qui aurait pu exercer son influence" (p. 665). 33) Notoriously weak and hesitant is Լ. Bloomfield's treatment of sporadic sound change in his book Language (1933); see Chap. 21, § 10.

268

MULTIPLE CAUSATION

Fr. (dial.) coronel 'colonel' reflects, in its central pillar, the corresponding phoneme of its trans-Pyrenean cognate; It. gonfalone and Sp. gonfalón '(ancient) banner' prove, upon closer inspection, to be mere Gallicisms (Fr. gonfa-lon < OFr. -non), etc. In other cases, possibly more numerous, Togeby's stimulating alternative hypoth eses lead to a richer orchestration of the earlier analyses rather than to their complete abandonment. Thus, in semi- or, better, almost-learned Sp. plegaria 'prayer' < PRE CARIA (p. 655) the dissimilation, invoked by Grammont, seems to be the decisive force behind the withdrawal of the first r in favor of /, but plegar 'to fold' < PLICĀRE may have been, on the semantic side, a contributing factor (by suggesting genuflexion). Furthermore, the switch in Spanish from an older, rustic form pregar 'to fold' (preserved in Portuguese with a more advanced scale of meanings: 'to nail, fasten, fix') to the socially more elegant variant plegar created an ideal setting for the simultaneous replacement of pregaria by plegaria. Our attention, then, is drawn to an ensemble of circumstances, by no means mutually exclusive: (1) a latent dissimilatory trend, (2) a semantic or imagerial bridge, (3) false regression, i.e., a typically so cio-linguistic reaction. Each single ingredient of this amalgam is isolable and can be charged with the full burden of responsibility for the crystallization of plegaria, but it is the joint momentum that most satisfactorily explains the out come. In a way, the aggregate of causes relates to a single cause as does a phoneme to one of its distinctive features, except that it is uncertain whether the subtraction of one cause would radically alter the total picture.34 To conclude: The prevalent notion of unicausality needs thorough revision much as did until recently the long-accepted belief in dominant monolingualism (or, for that matter, as did the earlier typological classification of languages on the basis of a single structural characteristic). By starting out with the expectation of over whelmingly plausible pluricausality we stand a chance of reaping two major bene fits: In terms of improved cognition, we shall do fuller justice to the complexities of reality, and in terms of academic tone and scholarly climate, we shall eschew that stridency of debate which, in the case of strict alternatives, the stern demand for a partisan choice has invariably carried with it, to the detriment of a serene and balanced appraisal. To supply just one example: The bitter controversy surrounding the postulated regularity ("Ausnahmslosigkeit") of sound change might have lost much of its acerbity and dogmatism if heavier emphasis had from' the outset been placed on the fact that ordinarily the 'sound law' describes but one ingredient (say, x) in the total situation at hand (say,  + y + z). If this is so, the law 'works' when y and z each equal zero, i.e., when there occurs no disturbance. It remains to be seen whether, on balance, the principles of linguistic etiology lend themselves to encoding in a rigid sequence of statements or formulas, approx imately along the tines selected for a different purpose by the late L. Hjelmslev in his Prolegomena and by some of the younger reformers of description, or whether a looser, discursive presentation, of the kind favored in historical research, is to be preferred as more germane to the analysis of events. 34) An analysis in miniature form of the four problems here outlined enters into my presiden tial address "Linguistics as a Genetic Science", to appear in Language.

THE FIVE SOURCES OF EPENTHETIC /J/ IN WESTERN HISPANO-ROMANCE: A Study in Multiple Causation At the fresh grave of Ramón Menéndez Fidai Over sixty years ago Menéndez Pidal for the first time clearly identified and described a parasitic or epenthetic serniconsonant / j / appearing in Asturo-Leonese dialect speech (and in related varieties of Hispano-Romance), typically wedged in between radical and final vowel, or radical and suffix, thus: muriu 'wall' for Sp. muro. The discoverer of the phenomenon supplied no genetic explanation on that occasion. An attempt is made now to account for the intercalary element present in some dialects and absent from others through the assumption of multiple causa tion, i.e., of interplays of forces varying from one area to another. Specific agencies thus appealed to include: (a) the flux between learned and vernacular transmission of suffixes; (b) hypercharacterization of gender; (c) local convergence of -ear and -iar verbs; (d) reverberations of the Hispano-Romance vicissitudes of Latin Consonant + ; and (e) certain special features of the conjugational paradigm. [Y. M.] One of the hallmarks of Menéndez Pidaľs incomparable scholarship was his ability, in some contexts, to carry through an elaborate investigation to its success ful conclusion and in others merely to sketch or adumbrate a problem, sometimes non-committally as to the underlying causes, but in any event so deftly that other scholars would confidently revert to it, if need be after a major lapse of time, and would take up the thread of the inquiry at the point where the master had left off. At relatively early stages of his career Don Ramón thus succeeded in isolating the so-called "unstressed suffixes" of Hispano-Romance (1905), a very elusive category of morphemes — if they deserve that label — whose full intricacy is gradually dawnint upon younger workers only now, more than sixty years later; and in briefly positing two consecutive redactions of Juan Ruiz's collection of poems (1901), a hypothesis which has at no time been debated more passionately and with a heavier arsenal of erudite weaponry than at the present moment. During the same pioneering period of his long and ceaselessly active life, the founder of our studies, on the occa sion of his trail-blazing monograph on Leonese, 1 drew the attention of fellowworkers to a peculiar phenomenon characteristic of the entire Northwest of the Iberian Peninsula rather than of Asturo-Leonese dialects alone: the insertion of a less 1) "El dialecto leonés," RABM, 3d Ser., X (1906), 128-172, 294-311 at 150, 152, and 301 (the three key-passages are not as closely integrated through cross-references as one might have wished). This monograph was re-issued as a separate book, with a preface, notes, and a supple ment (ed. Carmen Bobes, Oviedo, 1962); see pp. 50, 53, lOOf. For abbreviations see p. 274. Acevedo refers to B. Acevedo y Huelves and M. Fernández y Fernández, Vocabulario del bable de Occidente (Madrid, 1932); Rato to A. de Rato y Hevia [Rato de Argüellesl Vocabulario de palabras y frases bables (Madrid, 1891-92).

270

MULTIPLE CAUSATION

than clearly conditioned and never before tidily described, let alone analyzed,/j/ 2 say,  'wall' in lieu of muru and quiciás 'perhaps' in preference to quizás. The three short but pithy passages in question have been often referred to or epitomized, especially in the spate of dialect monographs issued in quick succession, in Madrid and in a few provincial research centers, after the mid-'forties.3 No delib erate attempt, however, seems so far to have been made to supply a framework for delimiting the scope of the process at issue and, above all, for establishing a cogent system of cause-and-effect sequences. The paper here offered aims at no such truly exhaustive treatment as the problem, in all its ramifications, richly deserves and invites: Using a modicum and, in some instances, a bare minimum of documentation, I have endeavored to limit my goal to the identification of those forces which must jointly have set in motion the lexical mass surveyed, practically for the first time sixty-three years ago, by that keen observer of living speech and that peerless archivist alike, D. Ramón Menéndez Pidal.

2) Note the silence of A. Morel-Fatio, "Recherches sur le texte et les sources du Libro de Alexandre,'՝ Rom., IV (1875), 7-90, at 30-32, who followed closely the line of E. Gessner's trailblazing pamphlet Das Leonesische; ein Beitrag zum Altspanischen (Berlin, 1867). True, MorelFatio's text (MS  of the Alexandre) shuns the crassest examples of epenthetic / j / ; thus, one finds enton(ça), eston(ças), estonce, mentre rather than the expanded variants in -ia(s). But alabancia, demonstrártela, matančia, segurancia all occur beside creença, fallença, ganança, and repentença՛, so áo for cia, (es) fordar, esforcio beside f o rça, esforçar, esfuerço, and faciana beside façanna 'deed' - witnesses to protracted wavering which Gessner, p. 19, had interpreted as evi dence of the surrender of Old Leonese now to Portuguese, now to Spanish influences. W. MeyerLiibke was more concerned with semiconsonant metathesis and kindred phenomena than with epenthesis of / j / ; see his Romanische Lautlehre ( — GRS, I; Leipzig, 1890), § § 295 and 385. P. de Mugica, Dialectos castellanos: montañés, vizcaíno aragonés (Berlín, 1892), scattered many noteworthy examples all over Part I of his booklet: cancia 'hacia' (pp. 4, 21); -encía for -iencia, ripugnancia, cubicia 'codicia' (p. 10); relambio 'relamido' (p. 12); enquinia 'inquina' (p. 23); the exclamation ¡coreia/ (p. 28); fanfarria (p. 30); juriaco 'hole' beside juriacar 'to bore', lombio, Ifljumiaco (p. 31); pación 'pasto' (p. 33); rustrio (p. 35); sur bia 'poison' (p. 36); zurriascar (p. 38), but drew no forceful conclusion from his own material. 3) For one participant's voice in the discussion, see M. J. Canellada, El bable de Cabranes {RFE, Suppi. XXXI; Madrid, 1944), p. 13. The subdialect in question has been weaned away from the Eastern group (around Villaviciosa) under the pressure of Oviedo. Canellada also offers stray data on -iar Fr. chance →Sp. chanza), and folgancia 'ease, leisure, idleness'. García Rey confirms alabancia and mudancia for El Bierzo, and adds adivinancia. To this list I can add from Central Asturias . recordancia 'memory, remembrance' as against C. umedanza 'humedad'. The general lingering of a descendant of -ANTIA beyond the limits of tolerance set by Castilian is an archaic and parochial trait, the more progressive (or, at least, more successful) verbal and adjectival abstracts being represented by Sp. comparación, hum-edad, recuerdo, segur-idad. The specific prevalence of -ancia is a matter of morphological leveUng and analogy, with socio-cultural overtones. The residual vestiges of -anza are more likely to represent stray intruders from Northern Castile than isolated local pockets of resistance. 10 The Ast. -encía derivatives pose fewer problems and show little divergence'from the course taken by Castilian. C. pa ce neia as against Sp. paciencia involves a (preventive or post-factum) dissimilatory elimination of /je/ . . . /ja/, with numerous parallels in older Spanish, ironically entailing one famous case of over-reaction (apar-iencia 'appearance' in lieu of puristically more justified -encia). If C. sencia 'common sense, judgment' reflects, as I suspect is true, Sp. ciencia < SCIENTIA, then it arose at the intersection of two dissimilatory lines: /Ө/ . . . /Ө/ and /je/ . . . /ja/. Mir. oufensia Offense', clashing with Sp. ofensa, caught Menéndez Pidal's attention in 1906; it may owe its / j / to the combined pressure of -encía and of other abstracts in -ia, typically Latinisms (cf. Sp. industria, molestia, etc.). . -, xi-rigoncia 'silly remark', as against Rato's xerigonza (cf. Sp. ¡er- 'jargon, gibberish'), shows affinity to both -ancia and -encía, as is exemplified by the striking proximity (in terms of shape and meaning) of the phrase  facer xigomencies ( = fey ures) 'to give winks, make grimaces or ugly faces'. Presumably  xuncia 'dry stalk used for stringing onions, maize cobs, etc.' < IUNCEA has been drawn into the process; cf. Sp. juncia 'sedge'. A parallel sideline is exemplified by M. granc-ia, -ea for Sp. granza (bot.) 'madder'; (pl.) 'siftings, chaff, refuse' and jarancios 'thick bramble-bushes' (cf. Sp. ¡aral). While the case of -ancia is transparent on account of the gratifyingly small number of factors involved, certain other, more complex instances of epenthetic / j / belong under the same rubric. Take -ACEU and -ICEU, which normally yielded -aço, -iço in the West and -azo, -izo in the Center of the Peninsula. The parallel learned transmis sion as ֊dceo and -iceo — at present familiar from modern scientific terminology — was very spottily represented in the Middle Ages; but through the alliance with semi-learned reflexes of -ÄTIÕ (nom. sg.), -ÏTIÕ (nom. sg. and abi. sg.), and -ĪTIA (nom. sg. and pl.) — an alliance made possible by the tendential convergence of intervocalic /kj/ and /tj/ in spoken Late Latin — Asturian and near-by dialects in the end did develop whole series of words in -aciu, -acia, -iciu, and -icia, cf. . pigaciu 10) This statement must be understood against the general background of dialect mixturein Asturian. On the state of affairs at Sistema see M. May [Vihman], "Preservation, Innovation, and Borrowing in a Newly Described Subdialect of Asturo-Leonese," RPh, XIX (1965-66), 48-58, a reassessment of the data crudely assembled by J. A. Fernandez (1960), and the earlier analysis by M. Menéndez García, "Cruce de dialectos en el habla de Sistema (Asturias)," RDTP, VI (1950), 355-402. I incline to associate with the spread of -ancia the transmutation of percançar

THE FIVE SOURCES OF EPENTHETIC /J/

275

'nap, light sleep' (= W.-Ast. pig-arzo, -azu); C. concias 'rustic sandals' (apparently made of leather); C. ñiciu 'seed prepared for sowing; trace, mark' beside (a)ñiciar 'to thrive', añiciá-se de 'to have something for the first time' (related to INITIU 1 1 ); C. oriciu 'hedgehog', 'thistle, burr' (= Sp. erizo); Cae. (Casas de MüTán) palicia 'beat ing' [with a stick] (RMP); cf. . rebiciar 'to recover from, bounce back after, a spell of sickness' (with a reference to Acevedo and Rato), akin to VITIUM in the sense of 'exuberance, vitality'. One is tempted to place here M. estrump-icio 'horrendous noise' = Sal. -ido; the starting point may well have been Sp. bullicio 'bustle, tumult, uproar', which descends in a straight line from BULLİTIÕ 'bubbling, ebullition'; cf. rebullicio 'great clamor'. Contrast these regionalisms with Sp. corizas and erizo. The insertion of / j / has not, however, become consistently obligatory in the area at issue; one still finds in C. tinaces 'pair of wooden pincers serving to extract chestnuts from their husks', also erbizu 'smell of grass', carnizu 'smell or taste of meat', llechizu 'flavor of milk', mortizu 'stench emitted by a corpse', podrizu 'smell of rotting' (alongside adj. podre 'rotten' and a cluster of rival offshoots 1 2 ), and seizu 'stench from leftover food, smell of tallow' (beside seu = Sp. sebo); see Canellada, p. 28. As happens not infrequently in dialects tolerant of weakly differ entiated variants, small subgroups begin to emerge — semantically controlled. Thus, it would seem that, at least in C , speakers favor -izu over -iciu for names of odors (see above); also, in humorous designations of 'blows, hits, jerks', and the like -azu has barred -aciu from infiltrating: carnerazu 'butt with one's head' (in ram-like fash ion), cebellazu 'blow with a rope made of osier-twigs' (cibiella), sofronazu 'crude answer, public display of contempt' (from sofrenar lit. 'to curb, restrain'; cf. Sp. sofrenada 'sudden check, or reining-in; severe reprimand'), xigazu 'blow, hit with a stone' (cf. Sp. chinazo, B. chinarrazo 'blow with a pebble'), etc. Since -azu so used has been independently shown to constitute a post-medieval innovation, 13 one cannot exclude the possibility that a Castilian form has been grafted onto the Asturian lexicon. Finally, in the transmission of verbal abstracts in -TIÕNE and -SIÕNE Old Spanish again distinguishes between two levels: One can set off -zón (as in armazón 'frame work', segazón 'harvest season, reaping', dial, nevazón 'snowfall') from commoner -cion (as in untold abstracts), and -sión (as in OSp. confisión 'confession' [ > mod. -festón], prisión 'seizure, jail') from -són, which may suggest a Gallo-Romance detour (mesón 'inn, hostelry' < MANSIONE 'abode'). Asturian seems to go at least as far as Castilian (which in this respect contrasts with Galícián-Portuguese) in its accep tance of ֊ción and -sión, but exceeds the standard in endowing words so derived with a concrete sense: C. pación 'green grass spread over the meadow or recently

(cf. Sp. percance 'mischance, misfortune', 'perquisite', related to Arajtfcized alcançar 'to reach [in hot pursuit]', alcance 'reach' < *INCALCIĀRE) into C. percanciar 'to get along, be put into effect'. 11) On this word-family's connection with INITIU see H. Lausberg's delayed review - brief but pithy ֊ of Canellada's dissertation in RF, LXIX (1953), 45Of. 12) The conspicuous form podren surely involves-AGINE, cf. OPtg. tristen 'spell of sadness', and the like, see Lg., XIX (1943), 256-258. Less opaque, since they harness the common suffix -ICĀRE, are podrigañá-se 'to rot' and podrigañu 'putrefaction' (in reference to trees, fruits, etc.). 13) On the different strands loosely woven together in -azo see my article "The Two Sources of the Hispanic Suffix -azo, -aço,"Lg., XXXV (1959), 193-258.

276

MULTIPLE CAUSATION

mowed' (also Acevedo, Rato); C. procisión 'procession' (with a metaphony shared by older Spanish; cf. Ptg. procissão, . complisión = Sp. complexión); . torción = torzón 'colic' and retorcían 'twist, sprain, dislocation'; prisión 'chain or collar used to tie calves to the crib'. From Extremadura I can add M. escurrición beside the verb escurriciar, recorded by Zamora Vicente in his grammatical prospectus ( § 19). Old Portuguese here represented the opposite end of the spectrum (confissom < CONFESSIONE), acquiescing even in -jom < -SIÕNE and -xom < -SSIÕNE, cf. (o)cajom < OCCASIONE, (com)paixom < ( C O M ) P A S S I Õ N E . 1 4 The idiosyncrasy the cases here considered {-ancia, -aciu¡-iciu, -ción/sión) have in common is the fact that the insertion of / j / is anything but phonological in nature. Fundamentally the process involves the rivalry of suffix variants, whose splitting was originally provoked by socio-cultural factors (diverse channels of transmission caused by different degrees of literacy and of exposure to Latin in medieval Europe). In a new geographic and linguistic context, however, there may very well have developed, on the local scene, a secondary affinity between this particular process and others, truly phonological in nature, which likewise happened to revolve around / j / in the given position. In other words, the rise of -anza beside -ancia, etc., is one thing; the local prevalence oí՝-ancia over -anza, or vice versa, is quite another. III. SOURCE B: EPENTHESIS AND HYPERCHARACTERIZATION OF GENDER While it is marginally possible for an Asturian noun in -e to absorb an epenthetic / j / ֊ witness (C.) rebitie < Sp. ribete 'edge, trimming' beside rebitiar (= W.-Ast. rebintiar, Sp. ribetear) 'to trim' - the typical development, especially if an oldstock feminine is involved or if the trend is toward feminization of a weakly marked masculine, is for the -e to be replaced by -ia. In this fashion, locally pleasing epenthe sis of / j / and hypercharacterization of gender, transcending by a wide margin the area under scrutiny, go hand in hand and support each other. 1 5 The geographic center of this particular development lies, incidentally, to the West of Asturias, in the Galícián-Portuguese zone. Examples include: dial. Ptg. alfada 'garden lettuce' = Ptg. alface; Gal. andias 'litter, stretcher, bier with shafts' = Sp. andas < OSp. andes; dial. Ptg. ást(r)ia 'rod, shaft, staff, flagpole' = Ptg. hasta; . bimbre (f.) ~ bimbria 'willow-twig' = Sp. mimbre (  . ) , Ptg. vime; . calambrìa 'spasm, cramp' = Sp. calambre; dial. Ptg. facia 'cheek' = Ptg. face; C.-Ast. güestria 'procession of ghosts' = OSp. huest(e) antigua; Aient, lágia 'slab, flat rock, paving stone' = Ptg. lage, laja, Gal. laxe; . landria 'acorn' = Sp. landre (fig. 'small tumor [in glands of neck, armpit, groin, etc.], small pocket'; otherwise replaced by the Arabism bellota), Ptg. lande; dial. Ptg. pélia 'skin' = Sp. piel/,Ptg. pele; dial. Ptg. rést-ea, -ia 'string' = Ptg. reste, Sp. ristra; . and Col. sebia = Ast. sebe 'wattle, stockade'; dial. Ptg. vestia 'dress, clothing' 14) See E. B. Williams, From Latin to Portuguese (2d ed., Philadelphia, 1968), § § 8 9 and 106. 15) For a broad-gauged discussion see my article "Diachronie Hypercharacterization in Romance/MrL, IX (1957), 79-113;X (1958), 1-36.


277

= Sp. veste.16 If prediction in these matters is at all admissible, it may be argued that C.-Ast. trabe 'large beam', 'ledge or sill in a wall' (C.) can at any time be ex pected to cast off the by-form *trabia, particularly in view of the coexistence of the dimin, trabiella 'shelf'. The specific advantage of the substitution of-ia, rather than -a, for ֊e (inconclusi ve as to gender) lies in this: While quite a few Hispano-Romance words in -a are masculine (the epicene type el, la gallina 'coward', the Hellenisms in -ma and -ta, etc.), unstressed -ia almost invariably points to a feminine. The heavier representation of Portuguese in this corner of the field may be due to the fact that the articulatory and acoustic distance between ֊e and -a is distinctly shorter along the Atlantic Coast than it is in the Center, a circumstance making the intercalation of /j/ doubly func tional, hence welcome. IV. SOURCE  : CONVERGENCE OF -ear AND -iar VERBS; IMPLICATIONS OF THE SPREAD OF THEIR BLEND

Hispano-Romance has received, as part of its Latin heritage, two special categories of -ar verbs which tend to converge in most dialects: (a) those in -iar, either inherited from -IĀRE, or extracted from descendants of -IGĀRE, or else newly developed from nouns - whatever their ancestry ֊ in -io, -ia, -ie;11 and (b) those in -ear (akin to the Ptg. -ejar, the Fr. -oyer, and the It. -eggiare series), whose kernel in the last analysis perpetuates Gr.-Lat. -IDIÄRE verbs — on a humbler level of transmission than do their -izar counterparts.18 Whereas speakers of Italian have no difficulty keeping apart their contingents of -iare and -eggiare verbs, their Hispanic cousins cannot help equating -ear and -iar, both pronounced alike /jar/ in the arrhizotonic forms of the paradigm, and have in some instances allowed the confusion to spread to radical-stressed forms. Because negoc-iar and pass-ear, -iando and -eando, etc. happen to rhyme in Portuguese, the interplay between members of a "proportion" has forced 'I negotiate' into the mould of 'I walk' {negoceio after passeio) ֊ though copio Ί copy', stressed on the i, and a solid block of other verbs continue to stand apart. 16) The article identified in the preceding footnote examines briefly alfâcia (fn. I36),andias (fn. 142), âst(r)ia (fnn. 135, 162), facia (fn. 136), lâgia (fn. 162), pélia (fnn. 135, 162), also réstia and vestia (ibid.). On güestria, which in the last analysis reflects HOSTE 'enemy (troop)', see RMP, Rato, and Canellada; on sebia, RMP and Canellada; on restìa, my own paper "The Etymology of Hispanic restolho, rastrojo, rostoll, "RPh, I (1947-48), 209-234, at 228f. 17) Cf. (AB)BREUIARE 'to shorten' > abreviar; ADLEUIĀRE 'to lighten' > aliviar; L I TIGARE 'to go to law, quarrel, dispute' > lidiar 'to struggle' but NĀUIGĀRE 'to sail across, swim' > OSp. nave-ar ~ -egar 'to sail'; RŪMIGĀRE (semantically influenced by RUMINARE 'to chew the cud') > rumiar 'to ruminate, muse'; apremiar 'to press, urge, compel' (based on OSp. premia 'oppression', which, in turn, echoes the -IA abstracts, such as SOLLERTIA 'shrewd ness, ingenuity', etc.). 18) Note the contrast between OSp. batear and mod. bautizar 'to christen', both reflexes of Gr.-Lat. BAPTIZARE. FASTIDIARE 'to loathe' > hastiar 'to disgust, cloy, sate' stands, of course, apart on account of its ; moreover, it has at all times been exposed to the influence of FASTIDII] ՝>hastío. Strictly learned are fastidi-ar 'to bore, vex', -oso 'annoying',- 'boredom, squeamishness, profound dislike'.

278

MULTIPLE CAUSATION

Conceivably as a result of this widespread tendential amalgam the -ear verbs have diffused tremendously in modern Spanish, on every level of discourse; so have the corresponding postverbal abstracts in -e(i)o, mandatorily masculine, on the model of pasear: paseo. Typical pairs of neologisms include telefonear 'to phone': telefoneo 'frequent phoning', as against Ptg. -nar, It. -nare, Fr. téléphoner. There are, in fact, instances of ostensibly deverbal abstracts having overtaken the corresponding verbs in this propensity toward the "augment" -e-; i.e., the verb still shows an infinitive in old-style -ar rather than in -ear, while the matching noun, as if in anticipation of the verb's further development, has already switched from -o to -eo (often display ing intensive or iterative overtones). 19 The exceptional productivity of the new, unified -iar in the Northwest can be illustrated by contrasting pertinent regionalisms with Spanish counterparts (a) ending in -ar, or (b) quite differently structured: (a) Ast. alienar = Sp. alterar 'to change, upset' (RMP); Mir. amansiar = Sp. amansar 'to tame, break, soothe'; B. atrasiar 'to delay' = Sp. atrasar՝, Ast. (inel. .) and Salm. (Torres Villarroel) cruciar = Sp. cruzar 'to cross' (RMP, Canellada); Salm. emplaciar = Sp. emplazar 'to summon(s), place, locate' (RMP); Ast. esforciar = Sp. esforzar 'to strengthen', refl. 'to exert .s.' (RMP); . filtriar - Sp. filtrar 'to filter'; . garrot-iar 'to beat the wheat or the spelt-wheat' vs. Sp. a-garrot-ar 'to bind with ropes, squeeze hard, pinch', refl. 'to turn stiff, numb' (cf. a-, en-garrar 'to seize'); C. gociá-se = Sp. gozar 'to enjoy'; Ast. rasiar = Sp. rasar 'to strickle, graze, skim' (RMP);  recost-i(y)á-se = Sp. recostarse (sobre algo) 'to repose, recline on'; . retembl-iar = Sp. retemblar 'to vibrate, shake, quiver'; Ast. (inel. .) trepiar 'to withstand violence, strain against' = Sp. trepar 'to climb'. (But B. and C. deliriar 'to rave' = Sp. delirar simply shows contamination by the noun delirio.) Or < costilla! (b) . a-prad-iar = Sp. reunir hacia el centro la hierba esparcida en el prado 'to assemble the grass strewn over a meadow'; 2 0  pal(ot)iar - Sp. trabajar un terreno a pala 'to dig up a patch of terrain with a shovel'; C. reboltiar = Sp. revolver, volver lo de abajo hacia arriba 'to turn upside down'; . retoñar = Sp. bramar un buey y ponerse enfurecido 'to roar and get furious' (speaking of a bull; cf. tor-ear 'to fight bulls; to mate a bull to cows'); C. sofaldiar = Sp. levantar las faldas 'to raise the skirt'; C. tastiar = Sp. tener tasto (la comida) 'to have a stale taste'; C. testenar = Sp. porfiar 'to be stubborn' (cf. testarudo 'hard-headed, obstinate'); C. trapiar = Sp. caer trapos de nieve 'to fall in flocks' (of snow). In some instances, of course, Asturian uses a primitive unavailable in Castilian, e.g., par b iar 'to have an earlymorning snack' (from parba, presumably < PARVA [pl.] 'a few small bites, a slight 19) See my paper "Fuentes indígenas y exóticas de los sustantivos y adjetivos verbales en -e" (I), RLiR, XXIII (1959), 80-111, at 109, where an attempt is made to set off such -eo ab stracts as flank ordinary -ar rather than -ear verbs (e.g., abaniqueo 'fanning, swinging motion, excessive gesturing', denodeo 'bold, audacious manners', esquileo 'shearing [season, place]', expedienteo 'procedure, red tape', pregoneo 'cry of street hawkers, crying of goods') from others which apparently have, in the end, paved the way for (secondary) -ear verbs: e.g., bojeo 'act of sailing round an island and measuring it', cosquilleo 'tickling', laboreo 'tilling, labor'; (naut.) 'reeving, running'; (min.) 'works, working' - against the background of boj(e)ar, cos quillear, and lab(o)r(e)ar. 20) Interestingly, some postverbal nouns preserve intact the original -e-, e.g.  apradeos 'restos de hierba que quedan sobre el prado después de atropar y hacer las cargas'.


279

intake'); or there may be a sharp difference in meaning, e.g.,  guar(i)ar 'to incubate' beside Sp. ponerse huero or güero 'to become addle(d)'; or the verb might have come into existence in Spanish but just failed to do so: C. regentiar 'to be domineer ing, boss around'. Over against these instances of incompatibility there are, needless to say, others of perfect or nearly perfect fit, e.g. C. gotiar = Sp. gotear 'to drip'; C. patiar = Sp. patear 'to kick, stamp'; C. repico-tiar = Sp. repiquetear 'to chime, ring a merry peal'; C. retorniar = Sp. tornear 'to shape by turning'; C. retranquiar 'to twist' = Sp. (archit.) retranquear 'to carve figures 'round columns'; . rodiar = Sp. rodear 'to circle'; . saboriar = Sp. saborear 'to savor'; C. solletrar (Acevedo: -iar; Rato: -ear) = Sp. deletrear 'to spell, combine letters'; C. tutiar 'to serenade' = Sp. tutear 'to use the familiar address'. The relevance of all these details to epenthetic / j / consists in this: The external and internal diffusion of -iar, prevalent in denominal and deadjectival verbs on a conspicuous scale (filtńar from filtru, garrotiar from garrote, a-pradìar from pradu, a-mansiar from mamu, etc.), has produced a setting which, through the interplay of various "proportions", has greatly favored the spread of / j / . Consider the follow ing scheme: Certain -iar verbs of old standing flank nouns in -ia or -iu (comparable to the familiar situation in Spanish: copiar 'to duplicate, imitate': copia, or acopiar 'to gather, store up': acopio). Alongside this traditional stock there have emerged new verbs in -iar (for the most part equivalents of Sp. -ear) accompanying nouns in -a or -u. The tendency will be for the last members of the proportions -iar: -ia:: -iar:-a or -iar:-iu::-iar:- to absorb an / j / increment. Whatever in each instance the hypothetical concatenation of events, developments of the type here outlined could very easily have taken place in the following families: C. farrapiu 'trace, remnant', farrampio s 'snow flakes' beside farrampiar 'to snow';  gradia 'harrow' (= Sp. grada) beside gradiar 'to harrow' (= Sp. gradar); Ast. gusmia 'snooping' (= Sp. husma) beside . (a)gusmiar 'to stir, mess up, poke'; 2 1 C. muriu 'wall' (= Sp. muro) and sobremuriu 'log placed on top of the muriu' beside muriar 'to build walls', muriador 'wall-builder'; C. rescampiu 'lightning' beside rescampiar 'to flash, sparkle', lit. 'to lighten';  tracamundi-u, -a 'confusion,error' beside tracamundiar 'to exchange, substitute inadvertently'. The model could very well have been furnished by, say,  ripia 'thin board between rafters (or beams), on top of which tiles are placed' alongside ripiar 'to place such thin boards', cf. Sp. ripia ~ ripiar (traceable to the mid-13th century but, unfortunately, of less than certain descent, see BDE, s.v.), and the fluidity of usage is dramatized by C. corrip(i)a 'temporary pigpen'. In a few extreme instances the verb (here credited with having triggered the process) may very well have secondarily fallen into desuetude, leaving its erstwhile partners, the noun or the adjective, as the sole recipients and repositories of epen thetic/j/. 21) Cf. Sp. husmear 'to scent, smell', orig. husmar (10th c), from a Greek stem widespread in Romance; see BDE. Amusingly, Canellada translates (a)gusmiar by 'hurguear', which must be a dialect word for hurgar disguised as a gloss; but then her style also allows for editorial, exegetic 'suerbe' as a substitute for Sp. sorbe.

280

MULTIPLE CAUSATION V. SOURCE D: SOME IMPLICATIONS OF THE HISPANO-ROMANCE DEVELOPMENT OF CONS. + /J/

It can be asserted, on a high level of abstraction, that intervocalic sequences of CONS. + / j / in the parent language lead to the following spectrum of results from the Atlantic Coast to the Ebro Valley: 22 (A) Complete disappearance of the group at issue: EXAGIU 'balance, weighting' > ensayo 'assay, test' > 'essay'; FASTIDIU 'loathing, squeamishness, dislike' > hastío 'disgust, tedium'; NÄVIG  'vessel, (large) ship' = navio 'id.'; (B) Amalgam of the consonant and the semiconsonant into a single, previously unknown phoneme: PUTEU 'well, pit' > OSp. pozo; GAUDIU 'joy, delight' > go z o (beside Arag. goyo); ŌSTIU/ŪSTIU 'door, entrance' > uço (Leon, uxo); FASCIA 'bandage, girdle, girth' > haca 'strip of land' (Arag. faja 'sash, girdle'); VĪNEA 'vine yard' > viña; VIRILIA (n. pl.) 'mark of a grown man' > verija 'region of the genitals'; (C) Preservation of the CONS. + / j / cluster (or, at least, of its skeletal structure),' without any influence on the preceding vowel: APIU 'parsley, celery' > apio 'celery'; CAMBIĀRE 'to exchange, barter' > OSp. cam(b)iar; CAVEA 'enclosure, den, cage' > gavia 'madman's cage, ditch for draining' (cf. Ptg. gaiv-ar 'to provide with drainage ditches', -agem 'drainage ditch', -el 'wall whose thickness tapers off toward the top'); ĒBRIACU 'drunk(en)' > embriago; FOVEA 'pit(fall), trap for game' > hoya 'hole, cavity, glen, pit'; NERVIU 'sinew, vigor' > nervio 'nerve, energy'; *NOVIU (recorded as a proper name, chiefly in South Italy) > novio 'person betrothed, sweetheart, fiancé, bridegroom'; (D) Preservation of the CONS. + / j / cluster, with concomitant or subsequent raising of the preceding vowel: CĒREU 'waxen' > cirio 'long, thick wax taper, church-candle'; Late Lat. OBVIĀRE 'to meet, withstand, hinder' > uviar 'to arrive' (Ruiz, 1704d: uyar) beside antuviar 'to act quickly, in anticipation' (cf. mod. de antuvión 'suddenly'); PLUVIA '(of, from, bringing) rain' > lluvia 'rain' (via *llovia, witness the verb llover; Ptg. chuva competes with dial. chuiva); VITREU 'of glass, glassy' > vidrio 'glass' (via * vedrio) ; (E) Metathesis within the CONS. + / j / cluster, often entailing (a) the monophthongization of the resultant diphthong (further subdivision into a) and β) reflects the difference in the preceding consonant); and, upon occasion, eventually provoking (b) — in the pretonic syllable — the raising by one step of the resultant monophthong: (a) a ) ĀREA 'level space, site, yard, playground' > era 'threshing floor, vegetable patch'; Low Lat. CABALLĀRIU Ht. 'man in charge of a packhorse' > cavaliero 'horseman, knight' > 'gentleman'; GLĀRE A 'gravel' > (g)lera 'gravel pit'; MĀTERIA 'matter, stuff, timber' > madera 'timber, wood, lumber' ; β) BĀSIU 'lover's kiss' > beso '(any) kiss'; CĀSEU 'cheese' > queso. Words that presumably belonged here include: Late Lat. CAMĪS(I)A 'shirt' > camisa; Gr.-Lat. CER- SEA, ֊ESEA (the former closer to Marcus Empiricus', the latter to Anthimus', 22) I owe several data, especially on the toponymic side, to Menéndez Pidal, Manual de gramática histórica española (6th ed.; Madrid, 1941), § 53; but my arrangement of the material is different from his, striving for a stronger dosage of abstractness. Additional information has been extracted from F. Hanssen, Gramática histórica de la lengua castellana (Halle, 1913), §140.


281

preference) 'cherry' > OSp. cere-sa, mod. -za; (Gallo-Lat.) CERV-ÉS(I)A, -IS(I)A 'kind of beer' > cervesa, mod. ֊za; and TENSIONE 'inclination, aiming-at, strivingafter' > tesón 'tenacity, firmness'; (b) CĒREOLA (PRUNA) 'waxen-skinned hard-core fruit' > OSp. ceruela 'plum' (A.D. 1438: BDE) > ciruela; ^VARIOLA (recorded as a woman's name, from VARIUS 'motley, variegated') > OSp. veruela 'pock, smallpox' > mod. viruela; (F) Disappearance of the / j / element, on condition that before vanishing, it has left its mark in raising the preceding (pretonic) vowel: TON S IONE 'cupping, shaving, shearing' > tusón 'sheep's fleece'. The ֊ frequently confusing ֊ details of the development need not detain us here, but the main factors controlling the diversity of the evolution stand out in sharp relief. They are: (a) The character of the consonant ushering in the / j / ; thus, /dj/ and /gj/ favor Solution A; / t j / , /kj/, /nj/, /lj/ invite Solution B; labials (including both bilabials and labio-dentals) harmonize with Solution  ; /rj/ and, less consistently, /sj/ demand Solution E ; (b) The character of the stem vowel preceding the cluster: While /i/ and /u/ cannot, by definition, be raised, /e/ and /o/ lend themselves to raising to /i/ and /u/, respectively, and /a/ can be simultaneously raised and fronted to /e/; (c) The lexical level of the word. Understandably, Gr.-Lat. EC(C)LĒSIA 'church' (lit. 'assembly') > OSp. e֊, i-glesia stands apart from CERVES(I)A > cervesa՛, Gr.Lat. *CHAMAETUSIU > cantueso 'French lavender, spike'; and Late Lat. ( sabueso 'hound' (cf. also anthrop. BÕSIU > OLeon. Boiso, Sp.Bueso)՛, (d) The general characteristics of the dialect involved. Thus Galician֊Portuguese and adjoining "Western" dialects clash, as far as these shifts are concerned, in three major respects with Castilian and its closest satellites: a ) They allow /sj/ to fuse into /ž/: BĀSIU > beijo, CĀSEU > queijo, CERASEA > cereja 'cherry', OCCASIONE 'opportunity, stroke of f o r t u n e ' > OPtg. cajom 'accident', THERESIA > OPtg. Tareija, even EC(C)LĒSIA > (e)igreja. Originally this sound change prevailed in numerous Central dialects as well, as is borne out by OSp. egrija 'church' (a variant which receives support from toponymy: Burg. Zam.Gn/fl/vfl<EC(C)LĒSIA ALBA, Grijota< EC(C)LĒSIA ALTA, etc.): j3) They maintain intact certain secondary diphthongs either monophthongized (ei > e) or replaced by tertiary diphthongs (oi > ue) in Castilian; γ) They practice metathesis on a more generous scale, esp. as regards the se quence LABIAL + / j / . Note the contrast RABI-ĒS, (Servius, glosses) -A 'fury, anger, madness' > OSp. ravia: Ptg. raiva; RUBEU 'red(dish)' > OSp. ruvio 'ruddy, blond': Ptg. ruivo 'strawberry-blond'. In the West, metathesis may result in the total absorp tion of / j / by the stem vowel, thus: *PREM-IA 'pressure, oppression' (on the model of ANGUST-IA, usually pi., 'narrowness, distress', and of Late Lat. A NX IA 'anxiety, anguish') > OPtg. prema: OSp. premia; SEPIA 'cuttle-fish' > Ptg. siba: OSp. (via Mozar.) xibia (A.D. 1335) > mod. jibia (see BDE); SUPERBIA 'pride, arrogance'> OSp. sobervia (mod. -bia): Ptg. soberba (rarely -bia); VINDĒMIA 'vintage' > Ptg. vindima: Sp. vendimia. (Historically, the case of Ptg. nervo vs. Sp. nervio 'nerve, energy' is slightly different, inasmuch as NERVUS and NERVIUM coexisted at

282

MULTIPLE CAUSATION

the starting point — witness OSp. niervo, dial, ñervo; but syn chronically the current geographic distribution of nervo [West] and nervio [Center] indubitably reinforces the impression of a phonic isogloss; the same is true of Ptg. amplo 'wide' < AMPLU beside Sp. amplio < [ո. compar., adv.] AMPLIU[S] and, broadly, speaking, of Ptg. agro 'sour' < Ä C R U alongside Mod. Sp. agrio, which represents older agro contami nated by agriar 'to sour'.) Outside the close-knit labial series one observes VITREU > Ptg. vidro: Sp. vidrio. Finally, one is free to add, from the ranks of the crumbling -IDUS series, such contributing elements — for the most part involving a radicalfinal bilabial or a labiodental — as Ptg. limpo: Sp. limpio 'clean' < LIMPID U 'translucid'; Ptg. sujo: OSp. suzio (mod. -cio) 'dirty' < SÖCIDU 'juicy, sappy'; OPtg. tibo: Sp. tibio (OLeon. tebio) 'lukewarm' < TEPIDU;Ptg. turvo: OSp. turvio (mod. -bio) 'turbid, troubled, unsettled' < TURBIDU. But Sp. lacio 'faded, withered' < FLACCID U 'flabby' apparently lacks a counterpart. All these factors, far from acting separately, may have worked together. It is, in fact, such ever-shifting and conflicting ensembles or alliances of circumstances that in the long run best explain the numerous complexities and, at first glance, baffling inconsistencies so characteristic of this entire domain. Thus, the coex istence in Spanish of (a) rubio (orig. ruvio), (b) royo 'red', roya 'rust, red blight, mildew' (scattered over Castile and Aragon; cf. Peñarroya [Cordova, Ciudad Real, Teruel], Villarroya [Logroño, Zaragoza], Monroyo [Teruel], and (c) ruyo (typical of Ávila, Burgos, Soria; cf. OSp. Monte Ruyo, Covasruyas 'Covarrubias') seems to be a matter of subdialectal differentiation, on the level of folk speech. But the fact that -orio has won out over -oiro in Central and Eastern Asturias and farther east in Santander (see below, Section VII), a victory - surprisingly ֊ not shared by -ario in its parallel rivalry with -efijro, may, in addition to certain strictly glottogeographic implications, also be ascribed to the concomitant circumstance that there existed, to begin with, many more "learnèd" (i.e., ecclesiastic and juridical) words in -orio than in -ario, a margin of preponderance which in the end drove a wedge between these twin suffixes. What matters most to us in the context of the present investigation is that the regional distribution, as between West and Center, of agro and agrio, chuva and lluvia, gaiv-ar, -el and gavia, igreja and iglesia, limpo and limpio, noivo and novio, prema and premia, raiva and ravia, ruivo and ruvio, siba and jibia, soberba and sobervia, sujo and suzio, tibo and tibio, turvo and turvio, vidro and vidrio, vindima and vendimia must have produced in border-dialects a sizable volume of confusion and a feeling for an unmistakable fluidity of usage, as regards both the metathesis and the epenthesis of / j / in this particular position. In itself, such a volume of in security might have been too slight to leave numerous vestiges or to have important repercussions; however, in conjunction with the other factors here isolated one by one it may have gathered sufficient strength to explain why in Asturian, which abuts on northwestern and northcentral dialects alike, speakers have toyed more than anywhere else with intercalary / j / . 2 3 23) Here are a few incidental remarks on details plus a scattering of hints as to bibliographic sources. On Ptg. chuiva/chuva see M. de Paiva Boléo, "Dialectologia e historia da língua: Isoglossas portuguesas," BF, XII (1951), 1-44 and 8 folding maps, cf. Lg., XXVIII (1952), 124-129;


283

One last example, to illustrate the precise mechanism regulating the transmission of certain words that fall under our rubric. Sp. savia 'sap, juice of plants', unlike its near-synonyms jugo and zumo, is extraneous to the old lexical stock of HispanoRomance. It is obviously somehow related to SAPA 'must, new wine', which struck root, from the start, in French, Provençal, and Catalan; but the discovery of its specific itinerary raises all manner of problems. On two occasions, Corominas has offered his readers alternative explanations (DCELC, IV [1957], 163a-164b, and BDE [1961], 513b): savia - absent, according to him, before 1765-83 - could have penetrated either through Catalan, where its congener is saba՛, or through French (sève), via Ptg. seiva (also sálvia), a complex of sound-sequence and imagery that untutored speakers could readily have associated with saiva 'spittle' < SALIVA. Though the former hypothesis sounds simpler, it is doubtless erroneous, since there are no precedents for a Catalan word shaped like saba to be transmuted on its way to Castilian into savia. The adoption of Ptg. seiva/saiva (whatever its own background) as savia, on the other hand, makes excellent sense, confirming our suspicion that the models of such interdialectal correspondences, or transformational rules, are deeply ingrained in the subconscious minds of speakers serving as transmitters. Such corre spondences operate most smoothly where metathesis is involved; they may be expected to work also, if a shade less regularly,i.e., less predictably,in the case of not quite so sharply silhouetted epenthesis. VI. SOURCE E: EPENTHESIS AND THE CONJUGATIONAL PARADIGM Clearly, the phonetic drift is refracted rather than reflected in the verbal para digm: Under varying circumstances, it can be reinforced and extended beyond its original confines, or it can run afoul of staunch opposition, to the point of being eroded or completely thwarted. If APIU 'celery', in Spanish, yields apio while SAPIA (M) '(that) I may taste' produces sepa (the Portuguese equivalents are aipo and saiba), it follows that morphology is governed by a lively interplay of (a) R. Posner, The Romance Languages (Garden City, N.Y., 1966), p. 46, wonders about the stem vowel of the base. Menéndez Pidal, Manual, § 5 3 , aligns camuesa 'pippin, sweet and aromatic apple' and cantueso with Bueso and sabueso; Corominas, BDE, pp. 120b and 123a,, traces the former to Ptg. camoesa (beside pero camoês), derived from a proper name, and the latter to Gr. khamài thýos, lit. 'earthy incense', Latinized through adaptation to Lat. T(H)ŪS. On prem(i)a see my article "Apretar, pr(i)eto, perto: historia de un cruce hispanolatino," BICC, IX (1953[-55]). 5-139, at 108-110. On agr(i), ampl(i)o, soberb(i)o, vidr(i)o - also necio 'silly', zafio 'uncouth', and a few additional adjectives (or original adjectives), see my aforementioned study on multiple vs. simple causation; a systematic confrontation of-EUS and -IDUS derivatives (e.g., juncia 'sedge' < IUNCEA 'rush-like' vs. escalio 'wasteland being put in cultivation' dial, meio), and where it, independently, performs an important antihiatic function. (In Leonese proper one observes no comparable extension of / j / ; thus V. Garcia Rey reports from El Bierzo: oya and tray a — reminis cent of Old Spanish —, plus haiga [cf. coll. Sp. huiga 'huya'], tienga; note the 2d pl. dedes, podades, requirades, and, surprisingly, teniades.) Complications superadded to this broad design involve the effects of diphthongization and of the intercalary velars on the / j / epenthesis. Subdialects at this cross roads probably each follow their own path. On Cabraniego usage we fortunately possess ample information: This subdialect offers its speakers a clear-cut choice between diphthongization and epenthesis. So it is either arrespuenda or arrespondia, either cuerra or corria, either cuesa or cosia, either esmuela or esmolia (but see below), either muerra or morria, either ruempa or rompia, either suerba or sorbia.25 The velar insert is ordinarily incompatible with the / j / epenthesis. So it is valga. . . válgamos /valgámos. . from valir (= Sp. ֊er)·; venga. . .vengamos/vengamos. . .from

24) Cabraniego villagers are divided on the stress pattern on Forms 5 and 6 of the paradigm. Some say cabíamos, cábiaes, favoring a single stressed syllable throughout the six-form set, exactly as speakers of Hispano-Romance have done in the case of the impf. ind. (see HR, XXVII [1959], 435-481; Proc. of the 9th Intern. Congr. of Linguists [The Hague, 1964], pp. 402-405); others cabíámos, cabiáes, etc., in faithful adherence to the model bequeathed by Latin. Among the "traditionalists", a more "progressive" wing compresses -áes to -és, thus: cabiés, comiés, dormiés, etc. This mechanism seems to have become completely standardized, to the point of safe predictability. 25) One discerns a subtle correlation between diphthong, / j / epenthesis, and stress pattern, where the latter is fluid (see preceding fn.). The choice of the diphthong - almost invariably /we/ rather than /je/ ֊ betokens the speaker's definite preference for stressing the same syllable throughout the set; on the other hand, epenthesis of / j / does not automatically preclude free dom of choice as to stress. Hence one hears either arrespuénda. . .arrespuéndamos. . . , or arres pòndia. . .arrespondiámo s; either cuérra. . .cuérramos, or córria. . .corríámos, or else córria. . . córríamos; either cuesa. . .cuésamos, or cosia. . .cosíámos, or else cosia. . .cósíamos. Canellada has been extremely clumsy in presenting these clear-cut facts (pp. 40f.).


285

venir, etc. Diphthongization is reconcilable with the velar, so far as /we/ is con cerned: dólga. . .dóIgamos. . .flank duelga. . .duéIgamos (cf., above, . tienga, which presses into service another diphthong); nevertheless ponga. . .pongamos¡ponga mos alone is on record, conceivably for historical reasons (DǑLEAM vs. PŌNAM). In a single instance, diphthongization, /j/ epenthesis, and velar insert have interacted to produce a uniquely luxuriant growth of parallel series: (a) esmuéla. . .esmuélamos. . . , (b-c) esmólia. . .esmòliamosl/esmoliamo s. .., (d) esmuélga. . .esmuélgamos. . . , (e) esmólguia. . .esmolguiámos. . . . 2 6 The Central Asturian paradigm is so tidy as to be even esthetically satisfactory, leaving as it does hardly any rough spot.27 It exemplifies with striking incisiveness the mutual repugnance of (a) diphthongization (especially one leading to ue) and (b) /j/ epenthesis, a situation independently observable in the ranks of nouns; see Section VIII, below. What remains to be accounted for is the source of the pres. subj. in ֊ia - especially where it is nonetymological, as is true of corria, cosia, etc. and the specific reason for its spectacular propagation. It seems obvious that the points of departure were such Latin verbs of the -EÕ/ -ĒRE, -IŌ/-ĚRE, and -ĪO/-ĪRE classes as had, to begin with, subjunctives in -EAM and -IAM (two varieties convergent from the start). A form like cabia < CAPIA-M, -T, then, causes no surprise, and the preservation of the etymological sequence of b and i flows from the general reluctance of Castilian and some other dialects to tamper with inherited LABIAL + /j/ (cf. OSp. ruvio, ravia, and the like); see Section V, above. It is not Ast. cabia that clamors for some plausible interpretation, but rather its metathesized counterparts Sp. quepa and Ptg. caiba (as against OPtg. cabha, a graphy which stands for /kabja/).

26) One minor local detail worthy of incidental mention: Form 6 of the set ordinarily ends in -an, but immediately preceding epenthetic / j / changes -an to -en, hence cuerran beside corrien, cuesan beside cosien, etc. Significantly, the intercalary velar also has the effect of changing -an to -en, thus: duelguen, ponguen, valguen, venguen. Since Portuguese, of proven affinity to Asturian, has here preserved the historically expected form: dolham < DOLEANT, venham ֊en must have occurred before analogical /lg/, /ng/ evicted etymological /λ/, /n/. If we make this assumption, then a single cause for -an > ֊en on the local scene, namely the presence of an immediately preceding palatal element, has been identified. On Ptg. dolha, ponha, valha, venha, etc. see Williams, From Latin to Portuguese, § § 34, 2; 43, 2A; 148, 1A; 193, 4 and 6; 201, 2 and 4; 203, 4 and 6. The same considerations apply to the suffix of the 2d sg.: ordinarily -as, but -es after / j / and /g/. Hence cuerras but conies; ruempas but rompies; sabies; pongués, both dolgues and duelgues; etc. In this connection it is noteworthy that oír generates oyas, oyan and coyer 'to pick' chimes in with cueyas, cueyan, while dir 'to go' goes its separate way with vayes, vayen. Apparently, the / j / traceable to /dj/ and /lj/ (AUDĪAS, COLLI(G)ĀS) and antihiatic / j / - the latter obviously more closely akin to the typical Asturo-Leonese epenthetic semiconsonant - produce entirely different results. But the dividing line is, admittedly, thin, and confusions are bound to occur; in fact, toyer < T O L L E R E 'to raise, remove, carry off' (here more resistant to erosion than its decayed Spanish cognate tullir 'to cripple') shows a unique divorce between Forms 2 and 6, regardless of monophthong or diphthong in the stem, toyes vs. toyan or tueyes vs. tueyan. 27) For one example of irregularity left unleveled see the plural of the paradigm of ver: veamos, vé(a)des, vé(a)den. The preservation of the -d- calls to mind certain peculiarities of the strong preterite of that same verb; see my article "Paradigmatic Resistance to Sound Change: The Old Spanish Preterite Forms vide, vido. . ., "Lg., XXXVI (1960), 281-346.

286

MULTIPLE CAUSATION

The only plausible reason for the local spread of the -/-element is the fact that the bifurcation basically produced by the treatment of CONS. + / j / - with / j / remaining intact after labials, r, and s, yet otherwise being absorbed into the stem or fused with the consonant on which it leaned - gave rise to a new and unwelcome situation, because it complicated morphology at the very point where the communi ty was striving for neatness and simplification. The options open to the speakers were either to maintain a multitude of patterns, as has actually occurred in Portu guese (pode ~ possa; quer ~ queira; faz ~ faça; diz ~ diga; bębe ~ beba; etc.) or to try to engage in some kind of leveling. The latter could have been conducive to the total elimination of the / j / element (its indirect reflexes included), i.e., in the direction of Sp. pueda beside ind. puede (as against Ptg. possa/pode); or in the direc tion of forcible absorption of the / j / into the stem (counter to the broad phonetic trend but in accord with Sp. quepa, sepa and Ptg. caiba, saiba), or else in the direc tion of the gradual generalization of cabia, sabia, which is exactly the solution Central Asturian ֊ in particular, Cabraniego ֊ eventually adopted. Two advantages accrued to this remodeling: First, the / j / placed in this position as a grammatical signal harmonized most gratifyingly with the epenthetic / j / of untold nouns, ad jectives, and adverbs, of the muriu, curtiu, anantias type; and, second, cab-ία, sab-ia, -ia, cos-ia still imply the hierarchical recognition of the cab-, sab-, com-, cos֊ stems shared by most other sectors of these verbs' paradigms, whereas the alter native choice of, say, *caib-a, *saib-a, *coim-a, *cois-a, and the like would unavoid ably have led to the establishment of rival stems (*coim-, *cois-, etc.) at a juncture when Hispano-Romance, I repeat, was resolutely veering away from apophony and related processes.28 Because the insertion of / j / in the pres. subj. involves purely morphological conditions, such as the speakers' recoil from any further proliferation of stem vari ants, it seems legitimate to postulate here a fifth, discrete source of the epenthetic semiconsonant. At this juncture, the inventory of autonomous factors contributing to the entrenchment and diffusion of intercalary / j / comes to an end. There remain to be tackled four side-issues raised by the analysis of the record. VII. THE CONFLICTING REFLEXES OF -Ōriu, -ǔfiu, AND ŏriu ι

Worthy of separate mention, in the framework of Asturo-Leonese dialects, is the fact, neatly established in Menéndez Pidal's 1906 monograph (p. 150), that in Old Asturian texts the familiar suffix -ŌRIU/-ŌRIA as well as the rarer wordfinal segments -ŮRIU (as in AUGURIU 'interpretation of omens') and -ORIU (as in CORIU 'leather') invariably yielded oi, through metathesis of /r/ and /j/ and, one gathers, through concomitant neutralization of the phonemic contrast /o/ : /ρ/; whereas in modern dialects West Asturian alone has kept /ojr/, allowing Central and East Asturian as well as contiguous Santanderino (Montañés) to switch to /orj/. Teberga marks (or, until recently, marked) the transition from /ojr/ to /orj/. Spanish itself has ue through pre-literary merger of the diphthongs ue (< Ǒ) and oi, in favor 28) On this point see my article "Range of Variation as a Clue to Dating (I)," RPh., XXI (1967-68),463-501.


287

of the former: cuero, salmuera 'brine, pickle' - except that where a neatly detach able suffix, bequeathed by -ÕRIU,is involved (and in a few other words, such as STOREA 'rush mat' > estera), the expected -uero, still observable in Berceo, before long yielded to -ero through contamination with more vigorous -ero (= Ptg. -eiro) < -ĀRIU. As a result of amalgamation of native oi and imported ue some varieties of Asturian display - or, half a century ago, displayed - the compromise form uei: cueyro (Santa Olaya), salmueira (Luarca, Santa Olaya, Curuena). Let me first cursorily illustrate these points, using Menéndez Pidal's own docu mentation (and slightly padding it in the process): W.-Ast. (Villaoril) abintadoiru — probably 'pitchfork' or 'winnowing device' (RMP); OAst.-Leon. agoiro (RMP); W.-Ast. (Villapedre) cobertoira (del puote), (Luarca) curbetoira (del pote) 'lid of an earthen pot' (RMP); W.-Ast. (Villapedre) coiro, (Santa Olalia) cueyro (RMP); C.-E.-Ast. corredoria 'kind of pulley' (RMP); W.-Ast. (Villaoril) culadoiru 'perforated stone placed on top of the laundry tub' (RMP), cf. . coladoriu; . componedoriu 'leaven, yeast'; . comprendori-u, -a 'brain, wits' (cf. coll. Sp. entendederas); OAst.-Leon. Doyro (hydron.) 'Duero' (RMP); C.-E.-Ast. estandoriu 'picket or stake placed at the side of a cart to secure a load' < *STĀTŌRIU (RMP), cf. . estan-, esten-doriu, with a reference to Rato's estadorio; W.-Ast. fesoira vs. C.-E.-Ast. fesoria, Ribadesella josoria 'spade, hoe' (RMP, whose base FOSSŌ RIA, from FODERE 'to dig', is, on balance, better than would be FISSÕRIA, from FINDERE 'to cleave', see REW3 § § 3 3 2 8 and 3462); C.-E.-Ast. (including C.) mesoria 'sickle for mowing or stick for picking the ears' (RMP: MES SORIA, from METO, MESSUS 'to harvest'; I have here adjusted the quantity of E to the standard of Ernout-Meillet; cf. REW3 §§5544f.); W.-Ast. (Vülaorü) paradoxra (RMP) vs. . paradoria 'part of the water mill' (cf. Canellada's Drawing 18); W.-Ast. (Villapedre) pasadoiro 'stepping-stones across a brook' (RMP; cf. Sp. pasadéra)', . pastoria 'rope, chain, or leather strap for tying the legs of a horse', flanked by the self-explanatory verb -riar; C. (nin) pulidoria '(to own) absolutely nothing', cf. pulir = esňidiar (based on NITIDU) 'to slip away'; OAst.-Leon. (Villapedre) salmoyrada (RMP), W.-Ast. (Luarca, etc.) salmueira (RMP), C.-E.-Ast. salmoria and Sant. (Cabuérniga) moria (RMP), . salmoria flanked by salmoriar 'to pickle'; C.-E.-Ast. sechoriu 'ploughshare' (RMP: SECTÕRIU, cf. REW3 §7769), see Canellada's Drawing 19 and her double entry sech-oria, -oriu; C. subidoria 'small step-ladder — suspended in the air — used for reaching the granary'; C.-E.-Ast. sumidoriu 'sewer, sink, drain' (RMP; endorsed by Canellada), while Acevedo's W.-Ast. sumideiro ~ instead of *-oiro - shows in its suffix the influence of Sp. -dero;  tortoria, - 'pig's snout' (tener t. 'to be furious', cf. murria from morro), alongside the C.-E.-Ast. phrase al retortoriu (RMP); W.-Ast. (Luarca, Villapedre) treitoira, (Santa Olaya) treitoiras 'clamp enclosing the axle of the cart' (RMP) beside C.-E.-Ast. trechoria and Sant. trichorias (RMP; the obvious base is TRACTÖRIU, see REW3 §8826 and E.-M., p. 698); 29  tresmontoriu 'enormous amount' (cf. PRŌMONTŌRIUM 'mountain ridge'); W.-Ast. (Valdés) visadoiro 'plough' (RMP: [*]VERSĀTŌRIU, an intensive-iterative elaboration on *VERSÕRIUM [REW3 §9245]; VERSĀRIA is recorded as a nautical term [E.-M., p. 725b]). Note such verbs as  mesoriar 'to harvest with sickles', (a)-, em-boltoriar 'to pile up or wrap up carelessly'. 29) For  the compiler furnishes two definitions (bolstered by references to Rato): trechoriu 'sendero estrecho en la falda de un monte, por donde se echan a rodar las cargas de lena' and -ies (f. pl.) 'piezas de madera del carro, entre las que gira el eje' (as against RMP's 'abrazadera en que gira el eje').

288

MULTIPLE CAUSATION

Since the general trend was for -orio to dislodge -oiro except in the extreme West of Asturias, the total acoustic impression given by this metathesis, during the long period of wavering, must have been extremely close to straight epenthesis of / j / , thus reinforcing the trend here surveyed. 30 Interestingly, Leonese proper has absorbed the Castilian substitute form -r: . apilladero 'part of the plough', (a)siestadero 'resting-place for cattle' (= Sp. sesteadero), bebedero 'roundish hole made in a cart-wheel, to let water pass through', cavadero 'dug-up earth', cebadero 'good harvest of barley', cernidero 'good patch of dug-up ground', coladera 'sink, scullery', etc. Is it mere coincidence that Leonese shows the agency of epenthetic /j/ on a far more modest scale than Asturian? VIII. DIPHTHONGIZATION AND EPENTHESIS OF / j /

Since diphthongization follows far hazier rules in Asturian than in, say, Portu guese or Castilian, it is challenging to determine whether there exists, outside the realm of conjugation, any possible link between the distribution of ue vs.  and of ie vs. e, on the one hand, and, on the other, the occurrence of epenthetic / j / . (On the relevant features in the comportment of the verb, especially in Cabraniego, see Section VI, above.) Tendentially, diphthongization of the stressed vowel and the presence of posttonic /j/ - whether merely preserved from time immemorial or newly wedged in — seem to block each other. How else can one justify the contrasts between the following forms: OAst. Decembrio (RMP) vs. Sp. Diciembre (but note the ending of Ptg. Dezembro); OAst. esforcio 'effort, stress' (RMP) vs. Sp. esfuerzo (OSp. -ço); Salm. (Masueco) estrundio 'crash, uproar, pomp' (RMP) vs. Sp. esìuendo, a distortion of TONITRU 'thunder', with the EX- of explosión, etc.; 3 1  forcia 'effort' vs. Sp. fuerza and esfuerzo (OSp. -ça, -ço); OAst. governio 'govern-ment, -ance' (RMP) vs. Sp. gobierno (OSp. -vierno);32 18th-c. Salm. (D. de Torres Vülarroel) lencio 'linen cloth' (RMP) 30) Note the radically different circumstances surrounding the transmission of -ĀRIU and -ÕRIU in Luso- and Hispano-Romance. While Spanish discriminates, within a single lexical system, between two levels (vernacular -ero vs. learnèd -ario) and while medieval Portuguese invites one to set off as many as three levels (vernacular -eiro vs. semi-learnèd -airo vs. strictly learnèd -ário; see Lg., XX [1944], 119), this suffix's pattern of metathesized vs. non-metathesized forms never serves to plot dialect frontiers on the geographic map; even Mozar. -air, in its basic design, harmonizes with neighboring reflexes. Conversely, throughout the Center, the rarer -ERIU (as in [Gr.-Lat.] COEMĒTĒRIUM, MINISTERIUM, and [Gr.-Lat.j MONASTĒRIUM) and its feminine counterpart invariably show the / j / kept in its original place, even where the suffixes join vernacular stems to yield unprecedented derivatives (cf. NRFH, VI [l952], 250-260); Old Portuguese alone tolerated -eiro beside -ério {Lg., XX, 119). Thus, within this set -ÕRIU is unique in helping one to trace the boundaries not, or not only, of socio-cultural (OSp. -orio vs. -uero), but also of regional, dialects (W.-Ast. -orio vs. C.-E.-Ast. -oriu). 31) Estrundio stands apart in that it shows the secondary metaphonic shift o > u. B. estruéndano, on the other hand, displays the familiar triple affinity of (a) stress on the antepenult, (b) diphthong, and (c) post-tonic a - an ideal setting for the spread of "unstressed suffixes". 32) Menéndez Pidal cites Ptg. (FS) invernio 'winter', which, for all its conspicuousness, does not fit the description of this series, as might have Ast. inverniu/inviernu < HĪBERNU. On this word see RPh., XXII (1968-69), 265-267.


289

vs. Sp. lienzo (OSp. lienço); Zam. (Sayago) melrriu 'blackbird' (RMP) vs. Sp. mirlo, orig. mierlo;33 OAst. Setembrio (RMP) vs. Sp. Se(p)tiembre, Ptg. -tembro! These cases form no strictly homogeneous group. In esforcio, farcia, lencio one observes the perpetuation - not purely graphic, as thoroughly modern and audible farcia demonstrates - of an etymological /j/, witness FORTIA and LINTEU/ *LENTEU.34 The other bases, however (e.g. MERULU, SEPTEMBRE, TONITRU; Gr.-Lat. GUBERNĀRE) lacked any / j / ingredient, so that Ast. / j / here is intrusive and actually competes in appeal of novelty with the diphthongs ie and ue, which in comparable forms have prevailed, or, at least, temporarily asserted themselves, in Castilian and closely allied dialects. The trend toward mutual exclusion of rising diphthong and /j/ is most prominent where the latter is adventitious, sporadic. For all its momentum, the tendency has fallen short of stiffening into a rigid rule, at least so far as ue is concerned. I can cite a few exceptions: the Ast. topon. Agüeria competing with Agüe(i)ra < AQUĀRIA (RMP) - note the unusual proven ience of ue; C. cuerr(i)a 'walled-in space used for storing chestnuts'; and C. tueriu 'thickest part of a tree trunk', which Canellada quite properly likens to Rato's tuero 'stalk of cabbage-like vegetables'; the topon. Tueya (Gijón) beside Toya (Villaviciosa) and Sigüeya (Cabrera Baja) beside Segovia included in the Catalán-Galmés dossier. Interestingly, the exceptions all seem to involve a margin of wavering; in this respect, they call to mind cases of vacillation like OSp. vergüe-nça ~ -na 'shame' < VER CUNDIA (beside clear-cut Ptg. vergonha, It. vergogna, etc.), also OSp. mas-, mestuerço 'water cress' < NASTURCIU, where the speakers, at the crucial points, seemed undecided as to whether to allow / j / (a) to affect (through assibilation) the contiguous consonant, (b) to penetrate into the stem - at no risk to the interposed consonant - and to join the stressed vowel in bringing forth a new diphthong, or (c) to accomplish both simultaneously. Also, in -ueya the /j/, atypically, does not lean on a consonant. How can one account for the widespread incompatibility of rising diphthong and intercalary /j/? One visualizes three possible avenues of approach, which, in volving as they do the assumption of disparate sequences of events and dissimilar factors of causation, can hardly ever coalesce into a single complex hypothesis. Either the / j / asserted its agency or wormed its way into the words early enough 33) On MERULU >mierlo > m i r l o and related changes, see my two papers: "The Inflectional Paradigm as an Occasional Determinant of Sound Change," Directions for Historical Linguistics, eds. W., P. Lehmann and Y. M. (Austin, 1968), pp. 21-64, and, in more searching detail, "Le nivellement morphologique comme point de départ d'une loi phonétique: la monophtongaison occasionnelle de ie et ue en ancien espagnol," Mélanges Jean Frappier (Paris: Klincksieck, to appear in 1970). 34) On the diphthongization in Asturo-Leonese there exist two major schools of thought. One hypothesis, traceable to A. Morel-Fatio, J. Cornu, and F. Hanssen and culminating in E. Staaff, Étude sur l'ancien dialecte léonais. . . (Upsala, 1907), pp. 189-202, views diphthongiza tion in this dialect group as superimposed by Castilian. The reverse idea of an old, autochthonous diphthongization originated with Menéndez Pidal and is championed most resolutely by D. Catalan and Á. Galmés, "La diptongación en leonés," Arch., IV (1954; Miscelánea. . Amado Alonso), 87-147, who explain away monophthongs as primarily due to Galician and Latin pres sures.

290

MULTIPLE CAUSATION

to have closed ę to e and Q to  and thus to have prevented diphthongization; this conjecture would obviously parallel the customary, indeed classic, explanation of OSp. teja (rather than *tieja) 'tile' < TEGULA, ojo (rather than *huejo) 'eye' < OCULU, techo (rather than *tiecho) 'roof, ceiling' < TECTU, ocho (rather than *huecho) 'eight' < OCTŌ. Inauspiciously for its advocates, this "rule", so neatly observable at work in Castilian, is famous for operating neither in Asturo-Leonese nor in Navarro-Aragonese. A second supposition - perhaps too sophisticated to carry conviction — would start from the premise that Northern Spain's rural popula tion has not been narrowly monodialectal and that such equations as (Sp.) cuero = (C.-E.-Ast.)  = (Gal.-Ptg., W.-Ast.) coiro 'leather' (< CORIU) have become deeply if dimly ingrained in the speakers' minds. On the analogy of a hard core of such equations those untutored innovators who allow a / j / to slip into one of their words will almost automatically replace any ie by e, and any ue by o. The margin of delay in such replacements is reflected in the aforementioned "exceptions" (tueriu, etc.). The corollary of this explanation is that / j / has here been inserted at a fairly late date, a process entailing the return from older ue (itself the product of primary Ǒ) to a secondary o. What is discomforting about this proposal is that the old palatals, released by -CT-, -C'L-, etc., are amply known to have stimulated rather than stifled diphthongization on the local scene.35 Perhaps the safest argu ment would be to declare forcia a straight Latinism; to analyze Setembrio, Decernbrio as hypercharacterized Latinisms (with the -rio of imperio, vituperio, etc.); and to view such doublets as (rustic) fuerça/ (learnèd) forcia as models which could have been seized upon to transform govierno into governio, mierlo into melrf(r)iu, etc. This last explanation seems far and away the most realistic, in terms of historical plausibility, and ties in with the prevalence of -ancia over -ança, etc. (see Section II, above). The train of thought acquires a touch of credibility against the background of extreme and protracted fluidity in the local use of diphthongs (especially as re gards their nonobligatory extension to pretonic syllables in the verbal paradigm and in suffixed derivatives). It is arguable that the transfer of this mechanism to the verbal paradigm in Cabraniego represents the last phase of a complicated process. IX. PATTERNS OF INTERFIXED / j /

Whereas typically the epenthetic / j / is intercalated between the stem and the verbal inflexion (or the ill-defined remnants of nominal inflexion: -a, -o, or else adverbial -s), one also detects a sharply characterized minority of cases in which /j/ intrudes between the nominal stem and a full-fledged derivational suffix. In recent years it has become customary to refer to such buffer segments as "interfixes". 36 Let me briefly illustrate the insertion of /j/ before -aco (-ago), -ada and -a(d)o ~ -au, -adura, -al, -án, -anco, -asco, -ato, -ay, and -ón.

35) See Section V of the well-documented paper by Catalán and Galmés (cited above). 36) I am here using this term, tentatively coined by H. Lausberg, in the sense given it in my own 1958 monograph {Estructuralismo e historia, II) as well as in later inquiries by F. Gonzalez Olle, J. R. Craddock, and others.


291

(a) -aco: Over against Ptg. buraco, OSp. huraco, Ast. furaco 'hole' (from a preLat. IE base contaminated by FORĀTUS, which has independently survived and sprouted in horadar 'to bore, perforate' via OSp. horado 'hole'? — see Corominas, BDE), one finds Sant. juriaco.37 For 'slug', the Asturian form is  llimiagu, match ing Sant. llumiaco (cf. It. lumaca); the starting point posited is *LĪM-ACCU (RMP, 1906), akin to *LĪM-ĀCEA presupposed by Sp. limaza - babosa 'id.', both inferred bases ultimately traceable, through suffix change, to LĪM-ĀΧ, -ĀCIS 'slug, (edible) snail'. (b) (a) -ada (f.): . alboriada 'dawn of day' = Sp. alborada; C. cuatriada 'good roll in bowling';  (pl.) xirigot(i)aes 'monkeyshine, cute behavior aimed at provok ing laughter'; (β) -ado > -au (  . ) : . sapiau 'porous ground undermined by toads'. The temporary emergence of such verbs as *cuatriar 'to hit four pins', *sapiar 'to behave Uke a toad, do a toad's work' is highly plausible as an intermediate stage (cf. B. borriquear 'to pretend to be stupid', lit. 'to act Uke an ass'; B. emburriar 'to push, shove', as one does a donkey; Sp. gatear 'to clamber, go on all fours' [intr.], 'to scratch, claw' [tr.]; vaquear 'to cover cows' [said of buUs], etc.). Note the optional status of / j / , a symptom of its recency, in xirigot(i)-aes, pl. of -ada. (c) -adura: . serp-, sirp-iadura 'herpetic humor', as against sebada 'herpetic eruption' (kind of rash); do these forms presuppose *serpiar 'to bite, do harm Uke a snake'? Cf. Sp. culebr-ear, serp(ent)-ear 'to crawl, wriggle, sUther (like a snake)'; see HR, XXIV (1956), 209. (d) -al: . berd-ial 'twig'; . ferr-ial subst, 'flint', adj. 'sturdy, healthy'. Related to Sp. verde 'green' and hierro (orig. fierro) 'iron', respectively. Cf. . cavial 'pole driven into the beam of a plough'. (e) -άη(ο): Cac. (Guijo de Granadilla) jolgac-ián 'bum, loafer', matching Sp. holgazán (which Corominas, BDE, decomposes into holgaz-ar 'to be idle, have a good time', in the last analysis an elaboration of folgar < FOLLICĀRE ' t o p a n t ' [like a pair of bellows or a leather bag], plus -án extracted from synonymous haragán). (f) -anco : . pilan- ~~ -ąuiu 'piece of wood, log, beam leaning over and threatening to fall down', with the metathesized local by-form pil-iancu; cf. the two correlated verbs, which may have triggered the metathesis: pilanquiar and piliancar 'to lean forward and fall down'. (g) -asco: C. berdiasc-a, -u 'twig, branch', from verde 'green' (similar, on the local scene, to berdial 'id.' in its hospitality to the semiconsonant, see above, but quite unlike berdugu, resistant to / j / ; Acevedo's vardasca matches more closely Rato's verdasca); zurr(i)ascu 'whip, lash' (reminiscent of Sp. zurriago beside zurrar), cf. C. zurriasc-azu 'whipping, lashing' (= Sp. zurriagazo). (h) -ato: C. flor-iatu alongside -itu '(wild) flower, blossoming plant', C. gol-iatu 'stench' (recalling Sp. olf-ato 'smell'; the g- stems from huele > gitele), C. sorbiatu 'small sip or gulp', surrounded by a phalanx of derivatives, 38 C. tus-iatu 'spell of Ught cough, hoarseness in the throat' beside C. tusiu 'attack of cough' (inf. tusir Sp. toser), cf. W.-Ast. tu sido. The occasional coexistence of dimin. -ato and of /-dominated near-synonyms (*floratu 'young flower' ~ floritu 'small flower', similar37) Menéndez Pidal, in 1906, actually operated with a base [*] FORACCU. 38) Sorbiat-ar 'to sip, slowly and loudly'; -azu 'noise made in slurping'; -ada 'the drink or liquid food gulped down at once', with lots of noise; -ón 'one who enjoys obstreperous sipping'; quite apart from the dimin. sorbicar 'to drink in very small sips', sorbites 'pears which, upon ripening, acquire extra softness'.

292

MULTIPLE CAUSATION

ly in the tusir family) must have produced, or stimulated to further growth, the combination -iatu. Yerb-atos, -ates (derog.) 'all manner of worthless weeds' refused to follow suit - within the same district of Cabranes - conceivably in a recoil from the irksome sequence /je/ . . . /ja/, or in an anticipation of such a recoil. 39 (i) -ay (the equivalent of Sp. -ajo): C. cociayu 'greens and offal mixed as pig feed' (beside coc-er 'to cook', dimin. -icar). (j) -ón: C. fam-ión, -ientu = W.-Ast. -ento 'hungry', also C. esfamiau, with an i unmistakably contributed by OSp. des-famnido, -fanbrido, -hambrido, OSay. deshambrinado, OSp. hambrido, cf. the contrast between ORioj. fam(n)iento and Ptg. faminto.40 Also,  pasar de raspión 'to scrape, graze'; sascudión 'jar, jolt, jerk' (beside the verb sascudir, doubtless a blend of sa- and es-codir < SVC-, EX-CUTERE beside QUATERE 'to shake'); 41 turrión (= torr-eňu, -endu) 'rasher, slice of bacon' (cf. Sp. torrezno), fig. 'ill-tempered, irascible person', pl. 'remnants left from heating grease' ( C ; also in Rato), over against turr-ada 'slice of toast' (Rato). Interfixation marks a deeper penetration of / j / into a word's inmost chamber than does its more common insertion at the border of stem and residual inflexion. The /j/ can be driven even more deeply into the kernel of a word only through metathesis, as in Ptg. saiba < sabha < SAPIAT; but, as a rule, metathesized / j / is etymological rather than adventitious.

X. ADVERBS IN -(i)as AND PREPOSITIONS IN -(i)a Aside from -mente and its variants (-miente, -ment, etc.) and alternates (guisa, etc.), Hispano-Romance uses -s (extracted from MAGIS, MINUS, POS(T), and the like) as an economical marker of certain adverbs; cf. antes 'before' vs. ANTE, OSp. en-, es-tonces 'then' vs. EX- (IN-) + TUNC + CE. With the passage of time, it was full-bodied -as rather than -s alone that assumed the burden of signaling adverbs as well as cognate prepositions and conjugations, witness the trajectory of DUM INTERI (M) 'while in the meantime' > OSp. (do)mientre > mod. mientras '(mean while'. Along this road Asturo-Leonese and its satellites have overtaken Castilian and, wedging in their characteristic / j / , have honed -ias into an even sharper-edged, unmistakable adverbial marker. In fact, of the three consecutive ingredients here involved, ljl and /a/ - stratigraphically, the "late-comers" - have by now become more vitally needed than the almost dispensable trail-blazer -s. The three best-known examples of adverbial -ia(s) are:

39) On -αtο, -ito in historical projection see J. R. Craddock, "A Critique of Recent Studies in Romance Diminutives," RPh., XIX (1965-66), 286-325, esp. 289-303. 40) For additional variants, a reference to C.-Ast. enganiu 'hungry, avid', and an analysis of -ido adjectives suggestive of 'want, lack, deprivation, inadequacy, etc.' (i.e., standing in polar opposition to the abundantial suffix -udo) see Lg., XXII (1946), 302-309, esp. 305f., and, from a higher vantage point, CTL, III (1966), 333-336. 41) On OSp. recodir 'to echo, retort, etc.' < RECUTERE 'to strike back, cause to rebound', sa- and se-codir 'to shake' < SUCCUTERE 'to shake from beneath, fling aloft', and their cognates see my paper in HR, XIV (1946), 104-159. On the descendants of EXCUTERE, fairly numer ous outside Spain, see REW3 § 2998; FEW, III (1928-34), 287 OF enjosque /nǧoskə/. By so stating we tacitly assume (1) the recurrence of the phrase inde usque; (2) a certain type of juncture between these two words conducive to coalescence of -de+ - into /ǧ/; and (3) the agency of various "sound laws" presiding over the chosen segment of evolution: (a, b) the normal lowering of ǐ to e and of ǔ to , () the loss of /w/ after /k/, and (d) the weakening of -e to /ə/. We next consider, in causal terms, the coexistence of the late medieval triad enjosque, enjusque, and enjesque rather than of enjosque alone; the u /ii/ of the second and most important member (4) is best explained as being due to associative pressure of sus < sū(r)su 'upward' (cf. mod. dessus) - a state of affairs not unexpected for those familiar with the aggressive actions of sus and its local counterparts elsewhere (e.g., the spread of  to its semantic opposites yuso in Old Spanish and giuso in Old Italian) — while the -e- of enjesque is most safely attributed to pressure of inde usque's own polar opposite, dē ex > OF dès 'from'. Judging from the outcome, en/usque fitted so smoothly into the lexical structure that, in the end, it (5) dislodged "regular" enjosque and "analogical" enjesque. Given the mobile design of certain adverbioprepositional phrases in Old French, (6) the opening syllable (en) could easily be sloughed off and (7) an attempt — abortive in the end — could be made to add -s in appropriate context, e.g., before a vowel, thus: jusque(s). Meanwhile, (8) an other interplay of erosive sound changes tended (a) to reduce /ğ/ to /ž/ through de-affrication, (b) to eliminate medial /s/ before /k/, and (c) to impair the status of final [], cf. mod. jusqu'ici 'until here'. After an interval of three centuries, (9) the /s/ before /k/ was restored, apparently, in the short view, on the analogy of lorsque 'when', presque 'almost', puisque 'since', and — optionally - tandis que 'while, whereas' and, in the long view, in harmony with the restoration of /s/ in numerous monosyllabic lexical morphemes (cf. mars 'March', ours 'bear', URSS 'USSR', and the like); the sequence ce que also had its share in the restoration of s. While a thin residue of doubt remains on the relative chronology of certain almost coöccurrent events (e.g., within the confines of Factor 3 and Factor 8), the general succession of changes here posited has acquired a sharp contour, and any radical reshuffling seems inadvisable. It is clear that, without the vital intermediate links in the chain, numerous alternative concatenations might have been speculatively tried out.8 The evolution can be profiled even more sharply, with regard to causation, in the case of the Luso-Hispanic synonyms of jusque, namely até and hasta. It is universally agreed that the sought-for etymon here is Ar. hattà. At this juncture we need, first of all, a cultural factor to support the severely isolated hypothesis of the borrowing of a preposition from a Semitic source: History teaches us that — no doubt, bilin gual — Arabs were employed as land surveyors in the Peninsula; the word for 'up to, till' may well have been almost a technical term in their parlance or trade jargon and (1) as such could have been absorbed into the mainstream of Hispano-Romance. 8) We owe the fundamental investigation into the vicissitudes oî jusque to P. Falk (1934); its main results have been incorporated into such etymological dictionaries as E. Gamillscheg's, in its revised edition (1966-69: 552), with a sprinkling of additional bibliographic clues.

ON HIERARCHIZING THE COMPONENTS. . .

303

Implied in the provenience of this lexeme is further its southern origin, while the preservation of usque in Provence, beyond the Pyrenees, alerts us to the possibility, however remote, that its vestigial remnants in the Peninsula might plausibly be found somewhere in the Northeast (see below); if this were so, (2) geographic configuration would be the second conditioning factor. The initial consonant of hattá lacked so much as an approximate equivalent in Romance; in observing and imitating its pronunciation in Arabic (and, similarly, in Germanic9), speakers of medieval Romance would either replace it by ƒ or altogether skip it: Witness this alternative in the rendition of Muhammad. This cleavage is reflected in (3) the contrast between OPtg. OSp. ata and OSp. fata; the confinement of the variant with ƒ- to the central dialects seems to coincide with the rise of a semantically kindred preposition, faz(i)a 'toward' (lit. 'with the face toward'), restricted to Old Spanish; Portuguese, to this day, uses contra or para com instead, while other cognate languages have recourse to vers(o) , etc. Hisp .-Ar. fata 'till' and Hisp.-Lat. faz(i)a 'toward' thus supported each other at the embryonic stage — a noteworthy instance of (4) lexical partner ship. Meanwhile, there developed, alongside the adv. desende < deinde 'thereupon', the conj. desque 'since', and other particles so shaped, the innovative preposition desde 'from'; and, given the pictorial affinity of 'from' and 'until', the cluster -sdof desde may have prepared speakers for (5) aspiring to an identical or, at least, similar cluster in fata.10 While there may have jelled such a latent polarization, the tendency was not sufficiently powerful for desde to directly impose its stressed vowel, or its -e, or even its central -sd- pillar on its assumed partner fata', or, viceversa, to accept fata's distinctive features (say, the -á-, the -a, or the -t-). Meanwhile, in medieval Northern Spain, there came into existence a word so far genetically unexplained, namely fascas 'almost', abundantly attested in 13th-century texts; taking the conspicuous nuclear /sk/ cluster as our cue, (6) we may interpret this quasi-preposition as a merger of the Arabic intruder fata and of the plausible residue of usque in that corner of the country — see Point 2, above. Going one step farther, we are free to argue that the coexistence of fascas 'almost' (='up to') and uncontaminated fata 'till', two words separated by but a short semantic distance, offered the perfect matrix for (7) the rise of the compromise form fasta, a conjecture supported by the fact that some of the oldest occurrences of fasta are traceable to texts (e.g., the Libro the Apolonio, composed in Old Aragonese) which also display fascas. The latent striving toward lexical polarization of desde and fata

9) Thus, Frank. *haunita 'shame' penetrated into Old French as honte [hõntə], which later emerged in Old Provençal as onta. Old Spanish borrowed both variants, substituting for the former fonta and leaving the latter intact. 10) One finds a mere suspicion of a possible link between the -sdof desde and the -st- of fasta in the Etymological Glossary prepared by J. D. M. Ford (1911:227): "The s is unexplained, but cf. the s of the correlative desde". J. Corominas (1955: 884) offers noteworthy attestations of fasta, asta, and hasta, but - unaware of Ford's conjecture errs in trying to explain away -st- as a lame imitation of Arabic geminate t; the point is, precisely, that fasta, characteristic of the Old Riojan dialect, was initially a mere variant of fata. On the effects of lexical polariza tion see my statement (1957: 79-113, at 103-6), with references to older literature; one eloquent parallel to deinde> > O S p . desende is dëiectâre > S p . desechar 'to reject', which I once examined in another context (1949: 201-14).

304

MULTIPLE CAUSATION

(Point 5, above) could well have abetted the rapprochement between fascas and fata. Only at this late stage did regular phonetic change make its appearance (8), allowing fasta to advance to hasta /hasta/ and eventually to /asta/, dial, /ahta/ or a h ta/,etc. The development of hattà in Portuguese went its separate way, with speakers - judging from texts11 - often stringing ata em [ataẽ], pronounced ať em. From this habitual combination até could before long have cut loose. Seeking next a common denominator for the trajectories of jusque, hasta, and até, we can (9) adduce the supplementary fact of the unusual diversity, in Romance, of expressions for 'until' (cf. It. fino a, sino a; Rum. pînǎ [la]12); behind this fact is the considerable amount of wavering on this particular point in other less closely related languages: G. bis, bis zu, bis nach; E. till, until, up to, down to, as far as, as late as, etc. Two afterthoughts. First, though the case history of hasta/até may in this one respect represent an atypical, if not an extreme, case, it illustrates the relatively modest part played by regular phonetic attrition in a word biography abounding in vicissitudes. Interestingly, the aggregate of the changes fasta > hasta > /asta/ > (dial.) /ahta/ represents merely the tail end of a long and intricate non-phonetic development. Second, the sequence of events here proposed and of the causes suspected behind these events, involves a fairly tight structure, though a certain reshuffling of its components is indeed possible. One can, e.g., argue that the lan guage universal or near-universal adumbrated under (9) should be placed at the top of the entire series, and one may, independently, plead for interverting (1) and (2) — actually, we are left wondering whether the recession of usque produced a near-vacuum into which hattà stepped before long, or whether the appeal and accept ance of hattà accelerated the decay of usque. Similarly, it is debatable whether the speakers' predisposition to polarize, however weakly, desde and fa(s)ta preceded, or followed upon, the contact, presumably established in Old Aragonese, between fascas and fata, to which we have chiefly credited the rise of fasta. Aside from these few loopholes, however, the thread_,of events has been rather solidly spun. Ill For all their relevance, the contextual and the chronological hierarchies are not the only ones worthy of a linguist's curiosity. Thus, one can, at least, make an 11) There further sprouted a cluster of variants ending in s: OPtg. atens, atees, etc., added on the analogy of a characteristic group of prefixes and prepositions (des-, es-, após, tras, etc.) rather than one of adverbs, as Corominas (1957: 1021) erroneously maintains. By far the best presentation and the most thorough documentation must be credited to E. K. Neuvonen (1951: 318-20). 12) Fino (a) is usually connected with Lat. finis 'end', finīre 'to end', fînītus 'ended'; it is less certain that sino (a) represents a blend of fino and (co)sì 'thus', the implication being a gesture like statement:'thus far'; on this score Migliorini and Duro (1964:525) are more cautious than is Devoto (1966: 395). If Rum. pînǎ 'until' is a reflex of Lat.paene ad, lit. 'almost to', as is not unlikely despite the existence of the widespread rival form para - a complication long ago studied by S. Puscariu (1905: 116) - then it would provide a remarkable typological parallel to the "knot" fascas/fasta here postulated.


305

attempt to rank the components of a causal relationship in the order of each constit uent's respective importance. For instance, in a context bearing on several cognate languages and involving a single multicausal process, it would seem to stand to reason that the recurrent element of causation — the one factor identified in two or more languages — should receive a higher rating than certain other elements, which seem to have operated only in the history of a single language. It is plausible to assume its preëminence in the actual flow of events, and it is almost mandatory to accord it special prominence in our account, or representation, of those happenings. Let me illustrate this point with an analysis of - almost irreconcilable - opinions on one notoriously controversial point of Romance morphology: the derivation of modern French adverbs in -ément. The heated discussion of this issue falls mainly into the period 1878-1932, stretching over a segment of time in excess of a half-century. Contemporary French — like most of its congeners — as a rule develops its modal adverbs in -ment from the feminine forms of the underlying adjectives, thus: grandement 'greatly, grandly' (from gran-d, -de), nouvellement 'newly' (from nouveau, -elle). Where the base form of the adjective happens to end in a vowel, modern French, unlike its congeners, has recourse to the masculine variant: ingénument 'frankly, ingenuously', joliment 'prettily', (iron.) 'nicely, finely', (coll.) 'extremely', vraiment 'truly'. Erratic relationships of adjective to adverb are, on the whole, few and far between; one oft-cited example, gentil 'pretty, nice, graceful' vs. gentiment 'nicely, gracefully', is irrelevant, referring as it does merely to spelling, since the -/ is mute. In a relatively small and fairly close-knit group of words, however, -ément rather than -ment is affixed to the masculine variant, with a few marginal and pre dictable adjustments. In the accepted literary language the number of such words appears not to have increased after 1900,13 and even in fluid (colloquial) and semicongealed (journalistic) usage reliable observers have of late recorded no significant new developments.14 The twenty-odd words at issue sprang into existence between, approximately, 1600 and 1900; here are the most familiar among them, presented at the outset in straight alphabetical (i.e., random) order: aveuglément 'blindly, rashly' (aveugle), commodément 'commodiously, comfort ably, suitably' (commode), communément 'commonly, usually' (commun), con formément 'suitably, in agreement with' (conforme), confusément 'confusedly, vaguely, dimly' (confus), diffusément 'diffusely, verbosely' (diffus), énormément 'enormously, beyond measure' (énorme), expressément 'expressly, positively, dis tinctly' (exprès), immensément 'immensely' (immense), importunément 'impor tunately, obtrusively' (importun), incommodément 'inconveniently' (incom13) I am relying here on accurate findings by J. Dubois (1962: 18, 19, 111), who collated two Larousse dictionaries (of the years 1900 and 1960, respectively), assessing thus the allomorphs: "Les variantes combinatoires sont d'une importance très diverse: -ement [as in mondiale ment] et -ment [as in épatamment] sont les seules vivantes, la forme en -ément n'existe que dans les recréations: mesurément" (19). Elsewhere he labeled aisément, assurément, aveuglé ment as "stable" (111). 14) There seems to be no mention at all of any -ément formations in the (unindexed) torso of É. Pichon's posthumous monograph (1942; the autlior died two years before), reflecting the preferences of the 'thirties.

306

MULTIPLE CAUSATION

mode), intensément 'intensely, violently' (intense), obscurément 'Obscurely, dimly' (obscur), opiniâtrément 'stubbornly' (opiniâtre), précisément 'precisely, exactly', 'quite, just (so)' (précis),profondément 'deeply' (profond),profusément 'profusely' (profus),uniformément 'uniformly' (uniforme)15. Authoritative older sources — e.g., K. Nyrop (1908: §608 ) and W. MeyerLübke (1921: §181) — sometimes blur the neat silhouette of this model by inter spersing their lists with such infelicitous examples as figurément 'figuratively' (figuré), insensément 'insanely, madly' (insensé), and sensément 'sensibly, judicious ly' (sensé), even though these three cases involve no apparent irregularity. It is appropriate, however, to invoke an extension of -ément in so isolated an instance as impunément 'with impunity', which is hardly based on impuni. Though the edges of the model, in general, are not fuzzy, one encounters isolated cases of protracted wavering, as when opiniâtrément coöccurs with -ement. Even before the discussion of -ément went into high gear, certain preliminary remarks were dropped by qualified observers; these stray comments appear, in retro spect, as a prelude to the ensuing debate. Thus, reflecting more on past-participial than on strictly adjectival adverbs, the French Renaissance grammarian H. Estienne, in his treatise Hypomneses, recommended light lengthening (symbolized in script by an acute accent mark) of the vowel preceding adverbial -ment. The chief advan tage of this pronunciation (known to us as conservative) was, he felt, the speaker's — and the writer's — ability to distinguish more neatly the adverbs at issue from their near-homonyms, the verbal abstracts, to the extent that the latter also hap pened to end in -ment: aveuglé(e)ment 'blind(ed)ly' vs. aveuglement 'blinding, blind ness', (fig.) 'infatuation, delusion'.16 In the historically-oriented 19th century certain pioneers of comparatism (e.g., F. Diez and G. Körting), being perfectly aware of the abnormality of the shift -ement /əmã/ >-ément /emã/, came up with nothing better than lame justifica-

15) This list, I repeat, is not exhaustive; there exists a small number of peripheral formations, a few of them ephemeral or facetious. K. Nyrop (1908: § 609) cited the 17th-century gram marian Vaugelas as militating in favor of extrêmément and Voltaire as inveighing against intimé ment, unanimément; in 1903 Th. Rosset overheard an illiterate Bourguignon residing in Paris use rondément 'roundly', perhaps an imitation of carrément. L. Spitzer (1928: 72) extracted from Ph. Plattner's grammar pertinément and timidément, clearly in lieu of pertinemment and timidement; significantly, pertinément tended to encroach on -(e)mment rather than -ement, while confinément, which Spitzer, on the authority of H. Vaganay, traced to the Renaissance writer Du Preau (1573), trespassed on -ument. Though their further growth was blocked, these two adverbs, along with the better-preserved impunément, in lieu of -i(e)ment, form an important triptych illustrating the many-pronged advance of -ément, in excess of what Tobler and his followers surmised. In his elaborations, Tobler documented mod. exquisément (1892), posthumément (1901). 16) Estienne's disquisition comes up for brief mention by K. Nyrop (1908: § 607). The prosodic verdict of another Renaissance grammarian, Jacques Pel(l)etier (1517-82), is also of relevance in this context. He expressly categorized as long the accented vowels in nommément, obstinément, privément all three, incidentally, participial rather than adjectival adverbs (pas sage quoted by Ch.-L. Livet [1859: 168]).

ON HIERARCHIZING THE COMPONENTS...

307

17

tions, while an otherwise well-informed E. Mätzner, as late as 1885, reported that this peculiarity was 'etymologically' (i.e., genetically or historically) unex plained.18 However, the lexicographer É. Littré — repeating, perhaps unwittingly, Estienne's example - remarked (after recording Fontenelle's 'regular' use of aveugle ment for 'blindly') that speakers had been well-advised in rejecting that form in favor of aveuglément 'à cause de la confusion avec le subst [antif] aveuglement'.19 Disputants closed ranks when A. Tobler (1878: 549-52), directly challenging Littré ('es ist indessen das Gefährliche der Homonymie dabei ohne Zweifel zu hoch angeschlagen'), offered the first nuanced, multicausal explanation for the hyper trophy of -ément.20 According to him, the principal reason for this extended use was, initially, the encroachment of past-participial -ément on the territory of ad jectival -ement where pairs of nearly identical participles and adjectives coexisted; e.g., conformément, at the start, flanked conformé 'conformed', but was secondarily associated with conforme and, as a result, dislodged conformement, paving the way for similarly contoured énormément', by the same token, infortuné(e)ment 'wretch edly' acted as an opening wedge for importunément; immensément echoed (in)sensémenť, expressément called up memories of Late OF espresseement, etc. Tobler recognized two subsidiary causal forces: In the individual case of communément, he argued, an older derivative in -elment (from commun-el, -al 'communal') — left drifting — had been attracted and absorbed. However, rather than applying this same atomistic lexical analysis to impunément, i.e., rather than reckoning with an erratic transfer of fairly isolated past-participial impuni(e)ment to the swelling ranks of the -ément group, he felt that a learnèd, or rather semilearned, ingredient, for once, was involved in a neologism like impunément: the combination of the Latin adv. impune (in French scholastic pronunciation) with vernacular -ment. Other such 'technische Ausdrücke der lateinischen Schulrhetorik' were, according to Tobler, confusément (favored by Amyot, a 16th-century classic), diffusément, and profusément. Gone was the slightest appeal to the negative role possibly played

17) Tobler (1878: 550) adduced - and rejected - such impressionistic statements as: "Hebung ['raising', 'salvaging'] des stummen e" (F. Diez) and "Rettung ['rescuing', 'saving'] des langen a des lateinischen Ablativs von dem Übergang in stummes e durch den Akzent" (G. Körting). In his digest of earlier pronouncements he also assessed certain statements by J. J. Ampère (1871: 280) and É. Littré (1863: 1.18). 18) See (1885: 230). True, the original edition - inaccessible to me - of Mätzner's historical ly slanted grammar of French appeared in 1856, and even the 2d ed. (1876) preceded the original version of Tobler's note by a narrow margin. The division - advocated by Mätzner — of all adverbs at issue into two major groups, depending on whether the underlying adjectives end in a vowel or a consonant, yielded no fringe benefits for the discussion of -ément. 19) This passage is one more proof, almost supererogatory, that Littré was the direct prede cessor of J. Gilliéron in matters of lexical pathology (induced by the threat of homonymy) and therapeutics. 20) Tobler's note exerted all the heavier influence as it was absorbed into the (oft-reprinted) opening volume of the author's collected grammatical papers - after radical expansion (1896, 1902, 1921: 96-106). The principal elaboration consisted in the newly marshaled evidence for the continued compositional, rather than derivational, status of the -ment adverbs in Old French and Old Provençal.

308

MULTIPLE CAUSATION

by the verbal abstracts, except for the hint (550) of a certain parallelism in the sound development of, say, éternument 'sneezing', remet-ciment 'thanks' and the participial adverbs in -u(e)ment, -i(e)ment. Tobler's note produced a powerful impact. From the late 'seventies until shortly after World War I a number of fine scholars (in France, in Germany, in Scandinavian countries) concerned themselves with the problem of -ément, offering more accurate datings, assembling the formations in more cogently devised subgroups, dredging up marginal items, and improving in many minor ways on Tobler's necessarily hasty and tentative formulation. Foremost among these continuators and refiners of Tobler's trend-setting statement were Arsène Darmesteter (1890: 2.287-94, at 291f.),21 Kr. Nyrop (1908: §608 ), 22 and W. Meyer-Lübke (1921 and 1966: §181). 23 While each author contributed a few original observations of his own (thus, Meyer-Lübke emphasized the connection of-ément with the relative length of the adjectives attracted to it), all three remained heavily dependent on the pioneer. The improvements were tactical rather than strategic, as when Tobler, to account for commodément, has extrapolated its forerunner *accom(m)od-ement from Sp. acomodadamente, It. accomodatamente, whereas Darmesteter was in a position to cite acomodé(e)ment from Amyot's Plutarch translation. All three linguists pro fessed their belief in multiple causation; thoroughly savored breaking up a single grammatical shift into its constituent lexical elements (Darmesteter declared: 'Chacun de ces mots doit être considéré à part' [292]); and remained oblivious of Estienne's and Littré's, all told, bolder thinking. Tobler himself, in revising his note (1886, etc.), also expatiated on the subject. An attempt to break new ground was made by Leo Spitzer (1925: 281-88; re vised, 1928: 70-84), in a rambling, self-contradictory essay, spiced with not a few sparklingly phrased, almost feuilletonistic remarks. After pruning the paper, one 21) Darmesteter had many opportunities to come to grips with the series of -ment words, examining them in context now of composition (1874: 69; the 1894 revision, by G. Paris, is unavailable to me), now of derivation (1877: 123). The chief value of his separate study on -ment (1882: 300-10; repr. 1890: 2.287-94, at 291-1) consists in the filigree quality of the philological groundwork, as when he showed, first, that diffusement, espresseement, impuniement, précisement, profusement, etc. did exist and were actually used by writers; second, that confusément may, counter to one's suspicion, have in fact preceded the (unsuccessful) neologism confusé; third, that it was obscurément which caused profondement to switch to -ément, etc. As regards the ultimate cause of the overspilling of -ément, Darmesteter remained agnostic: 'Par quelle action l'e muet est-il devenu é fermé? Nous ne pouvons le dire'. (True, the note was channeled through a pedagogical, not a philological, journal.) 22) Though less forceful in causal analysis than Tobler before him and Meyer-Lübke after him, Nyrop went farther along this path than Darmesteter. His strength lay in documentation from fairly modern sources, as when he extracted désirément from A. Daudet; conjoin this discovery with Darmesteter's evidence for débordément (1877:122). 23) Meyer-Lübke's analysis, though not flawless, derives its keenness from the fact that he previously examined the same issue on a pan-Romanic scale, if more summarily (1894: §620). On that earlier occasion he espoused Tobler's view of comunelment >-ément, but placed Fr. impunément, It. impunemente in a broader context of integration - what he later called 'Ein reihung' (OF com~comment'how', soef -soement 'gently', etc.). J. M. Piel's addendum (1966: 207) in the Supplement to the checked reprint of Meyer-Lübke's 1921 venture is, unfortunately, lacunary.


309

realizes that the author accepted, from among the three contributing factors isolated by Tobler, only one, namely the special appeal - and, consequently, influence of past-participial adverbs. On the stylistic side - at that turn, Spitzer's principal concern - he emphasized the learnèd provenience of most of those leader words as well as their active, dynamic character, traceable in the last analysis to their rooting in verbal soil, a circumstance which accounts for their "Beliebtheit" and "Affektbetontheit".24 By way of afterthought, as it were, Spitzer identified Fr. -é d'or en avant), Lunéville, and (ephemerous) empéreur. The first cycle of discussions came to an almost abrupt end with D. Scheludko's counterthrust (1932: 212-22). Long before Dubois, he established the fossilization of -ément in contemporary French. On the basis of a restricted, but methodically assembled corpus of 16th-century formations Scheludko proved that at the crucial period homonymy did play a pivotal role, with lexical items in -ement functioning usually either as verbal abstracts (e.g., affamement 'famine, exposure to hunger') or as adverbs (e.g., aigrement 'sourly'), but seldom either way (attentement, basse ment, bravement). Since there have been few verbal abstracts in -eement >-ément (of the type agrément, which is echoed by E. agreement), the gradual expansion of that suffix variant in the ranks of past-participial and, eventually, adjectival adverbs contributed effectively to disambiguation. The -eement variant underlying mod. -ément reached its peak in Middle French, thriving on translations from Latin. Though subscribing to Spitzer's critique of Tobler, Scheludko, in a brilliantly exe cuted maneuver, actually reverted to the pre-Toblerian position of Estienne and Littré, demonstrating that quasi-homonymous verbal abstracts and adverbs in -ment must have been in each other's way, an obstacle and a latent source of irritation which invited the remedial extension of adverbial -ément.25 As for the date of the crystallization of -ément, all parties are agreed — and agree, in turn, with F. Brunot (1909:346) - that the period of transition from the 16th 24) Spitzer's paper, though rich in allusions to all sorts of side-issues (parallels in German and English usage; references to and comments on short-lived, experimental French formations, such as entiérément, even ensemblément 'together'), was clearly half-baked. After flatly reject ing Tobler's interpretation of communément he himself cited one Old French derivative sup porting that conjecture: chamément 'carnally' (1928: 70, 84). In stressing the learnèd, almost scientific quality of -ément, he might have made it clearer that the medieval model as such was genuinely vernacular, acting as a substitute for, or circumlocution of, Class. Lat. -ē, -e, -ter, all three subject to phonic erosion; but that eventually it was seized upon by literalisticallyminded translators of Latin texts (legal, theological, etc.) and thus SECONDARILY acquired an erudite taint. 25) Among other accomplishments Scheludko calibrated (2l9nl) various obsolete -ément formations which Plattner had excerpted from unspecified sources; showed (220-1) that aveuglé ment, strictly, paralleled impunément, with -ément trespassing on the territory of -ir verbs (at the decisive stage the verb for 'to blind' had been aveugl-ir, not yet -er); and recognized already in 16th- and 17th-century formations a deliberately archaizing strain (222).

310

MULTIPLE CAUSATION

to the 17th century is involved. The following almost discrete reasons, then, have been adduced, over a century, for the genesis or, at least, the diffusion of -ément, in lieu of expected -ement (and, at rare intervals, also of-iment and -ument):26 (1) Avoidance of such ambiguity as may arise from homonymic overlaps with verbal abstracts in -ement (Estienne, Littré, Scheludko); (2) Extension of the past-participial adverbial model in -ément (Tobler recog nized the embryonic stage of this process, but provided no motivation for it), entail ing such stylistic advantages - enhanced dynamism and cultural sophistication as flow from any indirect connection with the verb (Spitzer); (3) Contagiousness of preëxistent instances of regular -ément in connection with Latin -eus adjectives, as in simultanément (Spitzer), (4) Confluence of scattered formations, including a few in -el-ment, left afloat in the lexicon — admittedly, little more than a concomitant (Tobler's communément and Spitzer's charnément); (5) Incorporation in a roundabout way of Latin adverbs in -e, such as impune 'with impunity', known to the élite through scholastic channels - once more, a concomitant (Tobler, Nyrop, Meyer-Lübke); (6) Intralexical spread as a consequence of (a) formal, (b) semantic, or (c) coin cident formal and semantic resemblances, as when énormément was coined after conformément (-orm-) and intensément echoed (in)sensément (-ens-) (Tobler and, with greater strength and specificity, Darmesteter, Nyrop, Meyer-Lübke); (7) Sheer length (polysyllabicism) of the underlying adjectives - again, a mere concomitant (a cautious Meyer-Lübke, with a faint response from Spitzer); (8) The appeal of learnedness — extending, paradoxically, to diverse categories of speakers (including the would-be-educated) and of writers (Nyrop, Spitzer, Scheludko); (9) An elusive phonostylistic quality increased liveliness? - found in a handful of other formations as well, e.g., in dorénavant (Spitzer). In revising his note, Tobler drew attention to the particular infelicity of-mement. All these disparate causes or factors, consecutively alleged, are noteworthy and involve samples of respectable reasoning; but the most intriguing of them is, I sup pose, (1), be it only because that explanation has been pleaded by three spokesmen: a distinguished Renaissance scholar, a hypersensitive mid-19th-century native expert, and the last (and, easily, best-prepared) discussant within the 1878-1932 cycle. Can one show that, piquancy apart, the underlying cause — avoidance of near-homonymy with a group of verbal abstracts - deserves to be placed at the top of the components of a causal hierarchy? We shall try to elucidate this issue by bringing in a clinching piece of evidence from a trans-Pyrenean medieval dialect.

26) To the various inventories of past-participial adverbs in -ément already tapped (Vaganay, Nyrop, etc.) add the helpful index (1941: 230-40) which H. Nilsson-Ehle attached to his splendidly documented Lund dissertation; it includes carrément, décidément, délibérément, désespérément, éloignément... réglément.


311

IV Medieval Hispanic texts, as a result of ceaseless interferences by scribes and copyists of different periods and varying dialectal backgrounds, generally show a high degree of confusion in the use of the variants of the adverbial suffix traceable to Lat. (abl.) mente 'in a mood or mind'. Thus, a single manuscript of the 13thcentury version of the "Libro de Apolonio" — which involves a Castilian overlay on an Old Aragonese foundation - wavers between -mient, -miente, and -mientre. The reflex of the nominal suffix -mentum in that same text is ordinarily -miento, but one encounters a few Latinisms in -mento and -menta, plus a scattering of Provençalisms (or Gallicisms) in -ment and -mente.27 As a result of such profusion of variants no clear-cut pattern for the avoidance, or non-avoidance, of the near-homophony between the two suffixes emerges from the available stock of data. However, there did exist near-by certain contemporary dialects or, at least, idiolects - judging from meticulously transcribed textual evidence - which displayed an astonishing degree of mutual dependence between the respective outcomes of -mente and -mentu(m). Thus, Vidal Mayor, once more an Old Aragonese text avail able in an exemplary edition (Tilander, 1956), exhibits the coexistence of adverbs in -ment: abastadament 'abundantly', acordablement 'through mutual agreement', acuitadament 'through heavy pressure', affirmadament 'affirmatively', agudament 'sharply', etc., on the one hand;28 and, on the other, of abstracts in -miento: abastamiento 'abundance, provision', abaxamiento 'lowering', aborrecimiento 'boredom', abreviamiento 'shortening', etc. Undoubtedly, Lat. stressed e, in any kind of syllable, should here have yielded ie, and in fact, the local word for 'mind', used separately, was mient, pl. mientes', when entering into an adverb, however, that same lexeme appeared, again and again, as -ment, with a monophthong. Could trans-Pyrenean influence have asserted itself? Are we to assume that, in this area alone, mente, when placed within an adverbial syntagm, before long forfeited its heavy stress, thus: fòrti mènte >fuerment — whereas in other texts the pattern fòrti ménte prevailed?29 We can hardly reach a decision on such scant evidence; but the advantage derived from a neater separation of such near-homonyms as abastamiento 'abundance' vs. abastadament 'abundantly', acuitamiento 'pressure' vs. acuitadament 'through 27) Here are a few illustrations, extracted from .  Mar den's careful edition (text and glos sary). The adverb makes its appearance in several sub-classes: (a) fieramient 'fiercely', (b) fieramiente, fuertemiente 'strongly', ricamiente 'richly'; (c) fieramientre, gravemientre 'gravely', unless speakers prefer radically different models, e.g., de alta guisa 'nobly', de fiera mariera, a escuso 'secretly', sobregent/e) 'beautifully'. The anisomorphism of the nominal suffix breaks through in the coexistence of (a) falagamiento 'flattery, cajolery', (b) fundamento 'reason, motive' beside the stereotyped phrase (c) de fundamenta 'thoroughly'; (d) cosiment 'favor', espirament 'breath (of life)'; (e) estrumente 'musical instrument'; and, by way of compromise, (sg.) argumente 'matter, topic' ~ (pl.) argumentos. Only in the case of -miento did there prevail a neatly profiled relationship between nominal abstract and underlying verb. 28) One runs across a very few exceptions: thus, ab(i)ertament 'openly' was in competition with abiertamient; fuerment, fuermientre, and fuertmientre all three vied for supremacy, with the diphthong -ie- and the final vowel going hand in hand. 29) One of the best observations in N. J. Dyer's perceptive note (1972: 304) is the relatively heavy stress on the adjective in the archaic (Alfonsine) construction de buena mient.

312

MULTIPLE CAUSATION

or under heavy pressure' is obvious. It is further plausible that the tendential use of the rival form with epenthetic r (-mientre) was also stimulated by the benefit it afforded of distinguishing more sharply between adverb and abstract, as I pointed out in my comment on Tilander years ago (1959: 689); miente alone at no time tended to appear with an intercalated r, neither did -miento. While one is free to advance different hypotheses, not all of them mutually incompatible, on the initial rise of -mientre,30 the — temporarily vigorous — spread of that fashion lends itself to the same broad interpretation as the baffling choice of the monophthong in OArag. -ment. The relevance of our finding consists in that, from a territorially and chronologi cally autonomous set of data, a parallel has been extrapolated to the interdependence of -menti (adverbs) and -ment2 (abstracts) in French, as posited, with varying degrees of authority, by Estienne, Littré, and Scheludko. The two situations under scrutiny were not entirely similar; e.g., the interplay of monophthongs (e) and diphthongs (ie), of variants apocopated (-nt) and nonapocopated (-nte), of forms with and without epenthetic r was for a while peculiar to Hispano-Romance, while remaining at all times alien to Gallo-Romance. Nevertheless, the one ingredient of near-homonymy and approaching ambiguity was common to both situations; so was, above all, the demonstrable eagerness of the two speech communities to find some remedy, though in the end the medications ministered turned out to be entirely different: extension of -ément in early modern French, polarization of -miento: -ment in at least one variety of Old Aragonese. In hierarchizing the causes adduced for justifying Fr. avauglément, obscurément, profondément, etc. it is consequently legitimate to give pride of place to this case and, in so doing, practically to overrule the verdict of almost all major experts (Tobler, Nyrop, Meyer-Lübke, Spitzer), though some of their earlier alternative choices can be reconciled with Estienne's and Littré's conjecture, if only assigned a position distinctly more modest than the one claimed by their original proponents. But after isolating this widespread latent malaise of overlap and conflict we must stop short of averring that reactions to it can be safely projected onto the level of the parent language. Staking out such claims would be contrary to fact. There was, strictly speaking, no direct connection between the events observed in medieval Aragon and those that occurred in post-medieval France. Fundamentally the same malaise acquired a degree of acuteness in two different locales at two diverse mo ments. The menace of clashes between the Romance descendants of Lat. -mente and -mentum falls into the class of chance convergences attributable to drift among genetically related languages - as redefined by J. H. Greenberg (1957: 46,49).

30) In all likelihood -mientre echoes, in a direct line, Cl. Lat. -menter, i.e., -ment+(t)er and, if this is so, represents the last vestige of the adverbial suffix -(i)ter, as Spitzer correctly sensed (1942: 5). But a modicum of influence could also have been exercised by (do)mientre Gater mientra, mientras) 'while' OSp. as-, es-conder and A(U)SCULTĀRE 'to listen, hear attentively' > OSp. as-, es-cuchar. Cf. "The Etymology of Spanish asperiega, esperiega", PhQ, XXVIII (1949), 294-311. 5 This derivation is still upheld; cf. the remark in Dauzat, Dubois, Mitterand's NDÉH, p. 3986: "Parce que l'ivraie cause une sorte d'ivresse". Similarly in E. Gamillscheg's EWFS2 (with a reference to REW 3 ) and in Bloch-Wartburg's DÉLF: "La voyelle initiale, qui est due à ivre, montre que le contact avec ce mot a été maintenu". — On traces of (obs.) Sp. briaguez 'drunk enness' see DHisL, I l (1936), 3566. 6 Ascoli was patently unaware of the fact that the standard form in medieval Spanish was egual (similarly egual-dad, -eza 'equality'), a situation which makes it plausible that the subsequent raising of e to  was stimulated by the /w/ ingredient of the immediately following -gu- segment. And he made no attempt to discriminate between the transparent Hispanism (top.) Iglesias in Sardinia and the indigenous reflexes of EC(C)LĒSIA which he adduced. 7 The message of Ascoli's article is occasionally obfuscated by his terminological quirks and proclivity toward poetizing; thus, for 'initial segments' he used "formole iniziali" (444); for 'pressures', "seduzioni" (445), etc. E. De Felice, La terminologia linguistica di G. I. Ascoli e della SVÆL scuola (Utrecht, 1954), ρ. 26α, caught only the former feature in his dragnet.

326

ACCENTOLOGY AND PHONOLOGY

the rarity of atonic chains like is v : , is c -, and iss- < Ics-, IPS- as a prime condi tion, a peculiarity which even subsequent confluence with ES-, AES- (whether followed by a vowel or a consonant) could not substantially modify. The dual novelty vis-à-vis Group I consisted, on the one hand, in the availability of a widely-represented segment EX- > ess- and, on the other, independently, in the rise and spread of a light prosthetic vowel before the s impurum (ST- est-, ist-) — preeminently, if not obligatorily, in unstressed syllables. The prosthetic vowel was vestigial ("sottilissima e quasi evanescente") — a mere soupçon ("un ultimo sentore") of a vowel —, distinctly weaker than the i of Sp. igual and Ptg. idade. Given this course of events in languages such as Italian, Raeto-Romance ("ladino"), and Eumanian, the opening segment of ispandere EXPANDERE 'to stretch out' or -albare EXALBĀRE 'to whiten' was clearly incapable of lending support to the corresponding endangered segment of, say, istesso, istade (whose initial vowel, we recall, had a prototype in the parent language). Above all, i S- could not give sufficient sustenance to is- for the latter to undergo contamina tion by ins-, except for the isolated case of EXIRE 'to go out (away, forth)', whose progeny, not least under the influence of the rhizotonic forms of the para digm (esci, esce in 2d and 3d sg.), appears as (inf.) escire, obs. iscire, with OGen. and OVen. insì 'uscì' exemplifying the expected occasional contamination by m-s-. 8 Conversely, in territories where a prosthetic minimum vowel was allowed to arise and to gather strength (STĒLLA 'star' > Logud, istella, Sp. estrella), the initial segment of iss-, iš- < EX- likewise gained in viability (Logud, ispandere, Prov. espandre). Here the sporadic encroachment of IN- on is-, es- loomed as a distinct possibility, unless the s was followed by a consonant. (This qualification is, strictly, not valid, except in Provençal, Catalan, Sardinian, and Classical French; there may be a few peripheral exceptions in Spanish9.) Lexical units 8 This isolated status of EXIRE, resulting in certain morphological "détresses" (Ascoli shuddered at the idea that uscì 'he left' might have been evicted by *si [447]) is, obviously, the prime reason for its gradual decay — often amenable to observation — in several cognate languages, including French and Spanish; cf. Lang., XLII (1966), 430-472, at 457f.; TLL, VIII :1 (1970; Mélanges A. Henry), 141-152, at 148; RPh, XXVII, 304-355, at 314-316. Ascoli broached the moot question — without actually coming to grips with it — of excessive erosion of a given word as a stimulus or pretext for innovative replacement; cf. Sp. comer 'to eat' in Heu of *er EDERE; Fr. avette or (borrowed) abeille '[cute, little] bee' for e(f) APE; VĀDERE ht. 'to rush, hasten' and, alternatively, *AMBITĀRE 'to circle, go in circles' as partial substitutes for ĪRE 'to go', etc. L. Bloomfield had his doubts on this score; see Language (N.Y., 1933) and Language History from "Language", ed. H. Hoijer (N.Y., 1965), p. 395. As with substitutions triggered by clashes between homonyms, the ready availability of a replacement may be the clinching factor. Also, the remedy need not involve total eviction, but may be confined to therapeutical change, as the Luso-Hispanic record of RĪGIDUS 'stiff' makes patent; see S. N. Dworkin's enlightening article in this issue. In any event, escape from inconvenience usually constitutes just one element in a bundle of coefficient forces; cf. my paper on simple vs. multiple causation in: To Honor Roman Jakobson (The Hague & Paris, 1967), pp. 1228-46. 9 On this side-issue Ascoli offered a lengthy footnote (448=450), which amounts to an excursus. By contrasting Sp. enmondar 'to remove knots from cloth' with EX-, Ē-MUNDĀRE 'to clean' (he might have included here enmendar 'to correct, change, amend' ĒMENDĀRE), Sp. enfriar 'to cool' with OSp. Ptg. esfriar [cf. Sp. resfriarse 'to catch cold'], Sp. enfaldar 'to lop off the lower branches of trees', refl, 'to tuck up the skirts' with It. sfaldare 'to cut into slices, cause to flake', Sp. (obs.)

CONFLICTING PROSODIC INFERENCES.. .

327

containing EX- > es(s)-, eš- followed by a vowel (or by s plus vowel) were, on the contrary, the perfect slices of material to provoke the intrusion of a nasal, witness Ptg. enxagoar, Sp. enjaguar ~ enjuagar 'to rinse' vs. It. sciacquare 'to rinse (out, through)', especially in view of the infrequency of the sets ess-a-, es-α-, etc. Logudorese may appear, at first glance, to contradict the evidence of Spanish, since it has warded off, under comparable circumstances, any potential interference by the nasal, as follows from comparison of Logud, is-ancare 'to break one's legs' (lit. 'hips') = It. sciancare, Logud, is-emņiare 'to lay waste' (cf. It. scempio 'simple, single, silly'), etc. with Sp. ens-, enj-. In reality, the local trend toward mandatory reduction of *in-sv- to is- has foiled any large-scale crystallization of ins-. Southern Sardinia's Campidanese dialect refuses to heed this rule and, as a result, not surprisingly, comes up with solutions such as insoru IPSÖRUM (gen.pl.m.) where the North has to make do with issóro. 2. Critique of Ascoli's "Law". It would be pointless and anticlimactic to reproach Ascoli at this late date with all sorts of minor inaccuracies and mis apprehensions.10 However, his paper — for all its brilliance and the enviable self-confidence it exudes — raises a number of serious, legitimate doubts; these must be ventilated, if only to retrieve the article's healthy core. A. THEORETICAL IMPLICATIONS. The author was aware of the methodological or theoretical import of his contribution ("questo tentativo teorico": 451), but did not spell out the implications as fully as today's readers might wish. Even if one discounts certain side-remarks (cf. fn. 8, above), as many as four broader issues emerge from his treatment of a fairly narrow and thinly-represented problem. In retrospect, the most noteworthy aspect of his "law" (a label he himself refrained from using) was easily its basic incompatibility with "phonetic laws" as defined by the Neogrammarians in reference to "Verschiebungen", since he reckoned with a cleavage, i.e., a dual development under essentially the enclarar [ = mod. aclarar 'to clear up'] with It. (ri)schiarare, Fr. éclairer [add éclaircir], and Sp. esclarecer, also Sp. entirar [an ephemeral formation?] with Ptg. estirar (found in Spanish, too: 'to draw, pull, tighten') and It. stirare 'to stretch (out), iron, press' (here he might have additionally invoked Fr. étirer 'to stretch, lengthen'), Ascoli posited, not without hesitation, such earlier stages as *ensfaldar, *ensfriar. The argument is unconvincing, since the older lexicon offered certain doublets (e.g., entonce [the predecessor of mod. entonces; cf. Ptg. entāo] ~ estonce 'then' IN-, -TUNC-, which provided an adequate model for large-scale prefix variation and eventu ally, in some instances, for irreversible prefix change. (Ascoli was aware of EX-TUNC-CE [449], but apparently not of vernacular estonce.) The protracted wavering between en- and es- in compound verbs seems, from today's vantage, to be largely due to the semantic, or functional, weakening of the descendants of IN- and Ē-/EX-, at one time sharply characterized in the parent language. IN- was diluted through its inherent ambiguity (adversative vs. ingressive or illative); -/Ethrough its debilitating contact with DĒ-, DĪ-/DIS-. 10 Thus, the discussion of Sp. isla, Ptg. ilka 'island' (448n) is marred by the author's unawareness of the Catalan provenience of ilha; the dominant OPtg. form was ĩsua. The Spanish word for 'whitewashing' is enjalbegar, involving the unsyncopated variant of the suffix -egar -ICĀRE familiar from other chromatic verbs, e.g., negregar 'to blacken' (see Lang., XXV [1949], 139-181, at 161, beside F. Lecoy's marginalia in Rom., LXXII [1951], 140-142; and BICC, I X [1953-55], 1-139, at 64f.). Enjalbelgar, cited by Ascoli (443), presupposes either an unusual, erratic variant or, worse, a ghost-word based on some misprint, possibly in an old dictionary. The reconstruction of the ad hoc base *EX-ALB-ELL-ICĀRE is, consequently, unfortunate (Meyer-Lübke wisely weeded it out).

328


same set of conditions, a possibility which the Neogrammarians — bent on the discovery of uniqueness — emphatically ruled out. 11 Other theoretical implica tions left un- or under-developed were: (a) the wisdom of operating with "seg ments" (as distinct from diphthongs, consonant clusters, and syllables);12 (b) the validity of the criteria of (a) lexical incidence and frequency, and (β) preëxistence of certain sound sequences (444, 446, 451: "forinole abituali"); 13 and () the justification, in the case under scrutiny, of Ascoli's repeated appeal to phonology alone, to the exclusion of morphology (esp. 451: "Ma sempre è non altro che una spinta fonetica, e non mai un'operazione  trasformazione morfo logica"), in general, and of prefixai and parasynthetic derivation, in particular. 14 B. ARBITRARY LIMITATION. Ascoli failed to provide any rationale for whole sale exclusion of such segments as (unstressed) EL-, ER-; EM-, EN-; F-, EV-, which shared with the specimens he dissected the tendency toward widespread loss of the dangling vowel, even though they showed no comparable affinity to the therapeutic intrusive nasal. Had Ascoli paid proper attention to the vicissitudes of words like (A)ERŪGINE 'rust (of copper)' and the corresponding adjective AERÜGINÖSUS,15 Ē-, *EXMERĀRE ' t o c l e a n ' , ĒRĀDĪCĀRE ' t o r o o t

Out'—Words

which show either apheresis at work in competition with some other solution (e.g., It. rame 'copper' vs. Sp. ar- > ai-ambre Vire' as alternative echoes of (A)ERĀMEN, or apheresis alone (Sp. roña 'scab, mange' alongside roñoso 'mangy', fig. 'stingy'), or else some alternative outcome(s), to the exclusion of apheresis, 11 There is a dark hint (447n2) of Aseoli's irritation at the Neogrammarians' presumptuousness or naïveté in the matter of analogy, apropos esco, esci, esce. On this point consult L Iordan and J. Orr, An Introduction to Romance Linguistics; its Schools and Scholars (1937; Oxford, 19702), pp. 27-30. 12 A theoretical justification for joint diachronic treatment of certain sound sequences may not have been supplied yet, at least not in Romance quarters. For practical purposes it has become customary to consider in this light a few special developments, e.g., -ACT-, -AX- (i.e., /aks/), -ULT- in Hispano-Romance. Otherwise, only diphthongs (AU, AE, OE) and consonant clusters (CL-, FL-, PL-; -NF-, -NS-, -NV-; -RS-; etc.) are usually reserved for such focus. A revival of Ascoli's intense concern with segments is conceivable, but only at the price of a generous quota of theoretical spade work. 13 It stands to reason that a preëxistent feature, within a sound system, that is itself not subject to change can serve as a potent magnet, attracting elements that have been left more or less freely drifting, as is true in Romance, for instance, of certain medial consonant clusters. It is further true that for extremely rare, especially unique, concatenations of sounds — in particular if they are structurally isolated, in terms of the pattern — one can expect several outcomes, from which it must at times be difficult to select one as the "norm" vis-à-vis a residue of "exceptions". Even if one makes these allowances, Ascoli's handling of such delicate balances often appears imprudent in retrospect. 14 I wonder whether any scholar has observed the fortuitous circumstance that no Latin pre fixes, with the exception of OB(S)- (doomed to blurring and eventual oblivion) and rare ULTRĀ-, began with a back vowel, over against the solid block of Ē-/EX-, EXTRĀ-, IN-, INTRĀ-/INTRO-, some of them highly productive. Does this lop-sided distribution not conjure up the danger that anal ysts may be tempted to lay at the door of phonology such developments as are moored, in reality, to morphological conditions ? Be that as it may, Ascoli was firmly convinced that he was facing a situation controlled solely by phonological forces. 15 See Consuelo López-Morillas, "A Midway Report on an Etymological Crux: Sp. roña", BPh, XXVIII, 488-496. Of the various hypotheses touched upon, the derivation of ronoso from AERŰGINŌSU (495f.), with rona jelling into existence as a back-formation, strikes me as by far the most realistic. Tendential apheresis would strike even a word of Arawak ancestry, cf. (e)naguas 'petticoat'.

CONFLICTING PROSODIC INFERENCES. . .

329

e.g. Sp. esmerar 'to brighten, polish', refl, 'to do one's best', OIt. smerare, OFr. esmerer (for details see below) —, then the cleavage he selected for micro scopic inspection would have been placed in a far bolder and more meaningful perspective. C. INADEQUATE ATTENTION TO AREAL CONFIGURATION. Ascoli's skill in juggling with a host of cognate languages, right down to Macedo-Rumanian (447n2), would have been impressive even in the twentieth century; it is most remarkable against the background of 1875 scholarship. But one salient feature eluded the author's vigilance: While he did, here and there, adduce (Northern) French evidence in grappling with side-issues,16 he failed to produce French examples in support of his central thesis, save for one isolated Berrichon form ; and Berry is, clearly, peripheral to the oïl zone.17 What seems most disap pointing today is not Ascoli's inability to supply illustrations, but his hesitation to draw any inference from the striking lacuna in the record he had so pains takingly assembled.18 D. INATTENTION TO SIGNIFICANT VARIANTS. Ascoli acknowledged the pro minence of Spanish, even vis-à-vis its Western neighbor, in fostering the crystallization of the parallel segments ens- and enx- > enj- (and shrewdly traced Sp. enseres 'fixtures, household goods, implements' to a plural nominalized *ESSE[RE] 'to be', cf. REW3 2917), but, strangely, did not pause to explain the rise and distribution of the /s/ and /š/ vars. 19 One would, off-hand, assume that EX- /eks/ was reduced to /es/, particularly before a consonant, in neatly segmentable formations, partly in recoil from undesirable clusters and partly under pressure from the block of prefixes ending in /s/, such as ABS-, BIS-, DIS-, MINUS-, O(B)S- (as in OSTENDERE 'to hold out, show'), SATIS-, SUBTUS-, TRĀ(N)S-, and TRES- (in preference to Class, TRĪ- as in TRĪDUUM 'space of three days', TRĪGINTĀ 'thirty'; TRI- as in TRIFOLIUM 'trefoil', TRIMODIUM 'triple cornmeasure', TRIPĒS 'tripod'; and TRĚ- as in TRECENTŌS 'three hundred'). Con versely, /eks/ > /es/ prevailed at the outset in unanalyzable or opaque words, chiefly if followed by a vowel, such as EXĀMEN 'swarm, throng', 'tongue of a 16 Note the scattered comments on Fr. ivrogne and its offshoots (442nl); on OFr. errer 'to wander, stray, range, ramble' ITINERĀRE (444n1); on top. Embrun EBURODUNU (445nl, according to Diez); on écouter, étriper, etc. (449n). 17 To be sure, Berrichon belongs to the "dialectes français proprement dits"; it is classified as such, e.g., in H. E. Keller and J. Renson's Supplement (SPRF, LV [1955]) to W. von Wartburg's Bibliographie des dictionnaires patois (1934) and in other reference works. But at a glance the map appended to Vol. I of G. Gröber's Grundriss (19062) suffices to show that Berry, adjacent to Occitania, is, starting with territorial conditions, one of the less sharply characterized northern Sprachlandschaflen. 18 Though Meyer-Lübke, in 1890 (loc. cit.), made a strenuous effort to supply French counter parts, his material, culled in large part from Anglo-Norman and Picard-Walloon sources, is more illustrative of tendential prefix change (es- > en-) than of the split (apheresis vs. epenthesis) that Ascoli had intended to dramatize. The shift of focus is particularly noticeable in the verbs sum moned as evidence: engacer, enjoir, enlongier ĒLONGĀRE, ensancier, enwarder [as against mod. égard, It. sguardo']. 19 For details of |the genesis and spread in Hispano-Romance of ens- and enx- — as prefixes (ensanchar) and prefixoids (enxienplo) — see Excursus A.

330


balance', EXEMPLUM 'sample, model', hence Sp. enjambre, OSp. enxiemplo — with ensuing confusion and eventual sporadic switches in either direction. Hence, counter to expectation, ensayo 'test, assay; essay' EXAGIU 'weight, weighing'; but, predictably, enjalbegar EX + ALBICĀRE 'to be white'. 20 E. NEGLECT OF THE ANTICIPATORY EPENTHESIS OF THE NASAL. One recogniz

able force that remained unmentioned in Ascoli's scheme is the — tendential — anticipatory epenthesis of the nasal. It has since become independently known that this tendency, though not narrowly confined to any one "daughter langu age", happens to be particularly well represented in Hispano-Romance; 21 contrast (a) Pad. OVen. negun, Comasc. negün, OW-Fr. negun, OProv. negun, also (b) Morv. dego, Prov. degun, even () Pad. legun, with Cat. ningún, OSp. nengún(o) ~ ningún(o), Cat. and Salm. dengún,22 all of them offshoots of NEC ŪNUS 'not even one'. Again, Sp. parangon clashes with It. paragone 'comparison, model, example', from the same Greek source: PARAKÓNE 'grinding-stone' (REWQ 6226); note also such compound suffixes as Sp. -anchín, -ancón, -anchón, -inclión, -enchín (e.g., parlanchín 'talkative, garrulous'; berrenchín 'odor omitted by furious wild boar' (from VERRĒS), later, through association with berrear 'to bleat, low, bellow', also 'cry of wayward children'; camaranclión 'garret, attic'; vejancón 'decrepit'), most of which are distinctly characteristic of Spanish and traceable in the last analysis to -acho + -in, etc.; finally, observe the equally typical intrusion of the nasal in reverse direction: Sp. intrincado '(en)tangled' vs. It. intricato 'tangled, intricate' (the English gloss also forms part of the corpus), from INTRICĀRE 'to tangle, confuse', based in turn on TRĪCAE 'trifles'.23 20 A study of s as a coda of Hispanic prefixes remains to be prepared. Some of these elements are only marginally represented, e.g., sos- solely in sostener 'to sustain'; but note that so-spechar became sos-pechar not only on the strength of a newly-favored syllabic division, but also through folk-etymological association with pecho 'breast, chest', cf. Lang., XXVIII (1952), 299-338. The history of prefixai 'three'- would gain from joint consideration with the near-parallel trajectory of 'two'- similarly used, cf. my note, "The Semantic Link Between Latin BIS- and Romance bes-, bis-", in: Studies Presented to ... Whatmough ... (The Hague, 1957), 165-171, and the additional analysis of bis- > biz- in RPh, XXIII, 188-200. The rivalry between DĒ-, DIS-, and EX- deserves a study distinctly superior to the mediocre French-oriented Berlin dissertation by W. Koenig in BBRPh, V:l (1935) — one paying sustained attention to G. Devoto's and Rosally Brøndal's conflicting interpretations of It. s-. Finally, the element TRĀ-, as in TRĀMITTERE 'to send across or over, to convey', which had its share of contacts with INTRA- and INTER-, was eventually absorbed by TRĀNS-, cf. Fr. transmettre, Sp. tra(n)smitir, etc. Cf. Consuelo López-Morillas' article included in this issue. 21 A monographic investigation of the nasal insert in Romance — beyond the confines of H. Schuchardt's pioneering exploration (1901) —remains an urgent desideratum. For a few stray data and bibliographic hints see the opening pages of my paper, "En torno a la etimología y evolución de eansar, eanso, cansa(n)cio" (I), NRFH, IX (1955), 225-276. Cornelia Rippere's Berkeley research paper, "Nasal Insertion in the Romance Languages: Its Underlying Regulari ties and Multiple Causes" (December, 1971), has so far, unfortunately, remained unpublished. 22 True, REW3 5875 lists isolated, scattered vestiges of the same phenomenon for other terri tories, e.g., Vegl. nenčojn, Istr. ningoun, Lyonn. dẽgö, and Foréz. lego. The smaller zones can be assigned to the Northeastern Adriatic and to the Middle Rhônian. More neatly pinpointed data on the regional and temporal distribution of ne(n)gun(o), ni(n)gun(o), de(n)gun(o) can be culled from my monograph, Hispanic "alguien" and Related Formations, in UCPL, 1:9 (1948), 357-442, see the Index. 23 Intrincado contrasts, significantly, with OPtg. trigar-se 'to make haste', trigoso 'fast', frigança 'speed', a family which, despite the semantic gap, also goes back to TRĪCAE, or rather to the cor-

CONFLICTING PROSODIC INFERENCES...

331

The shift OSp. ymer no > ynvierno, observable through the prism of medieval texts, can thus be studied in contexts radically different from those adduced by Ascoli, and it falls into place either way very smoothly indeed.24 Using this state of affairs as a background, one readily understands the rise of ensanchar 'to widen' (to this day, a satellite of ancho), from arch, AMPLĀRE (Pacuvius) 'to widen, expand' (based on AMPLUS) contaminated by (EX)AMPLIĀRE 'id.' (based on the comp, AMPLIUS); and of OSp. enxiemplo, OPtg. enxempro '(instructive, edifying) story, anecdote', from EXEMPLUM 'precedent, object lesson, sample'. And one further appreciates the pressure exerted by semantically kindred and formally resemblant verbs, such as, precisely, ensanchar and its referential opposite ensangostar 'to narrow' (its nasal again — at least in part — anticipatory in nature; based on angosto 'narrow, close', [fig.] 'insufficient' ANGUSTU) on the descendant of *EXALTIĀRE 'to raise, lift up' (from ALTUS 'high, lofty', orig. 'grown'), which was at the outset esalçar, eixalçar — these archaic forms are documented in Old Galician-Portuguese25 and jibe perfectly with OCat. exalçar, Prov. eisausar, OFr. esh-, es(s)-(h)aucier (cf. mod. exaucer, semanti cally influenced by Ch. L. EXAUDĪRE) 26 — and in the end became ensalçar (mod. spelling -zar) in Spanish. By placing ensalçar beside enaltecer — virtually a pair of synonyms —, one grows aware of a paradox overlooked by straight historians: While in diachronic perspective ens- can, demonstrably, be decomposed into EX- plus an infixed nasal, it sounds and looks to the native speaker and scribe rather like an — admittedly, unusual — sequence of en- (readily identified as a common pre fix) and an "unexplained", residual s. The reason for this synchronic interpreta tion is that compound prefixes, on the whole less frequent in Spanish than in French and Italian (two sister languages which abound in such amalgams: ra-, ren-; ra-, rin-, ris-), normally lend themselves to linear analysis: pesquerir 'to inquire' (PER + EX), rempujón 'push, thrust' (re + em), reajuste 'readjustment' (re + a), etc. The background allowing this situation to arise is the gradual semantic blurring of en- and es-, conducive to a state of near-interchangeability (thus, es-clar-ecer no longer necessarily suggests, either through the agency of es-, or through the instrumentality of the inchoative suffix, the connotation of responding verb TRĪCĀRĪ, -ĀRE 'to cause trouble, be evasive', but was protected from the danger of any intrusion of the nasal; see Three Hispanic Word Studies, in UCPL, 1:7 (1947), 227-296, at 244-247, 258-260, 283. From the suffixal formations in -anchin, -ancóri numerous words in -ancho, -anca, etc., some of which have become etymological cruxes, were secondarily extracted. 24 See my article, "Identification of Origin and Justification of Spread in Etymological Ana lysis: Studies in Sp. s(ol)ombra, en-sueňo, dial, em-berano", RPh, XXII, 259-280, at 265-267. 25 One discovers useful data in the glossary appended to M. Rodrigues Lapa's masterly edition (19702) of Cantigas ďescarnho e de mal dizer dos cancioneiros medievais galego-portugueses, pp. 37ò, 41α. Also worthy of attention are such forms, documented in that glossary, as e(i)xerdar 'to dis inherit' (pp. 376, 44a). Ex- before a vowel clearly yielded e(i)x- in primitive Luso- and HispanoRomance, unless an anticipatory nasal, destructive of / j / , wormed its way into the opening seg ments of words such as OPtg. e(n)xempro, OSp. e(n)xiemplo. 26 See REW 3 2935 and add the comment of Dauzat-Dubois-Mitterand, NDÉH, s.v. I fail to grasp the wisdom of operating with the interference of EXALTARE, posited by E. Gamillscheg (EWFS2, p. 406a), with a reference to [J.] Bruch and [0.] Bloch.

332


'sudden flood of light', and *endarecer would be perfectly conceivable as a rival or substitute). The identification of this supporting factor does not invalidate Ascoli's con jecture, but serves to remind one that, whereas pioneers were happiest when discovering proof of elegant simpHcity, moderns must be prepared to settle for the less exhilarating assumption of multiple causation.27 F. NEGLECT OF SWITCH TO A BACK OR CENTRAL VOWEL AS AN ESCAPE FROM, OR ALTERNATIVE TO, APHERESIS. One of Ascoli's major accomplishments was

his well-substantiated thesis that ''dangling" unstressed front vowels were ex posed to hazards which the other vowels, in that same position, were ordinarily spared. Isolated instances of tendential apheresis hitting vowels other than (A)E and I were, clearly, not within his purview; so he could afford to leave unmentioned the case history of, say, Gr.-Lat. APOTHĒCA 'store-room' > OSp. abdega, Ptg. adega ~ Sp. bodega 'wine cellar'.28 Less venial on his part was the omission of cleavages where the preservation of an endangered front vowel (and of a syllable), bought at the heavy price of its labialization, contrasted with the alternative of its total loss, as in certain provincial outcomes of Gr.-Lat. EPISCOPU 'bishop', lit. 'overseer, supervisor' > It. vescovo, OSp. Ptg. bispo, OProv. bispe, over against Sp. obispo; most of the contemporaneous speakers of Old French apparently faced no such dilemma: ebisque ~ evesque (> mod. évêque), 27 For a definition of this term and a discussion of the concepts it represents see my paper "Multiple Versus Simple Causation in Linguistic Change", in: To Honor Roman Jakobson (The Hague and Paris, 1967), II, 1228-46. 28 Note, however, that the standard historical grammars, which aim at providing a bird's-eye view of apheresis, usually pair off A- and - as the most vulnerable vowels. Thus, 0. SchultzGora, Altprovenzalisches Elementarbuch2 (Heidelberg, 1911), §§47f., lumps together: (a) Guiana AQUITANIA, Mabilha AMĀBILIA (with an erratic transmission of the suffix, reminiscent of Sp. maravilla 'marvel' MIRĀBILIA), Vïerna *AVIGERNA, and (b) gleiza EC(C)LĒSIA, Lienors < Elienors. F. Hanssen, Gramática histórica de la lengua castellana (Halle, 1913), §65, and R. Menéndez Pidal, Manual de gramática histórica6 (M., 1941), §22, similarly surround bodega with such near-parallels as (a)brótano 'southern-wood' -ONU, cetrero 'falconer' < acetrero (on the underlying primitive see my note, "Quelques avatars romans d'un zoonyme et d'un ornithonyme latins", in Études...offertes à Félix Lecoy, P., 1973, pp. 377-384), And. por mor de usted 'for the love of you, for your sake', Ast. güelo 'grandfather', placing them alongside either semilearned or vernacular formations shorn of e-: obs. pistolero EPISTOLĀRIU, bizma 'poultice' EPITHEMA, glesia 'church' EC(C)LĒSIA (analyzing mod. iglesia as due to false restoration of the precariously placed vowel), mellizos 'twins' *GEMELLĪCIOS [via *yemellizos, with a recoil from pretonic ye-, as in enero 'January', ermano 'brother'], obs. radio 'stray' (E)RRĀTĪVU, plus semilearned ingredients of the onomasticon, such as top. Mérida ĒMERITA and anthr. Millán EMILIĀNU. The voicing of the occlusive in bizma, bodega, Guiana, gleiza, glesia (cf. also the top. Grijota EC(C)LĒSIA ALTA) — as against pistolero — invites, of course, certain chronological and stratigraphic inferences. It is difficult to reach a conclusion regarding the trajectory Gr.-Lat. HOROLOGIU > OSp. relox (Escorial and Toledo Glossaries, Palencia, Nebrixa), except for selecting OCat. relotge as the immediate prototype, involving the same pattern of vowel dissimilation as redondo 'round' ROTUNDU. J. Corominas, DCE, III, 1083a, has assembled an interesting array of Old Catalan vars.: relotge (Eiximenis), relonge (Roig), arolotge, arelodge, even aberrant alalotja and alalotge; moreover, there existed in the Low Latinity of the Peninsula such forms as erologium, orelogium, testifying to the interplay of rival patterns of vowel dissimilation. It is, consequently, uncertain whether the vowel actually apheresized in the transition from provincial Latin to OCat. relotge was the ety mologicalo-, or one of its occasional substitutes a- and e-. Fr. horloge (OFr. oriloge: Book of Kings) reflects another Low Latin variant and may have drawn additional strength from association with HŌRA 'hour'.


333

cf. évêché 'bishopric' ; but vesque is on record. The primary tool in transmuting einto o- in obispo was, of course, the adjacency of bilabial b, later /β/; but the ultimate cause may very well have been stiff resistance to syllabic crumbling.29 The semilearned family of imagen and its offshoots imaginar, imaginación offers a striking analogue. OPtg. omagem represents, once more, the solution which achieves retention of the original syllabic structure through labialization of the vowel in jeopardy (β- > -). Sp. magín 'fancy, imagination' (cf. the phrase se le ha metido en el magín 'he has taken into his head') exemplifies the alter native — as expected, the bolder change of the form runs parallel to the freer, less inhibited growth of the word's semantic and phraseological ambit. PreClassical Spanish still tolerated such by-forms as maginar and maginança.30 One conspicuously instructive case history — though involving, e[-' > a[-' merely as a side-line — concerns the vicissitudes of Gr. ELMOSÝNĒ 'alms', 31 a Hellenism which, through its close association with charity, developed strong ties to the ecclesiastic and religious realm in early Christianity and, as a result, 29 The most entertaining word biographies are obviously those involving grains or crumbs of folk etymology. Take the history of the 'elephant': Schultz-Gora, loc. cit., credited the o- of OProv. olifan to the effect of the l, a conjecture that might have been more satisfactory in the context of Tuscan (see E. F. Tuttle, "Sedano, sener, prezzemolo, and the Intertonic Vowels in Tuscan", RPh, XXVII, 451 - 4 6 5 ; for clues to earlier literature see my own papers in RPh, XXVI, 306-334, and ALE, XIV : 2 [1973], 201-242). It seems more plausible to implicate a playful hint of 'smelling, sniffing, scenting' (OLĒRE, opportunely preserved in Provençal), given the prominence accorded in legend and folk-belief to the elephant's trunk. But even this interpretation would not, of itself, suffice to justify OSp. orifant (Koran translation, Bibl. N a c , MS 4938, ed. . LópezMorillas, 65v°8), whose intertonic -i- calls to mind the vowel structure of (sg.) alimaňa '(exotic) beast' (pl.) ANIMĀLIA, while the oris vaguely suggestive of flamboyance and outlandishness (cf. OFr. oriflamme, orig. orie flambe AUREA FLAMMA, as used in the Roland). One is almost tempted to construct a scale, ranging from faithful reproduction of the word used in Antiquity {elefante: Rimado, E, 1925b) to wild distortions (alifaut 'ivory': Emperador Ottas, xlii). On medieval reactions to the elephant see P. A. Bly and A. D. Deyermond, "The Use of figura in the Libro de Alexandre", JMRS, I I (1972), 151-181, at 171-175. While an etymologist of Spitzerian persuasion may easily be carried away by his search for culturally and artistically suggestive sources of interference with the expected development, the wiser strategy is to point out (a) the vulnerability of - in ELEPHANTE; (b) the threatening "flat ness", measured by the gaudiness of the pictorial message, of some such residual formation as Hefante; and (c) the stimulus thus exerted on the speakers' imagination to produce, through blends, an innovation at once phonically resistant and visually suggestive, whether through appeal to oler, to oro, to alimaña, or to any combination of these tempting frames of reference. On OSp. bispad(g)o, bispalia, bispete, bispiello, and bispo, found in MSS of Berceo, Don Juan Manuel, and Fuero Juzgo, see DHist., I I (1936), 234a. 30 A poem twice included in the Cancionero de Baena and attributed to a total of three 15th-c. writers contains the passage: "Tú le posyste tal maginança...". See M. R. Lida de Malkiel, Dido en la literatura espaňola (London, 1974), p. 11. On the derivational model see also my own mono graph, Development of the Latin Suffixes -ANTIA and -ENTIÅ..., UCPL, 1:4 (1945), 107, with addi tional examples of (i)-, e-maginança, mostly from 15th-c. sources. 31 Although Meyer-Lübke could, and did, fall back in his REW3 (§2839) on Wartburg's copious collection of data, his entry, for all its value as a repository of reflexes, is far from effectively architectured and fails to distinguish neatly between the four principal groups of variants. Corominas, DGE, I I I [1956], ΙΟΟαο, and IV [1957], 1039b, was entirely justified in selecting ELEMOSÝNA as a starting point. Alternative reconstructions, such as Menéndez Pidal's *ALEMOSINA (Manual6, §22) and Schultz-Gora's "V. Lat." *ALMOSINA, hardly hold water, especially as long as the qualifier remains undefined (common or provincial Vulgar Latin?). Meyer-Lübke, in his REW1, inconsistently starred ALEMÓSYNA, but not ALEMOSÍNA, which he viewed as the only vars. relevant to Romance; the slip was corrected in the revised edition.

334


invaded several circumjacent languages, including Cymric ( = Old Welsh), German, and English, in addition to the Basque enclave. The record shows that here (a) the retention of the e- bequeathed by Greek has been exceedingly rare; the pertinent reflexes that come to mind are (in frequent and almost learnèd) elemosna and elimosna in Old Spanish.32 Ptg. esmola exhibits, as if by compensation, so many distortions as to almost thwart any attempt to piece together its growth, step by step; 33 moreover, the e is no longer found "dangling" in a free syllable. The two relatively common solutions, entailing no stress shift, are: (b) the loss of the initial vowel: It. limosina, Log. limuzina, Friul. limuézine, Sp. limosna,34 and (c) the substitution of a- for endangered e- and, as a result, the eventual retention of the syllable — with the remote possibility of a lexical blend with ALIMENTUM 'nourishment' as a further stimulus: (a) Engad. almousna, Fr. aumône, Prov. almosna, OSp. almosna1,35 Cat. almoyna (latest orthographic 32 Aside from the strict "cultismo" elemósina which, as Corominas insists, is metrically sup ported (Berceo, Loores, I32d), one finds traces of (a) elemosna (Grail Fragments, ed. . Pietsch, 285v°), flanked by the agentive elemosnero (DCE, Suppl.), and (b) elimosna, which Corominas cites, it is true, from two less than reliable editions (DCE, III, 100a), but for which he offers the support of wavering in Low Latin: elemosyna ~ elymosina (9th-c. MS; CGL, V, 584.11). 33 Note, however, in Rodrigues Lapa's aforementioned glossary (p. 42a), the three entries esmolar 'pedir esmola', esmoleira 'bolsa de esmolas', and, most noteworthy, esmolnar 'dar esmola' (metathesized *elmosnar, which Corominas claims to have encountered in archaic sources?). One gathers that esmolnar and esmolar, originally mere variant forms, were in the end semantically polarized ('to give alms' vs. 'to beg' [lit. 'for alms']). In modern Spanish the contrast is signaled by the verbs accompanying the noun, in phrases like hacer (or dar) limosna vs. pedir limosna (== pordiosear). Ch.-L. eleemosynarius, OFr. au-, al-mosnier were also ambiguous and ended up by designating neither the donor, nor the receiver of charity, but the priest in charge of its administra tion. Esmoleira 'alms-chest, alms-box' is, of course, denominative. Perhaps not irrelevantly, the prefix es- seems to be more abundantly represented in the West than in the Center; contrast Ptg. escupir 'to spit' with Sp. cuspir EX-CŌN-SPUERE (with dissimilitory loss of the second s in the case of the longer compound), Ptg. esquecer-se 'to forget' (Spanish still keeps acaecer 'to happen', having jettisoned escaecer), esquentar beside aquecer 'to warm' (Spanish uses instead either acalorar, geared to the corresponding abstract, or acephalous calentar). Also, compare within the indigenous stratum of the lexicon, Ptg. esquerdo 'left' with Sp. izquierdo, whose segment iz- stands in complete isolation. Note further fam. esquipático 'strange, eccentric' and e(s)quipar 'to equip'. Thus, the survival of the (originally) genuine prefix es-, as in escupir, esquecer, esquentar, and the local predilection for the prefixoid es-, as in esmola once flanked by two verbs, clearly seem to go hand in hand. In Old French, one also finds stray vestiges of esmosnes; for one example, see H. Rheinfelder, Kultsprache und Profansprache in den romanischen Ländern, BAR, I I : 18 (Genève, 1933), pp. 226232, at 229 — a chapter rich in background information on the gradual differentiation of miseri cordia and eleemosyna (including the latter's descendants, esp. OProv. almorna). 34 G. Baist's experiment with the conjecture of an early Italianism ("Die spanische Sprache", in Gröber's Grundriss, I2, §34; cf. the faint echo in J. D. M. Ford, Old Spanish Readings2, Boston, 1911, p. 244a) must certainly be written off as a failure. In doing so, however, Corominas (DCE, III, 100a) leaps out of the frying-pan into the fire with this facile remark: "Por confusión con la a del artículo". Limosna(s) was the dominant form in medieval Spanish (Ruiz: 149a, 1308c, 1572a, 15906, 1628c, 1651a, etc.), blessed with a profusion of stereotyped phrases (demandar limosnas: Berceo, Santo Domingo, 105c; fazer grant limosna 'to do a noble deed': Carlos Maynes, xxxiii, and Emperador Ottas, ed. Amador de los Ríos, xlii) and surrounded by a system of satellites: limosn-adero, -ador, -era, -ero (on all four see the DCE), with limosnar 'to give alms' (Rimado, N, 576d), the forerunner of mod. -ear, serving as an important secondary center. Modern regionalisms include E.-Ast. llimosniegu 'generous donor of alms' (B. Vigón). 35 The fuller variant alimosna, involving a single syncope, occurs not only in the texts, for the most part archaic, adduced in the DCE, III, 100a (Berceo, Elena  Maria [v. 208], Proverbios de


335

preference: almoina), and (β) Cymr. alwysen, G. Almosen, E. alms, . arremusina.36 Finally, (d) a more energetic apheresis, lopping off two consecutive syllables, accompanies the accentual restructuring of the word in certain Italian and Raeto-Romance dialects: NE It. mozina, OMil. musina, Emil. (Reggio) molseina, Moden, mulseina, Gris. (Puschlav) moznina 'box for saving' ( < 'alms chest'), Trevisano far muzina 'to save, economize'.37 This case history, involving a key word which has been in the focus of interest since Diez, dramatically demonstrates that there existed powerful alternatives to the dilemma postulated by Ascoli, one such rival split being between apheresis and e- > a-. There are other cases, remotely or closely comparable, on record. Thus, Gr.-Lat. EVANGELIUM 'gospel' appears in Tuscan as vangelo, but in Aljamiado as abanjello38 One must expect numerous individual complications. Thus, ĒRĀDICĀRE 'to root out' apparently cast off, at the start, the rival forms arraigar1 'to uproot' (cf. Er. arracher 'to tear out') and desraigar (involving EX- > des-), which eventually coalesced into desarraigar39 The difficulty here Salomón [ed. . E. ], an emended passage in Calila e Dimna, ed. Allen), but also in Barlán e Josaphá, ed. Moldenhauer, 127 r°, 204v°. The corresponding verb alimosnar, equally venerable (Alexandre), as Corominas correctly remarks, matches Gasc, amoinà. The more radically con tracted, no doubt racier, var. almosna pertains to the same period (13th c ) ; association with alma 'soul' may have ensued, but is hardly needed to justify the second syncope. — This branch, overlooked by Meyer-Lübke but captured in Menéndez Pidal's Manual de gramática, could temporarily have received collateral support from Provençal usage, which made itself felt in the wake of the Cluny reform and at the peak of the pilgrimages to Santiago, only to decline later; cf. R. Lapesa, "La apócope de la vocal en castellano antiguo", EDMP, I I (1951), 185-226. 36 J. Jud, "Probleme der altromanischen Wortgeographie", ZRPh, XXXVIII (1914-17), 1-75, at 67, included ELEMOSYNA (pruning it to *ALMOSINA) among "einige christliche Wörter... deren Quelle offenbar in Frankreich liegt", citing as evidence of its radiation OHG alamuosan (which he undoubtedly extracted from Kluge) and OIr. almsan (on the authority of H. Pedersen, VGKS, I, 241). 37 The -l- of the medial cluster -ls- may conceal the last remnant of the lost syllable -LE-. This branch is regionally limited, to the extent of lacking any representatives in the Iberian peninsula. The preëminence of limosna in Hispano-Romance also shows in the influence which its vowel structure, as a result of many resemblances, seems to have exercised on lisonja 'flattery, adula tion'. The oldest form of this Provençalism or Gallicism is los(s)enja (14th-c. MS of Buenos proverbios, cited in DGE, IV, 1040a; De una emperatriz..., ed. Mussafia, xix); the corresponding verb was losenjar (Disputa del alma y el cuerpo, v. 31: "Te solién dar por to losenjar", cf. Ford, OSR, p. 245α; El emperador Ottas, vii), and the agentive was losenjero (Ottas, ix). Whichever the ultimate source of the word (for a constructive digest of M. Rodinson's Arabico-Persian conjec ture: LAUZÎNAG '[sweetened, sugar-coated] almond' see DGE, IV, 1040a), the result of the vocalic metathesis invoked by Corominas (DGE, III, 108b-110α) and other etymologists should, of course, have been lesonja, etc. In reality, all members of the family consistently display li- rather than He-, no matter how deeply they are split in other respects. Thus, MS S of Juan Ruiz's Libro de buen amor (183α, 419c, 638ā1, 1437d, 1443a) and MS E of Pero Lopez de Ayala's Rimado de palacio (the passage matching Ν, 663ā) exhibit traces of lij- rather than lis·, as a result of consonant assimilation and of the familiar "trueque de sibilantes"; the verb may be lisonjar (Rimado, N, 271d, 1234c), flanked by the agentive lisonjador 'fawner' .(ibid., 280d, 701b), or lisongear (Confisión del amante, ed. Knust, 313v°, 314v°, 315r°, 316r°; subst. inf. 317v°), accompanied by (a)lisongeador (Toledo Glossary, ed. Castro, 33), unless speakers prefer the "shortcut" lisongero (Rimado, N, 288c; Palacio Glossary, 353, etc.) — the core syllable is consistently li-, an interesting clue to the link connecting, in a Mediterranean country, the rituals of beggary and flattery. 38 This form and its var. in -eliyo occur, e.g., in the oft-cited Aljamiado translation of the Koran, lv°9, 59v°10. 39 The problem is, conceivably, less intricate than the record of relevant studies. Where F. Diez's flair, philological competence, common sense, and freedom from pedantry combined to

336


consists in the independent sprouting, in Spanish, of arraigar2,from the RĀDĪCE/ complex, a new verb endowed with the "positive" meaning 'to strike root' — cf. arraigo 'landed property', arraigue 'settling down, taking root', arraigado 'owning real estate'; (fig.) 'fixed, secure, inveterate'; (naut.) 'mooring', arraigadas (naut.) 'futtock-shrouds' — an entire family semantically irre concilable with the older "negative" arraigar1 and one whose pressure un doubtedly accelerated and clinched the acceptance of desarraigar] cf. destallar 'to prune useless branches', (prov.) desroñar 'to lop off decayed branches', desramar 'to strip off twigs', and the like.40 Particularly noteworthy are those situations where an unprotected e- made its appearance secondarily — e.g., through belated loss of G- > y-, a result of "de-diphthongization", as in the family of GEMELLŌS 'twins' (cf. Sp. mellizos 'id.', as against Fr. jumeaux). Take the case of Sp. amelgar 'to open furrows, mark boundaries with mounds', flanking the noun amelga 'ridge between two furrows'. It is tempting to link the verb directly to *GEMELLICĀRE 'to double, pair off' (cf. REW3 3720), in which case mellizos/amelgar would, once more, beautifully illustrate the shift e- > a- as an alternative to apheresis.41 RĀDĪCĀRE

help him subsume Fr. arracher, It. e-, s-radicare, and Prov. a-, es-raigar under E(X)RĀDĪCĀRE (EWBS3, II [1870], 207), his successors, confused by Neogrammatical worries, complicated the simple bifurcate picture of the ancestral family: RĀDĪCĀRE (or -ĀRĪ) 'to strike root' vs. ĒRĀDĪCĀRE 'to uproot', cf. the adv. (EX)RADÏCITUS 'from the root upward' (Ernout and Meillet, DÉLL4, pp. 562b563a). To justify arracher a scholar of Gröbere's caliber would invent of whole cloth *ABRĀDĪCĀRE, contrasting it with Sp. ar[r]aigar *ADRĀDĪCARE("Vulgärlateinische Substrate...", ALLG, I [1884], 233f., VII [1892], 34f.) — despite his awareness of OProv. esraigar, OFr. esrachier; bad as this analysis sounds, it is, on balance, less frustrating than Ε. Schwan's *ABRAPTICĀRE (Grammatik des Afrz., Leipzig, 1888, §228), deservedly rejected in F. Neumann's masterly rev. (ZBPh, XIV [1890], 566). Cf. also Meyer-Lübke's harsh critique, as early as his BEW1 (§§666, 2887). The views here espoused correspond, by and large, to those championed by Wartburg, FEW, I I I : 18 (1931), 235a. The existence of a primitive Hispano-Romance use of arraigar as a reflex of ĒRĀDICĀRE is based on such residua as Ptg. dial. (Barros) arreigar 'to root out' (BL, XX, 141) and Gal. arrigar, as produced by V. Garcia de Diego in his grammar. Rather than adopting the mod. Sp. meaning of arraigar: 'to strike root', irreconcilable with 'uprooting', today's speakers of Galician have re course to agarrdr, prender (Redondela), or botar raigaña (Montes); or else they use arraiganar, arraizar, just as their Portuguese neighbors to this day prefer enraizar (rare and poetic in Spanish) to (imported?) arraigar; cf. J. S. Crespo, Contributión a un vocabulario castellano-gallego (M., 1963), s.w. arraigar and desarraigar. 40 Several side-issues of the problem must be omitted from consideration, including the likeli hood of pressure of arrancar on both sides of the Pyrenees; see Meyer-Lübke, ZBPh, X X X I X (1917-19), 362f. The apheretic treatment of ĒRĀDĪCĀRE, as an alternative to e- > a-, is neatly shown by Lorr. rai 'to tear out'; true, A. Horning, "Zur Kunde der romanischen Dialekte der Vogesen und Lothringens", ZBPh, I X (1885), 497-512, at 510 (§70) — who dissipates A. Scheler's doubts as to the regionahsm's range and authenticity — justifies the apheresis by the avoidance of -sr-, citing OF fors ragier 'to rage, be beside oneself' as a favored substitute for esragier. MeyerLübke, who traced the a- var. also to Abruzzese (REW1 666) and the apheresized var. to RaetoRomance (REW3, 6692: Surs, riģau 'uprooted tree', Low. Eng. riar 'to tear out the flax with the root'), preferred to see in es- > a-rachier either assimilation of the first vowel to the second or the effect of the prefix a-, but in the end agreed with Wartburg (who had invoked the weeder's body posture to justify Ē- > a-) on the wisdom of transferring several patois forms from RĀDĪCĀRE to ĒRĀDĪCĀRE. I would go one step farther and consolidate, under a single entry, ĒRĀDĪCĀRE and *DĒRĀDĪCĀRE, if the latter is at all needed to account for It. diradicare. 41 The earlier literature on G- is digested, with the injection of some new ideas and fresh data, in two recent papers of my own: "Etiological Studies in Romance Diachronic Phonology", ALH, XIV:2 (1973[-74]), 201-242, at 229-234, and "In Search of Penultimate Causes of Language


337

In exceptional cases the availability of alternatives could serve as an antidote to the irksome effects of polysemy. Thus, if one projects, with Meyer-Liibke and other comparatists, the verb *Ē- or *EX-MERĀRE (from MERUS 'pure', -U 'undiluted wine') onto the plateau of Latin (cf. REW3 3024, where credit is given to Diez) and if one further endows that verb with the rival meanings 'to deprive of purity' (by mixing, adulterating) or 'to impart purity' (by scrub bing, polishing, tidying up), depending on the nuance of the prefix, then one understands the speaker's benefit from a sharp split. Thus one arrives, on the one hand, at Arag. amercar 'to mix water with wine' (J. Borao), Aljam, amerado 'mixed' and ameramiento 'mixture, mixing' (Koran translation, ed. LópezMorillas, 33v°l, 7), the implication being 'to deprive the wine of its purity', with adversative Ē- (cf. E. enucleate) disguised as α-; and, on the other hand, at the better-known OIt. smerare, OFr. esmerer 'to clean, polish' and, above all, Prov. Cat. Sp. Ptg. esmerar (flanked, in Hispano-Romance, by postverbal esmero 'care, tidiness'), with intensifying Ē- > es- ~ des- brought to bear on inherently negative concepts, as in OSp. (d)esnu(d)ar 'to bare, denude' (cf. Fr. dénuer; Ptg. desnudar gives the impression of being a Castilianism), Sp. desmenuzar ~ Ptg. esmiuçar 'to crumble' (based on MINŪTIAE 'small bits, details'). Initially, OSp. arambre 'wire', esp. 'copper wire' (cf. mod. alambre, arrived at through dissimilation of r-r), Ptg. ararne 'wire', 'alloy' (slang: 'money, dough'), from AERĀMEN 'copper or bronze object', with substitution of a- for β-, may similarly have involved the advantage of semantic specialization. The expected alternative, namely apheresis, is illustrated with It. rame 'copper (ore, coin, engraving)'; note the semantic distance. The regional distribution of the vari ants is hardly surprising in light of OIt. mal de San Rasmo, from ERASMU (REW3 2887: 'Bauchgrimmen oder Ruhr', with a reference to [. Wiese] ZRPh XXXV [1911], 232). A- prevails, then, in nouns before r (arambre) and l (almosna) — unless folk etymology overrides this trend (olifant) — and in verbs regardless of environ ment (amelgar); o- in nouns before bilabials (omagem, obispo), but less than consistently (abanjello). On some etymologically challenging problems, see Excursus B. G. NEED FOR BROADER VIEW OF THE RESTRUCTURING OF THE SYLLABLE. The

one major alternative to apheresis that Ascoli did recognize — as exemplified Change: Studies in the Avoidance of /ž/ in Proto-Spanish", offered (in absentia) at the 1974 Texas Symposium on Romance Linguistics. The case of amelga, which came to the attention of scholars through the efforts of Menéndez Pidal, "Etimologías españolas". Rom., XXIX (1900), 334-379, at 337, and of V. Garcia de Diego, Contribution al Diccionario hispánico etimològico (M., 1923), §278, continues to be puzzling, even if one credits the nautical term jimelga to the Galician strain and agrees to distinguish between (a) Cast, mielgo 'twin', with a diphthong, as a direct reflex of *GEMELLICU and (b) amelga 'faja que se señala al sembrar un terreno', with a monophthong, as post-verbal. Murc, chamelga 'surco para que el sembrado resuite uniforme' shows its Catalan provenience in its ch- < /g/-; is the immediately following a also a vestige of this migration, or does it testify to vowel dissimilation? If the alternative is true, my above interpretation of amelga would invite further nuancing.

338


by It. imbriaco ĒBRIĀCU — is best described not as an epenthesis of the nasal, but as a reorganization of the endangered initial syllable, in safer directions such as V → VC, or V → CV, or V → CVC, clearly a many-pronged pattern of which imbriacu, OSp. enxienplo, etc. represent merely one particularly conspicuous variety. Ascoli's analysis can be improved through rejection of his stubborn dis regard of the morphological component of the tangle. The tendential switch from Ē- to EX-, in the prefixai system of classical Latin, carried in itself a remedy against the hazard of a "dangling", unprotected vowel. There also existed a semantically-motivated rivalry between DĒ- 'down from' and DĪ-, DIS- 'asunder', with the latter apt to win on account of its superior graphic suggestiveness; DIS- and EX- also competed for attention, and in languages where a prosthetic front vowel sprouted contacts were established between, say, DĒ-SP. .. and e sp-, cf. the long distance between Ch.L. SPŌNSĀRE (Tertullian) 'to espouse' and Imp. L. DĒSPŌNSĀRE 'to pledge' as against the shorter distance between Sp. espos-o 'husband', -a 'wife' and despos-ar 'to perform the marriage ceremony for', -ado 'newly-married', -orio(s) 'act of betrothal'. 42 Against this background of fluidity, one finds, side by side, such reinforce ments as des-, des- and ens- (ins-), enx- (inx-). Thus, a single MS of an Aljamiado Koran translation wavers between desenp°lo (5r°7) and insenp°lo (23r°ll) 'exam ple'. 43 Technically, in terms of "surface causation", one can separate the two processes involved, by arguing that the speakers' vacillation between two rival prefixes, des- and es-, which here spilled over onto a prefixoid, and the optional anticipation, before /š/, of the nasal entering into the cluster -np°l-, constitute entirely unrelated situations. And yet the recoil from highly vulnerable eprovides a common denominator, in terms of "deep causation". Incidentally, since IM-, IN-, qua prefix, was subject to occasional atrophy in Romance (cf. OProv. genh 'mind' INGENIU, tro 'until' INTRŌ-), the coexistence of Sp. briago ~ embriago, It. briaco ~ imbriaco fitted a well-established mould.44 There were other solutions open to speakers eager to add strength to a crumbling syllable. Aljamiado texts favor ye(gu)walar 'to make equal, to level, arrange' (Koran, 30r°, 42r°7, 51v°10, etc.; Libro de Yūsuf, 85, 123, etc.), ye(gu)wal 'like, same' AEQUĀLE (Koran, 16v°13, 17r°4, et passim), as against OSp. egualar (i-), egual (i-). The diphthong could have been transferred from the paradigm of AEQUĀRE (REW3 239), which happens to be abundantly represented 42 T h e examples of H i s p a n o - R o m a n c e wavering between des- a n d es- are legion; t h e ofta d d u c e d K o r a n translation offers readings such as eskereer ' t o believe' (lv°14, cf. Recontamiento..., ed. A. R . Nykl, 63v°, a n d Libro de Yūsuf, 170), eškubrir 'to reveal' (24r°3-4), a n d ešmentir ' t o a d e n y ' (18v°6). More arresting are instances of overreaction, like dest rado 'platform' ( 3 3 r ° l l , 45v°8), as against Sp. estrado STRĀTUM, lit. 'sth. stretched-out, spread'. 43 Desenp°lo a n d deštarado (see t h e preceding fn.) represent near-parallels, except t h a t in EXEMPLUM t h e E - w a s etymological, while in STRĀTUM a n ancillary vowel was introduced. 44 I refrain from discussing here t h e apheresis of bi- a n d tri-phonemic syllables. Easily t h e bestk n o w n case, coll. P t g . mano ' b r o t h e r ' , is complicated b y t h e suspected link t o Spanish a n d b y t h e general t e n d e n c y of kinship t e r m s (effect of child language?) t o begin w i t h a consonant, cf. OFr. ante (—* E . aunt) > m o d . tante.


339

on the Romance level, in general, and, particularly, in Luso- and HispanoRomance, medieval and modern.45 Distinctly noteworthy, in this context, is OPtg. i- ~ in-guar, because it once more dramatizes the nasal epenthesis and the borrowing of a rising diphthong from a congener as being, in the last analysis, alternative answers to the same basic need. Yet another attempt at syllabic restructuring is observable in Ptg. her-dança 'bequest, heritage', her-dar 'to bequeathe, inherit', her-deiro 'heir'46 as against Sp. (semilearned) he-ren-cia, he-re-dar, he-re-dero, and Fr. (heavily learned) héritageIhérédité, hériter, héritier — with offshoots extending into English. Despite its vulnerability, the Spanish form was protected by its status from any such distortion as *redar or *arredar, the more so as the coexistence of red 'net' RĒTE, enredar 'to entangle, hamper' would have immediately blocked any such development.47 In sum, though the aggregate of flaws in Ascoli's high-aiming note tempts one to re-write it in its entirety, it remains a landmark in the history of scholarship. Its pervasive weakness is the author's failure to realize that the broad theme of his exploration was the apheresis of a subclass of vowels in certain vulnerable positions as well as the, minimally, half-dozen alternatives to apheresis; the case histories he focused upon represent, in the aggregate, just one particularly conspicuous and neatly detachable variety within a whole spectrum. The strength of Ascoli's paper, in the perspective of present-day knowledge, lies in the recognition of a basic difference between the reactions of front and back vowels, when exposed to this particular hazard; and in the ingenious inter twining of the vicissitudes of the (expanding) prosthetic vowels and of the (retreating) inherited vowels endangered by the trend toward apheresis. The one general impression that the reader carries away is that of the marked frailness of the unstressed word-initial vowel, especially if that sound happens to be a front vowel which of itself constitutes the opening syllable. 3. Darmesteter's "Law". One recently-published paper — of considerable, though indirect, relevance — that Ascoli neglected to take into account had been contributed by an aspiring young Frenchman, Arsène Darmesteter: "Phonétique française; la protonique non initiale, non en position", Rom., V 45 L. Spitzer chose to see in Ptg. iguaria 'tidbit', OJud.-Sp. yegüería 'mess' a reflex of AEQUĀRE, whereas I recognized as the word's kernel IECU(R)ĀRIA 'giblets, liver dish', recorded in a gloss. As a severely isolated relic, the descendant of IECU(R)ĀRIA could, of course, have undergone the secondary influence of AEQUĀRE. On this controversy see Lang., XX (1944), 108-130; X X I (1945), 98f. and 264f.; XXII (1946), 358f. 46 Somewhat similar to the treatment witnessed in herdeiro is the transmutation of 'heliotrop' into OSp. eltropio (Confisión del amante, ed. H. Knust, 297r°); Gower's text was, incidentally, transmitted through a (lost) Old Portuguese version. 47 One more possibility of retreat from an embarrassing situation involving an unattached vowel comes to mind: complete lexical loss. Thus, OSp. femencia, fimencia < arch, fimiença 'religious ecstasy', 'fervor, ardor' VEHEMENTIA X FIDE had no difficulty in holding its own until, through loss of/- > h- > 0, the word dwindled to emencia, ymencia Rimado, E, 1391b, 1453c, as against C, lb: femencia), at which point it began to decay. This consideration is to be added to those presented at earlier junctures: RR, XXXV (1944), 307-323, at 319-323; UCPL, I:4 (1945), 121; RLiR, XVIII (1954), 161-171, at 162-164.

340

ACCENTOLOGY A N D PHONOLOGY

(1876), 140-164.48 The title that G. Paris' star pupil had chosen for the an nouncement of his discovery was, admittedly, a bit clumsy; by way of short-cut, fellow scholars — including those toiling in neighboring fields, like the Provençalist O. Schultz-Gora — started referring to his main findings before long as "Lex Darmesteter". In a way, the credit thus given to a relative beginner was more justified than the parallel acknowledgment of Ascoli's preëminence, inas much as Darmesteter's formula, unlike Ascoli's, was actually compatible with the newly-pressed demand for uniquely valid cross-temporal relationships as defined by Neogrammarians;49 also, Darmesteter's Law, again in contradistinc tion to Ascoli's, accounted for the behavior of a sizable number of words, all of them confined to Old French. Strictly speaking, Darmesteter was not at all concerned with word-initial syllables, whatever their configuration. His topic, translated into more modern jargon, was the fate of the free intertonic syllable that preceded the main stress, as in NIĀ'goodness' or, to cite his own favorite example (143n1), 50 SACRĀMENTUM 'oath of allegiance, solemn promise'; not, by any chance, LABŌREM. Because, however, "tout se tient" in phonetic and, particularly, in accentological analysis, one is free to infer a high degree of mutual conditioning between, say, the - and the -NI- segments of the selected example. The central thesis of Darmesteter's Law contends that the intertonic vowel immediately preceding the main stress, assuming a free syllable is involved, develops exactly like the vowel of the final syllable. To put it differently, the vicissitudes of the ï in BONITĀTE and those of the Ā in SACRĀMENTUM are precisely the same as if BONI and TĀTE, SACRĀ and MENTUM were two separate words. The implication of this equality of results is near-equality of the driving forces 48 I n c l u d e d in t h e Reliques scientifiques recueillies par son frère [ J a m e s D.], 2 vols. (P., 1890), I I , 95-119. 49 D a r m e s t e t e r ' s almost reverential a t t i t u d e t o w a r d linguistic " l a w s " can be inferred from t h e titles of some of his writings; a case in point is his well-known monograph, De la création actuelle de mots nouveaux dans la langue française et des lois qui la régissent (P., 1877). I n this respect t h e a u t h o r reminds one of t h e scientism of t h e Belgian pioneer A. Scheler, who titled r a t h e r preten tiously his — all told, modestly progressive — dictionary: Dictionnaire ď'etymologie française d'après les résultats de la science moderne (Bruxelles, 1862; P . , 1873 2 ; Bruxelles, 1888 3 ). 50 Before D a r m e s t e t e r ' s arrival on t h e scene, t h e chief rival analyses e x t a n t were as follows. A. Brachet, " D u rôle des voyelles latines atones d a n s les langues r o m a n e s " , JRL, V I I (1866), 301-315, also Dictionnaire étymologique de la langue française (P., [1868]), Preface, p . lxxxi, a n d s . w . accointer a n d aider —, argued t h a t , in t h e given slot, long vowels were m a i n t a i n e d a n d t h e short ones dropped. Conversely, J . S t o r m , " R e m a r q u e s sur les voyelles atones d u latin, des dialectes italiques et de l'italien", MSLP, I I (1872), 81-144, asserted t h a t t h e clinching condition for t h e survival of an endangered unstressed vowel was t h e association of its lexical carrier with some word in which t h e same syllable was under h e a v y stress. N . Caix's severe critique of S t o r m unleashed a t e m p e s t in t h e Rivista Europea (1874-75); see R . A. Hall, J r . , BIL, Nos. 312-316. After establishing a long list of glaring errors (most of t h e m quite elementary) in B r a c h e t ' s allegedly supporting examples a n d after refusing t o project onto t h e level of t h e p a r e n t language suffixal derivatives which common sense would assign t o t h e O F period (see esp. 1427n3), D a r m e steter forcefully s t a t e d his own " l a w " in just five lapidary lines (p. 143 t o p , followed b y a n almost equally laconic technical elaboration): Nous allons essayer d'établir que le sort de la protonique en français repose non sur la quantité, mais sur la qualité de la voyelle, non sur sa durée, mais sur son timbre, tout comme pour l'atone finale; que l'accent tonique divise le mot en deux moitiés, et que les voyelles finales de ces deux moitiés sont soumises à des lois de même nature.


341

behind them: The syllable ò is likely to be almost as strong as Ắ, and the same holds for SÀ and MÉN, respectively. The word-initial syllable emerges from this analysis as, presumably, very strong in Old French, provided it is two steps removed from the main stress. The crucial consideration, however, — and here one is tempted to transcend the self-imposed bounds of Darmesteter's observation — is that even where a free syllable immediately before the main stress is involved, that syllable, judging from the treatment speakers meted out to its vowel in the transition from Latin to Old French, must have been very firmly articulated and deserves to be credited with a counterstress; this inference holds even if such a syllable comprised just one vowel, regardless of whether that vowel was articulated in the front, in the center, or in the back of the mouth. At this point the ominous specter of partial conflict between Ascoli's "Law" and the wider implications of Darmesteter's "Law" begins to loom. Not only did an initial vowel seemingly as vulnerable (according to Ascoli) as was the e- of OFr. eage 'age' *AETĀTICU (from AETĀS -TIS) remain safe from apheresis,51 but speakers, for centuries, distinguished rather tidily in such pretonic syllables between several qualities of monophthongs, although they might conceivably have reduced all such vowels to a sort οf Murmelvokal // spelled e-, as they demonstrably did, for instance, word-finally, where the reflexes of all such vowels as were, for one reason or another, protected from complete erosion coincided in a kind of neutral minimum vowel: FĒMINA feme 'woman', HOMINE ome 'man', GENERU gendre 'son-in-law', etc. In the word-initial pretonic syllable, by and large regardless of the specifics of its configuration, a fairly rich gamut of distinctive vowels was preserved beyond the peak of the Middle Ages — much as it was in the countertonic syllable, at two removes ahead of the main stress, in harmony with Darmesteter's startling discovery.52 Thus one encounters, in typical 12th-century texts, side by side: (a) an /ü/: jurer, Huon (reflecting either Lat. ü or its closest Germanic counterpart); (b) an /o/: (a) j ornée, mostier, sovent, (β) colour, dolour, reflecting either Lat. u/õ, or ŏ; (c) an /a/: äost, avril, baron, bataille, f arine, laron, naïf, pareit, saol, valoir, reflecting Lat. A (the failure of t h a t vowel to yield ground to e is particularly conspicuous in the broad context of Latin-Oïl phonological relations); (d) an /e/: (a) eage, leece; (β) crever, legier, seist; (γ) peser, reflecting either AE or Ě, or else Ě, in this order; (e) an /i/: citez, iver, reflecting ancestral Ī. 5 3 51 The intermediate link between OFr. eage and mod. âge was, incidentally, MiFr. aage; the gambit thus involved vowel assimilation followed by foreseeable contraction rather than spon taneous apheresis of the e-. 52 I presented the gist of this section of the paper at the Naples Congress of the Société de Linguistique et Philologie Romane, on April 15, 1974, under the title "Critères pour l'étude de la fragmentation du latin". 53 I have extracted most of my illustrations from the pertinent paragraphs of E. Schwan and D. Behrens, Grammatik des Altfranzösischen, and from G. Rohlfs' revision of K. Voretzsch's classic, Einführung in die altfranzösische Sprache (the section is somewhat clumsily titled: "Die neben tonigen Vokale in nichthaupttonigen Anlautsilben"). The information provided by Voretzsch

342


This diapason of five vowels carefully kept apart except under a few neatly definable conditions (before a nasal: DOMINIU 'rule, power' denteine, also in a checked syllable: TRUNCĀKE 'to shorten, maim' trenchier; in hiatus: A(U)GURIU 'augury, omen' eür, MĀTŪRU 'ripe', 'speedy, early' meür 'ripe' 54 ), in conjunction with similar conditions prevailing in countertonic syllables collocated farther away from the main stress, but still word-initially (as in: BÒNITATE), produced the impression that in Old French it was not the relative distance from the main stress that counted most, but rather the absolute fact of word-initial position. Some scholars were so impressed by this discovery that they extended its impli cations to southern Eomance languages (as is true of J. Cornu, whose original concern was with Franco-Provençal, but who later switched to Old Spanish and Old Portuguese), and even E. Menéndez Pidal and his numerous followers in the end adopted this criterion, which, at least in my view, should never have been extended to Hispano-Romance. 55 Just as Meyer-Lübke's generation erred in attempting to overextend Verner's Law,56 so — at approximately the same time — did Cornu and the Madrid School slip in generalizing Darmesteter's formulation and certain implications of and inferences from those findings, which were astonishingly accurate as far as Old French was concerned and to a certain extent were applicable to Provençal, but failed to lend themselves to rash generalization. 4. Germanic Pressure on Gallo-Romance Prosody ? Is it, on balance, correct or, at least, permissible to invoke a latent conflict between Ascoli's Law and Darmes teter's Law? Certainly not as long as one takes either "law" at its face value. Ascoli was concerned, as he never tired of repeating, with minimally-stressed word-initial vowels — subject, by definition, to all sorts of hazards in the free flow of speech; Darmesteter was interested, to a certain extent, in weaklystressed word-medial vowels, collocated inside the shells of polysyllabic words, hence effectively protected against any impact or abrasion through contact with the terminal segment of a preceding word. Ascoli concentrated his attention on front vowels; Darmesteter recognized the privileged status of A, but justifiably — within the chosen context — lumped together front and back vowels. Ascoli, a full-fledged comparatist, viewed the development selected for microscopic inspection in a bold pan-Romanic perspective, though, perhaps inadvertently, paid less attention to northern France than to any other region; Darmesteter, more modest in the range of his professed curiosity, but endowed with perfect is exhaustive, but his inability to tighten the presentation obfuscates the picture and hinders the neophyte, for whom the textbook was composed, from grasping the essentials. 54 One of the few cases left unaccounted for is that of GAL(L)ĪNA 'hen' > geline; the impulse for fronting the A could have come from initial /ž/ or /ğ/. 55 I have tried to cope with this problem in my article, "Toward a Unified System of Classifica tion of Latin-Spanish Vowel Correspondences", RPh, XVI, 153-169. 56 See my note, "Quelques fausses applications de la 'Loi de Verner' aux faits romans", CFS, X X I I I (1966), 75-87 ( = Mélanges A. Burger). Add the evidence presented on the OSp. sibilants (in criticism of Meyer-Lübke's earlier hypotheses) in RPh, XXV, 1-52, and, most recently, in "Etiological Studies in Romance Diachronic Phonology", ALE, XIV:2 (1973), 201-242.


343

vision, purposely centered his lens solely on Oïl territory. (One final remark: Ascoli produced an immensely stimulating note open to severe criticism; Darmesteter proffered a perfectly-worded hypothesis, so carefully circumscribed that it invites little further discussion.) Nevertheless, in the end one discerns a conflict between Ascoli's Law and important IMPLICATIONS and EXPANSIONS of Darmesteter's Law, beyond the

confines of its original formulation. The single presupposition of all develop ments envisaged by Ascoli is the striking weakness of the unstressed wordinitial front vowel, including a monophthongized diphthong, as in (AE)QUĀLE, (AE)STĀTE. Conversely, the one condition which makes Darmesteter's Law understandable is the remarkably heavy stress that must have hit the - of BÒNITÁTE, regardless of its consonantal onset; an equally heavy stress, to embroider on Darmesteter's observation, must have marked, in local pronun ciation, the LA- of LÀTRŌNE and the (A)E- of *ǼTĀTICU, judging from OF làrón, éáge. The simplest explanation of this prosodic paradox, in spatio-temporal terms, is the hypothesis that here again, as in so many other respects, medieval Northern France shows the impact of a Germanic language (Frankish), with which the local variety of provincial Latin was for several centuries in a state of connubium. While the prevalence of protracted bilingualism is no longer questioned, and while Darmesteter's Law has enjoyed, over almost a full century, virtually unanimous and unqualified recognition,57 its implications have hardly ever been pressed into service in an interpretive mood for the simple reason that the leading investigators of Germanic-Romance symbiosis, especially W. von Wartburg, 58 have concentrated their attention on the domain of stressed vowels, to the practical exclusion of their unstressed counterparts. Yet it suffices to 57 For a masterly summary and an excellent digest of later discussions — by E. C. Armstrong, W. Förster, A. Horning, F. Neumann, G. Rydberg, and others — of certain significant details (mostly chronological and stratificational), see §§126-130 of W. Meyer-Lübke's HGFS, I: Laut und Flexionslehre2-3 (Heidelberg, 1913); parallel treatments of varying originality are found in K. Nyrop, GHLF, I (Copenhague, etc., 1899), §§254-256; and in other standard manuals. The highly idiosyncratic fact that the countertonic a of, say, RĀDĪCĪNA 'root' racine and the pretonic a of MARĪTU 'husband' mari were treated alike in proto-French is inadequately stressed — indeed, unjustifiably taken for granted — in statements like this one: "La syllabe initiale d'un mot latin possède un accent secondaire qui permet à la voyelle de subsister en français" (F. Brunot and C. Bruneau, Précis de grammaire historique de la langue française3 [P., 1949], §55). The consistent survival of the original word-initial segment of mari — in preference to *meri — is doubly striking if one recalls that the propensities of Gallo-Romance speakers were split with parental post-tonic A in proparoxytones, e.g., lazdre 'leper' LAZARU (beside cosdre /kozdrə/ 'to sew' cõ(N)s(u)ERE),jatte 'bowl, platter' GABATA (beside dette 'debt' DĒBITA), Estiefne 'Steve' Gr.-Lat. STEPHANIT (beside juefne 'young' IUVENE), the dendronym plasne 'plane-tree' PLATANU (beside resne 'rein' *RE-TINA) — all of them placing A on the same level as the other vowels — pitted against ane 'duck' ANATE, foie 'liver' *FĪ-, *FĒ-CATU, lampe 'lamp' LAMPADA, top. Ourche ORCADA, all of which conceivably point in the opposite direction (preservation of A as e), see Meyer-Lübke, HGFS, I 2 -, §121. 58 On Wartburg's 1936 article and 1950 book, both bearing the title Die Ausgliederung der romanischen Sprachen, and on a miscellany of less familiar companion pieces from his pen, see my remarks in "Comparative Romance Linguistics", TL, I X (1972): Linguistics in Western Europe, pp. 835-925, at 863-868 and fnn. 95-99; also my paper presented at the Naples Conference (see fn. 52, above).

344


observe the accentual changes undergone by French words in the mouths of practically any group of Germanic speakers (Englishmen included)59 to realize that the main trend has invariably been toward a shift of stress in the direction of the word-initial syllable. Even where the goal was not reached, that syllable, in the process, gained in strength as regards both intensity and neatness of articulation. Thus Ascoli and Darmesteter were, in a way, both right — yet each one stopped short of thinking through his argument to its last possible consequence. One strong objection to the hypothesis here advanced that could conceivably be raised would point to the fairly thin spread of Germanic settlers over southern France, a situation that at first glance seems to conflict with the claim of the validity of Darmesteter's Law, as interpreted in this paper, over the whole domain of Old Provençal. However, the two-case system in certain categories of nominal declension also straddles Old French and Old Provençal and yet has never been more persuasively explained than through appeal to the symbiosis of Germanic-speaking and Romance-speaking nuclei of inhabitants. Although such innovative or conservative peculiarities (shift of accentual strength, resistance to morphological decay) are most likely to have crystallized in the North, they can perfectly well have subsequently extended powerful prongs toward the South, through selective diffusion.60 Is the emphasis on the pretonic or countertonic syllable, provided it is wordinitial, observable through any prisms other than those of vowels in polysyllabic formations? Conceivably, the characteristic stress on the personal object pro nouns at the beginning of a syntagm, as in Roland, v. 2834: "i ai perdut et trestute ma gent", is likewise traceable to a Germanic style of underscoring a vitally important ingredient in a given context; one is further reminded of the 59 This tendency breaks through in different ways; the original number of syllables may be pre served, but their hierarchy reversed (as in Fr. petit > E. pétty, including the widespread pronun ciation, clearly observable in America, of pétits fours; and in Ch.-Gr. papas, acc. papan 'minor cleric' > G. Pfaffe, according to F. Kluge, EWDS18, rev. W. Mitzka, Berlin, 1960, s.v.); or, in a Germanic language, there crystallizes a monosyllable reflecting a Gr.-Lat. or Romance poly syllable stressed in the middle or at the end, as in Pferd 'horse' from M.-Lat. parēarns, from hybrid PARÁ VERĒDU, also in Pfühl 'soft cushion' < PULVĪNU; radical shortening presides further over Pfingsten PENTĒKOSTÊ and Pfründe PRAEBENDA. Noteworthy is the prosodic bifurcation (a) E. pillar, G. Pfeiler vs. Fr. pilier (orig. piler), It. piliere (a likely Gallicism), Sp. pilar, from *PĪLĀRE based on PĪLA 'pillar, pier'. The difference between Chaucerian English and modern English as regards the stress of words borrowed from Old and Middle French is well-known; so is the local peculiarity that nouns and verbs can be differentiated through diverse responses to the temptation of a stress shift (the object ~ to objéct, though not yet in the case of consent) — here colloquial English often transcends the sanctioned distribution of doublets (tendentially, esp. among the younger speakers, the résearch ~ to research). See the classic analysis by O. Jespersen, A Modern English Grammar on Historical Principles, I: Sounds and Spellings (Heidelberg, 1909), pp. 160-186. 60 I realize that, in discussions of sub- and superstratal relationships (the classic case being that of the controversial Etruscan involvement in the gorgia toscana), the near-coincidence of the territories of the Wo suspected partners, Language χ and Language γ, has played a crucial rôle. While it is true that areal coincidence sharply raises the probability of such influence, I do not view it as a clinching argument, because diffusion, under favorable circumstances, could have secondarily modified the contour of the territory affected.


345

infinitives of reflexive verbs: soi aler = aler s'en 'partir', soi apercevoir 'prendre conscience de...', soi desjeüner 'se faire sortir de l'état de jeûne', soi escharguaitier 'se tenir sur ses gardes', etc. 61 The link to modem French emphatic stress, which hits the first syllable beginning with a consonant: épatant, embarrassant, involves a more remote possibility, for the simple reason that the early stages of this conversational idiosyncrasy elude close observation through extant texts. 62 Ascoli's and Darmesteter's "laws" are not, then, mutually exclusive — though Ascoli's note would indisputably have gained in weight through closer attention to the peculiar, not to say erratic, state of affairs in Old French, if that situation is measured by the standard of "average Romance". In fairness to Ascoli it should be borne in mind that he intended to write a note, by definition an appetizer, and not an exhaustive monograph on apheresis, the acute need for which has dawned on scholars of our own generation. The paradox consists in that Old French — the first Romance language studied with any degree of thoroughness in the 19th century — turns out, the closer it is examined from our own vantage point, to have been the least characteristic of all descendants of Latin.

61 See, among many other sources available, G. Moignet, Grammaire de Vancien français: morphologie-syntaxe, "Initiation à la linguistique", Série B, I I (P., 1973), pp. 185-187, from whom I have borrowed several illustrations; and R.-L. Wagner, L'ancien français: Points de vues, programmes (P., 1974), pp. 191, 199. Here are a few additional illustrations of emphatic and con trastive use: "...se deniers avoie / moi et vous en aisseroie"; "aus perdirent et nos salverent"; "et moi que chaut?". I t is, of course, true that in certain positions AFTER the verb the stressed forms of the pronouns were also used, as in: "Mesfaire soi por nul desroi" (Beroul, 820), cf. to this day: Tais-toi! A more "typical" Romance language, such as Old Spanish, would use only un stressed forms in such context (me, not mi, etc.). 62 Cf. M. Grammont, Traité pratique de prononciation française (P., 1948), pp. 139-151; K. Nyrop, Manuel phonétique du français parlé (Copenhague, etc., 19234), § 141 : 2; and many notes: by Marouzeau (BSLP, XXV), Dauzat (FM, IX), and others.

EXCURSUS A: GENESIS AND SPREAD OF THE SEGMENTS ens-, enx-

The rise of the prefixai segments ens-, eux- ( = mod. enj-) in Hispano-Romance, particu larly in Spanish, can be cogently placed in a context very different from the framework into which Ascoli attempted to squeeze this process a century ago. The alternative context here appealed to is the epenthesis of the nasal ("Nasaleinschub"). I t may be useful to review the principal varieties and ramifications of this phenomenon before assigning a specific niche, within t h e total edifice, to the parallel shifts es- > ens-, ex- > enx-.A The insertion of t h e nasal may involve: (a) a phonological process, (b) a morphological process ("Formenmischung"), () a cross of near-synonyms, (d) a confusion of near-homo nyms, or (e) false restitution. The last variety is, clearly, of a sociolinguistic nature and pre supposes a volitional act, a deliberate choice b y speakers of a speech community whose members are familiar with different educational levels or stylistic registers of t h e given A For full bibliographic data see my paper, "En torno a la etimología  evolúción de cansar, causo, cansa{n)cio", NRFH, I X : 3 (1955), 225-276, esp. 240-242, 265-275.

346


language. Conversely, the remaining varieties, (a)-(d), cast the speakers in far less active rôles, and the changes are likely to have occurred at lower plateaux of consciousness, (a) Where the insertion has been phonologically motivated, one can distinguish between: (a) The echoing of a preceding nasal: MULTU 'much' > Hisp.-Rom. muit(o) > P t g . /mữjtu/ beside Sp. mu(n)cho 'much' ~ muy 'very'; MACULA 'spot, mesh of a net' > P t g . mazela 'sore spot, blemish', OSp. manziella 'blot' > mod. mancilla; OSp. malatia 'ailment, sickness' (based on MALEHABTTUS lit. 'ill-disposed', see REW3 5264) > dial, malantia; IN + *DĪRĒCTIĀRE 'to send off in the right direction' ( χ DĒRĒCTU 'straight') > OSp. endere(n)çar and, through further infiltration of -n-, aderençar; NĪDU 'nest' > OSp. nio (cf. n-iego or ni-ego 'nestling') ~ nido, but Ptg. ninho; IN-TRĪC-ĀTU 'placed in an embarrassing situation' > Sp. intrincado beside E. intricate, I t . intrig-are, - —> Fr. intrigu-er, -. (β) The anticipation of a following nasal: *EX-AMPLÄRE 'to widen' (EXAMPLIFICĀRE is on record) > OSp. e(n)sanchar, and, through extension of ens-, also EXAGIU 'balance, weigh ing' > Sp. ensayo 'test, assay, essay'; EXEMPLU 'sample, example, precedent' > OSp. enx(i)enplo, OPtg. enxenpro 'exemplary story, anecdote' beside ensienplo through contact with words like ensayo; OSp. fazendera adj. 'industrious', subst, 'working woman, public work', etc. > dial, fanzendera; OSp. gerigonça 'jargon, slang' > mod. jerigonza, var. jeringoza (cf. REW3 3685); PALĀTIĀNU 'relating to the Imperial Palace on a Roman hill' > OSp. palaçiano (beside palaciego) 'courtly', var. palanciano; Gr.-Lat. PARAKÖNE 'grind stone' (REW3 6226) > Sp. parangón 'model, match, comparison', as against It. paragone,, E. paragon; RĒGULA 'rule(r), plank' > Sp. (semilearned) regla —> religión 'line'; (γ) Joint echoing and anticipation: (MĀLA) *MATTIĀNA '(apple) from the orchard of Mattius' > OPtg. maçãa > maça (but note madeira 'apple-tree') beside OSp. maçana > manzana; NEC ŪNU 'not even one' > OSp. neguno (hence, through dissimilation, Ast.-Leon. deguno), later nen-, nin-guno beside Ptg. nenhum. (b) The cases of morphological confusion, typically, involve verbal families, at whose center one finds a verb endowed with a nasal "infix", real or apparent, restricted to the present-tense stem (FRANGŌ 'I break' vs. FRĒGĪ ' I broke', FRĀCTU 'broken'; PREMŌ ' I press, squeeze' vs. PRESSI ' I squeezed', PRESSU 'squeezed') or, at most, to the combination of present tense and perfectum (PINGŌ ' I paint', PINXĪ 'I painted', PICTU 'painted'). Cf. Sp. pintura, F. peinture 'painting' vs. It. pittura; Sp. pintoresco vs. It. pittoresco (-> F. pitto resque, and, with a thin veneer of classical learning, E. picturesque); Sp. prensa 'press, clamp, vice, mill, printing press, press media' vs. imprenta 'print(ing)', 'printing-office' (vs. t h e Ptg. compromise form imprensa), as against OFr. presse, whose vernacular cognate in Spanish has been pri(e)ssa 'haste', lit. 'pressure'; OPtg. xofrango 'osprey' ( = Sp. quebrantahuesos), from OSSIFRAGU lit. 'bone-breaker' (from the killing habits of this big bird of prey), Via *(0)SS0FRAGU X FRANG-.

(c) The accurate tracing of blends of near-synonyms would plunge us deeply in wordbiographies. Suffice it to state t h a t ALAUDA 'lark' (OSp. aloa beside dial, aloda, aloya, cf. REW3 313), HIRUNDINE 'swallow' (which, structurally, reminds one of SANGUINE 'blood' > OSp. sangne > sangre alongside INGUINE 'groin' > Sp. ingle), and Gr. KALANDRA 'sorte d'alouette (huppée?)' (Ernout-Meillet, D E L L 4 , s.v. CALIANDRUM) = G. 'Haubenlerche' (cf. REW3 1486) all three influenced each other. Cf. the result in standard Spanish: alondra, golondrina, calandria, with the -ndr- cluster serving as a common denominator or unifying thread. (d) Counter to expectations, it does not always prove easy to segregate homonyms from synonyms, simply because it is pairs and triads of such synonyms as share certain con spicuous features of form t h a t are most likely to exert reciprocal influence; thus, ALAUDA /alawda/ and CALANDRA (accented the Latin way) each show intervocalic L and postconsonantal D, and the cross of plátano 'plane tree, sycamore, banana' and planta 'plant', which is responsible for dial, piántano, can be similarly interpreted, except that here multiple


347

causation is made even more patent through the additional agency of anticipation. Homonymy seems to have acted in isolation in such difficult cases as *cassar 'to fatigue, wear out, break down' (through repeated shaking) < (intensive-iterative) QUASSARE X cansar 'to dodge, shun' — cf. I t . (s)cansare —, of Graeco-Latin nautical background (CAMPSĀRE 'to circumnavigate'), a blend from which Sp. Ptg. cansar 'to tire' eventually sprouted; OSp. cansacio (based on t h e medical term QUASSĀTIŌ 'break-down, collapse, exhaustion') > mod. cansancio involves echoing. A comparably intricate case is t h a t of Sp. fincar, mod. hincar 'to thrust, stick, poke, dig in, drive into', which seems t o represent an instance of age-old confusion between (a) ficar (abundantly preserved in Portuguese: 'to stay, endure, pause'; cf. It. ficcare 'to thrust') < *FĪGICĀRE, from FIGERE 'to fix, fasten, attach' and (b) the family of FINGERE 'to shape, fashion, mould' (from which the obsol. Sp. heñir 'to knead' also branched off, as an autonomous side-line). FIGERE was flanked b y perf. FÏXÏ, p . ptc. FĪCTU/ FIXU [fīksu] ; t h e paradigm of FINGERE included t h e key forms F I N X Ī , FICTTJ (cf. FIG-ŪRA,

FIG-MENTUM). This degree of similarity in certain sectors of the respective paradigms ex plains how the nasal leaped from one family to the other. (e) False regression or restitution, appealed t o frequently b y workers in other Romance domains (in explanations of, e.g., R u m . opt 'eight' ō), played a certain part in HispanoRomance as well. Even if one dismisses as widespread whimsical graphies Late Lat. occansio and thensaurus, in lieu of occĀsIō 'opportunity' and Gr.-Lat. THĒSAURUS 'treasure, store, hoard; treasury', there remains t h e case of OSp. fonsario 'tomb, grave' FOSSĀRIU 'gravedigger' (semantically influenced by OSS(U)ĀRIU, cf. Fr. ossuaire 'charnel-house' ?) and, above all, OSp. fonsado 'army, expeditionary force' FOSSĀTU 'ditch, trench', best understood if placed in t h e context of fluctuations like MĒ(N)SA 'table', MĒ(N)SE 'month', TRA(N)S- 'be yond, behind', etc. Moreover, given the tendential reduction of -LS- and -RS- to -(s)s~, as in (a) ĪNSULSU 'insipid, tasteless' (lit. 'unsalted') > (en)soso and (b) MORSU 'bit' > mues(s)o, VERSU 'verse' > OSp. vies(s)o, and TRĀ(NS)VERSU 'oblique' > Sp. travieso 'naughty', there arose, through confusion with (c) the -(N)S- imbroglio, such regressions as URSU 'bear' > Sp. oso, Arag. onso, Ptg. (learned) ursoB and (ĪN)SULSU > Am.-Sp. sonso (zonso) 'silly'. Rivalry between -ns- and -ls- as parallel instances of regression from -ss- is observable in dial. (Salm. Segov.) sienso, sielso 'buttocks' SESSU lit. 'seat', 'act of seating oneself' (Apuleius), except t h a t the rise of sienso could, additionally, have been stimulated b y coexistence of (a)siento, from (a)sentar 'to seat' *SED-ENTĀRE ;C t h e adj. SEDENTĀRIUS, though rare a n d non-classical, is recorded. Note also vestigial OSp. siesto, a near-synonym. This handful of facts, seen in bird's-eye view, suffices to raise or sharpen a number of points. (1) Although a thin thread, indeed, leads from the Romance trajectories of the EX- words to apheresis (cf. E . sample EXEMPLU, I t . saggio EXAGIU), vindicating in part Ascoli's ana lysis, a much stronger link ties forms like OSp. enxiemplo, mod. Sp. ensayo to the aggregate of instances of nasal insertion, with the closest fit under (a) phonology, (β) anticipation (as above). Here words like ensanchar *EX-AMPLĀRE, enxuto 'dry' EXSŪCTU lit. 'sucked out', 'deprived of its sap or juice' smoothly fall into place. B The case of urso is noteworthy, because it presents a perfect parallel to the equally puzzling introduction of the strictly learnèd form dulce 'sweet' in Spanish, a choice which seems inexplicable at first glance against the background of Ptg. doce, I t . dolce, etc., all of them transmitted in the vernacular layer. The medieval record shows an astonishing profusion of vars. (duz, duce, doz, doce, etc.), an excess which may have become so irritating to speakers as to prompt them to toy with the remedial adoption of an unassailable Latinism. For further ideas on this score, see my paper, to appear in the memorial for Fernando Antonio Martínez (Bogotá): "La vacilación fonética como causa de una pérdida léxica; en torno al desarrollo de GAUDĒRE y GAUDIUM en hispanorrománico". c See S. N. Dworkin's forthcoming note in EPh: "The Etymology of Old Spanish siesto: A Return to the Family of SEDĒRE".

348


(2) Genetically, ens- stands much nearer to es- EX- than it does to en- IN-; ensanchar is, initially, a mere variant of esanchar, fully comparable in status to mançana as a variation on, or a satellite of, maçana. In a modern dialect, however, innovative enanchar 'to widen' can spring into existence, sporadically and spontaneously, and has indeed done so, on the model of, say, em-bot-ar 'to blunt', en-rubi-ar 'to bleach', en-tihi-ar 'to temper, make luke warm' (in the standard enanchar, despite a late-medieval prototype, still ranks as a collo quialism). Where (experimental) enanchar and (older, "authorized") ens-anch-ar are pitted in competition, ens- will sound to speakers and look to writers like a variation on en- (cf. Lat. ABS- beside 1-, AB-; also OBS- beside -). The synchronic problem then arises as to whether to class ens- as an autonomous prefix, or to declare it an allomorph of en-, or else to break it down into the prefix en- and the "interfix" -s-, a decision which would entail operating with a hitherto hazily identified subcategory of Romance interfixes. (3) While it is easy to account, historically, for ensanchar, and to explain the later — and less widely accepted — Sp. ensangostar 'to narrow down', from angosto ANGUSTTT 'short, scarce, narrow' (presence of -n- in the root morpheme, plus pressure of lexical polarization: 'wide' ~ 'narrow'), it is less comfortable to deal with ensalçar 'to raise, lift up' (against OPtg. eixalçar *EXALTIÃRE, from ALTUS), where no nasal in the root can be pinned down as the prime agent of attraction. Here it is the semantic affinity and formal resemblance of ancho AMPLU and alto ALTU t h a t must jointly have triggered the shift. Similarly, one readily understands that -n-, once absorbed into OSp. endere(n)çar, could not so easily have been dislodged as a consequence of prefix change, hence aderençar, with a seemingly unmotivated adventitious -n-. But could we satisfactorily explain ensalçar and aderençar in the event t h a t the words they are modeled on had, accidentally, been excluded from the records ? Are we actually certain which word ensayo was patterned on ? (4) While the segments ens- and enx- go back to separate sources, either neatly identifi able, e.g. enseres 'household goods, implements' ESSE(RE) 'to be' (substantivated and pluralized) beside enxambre 'swarm (of bees)' EXĀMINE 'swarm, throng, shoal', 'tongue of a balance', the widespread interchangeability of s(s) and x /š/ in Old Spanish, whichever its ultimate cause, D accounts for numerous switches, especially in the direction enx- > ens-, to the tune of stylistic or sociolinguistic overtones (s doubtless ranked as more refined t h a n x /š/). The switch occasionally involved a clear-cut decision (ensayo before long evicted *enxayo); cases of long-drawn-out wavering are likewise on record, e.g. ensienplo ~ exienplo (and vars., see below). (5) A question which may be formulated but cannot be answered before a bigger chunk of material becomes freely available is whether certain phonotactic environments, e.g. the character of the potentially following consonant, favor or obstruct the intercalation of the nasal. The sprinkling of examples immediately available points in the direction of /k/, /g/, /c/ and /č/ as ideal neighbors of an intrusive, supervenient nasal, hence the popularity of the suffixal segments -ancón, -anzón, -anchón in racy derivatives (e.g. vej-a[n]c-ón 'feeble old man') and of their equivalents geared to other vowels (-incón, -enchín, and the like). The problem invites microscopic inspection. Remembering, then, t h a t the starting points for the ens-, enx- series lay, (a) on one hand, in the preëxistence of inherited words like ĪNSERERE 'to graft in or onto, implant' and

D This is not the place to review the scattered literature on the subject, from R. J. Cuervo's "Disquisiciones..." (1895) to A. Alonso's "Trueques de sibilantes..." (1947) and beyond, often with special attention to the speech of the Moriscos; V. García de Diego's "Dialectalismos", RFE, I I I (1916), 301-318, at 307 ( = §9:2), marks the midpoint. For some preliminary comments see my 1947 note on Sp. cosecha, plus the article, projected in an entirely different perspective: "Derivational Transparency as an Occasional Co-determinant of Sound-Change: A New Causal Ingredient in the Distribution of ç and z...", RPh, XXV, 1-52.


349

of idiomatic sequences such as *IN-SIMUL, *IN-SEMEL, E plus characteristic verbal newformations from, say, sabón ~ xabón 'soap' SAPŌNE and xavorra ~ çahorra 'ballast' SABURRA, F and, on the other, (b) in the above-mentioned mechanism of nasal epenthesis, to which we may add, for good measure, (c) the agency of Ascoli's Law at its purest, as in (em)briago 'drunk' ĒBRIĀCU, G we can very succinctly examine a few relevant case histories, by way of mere sampling. 11 On internal evidence, it seems easiest to posit esanchar (its var. desanchar is to this day preserved in dialects) > ensanchar 'to widen' as the opening gambit, and to assume t h a t esalçar > ensalçar, enxalçar 'to raise', beside enxaltar, followed in its wake; these verbs were accompanied by satellite formations in -ador (agentive) and ֊amiento (abstract), transparent as to structure and meaning. 1 Enxaltar could very well have paved the way for E Enxerir, subject to a metaphony powerful enough to have raised its initial and its medial vowels (inxirió), occurs, e.g., in the Alfonsine General estoria, I, 2536; cf. Escorial Glossary, 2982, and Toledo Glossary, T 1681 and 1991. The p. ptc. enxierto mediated between enxerir and enxertar (El mariscal fñigo, Canc, de Baena, № 576: "...enxiertas / en el árbol del desdón"). On substanti vized Cat. eixart 'graft' see W. von Wartburg, "Die griechische Kolonisation in Südgallien", ZRPh, LXVIII (1952), 18f. (note the loss of -n-). For Fr. ensemble it suffices to operate with IN + SIMUL 'at once, together', as do, among others, the authors of the NDÉH, s.v., who note the rein forcing effect of IN-; however, archaic Sp. ensiemo seems to point to the substitution of SĚMĚL 'once' for SÏMUL, unless one prefers, gratuitously, to posit some such compromise form as *SĚMUL, as did J. D. M. Ford, Old Spanish Readings, ed. 1911, p. xxvi. There existed in Old Spanish, needless to say, numerous other words, of transparent structure, displaying the sequence en + s- : ensalada 'salad' (beside sal 'salt'), ensandecer 'to stultify' (beside sandio 'foolish, inane'), ensangrentar 'to stain with blood' (beside sangre 'blood'), ensartar 'to string, thread' (beside sarta 'string, file'), ensellar ~ ensillar 'to seal' (beside sello 'seal' — vestigially confused with siella ~ silla 'saddle'?), enseňar 'to teach', also -ador, -amiento (beside seña 'sign'), (reřl.) enseñorear 'to seize, become master of' (beside seňor 'master'), (refl.) ensoberveçer 'to become arrogant, embolden o.s.' (beside sobervio), ensonar 'to dream' (beside en-sueno 'dream', as against sueño 'sleep'), ensuziar 'to soil, smear, defile' (beside suzio 'dirty', lit. 'juicy, sappy'). In this series there is not a trace of any tendency to switch from ens- to enx-. The etymology of ensanar 'to madden' raises separate problems (see my paper in HR, XLII [1974], 1-32), but, at least synchronically, ensañar related to saña 'rage, fury' the way enseňar did to seña. F On enxabonar see Ford, OSR, p. vi, and J. E. Gillet, ed. Torres Naharro, Propalladia..., III, 447f., who also records enxabonador; enxavorrar occurs in Documentos linguisticos de España, ed. R. Menéndez Pidal, I: Reino de Castilla, № 23.30. G Embriago: Consejo e consejeros, ed. A. Rey, 109v°; enbriagar: Tratado, ed. B. Maler, 34.3.37. H Additional literature is hidden away in numerous — hopelessly scattered — statements, starting with Gessner's remark (1867) on the scarcity of ens- in Old Leonese (see fn. I, below) and reaching its peak in G. Baist's two successive attacks, apropos Sp. enclenque 'feeble, ailing' and entibo 'prop, shore, foundation': see "Spanische Etymologien", ZRPh, V (1881), 550-553, and "Etymologien", ZRPh, VI, 432-433, with mention of Sp. enjullo 'warp-rod', Ptg. enxalmar 'to cure by spells', Gal. enxeitar ASSECTĀRĪ 'to follow eagerly' (see my elaboration in HR, XVII [1949], 183-232). Almost a quarter-century later A. Morel-Fatio had to remind an inexperienced Menéndez Pidal, on the occasion of Manual elemental...1, that OSp. ensiemplo and ensayo parallel ponçoña and mançana (Rom., X X X I I I [1904], 272). D. Alonso briefly broached the problem apro pos Gal. enxebre (from SEPARARÉ, *-PERÀRE), in CEG, VIII (1947), 536f.; cf. Wagner, "Ety mologische Randbemerkungen...", ZRPh, LXIX (1953), 347-391, at 362! On ens-, enx-almar and on the practitioners of this quackery see Gillet, ed. Propalladia, III, 626-628, 792; though unrelated, ens-, (en)x-alma 'pack-saddle, mattress' SAGMA shows striking parallels. I Ensanchar is old and classical (Astronomia, II, 12; Rojas Zorrilla, Teatro, ed. Ruiz Morcuende, p. 65); but enanchar also occurs in medieval texts (El Corbacho, 21r°), while asanchar in Alfonso Onceno, 1930a, should be emended to ascuchar 'to listen' (see M. R. Lida de Malkiel, La idea de la fama..., p. 228n70). On desanchar see J. Gulsoy, RPh, XIV, 351, in reference to the "Lower Aragonese" subdialect of Villar del Arzobispo, as surveyed by V. Llatas (1959); the local spread of des- at the expense of es-, ex- has not stopped short of invading the domain of Latinisms: desagerar 'exagerar'. A quarter-century ago I drew attention to a- and (d)esas, occasionally, rival reflexes of Ë- (PhQ, XXVII [1948], 117n3lj. Ensanchar is used as the gloss of amplifico and dilato and as a synonym of prolongar in the Escorial Glossary, 2399, 2570. Esalçar figures in A. Castro's edition of Glosarios, s.v. ma[g]nifico;, exalçar, admittedly rare, can be traced to Gil Vicente's Auto dos quatro tempos, cf. E. Asensio, RFE, X X X I I I (1949), 350-375, at 366 (who also,

350


enxalb(eg)ar 'to whiten, whitewash', given the resemblance of ancho, alto, and albo (OSp. Ptg. alvo), the terminal segment to be credited to ALBICĀRE (REW3 321), NIGRICĀRE 'to blacken' (REW3 5920), and the like. J Old Galician-Portuguese wavered between eixalçar and esalçar, eventually supplanted by (Latinizing) exalçar.K The progeny of EXEMPLUM in all likelihood represents the single most exuberantly developed twig. The "ideal" reflex, (a) exiemplo, is found in a few texts, notably those of Navarro-Aragonese background; the monophthong of (b) exemplo can be laid at the door of Latinization, or one can think of x /š/ as having absorbed the adjacent, homorganic / j / ; (c) enxienplo shows the intrusion of the nasal; so does (d) enxenplo, with subsequent loss of / j / , this time indubitably on account of the preceding /š/, cf. () Moz. inšenp°lo; not un expectedly, one detects a trivial departure, namely the substitution of r for l, in (f)-(g) OPtg. e(n)xempro; finally, (h) ens{s)iemplo marks the tendential intrusion of ens- at the ex pense of etymological enx-, as in ensayo EXAGIU. L The Hispano-Romance history of EXAMEN (or -MINE) differs from t h a t of EXEMPLU not so much on account of contrastable formal configurations (these are, on the contrary, very similar, particularly where a buffer consonant changed -mne into -mbre) as by virtue of a split of EXAMEN into two separate units: learnèd EXÄME(N) 'test' and vernacular, indeed rural, EXAMINE 'swarm'. M Enjugar 'to dry, wipe off moisture', refl. 'to grow lean, dry up', once fairly widespread (cf. enxugar in Rey Guillelme, ed. Knust, 38v°; Emperador Ottas, Chaps. 50 and 51 ;Confisión del amante, 133v°), has drifted into such a paradoxical relation to jugo 'juice', orig. xugo at 356, documents enxalçar);, enxalçar is characteristic of the Grail Fragments, ed. K. Pietsch, 258v°, 269r°, 269v°, and of Barlán  Josaphá, ed. G. Moldenhauer, 104v°, 113r°, 131v°, 132r°, cf. Ptg. enxalço 'pequeno arco'; ensalçar was the predominant form in medieval texts (Santa Catalina, 16r°; Plácidas, 25v°, 28r°; Confisión del amante, 108r°, 132r°, etc.) and was flanked by ensalçador (Ottas, xx). Enxaltamiento has figured, since the days of E. Gessner (Das Leonesische..., p. 12), in the discussion of Alexandre, 0, but Moldenhaver's emendation of ensalçamiento in Barlán  Josaphå, 99r°, was pointless, since that derivative, aside from its plausibility, is authenti cated by Confisión del amante, 316v°. J Cf. lobregura, negr(eg)ura, tenebr(eg)ura 'darkness, gloom, murkiness' and related formations, to which I drew attention in Lg., XXV (1949), 139-181, at 159-166, and later in BICC, I X (1953 [-55]), 1-139, at 66, as well as in Estructuralismo  historia, ed. D. Catalán, II (1958), 107-199. K Although M. Podrigues Lapa, in the Glossary appended to his edition of Cantigas d'escarnho  de mal dizer, s.vv., vacillates between exaltar and exalçar, and even though the fuller dictionaries (e.g., C. de Figueiredo's) record and document the latter flanked by offshoots in -açao and -amento, it is not inaccurate to dub exalçar (unlike alçar, realçar) as obsolescent, in strong contrast to Sp. ensalzar. The Latin-Old Portuguese verb dictionary appealed on five occasions, unhesitatingly, to exalçar, as it did to exenprar and exerdar, while it was split in its sponsorship of e(y)xertar and favored, erratically, esugar; see H. H. Carter, RPh, VI, 101. L Exiemplo: Berceo, San Lorenzo, 26d, 74c, Vidal Mayor, ed. G. Tilander, see III, 132, Libro de miseria, 179a, and El Gorbacho, 7r°; exemplo Alexandre, 0 ; enxiemplo General Estoria, II, 184a, Libro de miseria, 2496, Grail Fragments, 258r°, Barlán  Josaphá, 95v°, El Gorbacho, 3v°, 6r°, 8r°, 9r°, etc.; enxenplo Confisión del amante, 21r°, 24r°; 389r°; El Gorbacho, 5v°, 15v°; Consejo  consejeros, 99r°, 99v°, 100v°; Misèria, 216a; J. A. de Baena, "Prólogo" to his Cancionero, cf. insenp°lo in Koran translation, ed.  López-Morillas, and, among derivatives, OSp. enxenplado (El Corbacho, 18r°) and Salm. enejemplar (NRFH, XIII, 263); Ptg. exempro: Letter of Diogo de Gouveia (A.D. 1538) to King João the Third, see M. Bataillon, Misc. Leite de Vasconcelos (Coimbra, 1934), p. 91; en(s)iemplo Poema del Cid, 2731, and Miseria, 133d The substitution of r for l is found, sporadically, also in Spanish, cf. ensiempro in Juan Puiz, 909b (enxienpro occurs in his corpus, too). Cf. J. J. Nunes, "Convergentes e divergentes", BSC, X (1915-16), 835. M Cf., in addition to Sp. enjambre (orig. enx-), the Portuguese doublets e(n)xame (F. A. Coelho, in Rom., I I [1873], 289),  Gal. eixame (Cantigas, 211:6), exambre in the Old Leonese version of the Alexandre (747b in reference to flies, and 957b), and the bizarre Extremeno variant — due to a blend of EXEMPLU and EXAMINE ? — enjamplao 'enjambrado' (A. Cabrera, in BRAE, I I I [1916], 654). Even here /s/ and /š/ are, at certain points, in competition: cf. Burg, an-sambre, -šambre (V. García de Diego, in RFE, I I I [1916], 307).

CONFLICTING PROSODIC INFERENCES,. .

351

SŪCTT/SǓCCU, t h a t the two words are unlikely to be associated any longer. Historically, the state of affairs is transparent once we remind ourselves that, counter to first impressions, not en-, but enx- EX- is involved; EXSÜCÄRE is fortunately recorded, in Caelius Aurelianus; cf. DÉLL4, p. 602b; REW3 3073, 6407); the verb is, strictly, enx-xugar. Of greater import ance t h a n the finite verb has been the congealed ptc. EXSŪCTUS 'dry' (REW3 3074), which underlies OSp. enxuto, seldom exuto, cf. Upper Arag. xuto. OGal.-Ptg. e(n)xuito typifies the transitional link. N Through semantic polarization,  presumably exerted sufficient pressure on IN- or EX-AQUĀRE (cf. It. innacquare, discussed in Excursus B) to have tendentially deflected enjaguar 'to rinse' in the startling direction of enjuagar — a metathesis oftobserved, but seldom if ever satisfactorily explained. 0 There exists a sizable literature on EXAGIUM (REW3 2932) and on the verb which this noun seems to have cast off at an early date — a verb (often endowed with technical mean ings) displaying the speakers' not unexpected vacillation between ens- and ass- ; p the speech community's resolute recoil from enx- in this one instance remains to be fully justified. Further research, which cannot be pressed into the straitjacket of an excursus, should focus on additional Hispano-Latin words, of diaphanous descent, involving EX- before vowel — including the special case of mute  plus vowel, such as the products of *EXHĀLIN Enxuto was common in medieval texts; cf. Barlán  Josaphá, 116r°: con los pies enxutos 'with dry feet'; Confisión del amante, 205r°: pasar por enxuto (in reference to the crossing of the Red Sea by the Israelites), 265r°. Don Juan Manuel's Libro de la caza, ed. G. Baist, offers examples of both enxuto (199v°) and exuto (203r°). Xuto was recorded by J. Saroïhandy at Ansó as early as the turn of the century; cf. . Torres Fornés' remark to this effect, in: Sobre voces aragonesas usadas en Segorbe (Valencia, 1903), p. 190. Ptg. enxuito (reminiscent of dial, fruito FRŪCTU, but less arresting than OGal. eixuito, preserved in one Alfonsine Cantiga [112:10]) ~ enxuto was caught in J. J. Nunes' dragnet of doublets (BSC, X, 828), while enxugar appears in folk lyric ("Veio o sol com seus raios / enxugar quern se molhou", quoted by A. C. Pires de Lima in Homenaje a F. Krüger, I [1952], 189); on one occasion Nunes contrasted innovative enxugado with traditional enxuto (LP, I [1930-31], 72). Regarding Lat. EXSÜC-ĀRE, -ĀTIŌ, -ĀTŪBA; OFr. (r)essuit 'dry, hardened'; It. asciutto 'dry, thirsty', prosciutto 'ham, bacon' [beside prosciugare 'to dry, drain'], see the elabo rations on the relevant entries in Wartburg's FEW by Ch. H. Livingston, "Old French essüer, ressüer in English", RPh, X I : 3 (1958), 254-267, at 255f. Enchaguazar in El Bierzo, captured by V. García Rey (86), exemplifies the by no means unfamiliar flux enx- /enš/ ~ ench- /enč/, cf. also SUB- > OSp. so-/sa-, ça-, cha-, and the like. ° R. de Sá Nogueira, "Curso de filologia portuguesa", §836, in LP, I (1930-31), 290, went to the extreme of confessing, apropos "popular" enxaguar: "Tenho sérias dúvidas àcerca do seu étimo". p Ensaio has often been briefly adduced on account of its form or examined on semantieocultural grounds; cf. Ford, OSR, p. xliii; S. Silva Neto, Fontes do latim vulgar (Rio, 1946), p. 209, and J. Jud's reaction in VRom, XI (1950), 236. Among the technical meanings of the verb note ensayar 'to prospect' [gold]; see B. E. Vidal de Battini, in Homenaje a F. Kruger, I, 322f. If OSp. ensayar means 'to beleaguer, lay siege' (lit. 'test one's resistance') in the Crònica de San Juan de la Pena, as quoted by A. Ubieto in RFE, XXXV (1951), 30, then a link can be established to the title of the misogynic narrative text, Enganos e asayamientos... (cf. line 17), on the assumption of widespread insecurity about en- ~ a-, cf. J. Jud's hint of Emil. assazare (VRom, XI [1950], 343) as well as It. assaggiare 'to taste, try, sample', older Sp. asayar, E. assay ~ essay (with semantic differentiation), Cat. assaig. On second thought, however, it seems more judicious to subsume OSp. asayar with Ptg. ensejar (also dial, ass-, dess-ejar), Gal. enjejar under ÏNSIDIÂEÎ 'to ambush', cf. REW3 4461 and M. L. Wagner's rev. of Lino Netto's Minhoto monograph, s.v. assejo (which he analyzed as a var. of ensejo 'opportunity, right moment'), in RPh, VI, 320. Of course, EXAGIĀRE 'to test' (the enemy's strength or intentions) and ĪNSIDIĀRĪ 'to attack from a concealed position' could very easily have fused, since /gj/ and /dj/ tended to merge, a semantic bridge was readily available to stimulate imagination, while EX- and ĪN-S- were irresistibly attracting each other. Certain side-lines peculiar to the French evolution (contamination of essayer by essuyer 'to wipe clean' ?) were pursued by J. Orr, "Two Cases of Pseudo-Semantic Development", RPh, VI:4 (1953), 294-298. The same fluidity underlines Fr. assembler alongside (or borrowed as) Sp. ensamblar 'to join, couple, dovetail'; see Wagner's aforementioned review, s.v. enxomeia, in RPh, VI, 328; the critic wondered parenthetically whether (en)xa(i)mel 'upright pole or stick used in the framework of a mud hut' was an Arabism difficult of genetic identifica tion. Ensayar 'to lay siege' cannot be completely separated from asedio 'siege' OBSIDIU or OBSIDIŌ (DCE); did this contact seal the defeat of enxA

352


TARE 'to exhale' (beside better-represented EXHĀLĀRE; cf. REW3 3011-12) and EXHĒRĒDĀRE 'to disinherit' (REW3 3012a). Q The extreme fringe of the expansion of enx- is marked by the type known to Ascoli, a century ago, only from AXUNGIA Sp. enjundia 'grease or fat of any animal', but in reality involving also equations like Sp. azada (OSp. açada, also açadón) vs. Ptg. enxada 'spade, hoe' *ASCIĀTA and, above all, various Arabisms (or Hellenisms with an Arabic veneer), cf. OPtg. enxeco ~ eixeco 'misery, predicament' AŠ-ŠIQQ; Sp. azufre, Ptg. enx-ofre, -ufre, Gal. (a)xufre ~ (a)xofre 'sulphur, brimstone' SULPHURE, and Sp. ajuar (orig. axuar), Cat. aixovar, OVal. eixahuar, Arag. ajobar, Ptg. enxoval 'bridal apparel, trousseau' AŠ-ŠUWÄR. R Q Sp. desheredar shows the substitution of des- as an alternative to eix- + enx-; OGal. eixerdar (Cantigas, 353:2) illustrates the original stage, as do the abovecited vars. assembled by M. Rodrigues Lapa. The General estoria, I I , 375b, 401a has preserved a plethora of variants, both phonetic and merely graphic, for 'to breathe' (ensaneldar) and 'breath' (enssan-ello, -eldo; ensaneldo, (en)(s)eneldo, in free competition with resollo, cf. mod. resuello 'respiration' vs. resoplo 'snorting'); see M. R. Lida de Malkiel, in RPh, XIII, 28n19. R *ASCIÄTA (REW3 697) is flanked by ASCiOLA 'small ax' (REW3 698), whose reflexes include Poit. ansol and — inopportunely omitted by Meyer-Lübke — enxó 'adze'. On SULPHUR there exist numerous sources of information, including Nunes, BSC, X, 828; on ajuar and its congeners see E. Veres d'Ocón, RV F, I:3 (1951), 14; also E. K. Neuvonen in BF, XII (1951), 344: OGal. axuar; and, above all, the noteworthy Balkanic Judaeo-Spanish forms involving antihiatic -g(surveyed by Wagner, "Espigueo...", RFE, XXXIV [1950], 30). OPtg. (e)yxeco ~ enxeco was al ready listed in the Elucidário; OGal. enxeco (Cantigas, 356 :3) caught the attention of the Marqués de Valmar, later captured the interest of Neuvonen (BF, XII, 328). Enxeco, having been shared by known troubadours and the unknown author of the Demanda do Santo graal, has figured prominently in discussions of that text's dubious authorship (M. Rodrigues Lapa, in LP, I [1930-31], 272); in fact, it was misunderstood by Spanish translators and clumsily rendered by excesso, see K. Pietsch, MPh, XIII, 636f.; Id., Spanish Grail Fragments, I, xxxiiif.; Rodrigues Lapa, LP, I, 312.

EXCURSUS  : F R O M O L D SPANISH enader  MODERN

anadir

(AND R E L A T E D PROBLEMS)

A NUMBER of compounds of DARE survived into Romance, but they no longer formed any close-knit, readily recognizable family: CONDERE 'to build, found, p u t away, store, hide', PERDERE 'to destroy, lose', REDDERE 'to restore, return, render', TRĀDERE 'to hand over, surrender, betray', VENDERE 'to sell' each went its own way. a One such former satellite t h a t cut loose from t h e system and started drifting through the lexicon of, a t least, one major provincial variety was ADDERE 'to give, bring, place; add, join', except t h a t t h e a PERDERE and VENDERE show a good record of uneventful survival; REDDERE has weathered many storms (cf. the inflectional contrast between Fr. rendre, Sp. rendir, and Ptg. render), but not before it received added strength, in most varieties of Romance, through blend with its semantic opposite PR(AE)-(H)ENDERE 'to grasp'; hence It. rendere, etc. TRĀDERE perished in many branches as a result of a disastrous homonymic clash with distinctly stronger TRAHERE (or *TRAGERE) 'to trail, pull along, drag', yielding ground to periphrastic substitutes: Sp. traicionar, Ptg. atraiçoar 'to betray' (but traidor 'traitor' TRÄDITÕRE has proved resistant). Since CONDIRE 'to pickle, preserve, season, temper' undoubtedly lingered on in Romance (REW3 2123; cf. my note, "Cundir: Historia de una palabra y de un problema etimológico", BIFCh, VIII [1954-55], 247-264), it is arguable that CONDIRE blocked CONDERE the way TRAHERE evicted TRADERE. Speakers were, of course, free to use for 'concealing' ABSCONDERE (OSp. asconder, later esconder; It. nascondere) and, in certain sections of Italy, RECONDERE (REW3 7128). The last joint action of the members of this rapidly disintegrating family was the impetus they gave in alliance with genetically unrelated CREDERE 'to believe', TENDERE 'to stretch, spread', etc., to the so-called "DEDÎ-/STETÎ preterite" in provincial Vulgar Latin. In modern Romance, even the cohesion of the relics of acephalous -CIPERE is, generally, better recognizable (cf. Fr. con-, dé-, per- and apper-, re-cevoir; but note Sp. concebir vs. recibir) than the common parentage of certain -DERE verbs.


353

meaning favored was the one most sharply profiled and t h a t the form actually transmitted was *INADDERE, with the nuance of E. 'to throw in for good measure'. b Note t h a t the distri bution of IN- and AD- in this reconstruction happens to be diametrically opposed to their sequence in ADIMPLĒRE 'to fill' (REW3 165: It. adempiere, Neap, denere, OFr. aemplir, OProv. azemplir).c The bidirectional alliance of AD- and IN- in these two verbs is hardly coincidental: 'adding (up)' and 'filling' show distinct semantic affinity. The type *INADDERE, postulated for the Iberian peninsula by Diez after certain wild gropings by pioneers d and confirmed by Meyer-Lübke in the two versions of his dictionary (4329; the macron over the A is due to a misprint), though apparently not in his earlier comparative grammar (cf. his silence in Lautlehre, §541),e can now be considered as definiti vely established/ the characteristic reflexes being OSp. eñader beside anoder, as against mod. anadir, and Ptg. enader, completely obsolete, in addition to OCat. enadir. Many more vari ants have, over the last few decades, been ferreted out, especially in medieval texts, and the purpose of this excursus is to organize them all into a cogent system. ADDERE, which underlies *INADDERE, has not been transmitted by word of mouth into the vernaculars; Ptg. adir is usually written off as a learnèd incrustation, which it indeed seems to be on the strength of its assignment to the -ir class. Portuguese has a handful of such striking cultismos unparalleled in or no longer shared by Spanish (cf. expandir 'to spread, unfold, expatiate'); the absorption of adir may have been stimulated by the earlier borrowing of the abstract adiçāo 'increment', whose Latin prototype had been a mere satellite of ADDERE. If so, the hierarchy of verb and verbal noun was, in a way, reversed within the layer of erudite borrowings. Even the scanty information provided by standard reference works at once raises two problems: (a) the split of *INADDERE as between the -er class and the -ir class of HispanoRomance verbs; and (b) the gradual transmutation of etymological en- into añ-. The first issue is indisputably morphological; the second seems to be phonological, though the exis tence of countless verbs ushered in by a- (atrasar 'to delay', avivar 'to enliven', etc.) may have been a concomitant. A third issue, by and large unexplored though it was clumsily pointed out by V. García de Diego half a century ago, is (c) the need to operate with *IN[N]ADDERE rather than *IN-ADDERE. This third difficulty cannot be attacked without a detour through Italian. On all three side-issues there exists at present a slightly elusive corpus of information; as one scrutinizes the material available, additional problems — e.g., the authenticity of the little-known variant enadir, with n for ñ — come to light. (a) The two facts t h a t Catalan (and even Navarro-Aragonese), on a large scale, favor -ir (cf. cullir as against Ptg. colher, Sp. coger 'to pick, pluck, seize') and that Galician-Portuguese b Cf. also, for the use of IN- in this semantic sector, Fr. enchérir and renchérir, both traceable to 12th-c. texts (NDÉH, s.v. cher); Ptg. encher/OSp.fenchir 'to fill' IMPLËRE; It. incremento and E. increase, related to INCRÉSCERE. c Other composite prefixes are represented in ADËRIGERE (REW3 162), ADIMPERĀRE (REW3 165a), and *INAFFLĀRE (REW3 4331). d R. Menéndez Pidal, ed. Cantar de Mio Cid, p. 637, reports that T. A. Sanchez and DamasHinard were so confused by the syncopated future enadrán as to have fabricated the "false infinitive" enandar. e The REW1, inexplicably, still listed an alleged Catalan reflex anater, which Meyer-Lübke later, fortunately, dropped. In the 'thirties he also became sensitized, through V. García de Diego's well-aimed critique (in Contributión..., §312), to the need for reckoning with an *-NNrather than an -N- in the base. f F. Hanssen, Gramática histórica de la lengua castellana (Halle, 1913), §431, briefly cites añadir to exemplify the blurring of a prefix (likening it to amparar, in which he fails to recognize the descendant of ANTEPARĀRE). Menéndez Pidal, Manual, now sees in en-ade the link between *IN-ADDIT and mod. anade (§45), now refers to the rivalry between enader and añadir (§111); his excellent treatment of the problem in the Cid edition contrasts with his later silence in the successive editions of Origenes.

354


(plus the adjoining Western Spanish dialects) often prefer -er where Castilian, not infre quently after long wavering, has opted for -ir (cf. morrer: morir, viver: vivir) have been known for decades and were not so long ago confirmed after scrupulous sifting of the evi denced Consequently, within this framework, the discovery t h a t older Portuguese used enuder (a form which later fell into desuetude), n and that Old Spanish wavered between -er and -ir, while Old Catalan gravitated toward enadir makes excellent sense. In Old Spanish, all other circumstances being equal, 1 one will tend to analyze the -er vars. as being older t h a n their -ir counterparts. (b) The long-drawn-out substitution of a- for e- has been the least well understood facet of the complex development. I t falls into place, however, within the context of the present critique of "Ascoli's Law", i.e., on the assumption t h a t the alternative was the constant threat of apheresis. Cf. OSp. enaspar beside Val. naspar (also It. naspo ~ aspo, Ast. Val. naspa), from a hybrid verb involving IN- and based, in the last analysis, on Goth, HASPA 'reel' (REW3 4069). Significantly, It. in(n)aspare has moved in the direction of annaspare, and innaffiare *IN-AEFLĀRE has yielded pride of place to annaffiare, surrounded by a galaxy of derivatives (-amento, -ata, -atura, -atore, -atoio). The reality of the hazard eluded is demon strated by It. namorato 'enamored', nemico 'enemy' (I)NIMÏCU, and nascondere 'to hide' < Late L. (I)NA(B)SCONDERE; OProv. enojar, from Late L. INODIARE, split, upon being ab sorbed by Italian, into the two branches annoiare and noiare (flanked by noia and noioso)? Where speakers remained undecided as between the switch from e- to a- and the abandon ment of the syllable, they sometimes altogether scuttled the lexical item, cf. the loss of OGal. OSp. enat(i)o 'ugly, repulsive', conceivably from INU recomposed as I N A P T U . K B See D. A. Nelson, "The Domain of Old Spanish -er and -ir Verbs: a Clue to the Provenience of the Alexandre', RPh, XXVI, 265-303, and Knud Togeby, "L'apophonie des verbes espagnols et portugais en -ir", ibid. 256-264. Nelson distinguishes sharply between two categories of verbs: (a) those which adopt the -er forms in the West and the ֊ir forms in the Center, so that their occurrence in either conjugation class has a diagnostic value for assigning a given text to a geographic area (aduzerjadozir 'to bring', viverjbevir 'to live', etc.) and (b) those which show pro tracted wavering even in the central dialects: (a-, com-) bat-er, -ir; con-, re-quer-er, -ir, etc., occa sionally to the tune of polarization between simple and compound verbs. He assigns anad-er, ֊ir (p. 275), an entry under which he subsumes also the en- vars., to the second group. h Modern Portuguese leans toward acrescentar. Speakers have also toyed with the substitute aditar (extracted from learnèd aditamentot), but here another occurrence of irksome homophony made itself felt: aditar happens to mean also 'to make happy', from dita ( = Sp. dicha) 'luck, happiness' < (pl.) DICTA (lit. 'prediction'). Agregar in Portuguese stands semantically apart (tr. 'to bring together, annex', refl. 'to join' [e.g., a party]), so that — unlike the state of affairs in Spanish, where agregar outranks anadir and agregado has supplanted both anadido and añadidura except in the set phrase por anadidura 'in addition' — it does not qualify for rivaling the local representatives of (IN)ADDERE. Note the alternatives selected by speakers of cognate languages: It. aggiungere 'to add, subjoin, reach' echoes ADITJNGERE, while F. ajouter, orig. ajoster 'to assemble' was initially tantamount to 'mettre auprès', from the adv.-prep. joste IUXTĀ. 1 That is, assuming that the copyists hailed from the same section of the country. The more "Western" their speech habits, the greater the chance of their adhering to traditional en-. There is, in other words, a constant interplay between the time axis and the space axis. j For Italian, I have, in general, been guided by B. Migliorini and A. Duro, Prontuario etimologico della lingua italiana (Torino, 19644), pp. la, s.v. acqua (innacquare ~ annacquare and various derivatives), 28b, 283b, 362a, 368a, et passim. k The REW lists neither INEPTUS nor *INAPTUS; the credibility of the latter is enhanced by ADand (Lucilius) EX-APTUS, by no means meagerly preserved in Romance (DÉLL4, p. 39a). In Old Galician one encounters side by side enato 'ugly' (Cantigas, No. 182:12) and enatio (recorded by J. Cuveiro Pinol, beside enateza 'desaliňo, desaseo'). If one recalls the range of meanings of INEPTUS (A. Ernout: 'impropre, maladroit, sot'), the fit is perfect. The stereotyped phrase "alguna cosa fea nin enatía" occurs twice in Barlán e Josaphá (102r°, 107v°); enatio is further peculiar to MS G of the Libro de buen amor (402c: "Al [sic] más astroso al enatio ajoba"; 403c: "Quier feo quier enatio, aguisado non catan"), whereas S wavers between enodio (402c; confusion with 'young stag', cf. REW3 2874?) and natyo (403c). Against this background, one begins to wonder whether OSp. enano '(male) dwarf' (Ruiz, 401b, 1279a), enana '(female) dwarf' (4316), traditionally traced


355

(c) The tendential gemination of the n is a process best amenable to observation in Italian, where -nn- is immune from development toward /ɲ/. Hence, innacquare 'to water', innaffiare 'to sprinkle with water, wash down [a dinner with excellent wines]', innalzare 'to raise, erect' (as against Sp. ensalzar), innamorare 'to inspire with love', innaspare 'to reel' (see above), innescare 'to fuse, arm (the lead), prime (a pump), strike (a discharged lamp)', from esca 'bait, lure, tinder, fuse', in(n)ondare 'to flood', innorare 'to pray, entreat'. 1 The main thrust of the evolution can be outlined approximately as follows. The charac teristic Latin prefixes transmitted into Romance had, or acquired, an s coda: DIS-, EX- = [eks], MINUS-, POS(T)-, S U B T U S - , TRĀ(N)S-, etc. Those ending in a vowel: D Ē- , R E - , SU(B)-, TRĀ-, as well as metathesized INTER- and SUPER- (cf. Sp. entre-, sobre-) eked out a precarious existence (SE- perished), and those ending in -R: PER-, PRÕ-/POR- survived marginally. A D - , reduced to a-, throve, b u t lost its semantic load irretrievably; other extra-short prefixes, including Ä-/AB-, Ë-, and O(B)-, were swept away. This chain of events left en- in an isolated position, particularly after the petrifaction of com-/co(ri)-, as in començar. To preserve and bolster its identity, speakers would generalize the canonic form en + C-, since the alter native, en + V-, was a t all times a p t to be reinterpreted as  + nV- and thus subject to losing its  in the process, cf. I t . nemico 'foe', the Spanish phytonym (e)neldo < aneldo < Gr.-Lat. ANĒT(H)ULU, preclass.-Sp. (e)nora(n)buena

I N HŌRĀ BONĀ, noramala

I N HŌRĀ

MALĀ beside polysemic P t g . embora,m also I t . (e)nel(lo), P t g . (e)no I N ILLÕ, and the like. The simplest way to ensure the desired canonic form en + C֊ in a derivative whose primitive happened to start with a vowel was to extend t h e length of the nasal beyond the morpheme boundary: en + [n]V. This adjustment was the easier to achieve as there existed from t h e outset numerous words with the — etymologically justified — initial segment INN-, t h u s : INNĀRE 'to swim in, flow or sail over', INNĀSCĪ 'to arise in', INNECTERE 'to weave or put together', INNÏTÏ 'to lean upon', INNOCËNS or INNOCUUS ~ INNOXIUS 'harmless', INNOVARE 'to renew',

INNÜBILUS 'unclouded', INNUERE 'to give a nod, make a sign to', INNUMERĀ(BI)LIS 'countless', INNÜTRÏRE 'to bring up in or among'.11

to a cross of NANUS (cf. Fr. nain) and INĀNIS 'empty, hollow' (cf. REW3 4334), does not actually present a case of false regression. Cf. also the jettisoning of enante de 'before' (Barlán e JosapM, 207v°, with counterparts in the Cantigas) in favor of antes de or delante de. 1 On OIt. innaspare and the increasingly more common annaspare 'to wind (thread) on a reel' see, in addition to the standard dictionaries, . . Livingston, Skein-Winding Reels; Studies in Word History and Etymology (Ann Arbor, 1957), p. 197. Speakers of Italian apparently used the momentum of in- > inn- to squeeze (obs. dial.) insetare *IN-SITÄRE (from INSITUS, an erratic member of the INSERERE family) into the mould of innestare 'to graft, insert'. Such is, at least, the interpretation of Migliorini and Duro, with which G. Devoto, Avviamento alia etimologia italiana: dizionario etimologico (Firenze, 1966), p. 233a, is basically in agreement. Devoto also records innastare from asta 'staff, shaft, spear' and places innant-e, -i alongside innanzi. There are a few older examples. In her forthcoming RPh rev. of J. Corsi, ed. Poesie musicali del Trecento, F. B. Ageno documents in{n)amorare, which also figures in Masuccio Salernitano's Il Novellino, ed. J. Petrocchi, 24.4. m On (e)neldo see Corominas, DCE, I I [1955], 272a; on noramala, etc., Joseph E. Gillet, ed. Propalladia, I I I : Notas, p. 862a. There is  good reason why in Old Spanish neither enemigo 'foe', 'evil' (LBA, 399b), 'devil' nor enemiga 'hatred, enmity, ill-will, malevolence, aversion, taunt' {LBA, 89b, 825c), rarely 'female enemy' {De una emperatriz..., ed. Mussafia, Ch. iv), were apheresized: This path was blocked by preexistent nemigaja {= OPtg. ne-, ni-migalha) 'nothing' — absent, in turn, from south-central Italian —, cf. Menéndez Pidal, "Etimologías espaňolas", Rom., X X I X (1900), 360, and Meyer-Lübke, REW3 5885. Hence enemistad or, through regression, enamistat {LBA, 1522b); enemistarse 'to become an enemy, fall out with'; the fixed phrase fazer {su) enemiga, etc. {El emperador Ottas, ed. Amador de los Ríos, Ch. XL; Barlån e JosapM, ed. Moldenhauer, 103r°, 121r°, 160r°, 164v°). n The otherwise all-important division into adversative and illative (or ingressive) IN- for once plays no major part in the process here surveyed. Observe INNUBA 'unmarried, without a husband' alongside INNÜBERE 'to marry into', a pair which dramatizes this cleavage.

356


Can one produce evidence of intercalary radical-initial ո after IN- from Hispano-Romance, aside from OSp. enader, OPtg. enader IN-[N]ADDERE ? There is, I think, no dearth of scattered traces. Consider the case of OLeon. ennartar 'to betray, seduce', as found in Elena y Maria, a dialogue meticulously edited by Menéndez Pidal: "Comer Z gastar / Z dormir Z folgar, /fijas de omnes bo[nos] ennartar" (lines 112-114). I n his glossary, the Madrid scholar offered this comment {RFE, I [1914], 87f.): " J u n t o al significado general de 'enganar', tenia el de 'seducir a una mujer'", and adduced an interesting parallel from General Estoria, MS 816, fol. 190d: "Si alguno enartare a virgen  a mugier non desposada con él". The meaning guarantees the derivation from arte 'art, craft, ruse' and makes it further plausible t h a t the speakers remained thoroughly aware of this connection. Such an assumption would explain both the coinage (on the model of enredar, enganar) of IN[N]ARTĀRE > ennartar, i.e. /ejiartar/, and its subsequent "correction", under relentless pressure from arte, to enartar, as used four times by J u a n Ruiz. 0 The process calls to mind the adjustment of archaic annochecer and allongar to anochecer, alongar; see below. One can further cite sporadic traces of ennamorar, plus the Graeco-Latin zoonym ennodio 'young stag', from ENODIS, with hypercharacterization of the gender by way of additional intricacy. p Despite the deceptive parallelism, one immediately recognizes the difference between the evolutions in the two peninsulas. I n Italian, the msertion of ո after in- (strictly, the exten sion of n) brought the process to a helpful conclusion (innaspare, etc.). In Spanish, however, the fairly late change enn- > en-, i.e. /en-/ > /eɲ/, in the wake of the elimination of all normally-lengthened consonants, in the end defeated the purpose of the epenthesis, since ewas once again left "dangling" in a severely endangered position. Hence its occasional loss and the consequent emergence of word-initial ñ (originally alien to the structure of the language), cf. nudo ~ nudo 'knot' NŌDTT, nublar ~ nublar 'to cloud' NŪBILĀRE 'to be cloudy, cover with clouds'; or, alternatively, the sporadic switch from en- to an-, facilitated by the préexistence of t h a t segment in ano 'year' ANNIT and its far-flung family. q Anorar 'to desire wistfully, in a nostalgic mood' and anoranza 'nostalgia' stand apart as fairly recent imports from Catalan (enyorar IGNŌRĀRE) — previously soledad was so used, cf. Ptg. saudade, OPtg. soidade — but are nevertheless relevant as witnesses to the continued vitality of eñ- > an-. An additional power behind the shift en- > an- was the attraction exercised by a solid block of Arabisms displaying the latter sequence, as in anafil 'Moorish musical pipe or trumpet' < AN-NAFÏR. In this sliver of the lexicon (consisting, typically, of nouns, a few of them accom panied by secondary verbs) the ñ represents the compression of two consecutive n's, the

° Juan Ruiz used enartar once reflexively: "Ansý muchas fermosas contigo se enartan" (S, 403a) and thrice transitively: "Eres mentiroso falso en muchos enartar" (S, 182c); "Guardat la que non fuya que todo el mundo enarta" (S, 1195c): "Desta guisa el malo sus amigos enarta" (S, 1457c). Enartar occurs repeatedly, and with exactly the same meaning, in the Alfonsine Cantigas (Nos. 7:1 and 366 :4; also, F estas de Santa Maria, No. 4:6). p A. Castro, ed. Glosarios latino-españoles de la Edad Media (M., 1936), lists in his Index (p. 327b) only enamor-ar, -ado; in reality, the Escorial Glossary contains the entries ardelio 'enamorado' (433) andfilocapio 'ennamorar' (2648), filocata 'enamorada' (2648), -us 'enamorado' (1199), from see pp. 167a and 221b. Even if it constitutes a mere slip of the pen, ennamorar is noteworthy. The typical Old Spanish forms were, admittedly, enamor-ar, -ado, -amiento; cf. Ruiz, passim, and Confisión del amante, 35r°, 164r°, 174r°, 402v°. How many *enn- forms, belatedly recomposed, actually underlie enardecer (Esc. Gl., 2321, 2620), enorgullecer (Grail Fragments, ed. K. Pietsch, 282r°), etc., is anybody's guess, given the lacunary shape of the record. It is further arguable, though hardly demonstrable, that in the course of recomposition folk etymology could have struck with special force, as when enervolar 'to poison' (based on HERBULA 'herb') tended to degenerate into enarvolar, with a hint of árvol (-6-) 'tree'; cf. Ruiz, 183b (S) beside 597a (G). q Cf. adj. cordero anal 'yearling lamb', n. 'anniversary', añalejo 'ecclesiastic almanac', anada 'good or bad season', añejo 'old, stale, musty', añinos 'unshorn coats of yearling lambs', aninero 'dealer in lambskins', añojal 'fallow land', añoso 'old, stricken in years', and many more. Cf., in Juan Ruiz, anal 'yearling' (1013a, 1016a) and añejo 'year-old, stale' (119a).


357

first forming part of a predictable variant of the def. art. al-, the second constituting the initial root consonant. 1r Corominas was, on balance, doubly in error in the 1961 summation (BDE) of his earlier analysis (1954) of anadir: After correctly identifying OSp. eñadir, OPtg. enader, and OCat. enadir as the three starting points, he chose for no valid reason a rare erratic var., OSp. anadir (possibly spurious, see below), as the immediate prototype of the classical and modern form, instead of seeking a defensible explanation for the mainstream of events, namely eñ- > añ-, to the accompaniment of -er > -ir. Also, he imputed to lexical analogy (wavering between anudar and anudar, anul·lar and añul·lar, even — of all conceivable models! — anascar and anascar) the alleged drifting from an- to añ-, whereas in reality the initial seg ments ejn- and ijn-, for strictly phonological reasons, became unacceptable in Spanish, as any quick check with a standard dictionary will at once confirm, while an-, benefiting from sever al tributaries, steadily gained in appeal. I t is difficult to establish an unequivocal medieval corpus of ariadir vars., because the wavering between ann- (añ-) and enn- (eñ-), normally transparent, is compounded by the oscillation -er ~ -ir, observable only in certain privileged sectors of the paradigm. s How can one, for instance, decide whether the author, or translator, of the Vida de Barlán  del rey Josaphá de India (ed. G. Moldenhauer, 1929) — to select a less than distinguished text almost at random — favored the -ir or the -er var., as long as the only potentially relevant passage (121r°), in MS P, is inconclusive: " E quando el Nuestro Sennor dixo estas cosas, annadio..." (similarly in MS G), whereas, through a twist of circumstances, the preference of the translator, via a lost intermediate Portuguese version, of John Gower's Confisión del amante (ed. H. Kjiust) happens to be clear-cut: "Somos aquí venidos a te suplicar que r Añafyl occurs, as one would expect, in Ruiz: 1096c (S, G) and 1234a (S, G, T). Supporting examples include: añacal 'carrier of wheat to mills', 'baker's board for carrying bread' (Cádiz: añacalero 'carrier of lime, tiles, roof-tiles, etc. to a building site') AN-NAQÄL; obs. añacea 'merri ment, fiesta' AN-NAZĀHA (anacear 'to take part in a cheerful, noisy party'); (papel de) añafea = esłraza 'rough, unbleached paper' AN-NAFĀYA 'offal'; anasco 'entanglement, muddle' AN-NAŠIQ (and, in its company, anascar 'to collect small trinkets by degrees', 'to entangle, muddle'); anil 'Indigo plant and dye-stuff' AN-NÏL (the var. anir, still listed by the Academy Dictionary, should be interpreted in light of such split decisions as alfil-er ~ -el 'pin', see Anita K. Levy, "Contrastive Developmentfs]...", BPh, XVIII, 399-429, and XX, 296-320). A few items cause difficulty. The Academy declared añicos 'bits, small pieces, smithereens' an Arabism in 1947, but J. Corominas, in 1961 (BDE), advocated the recalcitrant word's [indigenous] Ibero-Romance ancestry (*ANN-), citing Ptg. (a)naco 'shred'. Anagaça 'lure, decoy', 'allurement, enticement' was a straight Arabism for the Academicians, whereas Corominas, though well-aware of its equivalents, Ptg. negaça and Hisp.-Ar. NAQQĀZA, wondered — given the silence of extra-Hispanic Oriental sources — whether diffusion had not taken place in the opposite direction. Among other exoticisms, of marginal rele vance, one may cite ana 'wet nurse', of Basque ancestry, and añas 'kind of Peruvian fox', of Quechua background. Not directly related to an-, but worthy of mention here on account of its ties to the Arabic strain of Hispano-Romance, is OSp. anaziar, enaziar, naziar, OPtg. anazar 'to cross over to the enemy camp', on which there exists a copious etymological literature (R. Menéndez Pidal, C. Michaëlis de Vasconcelos, A. Steiger, and others), digested by J. Corominas, , II, 251a-253a. Whatever the root-morpheme, the rivalry of the segments an-, en-, n- falls into place within the frame of the present study. Enaziado seems to have been the favored variant; cf. PCG, 514a33; Escorial Glossary, 2167; Crònica de Alfonso XI, , LXVI, 317a, 347a. Ennaziar, interestingly enough, is also on record. 3 Particularly detrimental to our analysis is the accidental fact that the future and conditional tenses, in early Old Spanish, showed syncope of the intertonic vowel, thus: enadrán (Cid, 1112); en[a]dré (Alexandre, , 925a, but P, 953a: anadiré;, = ed. Willis, 972a); also, Buenos proverbios, ed. H. Knust, pp. 20, 35: ennadrá; Apolonio, 398d: enyadrié. See Menéndez Pidal, ed. Cantar de Mio Cid, pp. 285 (where comidrán, cadran, odredes are adduced as parallels) and 637. Again, had the root vowel been  or  rather than a, the functioning or inoperativeness of metaphony might, in certain medieval dialects, have cast light on the conjugation class involved, cf. tenié vs. mintié, cosió vs. murió.

358


quieras anadir...en los días de mi vida" (62v°) ?ť I n his authoritative Glossary (1922) to the Libro de Apolonio, . Carroll Marden listed two forms of the infinitive, anyader beside enyadir (II, 75); his grammatical comment was sketchy and noncommittal (II, 14f.). Since the ֊er infinitive occurs twice, once in rhyme (28c: "Aun treynta días le quiso anyader" [ : perder, asolver, cayer]; 525c: " E t te quiero aun anyader en soldada"), and in the absence of any traces of an -ir counterpart, one is tempted to claim the verb for the -er class within the author's idiolect (even though, generally, Old Navarro-Aragonese tended to gravitate to ward -ir). J u s t how the poet pronounced the opening segment we are unlikely to ever ascer tain, given the discrepant var. emerging in qu. 398d: " E l enyadrié veyente pesas de buen oro colado". As regards the two major versions of Pero López de Ayala's Bimado de palacio,u a certain amount of prudent extrapolation seems admissible in the case of MS N. Since two well-authenticated passages, qu. 572b: " E l santo confesor quiso más añoder" ( : fazer, aver, poner) and qu. 1178b: "Dixe una palabra, e quiero anoder" ( : saber, poder, fazer), contain the controversial infinitive in rhyme, it is highly unrealistic to doubt t h a t anoder also underlies añade in 971d, anaden in llOOd, once more anode in 1128c, and anadió in 1517c. But one can not be equally certain about the habits of the scribe(s) of MS E. True, in qu. 1627d, which lacks a counterpart in N, one reads: "Quinze le fueron anadidos..."; but where N has anaden, E offers enaden; where N reads añadió, E confuses us with its eňadió; where N, consistently, uses anade, E on one occasion offers andan (clearly spurious), and on another eñade. The chancelor himself, as distinct from his copyists, presumably used an -er verb whose root morpheme contained either enad- or anod-. Additional complications arise from unexpected, less familiar patterns of vowel distribu tion. Menéndez Pidal unearthed isolated traces of añidir in Golden Age writers (Santa Teresa, Aldrete; see ed. Cantar, p. 264); not coincidentally, there seem to be no medieval precedents, except for anedir, and the two must have related like pedir to mod. dial, pidir. J u a n Ruiz apparently used an -ir var., judging from qu. 1629b (S): "Más aý añadir e emendar si quisiere"; T prefers anedir.v I t is uncertam whether the radical, in his idiolect, was anad- (anad-), enad-, or añed-, given the conflicting evidence of scribal predilections. Specifically, in qu. 690a MS G (the only extant source for t h a t passage) reads: "Do anadieres la lena, creçe syn dubda el fuego", while for 1143d we have a choice between three divergent readings: "Quince años de vida anadjó al culpado" (S; b u t G: eňadió, T: añedió). Anod- is unlikely to represent more than an infelicitous slip of the calamus; cf., in Ducamin's authori tative edition, the graphies anejo (119a) for onejo, ano (762c) for ano, etc. The most delicate decision will often be between en- and en-. Mention has already been made of OSp. allongar ' to lengthen' Ě-, *AL-LONGĀRE gradually yielding ground to alongar — through recomposition, as it were (i.e., through renewed association with the primitive luengo 'long', against the background of the derivational model -ar, frequently appealed to in extracting verbs from adjectives, in rivalry with a-, en-, es cer). Along a parallel line, archaic annochecer (impers.) 'to grow dark', (pers.) 'be somewhere a t nightfall', pronounced  On the idiosyncratic construction eñader en 'to increase, augment' see Menéndez Pidal, ed. Cantar, p. 637, who, in addition to several Old Spanish passages (culled from Crònica de los Once Reyes, Libros de Astronomia, etc.), cites a remarkable Hispano-Latin prototype: "Ad[d]idit ipse comes... simul cum sua matre in superbia eorum" 'aumentaron su maldad'. Add to this dossier Apolonio, 525c, and Leomarte, Sumas de historia troyana, ed. A. Rey, §124 (p. 221): "Tomaran a annader en el vuestro plazer". Perhaps this striking construction echoed the ablative governed in Classical Latin by FRUĪ 'to enjoy, have the benefit of' and ÜTÏ 'to use, employ, enjoy'. u I am following here Marion A. Zeitlin's Vocabulary..., a finely-chiseled unpubl. Berkeley dis sertation (1931), based — after scrupulous collation with the MSS — on Alfred F. Kuersteiner's competent edition. v Unaccountably, H. B. Richardson, An Etymological Vocabulary to the "Libro de Buen Amor"... (New Haven & London, 1930), p. 20, apropos qu. 1629b remarks: "Here we have anadir in S, eñadir in T".


359

with an /ɲ/, under pressure from noche 'night' before long became anochecer.™ Also, ennartar 'to seduce with one's charms or wiles', bracketed with arte, was, we recall, transmuted into more neatly segmented en-art-ar. As a result, it is marginally possible t h a t enader — found, according to Menéndez Pidal (ed. Cantar, p . 264), in t h e Cid, also in a document A.D. 1285, and in Don J u a n Manuel's Libro de la caza (ed. Baist), p . 2.25 ("nin enader ninguna cosa más") — involves the shift enader > enader conducive to slightly sharper profiling. x Alternatively, it can be argued t h a t merely a few additional instances of omission of t h e tilde are a t issue. y Conversely, anadir is unlikely to have enjoyed any status higher than t h a t of a false graphy or an incidental slip of the tongue. 2 On this cumulative evidence, direct or circumstantial, one can hazard t h e following cross-dialectal break-down of the inventory of variants (with a minimum of repetition) : (1) (a) OSp. enader or enadir:* Berceo, Milagros, 587b (ennadió), and Santo Domingo, 221d (ennadieron, var. annadieron); La fazienda de ultramar, ed. M. Lazar, 118.19, 156.15; Fuero Juzgo, p. 14a; Partidas, I I.15 (var.); Buenos proverbios, 20, 35 (ennadió, ennadia);? w See Menéndez Pidal, ed. Cantar, pp. 229 and 466, with a dual reference to annochesciere in the Primera Crònica General, ed. 1906, p. 511a4, and to anochesca in the chosen epic (v. 432). Allongar, whose ultimate model could have been ĒLONGĀRE (Ernout and Meillet, DÉLL: 'allonger, éloigner'; REW 3, 2853, with references to Rum. alunga, OLomb. alongar, OIt. allongare, etc.), is particularly relevant, because it falls into a pattern abundantly represented; cf. Gal. emorçar (Cantigas, 278:3) beside OSp. almorçar, Ptg. almoçar, against the background of AD- ~ DË-MORDĒRE, and the like; also Sp. anegar 'to drown' vs. ĚNECARE 'to kill, drown'. x Admittedly, -ad- at that advanced phase hardly carried any semantic message of its own. But the substitution of en- for eñ- aligned eñader with a profusion of verbs so ushered in — of different degrees of transparency: embotar 'to blunt', embutir 'to stuff', entelerido 'numb, shiv ering', etc. y On the habits of the copyists of Juan Ruiz see above; on Per Abbat see Menéndez Pidal, ed. Cantar, pp. 226f.: dueno, Munoz, adel-inar ~ -inar 'to proceed', estr-ana ~ -ana 'strange' (f.), mont-ana ~ -ana 'mountain'. z One is free to fancy a speaker torn between the poles añadir and enader stumbling, on an iso lated occasion, upon anadir. Such a compromise form, however, would pertain to Saussure's

LANGAGE or Chomsky's PERFORMANCE, rather than to LANGUE or COMPETENCE. a Two or three data are due to the — rather uninspired — entry (I, 225a-226a) in Corominas' DCE, which contains a few useful references to Catalan sources. The author reports that afegir has replaced enadir (which, unlike OSp. enadir, presupposes -N- rather than *-NN-); records OCat. ennartar, but is oblivious of its Old Leonese counterpart; sharpens Garcia de Diego's hypothesis of the gemination of the nasal, without drawing from it any crucial inference; offers parallels to Meyer-Lübke's — probably spurious — datum enater, without either proving or dis proving its disputed authenticity. In enader the author sees an archaism rather than the change en- > en- here advocated. The process here studied has several ramifications inviting microscopic investigation, which cannot be here provided. Sp. anusgar 'to choke', fig. 'be displeased' has long been known as an etymological crux. The wavering, in Salmantino, between (refl.) anosgar and anus-, anuz-gar (J. de Lamano y Beneite) prompted V. García de Diego, Contribución, §325, to trace the entire cluster to *INNÕDICÄRE 'to knot', with a reference to the speakers' indecision between NŌDUS and *NŪDUS. L. Spitzer later opted in favor of *IN-OSSICÄRE; i.e., in IN[-N]-OSSICÅRE, from OS(SUM) 'bone'. See also J. Corominas, DCE, I, 229b-230a and (Supplement) IV, 9215, s.vv. anagaza and anuscar (polemic with García de Diego's analysis in the latter's DEER, §3461). Anusgar's possible link to amusgar MORSICĀRE also bears investigation; cf. PhQ, XXIV (1945), 233-254. García de Diego, Contribución, §328, further drew attention to Salm. añ-, an-udrido 'desnutrido', tracing them to IN-NÜTRÏTU, and to the truncated p.ptc. añodro 'id.'. Cl. Lat. NŪTRĪRE con trasts, as regards the quantity of the first vowel, with the prototype of Fr. nourrir, OSp. nodrecer. The same miscellany of etymological notes concerned itself with (§326) INNOVARÉ Gal. anovar 'to change' and (§327) INNŪBILĀRE -Ast. anublar (Rato y Hevia), documenting the latter in the compilations of A. de Palència and Nebrixa. 0 I am following, by and large, in the footsteps of Menéndez Pidal, ed. Cantar, pp. 264, 637, except with respect to the Apolonio, for which C. Carroll Marden's scrupulous edition provides superior readings, occasionally divergent from those previously accepted; I also owe some bits of

360


(b) OSp. eňader or enader, -ir: Alexandre, ed. Willis,972a (0);Cronicón Villarense, see BRAE, VI (1919), 199, with traces in Judeo-Spanish; (c) OGal.-Ptg. enader: Alfonso, Cantiga 4:1 for the 'Testas de Santa Maria": "Como Deus é comprida Trijdade / sem enader nen minguar de ssí nada" (Acad. ed. of Cantigas, II, 574); Frei Joaquim de Santa Rosa de Viterbo, Elucidário..., ed. 1798, I, 3936: en-, em-ader (credited to the 15th c.).Y (2) (a) OSp. a-hader: Bibi. E s c , fol. 186a (cf. Biblia Scio, Ecclesiasticus 18:5); Apolonio, 28c, 525c; Toledo Glossary, 1952 ('addo'); Escorial Glossary, 2917 ('addo'); E. de Villena (DH); cf. añadedura: Escorial Glossary, 487 ('aumentum'), 517 ('aumentarium'), also J. A. de Baena, Cancionero, 148v°; (b) OSp. añader or anadir: Crónica de Once Reyes, 3r° (aňadió) — Tercera crónica, 280d; (c) Cl. Sp. aňedir: J. del Encina, Cancionero, ed. 1516, 300v° I I (añedido); G. A. de Herrera, Agricultura general (1515), ed. 1818,1.1, Prologo (el añedir); F. Cervantes de Salazar, Crònica de la Nueva España, ed. 1914, p. 315 (añedi) — for all three passages see DH, I, 627b; abundantly represented in modem dialects, on both sides of the Atlantic (A. Alonso and Ճ. Rosenblat, BDHA, I, 88f.; see also 386f., 391nl). (d) CI. Sp. aňedir or añidir: . Casas, Apologética historia..., NBAE, XIII, 381 (aňide); J. de Arjona, La Tebaida, , XXXVI, 146b (aňide); Escalante, Diálogos de arte militar, ed. 1583, p. 66 (añidió); . de Balbuena, El Bernardo, , XVII, 347 (aňida) beside grammatically unequivocal aňididura in Juan de Tolosa, Discursos predicables, ed. 1589, fol. 60; all five passages cited by DH, I, 628a. (e) OSp. añadir: Alexandre, 972a (P): "aňadiré en riqueza"; Alfonso, Astronomia, 68, 6. Cf. aňad-ido, -idura, -imiento. (f) OSp. anoder [?]: Espéculo, 7, 2, in: Alfonso, Opúsculos legales, I (Madrid, 1838). Several discrete forces, then, presided over the unique Peninsular development of *IN[N]-ADDERE, away from eñader and in the direction of añadir.0 Chief among them were the relentless growth of the -ir class at the expense of the -er class; the steadily increasing commonness of the segment a- in verbs, as a prefix and, occasionally, as a prefixoid; the acceptability of an-, not least through the protection afforded by a solid block of Arabisms, and the concurrent decay of en-; plus the vulnerability of an e- left "dangling" — as estab lished by Ascoli in his brilliant intuition of 1878. But the kernel of the problem must be defined differently: I t revolves around the lengthening of the nasal, presumably in late Antiquity, as in Cat. OLeon. en-n-artar from arte, and in the Italian counterparts adduced, a move initiated to bolster up the endangered free syllable comprising a single front vowel; and in the eventual reductio ad absurdum of t h a t protective measure, after geminate n, at the peak of the Middle Ages, ironically became a palatal ո of standard length, incapable of standing guard over e-. information (and a few felicitous conjectures) to Dana A. Nelson, "The Domain ...", at p. 275. Nelson supposes that Berceo and the unknown author of the Alexandre used eňader (and, in the process, demolishes Corominas' unfounded web of hypotheses). Y The form emader, glossed 'addo', figures as No. 80 in "A Fourteenth-Century Latin-Old Portuguese Verb Dictionary", ed. H. H. Carter, in RPh, VI, 71-103, at 74 and 100. Addenda at proof—Ad ng: Ascoli's unrealistic reconstruction of proto-Sp. *ensfaldar, etc., could have been prompted by OIt. insforzare, see F. B. Ageno, RPh, XII, 323; ad p. 512: For one attempt to trace to a Greek base Ptg. enxárcia 'shrouds and stays of a vessel' (also enxarciar = Sp. enjarciar 'to rig'), Sp. zarpar 'to weigh (anchor)' see G. Alessio in RLiR, XVII (1950), 71f.; ad ng: T. Montgomery's paper, "Complementarity of Stem Vowels in the Spanish Second and Third Conjugations", is to appear in RPh.

ETIOLOGICAL STUDIES IN ROMANCE DIACHRONIC PHONOLOGY This collection of studies, at first glance heterogeneous, is tied together-less loosely than might initially appear ֊ by several emphases and deèmphases. Common to them all, aside from the chosen time perspective, the confinement to Romance material, and the concentration on problems of sound development, is the pre occupation with the causes of change, including the all-important interplay of causes at work. To this extent it is not inaccurate to call them etiological, at least in intent. Of each of the six problems involved only a rough sketch is presented. With more time and, above all, more editorial space available, it would have been possible, given the well-nigh inexhaustible wealth of documentation available in Romance, to transmute each vignette into a heavily illustrated monograph. In the course of such expansion, certain details would undoubtedly have been corrected; but, as Romanists know from past experience, elaboration, for all its benefits, also conjures up certain dangers: It is all to easy to get lost in details, to be carried away by the entertaining vicissitudes of individual words. Self-imposed selectivity and limitation, under certain circumstances, may be wholesome; I hope I have not gone too far in using very sparingly — and perhaps a shade arbitrarily — both primary data and learned sources. The languages chiefly tapped are Italian (in particular, Tuscan) and Spanish; as a foil to the latter, Portuguese occasionally makes its appearance, while other cognate languages merely serve as a background. Perhaps the resources of Italian and Spanish have not been sufficiently integrated; note, however, that the explana tion here favored of It. chiedere, from QUAERERE, receives support from dial. Sp. quedré (fut.), used in preference to standard querré, from querer. Observe also the benefit derived from contrastive comparison of the treatments of the intertonic syllable in the two languages. I would not be surprised if careful study of relevant conditions in Italian were someday to help scholars to solve the vexing riddle of the nominal augments infelicitously called "sufijos átonos" in Hispano-Romance. As often in the annals of scholarship, the paradox of seeming exceptions can provide a suitable starting point for an inquiry, on the assumption that one is willing to probe beyond the clarification of the immediate difficulty. It is nothing short of a paradox that, of all possible collocations, it is the position of unstressed  before r which has prompted speakers of the Florentine dialect to open that front vowel to a. Again, it is bizarre that, while the affricates ç [ts] and z [ z ] were freely available in Old Spanish, speakers should have transmuted the ancestral DJ (as in VERËCUNDIA 'shame') into ç rather than into z. Certain circumstances may, upon occasion, block the quick recognition of the causal relationships. One such circumstance may be the pattern of conventional

362


classification. For perfectly defensible reasons, our standard manuals of historical grammar record the development of G e,i - and QU- under initial consonants and consonant clusters, respectively. But in a causal inqmry the salient point is the speech community's vigorous reaction to a segment which sounded exactly like an "unwanted diphthong" or, more accurately, like a diphthong placed in an undesir able position. The process becomes transparent when examined in the context of diphthongization and secondary monophthongization, since it has causally noth ing to do with consonants. The speakers' reaction to a risk or an unpleasantness is what ties together all four Hispanic studies. The analysis here favored endows at least some speakers — the tone-setting group — with a considerable measure of initiative, rather than assigning to them the rôle of passive victims of historical accidents. The Italian studies are conceivably best characterized as explorations of multiple causation. In the past I have had ample opportunity to point out situations in which a (usually abnormal) morphological state of affairs leads to adjustments which, in the end, may result in a new sound change. Realistically, it must be assumed that such situations are uncommon and, moreover, that this infrequent move from morphology toward phonology is, as a rule, just one ingredient in a complex situa tion rather than the sole cause of change. The opening studies aim at illustrating just such situations.

I. The Change of Unstressed ,  into  before r in Florentine. This change, in certain positions, is highly characteristic of Tuscan, while in other positions it determines its subdialects, pitting Florentine against Pisan and Senese; it clearly contradicts the general Romance trend, which runs in the opposite direc tion.1 The change is least conspicuous in proparoxytones: lf bàvero ~ bàvaro 'collar of a coat' coexist, as variants of bavarese 'Bavarian', there is nothing unusual about this proliferation of forms once we recall the rivalry at the Latin stage between CAMERA and CAMARA 'vault' (a word of Greek background). Far more striking are the forms of the future tense {-rò, -erai, etc.), pertaining to the conjugation in -are, and changes like Gr.-Lat. MARGARITA > Flor. mar-, mal-gherita. Most of the facts have been known for a long time, and the evidence available is plentiful; but the matter of causation remains, by and large, unexplored. We shall start our

1) The state of affairs here summarized already caught the attention of such pioneers as N. Caix, A. Mussafia, and J. Storm, to say nothing of the comparatist F. Diez, and was force fully presented by Meyer-Lübke in his Italienische Grammatik (Leipzig, 1890), the ultimate version of which ֊ involving at once an elaboration, a compression, and a translation — is the Grammatica storica della lingua italiana e dei dialetti toscani (Torino, 1927), with sundry im provements suggested over the years by M. Bartoli, G. Braun, E. G. Parodi, and the author himself; see § 7 1 , which has served as my immediate source, although I have checked every datum against more recent reference works, such as Ernout-Meillet's Dictionnaire étymologique de la langue latine4 (Paris, 1959-60), Migliorini-Duro's Prontuario etimologico della lingua italiana3 (Torino, etc., 1958), and the like, and, as a result, have introduced numerous changes.

ETIOLOGICAL STUDIES

363

sketch with the less baffling cases {camera, albero 'tree' < ARBORE) and shall attempt to place this particular shift in a somewhat broader context (parallel changes, under different sets of conditions, in the directions of a and o). The more difficult and more challenging varieties of the shift can then be attacked with somewhat greater confidence.

A. On relatively few points of diachronic phonology do Florentine Italian and Castilian Spanish appear to be in really sharp contrast. One such neat polarization involves certain categories of unstressed (specifically, intertonic) vowels. In pro֊ paroxytones Spanish favors the retention and, on a modest scale, expansion of a single vowel, namely a, regardless of the details of that vowel's immediate environ ment. Hence, on the one hand, ánade 'duck' < ANATE, espárrago < Gr.-Lat. ASPARAGU (with substitution of -es for as-, as in esconder 'to hide', escuchar 'to listen'), (h)uérfano < Gr.-Lat. ORPHANU (also Leon.(h)uérgano

*-U, perhaps after the near-synonym MISER-U (cf. povew); CANC(E)RE, with concomitant hypercharacterization of gender and semantic differentiation of dou blets; GENERU, SOCERU, and, in their wake, *SOCERA for SOCRU (from SOCRUS -US [f.]); and (d) the jelling of certain other characteristic relationships between unstressed vowel and following consonant, such as (a) -ano, -amo, -, -ago, -afo, beside their counterparts in -a (where the influx of Hellenisms may have been the prime mover), and (β) -olo, -ola, where the residue of Latin -ULU diminu tives (i.e., those not syphoned off by -ELLU) may have been chiefly operative. Any alliance between, say, a and ո as well as  and / sets a powerful precedent for a newly emergent affinity similarly delimited; for instance, one between e and r. Each of these four forces in isolation may have proven inadequate to set in motion the tendency here examined. On the other hand, their convergence or coefficiency produced a situation in which the shift ՝֊ar- '('-or ' ) > ՝-er- ' begins to fall into place. B. For a variety of reasons the change from -er- to -ar- in such intertonic syllables as precede the main stress involves a situation thoroughly at variance with the one just surveyed. For one thing, the majority of the words this time concerned are not nouns (let alone technical terms of Hellenic background) but specific verb forms, pertaining to two tenses, the future and the conditional, of the steadilygrowing conjugation class in -are: bàlleró 'I shall dance', from ballare, màngeréi 'I should eat', from mangiare (but daw 'I shall give', farò 'I shall do', and starò 'I shall stand', from dare, fare, and stare, respectively, because no intertonic syllable is here, for once, involved).4 For another thing, the preference is narrowly Florentine and, in fact, clashes with trends observable, from the Middle Ages, in near-by Pisan and Senese, which both enter into the Tuscan bundle of dialects. One misses in the new context the agency of any such vocalic gamut as the familiar a-o-e {a before nasals, velars, and labials,  before the lateral, and e before r), but i reappears as a kind of neutral or unmarked vowel, on some occasions also as a Sprossvokal. This is how Charles H. Grandgent interpreted biasimare 'to blame, censure', caricare 'to load, pile up (with)' from older biasmare and carcare, coperire as a var. of prire, coricare 'to lay down, put to bed', desinare 'to lunch', manicare 'to eat' (by and large dislodged by the Gallicism mangiare, cf. Sp. manjar 'food, dish') from older cor-, *col-care (witness Fr. coucher), desnare, *mancare, the remote prototypes being, in this order, Gr.-Lat. BLASPHĒMĀRE, *CARRICĀRE, COOPERÎRE, COLLOCĀRE, DlS+IE(IŪ)NĀRE (haplologically compressed), and MANDŪCĀRE, lit.'to chew'.5 4) Sarò from essere 'to be' and andrò (optionally beside anderò, from andare 'to go') followed suit, on account of rhythmico-accentual affinity and syntactico-semantic proximity. 5) See that scholar's classic, From Latin to Italian: an historical outline of the phonology and morphology of the Italian language 55f. (= § 52) (Cambridge, Mass., 1927). The author further cites manimettere 'to broach, begin using' from MANUMITTERE (where, on second

368


Good examples, culled from among nouns, of ՝-ar- ' > '-er֊ ' are few and far be tween: guiderdone 'reward, recompense' < OProv. guizardon (older guid-) < Gmc. widarlon, lazzeretto alongside lazzaretto 'fever hospital, quarantine station' < (SANTA M A R I A D I ) N A Z A R E T H (near Venice), influenced by the name of the New Testament figure, Lazzaro 'Lazarus' (the beggar) — and, perhaps through folk-etymological contact with this family or through sheer phonic similarity, (bot.) lazzaruolo ~ lazzeruelo 'Neapolitan medlar' (with the -er֊ variant for once closer to the etymological base: Sp. acerola < Ar. azza'rûr, if this assumed route of trans mission is correct), mar- (rarely mal-) gherita 'pearl', (bot.) 'ox-eye daisy', 'Margaret' < Gr.-Lat. MARGARITA, zafferano 'saffron' < Ar. ZA'FARÂN. Perhaps the con trast between It. maccheroni and E. maccaroni would be an example in point, if only the controversial word's exact route of diffusion, to say nothing of its ultimate extraction on Italian soil, had been more firmly established. Labeńnto beside more learned (and more standard) labirinto, which handbooks sometimes adduce as an example of change,6 actually illustrates noteworthy preservation of the Y = Ĭ of Gr.-Lat. LABYRINTHU at its early Romance stage [e] and is thus, at best, compar able to cameretta 'small (bed)room or chamber' or to venerdi 'Friday' < VENERIS DIĒS 'day of Venus'. Obs. gue- beside mod. guarigione 'cure, recovery' and meraviglia 'wonder, marvel, amazement' < MĪRĀBILIA (pl.) 'marvelous events', which are sometimes lumped together with the preceding examples, actually stand apart on several counts.7 Closely related to the cases under study — but by no means identical with them — are the instances of pretonic, rather than intertonic,  either (a) preserved before r, as in periglio, the obsolete doublet of pericolo 'danger' < PERÏCULU, and as in (neg.) veruno 'any(body)' < VËRÈ ÚNU lit. 'truly one' ֊ both in significant contrast to sicuro 'sure, certain' < SËCÜRU, ֊ or (b) secondarily introduced before r, (a) at the expense of a: obs. ghe- ~ mod. ga-rofano 'dried flower-bud of clove tree' < Gr. KAR(Y)ÓPHYLLON, ser- beside more common sar-mento 'vine-branch, twig, tendril' < SARMENTU, smeraldo 'emerald' < Gr.Lat. SMARAGDU, or (β) as a substitute for some other vowel, e.g. obs. cerusico (beside more erudite and more modern chirurgo) 'surgeon' < Gr.-Lat. CHÏRURGthought, either the influence of manipolare or the use of z' in certain compositional schemata may be involved); neghittoso, from NEGLECTU 'carelessness' (the -oso derivative may go back to colloquial Latin); nevicare 'to snow' (from *NIVICARE, independently supported by Fr. neiger; here the rôle of the suffix -icare might have been individuated); and ubbidire 'to obey' -erfrom the intertonic vowel in proparoxytones (as in camera, attested directly in Latin) to the intertonic syllable before the main stress (as in margherita and in the verbal endings).

12) Further examples can be supplied from a profusion of handbooks (R. Menéndez Pidal's Manual de gramàtica històrica, etc.). The fact that, for a while, Meyer-Lübke toyed with the possibility of charging some of these forms to assimilation at a distance ("Fernassimilation"; see his Historische Grammatik der französischen Sprache, I, § 226) rather than to the contiguity of r to e, does not alter the main current of thought. 13) Thus, Grandgent, in a footnote to § 62, muses, apparently unaware of the circularity of this thinking: "The choice of e when r follows, and i elsewhere, is probably a phenomenon of association, e being in the whole mass of words the commonest intertonic vowel before r, i the commonest before other consonants".

372


(d) Understandable only in structural terms is the rise, on the local scene, of a special affinity between  and r, in response to no doubt preexistent affinities between a and ո (also m, c, g, ƒ) and between  and I. Once a speech community recognizes this category of relationship in two cases, it takes its members no inordinate effort to extend it to a third case. The general weakening of a to  in unstressed syllables would have been in any language a very trivial event calling for an entirely simple explanation. The fact that in a certain variety of Tuscan — on cultural grounds the most influential of all — a becomes  before, of all environments, r is something of a paradox, at least in the context of Romance linguistics, and is best grasped as a consequence of a most unusual concatenation of circumstances.

II. An Elusive Pattern of Consonant Dissimilation in Italian. The so-called sporadic, spontaneous, or saltatory sound changes — Ascoli's famous "accidenti generali" — have, after an extended period of relative eclipse, once again occupied a position of prominence in the very focus of scholarly discussion. This return to favor is, of course, due to the renewed general concern with "constants" and "universals" rather than with "particulars" and diversities these days; the volteface has been dramatized by many leading linguists' relative indifference to this category of change thirty or forty years ago. But return, in imaginative scholarship, can never be more than partial, lest it degenerate into repetition. While the pioneers either eagerly stressed, or vehemently denied, certain basic differences between sporadic and regular sound shift, today the interest has veered predominantly toward the detection of coincidences (better still, of compatibilities) between characteristic saltatory shifts, such as dissimilation,'on the one hand, and, on the other, various different kinds of change, down to lexical blends.14 Let me illustrate this state of affairs with an unusual, not yet cogently explained, pattern of consonant dissimila tion in Italian. (This time there will be no need to draw a sharp line between the literary language and the dialects, let alone to concentrate on a single dialect.) Italian has a handful of words in which the interrupted sequence r. . . r changes either into r. . . d or into d . . . r. Such monographs and handbooks as record the possibility do not distinguish sharply between these two treatments; specifically, they fail to observe that the former involves a "mixed bag" of nominal and adjectival formations, while the latter has been reserved strictly for verbs, thus: (a) ARMÄRIU 'chest, cupboard' > armadio (influenced by madia 'breadbin'?); CONTRĀRIU (adj.) 'opposed, opposite', (noun) 'opposite manner or direction' > contrario ~ obs. contradio; Med. Lat. PORPHYR(i)U, from Gr.-Lat. PORPHYR(ETIC)U 'hard rock containing coarse crystals in finer-grained ground-mass' > porfido (Dante); Gr.-Lat. PRŌRA 'bow, prow' >prora ~ (orig. dial.)prua 'id.' beside proda 'beach, foreshore', cf. the phrase andare a proda 'to hug the coast'; RÄRU 'thin, scattered, infrequent' > rado 'sparse, rare', cf. the adverbial phrase di rado 'seldom', beside raro 'rare, uncommon, exceptional' (there seems to be a partial overlap 14) For details see fnn. 21 and 27, below.

ETIOLOGICAL STUDIES

373

15

between these two doublets); (b) CONQUÏRERE 'to seek out' > Olt. conquidere 'to conquer'; FERÏRE 'to strike, knock, smite' > ferire beside Olt. fedire (rare var. fiedere); (IN)TERERE 'to rub, grind' (p.ptc. TRÏTU) > intridere 'to soak, make into a paste, knead, moist en, stain'; QUAERERE 'to look for, search' > chiedere beside Olt. cherere 'to ask'. Within this exiguous group, chiedere clearly outranks all the other verbs in importan ce (moreover, conquidere is its satellite); fiedere seems to have been minted in ephemerous imitation of chiedere, while intridere belongs to a notoriously difficult and capricious word family, witness the questions, formal and semantic, raised by OSp. (de)rretir and OPtg. (de)rreter 'to melt' as descendants of RETERERE 'to grind thoroughly'.16 What complicates the situation most is not only the conspicuous distance be tween these two subgroups (one discovers neither transition nor overlap), but, above all, the lack of any neat relationship to a far more plentifully represented pattern of dissimilation,17 again with two subgroups: r . . . / and I . . . . r, which this time do display a modicum of rivalry (observe in the following survey of the record the self-contradictory, rather than mutually conflicting, behavior ofARBORE and RŌBORE): (a) ARBORE 'tree' > Cal. arvule; CŪRĀTOR 'guardian, overseer' > Sic. Cal. curátulu 'supervising sheepherder'; M E R C U R T DÏE 'Mercury's day' > mercoledi (obs. var. mercoldï) 'Wednesday'; MORTĀRIU 'mortar' > mortaio ( bound form: mortar-), dial, -aro beside the dimin. mortar-, mortal-etto; RÄSÕRIU 'shaving gear' > Sit. rasolu 'shaving blade'; REMORĀRĪ 'to linger, remain behind' > remolare alongside the postverbal noun rémora 'delay, impediment; wake, slick'; RŌBRE 'common oak' > Sic. ruvulu ; (b) ARBITRU 'witness, spectator', 'judge' > arbitro ~ àlbitro 'judge'; ARBORE > It. albero; CEREBRU 'brain' > OIt. c(i)elabro beside mod. cerebro; Goth. HARI15) It is questionable whether we are dealing with r . . r or with r . . . rj in the case of ARMÄRIU . . . CONTRĀRIU. If the latter interpretation is applicable, one is prompted to adduce as a parallel the rare Portuguese by-form martidio < MARTYRIU (cited by Grammont and, in the wake of that scholar, by Togeby); a striking alternative is offered by OSp. contrallo. Un questionably, one must operate with / . . . / / (rather than with / . . / ) in trying to understand It. giglio 'lily' < LÏLIU and gioglio ~ loglio 'darnel' < L O L I U (as well as hypercorrection under lying Giuglio ~luglio 'July' < I U L I U ) , which once puzzled Grandgent ( § 84.4: "unexplained"). 16) On conquidere and its p. ptc. conquiso see the valuable documentation from older Italian uterature offered by  Delcorno in RPh 25.321 (1971-72). (RE)TERERE poses thorny issues for the student of proto-Spanish and -Portuguese as well; see my two - mutually complementary - papers (in press), "Two problems of Hispanic morpho-etymology", in Studies in honor of Tatiana Fotitch (Washington, D. C , 1972), and "Perspectives d'un renouvellement de l'étymologie romane", in the transactions of the 1971 Congress (in Québec) of the Société de linguisti que et philologie romane. 17) The r. . . / solution is familiar from other Romance languages, notably from Spanish, where we find not only the series árbol 'tree', cárcel 'jail', mármol 'marble', etc. and individual cases of cleavage like ralo 'sparse' beside raw 'infrequent, odd' < RÄRU, but also keen rivalry between -/ and -r in syllable-filling function, cf. alquile(r) 'rent' beside albaňi(l) 'mason'; in particular, there occurred protracted competition between alfiler (which ultimately won out) and alfilel 'pin', from older Hisp.-Ar. alfilé. See Anita Katz Levy's important article in RPh 18.399-429 (1964-65), 20.296-320 (1966-67).

374


BERGO or Frk. *HERIBERGA 'military lodging' > albergo 'inn, hostel, hotel'; Gr.-Lat. MARGARITA 'pearl, kind of flower (daisy), woman's name' > Mar- ~ Mal-gherita; Celto-Lat. (PARA)VERĒDU 'swift horse, hunter' > palafreno 'palfrey, saddle-horse' (transmitted through Gallo-Romance and influenced by freno 'bit, bridle'); PEREGRÏNU 'foreign, strange' (akin to AGER) > pellegrino 'vagrant, foreign', 'pilgrim' beside learnèd peregrino՝, *PURPURE 'purple' >pólpore (adjusted to native FURFUR 'bran, scurf) alongside the more conservative Hne of descent: Gr.-Lat. PORP(H)YRA > PURPURA > pórpora; RŌBORE > Sw.-Lomb. (Lugano) luvra- SORŌRĒS 'sisters' > Apul. selure (apparently with concurrent vowel dissimila tion); Celto-Lat. VERT(R)AGU, -A (var. VERTAGRA) 'greyhound' >veltro (pre sumably via OFr. or OProv. veltre). As if this degree of intricacy were not enough, there was further available to speakers of Italian a third, rather widespread escape from phonic infelicity, namely dissimilatory loss either of the first or of the second r: (a) DĒ RETRŌ 'from behind' > dietro 'behind' (also the compounds ad-, di-, in-dietro), undoubtedly supported by entro 'within' and contro 'against' (from INTRA and CONTRĀ, respectively, with the sharply etched ending of dopo 'after' < DË POST, sotto 'under' < SUBTUS, verso 'toward' < VERSU, etc.); Gmc.-Lat. FREDERĪC U > Federigo՛, է opon. (S. Lazio) PRIVE RN U >Piperno (through concur rent facetious association with the nickname Pippo from Filippo, in turn coined much like Sp.Pepe from [Jo]sepe and like E. Bob from Rob[ert] ); (b) ARÄTRU 'plough' > Sic. Cal. aratu (cf. Sp. arado) beside It. aratro; CICER 'chick-pea' > cece; CRÏBRU 'sieve' > Cal. crivu beside It. cribro, Fr. crible (Spanish tolerates the residual verb cribar, but prefers as nouns either cedazo < SAETÄCEU, originally a mere qualifier, or borrowed tamiz); DË RETRO >drieto, as an alternati ve to dietro, cf. *DËRETRIĀNU (adj.) 'behind', (noun) 'rear, bottom' > deretano; FRĀTER 'brother' > fra(te) 'friar'; MARMOR 'marble' > marmo; OPPROBRIU 'reproach, taunt, scandal' > obbrobrio ~ brobbfrjio 'disgrace, infamy', PROPRIU 'own' > proprio, var. propio (as in Spanish); RĀSTRU 'rake' > Cal. rastu, cf. Ptg. rasto 'trace, track'; SOROR (nom.) 'sister' > suoro, mod. suora. Even this modest specimen of data raises multifarious questions and throws noteworthy shafts of sidelight (e.g., on the close ties between South Italian and Hispano-. incl. Luso-, Romance); but such spells of alertness and such fresh insights do not contribute much to the clarification of r . . . r > d . . . r ~ r . . . d, which served as our starting point. One readily guesses the phonotactic conditions that may have counseled either modification or elimination of either r; but one does not immediately recognize the delimitations between the two competing remedial treatments short of total loss, as in r. . . d vs. r. . . I, etc. This impasse makes it seem advisable to inventory and reëxamine some of the older attempts at analysis. The original approach, exemplified by U. A. Canello's pioneering inquiry into doublets in Italian (with a kind of running commentary from the editorial pen of G. I. Ascoli),18 focused the reader's attention on clear-cut instances of consonant 18) "Gli allotropi italiani", AGI 3.285409 (1878), at 360 ( § 64), with a discussion of the hypothesis, championed by F. Diez, that proda may be an amalgam of two homonyms, one of which is traceable to Germanic. In a footnote Ascoli advocated the Genoese descent of prua.

ETIOLOGICAL STUDIES

375

dissimilation (proda and rado, say), leaving entirely out of account the -dere < -RERE verbs. It was Meyer-Lübke, in 1890, who ֊ in his Italian rather than his comparative Romance grammar — dragged in chiedere,19 apparently without asking himself why of the numerous Italian verbs with infinitives in -rare (e.g., arare 'to plough', divorare 'to devour', girare 'to swing round', tirare 'to pull', varare 'to launch', virare [naut.] 'to go about') not one has cast off a variant, let alone a success ful substitute form, in *֊dare. Such has been, for better or worse, Meyer-Lübke's prestige that, despite this obvious objection and despite other qualifications, yet to be mentioned, the equation QUAERERE > chiedere has figured among cases of dissimilation to this day, invading in particular the influential grammar by G. Rohlfs.20 Yet criticism of the vulnerable passage began at an early date. In writing his celebrated doctoral thesis on consonant dissimilation, M. Grammont cited the opinion of a friend and fellow-student — none other than A. Meillet — to the effect that chiedere and conquidere arose not through dissimilation, but in response to analogical pressure: Since the p. ptc. chiesto appeared to echo visto 'seen', the infini tives were also brought together, with chiedere bowing to the superiority of vedere, just as conquidere (p. ptc. conquiso, despite the noun conquista 'conquest', cf. Fr. conquis but conquête) fell under the spell of rather numerous -dere/-so verbs.21 Three years later, G. Paris, in reviewing Grammont's thesis, at once recognized the flaw in Meillet's argument.22 The stress pattern of vedere < VIDĒRE, a paroxy19) Italienische Grammatik 162 (Leipzig, 1890) ֊ an indirect reference. In his Romanische Lautlehre of that same year, under "Dissimilation", the author no longer adduced either chiedere or fiedere. 20) HGIS § 328, with the specific comment: "Dann auch auf andere Verbalformen übertragen". 21) La dissimilation consonantique dans les langues indo-européennes et dans les langues romanes 120f. (Dijon, 1895), under the chapter "Etymologie populaire, croisements, jeux de mots, etc." (113-26), which enters into Part 2 ("Mêmes effets, causes différentes") and pro vides the starting point for K. Togeby's eventual elaboration. Note that a very young Meillet here appears as Grammont's friend, as distinct from his five teachers Bréal, Saussure, Arbois de Jubainville, Schmidt, and Thurneysen. Acting on his own, Grammont declares himself unfamiliar with OIt. fieder e ; centers attention about conquisto, disregarding conquiso՛, and cites, as models for innovative intridere/intriso, such couples as chiudere 'to close'/chiuso, ledere 'to harm, injure'/ leso, redder e 'to cut (off), curtail'/reciso. 22) I owe to Togeby's article (see fn. 28, below), p. 661, the reference to this verdict, in cluded in the assessment which appeared in the Journal des Savants (1898). The rash association of chiedere (and sometimes also fiedere) haunts such later treatments as Grandgent's From Latin to Italian § 104; Bartoli's revision of Meyer-Lübke's Grammatica storica 118 [ § 1 4 2 = 283]; B. Wiese, Altitalienisches Elementarbuch2, 60. From here the path leads to Rohlfs (1949); see fn. 20, above. 23) In his REWi the author split the Romance material into (a) the descendants of [*]PLŪRÏRE: Lomb. spyüri, Emil. (Reggio) spyurir, Gal. kyurire, Sic. (Noti) Čuriri, with derivatives and back-formations in Mantua, Verona, and Parma, and with a blend (RODERE being the part ner) in Naples and Abruzzi, all of which amounts to a fairly even distribution over northern, northcentral, southern, and insular Italy; and (b) the descendants of PRŪDĪRE, found chiefly in the West and the East of the Iberian peninsula (Cat. Ptg. prut, Gal. proer), in ancient Provence (pruzir), in northcentral Sardinia (Log. prudire), and in standard Italian (prudere); remnants of PRÜRÏGO were assigned to Piedmont: priiizu, and to Lombardy: sp(y)üriz (according to

376


tone, made it an unlikely model for chiedere, a proparoxytone, whereas radere 'to scratch' /raso, prendere 'to take'/preso were more readily acceptable for conquidere -- except that it was hazardous to credit a form as isolated as the infinitive with the restructuring of a verb's entire paradigm. It is perhaps unnecessary to follow every step of the discussion over the subse quent period of sixty-odd years, a timespan during which the debate went on sim mering without ever actually reaching the boiling point. During these uneventful decades, however, a sizable volume of information was brought together on the Romance vicissitudes of PRÜRÏRE 'to itch, burn' (and, on a minor scale, on its satellite noun PRŪRĪGŌ 'itch'). Semantically, PRÜRÏRE was a flawless specimen, because the most contentious skeptic could not suspect the reflexes of a verb en dowed with such an earthy meaning to represent words imported from a neigh boring dialect or transmitted through (semi)learned channels. Moreover, in this verb - in stark contrast to FER-ÏRE, QUAER-ERE, and TER-ERE ֊ both r's appeared on the same side of the morpheme boundary, being crowded into the root mor pheme, a situation which opened up entirely different perspectives for the assump tion of consonant dissimilation. As scholars gradually became aware of It. prudere 'to itch, prick', surrounded by many striking dialect forms (Lomb. spyürí, Cal. kyurire, etc.), as well as of OProv. pruzir, Cat. Ptg. pruir, Gal. proer, they were led either to subsume this profusion of variants under the etymological triptych PRÜ RÏRE, *PLŪRĪRE, *PRŪD-ĪRE/-ĚRE, which is almost exactly what W. MeyerLübke did in two versions (1911-20, 1930-35) of his comparative etymological dictionary (§ 6802), or even to project these reconstructions, at least by implica tion, onto the temporal plateau of provincial spoken Latin.23 Less impressive on this particular score have been the recent accomplishments of two otherwise distinguished students of consonant dissimilation in Romance. Rebecca Posner, on two meaty pages of her Oxford dissertation,24 assembled many valuable stray data and bibliographic hints (however, the critically important contro versy between E. G. Wahlgren and T. Navarro, apropos of r ~ [δ], somehow eluded her attention),25 but fell short of achieving any cogent synthesis and even confessed C. Salvioni). In revising his dictionary, the author merely corrected the spelling of Parm. sparena and traced Log. prudire to Italian rather than directly to Latin. He also concerned himself with this verb in his Einführung2 147 (Heidelberg, 1909) and his Einführung3 176 (1920). 24) Consonantal Dissimilation in the Romance Languages 124-6 (Oxford, 1961;Publ. of the Philological Society 19). 25) For a brief analysis of this clash of opinions (Wahlgren's Uppsala monograph, Un pro blème de phonétique romane: Le développement "d"> "r", appeared in 1930; Navarro's review in RFE 18.383-5 one year thereafter), see my article, "Multiple Versus Simple Causation in Linguistic Change", To Honor Roman Jakobson 2.122846 (The Hague, 1967). I have stated and defended elsewhere my belief that Sp. Ptg. mentira 'lie' must be divorced from Cat. mentida and aligned with its semantic opposite vera(s) < V Ē R A , perhaps also with jura 'oath, sworn statement'. 26) Posner's conclusion (126) appears to lack a certain modicum of firmness: "It is probable, then, that /d/ is not normally substituted for dissimilated /r/. In the examples where such a dissimilation does seem to have taken place, there may be analogical influence [i.e., lexical blend], or epenthesis of [d] [i.e., original dissimilatory total loss of r, then insertion of antihiatic d] — or the [d] may be a dialectal continuator of more regular * [l]. . . or r-d falls into a time wrhen there was no intervocalic / available in proto-Italian".

ETIOLOGICAL STUDIES

377

to feeling uneasy about the fact that d - in Romance, a phoneme, as a rule, less frequent than r — should have quite often qualified as a substitute. On the positive side of the ledger, let me credit to her such partial findings and improvements as the careful distinction between [d] and [δ]; the discovery of the characteristically Italian predilection for -r- > -d-; the unearthing of very early, isolated attestations for RAD- < RÄR- and for quedere < QUAERERE (a fact which of course does not mean that at those archaic stages the variants were widespread, let alone domi nant). 26 In his brilliant and weighty review of Posner's book, Knud Togeby made numerous individual suggestions which either improve earlier knowledge or are apt to invite contradiction27 and thus to spark fresh discussions.28 His general thesis is to confine the agency of consonant dissimilation quite radically (still, he cites, under § 29, It. rado and mercoledi as authentic illustrations) and, in most instances, to shift to other forces the responsibility for the results. Thus, in speaking of It. and Sp. porfido, he emphasizes the special status of the word (borrowing); in accounting for It. prudere he appeals to a blend (§ 19), namely to conflation with RODERE 'to gnaw' allegedly recognizable also in PRÜRÏRE's shift to a different conjugation class (an argument, let me add on my own, which holds neither in Provençal, nor in Catalan, nor indeed in Portuguese); regarding chiedere, Togeby defers to the judgment of Grammont - i.e., in the last analysis, of Meillet (§ 23). Like the Posner book, the opinion voiced in Togeby's review article remains a valuable link in a chain of pronouncements, but does not preclude further study. In trying to sift the explanations so far advanced, to consider the wisdom of appealing to other agencies, and — after all the necessary additions and subtractions have been made — to arrange in a coherent whole the individual factors held respon sible for the unusual dissimilatory process, it will prove useful to separate far more sharply than has so far been done the (nominal) cases of r-r > r֊d from the (verbal) cases of r-r > d-r. The former represent a very motley crowd or crowdlet, defeating any attempt at quick discovery of a common denominator (porfido is a technical borrowing, contradio may be semilearned, armadio has been attributed to lexical blend, on proda there has been no complete unanimity of etymological expertise, 27) The individual vicissitudes of many of these words cannot properly be presented here. An analyst as sober as Grammont accepted Wölfflin's pioneering conjecture (ALLG 7.606) of a cross of *MEDÏDIËS 'mid-day' and MERUS 'pure, unmixed, sheer, complete'. The relation between the regional vars. párpado and párparo 'lid' cannot very well be laid down without previous etymological commitment; unfortunately, Cl. Lat. PALPEBRA (which has safely survived in Fr. paupière) was ringed by all sorts of variants (PALPE-BRU, -TRA, etc.), an unusual case of polymorphy, and if *PALPETRU, recommended by Corominas as the immediate proto type, were accepted, then párpado and párparo could be declared equidistant from it. In lámpara one has to allow for the agency of the "unstressed suffix" (better: augment) '-aro, -ara; on mentira 'lie' its semantic opposite vera(s) 'truth' may, I repeat, have left an imprint (see RPh 6.121-72 [1952-53]), etc. 28) "Qu'est-ce que la dissimilation?" RPh 17.642-67 (1963-64), with an excellent biblio graphy, especially a digest of the chorus of critical reactions to Grammont's thesis. Several pas sages of that paper are relevant to the present study; I cannot go into details except for noting that It. dietro 'behind' < DË RĚTRŌ, which baffled the author, in all likelihood echoes in its concluding segment such prepositions as contro 'against' and dentro 'inside'.

378


etc.), except that dissimilation of either r.. . r or, in two instances, of r. . . rj, actual ly seems to be at issue. The latter involves an indisputably dissimilatory process only in the case of PRŪR-ĪRE, where two R's, out of a total of three, form part of the root morpheme. As regards the progeny of QUAERERE and CONQUĪRERE, of FERÏRE, and of TERERE, the succession of r's characterizes only a certain part of the respective paradigms: the infinitives and the two tenses based thereupon (future and conditional). To some extent, at a certain evolutionary stage nouns and verbs undeniably could have supported one another, despite the cleavage r. . .d/ d . . . r, but, in terms of isolable factors of initial causation, it is advisable to keep the two groups apart. Once that much has been settled, one may invoke as the prime reason a basic phonic affinity between d (or [δ]) and r pronounced with the tip of the tongue - a panchronic, pantopic proximity observable in a variety of languages (cf. Lat. MEDI- 'half' + DIĒS 'day' > MERIDIĒS 'noon', Sp.párpado ~ dial.párparo 'eyelid', Gr.-Lat. LAMPADA 'lamp' > OSp. lámpara)27 This latent propinquity could have asserted itself where additional motivations for the choice of r were operative (in Spanish, the powerful attraction exerted by the unstressed final segments '-aro, '-ara) or where the change, through injection of new associations, bade fair to enrich the word's semantic load (MERIDIĒS calls to mind MERUS 'pure', suggesting suffusion in bright daylight), or where the alternative choice of I happened to be blocked (through menace of some culturally undesirable association). This trend — or at least one of its grooves — might have benefited rado, apparently the oldest word displaying the shift. A second, independent force is the one that has haunted the imagination of several form-conscious French scholars (Meillet, Grammont, Paris): Because several '-dere or ֊dére verbs in Italian had a p. ptc. in -sto or, more frequently, in -so, there could very well have existed a stimulus for speakers to place, beside chiesto, chiedere; also alongside conquiso, conquidere, etc. A third force which, I suspect, has not yet been placed in the limelight of dis cussion is the Italian speech community's tendential recoil from syncopated futures and conditionals. This is not the place to review the latest thinking on the origin of the Romance future tense, or, more narrowly, to examine the controversial relation of syncopated infinitives (such as scene 'to choose' vs. scegliere) and syncopated futures (such as scerrò vs. sceglierò).29 29) Some of the more recent pronouncements on the prehistory and early history of the Romance future, following upon E. Coseriu's essay (1957), include: É. Benveniste, "Mutations of linguistic categories", Directions for historical linguistics: a symposium 83-94 (eds. W. P. Lehmann and Y. Malkiel; Austin & London, 1968) and several critical reactions to this paper; P. Valesio, "The Romance synthetic future pattern and its first attestations", Lingua 20. 113-61, 279-307 (1968), cf. V. Pisani in Paideia 24.89 (1969); J. L. Butler, "Remarks on the Romance synthetic future", Lingua 24.163-80 (1969) - in criticism of Valesio's article - and the latter's rejoinder, "The synthetic future again: phonology and morphosyntax", ibid. 181-93, as well as an independent note from Valesio's pen, "La genesi del futuro romanzo",Z,eS 4.405-12 (1969), in turn reviewed by Ana Maria Barrenechea in RPh 25.326 (1971-72) - the critic's verdict: the argument is so far still inconclusive; finally, a lengthy section of Butler's review of W. Mańczak, Le développement phonétique des langues romanes et la fréquence (Kraków, 1969) in RPh 25.331-6, esp. 334-6. Benveniste credits a certain type of clause with having given biïth to the use

ETIOLOGICAL STUDIES

379

At the crest of the vogue of syncope  could be freely eliminated between two r's: dimorrò, liberrá, while metathesis accounted for enterrò < entrerò = mod. intrerò (Grandgent, § 201). In some dialectal and even literary varieties the vogue of an -rr- in future and conditional became so irresistibly strong as to have given rise to guiderrà 'he will lead', presterrò 'I shall lend or loan', crederrò 'I shall believe' (Boccaccio, Sacchetti, Pulci), in Heu of expected — and actually recorded - guiderà, etc., which ultimately won out; not infrequently, consonant assimilation (*-lr> -rr-, etc.), eminently characteristic of Italian, led to syncopated futures even where the infinitives themselves remained immune from such compression: dorrò T shall lament' (from dolere), rimarrò T shall remain' (from rimanere), terrò 'I shall hold' (from tenere), varrò 'I shall be worth' (from valere), venò 'I shall come' (from venire), vorrò 'I shall want' (from volere) — obviously on the model of morrò 'I shall die' (beside morire) and parrò 'I shall seem' (beside parere). All this is well-known — and, to boot, trivial. Less clearly understood has been the extent of the reaction against this very same drift toward syncope, a reaction dictated by the speakers' urge not to fragment beyond a certain degree the tightknit conjugational paradigm of a verb. The partial abandonment of radically syn copated forms, chiefly those exhibiting -rr-, has been achieved in a variety of ways, some of them immediately arresting the analyst's attention, others subtle and almost imperceptible. Thus alongside the overt rejection of certain excessively "short" infinitives, futures, and conditionals (in the end, scene has yielded to scegliere < EX ËLIGERE, corre to cogliere 'to pluck, gather' < COLLIGERE, sciorre to sciogliere 'to loose(n), undo' < EX SOLVERE, sverre to svellere 'to uproot' <E(X)VELLERE, torre to togliere 'to take away' < TOLLERE; on the other hand, bere 'to drink' < IBERE is still firmly entrenched), there have occurred all sorts of indirect retreats and evasions. The inconvenience of parrò can be avoided if recourse is had, instead, to such quasi-compounds as ap-, com-parire, i.e., to cognate verbs whose i is less subject to erosion than is the e of parere, hence apparirò; note also sparire 'to disappear'. The full importance of this bifurcation becomes understand able when one recalls such unbroken series as Fr. paraître, -, com-, dis-paraître; Sp. parecer, a-, com-, desa-parecer, to say nothing of their Latin prototypes. Where a verb is used in the future quite sparingly, as is true of solere 'to be wont to, in the habit of, the risk of syncope has been warded off from the start (solerò, not *sorrò). Even if sverre occasionally still flanks svellere as the less favored variant in the future, the choice lies clearly between svellerò and transparently analogical svelgerò, patterned after volgerò T shall turn' and, especially, svolgerò T shall unroll, unwind', whose -g-, in the last analysis, is also analogical and traceable to dissimila tion (VOLVERE). There also exist inconspicuous lexical replacements; thus, one outside observer of the Italian scene has recently remarked: "Mettere crowds out porre, which now occurs principally in compounds . . . ; for all practical purposes, menaré and portare have dislodged condurre. Trarre is a relic form, enjoying little currency among of HABĒRE as an auxiliary verb in Late Latin, Butler contradicts Meyer-Lübke in affirming that "irregular" infinitives (e.g., bere 'to drink', fare 'to do') arose in Italian from kregular futures. He and Valesio are at loggerheads as regards the heaviness of the stress hitting the auxiliary.

380


speakers".30 If this is so, what stands in the way of assuming that one major advan tage attaching to the shift OIt. ch(i)erere > chiedere was the ensuing freedom from the hazard of syncope (fut. *cherrò, cond. *chenei, inf. cherre), which menaced to blur the neatness of the verb's total paradigm — obstructing the link to pres. indic. chiero, impf. ch(i)ereva, etc.? There was no dearth of verbs in -dere (including those in -ndere and -rdere) that no doubt facilitated the transaction, casting a net of formal and semantic similarities; and the speakers' hesitancy demonstrably surrounding PRŪRĪRE provided another concomitant or, rather, catalyst. But while we must make allowances for such secondary circumstances, the growing aversion to -rre infinitives, -rrò futures, and -rrei conditionals may very well have been the clinching factor in the switch from ch(i)erere to chiedere (the same holds for its satellites).31 The history of Hispano-Romance inflection furnishes a striking parallel. Here syncope had likewise been at work, and a discernible recoil from it also made itself felt. Thus, OSp. é  shall die' in the end yielded ground to moriré (the last trace of the -n- survives in the Ptg. inf. moner); poné, tené, verré (from porter, tener, and venir, respectively) competed for centuries withponré, tenré, venré (i.e., variants lacking the element of assimilation) and with pondré, tendré, vendré (i.e., variants involving a homorganic buffer consonant) as well as with porné, terné, verné (i.e., counterparts exhibiting metathesis). Significantly, to this day, querria T should like' is being avoided in the standard, as long as one can — through a syntactic mechanism - substitute for it quisiera (and for symmetry's sake a native may prefer pudiera to podria), while certain dialects go so far as to operate with quedré, model ed on podré 'I shall be able to'. (Note that in Spanish, unlike Italian, a further compli cation arose through the crystallization of /R/ — often corresponding genetically to rr— as an independent phoneme, which in some regional varieties has drifted away so briskly as to have become phonetically unrelated to /r/.) In Old Portuguese, the anomaly morir 'to die' (moiro T die') ~ rré shall die' was in the end elimi nated by total restructuring of the paradigm: inf. morrer (its -rr- due to the medieval future, its -er provoked by polarization with the verb's semantic opposite viver 'to live' < VĪVERE) ~ morro ~ monerei, a perfect example of morphological leveling.32 But if It. chiedere and its satellites (conquidere, fiedere, intridere) at their initial stage are best accounted for in terms of conjugational analogy (neatness of the contour of the root morpheme) rather than in terms of sporadic sound change

30) See the aforecited book review in RPh 25.336. 31) It cannot be argued that the early appearance of chiedere (R. Posner cites one isolated 9th-century example) disqualifies it from any assumed participation in a "rebound". The -d and the -r- forms coexisted for several centuries; an initial mistake, possibly confined to a small area, became entrenched and later spread, because it had struck a chord in the speakers' hearts; had met a preexistent expectation or disposition; offered a distinct advantage (whichever meta phor one prefers). Note Sic. sediri (Rohlfs). 32) On the history of morir/morrer see my paper, "Español morir, portugués morrer . . .", BHi 57.84-128 (1955). For clues to earlier discussions of quedré (querré) (A. Bello, R. J. Cuervo, A. M. Espinosa-padre) see F. Hanssen's 1913 grammar, § 261. Though quedré is visibly patterned on podré, and is genetically distinct from It. chiederò, the adventitious d in both contexts admirably serves the same purpose.

ETIOLOGICAL STUDIES

381

(consonant dissimilation), the development went one step farther. The vicissitudes of PRŪRĪRE, the temporary coexistence of *chiero and chiedo, plus a minimum of articulatory and auditive affinity of (trilled) r and (post-dental) d seem to have created a propitious climate of fluidity of usage at which r could be replaced by d outside the verb, under conspicuously favorable circumstances. One such maximally favorable set of conditions was the speakers' recoil from the obnoxious sequence r-r. At this point chiero/chiedo may have started to exert its share of pressure on the nominal doublets raro/rado, prora /proda, etc. If the course of events here postulated is correct, then a close parallel has been furnished to the processes observed in Spanish over the last few years — sound changes rooted in, or favored by, constellations of morphological facts: -LG- > -lz-, -NG- > -nz-, -RG- > -rz-; ie > i and, secondarily, ue > e; s + Cons. > z + Cons.; -ç- > -z- in such derivational suffixes as -azar, -uzar; -azo, -izo; -azón; and -ez(a).33

III. Choice of a Niche in a Sound System and Search for Causation. Since the days of pioneering comparative grammars changes of sounds and forms have been recorded in some kind of system, and tacitly implied in the choice of the presentation was the awareness of bonds, cross-connections, and common causes. If, for instance, a language historian places together initial consonant clusters of the source (or parent) language, he will before long discover that those involving l as the second element (bl-, cl-, fl-, etc.) display a notable parallelism in their further growth. From there one is led to seek the cause of the shifts observed (say, It. -, chi-, etc.; Sp. (b)l-, ll-; Ptg. br-, ch-) in the configuration, position, and character of these clusters; perhaps also in their relation to the remainder of the sound system. In most instances this time-honored procedure leads to satisfactory results, and there is no need to abandon it. On certain occasions, however, the researcher ends up in a blind alley; the first, tentative causal explanation (if he is fortunate to hit upon one) may either involve a minor, superficial ingredient of the total complex of causa tion or may lead him completely astray. Under such circumstances, it is legitimate to ask oneself whether the transfer of the observation post to some point other than the one favored by those analysts to whom we owe the original inventories might not prove advantageous. Three such exceptional cases will be briefly examined here, with illustrations drawn from the history of Spanish. The contrast between Sp. cual and calidad, cuanto and cantidad, cuatro and catorce has been known for quite some time, and the deviation from such neat treatments of the Latin QUA- cluster as Fr. quel ~ qualité, quant ~ quantité, quatre ~ quatorze, consistently with /k/, on the one hand, and, on the other, It. quale, qualità, quanto, quantità, quattro, quattordici, consistently with /kw/, could not have been more conspicuous. The difficulty lies in the fact that the confusing state of 33) For a summary, with bibliographic hints, of my earlier inqmries so slanted, see the paper, "Morphological analogy as a stimulus for sound shift", LeS 4.305-27 (1969); add the latest piece, in RPh 25.1-52.

382


affairs in Spanish is not properly elucidated as long as one lists and discusses the process under word-initial consonants or consonant clusters, as has so far been rou tinely done.34 A further complication arose through unqualified application of "Verner's Law" to the distribution of ca- vs. cua՛: While the position of stress for once does seem to play a certain rôle in the split, it may represent a mere concomi tant. 35 It is the contention of this section of the paper that the prime reason for the seemingly erratic distribution of/k/ and /kw/ was the sudden emergence of the rising diphthongs /wa/ beside /we/ and /wo/ as a characteristic of the stressed syllable in central Hispano-Romance, to the extent of reqmring, as if by recoil, secondary monophthongization of a few old diphthong-like sequences which happened to occur in a pretonic syllable. (The countertonic syllable was not affected in Old Spanish: mod. Cuaresma 'Lent' < OSp. Quàraésma < QUADRĀGĒSIMA; mod. cuarenta 'forty' < OSp. quàraénîa < QUADRÄGINTÄ.) In other words, causally the process at issue belongs not under consonants (or combinations of consonant and semiconsonant), but under secondary reactions to diphthongization, in the chapter on vowels. One classic crux of Spanish phonology is the loss of G- before front vowels: GERMĀNU 'own brother' > ermano (later, for no good reason, spelled hermano), GINGIVA 'gum' > enzia. Again, the cognate languages do not go along with Spanish, displaying instead some such reflex as /ž/, /g/, or /j/; Portuguese joins its eastern neighbor on just a few occasions (irmão). In most instances, the syllable affected is pretonic; even GELU 'frost' > (h)ielo and GELAT 'it freezes' > (h)iela are flanked by GELÄRE > (h)elar, etc. If we apply our previous reasoning to the new situation and argue that the "ideal" forms *yermano, *yenzia, *yelar initially existed every where and displayed a sequence interpretable as a rising diphthong, this time /je/, in the least desirable place once stressed Ĕ had become ie, we implicate, once more, the trend away from diphthongization in pretonic syllables as the chief cause and thus shift the attention from the subsystem of consonants, a context in which the problem has ordinarily been discussed with limited success,36 to the subsystem 34) To cite at random a few older authorities: G. Baist, "Die spanische Sprache" § 46 (in: G. Gröber, ed., Grundriss der romanischen Philologie 1 [Strassburg, 1888]) and revised 2d ed. § 45 (1904-06); W. Meyer-Lübke, Romanische Lautlehre § 426 (Leipzig, 1890); A. Zauner, Romanische Sprachwissenschaft2 1.103 (Leipzig, 1905) and Altspanisches Elementarbuch § 67 (Heidelberg, 1908); R. Menéndez Pidal, Manual (elemental) de gramdtica històrica es֊ panola2 § 39.4 (Madrid, 1905); F. Hanssen, Gramática histórica de la lengua castellana § 145 (Halle, 1913). Examples can be multiplied almost indefinitely. 35) On earlier occasions I have argued the inapplicability of "Verner's Law" to Romance conditions; see "Quelques fausses applications de la 'Loi de Verner' aux faits romans", CFS 23.75-87 (1966; = Mélanges A. Burger), and, more recently, "Derivational transparency as an occasional co-determinant of sound change" (I), RPh 25.1-52 (1971-72). What sets apart the cases of q(u)a- and (y)e- discussed in the present paper is the fact that stress plays in them an indirect rather than direct rôle - namely to the extent that losses occur in certain unstressed syllables because speakers reject a few sound sequences which (secondarily) produce the effect of being misplaced diphthongs. 36) Again a selection of illustrations will do: Baist, "Die spanische Sprache" § 44 (2d ed.: § 43); Meyer-Lübke, Romanische Lautlehre § 407; Zauner, Romanische Sprachwissenschaft2 1.83f.; Menéndez Pidal, Manual elemental 2 § 38.3; Hanssen, Gramática històrica § 106. Note, however, that on account of the unusual perspective chosen, Zauner was for once forced to

ETIOLOGICAL STUDIES

383

of the vowels, a far more promising set of circumstances. Yermano(s) is, incidentally, a recorded anthroponym. Older and even more modern texts show such variants as suzidad (suc) 'dirt' beside suziedad, from suzio 'dirty' < SŪCIDU 'juicy'. Since in Spanish there have at all times coexisted variants of the Latin suffix -ITĀTE used in deriving adjectival abstracts, namely vernacular -dad (as in verdad 'truth') and -edad (as in vegedad 'old age' beside vejez) beside learnèd -idad (onestidad 'honesty' from onesto, along side onradez from onrado), one is at first tempted to discuss the antinomy suzivs. suzie-dad either under derivational suffixes or, worse, under channels of trans mission, i.e., levels of style. But suzidad has been by no stretch of the imagination more "refined" than the standard form in -iedad; what is involved is once more the latent tendency - not infrequently blocked - to monophthongize a diphthong like sequence in a pretonic syllable. A concomitant may very well have been the preëxistence of a few originally learnèd words in -idad, which had at a more advanced stage lost some of their pristine aura of solemnity.

A. The Contrast cuatro ~ catorce. The sharp cleavage /kw/ (seemingly in stressed syllables): /k/ (in unstressed syllables) ֊ observable in late medieval Spanish - has been severely blurred through a wide variety of secondary complications. The loss of the intertonic syllable in hiatus toward the close of the Middle Ages left speakers with such atypical forms as cuarenta, Cuaresma, instead of older quàraénta, Quàraésma, which had fallen snugly into place and had contrasted neatly with such examples of authentically pretonic development as camaño 'how big' */ dun dere/ > OSp. unzir 'to yoke' > surprising that, endowed with such keen flak, the great scholar should have refrained from applying a similar analysis to the pseudo-diphthongs q(u)a- and (y)e-. The contrast between Sp. cotidiano and It. quotidiano 'daily' does not lend itself to any parallel discussion in view of the protracted coexistence, on the level of Latin, of the adv. QUO- and CO-TIDIĚ. Exotic words in qua֊, such as the mineralogical term quartz or the tropical phytonym quassia, stand apart.

386


mod. uncir beside dial, uñir) further blur or distort the contour of growth. To recognize, amid this overgrowth, the mainstream of events it suffices to examine in some detail a mere handful of Latin-Romance word-biographies. The vignettes selected for study in this paper are those of GELŌ - ĀRE '(to cause) to freeze', GELU 'frost, chill', GEMELLU (beside *GEMELLICU, *-ÏCIU) and GEMINU 'twin', GEMŌ -ĚRE 'to sigh, groan, bemoan', 'to roar, coo, creak' alongside GEMÏTU 'sigh(ing), groan(ing)', GEMMA 'bud', 'gem, jewel', GENERŌ -ĀRE 'to produce' (plus the corresponding verbal abstract in -ÄTIÕ), GEN-EST(R)A, -IST(R)A 'Spanish broom', GENŪ 'knee' and its two offshoots GENICULU 'knot (in the stalk)' and GENUCULU '(Tittle) knee', GĒNS GENTIS 'clan, stock, folk', GER MĀNU (f. -A) 'sharing one's parents, own sibling',GINGĪVA 'gum',GYPSU 'gypsum', IÄ- ~ IĒ-NUĀRIU 'January', IACTŌ -ĀRE ~ (D)ĒIECTŌ -ĀRE 'to throw (away), fling', IÄ- ~ IË-IUNU 'fasting', IOCŌ -ĀRE 'to play (witty, sophisticated games)' and IOCU 'jest', IOVIS '[day] of Jove' > 'Thursday', TÜDICÖ -ARE 'to judge' alongside IÜDEX 'judge', IUGU 'yoke', IUNCTU 'that which has been conjoined', IU-, *IE-NIPERU 'junipertree', and IUVENE 'young'. As is well known, one can set off three evolutionary lines extending from Late Latin to Old Spanish: (a) One leading to /j/, (b) the other running one step farther to /ž/, which via /S/ - preserved in Asturian - eventually became /x/, and (c) the third involving total loss. It is the last-mentioned which in this context matters most. Line (a) is exemplified by IÜNCTA > yunta, DE(V)ORSU (pron. djorsu or djosu) > yuso 'down' (influenced in its vocalism by its semantic opposite suso 'up' < sŪ(R)SU) and conceivably by (h)ielo, yema,yente, OSp.yemdo (the archaic predecessor, we recall, of gemido). One can be in doubt as to whether in the tail section of this lexical series the word-initial diphthong /je/ represents just ancestral Ë, or an unavoidable merger of much older / j / < G with supervenient /je/ < Ě. The clinching argument, already familiar to Baist, is provided by yeso (var.ls, reminis cent of Valdivielso) < Gr.-Lat. GYPSU which, judging from Tusc. gesso, was pro nounced in Late Latin with a stressed close vowel, i.e., one not at all subject to diphthongization in Spanish. If this inference is correct, then the / j / of (h)ielo, etc. , historically, does contain a whiff of G-. The /ž/ is characteristic of such words as joven, judgar beside júez (later juéz), judío, jueves, and junta. It is traditionally assumed that this pronunciation, on Spanish soil, belonged to a more refined social dialect or has prevailed in a more elegant sector of the lexicon. Thus, aristocratic joven 'youth, adolescent' contrasts with rustic moço 'fellow, chap, guy'; jueves, though pagan in remote background (IOVIS 'of Jupiter'), long since attached itself to the ecclesiastically colored vocabu lary of the calendar; juntar 'to conjoin', junta 'assembly', etc. are, referentially and contextually, more elevated than their crassly rural congeners yugo 'yoke' and yunta 'team of oxen'; judio < IÜDAEU and OSp. judiego < IÜDAICU are redolent of the Bible; judgar and júez are juridical by definition. Any fine-grained study of this contrast between /- and y- must obviously be pursued on the level of micro scopic word-studies. The total loss of G- seems to be attributable to either one of two entirely un related causes. We have already ascribed uncir (var. uñir) to the most radical pattern

ETIOLOGICAL STUDIES

387

of dissimilation of */'dun'dere/, the presumably provincial Latin pronunciation of IUNGERE, and ayuno < IĀIŪNU, as well as echar (witness Ptg. geitar) < IACTÄRE, smoothly lend themselves to the same analysis.39 Conversely, (h)elar, OSp. emelgo beside (e)mellizo, enebro, enero, enzia, (h)ermano, (h)iniesta, and (h)inojo cannot be so explained away; neither can Elvira < Gmc. GELOIRA or (top.) Sant Ervás < SANCTĪ G E R V A S Ï (gen.). Let us assume that originally G- has here, as in the case of yeso, reached the stage of /j/. The reconstructed primitive forms would then be *yelar, *yemelgo, *yemelli֊ zo, *y enebro, *yenero, *yenzia, *yermano, etc. Such words were unimpeachably structured by the standard of proto-Spanish, before the crystallization of new rising diphthongs, esp. of ie, which hit the Spanish language during its formative period fairly late and with the impact of a cataclysm. One characteristic feature of the new state of affairs became the extra-close link between word stress and diphthongization: While in present-day Spanish mueble 'piece of furniture' is accom panied by muebleria 'furniture store or factory', mueblista 'furniture dealer', mueblero (adj.) 'pertaining to furniture', and while there is no objection to the use of, say, álguien 'somebody' and nádie 'nobody' (though a mild disinclination may linger on, since alguno and naide, naire to this day are the preferred dialect vars.), the older language would have unwaveringly favored mobl- in suffixal derivations (note one vestige: moblaje, beside more "advanced" mueblaje 'suite of furniture', not to mention the verb moblar 'to furnish'). If this is so, then *yemelgo, *yenero, *yermano, etc., must, at a certain point (perhaps ca. 800 A.D.), have become utterly distasteful to the speech community — all agog with its new diphthongs firmly riveted to words stress - , which in fact rid itself of the unwelcome diphthongs or diphthongoids in the pretonic syllable by unceremoniously sloughing off the /j/ segment. This sacrifice entailed all manner of minor consequences. The vowel newly ex posed to the highly vulnerable word-initial position was easily apt to be altogether eliminated (mellizo) or, conversely, to be bolstered up through substitution of a segment well represented as a prefix: dial, a-, em-belga 'ridge between plowed furrows' < * GEMELLICA ("Lex Ascoli"), also Ast. almelga through wavering between a֊ and ál- unleashed by the influx of Arabisms. It was also likely to be tinted by its environment, e.g., raised by a following diphthong: (h)iniesta (like sintieron 'they felt', from sentir) or by the vicinity of a word in a set phrase: (h)inojo, proba bly on the strength of the stereotyped sequence hincar los (h)inojos 'to bend one's knees' — also in an effort to differentiate the word for 'knee' from the descendants of FOENUCULU 'fennel', before the remedial generalization of rodilla 'knee-cap', lit. 'small wheel'? — The participation in this trend of a very few Portuguese words (irmão 'brother', iguaria 'tidbit' from Late Lat. [gloss] IECUÄRIA geindre subtly differentiated from gémir on the semantic scale, [OFr.] jamme, genou, genêt, gent, germain, jeudi, joindre); in Provençal (genh ~ ginh (d)z-, cf. Friul. dzelá, dzemí, dzímul, dzenoli, zint and Gen. zimi. Sard, bwas arrived at from "hard" G- frozen at its original stage (voiced velar stop): belare, binistra, benuyu. The clinching argument in favor of the hypothesis here formulated is the develop ment of IACĒRE 'to lie (flat)' in Old Leonese. While the reflexes of this verb in Old Portuguese (jazer) and in Old Spanish (yazer) are fully as expected, MS  of the Alexandre displays in as many as seven stanzas (14, 671, 784, 1004, 1314, 1703, 2413, on the authority of Corominas) the unique var. azer. Now it so happens that precisely in Asturo-Leonese the development Ě > ia, as an alternative to Ë > ie (that is, PEDE 'foot' > pia rather than pie), was conspicuously strong; it has left not a few scattered traces even in modern dialect speech. In this zone, then, pretonic monophthongization of ia (through shedding of /j/), by way of escape from an ill-placed diphthong, makes excellent sense indeed.40 40) I can discuss here neither the history of names, such as Yagu(o) < lACOBU,nor the record of unusual tell-tale words, such as essares ~ esseras 'gypsum fields' (both derivatives recorded in J. Roudil's Glossary [p. 644] appended to his recent edition of the city ordinances ["Fueros"] of Alcaraz and Alarcón), still less certain controversial etymologies, e.g., the origin of OSp. (y)engo 'free' - a formation perhaps best explained not as a direct heir to INGEN(U)US, but as an adjectival postverbal (or a truncated p. ptc.) extracted from engar Հ. * (IN)GEN-ICĀRE or -UÄRE, based on the nucleus of (IN)GEN(U)U. The idea here sketched out on the loss of [j] before e as a recoil from an unwanted diphthong is by no means new: It was briefly formulated by W. Meyer-Lübke, "Zur Geschichte von lat. Ge, Gi und J im Romanischen", VR 1.1-31 (1936), at 29, then taken up by V. García de Diego in his Gramàtica histórica españoh 67 (Madrid, 1951), but somewhat cavalierly rejected (p. 337) by E. Alarcos Llorach in his otherwise stimulating and well-written note, "Resultados de Ge,i en la Peninsula", Arch. 4.330-42 (1954), a paper in which dialectological and structural solutions

ETIOLOGICAL STUDIES

389

. Application to Composition and Suffixal Derivation in Spanish. It has been known for quite some time, through the researches of Menéndez Pidal and a few other pioneers, that in characteristic groups of compounds — especially those involving a short numeral as the first member — even originally countertonic ie would, with the passage of time, tend to become i. The sprouting of rival variants is in such instances the rule rather than the exception. Thus, to start out with toponyms, Sietmancas (based on ancestral SEPTIMANCA) has yielded Simancas; OSp. Sietcuendes (near Uclés)has been compressed to Skuen des, and Cienfuentes to Cifuentes, etc.; the concurrent simplification of unusual or endangered medial consonant clusters (-tm-, -tk-, -nf) seems not to have too strongly interfered with this process. In another set, even more close-knit, the medi eval compound numerals dizesiete 'seventeen' and dizeocho 'eighteen' (in which, through continuous intrusion of diez 'ten', i.e., through recomposition, the original diphthong was eventually restored in the standard), are to this day echoed on the dialect level by dizedó 'twelve', diz-e-trés 'thirteen', and even dismil 'ten thousand'. It is perhaps undemonstrable, but surely not implausible, to argue that the striking reduction of IÜLIÄNU to Illan(o), Ullan(o) may have started in the hagiographic forms Santillán and Santullan(o), to which one may add perillán 'sly fellow' < Pero Illán < PETRU IÜLIÄNU. Later development led to such groups as Il, Ullana, Illánez, etc. (tying in with yunque 'anvil' beside dial, unquera, mostly in toponyms), the alternative being consonantization, as seen in Ibáñez in preference to *Juáňez. In Santillán, perillán, etc. it is the second member of the original compound that has undergone "monophthongization". As for the wavering between i- and u-, one is reminded of Sp. -enta vs. Aljam. -anta (as in French and Italian cognates), from older -aenta < -ĀGINTĀ, in equivalents of 'forty', etc. The situation is probably more complicated on the side of suffixal derivation. M. A. Zeitlin, to whom we owe a superbly documented note — seldom consulted — on the "Reduction of ascending diphthongs in Spanish" (MLF 24. 84-90 [1939]), has drawn attention to such by-forms of familiar adjectival abstracts in Old and Classical Spanish as contraridad, propridad, and rrezidumbre, from contrario, proprio, and rrezio, respectively ; the dominant forms all end, as one would expect, in -iedad. But necio 'silly' < NESCIU has generated necedad, perhaps in tribute to its semantic proximity to OSp. torpedad 'clumsiness' (initially favored over torpeza) and the like. Generally speaking, Spanish words ending in -io, -ia and -uo, -ua — in most in stances, learned formations, unless -io reflects -IDU and -ia reflects -IDA — have two are occasionally contrasted as alternatives. The strength of Alarcos Llorach's note lies in the systematic comparison of Galician, Portuguese, Castilian, and Catalan (with some attention paid to Leonese and Mozarabic as well), and in the equally consistent confrontation of specific results in word-initial and word-medial position. He correctly recognizes the strategic rôle of  (I) so < GYPSU but diagnoses the case differently, appealing for help to Moz. algez as a possible cause of deflection from the norm. Our opinions coincide on the loss of/- in IUNGERE >uncir/uñmr, less so on the development of -ng- >-nz- > m o d . -nc-, and least of all on the trajectory of * RADIA > raça. For a later and more succinct presentation of Alarcos Llorach's views (basically unaltered) see his Fonologia española3 251-4 (Madrid, 1961).

390


root allomorphs available for derivation, one with the z' /j/ or u /w/ included at the end, and the other (not infrequently racier) without that tail segment, so that one can speak of two bound forms, real or at least latent: contrar- and contrari-, etc. This situation goes back to Antiquity: OSp. vazio 'empty' presupposes VAC- rather than VACU-U. Also, we must reckon with the coexistence of antigo 'old' Sp. mitad stands somewhat apart. For despite Menéndez Pidal's splendid collection (1926) of dated and localized examples of early usage (Origenes del espanol, § 48), it is not entirely clear whether archaic meytad or meatad (var. myatad) directly underlies mitad, or else whether one should posit conflation of these variants. All manner of concomitant processes may of course hinder or reinforce the tendency here sketched out. In Col. (Bogota) palitaria in lieu of parietarm, lit. 'relating to the wall', as observed by R. J. Cuervo, the dissimilatory trend, which is behind the change of the first r to I, may also be held responsible for the avoidance °f /j/ • • •/j/,> this time not through substitution but through total erosion of the first semiconsonant. The abovementioned var. meatad calls to mind a number of cases observable in the medieval texts, where pretonic ie has yielded ground to -ia- rather than under going monophthongization. Two explanations come to mind. It can be argued that those speakers who refused to settle on a monophthong translated their opposi tion to ie into the choice of a more resistant diphthong, one whose ingredients were more widely apart, more neatly polarized in articulatory, acoustic, and auditory terms, than are i and e, hence more likely to withstand the ever-present drift toward compression. A possibly more sophisticated argument would run thus: While pre tonic ie is against the grain of most speakers of the central dialects, being used morphophonemically in opposition to e (pie ~ peal, quiero ~ queremos), ia, as a rule, is not so involved and consequently remains more "neutral", hence apt to become more acceptable. As a result, friedad '[studied] coldness, coolness' (from frío) was doubtless pronounced in certain places friadad (with some assistance from the adv. friamente), just as feedad 'ugliness' (from feo) — which jelled alongside feeza - was here and there indisputably pronounced feadad. At this point the

ETIOLOGICAL STUDIES

391

influence of the numerous abstracts in -aidad, -aitad {maldad 'wickedness', lealtad 'loyalty') asserted itself, leading to such familiar, but still startling and startlingly successful, forms asfrialdad mâfealdad.41

IV. Proliferation of Variants as a Co-Determinant of Sound Change? The Case of Old Spanish ç < DJ. The study of variants has been the traditional domain of philology (including textual criticism), or, at most, of philological linguistics. A classic in this field, as conventionally defined, is, for instance, the Sanskritologist Maurice Bloomfield's posthumous magnum opus, salvaged by Franklin Edgerton in the early 'thirties: Vedic Variants; a Study of the Variant Readings in the Repeated Mantras of the Veda. And there is no need to elaborate here on the genre of variorum editions. It seems possible, however, to make the corpus of variants, properly distilled, ancillary to the needs of comparative-historical linguistics in certain ways not fore seen by the pioneers. A few years ago, I tried to show, in a paper titled "Range of variation as a clue to dating" (RPh 21.463-501 [1967-68]), that any striking clustering of inflectional variants within a morphological edifice otherwise marked by leanness or economy may have symptomatic significance and that, under certain circumstances, it is legitimate to draw crucial chronological inferences from such a contradiction between the whole and one of its sections. Today's analysis represents an experiment pointing in another direction. Suppose you discover a puzzling sound change - a shift that conflicts, first, with internal diachronic evidence supplied by successive stages of the language in question; and, second, with comparative evidence carved from the record of cognate languages. Suppose further that the seemingly erratic change can be illustrated with just a few unassailable examples - despite decades of assiduous search, by competent scholars, for additional documentation. Finally, by way of third independent supposition, assume that every single one of these isolated examples is flanked by a variant which falls into place rather smoothly as regards the relevant broader evolutionary trends. Does it not stand to reason that some tentative conclusions should be drawn from such a — rather sharply delineated — state of affairs? 41) I must curb the temptation to provide adequate philological underpinning - beyond the barest minimum - for the forms here cited. Torpedat occurs in Elena y María, v. 200; in Barlån e Josaphá, ed. Moldenhauer, fol. 98 v ; and in El Emperador Ottas, ed. Amador de los Ríos, Ch. 38. Suzidad prevails in Barlán e Josaphá, fols. 109r°; 115r°, 118v°, 127r°, 168v° 194v; the more normally convoluted suziedad is found in the same text, 109r (var.), 126v , 194v (var.) and in Conflsión del amante, ed. Knust, fol. 292v . Suziadad is peculiar to the Old Aragonese Vidal Mayor. In the case of propri(e)dad further proliferation of variants arose through the tendential dissimilatory loss of the second r. Necedat predominated already in medieval texts, even though ne(s)ciedat has left a few isolated traces (cf. A. Castro's 1936 edition of Glosarios latino-españoles de la Edad Media). To some extent, the diphthong in the pretonic syllable of limpiedumbre 'cleanliness' was protected by its coöccurrence in the stressed syllable of the far more common synonym limpieza. So also turbiedad beside turbieza (usually spelled torv-)՛, but reziedumbre was flanked by reziura. Numerous relevant data are documented in my paper, "The derivation of Hispanic fealdad(e), fleldad(e), and frialdad(e)" UCPL 1:5.189-214 (1945).

392


If we favor a complex sociolinguístic model for the description of the habits of the chosen speech community, we might, for instance, find it tempting to argue that the set of easily explained variants represents one tradition or transmission (say, allegro forms, or forms used in relaxed sub-standard style, or else forms preferred by rustics), while the set of their difficult-to-explain counterparts could represent the opposite pole (e.g., lento forms, or forms peculiar to a highly formal, standard style, or else forms adopted by an urban population, or by a group otherwise socioeconomically privileged, perhaps even culturally pretentious). The reference to some measure of self-consciousness, if not downright pretentiousness, is the cornerstone of the entire hypothesis, the main idea being that the erratic character of the evolu tionary side-line is due precisely to that group-of-speakers' reluctance to be mistaken for a gang of uncouth rustics and, consequently, to its readiness or willingness to exaggerate (and thus channel into a new groove) the feature which sets off its speech from that of less highly regarded persons. If our thinking so far has been correct, then the extent of this exaggeration should correspond to the degree of aberrancy or irregularity, i.e., of deviation from the expected course of events observed in the first place. Let us inspect a typical concrete situation. Old Spanish contained in its phono logical inventory two phonemes, ç and z, i.e., a voiceless and a voiced dental affrica te, on the whole rather carefully distinguished in the best available MSS ֊ those generally credited to scribes of local family background and upbringing. 42 Though there happened to be on record few if any tidy minimal pairs to support ç's and z's parallel claims to phonemic status, there exists overwhelming circumstantial evidence in favor of that contention. On balance, Hispano-Latin and Hispano-Arabic words alike fall into neat patterns as regards the distribution of sources of either phoneme, except for a few rather neatly circumscribed troublesome groups. One such group involves a small number of words displaying, quite consistently, a ç as a reflex of the ancestral intervocalic cluster -DJ-, as in BADIU 'bay-colored, brownish' and in RADIU 'staff, spoke, shuttle', fig. 'ray, beam'. Here is the small slice of material at issue : BADIU yielded baço 'yellowish-brown' which, if nominalized, acquired the ana tomical meaning 'spleen' on account of the peculiar color of that long-misunder stood organ; *INTERPEDIÄRE 'to trip'(from PĒS PEDIS 'foot'), a secure recon struction, underlies medieval entrepeçar 'to stumble', which later produced tr-, estro-, Ծtro-, even trom-peçar and from there trompicar, etc.;*FODlĀRE, a somewhat aberrant iterative of FODERE 'to dig out', fig. 'to prick, prod, jog', is a convenient starting point for hoçar 'to root, nuzzle', cf. hocico 'snout'; finally, *RADIA, a cogently posited feminine counterpart of RADIU 'rod', constitutes a plausible base for raça 'crack, slit; ray of light coming through a crack; cleft in horse's hoof; stripe in a fabric; line of descent', witness borrowed race in French and race in English. In all four instances the spelling with ç, indicative of the affricate's voicelessness, is secure and recurrent. A fifth case, representing the important family of MEDIU

42) For a preliminary reëxamination, from a new angle, of a representative specimen of the entire material so far garnered see my recent article, "Derivational transparency as an occasional co-determinant of sound change", RPh 25.1-52 (1971-72).

ETIOLOGICAL STUDIES

393

'half, is less conclusive as regards the detail of voice. One finds traces of an ancient toponym Mezanedo with a z, from *MEDIĀNĒTU (as it were), but the peculiar spatio-temporal circumstances surrounding its earliest appearance are such as to make one doubtful whether the local scribe's use of z implied any message concerning the voiced or voiceless character of the given consonant. The raw facts have been known for a long period of time, but it took a great and experienced scholar of the caliber of Ramón Menéndez Pidal to become at a certain point restless about the paradoxical situation43 long taken for granted. Why indeed, he wondered, did the speakers of Old Spanish reject voiced z, readily available to them and so conspicuously germane to ancestral DJ, a cluster presided over by a voiced dental, in favor of such a glaringly poor match as çl The question was the more relevant as Italian, which to this day distinguishes between one set of voiced and another set of voiceless affricates, has, reassuringly enough, used either its [ğ] or its [ d z], according to the dialectal transmission of each lexical item, as a reflex of parental intervocalic [dj], thus: mezzo 'half, from MEDIU; raggio 'beam', from RADIU. Although speakers of Italian have at their immediate command [Č] and [ts] as well, they have repudiated such alternative solutions as *mezzo and *raccio. How, then, should we account for the idiosyncrasy of Old Spanish? The key to the problem lies, I believe, in the fact that all the forms so far ad duced: baço, hoçar, raça, etc., represent mere variants. Either there has existed in close vicinity to them a straight doublet or, under less favorable circumstances, there has been preserved at least a close cognate showing a rival development. That rival evolutionary line involved, by way of first step, the change of D J to /j/, a semiconsonant which, depending on the dialect, could at later stages either be preserved unaltered, or be thickened to a full consonant (ž), or else be allowed to disappear. This far more familiar pattern of change is discernible in Sp. bay 'bay-colored', again from ADIU; in Ptg. peia 'snare, iron-shackle' and OCat. pija 'trap', both from *PEDIA, and in Sp. pi(h)uela 'jess' (a term of falconry), from *PEDIOLA, either base structurally reminiscent of *INTERMEDIARE; in Sp. hoya 'hole, pitch', hoyo 'pockmark', hoyuelo 'dimple', etc., all of which seem to call to mind *FODIÄRE; in Sp. rayo 'beam, spoke' alongside raya 'stripe, stroke, dash, mark', again from RADIU and from its assumed feminine counterpart. As regards the progeny of MEDIU 'half, we must invoke not the familiar outcome medio, originally learnèd, but the vernacular offshoots at present relegated to the dialects, meio and meo (e.g., meodia 'noon'), cf. Ptg. meio alongside studiedly erudite medio. Apart from these doublets, the reduction of DJ to / j / is found independently and uniquely in numerous words, e.g., fastio (mod. hastio) 'loathing, disgust' < FASTIDIU; moyo 'corn measure' < MODIU; porfia 'stubbornness' < PERFIDIA 'treachery'; poyo 'stone bench built against the wall' < PODIU 'balcony'; presea 'gem, jewel' < PRAESIDIA (pl.) 'supports, guards, posts'; also in toponymy, e.g. (San) Cloy  (Oviedo) < SANCTU C L A U D I U , and in certain widely-used verb froms, e.g.,Sp. sea, Ptg. seja T may be' (lit. T may sit') < SEDEAM (pronounced /sedja/) and Sp. veo, Ptg. vejo T see' < VIDEŌ (pronounced /vidjo/). There cannot be the 43) This account is based on the pertinent sections of the Manual de gramática histórica española (e.g., 6th éd., 1941) and of Orígenes del español.

394


slightest doubt, then, that this development, involving no such by-forms as display affricates, represents the norm, the preference of the overwhelming majority of speakers. Against this background, the rise of baço, hoçar, raça, etc. becomes psychologi cally better understandable. In a Hispano-Roman society — conceivably of the 3d, 4th, or 5th century ֊ in which part of the speakers tended to pronounce /dj/ laxly as / j / : baju, meju, raju, etc., another sector of the society — perhaps more conservative; at any rate, one gathers, severely dissatisfied with this pattern of laxity - may have tried to buttress the endangered dental by saying *batju, *metju, *ratju, etc., since tj, in contrast to dj, was exempt, in any layer of the speech com munity, from the hazard of reduction to a semi-consonant. The process envisioned recalls hypercorrection or hyperurbanism. Here as elsewhere, the intervocalic tj cluster eventually produced ç. as long as it was allowed to evolve without interference, cf. OSP. plaça 'square' from PLATEA, Gr. PLATE ÎA '(broad) street, square'. The processes here described are not entirely isolated. In their closest neighbor hood one might place the Hispano-Romance word-medial development — equally baffling at first glance — of Latin and Gothic dj, practically in every single instance either after ո or after r. The examples are few and far between. For classical Latin ndj yielding medieval Romance nç I can cite one probable etymology: grança 'mad der', pi. 'chaff, screenings, siftings, dross', apparently from GRANDIA, the neuter pl. of GRANDIS 'full-grown, large'; plus one absolutely safe etymology: OSp. vergüença, OPtg. vergonça 'shame' from VERËCUNDIA 'modesty, diffidence, bashfulness'. Add to this record the three Hispano-Gothic women's names Alduença (OLeon. OPtg. Aldonça) < ALDEGUNDIA, Ennegüença < HlNNEGUNDIA, and Tedgüença < THIUDEGUNDIA. At least in one of these instances (fortunately, the most transparent case of all) a variant displaying no sibilant whatever has been preserved: VERËCUNDIA is also reflected by OSp. vergüena, Ptg. vergonha. (In fact, through a certain historical caprice, vergüençahas survived into modern Spanish, where it is, of course, currently pronounced as -θa or -sa, depending on the region, while vergonha has lingered on in Portuguese.) The traditional explanation is that VERËCUNDIA was, at a certain point, deflected from its normal course by the many - semantically germane — abstracts in -ANTIA and -ENTIA (INFANTIA 'slowness of speech, childhood', INNOCENTIA 'harmlessness', etc.), giving way to *VERËCUNTIA. Such a narrowly defined influence may indeed have been operative, but it would account only for one out of five examples. Conversely, if we argue that the main reason for one faction-of-speakers' switch to *VERËCUNTIA was their reluctance to be caught up by the current that was leading to *VERËCUNNIA, we can immediately grasp the genesis of both variants. Once the sequence -nça had become habitual — in grança, in vergùença, and in the reflexes of the -ANTIA/ -ENTIA abstracts (e.g., simiença 'seed' from SEMENTIA), it could serve as a mould for the necessarily late Gothic proper names. Finally, ç in lieu of expected z appears word-medially after r in just two words that come readily to mind: orçuelo 'sty', 'snare, trap', for HORD-EOLU, -IOLU 'sty'j lit. 'small grain of barley', cf., on the semantic side, Gr. KQIΘBIΔOV, R.jačmen', Fr. orgelet, G. Gerstenkorn; and verça (mod. bena) 'cabbage', traced with unanimity to Lat. VIR(I)DIA (pi.), syncopated to *VIRDIA, lit. 'green stuff, cf. R. zelen'

ETIOLOGICAL STUDIES

395

'vegetables', G. griines Gemme, Sp. legumbres verdes. I suspect that the spread of the sibilant treatment from ndj to rdj is purely analogical and paradigmatic; i.e., because, in the given languages, ո and r clusters showed many mutual resemblances and affinities, one characteristic shift truly understandable only in the context of ndj was echoed, on account of those parallels, also by rdj. Though the stock of historical facts here tapped has its unavoidable share of gaps, these lacunae and uncertainties are fortunately few. As happens so often in Romance linguistics, the analyst at no time loses the pleasant feeling of treading solid ground.

. AFHXAL DERIVATION

ONE CHARACTERISTIC DERIVATIONAL SUFFIX OF LITERARY ITALIAN: -(t)aggine

1. The problem. The suffix selected here for the briefest defensible treatment can, without exaggeration, be called characteristic of, not to say highly idiosyn cratic to, standard Italian. This circumstance is all the more remarkable in that Italian otherwise shares virtually all of its abstract suffixes - e.g., -anza/-enza, -ena, -ezza, -mento, -orej-ura - with its sister languages, as it also shares the twin models of post-verbal action nouns. The ancestry of -aggine has always been known to scholars: The suffix descends in an almost straight line from Lat. -ÄGÕ/-ÄGINE. One important reason for -AGINE's distinctly better survival on Italian soil (specifically in Tuscany) than in the remaining Romance territories has, obviously, been the far stronger resistance to erosion of Late Lat. /g/ in a Tuscan word accented on the antepenultimate than in, for instance, that word's French and Spanish counterparts. The same force that accounts for the audible and visible lengthening of the key consonant, immediately after the stress, in abbaco, attimo, collera, femmina, and legittimo also justifies the transition of Lat. -ÄGINE to It. -aggine and helps us to understand the dramatic contrast between this outgrowth and Fr. -ain, Sp. -én, and the like 1 . However, by grasping this differentiating factor we have cleared only the first hurdle. Thus, it remains to be explained why -aggine developed with such remarkable vigor, while two fellow members of the same set of suffixes, apt to benefit from the same favorable circumstances (namely -iggine and -uggine, from -ÏGINE and -ŪGINE, respectively), barely succeeded in holding their ground, faring locally just a shade better than in the cognate languages - where a fatally weakened -ÜGINE, incidentally, gravitated toward a merger with the mass-noun suffix -ÜMINE. Apart from striking root with impressive strength, It. -aggine underwent a radical functional change, developing more and more from a nominal into a post-verbal (specifically, a post-participial) suffix. Its increasing proximity to the past participles in -ato of the A- conjugation has eventually led to the consolidation of the -(t)aggine or, better still, the -(at)aggine sequence. The entirely unequal record of -aggine, on the one hand, and, on the other, of the modest aggregate of -iggine and -uggine prompts one to wonder whether a joint consideration, both syntagmatic and paradigmatic, of -(at)aggine and of certain other A -dominated derivational suffixes in Italian may not supply the sought-after answer to our initial question. 1) I have on several occasions concerned myself with the Hispano-Romance vicissitudes of -ĀGŌ, -ĪGŌ, -ŪGŌ, starting with the short note published in "Language", XIX (1943), 256-258. More recent discussions include: Some Late-Twentieth-Century Options..., "Bulletin of Hispanic Studies", LII (1975), 1-11, esp. 4-5; In Search of Penultimate Causes of Language Change.... (in press); and Infinitive Endings, Conjugation Classes, Nominal Derivational Suffixes, and Vocalic Gamuts in Romance (in press).

400

AFFIXAL DERIVATION

2. Morphological analysis. It seems safest to open our attack on -aggine with a morphological break-down of the selectively assembled corpus. In this initial dissec tion, such criteria as the pattern of minor regional preferences, relative chronology of the date of appearance and continuity of use, semantic and stylistic contour, socio-educational level, channel of transmission, etc. will be provisionally ignored or, at least, deëmphasized 2 . In focusing attention on the relation of -aggine derivatives to the underlying primitives, one can set off a total of seven major blocks, some of which lend them selves to further subdivision. The first block comprises ADJECTIVAL primitives. Not infrequently, the adject ive at issue is primary or, at least, gives the impression of being so: astrusaggine 'gratuitously abstruse discourse or argument' (astruso), balordaggine 'dullness (of mind), stupidity' (balordo), bolsaggine 'broken-windedness' (esp. of horses), 'weakness' (bolso), caparbiaggine 'obstinacy' (caparbio), cecaggine 'blind ness, heaviness (of the eyes), drowsiness' (cieco), ebetaggine 'dullness, stupidity, foolish action' (ebete), goffaggine 'awkwardness, clumsiness, blunder' (goffo), grossaggine 'grossness, coarseness, ignorance, bluntness' (grosso), grullaggine 'foolish ness' (Tusc. grullo), lordaggine 'filth, filthy act' (lordo), lungaggine 'slow, tedious procedure; dallying, dawdling; long, drawn-out business' (lungo), melensaggine 'stupidity, silliness' (melenso), ridicolaggine 'absurdity' (ridicolo), rocaggine 'hoarse ness' (roco), seccaggine 'dryness, weariness, trouble, annoyance' (secco), sordaggine 'deafness' (sordo), tetraggine 'gloom, sadness' (tetro), votaggine 'emptiness, vacuity' (vuoto), zoppaggine 'lameness' (zoppo). In a closely-related subclass, the adjective, measured by a strictly descriptive yardstick, may still be unanalyzable into any neat sequence of constituents, but an untutored speaker will hazily recognize affixlike sequences which, in other contexts equally familiar to him, do function as genuine prefixes (e. g., in-) or suffixes (e. g., -ico, -ido):

2) I have used as a starting point of my collection GIOVANNI MONGELLI'S Rimario letterario della lingua italiana (Milano, 1952), checking it in dubious cases against standard dictio naries. Analytical information on -aggine and its peers is available in many historical grammars; long ago, B. WIESE, Altitalienisches Elementarbuch (Heidelberg, 1904), § 179, labeled -aggine as "semilearned", on the grounds that -i- had been preserved, rather than undergoing the expect ed change to a. The treatment of -aggine, -iggine, -uggine by G. ROHLFS, HGIS, III: Syntax und Wortbildung (Bern, 1954), § § 1058-59, is noteworthy on account of the profusion of dialect data, some of which involve evolutionary sidelines deliberately disregarded in the present paper (e. g., -ana and -ajna, with counterparts in Spanish, e. g., andana 'orbiť beside andén 'platform'). Rohlfs cites striking designations of sicknesses and physical defects in -aggine, àina, and -ania, from Tuscan, Calabrese, and Sicilian; but he is unable to justify the exuberant growth of -aggine - far in excess of its partners -iggine and uuggine; and his semantic analysis is faulty. I owe most of my glosses - in vernacular, indeed racy, British Fnglish - to the excellent bilingual dictionary of Barbara Reynolds. Though perfectly aware of the existence of the three fully learned series in -agine, -igine, and -ugine, I am forced to omit from my purview these, all told, unexciting groups of Latinisms.

ONE CHARACTERISTIC DERIVATIONAL SUFFIX

401

imbecillaggine 'foolishness, folly, stupidity' (imbecille), infingardaggine 'laziness, slackness, sloth, cowardice' (infingardo), insulsaggine 'silliness, boorish remark, boring conversation' (insulso), rusticaggine 'rudeness, roughness, simplicity in manners' (rustico), stupidaggine 'nonsense, act of stupidity' (stupido), zoticaggine 'boorishness, uncouth behavior' (zotico). In other instances, the underlying adjective may itself constitute a clear-cut derivative. Far and away the most characteristic contingent of such items is formed by those abstracts in -aggine which rest on -oso adjectives: accidiosaggine 'slothfulness', ambiziosaggine 'pretentiousness, foolish ambition', fastidiosaggine 'troublesomeness, vexation, annoyance', leziosaggine 'affected habit, mawkishness', meticolosaggine and minuziosaggine 'fastidiousness', oziosaggine 'sloth(fulness)', piccosaggine 'peevishness, touchiness, irritability', presuntuosaggine 'self-conceit', ritrosaggine 'shyness, aversion, reluctance', schifosaggine 'loathsome ness, nastiness', scontrosaggine 'peevishness, moroseness, coyness', sdegnosaggine 'scorn, contempt, haughtiness', spirito saggine 'forced wit. witticism'. Clearly, Italian has an excellent mechanism here at its disposal, to which the other Western languages offer no genuine counterpart in terms of uniformity, simplicity, and predictability. English usage, for instance, is split between rigid meticulosity, pomposity and relaxed meticulousness, pompousness՛, French suffers from complete gaps on the side of corresponding abstracts, in such cases as dédaigneux 'disdainful' and oublieux 'forgetful'; Old Spanish evinced a trend toward vernacular -osi beside learned -osidad, but the development of -osia failed to gain momentum 3 . Other suffixes can also be peeled off, though far less frequently, from such ad jectives as served as starting points for the rise of abstracts in -aggine, e. g., svenevolaggine 'affectation, sentimentality, simpering, languishing behavior' (from svenevole 'languishing', which can be separated into svenire 'to faint, swoon' and -evole), or testardaggine 'obstinacy, stubbornness, headstrong behavior' (from test-ardo, behind which one recognizes testa; cf. the French equivalent têtu). Compounds are rather exceptional; Francesco Berni, in a facetious vein, used the nonce magnificaggine in lieu of the transparent near-Latinism magnificenza. Certain phrases that function as qualifiers and are thus comparable to adjectives have in the past occasionally cast off substantives in -aggine; two fairly isolated examples are: dabbenaggine 'simple-mindedness, naivete, good-heartedness', from indeclinable dabbene 'decent, upright, honest' (da plus bene); and dappocaggine 'worthlessness, inefficiency, ineptitude, stupid action', from comparably fossilized dappoco 'worthless, useless' (da plus poco). The second major block in our analysis comprises the subcategory — typical of conditions in Latin and Romance - of primitives doing service as BOTH A D J E C T I V E S AND SUBSTANTIVES (for our limited purposes it is irrelevant which of these two functions, in every individual case, preceded the other):

3) This issue is examined in depth in my forthcoming article (to appear in "Romance Philology"): Medieval Roots of the Spanish Derivational Model: sabid-or ~" sabid-uria.

402

AFFIXAL DERIVATION

citrullaggine 'foolishness' (citrullo: [a] 'stupid', [b] 'silly-billy, fool'), gaglioffaggine 'loutishness, laziness' (gaglioffo: [a] 'lazy, brutish, clumsy'), [b] 'lout, scoundrel'), minchionaggine 'foolishness, silliness' (minchione [a] 'silly, stupid', [b] 'fool, simpleton, ninny; dunce'), nullaggine 'emptiness, vanity, nothingness' (nullo [adj. > adv.] 'not at all, by no means'vs. nulla 'nothing'),poltronaggine 'laziness' {poltrone[a] 'lazy, cowardly', [b] 'lazy fellow, slacker, poltroon'), trullaggine 'silli ness, stupidity' {trulio [a] 'silly', [b] 'simpleton, nincompoop'). From here a smooth road leads once more to a big block, the third in our array: strictly (or predominantly) SUBSTANTIVAL bases. A surprisingly large number of them have ո as their last consonant — either root-finally, or as a result of the occurrence of some such suffix, or suffix oid, as -one, or hypocoristic -ino:. (a) birbonaggine 'roguery' (birbone 'scamp'), bricconaggine 'rascality, knavery' {briccone 'scallywag'), buffonaggine 'tomfoolery, buffoonery' {buffone 'jester'), caponaggine 'obstinacy, obduracy' (capone 'blockhead, thickhead', augm. of capo), castronaggine, 'stupidity' (castrone 'castrated colt or wether, blockhead'), mellonaggine 'silliness' (mel-, mell-one 'melon'), zucconaggine 'pigheadedness' {zuccone 'stupid individual', lit. 'big pumpkin', cf. Sp. sandio from sandia 'watermelon'); see above on minchionaggine and poltronaggine; (b) asinaggine 'stupidity, silliness' {asino 'donkey, fool'), bambinaggine 'childish action, childishness' {bambino, -a 'baby'), cervellinaggine 'stupid action' {cervellino 'silly individual, weak-minded person'). Another sharply silhouetted group of subjacent primitives is the ֊ relatively few - nominalized present participles in -ante (practically never any in -ente), plus partial imitations of this model: birbantaggine 'roguery' (birbante), furfantaggine gine 'illiteracy, boorishness' (ignorante).

'rascality' (furfante),

ignorantag-

The bulk of the substantives attracted by -aggine, many of them names of the bigger domestic animals, seem to lack any salient formal features, such as stress, number of syllables, etc., except that in most instances the vowel that presides over the syllable immediately preceding -aggine is either  or a. This phonostylistic condition prevails in: bietolaggine 'foolishness, silliness' (bietola 'beet [root]' beside [slang] bietolone 'fool, booby, tearful individual'), buaggine 'slow-wittedness, doltishness' {bue 'ox'), ciucaggine 'stupidity, stubbornness' ([pop.] ciuco 'donkey', 'dunce, idiot, badmannered person'), (joc.) dottoraggine 'donnishness, professorial pose' {dottore 'doctor'), fanciullaggine 'childishness, puerility' (fanciullo, -a 'child'), golaggine 'gluttony' {gola 'gullet, gorge, throat'), gufaggine 'unsociableness, habits of a recluse' {gufo 'owl', [fig.] 'unsociable person'), idiotaggine 'ignorance, stupidity, stupid action' {idiota 'stupid person'), mul(l)aggine 'obstinacy', lit. 'mulishness' {mulo 'mule'), pecoraggine 'servility, stupidity, sheepishness' (pecora 'sheep, ewe, mutton'), vagabondaggine 'state of vagabondage' {vagabondo 'vagrant').


403

In a small residue of words, complete randomness seems to predominate: cervellaggine 'odd caprice, freakish whim' {cervello 'brain, brains'), grammaticaggine 'pedantic detail' {grammatica 'grammar' alongside -o 'grammarian'), (obs.) ribaldaggine 'knavish, dirty trick' {ribaldo 'scoundrel, rascal'; semantically differen tiated from E. ribald), somaraggine 'asininity' (somaro 'donkey'). The fourth block is, again, transitional and, once more, modest in size; to form it, we shall segregate, from the bulk of our collection, such items ending in -aggine as are flanked by A NOUN AND A VERB of comparable prominence ("bicuspid families"), regardless of the hierarchical relation between noun and verb in an etymological perspective. Also, for the purposes of this operation we shall refrain from distinguishing between substantives and adjectives, and shall not ask ourselves whether the former are, at root, post-verbal abstracts and the latter merely constitute truncated past participles: abbagliaggine 'glare, dazzle' {abbaglio 'dazzle, glare', fig. 'error' ~ abbagliare 'to dazzle, blind, fascinate, deceive'); corbellaggine 'stupid action' {corbello 'dolt, blockhead' ~ corbellare 'to make fun of, ridicule'); cornaggine 'obstinacy, pigheadedness' {corno 'horn' ~ comare 'to butt with horns'); gonfiaggine 'swollen-headedness, swaggery' {gonfio [a] 'swelling, lump, bulge', [b] 'swollen, filled out, bombastic' ~ gonfiare 'to inflate, blow up, fill with air'); sciattaggine 'slovenliness' {sciatto 'untidy, slovenly' ~ sciattare 'to crumple, spoil', dial, 'to weaken'); straccaggine 'weariness, slackness' {stracca 'fatigue' ~ stracco 'tired, worn out' ~ straccare 'to wear out'). This gambit has left us at the threshold of the realm of VERBS. At the start, we shall concentrate on -aggine's relation to fully-developed verbal paradigms, best identified through the respective infinitives (the fifth block in our edifice). To avoid superfluous complications, we shall abstain from asking whether the given verb, in the last analysis, boasts a substantival or an adjectival background: assordaggine 'deafening row' {assordare 'to deafen, stun'), cascaggine 'sleepiness, muscular fatigue, drooping of the limbs, nodding of the head' {cascare dal sonno 'to be dog-tired'), dimenticaggine 'habitual forgetfulness, absent-mindedness' {di֊ menticare 'to forget'). An incomparably bigger block — the sixth according to our count - awaits us as we approach such derivatives in -aggine as have been attached to PAST PARTICIP LES. Further subdivision is possible if we discriminate between the many participles in -ato, the very few in -ito and -uto plus an occasional representative of the radicalstressed category of participles bequeathed by Latin. We shall further set aside those cases in which the prefixation of s and the suffixation of -at-, etc., go hand in hand (a procedure traditionally labeled as parasynthesis) as well as others in which the critically important -at- segment now makes its appearance, now vanishes. Only incidental mention will be made of those border-line cases in which -ato, -ito, or

404

AFFIXAL DERIVATION

-uto merely represent pseudo-participles, i. ., are not accompanied by full-fledged verbs4. The suffix -aggine is attached to the participial segment -at- in a sizable number of cases: accorataggine 'grief' (accorato 'stabbed, pierced to or through one's heart'), avventataggine 'rashness, recklessness, imprudence', insensataggine 'senselessness, foolishness', sciagurataggine 'spiteful action, caddish trick' (sciagurato [a] 'ill-starred, wretched', [b] 'evil person' is a mock-participle, acting as the satellite of a noun [sciagura 'calamity'] rather than of any verb). Conspicuously frequent has been the coöccurrence of the negative (or, to be more accurate, adversative) prefix s- and the -at- segment in conjunction with -aggine. The items clearly form a close-knit whole: sbadataggine 'carelessness, heedlessness', sboccataggine 'coarseness of speech, foul language', scapataggine 'recklessness, light-headed behavior', (pseudo-participial) scape strataggine 'dissoluteness, wild behavior', sfacciataggine 'impudence, cheek', sfrontataggine 'shamelessness', sgangherataggine 'awkwardness, 'manners of an unhinged, disjointed individual', sgarbataggine 'rudeness', sgraziataggine 'clumsiness, ungainliness', sąuaiataggine 'coarseness, rudeness', 'awkwardness, slovenliness', spensierataggine 'thoughtlessness', svagataggine 'love of amusement, lack of con centration, absent-mindedness', sventataggine 'thoughtlessness, heedlessness', (pseudo-participial) sversataggine 'ungraciousness', svogliataggine 'listlessness, lazi ness, indifference'. In not a few of these intricately structured formations, the sensitive speaker and listener will still recognize the subjacent noun: bocca 'mouth', capestro 'halter', faccia 'face', fronte 'forehead', ganghero 'hook for an eye', garbo 'charm, courtesy', grazia 'grace, favor', pensiero 'thought', voglia 'will', or, less frequently, the under lying verb: badare 'to pay heed, be careful', vagare 'to ramble, rove' ֊ a circumstance which here, significantly, for once narrows or even bridges the gap between nouns and verbs. There is an astonishing paucity of examples of -it- plus -aggine, despite the ready availability, one would think, of numerous -ire verbs, and of the few abstracts in -itaggine actually caught in our dragnet practically all represent special cases. In sbalorditaggine 'bewilderment, amazement' and in storditaggine 'mistake, stupid action, folly' one observes the interplay of s- and suffix. Scimunitaggine 'silliness', based on -ito (a) 'silly', (b) 'blockhead', and scipitaggine 'fatuity, dullness', 'insipidi ty', involving scipito 'tasteless' (in lieu of expected * -ido), lack any underlying verbs. Even more astounding is the practical absence of the succession -ut- plus -aggine, even though paradigms of Italian verbs abound in -uto participles. Our surprise will grow once we remember that straight substantives with  or w as their stem vowel nearest the word boundary show a pronounced affinity for -aggine՛, 4) Cf. an early note from my pen, A Lexicographic Mirage, "Modern Language Notes", LVI (1941), 34-42, with Hispanic illustrations.


405

witness dottoraggine, etc. In fact, the sole example of -utaggine represented in our corpus, namely cocciutaggine 'pig-headedness, pride', involves nominal rather than verbal -uto, i. ., the suffix marking excessive endowment with a quality: cocciuto 'stubborn, stiff-necked', from coccio 'potsherd, cracked pot' (cf. colloquial E. crack pot). Finally, one can cite a few examples involving -aggine attached to a radicalstressed participle, provided it ends in -t: astrattaggine 'absent-mindedness, vague ness, inattention' and fintaggine 'deceitfulness, falseness, duplicity'. One observes in Italian widespread alternation between forms with and those without the -at- interfix. Typical examples include: sceller (at jaggine 'wickedness, villainy', scempi(at)aggine 'foolishness', scioper(at)aggine 'idleness due to a strike', smemor(atjaggine 'forgetfulness, bewilderment', trascurfat jaggine 'indifference, heedlessness, negligence'. There is no perfect balance between the rival variants (e. g., scempiaggine, geared to scempio 'simple, single, silly', outranks scempiataggine, while smemorataggine is, conversely, far ahead of smemoraggine). The reasons for the cleavage will vary from case to case; thus, scioperaggine is best linked to the subst. sciopero 'strike', while scioperataggine hinges on the corresponding verb, 'to (go on) strike', apparent ly best represented by its past participle rather than by the conventional alliance of infinitive and present tense. Haplological tendencies may also be involved; is it haplology or some lexical blend that most smoothly explains the rise of the vars. traf s jcutaggine beside the inf. trascutare — two satellites of -curataggine and -curarel On balance, the general impression given is that -ataggine has gradually been expanding its scope. The sum total of these deductions leaves a tiny residue of words in which -aggine performs NO NEATLY IDENTIFIABLE SUFFIXAL ROLE. Words containing such a hazily recognizable or totally non-segmentable sequence as -aggine, are just as difficult to describe on the semantic side; they will henceforth be known as constituents of the seventh block; (bot.) borraggine 'borage' (= Borrago officinalis), (bot.) capraggine 'goat's rue' (= Galega officinalis), (med.) lombaggine 'lumbago', mucil(l)ag(g)ine 'mucilage', (bot.) piantaggine 'plantain' (= Plantago major), (min.) piombaggine 'plumbage, graphite, black lead', propaggine 'slip, layer' (= 'shoot or twig of a plant'), 'issue, offspring'. These words, which seem to lack any common denominator except for the positive fact that they often pertain to the technical vocabulary of natural sciences and for the negative fact that they most certainly involve no humorously tinted abstracts, represent historically the hard core of the entire development, going back directly to the Latin stock (PLANTĀGŌ, PLUM BĀGŌ, PROPĀGŌ, etc.). 3. Semantic analysis. A brief reëxamination of the definitions so far cited prom ises to take us even farther, in certain respects, that the step-by-step morphological dissection already provided. Given the fact that -aggine, from the start, straddled nominal and verbal families (PLANTÄGINE related to both PLANTA and PLANTĀRE) and developed its maximum potential at the intersection of the two, namely in the participial domain, it hardly comes as a surprise to learn that the suffix at

406

AFFIXAL DERIVATION

issue, not unlike -eria, designates both conspicuous qualities (invariably those ob served in living beings, i. ., characteristics) and typical actions in which individuals endowed with such qualities would engage. Thus, one encounters, side by side, (a) 'absent-mindedness', 'stubbornness', 'duplicity', and the like (the English counter parts ordinarily end in either -ness or in -ty), and (b) 'stupid action', '[use of] foul language', 'light-headed behavior'. One can perhaps assign an intermediate position, along this axis, to such English glosses as 'folly' and 'villainy'. (In a small minority of cases, the translation into English has opened up the possiblity of the speakers' reference to some physical or mental condition resulting from either a previous action: 'amazement', 'bewilderment', 'grief, 'fatigue', or some involuntary action: 'drooping of the lines', 'nodding of the head'). Distinctly more revealing, and less expected, is the discovery of the specific kind of character traits, and of actions resulting therefrom, that can be suggested by -aggine. To be sure, one runs into a wide spectrum of uniquely attested, hence atypical, meanings: 'deceitfulness', 'sleepiness', 'wickedness', etc. What matters most in the present context, however, is the range of the recurrent denotations and connotations; and, in search of these, one runs, again and again, into such equivalents as 'stubbornness, pig-headedness'; 'rashness, recklessness, heedlessness, mistake, inattention, absent-mindedness'; 'coarseness, dissolution, untidiness'; 'silliness, foolishness, stupidity, mistake'; 'swaggery, pride'. Equally noteworthy is the record of semantic lacunae; one looks in vain for Italian counterparts of, say, 'delicacy, gingerliness, tidiness, tenderness, sweetness', or, at the opposite end of the line, 'bitterness, moroseness, garrulousness, avarice, sourness, feeble mindedness, senility'. The qualities involved do not evoke childhood or old age; neither, for that matter, do they suggest discipline, orderliness, elegance. The indi vidual portrayed (or the young domestic animal sketched) is a young adult, usually male, still a bit wild or, at least, very gauche, displaying his newly-acquired strength and size in many irritating ways;he is inexperienced, big, rude, swaggering, stubborn, and foolish. The picture is unflattering — but an equally damaging, laughable image might, let us admit, have been drawn of prissy, talkative old women or of dirty old men. However, -aggine has at no time been geared to such alternative purposes. 4. Rival suffixes. Given the considerable functional and semantic range of -aggine, one would expect it to be locked in constant struggle with members of other derivational suffix groups, which the Italian speech community presses into service to mint adjectival and verbal abstracts, mass nouns, and the like. There exist, of course, isolated instances of complete, irreversible semantic differentiation, as with aforementioned sboccataggine 'coarseness of speech' vs. ֊atura 'the removal of a layer of oil from the top of a flask of wine', -amento 'outflow, outfall, outlet'. These cases are, all told, less .challenging than those - the vast majority — which exhibit very fine-meshed distinctions, and thus invite microscopic examination, which cannot be here provided. Labeling the bulk of competing variants as so many near-synonyms, one can, impressionistically, set off cases in which -aggine has been involved in a contest, ephemerous or protracted, with -ata (buffon-, corn-), -eria (balord-, bribant-, capon-, casfton֊, ciuc(h)-, gol-, ribald-), -ezza (fanciull-, ridicol-), -ità (lezios-), -mento (abbaglia-, assorda-), -ume (pecor-). In more complexly architec-


407

tured word families conflicts have occurred, over the centuries, between -aggine and a variety of combinations: -anza and -agione (dimentic-); -agione and -atura (gonfî-); -eria and -ità (asin- and astrus-); -ezza and -ismo (astratt-); -ata and -eria (dottor-); -ismo and -udine (ebet-); -eha and -ina (furfant-); -eria and -ezza (goff-); -ia and -ità (ritios-); etc. In extreme cases, as many as four suffixes have competed for the favor of a single adjective; thus in addition to -aggine, such foreseeable rivals as -eha, -ezza, and -ura have courted grosso, while -aggine has shared the privilege of joining lordo with -ezza, -ume, and -ura. In still other contexts, very unusual suffixes have made their appearance as contenders, as when grammat-isia 'booklearning, clerkship' came close to endangering -aggine. There is, practically, no end to the variations. Those cases where -aggine has reigned supreme — not infrequently in cooccurrence with -os- and -(a)t-: accidiosaggine, ambiziosaggine, fastidiosaggine; avventataggine, idiotaggine; gaglioffaggine, grullaggine, gufaggine; infingardaggine, melensaggine, mellonaggine; dabbenaggine, dappocaggine — are all the more arresting, inasmuch as they disclose the actual semantic kernel of the suffix under scrutiny. This derivational analysis, however sketchily executed, fully corroborates the results of earlier semantic dissections, which show that the irreducible nucleus of -aggine is the signaling of exuberantly youthful, thoughtless, self-confident behavior. 5. Suffixal affinity. Of even greater potential relevance to our quest than the emulation between similarly oriented suffixes is the mutual affinity — observable through coöccurrence — among this class of grammatical morphemes. Thus, nominal -aggine and verbal -eggiare may here and there have supported each other; witness zoppaggine 'lameness' alongside zoppeggiare 'to limp a bit, have a slight limp', (fig.) 'not to go quite straight'. More complicated is the case of vagabond-aggine which, in its struggle with -eria, enlisted the support of both -eggiare 'to loaf, lounge' and -aggio 'vagrancy'. Despite the disparity in backgrounds (-aggine is of straight Latin descent, -eggiare is Graeco-Latin, while -aggio was imported from GalloRomance), the three elements, on the level of modern Italian, have been capable of forming a viable alliance. Similarly, corbellaggine 'stupid action' and corbellonaggio 'extreme stupidity' have become natural allies; the latter obviously presupposes the existence of a humorous augmentative in -one. Fintaggine 'deceitfulness' and fîntacchifu)olo 'sharp, cunning boy' have in common the central vowel placed in strategic position after the dental consonant; this partnership bolsters the chances of -aggine against its arch-rival -eha. Far and away most noteworthy is the mutual attraction between -aggine and the local descendants of -ÄCEU: Tusc. -accio, dial, -azzo, -asso, and the like, inasmuch as, in addition to sharing the tone-setting vowel a, most of these forms also display a characteristic affricate: [ d ž], (tg], or [ d z]. One recognizes a thread leading from fanciull-accia 'wicked girl' and -accio 'bad boy, naughty boy', 'man who behaves childishly' to fanciullaggine 'childishness, puerility'. Lungaccio 'tediously long, end less, boring, slow', in a way, feeds lungaggine 'slow, tedious procedure', 'dallying, dawdling, long-drawn-out business', strengthening the prospects of the latter against neutral lunghezza 'length'. Most persuasive is, presumably, the case of buaggine 'slow-wittedness, doltishness', extracted ostensibly from bue 'ox', but undoubtedly nourished by its secondary link to pejorative-augmentative buaccio 'dolt, blockhead,

408

AFFIXAL DERIVATION

ignoramus'; there is also on record a by ֊form buass-aggine 'dullness, stupidity', which seems to involve dissimilatory deaffrication and concurrent dentalization of /Č/ (through appeal to a northern dialect form?). Interestingly, one finds no connecting lines to -iccio, -izzo (and other z'- suffixes) or -uccio, -uzzo (and other u- suffixes). We have thus encountered an independent confirmation of our earlier suspicion that -aggine is far removed from overtones of sentimentality (whether endearment or commiseration bordering on contempt). The suffix under study implies bigness, coarseness, impetuosity, which one associates with unbridled adolescence and aggressive young adulthood - and the phonosymbolic correlate of this reference happens to be the vowel a in Italian. 6. Glimpses of the historical perspective. Within the space allotted to this note, it is impossible for the analyst to attempt any step-by-step reconstruction of the events that have led to the crystallization of present-day -aggine. By way of a substitute for a thorough investigation, one can tap, for a randomly selected portion of the corpus (from Me- to Su-), the chronological information supplied by C. Battisti and G. Alessio's Dizionario etimologico italiano (1950-57). One question that can be raised and tentatively answered, even with drastically reduced resources, is whether -aggine has continued to be productive in the 19th and, by implication, in the 20th century. The answer to this query is affirmative, as we can see from this sampling of relevant derivatives coined, in all likelihood, after 1800: meticolosaggine, minchionaggine, minuziosaggine, nullaggine, piccosaggine, poltronaggine, presuntuosaggine, sbalorditaggine, sboccataggine, scapataggine, scapestraggine, sfrontataggine, sgarbataggine, sguaiataggine, spiritosaggine, stordi՛ taggine, stupidaggine, svagataggine, sventataggine, sversataggine. The suffix, then, appears to be in full bloom, and its association with the pre ceding -at- segment plus the s- prefix (or, alternatively, with -os-) could hardly have been closer at any preceding period. The second question that comes to mind is whether there has been any steady continuity in the growth of -aggine. Again, the answer is positive, judging from the following record of first attestations: (a) 15th century — melensaggine (Bembo), mulaggine (Doni), (miner.) piombaggine, scellerataggine (Gelli), scimunitaggine (Allegri), sdegnosaggine (Tratt. segr. cose), sfacciataggine, sgangherataggine, sgraziataggine (Allegri), smemorataggine (Caro), spensierataggine (Borghini). (b) 17th century - ritrosaggine, sbadataggine (Redi), scempiataggine (Magalotti), sciattaggine, scioper(at)aggine (both variants in Salvini), svenevolaggine (Redi), svogliataggine (Segneri). (c) 18th century — scempiaggine, straccaggine. This stratificational break-down is particularly helpful in demonstrating the gradual expansion of -ataggine, which was previously conjectured on circumstantial evidence. Where doublets are available, the DEI makes it highly probable that the longer


409

variant involving -at- generally trailed its shorter counterpart by a margin of two centuries or more, as is true, specifically, of scellerf'atjaggine and smemorfatjaggine, though apparently not of scempi(at)aggine. The last question to which we may address ourselves is what the characteristic -aggine derivatives of the formative period, as reflected in a few 13th- and in numer ous 14th-century texts, actually were and what tentative conclusions one may draw from their record for the purposes of genetic reconstruction. Here are some specimens of Italian words already adduced, which happen to be traceable to pre1400 sources: mellonaggine, muc(c)il(l)ag(g)ine ~ mugellagine, pecoraggine (Fra Giordano), piantaggine (P. de' Crescenzi), (bot.) piombaggine, propaggine (with several mean ings), rusticaggine, scelleraggine, sciagurataggine, scipitaggine, seccaggine ('aridity' and, in Boccaccio, 'nuisance'), smemoraggine (Passavanti), sordaggine (P. de' Cre scenzi). Judging from this woefully lacunary inventory, which contains both petrifacts (e. g., mucillaggine, piantaggine, piombaggine) and, in larger numbers, innovations, the words closest to the Latin legacy were, in part, terms for 'sickness', 'defect', or 'default': cf. seccaggine and sordaggine, which, referentially, call to mind LUMBĀGŌ ; in part, words for 'transplant', 'germination', 'growth', e. g., propaggine — attested as early as the 14th century. Admittedly, these two semantic sprouts might have worked at cross-purposes and thus have neutralized and eventually stalled each other's development. But since 'physical defect' easily lends itself to extension to 'mental weakness or aberrancy', an entirely different course was taken before long: At the intersection of the two lines of 'mental (or behavioral) aberration' and '(rapid) growth, buoyancy' speakers discovered a rich mine — namely the lexical field of 'silliness, stupidity, stubbornness, conceit', or 'silly, stubborn action, etc.', which thus became the semantic center of a rapidly lengthening line of -aggine formations. In terms of phonosymbolism, the center thus arrived at effectively reconciled -aggine with a number of other derivational suffixes characterized by a, on the side of vowels, and sometimes by an affricate as well, on the side of consonants.

THE DOUBLE AFFIXATION IN OLD FRENCH GENS-ES-OR, BEL-EZ-OR, OLD PROVENÇAL BEL-AZ-OR

Old French bel-ez-or 'more beautiful' is familiar to every Romance scholar from line 2 of the Cantilene de Sainte Eulalie, a passage so lapidary in its paratactic con struction as to leave no margin of doubt as to the exact meaning of the unusual qualifier {Bel avret cors, bêliez our anima 'she had a beautiful body, [but] an even more beautiful soul' — in reference to the martyred Hispano-Roman virgin). Gram matically, belezor (obi. sg.) is a very striking formation, whether one examines it in reference to the Latin prototype of bel ( b e l l u s 'pretty, handsome, cute'; = Fr. 'bellot, joli' [Ernout-Meillet], applied at the outset solely to women and children and genetically related to b o n u s , b e n e ) , or to cognates of the Old French adjectives in sister languages1 - with the unique exception of Old Provençal — or to its substantivized adverbial neuter form (le bel-ais, -ois, like le mieux), which in turn poses challenging problems for the language historian. Prov. bel-az-or (obl. sg.) is, true enough, reconcilable with OFr. bel-ez-or, but this mutual compatibility does not of itself make the two forms any more transparent in derivational terms. Moreover, the relation of bel-az-or to its subj. case bel-aire constitutes another puzzle, the more so as OProv. bel-aire and OFr. bel-ais, -ois do not match. There existed in Old French a small group of comparatives in -our, a few of them traceable to Latin "irregular" comparatives in -(i) ōre and, more often than not, involving suppletion (gignour i ū n i ō r e, maour m a i ō r e, meillour m e 1 i ō r e, menour m i n ō r e , peour p e i õ r e).2 Other items in this close-knit group repre sent analogical extensions of the pattern: forço(u)r 'stronger', graigno(u)r 'bigger', noaillour 'smaller, more mediocre' (traceable in the last analysis to ոūg-â 1 i s 'frivolous, trifling, worthless, empty', from n ū g a e [pl.] 'trifles, nonsense', cf. n ū g-ā t o r 'joker, fibber, babbler', n ūg - ā x beside - ā t ō r i u s 'worthless, non sensical'); and all of them flanked by differently accented nominatives: fortre, graindre, juindre, maire, me indre, mieudre, noaudre, and pire. There also existed an even tinier group of innovative -o(u)r formations lacking subject-case variants, hence completely lexicalized: hauçour 'higher, taller' (reminiscent, as regards the assibilation of the dental, of forçour), joveignour 'younger' (involving recomposition of 1) A side-issue is raised by OSp. bellido, which lends itself to analysis as a blend of b e 11 u s and m e HT t u s 'honey-sweet', as I tried to show, perhaps somewhat clumsily, in one of my etymological juvenilia; see Language, XXII (1946), 284-316, and the post-script, ibid., XXIII (1947), 429 f. The hypothesis may derive some collateral support from the equation remilgarse 'to be prim and finicky, to smirk' < *r e-m e 11-i  à r e, but hardly accounts for the problem in all its ramifications, as the evidence of OPtg. velido cautions us to recognize. 2) I am falling back for the following statements on the material expertly marshaled by the authors of standard historical grammars (W. Meyer-Lübke, HGFS; O. Schultz-Gora, Altproven֊ zalisches Elementarbuch ; and others).

412

AFFIXAL DERIVATION

i û ո i  r into *i u v e n i o r), and sordeio(u)r 'worse' (based o n s o r d i d u s 'dirty, filthy, shabby; low, vile, vulgar').3 The crystallization, in this highly selective list, of future petrifacts based on n ū g ā 1 i s and s  r d i d u s makes it plausible that these two adjectives, at a certain remote stage, played a major rôle in the GalloRomance lexicon.4 The record of Provençal by and large confirms the impressions so far gained; such differences between the two sister languages as one detects by dint of search amount to little more than "accidental" lexical details and morphological rather than semantic nuances. One encounters, once more, a small solid block of neatly paired-off archaisms: máier beside máire (subj.) ~ maiór (oty.),mélher ~ melhór, ménre ~ menor, pleier ~ peiór; an occasional example of the spread of this pattern: génser ~ gensór, plus a residue of completely isolated -or forms: aussór, lonhór, nualhór, sordeiór.5 An idiosyncrasy of lonhor is that it seems, at first glance, struc turally closer to the adv. lonh l o n g e (cf. Fr. loin, It. lungi, Cl.-Sp. lueňe, Ptg. longe) than to the adj. lone l o n g u ; genetically, it simply reflects (unleveled) 1  ո g i õ  . The preference accorded to [] over [ո] may have received collateral support from the pressure of [X] in melhor, nualhor and of [j] in maior, peior, sordeior; see my remark on menhs, next page (fn. 7). The relative prominence of a neatly detachable suffix -o(u)r used in grading - both in the North and in the South of Gallo-Romania ֊ makes it highly plausible that belezor lends itself, at least historically, to the segmentation bel-ez-or (neatly paralleling OProv. bel-az-or), all of which would tend to make -ez- and-az-, respect ively, typical "interfixes" in the sense currently given to this term on both sides of the Atlantic.6 It is, consequently, the elusive -ez-j-az- element rather than the 3) The survival of s  r d i d u s on the vernacular level is an idiosyncrasy of Gallo-Romance; Sp. and Ptg. sórdido are strictly learned formations, absorbed at a late date and semantically congealed. 4) N ū g-ā 1 i s is recorded at a late date, and the glosses carry even the abstract ո ū g-ā 1i է â s. For a bird's-eye view of ո ü g ä 1 i s and s o r d i d u s see REW 3 5989 and 8096. While both words seem to have lingered to the north of the Pyrenees and to the west of the Alps as downright provincialisms, other members of their respective families struck root outside Gaul; this is true of *nū-, *n o-g i ո a 'trifle' (REW3 5990), a base pieced together from Italian dialect words, and of (sg. and pl.) s o r d ē s 'dirt' (preserved in Sardinian) beside s  r d i է i a 'id.' (perpetuated in Old Italian); Meyer-Liibke's  seems not to hold for Classical Latin. N u g a e itself perished, like numerous pluralia tantum: d ï v i t i a e , t r ī c a e - a s against the erstwhile satellites OPtg. trigança 'haste', nigar-se 'to hasten', trigoso 'hasty' - etc. One isolated attempt to deny that OFr. OProv. anceis was a comparative or, speaking with Karl Jaberg, an "elative" must be written off as unconvincing: G. G. Nicholson, "La terminaison adverbiale non comparative -eis en provençal et en ancien français", RLiR, VI (1930), 189-202. 5) This distributional pattern coincides with the gradual prevalence, among the agentives, of -adór over -aire and, among imparisyllabic abstracts in cognate languages, of descendants of -ä է i õ ո e, -ī t i ō n e over those of -â է i ô, -ï է i 5 (Ptg. cans-aço 'fatigue', inchaço 'swelling', Sp. bollicio, later bullicio 'bustle, rumble, wrangle', lit. 'ebullition', etc. are mere relics). 6) I hardly need go here into the history of this no longer controversial term (and underlying concept), coined almost casually by H. Lausberg, put to use shortly afterward in a lengthy monograph of my own (1958), and favored not only by Gonzalez Ollé in his doctoral disserta tion on diminutives, but also, of late, by B. Hasselrot in his inquiry Étude sur la vitalité de la formation diminutive française au XXe siècle, Studia Romànica Upsaliensia, VIII (1972), 13, 67.

THE DOUBLE AFFIXATION

413

root morpheme or the terminal suffix ֊or that clamors for cogent interpretation. Before proceeding any farther, we shall be well-advised to record the survival of a small set of adverbial comparatives traceable, in the last analysis, to the Latin neuters: OFr. mieus, moins, pis, plus — all four splendidly preserved to this day, with counterparts outside Gallo-Romania (e.g., It. meglio, meno, peggio, piu; Sp. menos, obs. chus ~ plus beside más m a g i s , etc.) — plus anceis 'previously', related to a ո է e(a), and sordois, from s o r d ĭ(d)i u s. Sord-ois has sometimes been credited with the rôle of a model in the coinage of bel-ois, though the contour of the semantic polarization is less than perfect. Practically the same series exists in the South: anceis, mais, menfhjs,1 mielhs, pieitz, so rdeis, except that the last -mentioned forma tion seems here to have given rise to a somewhat longer string of imitations: force(i)s 'stronger', gens-e(i)s 'more beautiful', long-e(i)s 'longer'. The slightly greater vitality of ֊e(i)s in Old Provençal is not irrelevant to the problem of dual affixation here under study. The lexicon of the Old French romance Horn, authoritatively edited and annotated by a philologist as experienced as was, especially toward the end of her career, Mildred K. Pope, seems to go back to Poitou, i.e., the South west of the Oïl territory.8 It may be controversial, as Jacques Monfrin has just, appropriately, reminded us,9 to press the point of the given text's dialectal coloring as energetically as the editor saw fit to do. Even so, in principle, scattered traces of regional speech may very well have survived, and that speech was, of course, characterized by a heavy overlay of Oil on Oc; to put it differently, Old Poitevin, for historical reasons entirely transparent, at a certain cut-off point switched allegi ance from the Old Provençal to the Old French alliance of dialects.10 If this is so, small wonder that gens-e(i)s, its exact function barely grasped in the new environ ment where -e(i)s was underdeveloped, had to be explicated as comparative through the addition of a by then more familiar suffix, namely -or. Gensesor in Horn, placed in its spatio-temporal context, thus poses no insuperable difficulty. No comparably neat solution can be expected for belezor and its congeners. A good starting point for the final phase of the analysis is the classification of vari-

7) The [ɲ] of menhs is at first glance baffling, but falls into place once we observe that in the other members of this tightly-organized set the ֊s (or -tz) is invariably preceded either by palatal I (mielhs) or by / j / . 8) The Romance of Horn, by Thomas, ed. Mildred K. Pope, Vol. II: Descriptive Infroduction, Explicative Notes, and Glossary, rev. by T. B. W. Reid (Oxford, 1964) = Anglo-Norman Texts, XII-XHI. 9) See RPh, XXVI (1972-73), 602-612. The critic accepts (p. 611) the reconstructed reading gencesor (line 147), not exactly ahapax legomenon (Tobler-Lommatzsch, AFW, IV, 269, adduce illustrations of the trisyllabic derivative, from Blancandin and the Roman d'Alexandre), but, decidedly, an uncommon form. Monfrin doubts that gencesor could have been extracted from the OProv. comparative genees and, on that same page, declares, on summing up in counterview his own linguistic analysis of the chosen text: "Je crois donc, au terme de cet examen, qu'il n'y a rien de 'poitevin' dans Horn". As regards gencesor, I happen not to share his skepti cism. 10) "Zur sprachlichen Gliederung Frankreichs" - a paper originally written for the Ph. A. Becker Festschrift: Hauptfragen der Romanistik (1922) and eventually absorbed into the first miscellany (1937) of the author's own Ausgewählte Aufsätze.

414

AFFIXAL DERIVATION

ants, as assembled and philologically filtered by A. Tobler and E. Lommatzsch: 11 (a) bellezour: the best-known form, thanks to its conspicuous occurrence in Eulalie — but, paradoxically, the one least frequently recorded. (b) belisor, with minor variations in the terminal suffix (easy to account for in spatio-temporal terms): (a) belisor: "Ainc belisor(s) ne porent estre" {Brut, eds. K. Hofmann and . Vollmöller, Halle, 1877, v. 3944); also in De saint Alexis, ed. J. Herz (Frankfurt a. M., 1879), w . 131, 493; La vie de saint Alexis (rédaction interpolée du XII e siècle), eds. G. Paris and L. Pannier (Paris, 1887), v. 53 ; (β) belisour: "Un chevalier encontre . . . / de membres et de vis ne vit nus belisour" (Li romans d'Alixandre, ed. H. Michelant, Stuttgart, 1844, 459.9); (7) beliseur "Eslire doit le beliseur / et le plus fine et le milleur" (Gautier d'Arras, Œuvres, ed. E. Löseth, I: Êracle, Paris, 1890, v. 2712); (c) belissour: ". . . ot paint en un destor / Elaine la röine, n u s n e vit belissor"(Le Chevalier au cygne, ed. C. Hippeau, Paris, 1874, v. 115); also Durmart le Galois, ed. E. Stengel (Tübingen, 1873), v. 40; Aiol et Mirabel, ed. W. Förster (Heilbronn, 1876-82), v. 3374, significantly, in rhyme with its semantic near-opposite nöelors; (d) belïor: "Si biaus estoit [li fils] qu'en nule terre / ne couvenoit belior querre" (Gui de Cambrai, Balaham und Josaphas, ed. C Appel, Halle, 1907, v. 380). Of these forms, bellezour is doubtless the original, by virtue of its occurrence in a text as archaic as Eulalie and on the strength of comparative evidence (OProv. belazor). Because -ez- was an isolated segment and, worse, one devoid of any identifi able meaning (the feature which, in the first place, makes it an interfix), it was subject to all sorts of rather whimsical distortions: -is-or may have been suggested by abstracts in -ison -i է i o n e (like traïson 'betrayal'), -iss-or by agentives (like envahisseur 'invader'). Belïor stands somewhat apart and may, indeed, represent a hapax legomenon — a variant minted in imitation of straight Latinisms in ֊or, -eur {supérieur, etc.)? But if OFr. bel-ez-ofujr/'OProv. bel-az-or thus emerge as the indisputable kernel, the need to trace the interfix, now more sharply delimited, to a defensible prototype remains all the more urgent. Meyer-Lübke, in Part I of his French grammar, very hesitantly invoked [*]b e 11 āէ u s. Perhaps the diffidence was exaggerated: b e 11 à է u 1 u s is on record in Plautus {Casina, v. 254) and may very well represent a compromise between b e 11 u 1 u s and *b e 11 ā t u s. Moreover, given the proven coexistence of â է e r 'black' and ä է r ä է u s 'blackened, wearing black' as well as o f s o r d i d u s 'dirty' and s o r d i d ā t u s 'sullied, dirtied', *b e 11 ā t u s acquires a clearer semantic physiognomy: It could plausibly have meant at the start 'prettied up, nicely dressed up', and the like. But *b e 11 ā t u s, though a suitable point of departure, does not suffice to ex plain belazor'/bel(l)ezo(u)r. We are free to extrapolate, *b e 11 ā t i ō r e, with a 11) Altfranzösisches Wörterbuch, 1:7 (1924), 911 f., s. w. belesor (belisor) and belïor; belais appears in col. 910, without formal mention of belois. Limitation of space makes inadvisable a systematic and critical survey of numerous earlier pronouncements, which include A. Hammer sfahr, Zur Komparation im Altfranzösischen (diss. Strassburg, 1881; unavailable to me), p. 15, and A. Horning, Zur Geschichte des lateinischen C vor E und I im Romanischen (Halle, 1883), p. 23.


415

development of -tj- > ֊z- paralleled by r a t i ō n e 'speech, reason' > OProv. razo\ but given the meager textual support for *b  11 ä է u s and the severely limited vitality of the traditional -i  r, -i ō r  type, a more realistic context is needed to lend credence to a chain of hypotheses. At this point it is high time to introduce the collateral evidence of OProv. viatz, OFr. viaz lively, quick(ly)', a word we have so far deliberately disregarded, though, as a reflex of v ï v ä  i u s, lit. 'quicker, more lively' (from vi v-ä x [a] long-lived, long-lasting', [b] 'quick to do something, e.g., to learn') it might have deserved mention at an earlier juncture.12 In real life, v i v ä  i u s! could very well have been both a command form (addressed to pupils, apprentices, workers, sailors, soldiers; cf. Russ. živo! 'quick', lit. 'alive, lively') and a qualifier, especially since on the collo quial level of Latin the suffix -āce (nom. -ä x) was on a wide scale expanded to -ā  e u, identical with - ā c i u in actual pronunciation. The type v i v ā k j u, of demonstrable viability both in the northern and the southern zone of Gallo-Romance, could plausibly have provided the stimulus for the elaboration of *b  11 ā t u s into *b  11 ā t i u and, under the pressure of m  1 i ō r , etc., into *b  11 ā t i õ r e, all the more easily as -kj- and -tj- had meanwhile likewise all but coincided. The influence hypothesized would have been most potent at an early stage, before b e l l u had developed into a full-blown rival of, and, in certain territories, substitu te for, p u l c h e r 'beautiful, handsome' and f ō r m ō s u s 'shapely': 'prettiness,' 'cuteness', 'youthfulness', and 'liveliness' all go together! The intermediate link *b  11 ā t i u receives powerful support from the OFr. comparative adverbs belais, belois. Though -akju normally yields OFr. -as (var. -az), witness b r a(c)c(h)i u braz 'arm',la q(u)eu laz 'noose, snare, trap', (c r ï b r u) s a e t ä c e u seaz Ç> sas) 'sieve (made of bristles)', it is a fact that under analogical pressure ("Systemzwang") -ais may crystallize instead, e.g., in the paradigms of certain verbs: fais 'I do', plais 'I please', etc. It is, therefore, perfectly safe to argue that the contiguity of pis, etc. could have led to bel-ais (via ephemeral *bel-as, -az) and, further, that the example of sord-ois could have stimulated speakers or writers to toy occasionally with bel-ois. In this light it is rewarding to examine the illustra tions of AFW, I, 910: 12) E. Gamillscheg, in his EWFS1 (Heidelberg, 1926-28), p. 894b, distinguishes between the French Renaissance word vivace, which he views as a learned borrowing from Classical Latin, and viaz, traceable to a VLat. offshoot in -  e u . O. Bloch and W. von Wartburg later added the observation that vivace can refer to Vitality', whereas vivacité suggests only Vivaciousness'. Interesting to me are the two facts that, first, on the familiar level, as in Russian (see above, in text), the connotation of 'speed' became the central meaning and, second, as in E. fast, Sp. ligero, and Fr. vite, the use of the formal adverbializer (-ly, -mente, -ment) has been normally dispensed with, so as not to slow down the command word for 'quick!'. Cf. Bernart de Ventadorn, no. 18: "Messatger . . . / viatz ven e viatz vai" (. Appel: 'Geh schnell und komme schnelT). I cannot possibly go here into the highly controversial etymology of Fr. vite, orig. viste (both adj. and adv.); but since the traditional tentative association with the past-participial stem of v i d e r e has turned out to be entirely unsatisfactory, I incline at least experimentally to favor connection with the segment v ï- 'life, live', either via a blend with the Germanic word underlying It. lesto 'agile, nimble, quick, brisk', or on the assumption of some such participial variant within regional colloquial Latin as *v T s c ǐ t u (cf. vécu 'lived' < *v ǐ s c-ūէ u) - through association with v i s c e r a 'entrails, guts'?

416

AFFIXAL DERIVATION

Et de Loon fu il nez et estrais, / et de paraige del mieux et del belais {Raoul de Cambrai, eds. P. Meyer and A. Longnon, Paris, 1882, v. 2446); C'est des barons de France, del miauz et del belais {Orson de Beauvais, ed. G. Paris, Paris, 1899, v. 2); Je suis de France do chastel de Peviers, / do Gastinois, do balois13 et do miez {La Prise de Cordre et de Sebille, ed. O. Densusianu, Paris, 1896, v. 62); Et des plus haus de Rome le belais et la flor ("Fragment d'une Vie de saint Eustache", ed. P. Meyer, Rom., XXXVI [1907], 12 ff.; at p. 27, line 323). Lommatzsch, in listing these passages fifty years ago, labeled them as examples of a comparative neuter substantive. I am far more impressed by the consistently formulaic character of the binomials del miauz et del belais or le belais et la flor. Grammatically, this limitation stamps bel-ais, -ois as a transparent archaism, by the standard of medieval French; stylistically, belais is a distinctly epic word. This latter characterization is hardly applicable on the same scale to belazorjbelezo(u)r, which, though undeniably poetic, are equally compatible with a lyrical (amorous), epic (heroic), and even hagiographie tenor. In any event, whichever the literary context, troubadouric, narrative, or martyrological, the last remnants of the family of *b e 11 ā t-(i)u, - i ō r e caught in the dragnet of the medievalist show complete semantic divorce from the frolicsome meanings that once attached to the twinformations b e 11 ( ā t ) i u and v i v ā  i u. This inventory leaves solely OProv. belaire unaccounted for. The suffix -aire as a reflex of -ā r i u did undergo a certain expansion in Provençal (the best-remem bered example is veiaire 'visible, clear, obvious' from v i d ē r e ) , but no red thread, not even one attached to problematic verai 'truthful', seems to lead from here to belaire. Unacceptable on every account would be any attempt to invoke b e 11 ā t   'warlike, militant, valorous, spirited', given the early decline of b e 11 u m 'war' (precisely — at least in Gaul — in conjunction with the spread of the adj. b e 11 u s). We are thus tempted to isolate maire (nom.) m a i   as the least implau sible immediate model. If our analysis so far has been felicitous, we are free to assume that adv. *b e 11 ā t i u s originally yielded the adjectival pair *b e 11 ä է-i  r/ -i õ r e (on the pattern of m a i u s or m a g i s — m á i o r / m a i ō r e ) ; and that the uncharacteristic "regular" outcome of the word-final segment -a t i o r, in the subject case, was eventually replaced by the highly suggestive (Meyer-Lübke's "einreihend") -aire, whereas belazor was allowed to echo faithfully the corresponding object case. Thus far the details; but the involved history of belazor/belezo('u)r also enshrines a few messages for the student of general linguistics. 14 It shows anew the formidable

13) In this particular passage the preceding Gastinois could have been an additional factor in swaying the scribe's leanings in the direction of -ois. The wavering between bel- and bal-, though striking, is not unparalleled (cf. the split of the prefix b i s- 'twice, badly-' into bes ana bar-); it seems to indicate a certain loosening of ties between the primitive and the derivative - by then a mere petrifact. 14) Some problems of indirect relevance are attacked by Jaan Puhvel, "Nature and Means of Comparison in Proto-Indo-European Grammar", to appear in the newly-founded Journal of Indo-European Studies. (Dr. Puhvel presented this stimulating paper orally at Berkeley in March 1973.)


417

influence potentially exerted within a small, close-knit group — morphologically defined — by one member on the other(s), calling to mind the vicissitudes of the shrinking strong preterites and past participles in various branches of Romance.15 It raises the disquieting problem as to why two relatively innovative, progressive varieties of paleo-Romance, such as Old French and Old Provençal, should for once have preserved a moribund category of Latin incomparably better than such ordinari ly conservative sister-languages as Old Spanish and Old Portuguese. Could symbiosis with Germanic — expected to be particularly strong in the North — have been the clinching factor, as was doubtless true of the preservation of a two-case nominal declension?16 The reinforcement of a shrinking Latin-Romance category through secondary identification with its Germanic counterpart would explain the temporary resistance (and even small-scale extension) of the adjectival -r- forms — a series which had a pendant in Germanic — as against the unrelieved crumbling of the posterity of the superlatives m i n i m u , p e s s i m u, and p   x i m u (OProv. pesme, pro(i)sme/pruesme; OFr. merme, pesme were lone remnants doomed to quick disappearan ce), where functionally corresponding Gmc. -st- was, formally, no match for Lat.Rom. ֊m~. Finally, on the stylistic level, it seems permissible to compare the -o(u)r of belezo(u)r (from Lat. - o r e ) with the celebrated -o(u)r of ancieno(u)r 'forefathers', Christieno(u)r 'Christians', paieno(u)r 'heathen, infidels', Sarrazino(u)r 'Saracens', restricted to certain formulaic sequences (from Lat. -õ r u m , mase. gen. pl.); both devices, appropriately homophonous, served as parts of a fixed chivalric-ecclesiastical décor. On a still higher plateau of abstraction, both can also be jointly likened to certain syntactic devices traditionally open to such sophisticated writers of a Romance language as have been eager to imitate or emulate the more memorable uses of Latin cases.17

15) I broach a few of these issues in an article all too short: "Deux problèmes de linguistique générale illustrés par le parfait fort de l'ancien hispano-roman", Mélanges offerts à Rita Lejeune (Gembloux, 1969), I, 471-483. The situation of mutual conditioning is equally observable with pronouns and prepositions. 16) Rumanian, mutatis mutandis, would illustrate almost the same state of affairs, with Slavic lending much-needed support to a system of Late Latin declension in a state of advanced dilapidation. Returning to the scene of early medieval France, one can adduce, among other parallels, the amalgamation of Lat. -  i u and a Germanic derivational suffix into -ier (as against expected *-air, cf. v a r i u vair), and individual word blends such as Fr. haut from a 11 u X Frk. *hauh or *hôh 'high'. 17) This last point has been elucidated by a connoisseur of older and classical language and literature as experienced as R. Lapesa: "Los casos latinos: restos sintacticos y sustitutos en españo", BRAE,XLIV: 171 (1964), 57-105.

THE RISE OF THE NOMINAL AUGMENTS IN ROMANCE GRAECO-LATIN AND TUSCAN CLUES TO THE PREHISTORY OF HISPANO-ROMANCE

in linguistics, no doubt as in other sciences and scholarly disciplines, that certain more or less loosely connected problems seem, to the analyst's disappointment, hazily outlined, inadequately integrated, even parsimoniously documented until they are brought together on an experimental basis. A stagnant discussion may then be energized through injection of a new argu ment or through introduction of a fresh set of data. Under such conditions the individual analyses which over a long period of doldrums gave the impression of having been left ''dangling'' inconclusively fall suddenly into place, and the contour of a single major movement becomes recognizable behind all manner of pale reflections and intermittent refractions. This paper attempts to accomplish such a tightening of resources by collocat ing, either for the first time or, at least, more closely than had been done on earlier occasions, such diachronic phenomena observable in Hispano-Romance as IT HAPPENS

(a) the rise of "unstressed suffixes" (e.g., the expansion of OSp. murciego 'bat', lit. 'blind mouse', into murciégalo, later murciéLago) ; (b) the tendency to stress the antepenult, often against the etymological canon, even in learned words ("esdrujulismo", as in — abnormal — atmósfera, [widespread] ópimo — in contrast to, say, It. atmosfera, opimo); (c) the preserva tion of a — and of t h a t vowel alone — in the weakly stressed post-tonic penult (as in ánade 'duck', huérfano 'orphan'); (d) the spontaneous substitution of a for some other vowel just so placed (as in OSp. páxaro 'sparrow, small singing bird' from PASSERE, or in cuévano 'hamper' from COPHINU, or in pámpano 'vine tendril' from PAMPINU, in contradistinction to alegre 'mirthful' < ALÄCRE 'lively, animated'); (e) the exuberant spread of certain interfixes, esp. ֊al- and -ar- (as in burg-al-és '[citizen] of Burgos' or in llam-ar-ada 'flush, flare-up, outburst' from llama 'flame' < FLAMMA); (f) the protracted, pandemic wavering between the suffixes and suffixoids -ego, -igo, -ago < -Icu (and their feminine counterparts), asin cién-aga ~ -ega 'marsh, moor, mudhole' from CAENU, or as in (al)muérdago 'mistletoe', from MORDICUS 'biting'.

A few of these processes have been tentatively juxtaposed before ; in any case a study, however brief, of the interplay of all six is overdue. The structure of this edifice is intricate enough, but despite the advantages accruing to the observer from the separate examination of the Luso-Hispanic state of affairs, as has been the tacit consensus among authoritative scholars

420

AFFIXAL DERIVATION

over almost seventy years, we shall at a certain point attempt to break away from this tradition and introduce, for the sake of contrast and complementation, the independent evidence of Italian, particularly of Tuscan, which displays a very characteristic ''conditioned" development of the next-to-last vowel in proparoxytones, especially in words bequeathed by Antiquity. The lastmentioned term is used here advisedly, because the Italian material shows more dramatically than the data culled from Luso- and Hispano-Romance sources the peculiar rôle played by the isolable Graeco-Latin streak, within the Latin corpus, in the rehabilitation of post-tonic a in the penult. Our grand tour promises to afford us a casual glance at some hidden features of the Hellenization of the Latin-Romance lexicon and may, along a different axis, prompt us to reconsider the status of a Romance "sufijo átono" (an element devoid of any recognizable meaning) as a controversial point of linguistic theory. The entire discussion will be preceded by a cursory survey of the highlights of earlier probings — which, as will be shown, extend over a period of three quarters of a century. 1. The Start of the Discussion. Through a strange twist of events, the year 1905 witnessed the publication of two articles, each by an accomplished master of Romance linguistics in a philological key, 1 papers which at once placed in the focus of attention a phenomenon theretofore almost unknown : 2 the addition to a (predominantly nominal) stem of an unstressed appendage or augment, which one of its co-discoverers somewhat rashly categorized as a suffix and which in 1 R. Menéndez Pidal, "Sufijos átonos en espanol" [dated December 1903], Bausteine zur romanischen Philologie; Festgabe für Adolfo Mussafia... (Halle, 1905), 386-400; C. Michaëlis de Vasconcelos, "Algumas palavras a respeito de púcaros de Portugal", BH, VII (1905), 140-196. Menéndez Pidal was avowedly aware of . Michaëlis being at work on púcaro/búcaro and other -ARO formations, and vice versa, but apparently neither friend actually saw the other's manuscript. 2 Judging from Menéndez Pidal's opening footnotes, the trail-blazing statements were made by the Swedish pioneer dialectologist Å. W:son Munthe, "Ein neuer Beitrag zur Kenntnis der asturischen Mundarten" [review of B. Vigón, Juegos y rimas infantiles..., Villaviciosa, 1895], ZRPh, X X I I I (1899), 321-325, and, that same year, by C. Michaëlis in "Estatinga, estantiga", A Tradiçao (Lisboa), I, 161-173, fn. 52; on the sequel to this article see fh. 3, below. Craddock's dissertation (see fn. 5, below) identifies earlier fleeting references to the periphery of our problem : by H. Schuchardt (1866-68) and J. Storm (1876). Munthe had become sensitized to the regional abundance of proparoxytones in preparing his Uppsala dissertation (1887) based on field-work in two West Asturian communities. Twelve years later, in writing his essay review, he selected the speakers' predilection — observed among East Asturians — for words in -anu, -aru (which he subsumed under a single "suffix") as the 14th among 21 "gemeinasturische Ziige". Here are his examples (I have supplied most of the glosses from M. J. Canellada, El bable de Cabranes, M., 1944, and from the posthumous 2d ed. of B. Vigón, Vocabulario dialecłológico del Concejo de Colunga, M., 1955):

áscuara 'ember', (a)viéspara 'wasp', beriénzanu 'extra-fine heather' ( = Sp. brezo), bígaru 'sea snail', cáncanu 'cancer', carápanu 'medlar', cóngaru 'conger eel' (= Sp. congrio), entruénzanu 'la grasa que se quita al desurdir', giiévara 'roe', mirándanu 'wild strawberry', piétana 'foot of a hide' (= Sp. piezgo), pómpara 'bubble'[with "expressive" nasal before bilabial], rucáncanu 'louse', yérganu 'blight suffered by maize'. Meyer-Lübke's silence in Vol. I I (1894) of his Romanische Grammatik is indeed baffling, especi ally if one recalls his many felicitous statements on the phonological facet of the problem: "Die Behandlung tonloser Paenultima", ZRPh, VIII (1884), 205-242, and GRS, I (1890), 261ff. The former study, completed almost immediately after his dissertation, is characterized by heightened attention to the Greek ingredients of the Latin lexicon ; throughout the 'eighties, the author was, of course, still immersed in Middle and Modern Greek studies, witness his annotated edition of Simone Porzio's Grammatica linguae Graecaevulgaris[a 17th-c. text] in BÉHÉ, LXXVIII (1889).

THE RISE OF THE NOMINAL AUGMENTS

421

the ensuing debate has ever since figured under the tag "sufijo átono". The two participants, Ramon Menéndez Pidal and Carolina Michaëlis de Vasconcelos, were friends, a circumstance which may in part explain why, despite the diver gence of the results obtained, they refrained from engaging in any dispute.3 As a matter of fact, the dialogue was to pick up momentum very slowly : Though Menéndez Pidal, the younger of the two potential "contrincantes", reverted to the problem at intervals,4 he never managed to expand the stimulating article (as he had publicly announced that he might do) into a full-blown, book-length monograph. D. Carolina's concern was, from the outset, incidental, and her interests before long started moving in other grooves, though she reverted to the problem in her long-unpublished Coimbra lectures. Only after an extended break, upon the conclusion of the Second World War, did the rather strident contention — characteristic of the present Central European scene — between, to use G. Eohlfs' scathing label, "substratomaniacs" (such as Johannes Hubschmid) and, if I may coin an antonym, "substratophobes" (such as Harri Meier) revive and intensify the latent division of views on the augments.5 I n connection with the widely-used ticket "sufijos átonos" note this background feature : Around 1900, it was axiomatic with Romanists t h a t the tonicity of a derivational suffix was a sine qua non condition of its productivity in the vernaculars, the prime example being the gradual replacement, in colloquial Latin, of unstressed diminutive -TTLTJS by heavily accented -ELLUS. 6 Since suffixes — a subclass of morphemes — by definition should be carriers of an 3 There are, in fact, on record few clashes and confrontations between the two scholars, of whom the younger was chivalrous and the older ladylike. The high level of tone in their infrequent public exchanges of opinion is exemplified by their masterly notes on Sp. estantigua 'phantom, hobgoblin, scarecrow', Ptg. estatinga < OSp. huest antigua, see RH, VII (1900), 5-9, 10-19. One of the rare collisions occurred apropos of OSp. yengo 'free', whose ancestry is the subject of a forthcoming vignette by a Berkeley team of Hispanic etymologists. 4 For instance, in the consecutive editions of his Manual (elemental) de gramàtica històrica espanola, cf. the rev. 6th ed. (1941), §83:1; and in Origenes del espanol (Madrid, 1926, 19292), §61 bis, pp. 337-343. For full details and additional clues see below. Menéndez Pidal was not afraid to execute full-scale retreats from earlier positions and even complete volte-faces, at least not in linguistics; for illustrations, see my papers: "Toward a Reconsideration of the Old Spanish Imperfect...", HR, XXVI (1959), 435-481, and "The Distribution of -ç- and -z- in Ancient Hispano-Romance", RPh, XXV (1971-72), 1-52. 5 For a thorough and balanced presentation of the various attacks on the problem see J. R. Craddock's revised Berkeley dissertation (orig. submitted in 1967), Latin Legacy Versus Sub stratum Residue: The Unstressed "Derivational" Suffixes in the Romance Vernaculars of the Western Mediterranean, UCPL, LIII (1969), esp. Ch. II: "The Unstressed Suffixes : Definition and History of the Problem" (48-86), as well as a companion piece : "Latin Diminutive Versus Latin-'Mediter ranean' Hybrid; on Proparoxytonic Derivatives of GALLA in Hispano-Romance and Sardinian", RPh, XXI, 436-449 ; also, quite recently, a rather far-ranging review, in Lingua, XXVI : 4 (Febr. 1972), 383-392, of my own 1970 monograph, Patterns of Derivational Affixation in Cabraniego. That review announces one more forthcoming paper, from the pen of the critic, which seems to be relevant: "Las categorias derivacionales de los sufijos átonos: picaro, páparo y afines" (to appear in Estudiós hispánicos: Homenaje a Rafael Lapesa, a miscellany which is also to contain a monographic study by the aforementioned Berkeley Team of Etymologists: "El nucleo del problema etimológico de pícaro ~ picardia; en torno al proceso del présta-mo doble"). Craddock's dissertation has provoked a number of critical responses, some of which bear, directly or obliquely, on the problem here at issue; see A. Greive, RF, LXXXII (1970), 142-145 (par ticularly severe in judging Craddock's suffixal analysis : 143); F. R. Hamlin, RLR, 1970, pp. 309314; H. Meier, ASNS, CCVI (1969-70), 291-295. 6 Apparently, neither the availability of such, for us perplexing, cases as ORioj. tienlla < *TEN-ULA (Berceo, Milagros, 246a, 273c; A. G. Solalinde: 'cuerda, lazo'), nor the long coexistence

422

AFFIXAL DERIVATION

identifiable meaning, which the "sufijos átonos" extracted by Menéndez Pidal in the over whelming majority of cases were not, some experts before long evinced skepticism as to the wisdom of the term championed. Thus, F . Hanssen, without attempting to arbitrate the disagreement between D. Ramon and D. Carolina, discreetly proposed an amended label, involving a grammatically indeterminate key-word: "terminaciones esdrújulas". 7 2. The Prelude. Years before Menéndez Pidal, in Å. W. Munthe's wake, had formulated his suffixal hypothesis, there was a dim awareness, among perceptive scholars, of the diffi culties raised by some of the lexical items at issue. I n his etymological dictionary Diez recognized the contrast between Sp. mascara, It. máschera 'mask', on the one hand, and, on the other, Fr. masque, and likened the former branch to such developments as Sp. cascarà 'shell' top. BRAGA(A) (rather than BRACARA) and for LAVALOS >

*-NOS > top. Lavãos (a hamlet near Figueira) the author appeals to the agency of consonant dissimilation, citing as a parallel BARBARU > bravo 'wild'. One is inclined to place here Cristóvao
Sp. tierno, Ptg. tenro and GENERU > Sp. yerno, Ptg. genro. 14 For two attempts to settle the long-drawn-out discussion of picaro (and its presumed cog nates) see the forthcoming papers alluded to in fn. 5, above. 15 As distinct from Peninsular expressions for 'whipping top', studied in ICC, VII (1951), 201-244, with a post-script in UCPL, XI (1954), 68.

426

AFFIXAL DERIVATION

From the record of anthroponymy the author cites the local family name Cuétara, tracing it to cueto 'crag, hillock', also used onomastically. 16

We cannot afford to examine here in equally lavish detail the remainder of the article, which clearly marked a major progress, not to say break-through, in the author's own growth, as regards both forcefulness of analysis and skillful tapping and blending of heterogeneous sources ; which would rank as remarkable even if judged by today's exacting standards ; and which, almost seventy years ago, represented nothing short of a masterpiece. 17 Nevertheless, one recognizes certain flaws, from the late-20th-eentury's vantage. Thus, it might have been more advisable for the author to exploit to the hilt the strictly formal analysis and subsequently to bring in temporal, regional, social, and stylistic con siderations as a kind of counterVÍEW (the way he himself learned to do things W e n t y years later). Rigorously formal analysis, pushed to the ultimate limit, might, for instance, have prompted him, in confronting (a)galla and gállara, or alizace and lizázara, to raise the ques tion of the compatibility of two accretions : a- and -ra. Menéndez Pidal's lukewarm and only partial acceptance of the Diez-Leite hypothesis failed to spur him on to raise the crucial question as to whether words like cóngaru, pif aro, sôbaro, Vitaro, endowed with "etymo logical" r, b u t devoid of "etymological" a, paved the way for the fully developed -aro "suffix" or, by way of alternative, were secondarily squeezed into a preëxistent mold. I n his lively opening comments the author gives the impression of sensing a cross-connection between the phenomenon a t hand and (a) the widespread "esdrujulismo" (farrago 'hodge podge', médida 'pith, gist, marrow', vis-à-vis FARRAGO and MEDULLA) as well as (b) the locally rather limited "antiesdrujulismo" of (a) the hill-dwellers around Huesca and Barbastro (cantáro 'jug', jicára 'chocolate cup', sahána 'sheet, altar cloth') and (/3) Salamanca's "charros" (Geronimo 'Jerome', fabrica 'factory', xacára 'merry tune'), but he stops short of building any solid bridge between the two, or rather three, sets of circumstances. Again, there is a tantalizing hint (but little more) of the rôle t h a t the conspicuous preservation of a in the post-tonic penult ("postónica interna") may have played [huêrfano, etc.], b u t any particulars are withheld.

As regards date and origin, Menéndez Pidal, in the concluding section (§9), overrules J. Storm's assumption of relative recency; using, for the sake of 16 This, I believe, was the starting-point for the author's active curiosity about the cuet- root, a fascination which led to more elaborate treatment in a section of Origenes del espaňol and in a separate note, "CQTTO, CQTTA", in RPh, VI (1952-53), 1-4 ("A. G. Solalinde Memorial"), even tually absorbed into the collectanea Toponímiaprerrománica hispana (M., 1952), pp. 267-275 ; cf. J. Hubschmid's more radical counterproposal of a non-IE etymon: "Asturisch cuetu — baskisch lcotor 'felsiger Hügel, Fels': ein Beitrag zur Erforschung des hispano-kaukasischenSubstrates", RPh, VI, 190-198. 17 The advance and self-confidence become doubly perceptible against the backdrop of the author's groping "Etimologias espanolas", Rom., X X I X (1900), 334-379, a paper in which he felt compelled to admit (351n2) his ignorance of Galician-Portuguese, whereas "Sufijos átonos" shows him in full command of Western dialectology. As regards systematic use of rural speech, the later paper marks the mid-point between the neophyte's "Notas sobre el babie hablado en el Concejo de Lena" (originally a section of the miscellany Asturias, eds. O. Bellmunt and F. Canella, Gijón, 1897 ; separatum, Gijón, 1899 ; funneled into the expanded version of El dialecto leones, ed. C. Bobes, Oviedo, 1962) and the accomplished master's monograph, "El dialecto leones", RABM3, XIV (1906), 128-172, 294-311. Impressive is the interfiling of personal fieldnotes, of regional data elicited — through correspondence — from friends (including one of the caliber of M. de Unamuno), and of lexical items extracted, by dint of reading, from medieval texts.


427

reciprocal corroboration, bits of Latin onomàstic material (BRÄCARA), shreds of internal analysis,18 and slivers of evidence culled from medieval vernacular texts (cuérnago 'irrigation ditch', gállara, murciégalo), he traces a hard core of the formations to the "periodo primitivo romance", without yet formally committing himself to ultimate sub strata! provenience (he goes farther in this direction in the 2d ed. of his Manual, published that same year but presumably prepared somewhat later). He admits his inability to pin down any "concrete" meaning or gamut of meanings, but readily recognizes certain stylistic over tones: sparks of humor and folksiness, here and there a scintilla of phonosymbolism. Very brilliantly and cogently, the author observes: "Tales sufijos no son mas que un adorno morfológico" — i . e . , we are here dealing with such suffixes as are not ordinarily grammatical tools, but suggestive elaborative adjuncts. I n a final observation, again remarkably subtle, he makes an attempt to link the -igo, -iga sub-branch (distinctly Western) to epenthetic / j / , likewise characteristic of the West, on the assumption t h a t the g may have been antihiatic, as in OSp. feguza 'faith, confidence' beside feiiza < FÏDÜCIA, or as in OArag. cadaguno beside cada uno 'each one'. 1 9 He invokes this situation as a partial cause — though, let me add on my own, it might just as well represent a mere side effect. 5. Menéndez

PidaPs

Second

Attack

o n the Problem ( 1 9 2 6 ) . T h e

seven

compactly organized pages (337-344 ; §61 bis — presumably a last-minute insert) which the Madrid philologist dedicated to our problem in the original edition of his Oxigenes del espaíïol show his scholarship at the highest plateau of per formance. His terminology has become neater (even though he has not discarded the label "sufijos átonos", the concluding paragraph refers to them, far more appropriately, as "incrementos sufijados"); also, in the twenty-year interval, he has scrutinized a wealth of archaic charters and has tapped reservoirs of toponymie information heretofore neglected. Several hazily recognizable word-families receive — in part, for the first time — a generous share of his attention : ORCA 'container, cavity' (top. Huércanos beside Balluércanes), sótano ~ sótalo 'cellar', bárgano 'stake used in fencing off a country property', Hisp.-Ar. abrécano ~ abricano 'terebinth', related to ábrego < AFRICU, cuérnago 'irrigation ditch', from CORNÜ 'horn', etc. More important, certain facets of the problem until then hidden from purview come into the open, e.g. apocopated nouns ending in ֊I rather than in -lo or -la (the truncation being a sign of Arabic transmission): Huércal (Almeria), murzíkal 18 These arguments are not wholly persuasive. Thus, nuégado 'dough containing walnuts and honey mixed with flour', assuming it stems directly from NXJCE/*NOCE 'walnut' (the author himself supplies an alternative), need not antedate the assibilation of 1, since the influence of -al, -u-era 'walnut-tree' and of other derivatives involving this root morpheme might have made itself felt at a later date. Similarly, the  in cenagal 'quagmire' provides no entirely reliable clue to chronology. 19 I plead guilty to having overlooked this passage in my article "The Five Sources of Epen thetic / j / in Western Hispano-Romance: A Study in Multiple Causation", HR, XXXVII (1969), 239-275 — written to honor, of all scholars, Menéndez Pidal. The g in cadaguno is not exclusively antihiatic, being due, at least in part, to the transfer of the segment -guno, as found in alguno 'some(one)', nen- or nin-guno 'no one'. The -g֊ \y\ in feguza, rather than being antihiatic, may represent a substitute for the etymological -d- /8/ of feduza, a variant once peculiar to Eastern Spain.

428

AFFIXAL DERIVATION

'bat' = OSp. murciégalo; an occasional place-name in Leno rather than in iano, e.g. Solór-çeno, -zeno; instances of wavering between -ano and -amo: várgano ~ hárgamo 'stake'; pristine formations in -olo, -ola: top. Piédrola (beside Piedra), top. Espuéndolas (Huesca) alongside Arag. espuenda 'slope, embankment', also Hisp.-Ar. Toléitola, Tolédola < ToLĚTULA in rivalry with Toledo; numerous meticulously pinpointed toponyms in ֊ega, -iga (Piérnegas, Padiérn-ega, ֊iga).

With an implied allusion to Gilliéron's thinking — then in the foreground of interest —, the author declares from the start that the "unstressed suffixes" represent a morphological device apt to give the words involved a certain formal buttressing for its own sake (''sonora amplitud"), without there being any therapeutic need to protect them from phonetic erosion. He still recognizes a close link between the emergence of the augments and the "regular" preserva tion of a — and of that vowel alone — in the inner chamber of proparoxytones (CAMARA, LAMPADA), but at this juncture posits the following chain of events patterned in a cause-and-effect structure (344) : The complexly layered suffixes are in part of indigenous, in part of Latin provenience. Their autochthonous ingredient explains the speakers' deep-rooted preference for the preservation of a in the "postónica interna" as a peculiarly Hispanic norm of sound develop ment, one which sets off Spanish from its sister languages, though transparent articulatory conditions may very well have reinforced this culturally or histori cally motivated favorite treatment of the central vowel. In this new hierarchy of forces, then, the indigenismo begins to win out over the Latin legacy.20 Since Menéndez Pidal at this point does not yet irreversibly jettison his older position, fundamentally noncommittal if not pro-Latin, it is stimulating to watch him draw a line of demarcation between the two converging elements of the eventual amalgam. (a) I n Section 4 of his intercalary chapter, in which he draws the balance, he recognizes a nucleus of (semilearned) formations traceable to dimin. -ULU, -ULA, the model being OSp. pénola 'quill, pen' < PENNTJLA; the process of step-by-step adjustment (-o- > ֊a-) is best exemplified with OSp. cernicolo 'sparrow hawk' > mod. -alo; further diffusion is observable in gàrgola 'head of flax' and in MESPILU 'medlar' > niésp-olo;21 onomàstic counterparts include Espuéndolas, Piédrola, Puértolas, Tolédola ; and the general character of such items, measured along the sociolinguístic (or educational) axis of either learned retardation or deliberate regression ("cultismo"), is shown by the archaic series artígulo, paupertágula, ribolo < KÏVULU 'small brook', siéculo, tídulus, vogábulo, often — but less than consistently — characterized by a mutually contradictory transcription of vowels and consonants. (b) A further source individuated is the progeny of adjectives in -icu, many of them 20 T h e passage is so crucially i m p o r t a n t as t o deserve u n a b r i d g e d q u o t a t i o n : Estos sufijos primitivos, que yo creo de origen prelatino, apoyados por la tendència del latin vulgar hacia los incrementos sufijados, pudieron ser la causa tradicional de la preferència que el espanol siente por la a postónica interna, la cual permanece a diferencia de las demás vocales que se pierden. Esta preferència se apoya, es verdad, en fundamentos fonéticos, ya que la mayor claridad de la a la hace apta para resistir en esa posición dèbil intertónica; pero no es una preferència general, y por ella se distingue el espaňol tanto del francès que pierde la a con las demás vocales, como del italiano que conserva las otras vocales lo mismo que la a; la preferència  la a reconocerá, pues, en espanol una causa tradicional indígena.

21 The author lists niéspola [mod. níspola], but its masc. counterpart, designating especially the fruit-tree, seems to me more pertinent. There exists the var. níspero. The shift m- > n- is dissimilatory ; was the choice of  -determined by labialization ?


429

secondarily substantivated. Here the author, first, uses to good advantage the testimony of such archaic graphies as basêliga, dómnigo, eglesiástigo, gótigo; in later lexical deposits he then recognizes products (a) either of Latin formations : fésigo 'doctor, scientist' < PHYSICH, lóbrego 'dark, gloomy, dank', (bot.) rapónchigo 'rampion', tósigo 'poison' < TOXICII or (/3) of post-Latin innovations : Ast. huélliga 'trace, scent' [cf. REW3 3417], rechónch(ig)o 'chubby', salómbrigo, zámbigo 'knock-kneed'. 2 2 (c) Menéndez Pidal exercises severer restraint in setting off a third group of archetypes, comprising Latin formations in -ĂNU, predominantly botanic in content. The leader words (GALBANU 'resinous sap of a Syrian plant', LADANU 'gum mastic', ORĪGANU 'marjoram', ORPHANU ' O r p h a n ' , PLATANU 'plane-tree', RAPHANU

'radish', TY(M)PANU

'tambourine,

kettle-drum') are credited with having deflected from their straight course the local descen dants of RNU 'southern wood', CAUPOLU 'small boat' [Ernout-Meillet prefer CAUPULU], INU 'basket, hamper', PAMPINU 'vine-tendril' , and SARCINA 'bundle, pack, burden' — witness OSp. abrótano, cópano, cuévano, pámpano, sárçano (A.D. 974) beside I t . abrotano and cófano. Conversely, archaic bárgano, Sp. burdégano 'hinny', búzano 'diver', retruêcano 'pun', tángano as well as place-names displaying this characteristic contour (Huércanos, Solórzano) are — probabilistically speaking — exempted from their influence. Similarly, the phytonym ASARU, BARBARU 'barbarian', CAMARA 'vault', CAMMARU 'crayfish', CANTHARU 'tankard', and COMARU 'strawberry-tree' ( = ARB???TU) m a y account for the divergence of chicharo 'pea', lámpara, páxaro from the straight course of CICERE [not -RU], LAMPADA, PASSERE, b u t — so the author argues — hardly suffice to explain away alicántara, cáscara, gállar a [not ag-] gárgaro, guácharo. Least of all should BŪFALU [or BŪBALU ; not ???-] 'gazelle' and SCANDALU lit. 'stumbling block' be invoked t o justify Crist-uébalo/-óbal, (ichth.) bonítalo 'bonito, striped tunny', cernícalo 'sparrow hawk', OSp. murciégalo 'bat', pezpítalo 'wagtail', (coll.) trápala 'noise, uproar, chattering', or should ASPARAGUS, MONACUS 'monk', PELAGUS 'open sea, main' be appealed to in an effort to fit into a suitable context bálago 'grain stalk', buétago 'lung', ciénaga 'mudhole', cuérnago, luciérnaga 'glow-worm', muérdago 'mistletoe', ráfaga 'gust of wind', fig. 'burst, flash', rázago 'burlap', etc. Finally, how could one (Menén dez Pidal wonders) lay at the door of BALSAMU 'balsam-tree' (also its gum) or CANNABU > cariamo 'hemp' such items as bála-mo ~ -go 'grain stalk', légamo 'slime, ooze', préstamo 'loan'? The clinching arguments in favor of autochthony are the local toponyms known to Antiquity: BRACĂRA, CAPARA, LEDISAMA, NAIARA, UXAMA (cf. Sp. Ledesma, Osma, etc.),

plus an occasional common noun of demonstrably Iberian ancestry, such as PARĂMU > Sp. páramo 'high barren plain'. 6. Menéndez Pidal's Third Attack o n the Problem ( 1 9 5 2 - 5 3 ) . T h i s s o p h i s t i c a t e d

pronouncement was not to remain the last made known in the author's lifetime. True, the Addenda to the 1929 reprint of his Orígenes contained no elaboration 22 Some of these words involve thorny etymological problems, as the author might have candidly admitted. The equation of lóbrego and L???BRICUS 'slippery' invites serious qualifications — should one posit concomitant influence of lobo 'wolf' and/or of ĒLŪCUBRĀRE 'to compose by lamplight'? To explain zambo, J. Corominas, BDE2 (1967), 620b, starts out from STRA(M)BUS 'squinting', invoking Mozarabic transmission (*eçrambo, cf. Ptg. zambro, with metathesized r; reminiscent of INCASTRĀRE > Sp. enga(r)zar 'to link, wire (jewels), enchase'?). Rechoncho, unrecorded before 1750 and obscure in ancestry (see BDE), hardly deserves a place of honor among the leader words. Rapónchigo, likewise of fairly recent vintage (1780) and, again, geneti cally opaque (cf. the same source), appears to presuppose a blend of ruponce (1505, from It. raponzo?) and ruipóntigo 'rhubarb' < RHEU PONTICU. In any event, these examples illustrate the recency and, by implication, the continued productivity of -igo, rather than its ancient roots, which the author is visibly eager to lay bare. (Rui- is folk-etymological, cf. the case of ruisenor 'nightingale'.)

430

AFFIXAL DERIVATION

on this particular chapter. 23 From the 3d ed. (1950), thoroughly recast, of his masterpiece, Menéndez Pidal — with a rather dramatic flourish — altogether withdrew this chapter, only to insert it, two years later, with very trivial changes (both cuts and inserts), in a miscellany of slightly revised toponymic studies from his own pen. 24 Weightier than the changes wrought on that occasion in the body of the text was an introductory editorial remark barely hinting at an aged Menéndez Pidal's latest opinion : Announcing an expansion of his analysis in a forthcoming Historia general del español, he capsulized his revised conclu sion in the comment incidentally tossed out : "Estos sufijos âtonos proceden de lenguas primitīvas extranas al latin". Thus his thinking, in almost exactly fifty years, had come full circle : From a predominantly Latin hypothesis (1903-05) he advanced to a mixed Latin-indigenous schema (1926), only to end up with a well-nigh entirely autochthonous conjecture.25 23 One finds a brief elaboration, less t h a n startling as regards d a t a a n d analyses, in Section 2 of " N o t a s de t o p o n i m i a " , in Mélanges de philologie et d'histoire offerts à M. Antoine Thomas (P., 1927), 295-300, a p a p e r channeled, w i t h slight changes, into our a u t h o r ' s Toponimia pre r o m á n i c a hispana, 4 9 - 5 8 , a t 5 3 - 5 7 . I refrain from probing such minor alterations as Menéndez P i d a l m a y , over t h e years, h a v e i n t r o d u c e d into t h e tail section of §83 :1 of his Manual de gram tica histórica, s t a r t i n g w i t h t h e 2d ed. 24 See Toponimia, 59-70. As regards Origenes, t h e 1950 t e x t , in general vigorously recast, shows j u s t a few innovations :

the insertion, at the start, of the qualifying phrase, "A primera vista" immediately before "parece un simple recurso morfológico para dar mayor amplitud a las palabras" (61); an elaboration (62, 65) on one detail: the re construction of *Búrgalos (a by-form of the top. Burgos), which alone, one reads, would have justified burgalés 'resident of Burgos'; a compression (66) of the statement on ORNA ( > Huerna, Vid-uerna), so as to obviate the risk of otiose repetition. Is Salúrceno (62) a misprint for Solórçeno (cf. 55)? Soïómbrigo (68) may involve a deliberate substitution, if one may judge from a remark dropped in "Sufijos átonos..." (p. 397 and fn. 2); sa- and so-lombra 'shade' are coexistent vars. (cf. RPh, XXII, 267-279, at 272f.). One m a y disagree w i t h t h e a u t h o r on t h e derivation of burg-al-és, provided one reckons w i t h t h e agency of a n interfix -al-, which could v e r y well h a v e been wedged in t o discriminate between t h e 'denizen of Burgos' a n d t h e ' b u r g h e r ' in general (burzés, later burgués) — cf. t h e function of -ar- in llamarada ' s u d d e n b u r s t of flame' (from llama) vs. llamada 'call' (from Itamar). On t h e rôle -al-, -ar- p l a y e d in such contexts see m y article, " L o s interfijos hispánicos : p r o b l e m a de lingüística histórica y e s t r u c t u r a l " , Estructuralismo e historia, ed. D . Catalan, I I (La L a g u n a , 1958), 107-199, a t 177-184 a n d fnn. 93, 114-119 (and cf. t h e concluding section of t h e present article); on t h e k e y feature of t h e sound development of burzés see " T h e Inflectional P a r a d i g m as a n Occasional D e t e r m i n a n t of S o u n d C h a n g e " , in Directions for Historical Linguistics: A Symposium, eds. W . P . L e h m a n n a n d Y . Malkiel (Austin, 1968), 2 1 - 6 4 , a t 48. 25 N o t t h e least interesting facet of t h e carefully executed revision (1926, 1952) is t h e m a t t e r of t a c t i c subtractions. T h u s t h e a u t h o r a b s t a i n e d from pressing his hypothesis of t h e antihiatic function of -g- [y] against t h e b a c k d r o p of W e s t e r n epenthetic [j], despite t h e t e m p t i n g oppor t u n i t y offered h i m b y selómbri(g)o 'shaded' in addition t o dial. güésti(g)a 'apparition', musti(g)o 'withered, gloomy', a n d nútri(g)a ' o t t e r ' . Also, he refrained from e x p a t i a t i n g on t h e intervocalic a l t e r n a t i o n of (α) g, n, r a n d (β) d, as in E.-Ast. Güévara (Colunga) vs. C.-Ast. Güévada (Oviedo), from hueva 'roe' < OVA 'eggs', or in P t g . relâmp-ago ~ -ado 'lightning', or in Sp. lóbado (beside lobádo) ~ Leon. llóbadu ~ Sp. *Hóbano (supported b y dim. lobanillo) ' t u m o r ' [note, a m o n g r e m o t e parallels, tanganillo ' t e m p o r a r y p r o p ' , en tanganillas 'shaky, t o t t e r y ' , a n d even longaniza 'sau sage']. One reason for this reticence m a y well h a v e been t h e a u t h o r ' s newly-fanned e n t h u s i a s m a b o u t "acoustic equivalence". T h u s , if in 1905 he barely h i n t e d a t t h e p r e s u m e d relation of párpado 'eyelid' t o PALPEBRA, V . - L a t . PALPETRA, t w e n t y years later he viewed t h e wavering párpa-do ~ -ro in a radically new light (and even infected w i t h this fervor T. N a v a r r o on t h e occasion of t h e l a t t e r ' s severe critique, in R F E , of E . G. W a h l g r e n ' s refreshingly i n d e p e n d e n t m o n o g r a p h [1930]). W i t h respect t o t h e molecule of formations leaning on a next-to-final " d " , t h e p a r t p l a y e d b y certain p a s t participles m i g h t h a v e deserved passing m e n t i o n . Sp. pérdida 'loss', t h r o u g h s e m a n t i c polarization, gave rise t o busqueda 'search' beside t h e straight post-verbal busca (note also


431

As a major surprise to the readers of his Toponimia, Menéndez Pidal, then approaching his mid-eighties, in just a few happy months of intensive labor mustered the strength to complete the announced project rather than delaying it ad kalendas Graecas. The third formulation of his views and array of support ing data marks indeed the radical volte-face for which he had prepared his followers. The very title of this third attack on the problem: ''Sufijos átonos en el Mediterráneo occidental", 26 indicates (1) a spatial rather than familial widening of the scope, with full inclusion of the southern coast of the Western Mediterranean basin; (2) the resolute retention of the controversial label "suffix"; (3) the distinctly heavier emphasis on toponymic material, initially used as a mere concomitant ; 27 and, by implication, (4) an eagerness to set store by the substratai ingredients to the disadvantage of the Latin stock. The docu mentation has by now become exuberant, almost baroquely overflowing, while the analysis of many key issues seems lopsided — in some respects, indeed, less satisfactory than were the preceding dissections. With regard to the onomastic corpus, the author is willing to make allowances for some instances of fortuitous homonymy, but, in general, asserts his optimistic readiness to trace at least four classes of isoglosses: Afro-Hispanic, Sardo-African, Corso-African, and SardoHispanic (I am once more merely paraphrasing his own summary). The clinching argument is the spread of '-ar, '-an, '-al, '-ag, and even '-am "suffixes" [actually, word-final segments] over the entire coastal and insular territory, the hinterland often included. The author is correct in admitting that, strictly, the atonicity of '-aro, etc. need not go back to the obscure ancestral languages hypothesized, but may be a pervasive feature of secondary Latinization ; 2 8 but this sober realization — a mere afterthought ? — hardly deters him bóveda 'vault' < *VOLVITA); as a masculine counterpart one may cite obs. empréstido 'loan'. Italian goes farther along this line, exhibiting two loose groups of "mercantile" terms: débito 'debt' and, in its wake, láscito 'legacy, bequest', préstito 'loan' (dial. accáttitu 'purchase', vúschitu 'gain, profit'); pèrdita 'loss', rèndita 'income, revenue', spèndita 'expense' and, on their model, (dial.) cérchila 'request', còmprita 'purchase', créscita 'increase'. See E. S. Georges, Studies in Romance Nouns Extracted from Past Participles, rev. J. R. Craddock & Y. Malkiel, UCPL, LXIII (1970), 14-16, 44f., and the corresponding sections in Annegret Alsdorf-Bollée, Die lateinischen Verbalabstrakta der U-Deklination und ihre Umbildungen im Romanischen, RVV-34 (1970), a Bonn dissertation to be assessed in RPh by Anita Katz Levy. Of the longer reviews of Toponimia (I have not seen those by E. Coseriu), A. Badía Margarit's in RFE, XXXVII (1953), 244-247 sheds no fresh light on the "suffixes". J. Hubschmid, in RPh, VIII (1954-55), 221-225, at 223, analyzes páramo as a hybrid involving PĀR PARE 'equal, even' plus a substratai suffix; he refers the reader to RF, LXV (1953), 280. (If this is so, mor-eno 'dark brown, dark-complexioned' would represent a neat derivational parallel.) My own critique (Spec, XXIX [1954], 588-594, at 591n6) calls attention to the proximity of (a) top. Padiérniga, Cistérniga (from PATERNA, CISTERNA, through the instrumentality of -ICA) and (b) OSp. noch֊ arn(-ern)-iego, orig. *noch֊orn-iego 'nightly' (from NOCTURNU, entailing the services of substratai -AECU), with a reference to SPh, XLVI (1949), 497-513. 26 The impact of the article was enhanced through its inclusion in the prestigious "Homenaje a Amado Alonso" (I) = NRFH, VII (1953), 34-55. 27 This trend runs parallel to the emphasis Menéndez Pidal, in the concluding decades of his life, placed on anthroponymy, witness his study of -z patronymics in BRAE, XLII (1962), 371-460. 28 One striking parallel comes to mind : Exotic names in -i and ֊u (specifically, those of African and Amer-Indian background) are, I believe, obligatorily stressed on that vowel in modern Spanish : Misisipi, Misuri, Timbuctú ; much less consistently those ending in -a or -o : Canadá, Paraná, but Angola, Congo. The reasons for this peculiarity may be manifold : Nominal formations in -i and -u are, practically, unassimilable to the native stock (words like metrópoli and ímpetu,

432

AFFIXAL DERIVATION

from toying with genetic kinship between those vestigially-known tongues. I n the parent languages, we learn, these suffixes were surely appended to common nouns, even though their traces happen to be mostly onomastic; a few of these non-IE lexical items were absorbed into Latin (say, SCALDALA 'spelt' 2 9 ); in yet another lexical layer the suffixes, detached from their exotic moorings, were freely added to Latin nouns (ALBĂRUS, CORNĂGUS), often such as refer to the configuration of the terrain. 3 0 The further extension to verbal stems is peculiar to the Romance phase; witness arrepápalo 'fritter' beside afore mentioned bállava and pícaro. Culturally significant is the confinement of the "suffixes" to micro-toponymy and -hydronymy, to names of unprofitable plants, or parts of plants, as well as of undomesticated animals and birds, and to assorted characteristic features of rural life (gállara, cernícalo; [Sant.] chátara 'footgear made of untanned leather', truébano 'bee-hive or container made from a hollow tree trunk', várgano). Elements of the standard lexicon (relámpago, sótano, vástago) and names of better-known cities (BRACĂRA > Ptg. Braga, orig. -aa) constitute exceptions from the bucolic streak; the fashion for affective diminutives of the type Tolédola, *Búrgalos remained ephemeral.

Conceivably the most perplexing statement in the summation relates to the semantic ambit of the "suffixes". They are now credited with a pristine abundantial meaning, allegedly observable in Cuétara, Cuéllar, ciénaga. Sometimes one perceives a phonosymbolic ("imitative") value suggestive of repetition, swiftness, or volatility, as in pezpitalo, ráfaga, trápala. In all likelihood, we further learn, they carried an emotional-hypocoristic message (affectivity, bordering on diminutiveness) : luciérnaga, murciélago. Given their essential raciness ("uso principalmente vulgar"), they have been occasionally applicable in pejorative contexts : bichángano, paparo, picaro. Ordinarily, these particular overtones evaporate, leaving the "suffixes" deprived of any identifiable mean ing: If cascara was once a mass-noun flanking casca, the two words have ultimately been reduced to the status of plain synonyms. — The author seems unaware of the fact that he is referring to ever-changing, iridescent CONNOTA TIONS and that, so far as straight DENOTATION is concerned, the elements at issue simply lack any that could have been captured in fifty years of intensive study — a circumstance which might well have encouraged him to question if not downright deny their suffixal rank. tribu are peripheral, and CI.-Sp. crisi, favored by B. Gracian, has changed to crisis); in some instances, transmission through colonial French may have played the decisive part (this surely holds for Canada, Mississippi, and Missouri against the background of 18th-c. conditions); then again, the indigenous stress pattern may quite often have indeed been oxytonic ; in the residuum, a kind of phono-stylistic hypercharacterization of exoticism may be operative. Whatever the ultimate cause, or interplay of causes, it surely would be hazardous to reconstruct on such flimsy evidence the nearly-ubiquitous oxytonicity of -i and -u in the source languages ! 29 Ernout-Meillet, DÉLL4, 599a, record the vars. SANDULA and SCANDULA of this word, doubtless an intruder, distinguishing the small cluster of formations from two homonyms: the plural of SCANDULUM, lit. 'stumbling block', a Hellenism (σκάνδαλον), and SCAND-/SCIND-ULA 'shingle' (the older ï var. being also the one better preserved, esp. in Irish and OHG). One suspects a measure of reciprocal influence between these three words, genetically unrelated as they obviously were. 30 The D É L L 4 draws a slightly different picture of ALBARUS (CGL, III, 264.33), listing it under ALBUS 'white' without any formal commitment as to its precise relation to *ALBARU 'white poplar', reconstructed from Romance material (REW3 318). It fails to identify CORNĂGUS. Did cuérnago mean 'horn-shaped ditch' ?


433

7. A Balance Sheet of the First Fifty Years of Discussion. I t is b y no means coincidental t h a t so much attention has here been lavished on t h e earlier phases of t h e discussion of t h e nominal augments, because, in sober fact, the debate, at a certain juncture, got side-tracked, so t h a t , details apart, t h e earlier rather t h a n t h e later pronouncements seem t o point in t h e right (or, a t least, in t h e most promising) direction. E v e n so, t h e major implications of t h e strictly linguistic problems — accentual-syllabic, semantic, afflxal — of the "sufijos átonos" were hardly understood b y the contemporaries and critics of the young Menéndez Pidal (over whom he towered), judging from the available reactions — consistently trivial — of G. Baist, J . Cornu, A. dos R. Gonçalves Viana, V. Garcia de Diego, F . Hanssen, and J . Alemany Bolufer, all of whom, potentially, h a d t h e technical knowledge to subject t h e rising star's exploratory paper (1905) to a truly searching critique. Through a coincidence, which t o some observers m a y appear fatal, t h e years separating the publication dates of "Sufijos átonos" (1905) and Origenes del espandi in its original garb (1926) witnessed an enormous upsurge of scholarly activity bearing on toponymy in con junction with glotto-archaeology — i.e., involving a t t e m p t upon a t t e m p t a t reconstructing prehistoric cultures and settlements along t h e Mediterranean coasts and on t h e respective islands and archipelagos through joint excavations and philological dissections of t h e lexical material retrieved. Milestones along this tempting road were t h e influential monographs b y A. Schulten (1914, 1923), F . Ribezzo (1920, 1927), E . Philipon (1922), W. Meyer-Lübke (1925), and A. Trombetti (1926). After t h e appearance of t h e Origenes t h e crescendo of this output continued unabated, with B . Terracini (1927), V. Bertoldi (1932, 1937), P. Chantraine (1933), and m a n y others joining t h e ranks of discussants and performing with everincreasing zest and self-confidence. 31 Conversely, t h e communication lines between data31 F o r all bibliographic details a n d a string of judicious digests a n d assessments see Craddock, Latin Legacy Versus Substratum Residue, 5 9 - 7 2 , w h o justifiably omits A. Zauner, Altspanisches Elementarbuch (Heidelberg, 1908), §§144-146; (1921 2 ), §§138-140, where t h e relevant formations r e m a i n u n m e n t i o n e d . However, t h e revised version of Z a u n e r ' s g r a m m a r (§25) d r a m a t i z e s a fact v e r y d a m a g i n g t o Menéndez P i d a l ' s final view, because t h e uniqueness of unstressed a in dodging erosion characterizes n o t only (a) uérfano < ORPHANU VS. t h e aggregate of cobdo 'elbow' < CUBITU, liebre ' h a r e ' < L E P O R E , omne ' m a n ' < ΗΟΜI ΝΕ, pueblo 'people, t o w n ' < POPULIT, tredze ' t h i r t e e n ' < T R Ē D E C I ( M ) , a n d yerno 'son-in-law' < G E N E R U , b u t also (b) caramiello ' s a l t w o r t ' < CALAMELLU 'small reed', para(d)iso < PARADĪSU, a n d quaraenta 'forty' < QUADRĀGINTĀ V S . cadnado-candado 'padlock' < CATĒNĀTU lit. 'chained', costumne 'custom, h a b i t ' < C Ō( N)S(U)

ETUDINE χ

- Ū M I N E , pulgar

' t h u m b ' < (DIGITU)

POLLICĀRE

(in

lieu

of

POLLICE),

and

petrol

[orig. peytral] ' b r e a s t b a n d ' [mod. pretal] < PECTORĀLE. Stray addenda to the bibliographic survey: (a) G. Bertoni, " I nomi spagnuoli dei colori del cavallo...", ΞΜΡ, Ι (M., 1925), 151-154, traced Sp. cópano to LL. CAUPILU 'small boat', It. tonfano 'deep pole in a stream' to Gmc. TUJNPHILO, Fr. marne 'marl, chalk, clay' to Gaul. MARGILA, but decided in the case of OSp. ro(d)ano 'reddish' [= Ar. ašqar] in favor of *RĀVID-ĀNUS 'greyish' rather than RUTILUS 'bright red'. (b) F. Krüger, "Die nordwestiberische Volkskultur", WS, Χ (1927), linked Ast. truébano (Munthe) to Berc. trobo, [torβo] 'powerful tree trunk' (with a reference to his own Die Gegenstandskultur Sanabrias..., 1925) and documented tárzano 'part of the hearth', see El léxico rural del Noroeste ibérico, tr. E. Lorenzo y Criado (M., 1947), 90, 100. (c) D. Alonso, "Las rodelas", RFE, XXXIV (1950), 248, described W.-Ast. ápago (Castropol) as part of a rodela or carreta, which is a rural appliance on wheels. (d) G. Alessio, "Problemi di etimologia romanza" (II), examined jointly Sp. caramba 'phallus' and carámbano 'icicle' (RLiR, XVII [1950], 58f.). (e) H. Meier, "Mirages prélatins", RF, LXIV (1952), 1-42, at 10, juxtaposed Gal. sutámbaro and Leon. (Astorga) sotambanado. (f) H. Janner, "Zum Namen Góngora",ASNS, CLXXXIX (1952-53), 28f., traced to Low L. CONCHULA 'small muscle, tiny container' the top. Góngora, Guip. konkorra, as well as C.-Ast. (Cabranes) cuéncano 'cavity in a human or animal body' and Extr. (Mérida) recuéncano 'cavity, hollow space, ravine'. (g) J. Jud examined the relation of -ULA to -ara in his (posthumously published) rev. of M. P. da Silva Pereira, "A nespereira" (Biblos, XXIV); see VR, XIII (1953), 212f. C. Michaëlis' article, w r i t t e n w i t h w a r m t h a n d élan, even a t a high p i t c h of excitement, a t t r a c t e d disappointingly scant a t t e n t i o n (for one voice in t h e wilderness see Matos Sequeira, " P ú c a r o s " , Atlántida, Y e a r I V , Nos. 4 2 - 4 3 , p p . 700-707). It was re-issued as a slender book, with charming iconographic illustrations, but—to the author's avowed regrets— practically no changes in wording (Coimbra U.P., 1921 ; L. : Rev. Ocidente, 19572). The author reverted to the gram matical facet of the problem in her Coimbra lectures (1911-12), made widely available over twenty years after her

434

AFFIXAL DERIVATION

h u n g r y R o m a n i s t s a n d theoretically-oriented general linguists temporarily ceased function.

to

The foreseeable result is visible in Menéndez Pidaľs 1953 paper, where a crushing weight of palaeo-onomastic data is not in the least counterbalanced by any renewed or enhanced concern about the status of a far-flung family of "suffixes" devoid of any isolable meaning, or even any tight gamut of meanings. The lack of rigor in morphological analysis was compounded by a certain laxity in the theoretical foundation of the author's clearly emotional substratophilia. 32 The evidence of cognate Eomance languages was examined with scrupulous care only where the stimulus of a beckoning discovery of common substratum influence was operative (as when, after the mid-'twenties, South Italian dialects again and again arrested Menéndez Pidaľs attention for the sake of the che rished Oscan ingredient).33 Undeniably, there may exist loose and remote causal connections between pristine onomastic proclivities and productive derivational devices. But the two sets of data must not be rashly lumped together, and the threads between them, if any can be detected, are unlikely to assume the shape of straight lines. There exist countless Northern French toponyms in ֊y (Neuilly, Vichy, etc.) and numerous Southern French counterparts in -ac (Cadillac, Cardaillac, death (L., 1946); this time with full knowledge of her friend Menéndez Pidaľs 1905 article, which she enthusiasti cally commended to her auditors and whose terminology ("sufixos átonos") she now unhesitatingly adopted (56-65). In the main, the classroom presentation involved the same examples as those previously offered in the ceramic study; but she added a sprinkling of new data, e.g. cóbr(eg)a 'viper', limách(cg)a 'snail', lônlr(eg)a 'otter', salamântega ~ ֊dra 'salamander'; elaborated on unstressed learnèd suffixes ('ico, '-ido, '-ulo); paid closer attention to epenthetic /j/; and included in her purview "macaronic" poetry and all kinds of deliberate playful distortions of speech, thus stressing buffoonery as the stylistic dimension of the problem. Ün the social implications of stress on the antepenult see also J. da Silveira, "Erros de prosódia: o prestígio dos esdrúxulos", RP, IV (19-44), 33-37. 32 T h e a u t h o r ' s almost doctrinaire c o m m i t m e n t t o t h e s u b s t r a t a i a p p r o a c h is, independently, familiar from such writings of his as t h e n o t e (all told, less t h a n felicitous), "Modo de o b r a r el s u b s t r a t o lingüístico", RFE, X X X I V (1950), 1-8, and, above all, t h e ambitious i n t r o d u c t i o n ("Dos problemas iniciales relativos a los romances hispánicos", p p . xxix-cxxxviii) t o ELH, I . A m a t t e r sometimes overlooked is t h a t , in those same years, he became, irreversibly, a s t a u n c h believer in t h e " e s t a d o l a t e n t e " thesis as regards other manifestations of culture, including straight history of literature ; t h e affinity of this assumption of latency to t h e hypotheses bearing on linguistic s u b s t r a t u m was m a d e crystal-clear b y P . L e Centii in his review article (BH, L X I [1959], 183-214) d e v o t e d to t h e powerful concluding chapters of t h e 1957 revision of Poesía juglaresca (orig. 1924). F o r full appreciation of Menéndez P i d a ľ s final s t a n d on t h e issue of s u b s t r a t a i suffixes it is advisable to t a k e into account such p r o n o u n c e m e n t s of his (at first glance, n o t directly related) as " L o s españoles en la h i s t o r i a " (1947), "Caracteres primordiales de la l i t e r a t u r a e s p a ñ o l a " (1949), " T r a d i c i o n a l i d a d en la épica española" (1950) all t h e w a y d o w n t o one of his last a n d b r o a d e s t v e r d i c t s : " E l estado l a t e n t e en la v i d a t r a d i c i o n a l " , Rev. de 0cc2, I I (1963), 129-153. 33 I n t h e mid-'twenties Menéndez P i d a ľ s curiosity a b o u t Italian a n d Italic models or parallels ranged over b o t h select Osco-Umbrian features a n d t r a i t s mediating between provincial L a t i n a n d early R o m a n c e (Origenes 1 , 303-305, 306f., 527f.). This fine balance was later impaired in favor of O s c o - U m b r i a n ; see t h e twin studies, "Pasiegos y vaqueiros; dos cuestiones de geografía lingüística", Arch, I V (1954), 7-44, a n d " A proposito de -ll- y l- l a t i n a ; colonización suditálica en E s p a ñ a " , BRAE, X L I I I (1954), 165-216. Outside t h e m a s t e r ' s direct sphere of influence in and a r o u n d Madrid, t h e younger workers t h r o u g h o u t those same years were gradually a d o p t i n g a more skeptical a t t i t u d e t o w a r d t h e m a x i m a l i s t claims of s u b s t r a t u m militants, a t r e n d of opinion shown by F . H . J u n g e m a n n , La teoría del sustrato y los dialectos hispano-romances y gascones (M., 1955; Columbia diss., 1952); E . Coseriu, El llamado "latín vulgar" y las primeras diferencias romances; breve introducción a la lingüística romance (Montevideo, 1954), 160-162, a n d , j u s t a few years later, various articles b y Curtis Blaylock. T h e orthodox followers (e.g., R . L a p e s a as t h e a u t h o r of n u m e r o u s successive editions of Historia de la lengua española) persevered.


435

Cognac...), but neither the -y nor the -oc items influenced decisively the two systems of Gallo-Romance derivational suffixes in their evolution away from Latin; still less have they been suspected of interference with any isolated features of the general sound development. 8. Some Currently Conflicting Trends of Thought. While, under present condi tions, it would be surprising to discover consensus on any potentially contro versial problem of linguistics, one clearly discerns, in this particular province of knowledge, two voices in the chorus of opinions.34 On the one hand, J. Hub schmid, as the undaunted spokesman for the substratist cause, has painstakingly reconstructed a great many obscure word-histories, using every imaginable scrap of information, and while the lexical biographies so meticulously (if at times, far too daringly) pieced together retain their inherent value, the semantic ambit of the words involved — typically, designations of features of the ter rain — is hopelessly far removed from the world of lively, humorous derivatives (referring, more often than not, to small rodents, reptiles, fish, insects, etc., also to toys) that one has come to associate, since the early years of this century, with the "sufijos átonos". 35 The opposite pole of genetic interpretation is represented by a long chain of, almost invariably, polemic, flamboyantly-styled articles from the pen of H. Meier,36 who prefers to start out from Lat. -ULU, -ULA — preponderantly, if not exclusively, a diminutive suffix. Now it is a fact that a certain stringed instrument with which Greek musicians acquainted the Romans ('lute, lyre') was accepted either in its near-etymological form, as CITHARA, -ERA (witness OSp. cedra), or in its adjusted, Latinized form (perhaps with diminutive or hypocoristic overtones), as CIT(H)ULA (note OSp. çitola) — quite apart from intentionally and exquisitely learned citara 'cithara, cither, zithern' 37 — and this neat two- or even three-way split need not have constituted an isolated case. But a thorough study of this type of relationship can become rewarding only against the backdrop of a methodic inquiry into the general vicissitudes of -ULU, 34 Craddock, Latin Legacy..., pp. 73-83, qualifies as a guide not only to several key studies by the chief protagonists of the drama, J. Hubschmid and H. Meier, but also to the thinking of J. Corominas, as can be extrapolated from individual entries in his DCE (1954-57), and to various writings by V. Bertoldi, G. Rohlfs, and Menéndez Pidal (the last-mentioned in collabora tion with A. Tovar). 35 Typical objects of Hubschmid's curiosity are such words as senara 'land assigned for cultiva tion in lieu of wages' (which he studied monographically in 1951), gandara 'low wasteland', *lámara 'rubble, boulders', Gal.-Ptg. támaro 'land-slide, cave-in'; Sard. (Campid.) tséppara 'rocky plain'. The usefulness of this sort of material for research in toponymy is, of course, incontrovertible. 36 Perhaps the best-remembered of H. Meier's articles so biased is "Mirages prélatins" [a title chosen in pretentious imitation of Gilliéron's trail-blazing piece], RF, LXIV (1952), 1-42, esp. 6-9. The major stumbling-block has been not the author's broad-gauged assumption of the superiority of Latin bases, but his utter ineptitude both as theorist and as etymologist in testing and implementing it. 37 This classic instance has long figured in brief statements; cf. J. J. Nunes, "Convergentes e divergentes", BSC, Χ (1915-16), 840: Ptg. citola ~ citara; J. . De Forest, "Old French Bor rowed Words in the Old Spanish...", RR, VII (1916), 369-413, at 387.

436

AFFIXAL DERIVATION

-ULA as recorded in texts, or as safely reconstructed ; and, amid all the clamor, such a vitally-needed large-scale investigation has not yet been undertaken by the advocates of antisubstratism. 38 Only the haziest and hastiest outline of such a highly desirable monograph can here be supplied. 9. Late Survival of -1u, -ulu? For the sake of operational tidiness, it seems advisable to cut out, from the start, three disturbing shreds of material: (a) words washed ashore through cultural diffusion, i.e. borrowings, as when Ptg. ilka 'island', in contrast to OPtg. ins-oa, -ua, turns out to be a fairly late Catalanism ;39 (b) words more or less wilfully distorted through superstition ('taboo'), e.g. certain names of the 'devil' (from Gr.-Lat. DIABOLU): Sp. diache, diantre, Ptg. diacho, etc.; 40 (c) words in which, for any reason, l, the consonantal pillar of the erstwhile suffix, was violently displaced, entering into some segment of the root morpheme, e.g. Ptg. brinco 'ear-ring' < VINCULU (presumably via *blinco). The first important decision to be reached involves a point of chronology : How accurate is the widely-accepted assumption of the early extinction of -ULU, -ULA, entailing, often, but not to the point of predictability, their replace ment by -ELLU, -ELLA — particularly in the ranks of diminutives ? Even random consultation of scattered etymological literature discloses the plausibility — on the authority of trustworthy scholars — of such reconstructions (in addition t o afore mentioned *TENTTLA 'snare', from TENĒRE 'to hold') as *BUXULA, *CAECULA, *IUXTULA,

*PETRULA, *SUBTULUS, hidden in various branches of Romance; 4 1 and this impression of 38 There exists, of course, a corpus of writings on -ULU, including an antiquated Uppsala dis sertation overtly tilted in the direction of Romance (Å. L. Blomgren, 1913); R. Hakamies' Helsinki monograph (1951) on the rise and growth of diminutive suffixes, an inquiry which also straddles Latin and Romance ; and countless shorter or incidental references. To the bibliographic clues assembled in Lg., XXVIII (1952), 303-4nl7 add, on the I-Ε side, W. Schulze's remark (337) in the Jagić FS and G. Bottiglioni's contribution ("Accento, anaptissi e sincope vocalica nel l'antico italico") in FS F. Krüger, I (Mendoza, 1952), 1-7 ; and, on the Romance side, L. Spitzer's fleeting comment (189) in Spitzer and Gamillscheg, Studien zur romanischen Wortbildung, Bibl. dell' AR, II:2 (Genève, 1921). 39 Insua, insula, ilha were recognized as triplets by [F.] A. Coelho, "Formes divergentes de mots portugais", Rom. I I (1873), 281-294, at 290, but the early analyses of the reflexes of INSULA were crude, cf. . von Reinhardstoettner, Grammatik... (Strassburg & London, 1878), p. 91. Ilha, its abnormality now diagnosed as a consequence of its borrowing from Catalan (L. F. Lindley Cintra in BF, X I I [1951], 196), appears as early as Castanheda's História da Índia, Bk. II (cf. J. Guimarães Daupiás, LP, I [1930-31], 55), while the native form, insoa, prevailed in medieval texts, e.g. the Arthurian Demanda (cf. M. Rodrigues Lapa in LP, I, 266n3); the erratic survival of -NS-, as if to compensate for the loss of -L-, reminds one of the history of OSp. feo/hedo 'ugly' < FOEDU. See also Lg., XXVIII (1952), 313n74. Spanish has wavered between vernacular isla and learnèd insula (Don Quijote, Part I I , Chap. XLVII = ed. Bonilla & Schevill, IV, 97, lOOf.). 40 However, for derivational purposes only diabl- seems to have lent services in Spanish: diabl-ear 'to play pranks', -esa 'she-devil', -esco 'devilish', -illo 'imp', -ura 'bedevilment, deviltry', etc. In Portuguese the free forms (di-abo, -acho) contrast with the bound form diabr-: -ete 'imp', -ura. Earlier analyses were often infelicitous (e.g., Coelho, Rom., II, 289f.). 41 Thus, to account for It. bus-, bos-sola 'compass', bossolo (a) 'small box', 'powder-box', (b) 'box-tree' ( = bosso), W. von Wartburg posits the type *BUXULA, a putative satellite of BUXA which itself represents a blend of two Hellenisms in Romance : PYXIS -IDIS 'box' and BUX-US or -U 'box-tree', an adaptation of πύξος; see 'Die griechische Kolonisation in Südgallièn...", ΖRPh, LXVIII (1952), 7. According to Ernout-Meillet, tree and dendronym, in all likelihood, are traceable to Asia Minor. J. Hubschmid, from Central Raeto-Romance material, infers *CAECULA


437

exuberant growth is reinforced by the parallel discovery of numerous equally solid inferences of verbs in *-???LĀRE — among those posited or reconfirmed not too long ago let me cite the triad *BRAGULĀRE, *FRAGULĀRE, *TRAGUJLĀRE ; also *ROTULĀRE (by now, a celebrated case

of semantico-pictorial diversification) and *SUB-MICULĀRE — the prototype for the entire series being ŪSTULĀRE (in Catullus) 'to burn, scorch, singe' (p.ptc. AMBŪSTULĀTUS as early as Plautus), alongside the primitive ŪRERE, p.ptc. ŪSTUS. 4 2

These, I repeat, are just a few samples of sustained productivity, and it would be unrealistic to argue that all of them involve archaic formations consistently shunned by Roman writers of all persuasions and almost miracu lously salvaged in the vernaculars. Why not champion the far more convincing alternative hypothesis of protracted fertility of -ULU, -ULA in colloquial Latin ? Beyond the shadow of a doubt, the spontaneous phonological trend was for theUin -???LU, -???LA to be syncopated at a distinctly early stage — in practically all Romance languages (though not everywhere or in every environment on the same scale), which means, in the last analysis, on the level of spoken Late Latin. This situation carried with it the risk of the formation — and, in some particularly hazardous phonotactic contexts, t h e radical resolution — of novel post-tonic consonant clusters, witness the well-known cases involving velar + L, such as ARTICULU 'small joint', AURICULA '(little, cute) ear', MACULA 'mark, stain, mesh', OCULU 'eye(let)', OVICULA '(young) sheep', ROTULU 'roll, cylinder', SPECULU 'mirror', TĒGULA 'roof-tile', VERMICULU 'grub, little worm'. 4 3 Partial similarity to the well-represented ranks of words in -CULU, -GULU is offered, in Hispano-Romance, b y TRĪBULU 'threshing machine', on the one hand, and, chiefly, in Italian, by such evolutionary lines as culla 'cradle' < CŪNULA and spilla 'brooch, tie-pin' < SPĪNULA, lit. 'little thorn', on the other. 4 4 'intermittently flowing stream' (lit. 'the small blind one'); see "Zur Charakteristik des Fassatals mit Ausblicken auf andere ladinische Mundarten", ZRPh, LXVI (1950), 338-350. On *IUXTULA 'strap attached to the yoke' (a cross of IUNGULA [IUNGLA : CGL, II, 94.5] and *IUXTĀRE) see P. Gardette, "De quelques mots francoprovençaux", Melanges de philologie romane... K. Michaëlsson (Göteborg, 1952), pp. 166-172, at 167: joucle. On Moz. Pétrola (Albacete), OSp. Piédrola (Álava, A.D. 1085; Ciudad Real) see Menéndez Pidal, Origenes2, §§32:3e and 61 bis; also, Topo nimia prerrománica, pp. 55, 64, 68. 42 On vestiges of ŪSTULĀRE in NAV toponymy: Gal. Ostulata, A.D. 818; Ucha(da), Vil- (La Coruña), Vel-ouchada (Lugo) beside Sp. Villoslada, ORioj. Villa U slada see J. M. Piel in RF, LXIV (1952), 242f. On the entire complex of Fr. frôler ~ frailler; trau-, tro-, treu-ler beside trou(il)ler; brailler as against brô-, brau-ler and brouiller see G. Tilander, "Origine et évolution de frôler...", ZRPh, LXVII (1951), 174-178. G. Rohlfs, "Über Blitz und Wetterleuchten", ZRPh, LXVIII (1952), 299, traces Lomb. (Bergam.) sümelegá 'to flash, lighten' to *SUBMICULĀRE. *ROTULĀRE 'to roll' has long intrigued scholars. Earlier attempts at the interpretation of pre sumed reflexes (e.g., J. J. Nunes in BSC, X, 831) have been superseded, so far as HispanoRomance is concerned, by Menéndez Pidal, "Notas para el léxico romanico", RFE, VII (1920), 16-19 and, especially, by D. Alonso, "Representantes no sincopados de *ROTULĀRE", RFE, XXVII (1943), 153-180, cf. . Pottier in Rom., LXXIII (1952), 276. 43 The cases of Sp. artejo 'knuckle', espejo, ojo, oreja, oveja, teja, vermejo (mod. b-), and of their congeners are too trivial to hold our attention. Diez and his successors, all the way down to J. J. Nunes ("Reacção literaria na língua", LP, I [1930-31], 10), traced Sp. vencejo, Ptg. vencelho 'band, string' to a similar formation, but Meyer-Lübke later rightly preferred an alternative in *-ĪLIA (REW 3 9339; a more accurate statement would have been: *-ILIA/*-ĬLIA). Students of doublets and triplets have contrasted, in traditional fashion, Ptg. malha ~ mancha ~ mágoa, rolha ~ róttila (Coelho, Rom., II, 283, 290), artelho ~ artigo ~ artículo, rollio ~ rótulo beside rol{o), r odela (do joelho) ~ rótula (Nunes, "Convergentes e divergentes", pp. 822, 838). 44 On It. spilla and Fr. épingle (in lieu of ideal *épindle ; but the clusters -nl- and -ndl- were unavailable for this purpose) and on It. culla see H, Lausberg, RF, LXIV (1952), 157, in criticism

438

AFFIXAL DERIVATION

I t has long been known t h a t some of the resultant secondary clusters, e.g. -SL-, turned out to be misalliances ; while a minority of speakers of Old Spanish accepted maslo 'male' < MASCULU, ASSULA 'chip, shaving' went through a long, tortured process of adjustments and compromises, including tne experimental insertion of a buffer consonant : ASCLA, *ASTLA and the blend (Schuchardt's "innere Formenmischung") ASTULA; 4 5 the history of PESSULU 'bolt' is entirely similar. More interesting is the fact t h a t the widespread substitution of -ELLU for -ULU and of -ELLA for -ULA, as visible in the history of FIBULA 'buckle, brooch, clasp, iron clamp' ( > OSp. fiviella, mod. hebilla), can also be concomitantly interpreted, a t least in its final stage, as a tentative evasion of a phonological embarrassment. 4 6

With our thoughts riveted to the strong possibility that the suffix -ULU, -ULA may for a while have maintained its momentum as an identifiable, even a moderately productive, morpheme even though the tendential syncope of ??? had already started to erode numerous words into which it entered, the speakers' eagerness to save the original syllabic contour of many endangered formations becomes readily understandable. We are simply once more facing the pheno menon of morphological (and syllabico-accentual) resistance to sound change. The series of characteristic examples stands out with particular neatness in Old Portu guese, and the feminine group in -oa < -ULA is silhouetted more tidily than the group of masculine counterparts in -0(0) < -???LU, where vocalic contraction threatened to, and often did, produce all sorts of secondary alterations : amêndoa < Gr.-Lat. AMYG-, AMYN-DULA 'almond', artigoo < ARTICULU 'article', bagoo < BACULU 'staff, walking-stick', cabidoo < CAPITULU 'chapter', diaboo < Gr.-Lat. DIABOLU 'devil', espadoa < SPATHULA 'shoulderblade', lit. 'small two-edged sword', encreo < INCRĒDULU 'incredulous', lândoa < GLAN DULA 'small gland', mágoa (var. mágna) 'pain, sadness', 'bruise, black-and-blue mark' < MACULA, mámoa < MAMMULA 'breast, nipple', névoa < NEBULA 'fog, cloud', parávoa 'word' < Gr.-Lat. PARABOLA (assimilated to -ULA) 'comparison, anecdotal account', perigoo < PERĪCULU 'danger, hazard', 'trial, proof, test', povoo < POPULU 'people' (cf. topon. Ρόνοα), régua 'straight edge' < RĒGULA 'ruler, plank' (as against telha 'tile' < TĒGULA), tâboa < TABULA 'board', tomboo < TUMULU 'mound of earth, hill(ock)'. The parallel series in (Old) Spanish shows a far less sharp-edged profile, as a result of syncope, intercalation of buffer consonants, metathesis, and other secondary interferences: almendra, cabildo, diablo, espalda, mamhla, iebla, palabra, periglo > peligro, pueblo, regla, tabla.

These words have long been known to scholarship,47 but any successful of the Schwan-Behrens manual and of Bally, Linguistique générale et linguistique française3, §163. 45 ASTULA alone explains the further adoption of *ASTELLA>Sp. astilla, through routine suffix change. The authenticity of the alternative midway solution, ASCLA, is supported by plentiful Romance evidence. (In the REW3, Nos. 736 and 740 are mutually illuminative and should there fore have been consolidated into a single entry.) In this sector of the domain one is tempted to place the difficult Romance phase of the history of VETULU 'old and decrepit' ( = Fr. 'vieillot'), which may involve transition of -T'L- to -C'L-. Cf. also VITULU 'calf' > Sard, vikru, viglu, vićću, as examined by Rohlfs in ZRPh LXVIII (1952), 296. 46 Granted that NĀUICULA (NAUCULA) and NĀUICELLA (NAUCELLA) 'boat' and the like coexisted in Latin, the spread of the substitution at issue could have been greatly accelerated by the diffi culties the speakers encountered with the pronunciation of newly-emerging consonant clusters. 47 From the immense literature available on these words let me cite just a few highlights. M. L. Wagner, in reviewing (1953) Lino Netto's Minhoto monograph (1947), clearly re-defined the status of amêndoa (which had been independently investigated by P. Aebischer in EDMP) and glândoa, classing dial. landra 'acorn' with land(r)e (pp. 329f.). On artigo, bago, diabo, perigo, etc. see, in addition to the standard historical grammars, Nunes, "Reacção...", p. 10, who, however,


439

attack on them has been blocked by the preconceived idea that syncope was obviated in the West and critically delayed in the Center through the semilearned status of the words. While a few of the items involved, admittedly, do have an ecclesiastic ring and sometimes show additional symptomatic sound features (artigoo, cabidoo, diaboo, paravoa), the majority seem to be completely free from this semantic or cultural connotation and in other phonetic respects behave in thoroughly normal manner (observe, in particular, lândoa, mágoa, mámoa, névoa, régua; also tomboo with its folksy substitution of -mb- for -m-, on the analogy of amos ~ ambos, etc.). By agreeing to consider as the prime mover a powerful morphological rebellion against a phonological tide, we place this entire development in the mainstream of events, withdrawing it from the isolated niche of "semicultismos". Once adequate allowance has been made for the long survival of '-(o)lthrough joint learnèd (i.e., broadly cultural) and morphological (i.e., lexical and grammatical) resistance to the erosion of the proparoxytonic words' weakest syllable, it becomes apparent that under special circumstances the l was apt to yield to r. (There is no need to document the frequency of interchanges between the two liquids in Romance.) In a language like Portuguese, generally known to favor — both initially and medially — such clusters as br, cr, fr, etc. at the expense of , cl, fl, it is doubtless possible to argue t h a t diabro 'devil' represents a lame adaptation of Sp. Leon. diablo. Stretching facts, one could incline to attribute to late diffusion of typically "Central" forms the rise of such "Western" counterparts as OPtg. estabro ( > estrabo) 'stable' < STABULU 'standing-room, quarters' and pobrar 'to populate', despobrar 'to depopulate', pobra 'settlement, colonization' (cf. Sp. establo, poblar/despoblar, pueblo), an assumption which, among other advantages, would account for the coexistence of native diaboo, povoar/póvoa and imported diabro and pobra.48 Régua could similarly be pitted against regra 'rule, order' (Sp. regla beside renglón). does not cogently account for the contrast diabo(o): diabro; on bago vs. báculo see Id., "Con vergentes...", p. 833, where the pair diabo-.diabro is again barely listed (828) and where [obs.] távua is contrasted with tábua 'board, plank' (831; dictionaries also record tábula, távola 'gaming table, round piece used in playing checkers') and (obs.) encreo [ = mod. increu] with incrédulo (835). Cabido is contrasted with capitulo by Coelho, Rom., II, 284, 286; and by Nunes, "Conver gentes...", p. 843; J. Leite de Vasconcelos, Lições de Filologia Portuguesa2, p. 91, traces it to Church Latin. Coelho records espádoa, táboa, and other items pertaining to this series in his oft-quoted study of doublets (285-287), with few if any helpful comments. OGal. parávoa occurs in the Alfonsine Cantigas (65.34b ; see E. K. Neuvonen in BF, XII [1951], 318), a situation which casts suspicion on the indigenous status of Ptg. palavra (a Castilianism or a Leonesism?). It is equally possible that E.-Ast. (Colunga) cabildu 'porch or portico of a parochial church' (B. Vigón) has been imported from, or influenced by, Castilian, even though Sp. cabildo ordinarily means 'cathedral chapter', 'municipal council', 'council room', 'town hall'. On Sp. mambla 'funerary hill', Gal.-Ptg. mámoa 'hill' < MAMMULA 'nipple, tit' (> 'female breast') see W. Meyer-Lübke, "Neubenennungen von Körperteilen im Romanischen", WS, XII (1929), 3, and RÉW3 52776 (with mention of Cat. mambla 'hill' and B. mam-ola, -ul 'fleshy part' and with a reference to a toponymically-slanted note by J. Leite de Vasconcelos). The secondary complications produced by syncope are not, of course, uniquely peculiar to Castilian; E. Gamillscheg, "Germanisches im Französischen", FS F. Krüger, I (Mendoza, 1952), 26, examines the triad of esp-aude ~ -alle ~ -aule from SPATHULA via *espadle. 48 On povoar, (des)pobrar, etc. see Nunes, "Convergentes...", pp. 828, 830, who also adduces poborar — patently, a compromise form. The same author cites (830) OPtg. cobra, copia < CŌPULA 'link, leash', a term of medieval versification (cf. Fr. couplet) which, on balance, might be a

440

AFFIXAL DERIVATION

Other cases do not quite so easily lend themselves to such explanation. Ptg. sa(i)bro 'sand' < SABULU 'gravel' cannot very well be laid at the door of Castilian pressure, since the Spanish word for 'sand' happens to be a descendant of ARĒNA. One readily understands the interplay of forces t h a t could have generated *LAMULA as a by-form of LAMELLA 'thin metal plate'; but, to account for Gasc, lámbros 'éboulis' (i.e., 'debris, fallen rock'), Hubschmid reconstructs *LAMURA, implicating consonant dissimilation. Then there are on record such clusters of variants as Ptg. combro ~ cômoro 'mound, knoll, hummock' < CUMULU 'heap, pile, mass' and, if Nunes' conjecture is correct, the reflexes of TUMULU 'sepulchral mound': tombo(o), tômboro, (pop.) tumblo beside erudite túmulo; all this without taking into account unusual channels of transmission such as Albanian, where GRŪMULUS 'small clod, lump, or eleva tion' (REW 3 3887), MASCULU 'male', RĒGULA 'plank, ruler', and SACCULU 'small bag' yield, to be sure, grumull, maškull, regull, and šakull, respectively, but GLANDULA lit. 'small acorn', apparently through dissimilation of the laterals, produces and such as Sp. tafurea 'ship for horses', which may go back to a satellite of TABULA via Ar. taifurîja.49

In themselves, these isolated instances of root-morpheme-final -r- as a residuum of -UL- were too few to harden into a pattern, and, more important, there is no concomitancy of a to an immediately following r yet in sight ; in conjunction with another, superior force, formations of this kind could very well have played a distinctive rôle in producing Hisp.-Kom.'-aro,'-ara. Certain evolutionary side-lines, e.g., the development of syncopated -C(U)L- immediately after a consonant (typically, L, N, or s), require no detailed discussion, since they seem not to be implicated in the genesis and growth of nominal augments, as here visualized. 50 Provençaiism rather than a Castilianism. OGal.-Ptg. esta-bro, -vro occurs in the Alfonsine Cantigas and in the Arthurian Demanda, while the authors of treatises on surgery and falconry preferred estrabo; the derivative estrabeiro 'groom' was eventually influenced by estribo 'stirrup', hence estribeiro, and such a servant's position was described as estrebaria. See J. M. Piel, "Etimologias portuguesas" [I], Biblos, XX (1944), 121-130, No. 8 — more explicit than Nunes, "Convergentes ...", p. 836. 49 On sa(i)bro see Piel, "Apostilas de etimologia e lexicologia portuguesa" (I), RPF, I (1947), 448-462; on lámbros, J. Hubschmid, "Ibero-romanische Wörter für 'Steinplatte'", RF, LXIV (1952), 43-56, at 46; on combro/cômboro, Coelho, Rom., II, 282, 291; on tombo(o), e t c . — n o t identical, I assume, with tombo 'archive, title deeds, register of charters' < Gr.-Lat. U 'slice, roll of paper, book' (Gr. Nunes, "Convergentes...", p. 839; on tafurea, H. and R. Kahane, "El término mediterráneo...", EDMP, I (1950), 75-89. The preëxistent relationship between -r, -r- and (dimin.) -ll-, in words both Latin (PUER ~ PUELLA) and Graeco-Latin (AMP(H)ORA 'two-handled jar' ~ AMPULLA), may have served as a catalyst. 50 The characteristic outcome is the affricate /č/ — a result from which it is difficult to infer whether the immediate prototype ended in -CULU or -TULU (thus, for 'hammer' one finds in the parent language MARCULUS, from MARCUS, alongside MART-ULUS, which in turn gave rise to a byform in -ELLUS, see Ernout-Meillet, DÉLL4, 387α, with a reference to an arresting conjecture by M. Niedermann; cf. Sp. macho 'sledge hammer' beside martillo 'hammer'). The first step in the development of MASCULU 'male', namely MASCLU, is recorded within the confines of Latin; it led smoothly to Cat. mascle (and to Fr. male), also to It. maschio, whereas OSp. maslo (rare) ~ macho (common) exemplify a bifurcation (different degrees of learnedness?). The equations Ptg. fruncho 'boil' < F(U)RUNCULU (Nunes, "Convergentes...", p. 836) and mocho 'hornless' (cf. Sp. mocho 'blunt, flat, shorn') < *MURCULU (an assumed dim. of MURCUS 'id.', reminiscent of CURTUS 'short') make better sense, at least phonologically, than does the derivation of the latter item from MUTILU 'maimed', as championed by Meyer-Lübke (REW3 5791; note his own misgivings and afterthoughts); [*]MURCULU was favored by P. Barbier, DR, IV, 68, and, again, by Nunes ("Convergentes...", p. 819). Relevant are further: bochecha 'cheek', from bocha < BUCC(U)LA '(little) mouth' (cf. AURICULA, OCULU, GENI- ~ GENU-CULU), as explained, after D'Ovidio, by Piel, "Etimologias portuguesas" [I], Biblos, XX, 126f., with a reference to ventrecha 'slice of fish contiguous to the head' < *VENTRISCULA (cf. REW3 9210: *VENTRISCA > It. Cat. Sp. ventresca,


441

10. A Side-Glance at Tuscan. In our search for the source of the atonic penult vowel a so highly characteristic of the Hispanic nominal augments, we may wend our way back to a point where a young Menéndez Pidal stood seventy years ago, perhaps without fully grasping the implications of his minor discovery. He did come across, and faithfully record, It. cofano 'casket' vs. COPHINU and It. pampano [~ ino] Vine-leaf' vs. PAMPINU as cognates of Sp. cuévano, pámpano sharing the uncommon shift from ï to a, but failed to follow up the lead —speci fically, to inquire into the relation of a to n in Tuscan and to probe, beyond a Romanist's narrower horizon, into the status of such words as COPHINU, PAM PINU within the Latin lexicon. Because Tuscan shows certain relations and developments far more graphically and even dramatically than does any known variety of Hispano-Romance, a return to the forgotten starting point becomes imperative. 51 On the Italian scene, one fact has been familiar since the days of N. Caix : The choice of the vowel (a, e, i, or ) in the post-tonic penult is, by and large, controlled by the following consonant — although for each development there exist specific models in Antiquity. Thus, a is preserved and allowed to spread before (a) nasals (starting point : orfano < ORPHANU ; extensions : abrotano 'southern wood' < ABROTONU, Gerolamo 'Jerome' < (H)IERONYMU); (b) velar stops (starting point: monaco 'monk' < MONACHU; extensions: cronaca < CHRONICA, sindaco 'mayor, auditor' < SYNDICU, folaga 'coot' < FIJLICA); and (c) labials (starting point: calamo 'reed, pen' < CALAMU, a case which can also be classed under nasals ; extension : [obs.] celabro 'brain' < CEREBRU, with concurrent consonant dis similation, and orafo 'goldsmith' < AURIFE[X]). In Florentine, a similar status is accorded to e before r as regards both preservation: canchero 'nuisance' beside cancro 'cancer' < CANC(E)RU, cenere 'ashes' < CINERE, and extension: gambero 'lobster' < CAMMARU, Ga spero < CASPARU, Lazzero < LAZARIT, but in other varieties of Tuscan and, a fortiori, in other dialects, as indeed in the vast majority of cognate languages, the generalization of a rather than of e is to be expected in this position, so that in Italo-Romance as a whole a prevails before n, m; c, g; b, f; and r — an enormous range of environments. Then again,  is favored by the adjacency of l, thus : tavola 'board, plank' < TABULA leads to (obs.) cembolo—dislodged by more "learnèd" -alo— < CYMBALU 'tambourine', pl. 'cymbals', mando(r)la 'almond' < AMY(G)DALA (p1.). The residual unmarked vowel i is fitted into all other contexts: carice 'sedge' < CARICE, cespite 'tuft' < CAESPITE (cf. Sp. césped 'lawn').

Once we compare the Italian and the Hispano-Romance evidence, we notice, OFr. ventresche) and OPtg. esfachado, Braz.-Ptg. fachear 'to fish at night by torchlight' < *EXFASCULĀRE, based on a cross of FACULA 'little torch' and FASCE 'bundle', see Piel, "Etimologias portuguesas" (II), Biblos, XXI (1945), 489-499, at 493f. The classic case of differentiation is offered by MACULA (Ptg. mágoa, see above) vs. *MACELLA (with a nasal insert : OSp. manziella > mod. mancilla; without the insert: Ptg. mazela 'sore spot', desmazelado 'careless' —> Sp. desmazalado 'hapless', semantically deflected under the influence of the Hebrew word for 'star, luck') vs. *MA(N)CULA (> Sp. Ptg. mancha 'blot, blemish'). See my earlier analysis in UCPL, I:7 (1947), 227-243, 269-282, and HR, XV (1947), 272-301. The descendants of Gr. MÁKELLA/ MAKÉLĒ 'hoe used in the vineyard' stand, of course, apart (see W. von Wartburg, in ZRPh. LXVIII [1952], 17). 51 I am here summarizing — with consistent omission of bibliographic detail — certain sec tions of my article "Etiological Studies in Romance Diachronic Phonology", which is to appear in ALH.

442

AFFIXAL DERIVATION

especially in Spanish — as so often — a more sweeping generalization of t h e common trend : If in Tuscan, particularly outside t h e city dialect of Florence, the a has already moved into MOST of the available positions, Sp. a has practi cally overrun them ALL. Next, we observe t h a t t h e affinity of  t o l on Italian ground is clearly due t o t h e nucleus of the diminutives traceable t o -ŬLU, -ŬLA ; in Spanish this thin border-line was no doubt first blurred, then resolutely erased — in particular, one finds no such transfer of t h e  as in I t . fievole 'weak' < FLĒBILE 'lamentable, wretched', pesolo 'hanging, dangling' < PĒ(N)SILE, prezzemolo 'parsley' < PETROSELINU, (obs.) torbolo 'muddy' < TURBIDU, (obs.) idolo 'useful, profitable' < ŪTILE. Another major difference between speech conditions in t h e two Peninsulas consists in this : While both Italian and Hispano-Romance remodel or sharpen the concluding segments of certain words (It. cronaca and sindaco closely parallel Sp. sólano and Ptg. Cristóvão, while abròtano/abrótano, cofano/cuévano, and pámpano show complete coincidence in t h e innovative t r e a t m e n t of t h e tail segment), Hispano-Romance transcends Italian b y a n astoundingly wide margin in often using word-final segments of such wide appeal as augments or appendages rather t h a n mere substitutes, as in murciego > murciégalo (later murcielago, etc.). This bold extension of t h e originally modest scope of -ano, -aro, -ago, etc. constitutes t h e decisive stage in t h e crystallization of t h e uniquely Hispanic "sufijos átonos". One all-important point insufficiently stressed even b y otherwise wellinformed Italianists is t h a t a very high percentage, if not t h e majority, of t h e formations involved are transparent Hellenisms of old stock (and humble cultural status): Of t h e words so far adduced ABROTONU, AMY(G)DALA, CALAMU,

CAMMARU, CANCERE or CANCRU (indirectly; Ernout-Meillet: " . . . a pris tous les sens d u gr. καρκίνος"),

CASPARU, CHRONICA, COPHINU, CYMBALU, ( H ) I E R O N Y M U ,

LAZARU, MONACHU, ORPHANU, PETROSELINU, SYNDICU are downright Hellenisms,

though it is of course true t h a t behind t h e Greek intruder there m a y lurk a still older Orientalism, as in t h e case of LAZARUS. A certain fluidity of t h e vowel, as regards stress a n d quality, m a y have started in t h e Graeco-Latin sector of t h e lexicon what through systemically superimposed stress shifts a n d imagina tive association with t h e indigenous diminutive suffix -ŬLU, -ŬLA. Both factors came together in άμνγδάλη > AMIDDULA ~ AMY(G)DALA 'almond', depending on t h e socio-educational level of the given speakers or the "realism" of the writer involved. I t is, on balance, in this sector t h a t t h e A in this kind of weaklystressed syllable became a sharply delineated feature (very much in contrast t o what was happening in t h e autochthonous hard core of t h e Latin lexicon 52 ) 52 Significantly, in native formations the reverse trend predominated : a, at least under certain circumstances, gravitated toward i. Thus, not ALACRE 'quick, lively, animated', but *ALICRE, underlies Sp. Ptg. alegre, It. allegro 'mirthful' and the corresponding verbs and adjectival abs tracts. ANAS -ATIS 'duck' (Varro) did survive in Hispano-Romance (Sp. ánode beside pato ; per haps also in Low-G. Ante), but apophonic ANITIS, etc. is on record (Plautus, Cicero), the diminu-


443

and that the consonants g, l, m, n, r became conspicuously prominent as the recurrent last pillars supporting the tri- or poly-syllabic structure of the given word. Semantically, many of these lexical items referred to plants, berries, fruits, vegetables, tools, containers, small animals (including seafood), much less frequently to persons, practically never to features of the landscape (configura tion of the terrain, etc.). Consequently, the fashion for terminal word segments to all intents and purposes identical with the "sufijos átonos" may very well have started with the Greeks in Rome — artisans, gardeners, fisherfolk, chefs — and the specific point of impact, as is beautifully illustrated by the "conditioned sound development" of Tuscan, may have been at the intersection of (a) ''adjusted" Hellenisms and (b) native diminutives in -ULU, -ULA. 11. Six Interweaving Strands. With this new slice of knowledge at our command, we may more confidently revert to the set of six possibly interrelated problems identified at the outset. The first, the problem of NOMINAL AUGMENTS, need not be rehashed, except for remarking that there exist defensible alternatives to the patterns of descrip tive presentation advocated by Menéndez Pidal. 53 The second, the tendency toward "ESDRUJULISMO", clamors for more search ing and broader-gauged treatment than the sketchy analysis offered by A. Alonso shortly after the start of his career.54 As a facetious, marginal device practiced on a small scale, shift of stress pattern exists, for instance, also in tives end in -ATICULA and -ITICULA, respectively; the Italian forms presuppose *ANITRA; and several Romance languages have completely veered away from the tradition (Fr. canard, etc.), conceivably in recoil from such exuberance of wavering. 53 One can imagine a reorganization along, e.g., the following lines, in the order of increasing complexity : (a) simple relationship of "primitive" to "derivative", without gender switch: bast-o (~ -ón) 'cane, staff' ~ vást-ago 'twig, sapling', 'rod, stem', 'scion'; buzo 'diver' ~ (dial.) búz-ano 'id.'; (OSp.) relamp-o '(flash of) lightning' ~ relámpago 'id.'; (b) simple relationship, as above, with additional feature of gender switch: cien-o 'mud, slime, silt' ~ cién-aga 'marsh, moor, mudhole'; pezpit-a 'wagtail' ~ pezpit-alo 'id.'; raz-α (OSp. raça) 'crack, slit, ray (of light), spoke (of a wheel), light stripe of a fabric' ~ ráz-ago 'burlap, sackcloth' ( = 'tejido ralo y basto'); (c) all manner of intricate relationships : borde 'wild, bastard, uncultivated' ~ burdé-gano 'hinny' (via *bord-, burd-iego?); cerner 'to sift, scan' ~ cernicalo 'sparrow hawk'; LUCERNA 'lamp' beside LŪX LŪCIS 'light' > luc- (orig. luz-)iérnaga 'glowworm, fire-fly'. 54 The classic examples, assembled and examined by A. Alonso, "Cambios acentuales", BDHA, I (1930), 317-370, are atmósfera (clashing with esfera), cónclave (eccl.) 'conclave' (as against CONCLAVE 'room, chamber'), farrago 'hodgepodge' (as against FARRĀGŌ 'mixed fodder for cattle'), hipógrifo (based on a Greek compound; as against G. Hippogryph, made famous by Wieland's verse), médula 'marrow, pith' (as against MEDULLA), parásito (as against PARASĪTUS), presago 'foreboding, betokening' (as against PRAESĀGUS), rúbrica 'heading, flourish' (as against RUBRICA, lit. 'red earth'), túbano 'gadfly, horsefly' (as against TABĀNUS, witness Ptg. tavão, beside *TAFĀNUS and TABŌ -ŌNE > Fr. taon). As is well known, óptimo may have tendentially deflected ópimo 'rich, fruitful' (OPĪMUS 'fat'); PROCERĒS 'noblemen' may have influenced prócero 'high, lofty' (PRŌCĒRUS 'tall, long'); pábulo 'nourishment' (PĀBULUM) may have colored the pro nunciation of pábilo 'wick, snuff (of candle)', traceable to V.-Lat. PAPILU < Gr.-Lat. PAPȲRU; vértigo 'dizziness, fit of insanity' (VĒRTĪGO 'whirling round, giddiness') may conceal a strong measure of attraction by vértice 'vertex' (while Ptg. vertigem parallels imagem; note also Sp. vorágine 'whirlpool').

444

AFFIXAL DERIVATION

Italian. Independently and with entirely different intensity, it has become a distinctive feature of Sardinian, a point not devoid of importance in view of our recently improved understanding of Sardo-Hispanic isoglosses. Finally, the "esdrujulismo" is intimately tied up with the issue of the percolation of learnèd words into the general vocabulary, including the untutored speakers' reaction to their intrusion (a problem, incidentally, transcending the limits of Eomance, let alone of Spanish ; cf. L. Bloomfield's cogent description of some of its facets in his — undeservedly neglected — superb article on "Physigunkus"). 55 Obvi ously, the latent trend toward stress on the antepenultimate, which engulfed even the conjugational paradigm in Old Spanish (aviamos, oviéssedes, etc.), met the tide of Hellenisms in '-aro, etc. and the remnant of Latin diminutives in -ÜLU, -ŬLA the way the crests of two or three waves are apt to converge and to reinforce one another. The third, the PREFERENTIAL TREATMENT OF A (shown in the preservation of that vowel and in its occasional spread beyond the expected norm), cannot be squarely laid at the doors of prehistorical languages allegedly exerting their influence through a network of semantically protean suffixes. For one thing, Old Provençal accorded the same privileges to a — prerogatives rooted in universal physiological conditions. For another thing — striking closer to home —, weakly-stressed a was conspicuously preserved in Old Spanish in certain other positions, for which it would be quite unrealistic to enlist any substratai languages as prime movers, e.g. intertonically BEFORE the main stress — as in caramillo, maravilla, primavera.56 Again, we are free to argue that the conjunction of this perfectly normal feature with the influx of Hel lenisms (also, secondarily, with indigenismos, especially those of the toponymic class, in individual provinces — those constituting Spain included) may very plausibly have had a powerful effect on the language as a whole. 55 On the (severely limited) scope of "esdrujulismo" in Italian see R. Stefanini's review article in this issue, "Ars rhetorica come linguistica generale", at §5.1. As regards Sardinian, let me cite, from among numerous sources of information, the revised doctoral dissertation of Jonathan L. Butler, Latin -ĪNUS, -ĪNA, -ĪNUS, and -ĪNEUS : From Proto-Indo-European to the Romance Languages, UCPL, LXVIII (1971), esp. pp. 116-126. Such OSp. "cultismos" in -ulo, '-ula as I can readily adduce (báculo, discussed by 0 . H. Hauptmann in his rev. — RPh, VII, 245 — of A. Rey's ed. of Castigos y documentos; célula, used by J. Rodriguez del Padrón, cf. M. R. Lida de Malkiel in NRFH, VI [1952], 336) seem to be divested of any stylistic value; not so Lope de Rueda's nigromántulo (see RFE, XXXIV, 219), which points to a certain productivity of mocklearned -ulo. Regardless of the element of hilarity, it is words of this type that must have in duced speakers to switch, again and again, from medula, the learnèd doublet of meollo(s) 'marrow, pith, brain(s)', to médula. La medula de los huesos has a finer ring than un hombre de escaso meollo, just as hasta la médula is more elegant than its racy counterpart, hasta los tuétanos; cf. W. Bein hauer, "Beiträge zu einer spanischen Metaphorik", RF, LV (1941), 14f., 19. Bloomfield's almost forgotten paper on ┌physigunkus ┐ appeared in MPh, XV (1917-18), 577-602. 56 See above (fn. 31) for the statement of Zauner. A practically unlimited number of examples can be supplied; to those furnished by F. Hanssen (Gramática..., §§62, 77: calavera 'skull', OSp. meatad 'half' [beside meytad], antiguadad 'old age' [beside antigüedad] and by Menéndez Pidal (l6, §23 : canaherla 'giant fennel' [orig. a compound], caramillo 'saltwort', paraiso 'paradise', romadizo 'cold [in the head]'), let me add cascab-el ~ -ilio 'tinkle bell', calabaza 'pumpkin, squash', calabozo 'dungeon', caracol 'snail', primavera 'spring', to say nothing of such exotic words as caravana and carabela.


445

Fourth, the SUBSTITUTION OF A FOR *E OR ZERO in such cases as cuévano (from and pompano (from PAMPINU) is, unquestionably, an important side issue. We may grapple with it more successfully if, adopting Gilliéron's style of analysis, we ask ourselves just what might have been the "regular" outcome of the bases, after the tidal irruption of vowel syncope. If the answer is : *cuevno} *pampno, then it is clear that the switch to the familiar, easily pronounceable terminal segment -ano (cf. m. huérfano, Leon, huérgano, ravano, témpano; f. savana = mod. rábano, sobana) was a pleasing alternative to a whole bundle of unsatisfactory prospects, such as (a) learning how to utter clusters like *-vn-, *-mpn-, theretofore unknown; or (b) devising new ways or "rules" of modifying these unwelcome clusters; or (c) settling on an addition to the severely isolated -eno formations (of the type cárdeno 'purple, violet' < COPHINU)

CARDINU). 57

Fifth, though there exists at present a considerable corpus of studies on Hispanic INTERFIXES, which leaves no doubt as to the preëminence, by a wide margin, of -ar- (var. -al-) over its competitors, 58 joint consideration of inter calary -ar- and of the augments'-aro,'-arabids fair to array these isolated and haphazard observations in a more meaningful perspective. Sporadic attempts have already been made to examine the interplay of "suffixoids" and interfixes ; 59 the step ahead of us is the recognition of an even subtler interaction between augments and the two aforementioned categories of elements. Because, on the one hand, speakers feel free to develop from, say, cáscara a whole family of new derivatives: cascar-illa 'Peruvian bark', cascar-on '(broken) eggshell', cascar-udo 'thick-shelled', and because, on the other, the protracted coexistence of casca and cáscara favors such (subliminal) patterns of segmentation as casc-ar-illa, casc-ar-ón, casc-ar-udo, it becomes easy to wedge in an "empty" -ar- between a root morpheme and a genuine suffix, as in llam-ar-ada 'flare-up' 57 One moot point deserves to be explicated : As the example of Tuscan (cofano, pampano) irre futably shows, resistance to the threat of syncope need not have actually preceded the switch from the front vowel to a. The exact "anatomy" of the Hispano-Romance process can no longer be reconstructed step by step ; one plausible sequence of events was, first, the tentative rise of appealing variants in -ano {cuévano, pámpano) as mere by-forms, then their sweeping generaliza tion once the coexistent *'-eno formations were in acute danger of losing their unprotected vowel. 58 See my paper "Los interfijos hispánicos, problema de lingüística histórica y estructural" in the miscellany Estructuralismo e historia, II, 107-199; also F. H. Jungemann, Word, XV (1959), 482f.; F. Gonzalez Ollé, Los sufijos diminutivos en castellano medieval (Madrid, 1962), passim; and note the recent disagreement between H. H. Baumann, ASNS, CCVTI:5 (1971), 385-389, and J. R. Craddock in Lingua, XXVIII :4 (1972), 388-391. An earlier paper of mine, in which I used the (inadequate) label "infix", busies itself with situations like c'.én-ega ~ -aga; see "Studies in the Hispanic Infix -eg-", Lg., XXV (1949), 139-181, and cf. the discrepant reactions of F. Lecoy, Rom., LXXII (1951), 140-142, and L. Mourin, RBPhH, X X I X (1951), 864f. 59 This interplay is explicitly identified and subjected to leisurely inspection in my mono graph: Patterns of Derivational Affixation in the Cabraniego Dialect of East-Central Asturian, UCPL, LXIV (1970), esp. in the introductory chapter. There have been several critical reactions, in addition to the aforementioned appraisals by Baumann and Craddock; see A. Alsdorf-Bollée, RF, LXXXIII (1971), 340f.; P. Charaudeau, BSLP, LXVI:2 (1971-72), 137; W. H. Haverkate, Erasmus, Aug. 10, 1971. Vol. XXIII, cols. 738-741 ; J. J. Montes Giraldo, BICC, XXVII (1972),

142f.; and M. Popescu-Marin, SCL, XXII (1971), 431-434.

446

AFFIXAL DERIVATION

(from llama 'flame'), a derivative which thus acquires a minimum of differen tiation from llam-ada 'call' (based on the verb Hamar).60 Sixth and last, there undoubtedly exists a tenuous connection between the rise of the augments and the VICISSITUDES OF THE SUFFIX -ICUS, but the formula tion proffered by Menéndez Pidal need not be regarded as definitive. To under stand the wavering between cién-ega and cién-aga, or the transmutation of MORDICUS 'biting' into — superficially Arabicized — (al)muérdago 'mistletoe', one must bear in mind two processes : first, the diffusion of the triumphant a in this kind of syllable from its strongholds before r and n to a prevelar position, exactly as in Tusc. cronaca < Gr.-Lat. CHRONICA; second, the crystallization within Latin, from time immemorial, of a small number of formations, involving the segment -Āc-, usually as an offshoot of -Āx -ĀCIS; witness FORNĀX -CIS (f.) 'oven, furnace, kiln, crater' beside dim. FORNĀCULA and adj. FORNĀCĀLIS (p1. n. -ĀCĀLIA, in reference to a ritualistic sacrifice); also the two agentives -ĀCĀTOR (indicative of a submerged verb?) and -ĀCĀRIUS, all of which justifies, at least on the side of form, such semantically aberrant offshoots as Sp. horn-agu-ear 'to dig [the earth] for coal', horn-agu-ero adj. 'wide, spacious', f. 'coal', etc. — over against Ptg. negr-eg-ar, Cl.-Sp. -egu-ear 'to project black color' ( < NIGRICĀRE ; see REW3 5920 and BICC, IX [1953], 64f., with further clues to negr-egor, -ura 'blackness'). 12. Two Sets of Conclusions. The material here displayed and the ideas ventilated seem to invite two separate sets of conclusions — one geared to the special needs and demands of Romance scholarship, the other lending itself, potentially, to good use on the level of general linguistics. Within the framework of Romance studies, the major break-through at tempted was the rapprochement between certain discoveries bearing on Italian and of others traditionally associated with Hispano-Romance. The contrastive treatment of Tuscan and Spanish turned out to be revealing, because Tuscan (especially as used outside the precinct of Florence) shows the interplay of -ano, -aro, -ago {-ana, etc.) formations, in whose ranks characteristic Hellenisms played the dominant rôle, with the -olo, -ola words traceable in the last analysis to Latin diminutives. The eventual twist given to this interplay in Tuscan, to the effect that the choice of the weakest vowel was made dependent not on the genetic source of the word, but on the character of the following consonant, represents an elegant local solution without any close parallel in HispanoRomance. In the Iberian peninsula, the generalization of a was more sweeping. Menéndez 60 One detects a certain parallelism between the occasional rise of -(ar)ada and the (sweeping) expansion of -(er)ía at the cost of -ία, through a concatenation of circumstances established long ago and briefly restated in my paper: "Genetic Analysis of Word Formation", CTL, I I I (1966), 305-364, esp. 316f. ("Tripartite Derivational Proportions"). See also M. S. Breslin's comments on the rivalry of -ise and -(er)ie in Old French (BPh, XXII, 408-20).


447

Pidal's own favorite illustration, *SUBTULUS 'cellar', is more eloquent than are his hypotheses ; for it implies the — frequently denied — late productivity of -ŬLU ; it shows, through the example of the Provençal offshoot (sotol), what the general trend in Romance, familiar from Italian, initially must have been every where ; and it adumbrates the first phase of the process that must have taken place to the south of the Pyrenees : the replacement of *-olo by '-alo (archaic sótalo) and -ano (sótano). The spread of a to the position before l (which Tuscan to this day has denied to that vowel) must have been the clinching event : cf. Christóval(o) from CHRISTOPHORU, with the concomitant dissimilation of the two r's (whereas Ptg. Cristóvao achieves the same dissimilatory effect by apply ing the remedy familiar from sótano). The parallelism of r and l in certain derivational schemata (cf. the classic instance of -al ~ -ar in designations of groves and orchards and in other mass-nouns) must have greatly facilitated and accelerated this process. The second phase, even more momentous, was the tendential transmutation of appealing word-final segments into freely attach able and detachable augments, as in mur-ciég-al-o. The generous share of responsibility assigned to Greek need not cause any surprise, still less alarm. The assumed Greek streak involves here not the prestigious discourse of rhetoricians and philosophers, but the parlance of fisher folk and artisans, from whom provincial spoken Latin borrowed, among other tools, -iscu, -escu as a diminutive suffix. The intertonic a in proparoxytones was highly typical of Graeco-Latin formations (CITHARA, TYMPANUM, etc.) and could very well have spread from this nucleus, with varying momentum and direction. The Hellenisms clearly outrank in their impact the oft-mentioned substratai formations, whose concomitant agency need not, however, be per emptorily ruled out. Students of general linguistics are bound to focus their attention on other problems. The semantic weakness, bordering on emptiness, of the nominal augments places these increments at the very periphery of affixation, in the vicinity of suffixoids, interfixes (the semantic deadwood par excellence), and integrating or serializing ("einreihend") suffixes, such as the -al of Sp. robl-ed-al 'grove of oak trees' and, at least initially, of E. histor-ic-al (originally a synonym of histor-ic). Like the suffixoids, the augments are not always neatly detachable from the ''primitives"; in fact, the words into which they enter seldom qualify for the rôle of "derivatives", but rather represent re-shaped, stylized variants of the older formations — variants exhibiting special elaboration or sharpen ing of the word-final segment. Of the three basic goals or functions of language : achievement of clarity, attainment of economy, and gratification of the speakers' craving for pleasure, it is clear that, with rare exceptions, only the third is compatible with the inner diffusion of nominal augments in Romance.

L. ETYMOLOGY

IDENTIFICATION OF ORIGIN AND JUSTIFICATION OF SPREAD IN ETYMOLOGICAL ANALYSIS Studies in Spanish s(ol)ombra, en-sueño, dial. em-berano

In memory of Ramón Menéndez Pidal I. The General Problem. In the rather slim corpus of writings on etymological theory insufficient emphasis has been placed on the fact that the proponent of a new hypothesis usually owes his readers two distinct explanations: a defensible conjecture concerning the genesis of the word at issue and an equally cogent argument accounting for its subsequent diffusion and, sometimes, ultimate triumph — or else, for its eventual withdrawal and defeat. Typically, these two mutually corroborative trains of thought require disparate slices of supporting material and involve diverse modes of imaginative thinking. Also, it may hap pen that the degrees of transparency of the two problems jointly constituting the composite operation clash quite sharply : One readily visualizes situations in which the actual word origin poses no difficulty (e.g., where borrowing from a neighboring language, or either derivation or composition harnessing familiar native elements, has been at work), but where the ulterior success of a seemingly unpromising "experiment" is truly astounding. Conversely, the specific condi tions of a word's crystallization may be unverifiable, but — once that word ap pears on the scene "out of nowhere'' — the reason for its further progress may be readily understood, as when the closest rival it succeeds in dislodging was afflicted with one of the classic "ailments" diagnosed by lexicologists (excessive shrinkage, morphological complexity, phonological oddity, taboo, polysemy, homonymy, social degradation, broadly cultural wear and tear, etc.). Of course, a word may very well raise difficulties on both accounts — such items form the hard core of cruxes and the source of the explorer's despair — or, if its history is trivial, it may offer no resistance at all to either phase of the etymological analysis. II. The Specific Problem : Coalescence of Preposition (Prefix) and Noun. The illustrative problem at hand involves a handful of Romance (specifically, Spanish) word biographies selected to test the general proposition outlined

452

ETYMOLOGY

above. Each word shows one recurrent feature : It comprises a prefix, or quasiprefix — which genetically may go back to a preposition used in a close-knit syntagm — attached or even welded to a noun of Latin extraction. The three nouns which historically form the nuclei of the chosen formations are in no way refractory to analysis. Of the prefixes or quasi-prefixes, some do and others do not invite discussion; the congealment of the syntagm ┌preposition + noun┐ into a single indivisible word is apt to whet a Romanist's curiosity, especially if that noun concurrently disappears from free use. Between these isolable facets of the problem one expects to discover significant cross-connections ; thus, the rise of unusual by-forms of prefixes may be related to extra-tight patterns of attach ment, which in turn may abound in semantic implications and could have a bearing on the survival of the primitive. Once again, "tout se tient" in language, from whichever angle we elect to observe its functioning. The three words under examination are Sp. ensueno 'dream', tendentially differentiated from sueno 'sleep'; dial. (Jud.-Sp.) emberano 'summer'; and, above all, Sp. and Ptg. sombra, dial. (Leon. and Jud.-Sp.) so-, sa-, se-lombra, OPtg. soombra 'shade'. By way of background let me state that, although prefixes generally tend to become preverbs in Latin and Romance, a few of them — particularly those conveying a neat spatio-temporal message (front-, side-, back-position, con frontation, midway point, or circling; superiority or inferiority; anteriority, posteriority, iteration) — have at all times been called upon to produce new nouns and, on a more limited scale, adjectives and even adverbs, cf. Sp. antepuerto 'entrance to a mountain pass', corn-padre 'godfather', coll. 'companion', pos-pierna 'animal's thigh', tras-corral 'back court' beside redo-pelo 'rubbing the wrong way', redro-viento 'wind blowing from hunter's position', reta-guardia 'rear-guard'; contra-calle 'parallel side street', entre-cuesto '(sir)loin, backbone of an animal', circun-vecino 'neighboring, surrounding'; sobre-mesa 'tablecloth, table cover', so- 'cavity under a rock'; ante-ayer 'day before yesterday', re(quete)-bueno 'extra-good', obs. sobra-bién 'extra-well'. Verbal nouns have, of course, ceaselessly mediated between pre-nouns and pre-verbs, effectively bridging the gap: res-quemar 'to bite, sting' supports both res-quem-o ~ -azόn 'burnt taste (of food)', 'parching' and res-quem-or 'sorrow, grief'. Most of these prefixes are vernacular, a few are learnèd (esp. circuri-, to some extent ante- and pos-) ; several have lingered on unimpaired and sharply contoured, others show clusters and even strange proliferations of variants (am- beside ante-', pes- along side pos-;1 sa-, som-, sam-, za-, zam- surrounding so-) and display contaminations 1 On pcs-cuezo 'neck, haughtiness', pest-orejo 'thick (nape of the) neck', a n d t h e like see m y s t u d y of OSp. pos-, por-, pro-façar 'to slander, scoff' (OGal. posfaz 'mockery') in RPh, III (194950), 27-72, esp. 4 9 - 5 2 ; on pes-puntar 'to backstitch' see J . Corominas, DCE, I I I , 924. Pes-uña 'hoof' and -uno 'half of a cloven hoof' s t a n d a p a r t , involving as t h e y do P E D I S (gen. of PĒS 'foot'); pes-quisa 'inquiry' a n d -quisar 'to investigate' illustrate a unique conflation of es- < EX- a n d per-; per-punte, OSp. -o is a Catalanism obliquely related to Fr. pourpoint. A t t h e opposite end of t h e scale. Sp. amparar 'to protect, shelter', for centuries flanked b y mamparar, now obsolete or dialectal (cf. its r e m n a n t s : mod. mampar-a 'screen', -o [naut.] ' b u l k h e a d ' ) , reflects L a t e L a t .

IDENTIFICATION OF ORIGIN

453

(there has developed a hidden contact between [a] apheresized [en]tre- and [b] tra-, a by-form of tras-). What all the unexciting cases here briefly surveyed have in common, and what sets them off so sharply from the three problematic words, is the peaceful survival of primitive and derivative : puerto flanks ante-puerto, corn-padre is reconcilable with padre, sobre-mesa has not obliterated or even en dangered mesa. On the other hand, sombra has evicted *ombra in HispanoRomance; emberano, on a local scale, is a straight substitute for verano; the coinage of ensueño 'dream' marks a feeble attempt to split the semantic scope of sueno. Such is, exactly, the extent of these three words' (and their satellites') claim on our attention. 2 We shall discuss the two simpler cases first, relegating the controversial trajectory of sombra to the end. III. Spanish ensueno vs. Its Central Romance Counterparts. The starting point for tracing the trajectory of Sp. ensueno 'dream' is Lat. INSOMNIS 'sleepless' from SOMNUS -Ī 'sleep', built after the same pattern as IM-BERB-IS 'beardless' (beside BARBA), IN-ERM-IS 'unarmed, defenseless' (beside ARMA), and IN-FĀM-IS 'disgraced, disreputable' (beside FĀMA). Note that INSOMNIS — a post-Classical derivative endowed with poetic overtones — denoted a person eager but unable to fall asleep ; over against it stood EXSOMNIS, also 'sleepless', but in the sense of 'feeling wide awake and rested, requiring or desiring no sleep'. From INSOMNIS there branched off a normally structured abstract in -IA : INSOMNIA, which en tered into many set phrases — typically, in its plural form : INSOMNIĪS FATĪGĀRĪ, ALQM. OCCĪDERE, CARĒRE. The frequent and formulaic use of this abl. plur. — displaying an ending undifferentiated as to gender — plus, far more impor tant, the lateral pressure of SOMNIUM -Ī 'dream', an independent offshoot of SOMNUS, and the preëxistence of the model GAUDIUM 'joy', INGENIUM 'natural quality, mental power', etc., figured among those factors which made it possible for INSOMNIUM -Ī 'sleeplessness' to emerge alongside the better entrenched -IA abstract. The European languages, which have absorbed as part of their medical ANTEPARĀRE reminiscent on the semantic side of G. vorbeugen ('to build fortresses in anticipation of attacks by barbarians'); in Hispano-Romance there ensued a—temporarily successful— contamination by such compounds of MANU + verb as man-tener 'to keep (up), uphold', refl. 'to remain firm', lit. 'to keep the control in one's hands', while elsewhere (e.g., in Old Provençal), emparar, a descendant of *IMPARĀRE — independently authenticated, in a wide latitude of mean ings, by French and Italian — was secondarily dragged in. 2 We shall not be concerned here with the functionally gratuitous addition of a prefix as a result of close association with a powerful near-synonym, as when Sp. endeble 'feeble, fragile, flimsy', unaccompanied by any verb, and conceivably also equally isolated enclenque 'weak, sickly' (not recorded before the 18th c ; to the dubious bases mentioned in DCE, II, 261f. one might add, in a realistic vein, G. kränk-lich, -ein — an echo of Thirty-Year-War lazarets in Central Europe?) seem to owe their en- to attraction by en-fermo 'sick' < INFIRMII 'feeble, faint', with which they also share a triad of prime characteristics : the number of syllables, the stress pattern, and the tonic vowel. Nor shall we dwell on the occasional transfer of prefixes (typically, from denominal verbs) either to the very same primitives from which they were originally derived or to other postverbals ("Präfixverschleppung"), as in Sp. en-grudo 'glue, paste', patently influenced by en-grud-ar 'to paste', -amiento 'pasting', -ador 'paster, pasting brush' (etymon: GLŪTEN, cf. Ptg. grude, mor phologically unadulterated), or as in Sp. des-nudo 'bare, naked', OSp. (d)es-nu(d)o,_beside (d)esnu (d)ar < DĒ-NŪ-DĀRE (from Ennius to the Vulgate), alongside rare and late E-NU-DĀRE; again, the leaner Ptg. counterpart nu perpetuates NŪDU.

454

ETYMOLOGY

terminology words of such referential content, reflect this duality : Fr. insomnie, Ptg. insônia, and E. insomnia echo the feminine, Sp. insomnio the neuter abstract of Antiquity ; Italian is conciliatory in tolerating insonn-ia beside -io.3 Far less familiar to students of Latin, but of heightened concern to Romanists, is a homonym, to be known as INSOMNIUM2, which has been likened to G. 'Traumbild' ; it may represent a learnèd imitation of ενύπνιον, but in any event is much closer to the domain of 'dream' than to the semantic ambit of 'sleep lessness'. It is this "double" of the better known INSOMNIUM1, a late straggler, that has seeped into several provincial varieties of colloquial Latin, emerging in medieval and modern disguises throughout Northern and even Central Italy, in the Raeto-Romance speech islets, in Catalonia, and in Spain proper, frequently locked in keen rivalry with the descendants of SOMNIUM. Also, it is from INSOMNIUM 2 that new verbs have taken their cue : It. insognare, Sp. ensonar, unless of course one prefers to extract them from the respective primitives (It. sogno, Sp. sueño), arguing that it is these verbs which, in the first place, gave rise to vernacular abstracts burdened with the prefix in-, en-.4 The record of pertinent research can be very briefly summarized. Combing through his three prized North-Italian 15th-century dictionaries with glosses in German, A. Mussafia, in a classic monograph (1873), extrapolated the verb insomniare 'to dream' (ensuniate, me ha insuniado) and the noun insonio 'dream'. Half a century later, in the original edition of his REW (§4469), Meyer-Lübke paraded a distinctly larger cluster of formations, without speci fying his additional dialectological sources: It. insogno, OVen. insomnio, Reggio, Moden. insoni, Engad. insömmi, Friul. insium, Sp. ensueño, adding for good measure, in the rubric of derivatives, Engad. ensömger 'to dream'. 5 The 3 W. von Wartburg has assembled some noteworthy scattered data in his FEW, IV (1947-52), 719b, all of them pertaining to the learnèd strain in the French lexicon. Of INSOMNIUM 'dream', which is here at the center of our attention, he found a single isolated vestige (Cretin, A.D. 1525 : insompne 'rêve'). Fr. insomnie (fem.) has been constantly in use since 1581 ; a harbinger of its eventual adoption is an isolated record traceable to 1555 ; in 1680 Richelet remarked : "Quelques médecins font ce mot masculin". The adj. insomne 'deprived of sleep' appears in a single Renais sance source (Papou, A.D. 1581); the abstract insompnieté (1495) reflects INSOMNIETĀTE. The spel ling with -mpn-, reminiscent of OSp. columpna and the like, marks an effort at careful differentia tion of the two nasal consonants — in constant danger of merger —, with homorganic -p- serving as a buffer. 4 In his celebrated Beitrag zur Kunde der norditalienischen Mundarten im 15. Jahrhundert (Vienna Acad., Ph.-H. K1., Memoir XXII [1873], p. 71 ; photographically reproduced, with F. Gysling's Indices and C. Tagliavini's Preface, in Bologna, 1964), A. Mussafia contended that N.-It. in-, en-suniar was a vernacular derivative from insonio 'dream' rather than a genuinely "'neue Bildung". Conversely, G. I. Ascoli, toward the end (pp. 451f.) of his brilliantly argued note "Le doppie figure neolatine del tipo briaco imbriaco", AGI, I I I (1878), 442-452, maintained that the equation insogno = sogno, due to the coexistence of insognare, must not be construed as proof that the adj. briaco 'drunk(en)' < ĒBRIĀCU first gave rise to im-briac-are 'to intoxicate', from which imbriaco 'drunk(en)' was then secondarily extracted; rather do the doublets briaco and imbriaco exemplify predictable rival solutions of the difficulty caused by such words as begin with a loosely dangling, unstressed, syllable-filling vowel. 5 There exists a sizable corpus of data on INSOMNI-U, -ĀRE in Raeto-Romance and in Italian dialects. No lesser figure than Ascoli commented in his "Saggi ladini", AGI, I (1873), 283 (fn. 2) on Val Poschiavo in-soeumi 'dream', as identified in P. Monti's pioneering Saggio del vocabolario... (Milano, 1856). Among this century's students of Western Raeto-Romance, spoken in Grisons, let me single out for mention C. M. Lutta, Der Dialekt von Bergün... (ZBPh, Suppl. LXXI [1923]),


455

definitive version (1930-35) of that comparative dictionary repeated, unchanged, this assortment of information, adding Tortos. (Prov. of Tarragona) insomit ; the Spanish and the Itahan verbs remained absent from the roster, nor did Meyer-Lübke make any attempt to calibrate the items adduced as to date of coinage, degree of continued vitality, semantic range, or stylistic level. The salient difference between Italian and Spanish usage consists in this: While Tusc. insogno has remained so peripheral even to the literary lexicon that relatively few lexicographers have bothered to record it, ensueno was seized upon by the architects of the Golden Age vocabulary as a most welcome and exquisitely elegant ingredient, witness Cervantes' authoritative and astonish ingly explicit statement : "De una de tres causas los ensuenos / se causan, o los suenos, / que este nombre / les dan los que del bien hablar son duenos". 6 From Classicism the meandering path leads to Romanticism, for which G. A. Be'cquer may well serve as the spokesman — though this time qua practitioner rather than qua arbiter : "La vi como la imagen / que en un ensueno pasa" (Rimas 24/LXXIV; Obras: Que en leve ensueno pasa). 7 From here the road takes us to the modernists of all stripes, for whom ensueno is a poetic word par excellence.8 §§97, 299, and, especially, Mena Grisch's exemplary Zurich thesis, Die Mundart von Surmeir (Ober- und Unterhalbstein) (RH, XII [1939]), pp. 39-176, who distinguishes between (a) archaic [sαmnęar], preserved in Plaun, Muntogna, and Schons; (b) more advanced [z'αsimğīər], charac teristic of most of Surmeir and Engadin ; (c) very advanced [sαmjá], restricted to westernmost Disentis; and (d) isolated [z'αsumαłiαr] beside the noun [sum???ł] 'dream', confined to Vaz (simil arly in Marmorera), perhaps presupposing *SIMILIĀRE. AS regards the East of Grisons, A. Velleman, Grammatica... della lingua ladina d'Engiadin'Ota (Zurich, 1915-24), exemplifies the use of s'in-sōmğer (p. 646) and discusses its conjugation (pp. 460, 645, 888), whereas A. Schorta, Lautlehre der Mundart von Müstair (RH, VII [1938]), §§69, 80, 181, 198),has recorded in his field-notes sömi < SOMNIU (flanking the verb sömyar), as against sön < SOMNU. On the Italian side we have a profusion of testimonies, esp. from the North. Cf. G. . Melchiori, Vocabolario bresciano-italiano (Brescia, 1817), p. 222b: en-sòme 'sogno', fà dei e[n]-some — ensomiàs 'sognare' (as against en-sonolent or en-sorgnat 'sonnacchioso', harking back to SOMN-US, -OLENTUS); A. Tiraboschi, Vocabolario dei dialetti bergamaschi antichi e moderni2 (Bergamo, 1873), ρ. 674α: in-sognàs, -somiàs, -sōmiàs '(in)sognarsi' = R.-Rom. in-sömgias; G. Patriarchi, Voca bolario veneziano e padovano...3 (Padova, 1821), p. 111b: in-soniarse 'sognare', in-sonio 'sogno' (as against in-sonà = in-sonolio 'addormentato, sonnacchioso, balordo dal sonno'); G. Ungarelli, Vocabolario del dialetto bolognese (Bologna [1901]), p. 1476 : insonni 'sogno'. The area of INSOMNIU clearly extends to the South-Central dialects; note the record in F. Chiappini, Vocabolario romanesco2, ed. . Migliorini (Roma, 1945), p. 166 : insogno 'sogno' vs. insonnolito (reminiscent of Patriarchi's in-sonolio) 'assonnato'. 6 Viaje del Parnaso, VI, 46v°; see . Fernández Gómez, Vocabulario de Cervantes (Madrid, 1962), p. 4006. Aside from adducing this passage the original Academy dictionary, I I I (1732), 4966, also quotes F. Herrera's commentary on Garcilaso's Second Eclogue: "Y sola la mente, no enlazada con algúm organo, se fatiga  congoxa con los ensuenos que finge presaga de lo futuro". 7 I owe this example and some of the following quotations to E. L. King's concordance ap pended to his book Gustavo Adolfo Bécquer: From Painter to Poet... (Mexico, D.F., 1953), pp. 302304. 8 In post-Romantic literature, on both sides of the Atlantic, ensueno has been used increas ingly in its figurative sense of 'daydream, illusion, fantasy'; M. Alonso, Enciclopedia del idioma (Madrid, 1958), p. 17426, refers the reader to specific passages in Valera's Pepita Jiménez, Rodó's Ariel, and Ortega  Gasset's El espectador. (Raimundo Lida reminds me that Rubén Dario chose "Ensuenos" as the title for a group of poems included in El canto errante and "Suenos" as the title for another such group ; see the Aguilar ed. [1967], pp. 732-740 and 1074f.) On the Peninsular dialect level ensueno is characteristic of the East; note Sant ni por ensueno 'in no way' (cf. E. by no stretch of the imagination and — idiosyncratic to the speech of women — I would not dream of doing it). Jud.-Sp. es-, is-fueno perpetuates sueno rather than ensueno: The prosthetic vowel has

456

ETYMOLOGY

Ensonar has meanwhile remained quite rare and confined to the Eastern portion of the Peninsula.9 The reason for the alacrity with which the masterminds and codifiers of Spanish usage have accepted ensueno is that it affords an excellent means of eschewing a potentially irksome polysemy.10 Through a whim of circumstances, Sp. sueno means both 'sleep' and 'dream', reflecting — in strict conformity with the canons of regular sound development — both SOMNUS and SOMNIUM. No such homonymic collision (within the bounds of the same word family : Schuchardt's "innere Formenmischung") has occurred in the other major Romance languages: French contrasts sommeil (Diez half-jokingly likened its prototype to 'Schläfchen'11) with songe, whose verbal counterpart in the end underwent a noteworthy semantic metamorphosis ('to dream' > 'to brood, muse' > 'to reflect, think'), perhaps in conjunction with the vagaries of resver, derver. Old Provençal favored for 'sleep' either som, or son, or son-elh, which, opportunely, clashed with somnh-ar 'to dream' (beside som-, -jar), flanked by the noun somnhe, som(n)i.12 Portuguese distinguishes adequately between sono and sonho, much as does, with superior efficacy, Italian (sonno:sogno). Ruma nian has completely extricated itself from the predicament by reserving masc. somn for 'sleep' and either ambigen. vis or fem. vedenie (its exact Slavic replica) for 'dream'. By preserving — if only marginally — ensueno, Spanish, as a vehicle of poetic discourse, has a residual means of separating 'dream' from 'sleep' where been added in response to the newly formed initial consonant cluster /sf/. See A. Alonso, "Pro blemas de dialectologia hispanoamericana", BDHA, I (1930), 409f., with a reference to J. Subak. 9 Ensonar has had a distinctly "Eastern" flavor in the Peninsula for at least seven centuries, a feature which enables us to link it geographically to the string of Italo-Raetian descendants of INSOMNI-U, -ĀRE. It is traceable to Berceo's venerable Vida de Santa Oria — a poem in which dreams provide the dominant theme —, st. 188cd: "Ensonnó esta duenna un suenno de(s)seado, / por qual muchas vegadas ovo a Dios rogado" (on dreams and visions in that saint's life see T. A. Perry, Art and Meaning in Berceo's "Vida de Santa Oria" [Yale University Press, 1968], pp. 20, 53-55, 68-88, 96-103, 111-114, 130f.); indirectly, it has given rise, in Navarre and Murcia, to the familiar phrase ni por ensoñación 'not even in one's wildest dreams' (cf. the comment, above, on its counterpart Sant. ni por ensueno), whereas Murc. refl. en-soñ-at-ar, -isc-ar 'adormecer' show a tangle with sueno 'sleep'; cf. J. Garcia Soriano, Vocabulario del dialeclo murciano (Madrid, 1932), p. 49. That indefatigable experimenter Azorín coined ensonador 'endowed or burdened with illu sions'; see Obras, IV (1947), 269. 10 Gilliéron was not the first to discover the full scope of the agency of anti-homonymy. Thus, F. Diez, in EWRS, Section IIC, s.v. sommeil, argued: "Gleichsam SOMNICULUS ..., eine Ableitung, wozu die Sprache genötigt war, um som (SOMNUS) von son (SONUS) ZU scheiden"; the author might also have included in his confrontation *som < SUMMU 'top', for which French has success fully substituted unequivocal somm-et. 11 The suspicion of an anti-homonymic gambit in the acceptance of ensueno beside sueno was initially voiced by F. Schalk in his article "Somnium und verwandte Wörter im Romanischen", which also seeks to encompass the vicissitudes of VISUM, VĪSIŌ, ŌRĀCULU, of Fr. rêver and its vars., of Sp. desvario, etc. This paper  forms part of Schalk's miscellany Exempla romanischer Wortgeschichte (Frankfurt a /M., 1966); see J. H. R. Polt's perceptive review in RPh, XXI, 560564, at 563. 12 The proliferation of reflexes of SOMNI-U, -ĀRE was well known to leading Provençalists half a century ago; they refrained, however, from proffering any explanation of this luxuriant growth. See 0. Schultz-Gora, Altprovenzalisches Elementarbuch2 (Heidelberg, 1911), §89; C. Appel, Provenzalische Lautlehre (Leipzig, 1918), §§30, 33a, 45, 59a. In Old Catalan the form somni was made famous by the title of Bernat Metge's masterpiece.


457

such differentiation is desired (quite apart from the metrical advantage that, in certain contexts, may accrue to a trisyllabic word). On the other hand, one readily envisions passages where a mood of drowsiness or the hazy contour of a dream-like landscape is deftly evoked by sueno on account of its mild ambiguity. While it would be unthinkable to confuse the concepts 'sleep' and 'dream' in grappling with the meaning of Calderón's La vida es sueno, there is no dearth of passages in Bécquer's Rimas where the two meanings seem to abut on each other and even to merge, where indeed any attempt to draw a razor-sharp line between them would be, esthetically, of no avail:"???Cuandopodré dormir con ese sueno / en que acaba el sonar!" (1/XLVIII);" j Oh, si las flores duermen, / que dulcísimo sueño!" (6/XVIII). Of course, a word magician of Bécquer's spell binding power (and even artists of far lesser skill) can at all times "disambi guate" sueno by a variety of procedures, such as the use of the plural, which mandatorily narrows down the meaning to 'dreams': "Isla de suenos donde reposa / el alma ansiosa" (5/LXXII), or such phraseological, lexicotactic features as the stereotyped combination with hay ("???Qué hermoso es, cuando hay sueno, / dormir bien...!", 18/LXVII) or with tocar ("¿Sera verdad que, cuando toca el sueno / con sus dedos de rosa nuestros ojos / ...?", 23/LXXV). which both peremptorily stamp sueno as the equivalent of 'sleep'. IV. Dial. (Jud.-Sp.) emberano 'summer'. The Latin words for 'summer' and 'winter' were, respectively, AESTĀS -ĀTIS and HIEMS -EMIS. While AESTĀTE has been relatively well preserved (cf. Fr. été, assigned to the masculine gender, and It. state), there has been a certain trend in Romance to replace the noun by descendants of the neuter-masculine of the corresponding adjective, AESTĪVU 'relating to summer' (cf. Sp. estio 'late summer'). To appreciate the rationale behind this shift one should perhaps start from the equation AESTĀTE 'summer weather, summer heat' (rather than the season as an abstract temporal seg ment) and (TEMPUS) AESTĪVU(M). Of HIEMS there seem to have remained no direct traces in vernacular speech — so sweeping was the advance of HĪBERNU and the transmutation of the erstwhile adjective into a noun (see REW3, §4126 ; for HĪBERNĀRE 'to spend the winter', also astonishingly well entrenched, from Rumania to Portugal, see ibid., §4124). The shifts here observed closely parallel the classic substitution of DIURNU 'daily' for phonically impoverished, grammatically aberrant DIE 'day' — a process on which there is no longer any lack of expert information ; note that the full-bodied, grammatically normal NOCTE has not budged, despite the ready availability of NOCTURNU.13 VER -IS (n.) 'spring' has bifurcated, in Hispano-Romance, into primavera 'spring', presaged by Class. PRĪMŌ VĒRE 'in the beginning of spring' and shared by 13 On the Romance offshoots of DIURNU see Karin Ringenson's classic study " D I E S et DIURNU ; étude de lexicographie et de stylistique", StN, X (1936), 3-53 and map; to this inquiry one may add the volume of later research, by H. P. Bruppacher and others, on the names of the week-days. On the modest-sized progeny of that word's polar opposite see my own note "Romance Descen dants of Latin NOCTURNUS, NOCTURNĀLIS", SPh, XLVI (1949), 497-513.

458

ETYMOLOGY

Italian and Eumanian (lit. 'first tokens of spring', from p1. VĒRA? But French has opted for printemps) and verano '(early) summer', based on another sub stantivized adjective reminiscent of R. vesennij, poet. vešnij (from vesna 'spring'), G. frühlingsmässig, Fr. printanier, Sp. primaveral, Ptg. primaver-al, -il, E. spring-like, (poet.) vernal. Conversely, Sp. otono < AUTUMNU has eluded any displacement, possibly because, in syllabic and accentual terms, it enters into a close alliance with verano, estío, and invierno. In some dialectal varieties of Spanish, e.g., in the Judeo-Spanish of Saloniki, the word for 'summer' is emberano — which elsewhere represents a very com mon, almost formulaic sequence, serving as a vehicle for a noun phrase: 'in summer'. It is the solid coalescence, not the loose concatenation as such, that is at the heart of the problem. As Ascoli shrewdly observed in 1878, French offers one striking parallel : en-de-main, cf. mod. le lendemain 'next day' — with a repetition of l(e) — which, let me add on my own, recalls the double d(e) in mod. Sp. de donde (as against OSp. and Ptg. donde) 'whence'. Conceivably, in the pre history of the French word the differentiation of 'morning' and 'tomorrow' played a certain rôle ; the elucidation of this point would demand a full-scale reconstruction of the relations between MĀNE and MĀTŪTĪNU (and their local products). As regards Jud.-Sp. emberano, one can best explain the welding-on of em- through reference to invierno, in view of the very sharp "lexical polariza tion" of 'summer' and 'winter'. (Lexical polarization in this context means the far-reaching solidarity of two semantic opposites, with a stronger, "active" partner affecting, in varying degrees, the form or grammatical status of its weaker, "passive" ally. 14) In other words, the key to the complete concrescence of en and verano may lie in the vicissitudes of its semantic counterpart. The original outgrowth of HĪBERNU in Spanish was yvierno, amply attested in medieval texts; 1 5 what with the familiar anticipation of the nasal (cf. parangón 'model', as against It. paragone < Gr.-Lat. PARAKONÊ, lit. 'touchstone, whet stone'), 16 what with the agency of "Ascoli's Law" (1878) which presided over 14 My own central thrust into this uncharted territory ("Lexical Polarization in Romance", Lg, XXVII [1951], 485-518) was accompanied by several satellite studies, of which the most ambi tious, perhaps, were "Estudios de léxico pastoril: piara y manada", BH, LIII (1951), 41-80; "Ancient Hispanic vera(s) and mentira(s)", RPh, VI (1952-53), 121-172 ; and Section II ("HispanoLatin *PEDIA and *NI") of the book-length monograph Studies in the Reconstruction of Hispano-Latin Word Families, UCPL, XI (1954). For another view of the phenomenon, in distinctly broader, panoramic perspective, see my paper "Diachronic Hypercharacterization in Romance", ArL, IX (1957), 79-113 (esp. 103-106); X (1958), 1-36. 15 The phonologically normal outcome yvierno was very common in medieval texts; it is the only form encountered in the Poema de Mio Cid, see v. 1619 in Menéndez Pidal's edition: " E l ivierno es exido, que el março quiere entrar". As the editor remarks in his Vocabulary, s.v., the concluding month of the winter season, by the 12th-c. calendar, was February. For a generous collection of supporting examples of yvierno see M. J. Canellada, El bable de Cabranes (RFE, Suppl. X X X I ; Madrid, 1944), p. 242. On a possible fleeting contact of yvierno with yfierno 'hell' see D. Catalán Menéndez-Pidal, "La pronunciación [ihante], por /iffante/, en la Rioja del siglo XI", RPh, XXI, 433. 16 For a glimpse of the nasal infix in Old Spanish — a sprinkling of illustrations and a guide to earlier discussions — see my paper "En torno a la etimología y evolución de causar, causo, causa(n)cio" (I), NRFH, IX (1955), 225-276, esp. 236-242. These findings must now be collated with D. Catalán Menéndez-Pidal's forceful view of the counter-process : the tendential loss of the


459

the cleavage of HĪBERNU into It. verno ~ inverno and explains Sp. (obs.) briaguez beside embriaguez 'drunkenness' and the like, the initially sporadic var. invierno finally evicted yvierno.17 The force that prevented word-initial i-, etymologically justified, from yielding to e under pressure from the prefix en{cm-) was primarily the metaphonic effect of ie, a phenomenon familiar from the verbal paradigm (pid-ieron, -iese, -iendo beside ped-ir, -). In arrhizotonic derivatives from which this diphthong was absent, e.g., invern-izo 'winterly', by-forms displaying an e (say, envern-izo) have at all times been exceedingly common, esp. in dialect speech. 18 It was, then, the semantic pressure of inviern-ļenvern- on verano, plus the concomitant fact that, through a strange historical accident, the kernels -v(i)ern- and veran- happen to display an arrestingly similar consonantal struc ture, and, in addition, the phrasal commonness of the sequence en verano, which, all three, conjoined to spark the transmutation of verano into emberano. What in the case of ensueno was accomplished in the literary standard by half hearted flight from polysemy, was here unobtrusively achieved, in a far corner of the Spanish-speaking domain, by lexical polarization. V. Sp. Ptg. sombra, dial. (Ast.-Leon.) so-, se-lombra. For at least a century and a half historically-minded Hispanists have been actively concerned with the s- of Sp. Ptg. sombra 'shade, shadow', a feature which sets off that word so sharply both from its Latin counterpart and obvious—if not necessarily direct — source UMBRA and from those cognates which come to mind first: Fr. ombre, It. ombra, Rum. umbră (all of them displaying unimpeachably normal develop ment). The pioneers did not worry about the accretion of s-; the Academicians, in 1739, were satisfied with the crude equation sombra < UMBRA, and long before them, S. de Covarrubias, in 1611, had recourse to the same derivation, merely remarking, apropos sombrero '(wide-rimmed, shade-providing) hat': "Quasi SOLIS UMBRA". It was the philological activity of T. A. Sánchez which, in the concluding decades of the 18th century, led to the discovery, in medieval nasal in such groups as -NS-, -NF-, and -NV- on their way from Latin to Old Spanish; see his above-cited scrupulously documented article in BPh, XXI, 410-435. 17 For centuries, ynvierno and yvierno were locked in rivalry in late medieval texts. Note that Portuguese boasts inverno, surrounded by a galaxy of similarly structured derivatives. There was, to be sure, no rising diphthong along the Atlantic Coast to prevent a shift of in- to en- under the pressure of prefix en-; but the cluster NASAL + CONS. is independently known to favor the reten tion or restoration of i, as against e, far more consistently in Portuguese than it does in Spanish (witness língua 'tongue', míngua 'dearth', as against Sp. lengua, mengua, from LĬNGUA and MĬNUERE, *-ǍRE, respectively ; equally worthy of contrast are Ptg. vinte < viinte and Sp. véinte < veinte < VĪGINTĪ 'twenty'). Italian, I repeat, likewise favors inverno alongside verno, a pair of doublets which inevitably caught the attention of Ascoli in 1878. 18 To cite just one striking example, the Cabraniego dialect of East-Central Asturias favors ibiernu (see fn. 15, above) for 'winter', but reconciles this preference with the choice of embern-ar 'to hibernate, feed animals fodder during the cold winter months', embern-era 'winter season', embernices (p1. fem.) 'winter crop of potatoes', embern-izar 'to turn chilly and gloomy, as in winter' (in reference to weather), embern-izu 'relating to winter' (a form anticipated by Juan Ruiz, 829d, S; G favors yvern-izu); see Canellada's monograph, p. 174, OSp. en-vern-iego 'wintry' (vs. veran-iego 'summerly') and additional offshoots are reviewed in Lg, XXV (1949), 172f.; UCPL, I V : 3 (1951), 135f. and 186.

460

ETYMOLOGY

texts, of a variant long-forgotten by the literate, solombra. Confronted with this by-form and alert to Covarrubjas' remark on sombrero, E. Cabrera — whose active years fall into the Napoleonic period — went one step further and ana lyzed solombra as the intermediate stage between ancestral UMBRA and mod. sombra.19 At this point comparativism appeared on the scene, tempting its practi tioners to use "lateral" evidence to fill any residual gaps in the direct line of transmission. As early as the year 1853, which witnessed the publication of the original version of his dictionary, F. Diez, though oblivious neither of Cabrera's conjecture nor of its antecedent in Covarrubias' formulation, resolutely veered away from any involvement with SŌL -IS 'sun'. 20 According to him, it was SUBor SUBTUS- that had yielded the s-, and the proof lay in OProv. sotz-ombrar 'to cast a shade', from which *SUB-UMBRĀRE — so his argument ran — could be safely extrapolated; the exact semantic hue posited was 'UNTER Schatten setzen'. Diez welcomed the support of OSp. solombra, segmenting it so l'ombra, on the model of Fr. à ľombre; called attention to strikingly similar Prov. Dauph. solombrar 'to cast a shade', mod. Prov. souloumbrous 'shady', granting that these regionalisms may represent mere distortions of sotzombrar ; and tenta tively encompassed in his bold purview the French adj. sombre 'gloomy, dismal, murky' (which penetrated, as somber, into Dutch and, he might have remem bered, likewise into English) and, more cautiously, Rutebeuf's essombre 'shady place' — a generous assortment of offerings indeed for a ground-breaking operation. 21 Such were the Founding Father's personal prestige and the reputation of the discipline which his grammar and dictionary represented that his hypothesis, 19 S. de Covarrubias Orozco, Tesoro de la lengua eastellana o española (1611), ed. M. de R i q u e r (Barcelona, 1943), p p . 943f.; "Diccionario de Autoridades", V I (1739), 147α; R. Cabrera, Diccio nario de etimologías de la lengua castellana, ed. J . P . Ayegui (Madrid, 1837), I I , 644. 20 T h e revised 3d ed. of EWRS contains b u t few a l t e r a t i o n s : T h e hypothesized shift [*]SOombrar > sombrar is now characterized as " c o n t r a c t i o n " , O P t g . soombra is mentioned (but no inference is d r a w n from its -oo-, so clearly a t variance with OSp. -o-), and, a t t h e v e r y end, sombr-ero ' h a t ' beside [OLeon.] solombrero, favored by MS  of t h e Alexandre, is a d d e d to t h e i n v e n t o r y . I n his S u p p l e m e n t t o t h e p o s t h u m o u s 5 t h ed. (1887), p . 777, t h e reviser A. Scheler a u t h e n t i c a t e s SUB-UMBRĀRE by referring t h e reader t o H . R ö n s c h ' s n o t e in ZRPh, I I I (1879), 104 a n d inclines to trace F r . sombrer ' t o founder' t o t h e same source, despite his familiarity w i t h H . W e d g w o o d ' s semantically persuasive rival e t y m o n (ON sumbla ' t o overwhelm, swallow d o w n ' ; see t h e concluding p a r a g r a p h of his " F r e n c h E t y m o l o g i e s " , Rom, V I I I [1879], 435-439). G. K ö r t i n g — as confused as ever — eagerly credit3 sombra to SUB-UMBRĀRE and, a t t h e s a m e time a n d w i t h equal alacrity, c o m m e n t s on s(ol)ombra: " V e r m u t l i c h beruhen diese Bildungen auf a n t o n y m e r A n l e h n u n g a n sol: ' S c h a t t e n vor der S o n n e ' " ; he a d d s P t g . sombreiro to t h e dossier a n d overrules Scheler in assigning F r . sombrer t o Old Norse. See Lateinisch-romanisches Wōrter buch 3 ( P a d e r b o r n , 1907). F o r one specimen of a fresh a t t a c k see R. Lathuillère's rev. (in RPh, X X I I : 2 ) of R . Arveiller, Contribution à l'étude des termes de voyages en français (1505-1722): "Sombrer vient de ľ e s p . zozobrar ou d u p o r t . sossobrar et, après avoir subi l ' a t t r a c t i o n de sombre, d'où soussombrer, a été r é d u i t à u n «simple» sombrer". 21 W i t h o u t sifting one b y one all t h e d a t a a d d u c e d b y Diez and his i m m e d i a t e followers, one can briefly s t a t e t h a t t h e y all erred in e x t e n d i n g sombra to t h e area of Catalan a n d also u n d e r e s t i m a t e d t h e power of diffusion (hierarchically, I t . ombraggio is n o t on a p a r with F r . ombrage b u t represents merely its offshoot, etc.). T h e evidence of Sard. umbra was i n t r o d u c e d b y G. Gröber, "Vulgär lateinische S u b s t r a t e romanischer W ö r t e r " , A L L G , V I (1889), 145f. ; -Lübke's e t y m o logical dictionary recorded D a l m . sombreya.


461

polished by dint of minor adjustments, improvements, and primings, has ever since been accepted almost universally — if with varying degrees of enthusiasm. Conceivably his first follower was P. F. Monlau ; the roster of supporters subse quently recruited, whether heavily committed or half-hearted, includes C. Michaëlis, P. Förster, J. D. M. Ford (who, however, was willing to credit SÒL with a measure of influence on solombra), W. Meyer-Lübke (first in his compara tive grammar, later in the original as well as the revised text of his dictionary), R. Menéndez Pidal (obliquely, and perhaps solely in the context of the Cid grammar), and W. von Wartburg in a fairly recent fascicle of his FEW — to cite just a handful of familiar names. 22 On the other hand, J. Corominas has, with impressive consistency, elaborated on the old Covarrubias-Cabrera formula, by abandoning the infelicitous syntagm SŌLIS UMBRA and pairing off, instead, SŌL and UMBRA as semantic

opposites — the former being, by implication, the stronger ("active", "posi tive") partner, endowed with sufficient strength to have attracted and partially reshaped its weaker ("passive", "negative") ally. 23 The tactical advantage of 22 P. F. Monlau, Diccionario etimológicode la lengua española2, ed. J. Monlau (1881 ; here quoted from the reprint, Buenos Aires, 1941 ; the 1st ed., dating back to 1856, is unavailable to me), pp. 369b-370a (s.v. asombrar) and 1052b: : "contracción del1. SUBUMBRA, so-umbra" (semantically, the author likens asombrar to OBUMBRĀRE); C. Michaëlis [de Vasconcelos], Studien zur romanischen Wortschöpfung (Leipzig, 1876), p. 62: "Wie lassen [frz.] porche < PORT-ICUS, [sp.] percha < PERT ICA, [cat.] mege < MED-ICUS sich noch in ihre Grundbestandteile zerlegen? Wer erkennt in [sp.] sombra, sondar noch ohne weiteres sub-ombra, sub-ondare [sic]? Wer nennt in ihnen noch ombr ond den Stamm?"; P. Förster, Spanische Sprachlehre (Berlin, 1880), p. 146 (with a sobering questionmark); H. A. Todd, in MLN, I (1886), col. 289, in criticism of W. I. Knapp's adherence to SŌLIS UMBRA (1883); W. Meyer-Lübke, Grammatik der romanischen Sprachen : Lautlehre (Leipzig, 1890), §446 (sombra and sondar are again paired off); REW 1 §8405 (note the terse comment anent Dauph. sulumbrá and OSp. solombra: "mit unerklärtem Z"); REW 3 §8405 (with a cross-reference to §9046; Engad. sumbria has been tacitly emended to -iva); R. Menéndez Pidal — pointedly silent on asombrar, sombra, so-, sa-, and se-lombra, umbrio in his Manual (elemental) de gramática histórica, from its 1904 to its 1941 version — agrees to operate with *SUBUMBRĀRIU as the base of sonbrero in his monumental Cid ed. (Madrid, 1908-11), p. 858; J. Alemany Bolufer, Estudio elemental de gramática histórica3 (Madrid, 1911), p. 33; J. D. M. Ford — who cited assombrar incidentally and noncommittally on p. 106 of his 1897 thesis ("The Old Spanish Sibilants", HSN, VII [1900], 1-182) — had more to say in the Vocabulary to his Old Spanish Readings (rev. ed., Boston, 1911), p. 292ab, favoring SUB + UMBRĀRE > [*]so.ombrar > sombrar, but also reckoning with the influence of sol on solombra ; F. Hanssen, Gramática histórica de la lengua castellana (Halle, 1913), §444, by implication analyzed s- as a reflex of SUB-, not SOLE; J. Huber, Altportugiesisches Elementarbuch (Heidelberg, 1933), §171.1; . Bloch and W. von Wartburg, DÉLF2 (Paris, 1950), p. 422b, s.v. ombreux ("l'esp. sombroso est refait sur sombra 'ombre', luimême refait sur sombrar 'faire de l'ombre', lat. pop. *SUBUMBRĀRE); W. von Wartburg, FEW, XIV (1957-61), 25. —With characteristic restraint, R. J. Cuervo, in his DCR, I (1886), 718b-720a, delineated the syntactic peculiarities of asombrar, while stopping short at decomposing it into a + sombra in the appended etymological vignette. —I owe to the vigilance of J. E. Gillet (see fn. 27, below) my knowledge of M. de Unamuno's modified endorsement of Diez's conjecture: SUBTUUMBRA, an advocacy which deserves mention merely as a matter of curiosity {Ensayos, I I I [Ma drid, 1916], 212 ; SUBTUS- is here a reminder of OProv. sotz·). —On the school of thought operating with su( IL)LA UMBRA (as formulated by J. J. Nunes, Com.pendio de gramática histórica portuguesa3 [Lisboa, 1945], p. 121), see fnn. 28 (last sentence) and 37, below. Salm. sonfria 'umbría' (Sierra de Francia: see J. de Lamano  Beneite) was correctly classed by Meyer-Lübke as a blend with fria 'cold' (fem.). 23 See Breve diccionario etimológico (Madrid, 1961), p. 529a, with interesting provisional datings of first appearances: umbr-ío: 1513; -í: 1739; sombr-ajo: 1495; -aje: 1739; -ear: 1739; -í: 1490; -ilia: 1817 (an adaptation of Fr. ombrelle, 1588). The key sentence runs thus: "La s-, agregada sólo en portugués y castellano, es probable que se deba al influjo de SOL y sus derivados

462

ETYMOLOGY

this position lies in its advocate's ability to trace both s-ombra and sol-ombra to a single deflection from the norm, attributing them to two different degrees — two consecutive spurts, as it were — of the same associative interference. Both major hypotheses, taken in isolation, are quite vulnerable. Diez's idea hinges on the key rôle played by the verb sombrar as the link between *SUBUMBRĀRE and sombra. Now, sombrar is a rare and spottily recorded verb, by no means particularly old ; it could at any time have been secondarily coined as a variation on asombrar (itself a 14th-c. "late-comer") or on sombrear; moreover it is hazardous to project *SUB-UMBRĀRE onto the level of (unified?) Latin on the strength of the reputed Provençal offshoot alone. As for the attractive "Einfall" of relying on the contaminative agency of sol, it so happens that lexi cal polarization in Late Latin and in Romance typically affects the stressed vowel (SINEXTER 'left' for SINISTER due to DEXTER 'right'; *NǑRA 'daughter-in-law' for NŪRUS due to *SǑC(E)RA 'mother-in-law', orig. SOCRUS), occasionally the gender, the final vowel, or the number of syllables, but seldom, if ever, the primitive's opening segment — prefixes excepted. Both major conjectures suffer from yet another, not uncommon ailment : They fail to account for the circumstances that either SUB (TUS) or SŌL exerted so strong an influence in certain Romance languages, esp. in Hispano-Romance, while being utterly ineffectual elsewhere — despite the fact that a merciless sun can be equally singeing in (say) Spain and arid Southern Italy, to say nothing of the universal appeal of a locative prefix. Also, in their most powerful formulations both hypotheses barely conceal their respective spokesmen's tacit belief in "simple causation"; the subtler and more realistic possibility of "multiple causation" has, it would seem, never been seriously weighed. I cannot examine here all facets of the problem or so much as survey the various opinions voiced on Fr. sombre and on the verb sombrer 'to founder' — another etymological enigma — and their remotely possible links to Sp. Ptg. sombra. But I wish to draw attention to a few neglected ingredients of the nuclear issue. It may be appropriate to reopen the discussion by casting an un hurried glance at UMBRA, as it emerges from its Latin context. If we let Ernout and Meillet guide us firmly through the maze of the Latin lexicon, 24 the following picture unfolds : UMBRA is surrounded by a verb, UMBRĀRE (which is flanked by an abstract in -ĀTIŌ and can be nuanced by a gamut of prefixes : AD-, itself the head of a small family, or IN-, or OB-, or PRAE-); by a series of adjectives: in -ōsus, -ĀTICUS (dimin. por ser sol y sombra, solano y sombrío, solear y sombrear conceptos correlativos, opu estos y acoplados constantemente. La variante solombra... comprueba la certeza de esta explicación". For earlier formulations of this view by the same author see "Indianoromanica ; estudios de lexi cologia hispano-americana", EFH, VI (1944), 157f., and DCE, IV [1957], 271δ-274α, esp. 272b: "La alteración de ombra en sombra  solombra pudo nacer al principio de un error: por una especie de trabalenguas, en lugar de decir sol y ombra, se diría sol y sombra, y luego, agravándose la alteración, sol y solombra". (Note that Corominas operates with the assumption of a mechanical slip, accounting for it in strictly syntagmatic terms ; I doubt that in actual speech sol y sombra is a strikingly frequent sequence and am led to suspect the latent, paradigmatic affinity of the two polar opposites, which, to use a felicitous French phrase, psychologically "s'appellent".) 24 See the unrivalled presentation in their D É L L 4 (Paris, 1959-60), p. 745b.


463

-ĀTICULUS in Plautus), -ĀTILIS, a n d -ĀLIS (the latter amenable t o observation through t h e

prism of St. Augustine's adverb in -ĀLITER 'figuratively', while Cassiodorus favored -ĀTICĒ 'in appearance'); b y two neatly differentiated diminutives: UMB(R)-ELLA 'parasol', lit. 'little shade' (Martial, Juvenal), and UMBR-ILLA 'species of fish' (glosses; cf. Gr. σκίαινα); by a derivative suggesting alternately a tool or a place : UMBR-VCULUM ( = σκίας); and by an occasional compound, such as poet. UMBRI-FER 'affording shade'. Five different nuances of meaning can be set off: 'shade projected by a body'; 'shady, secluded spot' → 'Object afford ing shade' → 'shelter'; 'bodiless contour, elusive or fleeing image' (p1. 'the shadows of the dead'); 'uninvited person making his appearance in the company of an expected guest' ( = σκία); 'species of fish' ( = Fr. ombre, ombrine; cf. UMBRILLA, above). 2 5 (Modern ich thyology uses umbrina, a Neo-Latin term, for a widely distributed genus of croakers [fam. Sciaenidae], which includes the European umbra [U. cirrhosa] — a Mediterranean food and market fish.)

To revert to the state of affairs in Hispano-Romance : In the not inconsider able number of pronouncements on s(ol)ombra now available certain issues seem to have been either entirely overlooked or, at least, sidetracked. Here are some examples : (A) Though in the spelling of certain late-medieval texts from Portugal the gemination of vowels often conveys no clearly identifiable message, the fairly consistent use, from the start, of the graphy soombra points to a trisyllabic pronunciation.26 If so, the chances are that the subjacent, proto-Portuguese type was solombra, since -l- is, on good authority, assumed to have disappeared late enough for the hiatus (soömbra)to have been temporarily maintained. This assumption would retroactively consolidate most of the Galician-Portuguese and Asturo-Leonese zones into a single major area — a pattern of territorial dis tribution reassuringly familiar to Romance scholars — and would further satisfactorily explain the contrast between OSp. sombra and OPtg. soombra, which has so far been swept under the rug. A much more finely brushed picture than has so far been available can also be provided of Ast.֊Leon. solombra, including its variants and offshoots, as a result of the unique meticulosity of J. E. Gillet. 27 The traditional form, known 25 On balance the semantic range of UMBRA appears modest. Such latent possibilities of trans fer as are observable in E. (window-, lamp-)shade have not materialized, and there has been no perceptible movement in the direction of shady 'dubious, suspicious'. I t might be rewarding to study cross-culturally and from a single vantage point certain secondary uses of blind 'un acknowledged, previously unknown' (cf. coll. E. blind date, blind carbon copy), ghost (E. ghost writer), phantom (E. phantom ship), and shade/shadow (E. shadow cabinet, shadow boxing). 26 Huber, Altptg. Elem., §337, cites from O Livro de Esopo, ed. J. Leite de Vasconcelos, this pass age : " E veendo a soombra, deytou-se na augua". See also Maria Egipciaca, ed. J. Cornu, fol. 54r°: ssoonbra. Those Old Spanish texts that shun the sol- var. consistently use a single  (Libro de miseria, ed. M. Artigas, 306d: sombra, etc.). The only noteworthy var. is xombra in the racy speech of Moriscos, as observed by early dramatists; cf. E. Veres d'Ocón, "Juegos idiomáticos en las obras de Lope de Rueda", RFE, XXXIV (1950), 195-237 (in particular the section occupying pp. 197-206), at p. 203. 27 See his ed. of B. de Torres Naharro, Propalladia and Other Works, I I I : Notes (Bryn Mawr, Pa., 1951), pp. 125f. (apropos solonbrera in "Romance" III, 23) and 313 (apropos solombrero in "Comedia Trophea", Intr., v. 53). Not unexpectedly, the exegete, despite a crushing congeries of tidily distilled data, failed to arrive at any clear-cut decision of his own or even to indicate his preference when confronted by alternatives. From his comments on certain medieval texts one gathers that he applied to the relation of solombra to sombra a TEMPORAL rather than REGIONAL yardstick.

464

ETYMOLOGY

from medieval sources and from Judeo-Spanish, 28 seems to have been uni formly solombra, and this archaic variety was also the one initially discovered by pioneer philologists and antiquarians. 29 As has been independently estab lished with other slices of comparable material, there has for centuries operated in Hispano-Romance a gentle dissimilatory mechanism tending to shift o-ó sequences to either e-ó ora-ó; the classic illustrations of this trend are ROTUNDU 'round' > Sp. redondo and SUCCUTIT 'he shakes from beneath, flings aloft' > Sp. sacude (Ptg. and proto-Sp. sacode).30 Once we heed this tendency, we shall hardly be surprised to learn that in modern Asturo-Leonese dialects and in their prongs and prolongations the once solid domain of solombra has disintegrated into much smaller entrenchments of : (a) salombra, (b) selombra, and (c) residual so- beside su-lombra. (On the other hand, Jud.-Sp. asolambrarse strikes one as a spurious formation, a ghost-word.31) Here is a skeleton outline of the record: (a) Salombra, the least frequent of the three, has been recorded in the Astorga and La Maragatería zones by S. Alonso Garrote. 3 2 28 For Jud.-Sp. solombr-a and -oso see the excerpts from an Old Testament translation (Con stantinople, 1873) adduced by M. Gaspar Remiro, "Vocablos  frases del judeo-espanol" (II), BRAE, V (1918), 350-364, at 361 (Gen. 19:8, Judges, 9:15, Is. 4:6, 18:1, etc.); also M. L. Wagner, Beiträge zur Kenntnis des Judenspanischen von Konstantinopel (Wien, 1914), §134, and, for Saloniki, C. M. Crews, VR, XII (1951), 195. Interestingly, among Sephardic Jews trans planted to Rumania the compromise form ombra has been introduced in deference to Rum. umbră; see Crews, Recherches sur le judéo-espagnol dans les pays balkaniques (Paris, 1935), p. 187. Gillet cites R. Fouché-Delbosc's collection of "Proverbes judéo-espagnols", No. 795, in RH, II (1895), 312-352. As regards Old Leonese texts, we have the testimony of the Libro de Alexandre (MS 0), already scanned by Morel-Fatio; checked against R. S. Willis' composite edition (1934) and adjusted to the new numbering of stanzas, the key-passages read: "Las alas espandidas por fazer solombra maor" (862c ; P : por fer sonbra mayor) ; "Que por buena solombra, que por la fontana / ally venién la[s] aves..." (939ab ; Ρ : que por la buena sonbra); "quando el sol va so la ribera / ela solombra de la tierra es comedianera / ..." (0,1179bc ; Ρ : la sonbra de la tierra yes entre medianera); "tanto puedc su solombra, nol auriedes sabor" (1979c ; P : tanto que de su sombra non avriades sabor) ; [speaking of Queen Calectrix] "trahe solombrera tan mansa e tan queda" (1875c; P : fazié una sonbriella tan mansa...); "por veer mås lexos tollían los solonbreros" (267d i P : por veyer mas alexos tolliénse los sonbreros). Other revealing texts are : Historia troyana en prosa  verso; texto de hacia 1270, eds. R. Menéndez Pidal and E. Varón Vallejo (Madrid, 1934), pp. 141 (twice) and 186 ; and the Poema de Alfonso XI, ed. J. P. ten Cate (Madrid, 1956), st. 1683d ('producto de la imaginación') and st. 2266d ('amparo, cobijo'), cf. the editor's Estudio preliminar y vocabulario (Amsterdam, 1942), p. 110. The word reappears in Gil Vicente's "Auto do Viúvo", cf. the relevant comment by A. dos R. Gonçalves Viana, Palestras filolójicas2 (Lisbon, 1931; orig. ed., 1910), pp. 258f. 29 Cf. E. Gessner, Das Leonesische; ein Beitrag zur Kenntnis des Altspanischen (Berlin, 1867), p. 34; and A. Morel-Fatio, "Recherches sur le texte et les sources du Libro de Alexandre", Rom, IV (1875), 7-90, at 50. Before them R. Cabrera cited MS  of the Alexandre; Diez, in his dic tionary, drew his readers' attention to a passage in J. N. Böhl de Faber's Teatro espanol anterior a Lope de Vega (Hamburg, 1832), p. 83. 30 On this moot point of phonology see my two articles: "The Word Family of Old Spanish recudir", HR, XIV (1946), 104-159, esp. 130-137, and "The Etymology of Hispanic restolho, rastrojo, rostoll ['stubble']", RPh, I (1947-48), 209-234. 31 As many as four times Gaspar Remiro, "Vocablos  frases..." (II), BRAE, I I I (1916), 498-509, at 507, so transliterates the verb (e.g., Dan. 4 : 9 : "Debajo de él se asolambravan las animallas"), but the Hebrew letters reproduced clearly point to the expected asolombrar. (The author, incidentally, supports Diez's etymology.) 32 For salombra my sole source of information is S. Alonso Garrote, El dialecto leonis hablado en Maragateria y tierra de Astorga2 (Madrid, 1947), p. 318 (without localization). —Generally speaking, analysts reconstructing the map from written dialectological records are hampered by the fact that most collectors of regionalisme have listed deviations from, but not coincidences


465

(b) Selombra was identified in the Leonese enclave of Miranda del Duero — politically a part of Portugal — and again in the sections of L e ó n explored by Alonso Garrote. 3 3 (c) [α] — Solombra has once again been observed in living speech by Alonso Garrote and by Rato  Hevia and Lamano  Beneite as well. Its all-important twin offshoots are : on the one hand, solombrio (Old Portuguese and Old Judeo-Spanish ; Correas' refranero ; mod. Salamancan, Leonese, West-Asturian: 'shady place'); on the other, solombria (H. Núñez's Refranes : 'id.'). A rival formation for 'shady place' has been solombrera (Enzina) or solumbrera, while solombrero 'hat, headgear' is a regionalism of conspicuously old standing. There are stray vestiges of adj. solombrero.34 [β] — Sulombra is characteristic of Western Asturias (Å. W. Munthe) and, a long stretch to the south, of Sanabria ; sulombre has also been recorded in Yugoslavia's Monastir dialect of Judeo-Spanish, which is equally hospitable to sulumbreru 'umbrella, parasol'. 3 5

(B) Without denying the possible link to sol (dramatized — with varying arguments and in different perspectives — by Covarrubias, Cabrera, Rato  Hevia, Alonso Garrote, 36 and Corominas) and without waiving Diez's appeal to the intercalary definite article (on the analogy of Fr. à l'ombre), a motion with, the Castilian usage. Thus, there is a strong presumption that sombra prevails in Santanderino, on the strength of sombr-iego 'sombrío' (J. Calderón Escalada, "Voces .. de Santander". BRAE, XXV [1946], 379-397, at 395); that it has struck root in El Bierzo, on the evidence of the local phrase (used by carpenters) poner el sombrero 'to start putting on or fitting the top' (V. Garcia Rey, Vocabulario del Bierzo [Madrid, 1934], p. 146); and that it has entrenched itself in many sections of Asturias proper, witness sombrao — Cast. sobra(d)o 'garret' (used for storing potatoes and all kinds of junk ; lit. 'dark room'?), best explained as a blend of sobra(d)o and sombra (L. Rodríguez-Castellano, La variedad dialectal del Alto Aller [Oviedo, 1952], p. 248; J. Neira Martinez, El habla de Lena [Oviedo, 1955], §§25, 79). But these are mere guesses. 33 Alonso Garrote, pp. 319f., cites the phrase estar a la selombra, musing : " resguardarse de los rayos del sol", and pinpoints on the map (Silván) selumbreiro 'sombrero'. An older record is J. Leite de Vasconcelos, Estudos de filologia mirandesa, II (Lisboa, 1901), 271, confirmed by Gon çaives Viana, loc. cit., and reëxamined by Gillet. 34 Solombra has, by all odds, remained the dominant form. I t has been observed by fieldworkers (Alonso Garrote, p. 323, with a proud side-glance at the Alexandre) and put to use in regionally colored fiction (e.g., by Concha Espina de la Sema in her novel La esfinge maragata; see J. Alemany [Bolufer], BRAE, I I I [1916], 45 ("la solombra gustable") and 61 ("¿Por qué no busca la solombraV). In all likelihood, Salm. solibrigada 'shelter' and solóbrigo 'worn, stale', fig. 'cunning, underhanded' (J. de Lamano y Beneite, El dialecto vulgar salmantino [Salamanca, 1915]. p. 634), involve crosses with abrigada and lóbr-ego,dial.-igo 'murky'. Note the eccentric (etymolo gizing?) spellings soV ombra, soV ombrerus 'sombrero[s]' in A. de Rato y Hevia, Vocabulario de las palabras y frases babies (Madrid, 1891-92), p. 112a. Lamano, pp. 634f., also lists the adj. solombrio 'shadowy', used as a masc. or a fem. substantive in proverbs already caught from the lips of rustics by G. Correas ; see the latter's Vocabulario de refranes, pp. 214 ("Ni trigo de valle, ni leña de solombrio lo vendas a tu amigo") and 396 ("Por San Matías, cantan las cotovías y entra el sol por las solombrías" : 'lugares baxos y sombríos'; Gillet traces this item to H. Nunez). There exists near-by a var. solumbrío ; the phrase al solombrio matches Cast. a la umbría. In his lavish documentation Gillet includes adj. solombrero ("una zarza solombrera": Juan de Linares, A.D. 1573) and its substantivized fem. counterpart: "Vente a estas solombreras" (J. del Enzina, Cancionero, 1496) and "So las verdes solumbreras" (Anon.). 35 Sulombra, a fairly rare and "accidental" variant (from metathesized solumbra?), seems to be thinly scattered over the entire Asturo-Leonese zone rather than concentrated in any clearly bounded area. Gillet, in his austerely factual inventory, lists the testimony of Å. W. Munthe, Anteckningar om folkmålet... (Uppsala, 1887), p. 89, and F. Krüger, El dialecto de San Ciprián de Sanabria: monografia leonesa (Madrid, 1923), p. 128a, beside Jud.-Sp. (Monastir) sulombre and s(ul)umbrero 'parasol' (M. A. Luria, RH, LXXIX [1930], 542). Krüger, p. 68, fails to ask himself whether the incipient displacement of native llume 'light, fire' by Cast. lumbre (transitional form : llumbre) could have been prompted by preëxistent solombra. 36 Solix is clearly a misprint for SŌLIS, a slip which, piquantly enough, goes back to A. de Rato y Hevia's Vocabulario, ρ. 112α. Conversely, Alemany favors SUBUMBRĀRE ( B R A E , III, 45). while Lamano wavers between the two possibilities (p. 635).

466

ETYMOLOGY

seconded by Gonçalves Viana and his followers,37 one may aver that it was, primarily, the tendency toward consonant dissimilation which must have prompted speakers to rid SUBUMBRĀ of its first b, by far the more vulnerable of the two on account of its immediate environment and its position vis-à-vis the word's accentual peak. The excellent contributions to the elucidation of this process so far made by M. Grammont, R. Posner, and K. Togeby (among others) have been predominantly cross-linguistic,38 with the result that we do not know with any degree of accuracy which languages and which dialects responded most fully to this sort of challenge. The last two experts have separately debated the issue of Romance imperfects, a tense in whose history HAB-ĒBAM, DĒB-ĒBAM (note the b-b sequence) may have played a prominent rôle; 39 but if in their case a different solution, namely the dissimilatory loss of the second 6, occurred, it must be remembered that the two starting points showed only limited resemblance. Generally, there can be no doubt of the major services performed by l in Spanish dissimilatory processes.40 Granted, on principle, that the intensity of response may have varied greatly 37 In a short and rather lame post-script (ZRPh, XLI [1921-22], 757) to an earlier study J. Brüch traces OProv. solombrar to sob l'ombra — a return to Diez and Gonçalves Viana (see the concluding sentence of fn. 28, above) —, assigns sotzombrar to a later stratum, and analyzes Fr. sombre, Sp. Ptg. sombra, Engad. sombriva, Vegl. sombreya as so many products of haplologized VLat. *S(UB)UMBRA. This sharp about-face marks the abandonment of a view he held only a year or two before ("Zu Meyer-Lübkes etymologischem Wörterbuch", ibid., XL [1919-20], 641-654, at 652f.), according to which Sp. s(ol)ombra and Occ. souloumbrado (A. Mathieu, from Châteauneuf du Pape, Dépt. Vaucluse), soulombre (C. Rieu, from Paradou, Dépt. Bouches-du-Rhône), souloumbreja (L. Achard, from Marseille), and souloumbru (M. Bourelly, from Aix; all four authorities cited by F. Mistral) involved an amalgam of SXJBUMBRĀ and SUBLŪSTRE 'dim'. 38 For full bibliographic data, including a digest of reactions — staggered over a half-century — to Grammont's original monograph and "follow-up" studies, see K. Togeby, "Qu'est-ce que la dissimilation?", RPh, XVII (1963-64), 642-667. Critical reactions to Posner's dissertation (Consonant Dissimilation in the Romance Languages [Oxford, 1961]) include: F. J. Barnett, FS, XVIII (1963), 85-87; J. L. D[ambiertmont], MaR, XIII (1963), 207f.; G. Gougenheim, BSLP, LVIII:2 (1963), 95f.; X. Mignot, RLaR, LXXXV (1962-65), 87-89; J. Mondéjar, ZRPh, LXXXI (1965), 394-401; L. Nauton, RLiR, XXVII (1963), 212-215; A. Rosetti, Phon, XII (1965), 25-28; W. Rothwell, ArL, XV (1963), 97-99; H. G. Tuchel, RF, LXXV (1963), 169171 —all favorable. While these assessors place the monograph in a succession of interesting perspectives, they fail to cast light on the specific problem at hand. See further H. M. Hoenigswald, "Graduality, Sporadicity, and the Minor Sound Change Processes", Phon, X I (1964), 202215. 39 The claim of a dissimilatory loss of  in HAB-ĒBAT, DĒB- entailing the reorganization of the Romance imperfect was initially pressed by R. Thurneysen, Das Verbum "être" und die französische Konjugation (Halle, 1882), pp. 30-32. In recent years, the entire issue has been re opened through R. Posner's provocative paper, "The Imperfect Ending in Romance", TPS, 1961, pp. 17-55, which views the fall of intervocalic -b- as normal, attributing the triad It. -ava, -eva, -iva to morphological pressure, while her critic K. Togeby, "Les désinences de l'imparfait (et du parfait) dans les langues romanes", StN, XXXVI (1964), 3-8, visualizes a transfer of endings from the conditional to the imperfect; cf. Posner's spirited rejoinder: "Romance Imperfect and Condi tional Ending — a Further Contribution", ibid., XXXVII (1965), 3-10. Whichever hypothesis proves more persuasive, what matters to us is the instability that must have surrounded, at the critical juncture, VbV — a state of affairs practically inviting saltatory changes, such as con sonant dissimilation at a distance. 40 There is hardly any need to pass in review the numerous Spanish words containing l as a dissimilatory substitute, typically for either n or r. Medieval examples cited as early as E. Gorra, Lingua e letteratura spagnuola delle origini (Milano, 1898), §113 — characteristically, under "Acci denti generali", in the tradition of Ascoli — include ableviar 'to abbreviate', alambre 'wire', Antolín(ez), árbol 'tree', Bernalt, calonge 'canon' (title), celebro 'brain', mármol 'marble', miércoles


467

from one regional dialect to another, it is little wonder that both in Hispania and in Gaul there should have occurred cleavages, pitting (a) Sp. sombra against Leon. solombra, OPtg. soömbra and (b) Fr. sombre (adj.) against dial. (Dauph.) solombrar (v.), Occit. souloumbrous (adj.). The particular choice of l-b as an escape from the unattractive sequence b-b may, let us admit, have been motivated by the engaging association with sol ; alternatively, one may point to antihiatic l often intercalated in the vicinity of an  (Fr. hugo-l-esque, -âtre, after idol-âtre; congo-l-ais and togo-l-ais, echoing angol-ais, etc.). 41 (C) Adequate attention has not as yet been paid to the fact that, in addition to Cat. ombra, the adj. umbrío 'shady' and the corresponding noun umbría 'shade, shady place' (which jointly constitute perfect matches for sombr-io and -í) represent relics of an older usage, which involved the product of UMBRA devoid of any accretion. While Sp. umbroso, a poetic word, may be dismissed as a learnèd formation deliberately imitative of elegant Latin precedents, 42 umbrio cannot under any circumstances be classed as a ''cultismo", because no -īvus derivative is detectable in either Classical or post-Classical Hterature, even though, on the evidence of Romance vernaculars, *UMBRĪVUS may very well have thrived in provincial ramifications of colloquial Latin. 43 Umbrio makes its ' W e d n e s d a y ' , pelegrino 'pilgrim', roble 'oak', a n d tórtola ' t u r t l e d o v e ' . On t h e special difficulties besetting Old Spanish words ending in -er a n d -el see A. K . Levy,"Contrasting D e v e l o p m e n t ( s ) in H i s p a n o - R o m a n c e of Borrowed Gallo-Romance Suffixes", RPh, X V I I I , 3 9 9 - 4 2 9 ; X X , 296-320. I n t h e c o n t e x t of this flux a sudden " l e a p " from SŬB-ŬMBRĀRE to *solombrar becomes less s t a r t ling, especially if allowances are m a d e for c o n c o m i t a n t factors (attraction b y sol ; t h e new definite article worming its w a y into this p a r t i c u l a r verb). 41 E v e r since t h e late 'fifties I have, in different contexts, emphasized t h e need to reckon w i t h t h e joint agency of compatible forces in language c h a n g e ; see, in addition t o t h e inquiries i n t o diachronic hypercharacterization a n d into interfixes, t h e t w o m u t u a l l y related p a p e r s : "Multiple Versus Simple Causation in Linguistic C h a n g e " , To Honor Roman Jakobson (The H a g u e , 1967), I I , 1228-46, a n d "Linguistics as a Genetic Science" [LSA Presidential Address, 1965], Lg, X L I I I (1967[-68]), 223-245. 42 R a t h e r typically, Sp. umbroso is in far closer p r o x i m i t y to t h e L a t i n p r o t o t y p e t h a n are, say, F r . ombreux, OProv. ombros, I t . ombroso; cf. t h e behavior of S p . -encia < -EXTIA as against P t g . -ença, P r . -ance, P r o v . -ensa, I t . -enza. T h e gulf between learned a n d vernacular words, t h o u g h n o t unbridgeable in Spanish (cf. iglesia), has in fact seldom been bridged — in m a r k e d contrast t o t h e situation in P r o v e n ç a l a n d I t a l i a n . Other learnèd borrowings, of peripheral i m p o r t a n c e t o our problem a n d , for t h e most p a r t , ephemeral, include umbra itself, plus derivatives in -áculo (along side v e r n a c u l a r sombr-ajo, -aje — t h e l a t t e r t h r o u g h c o n t a m i n a t i o n with a borrowed t r a n s P y r e n e a n reflex of -ĀTICU : ' s o m b r a que hace u n o poniéndose d e l a n t e de la luz y moviéndose de modo que estorbe al que la necesita', cf. Sp. vent-aja : P t g . vant-agem), -al, a n d -átil, quite a p a r t from umbela, umbelífero, a n d penumbra. 43 T h e -ío suffix a t issue is t h e same as t h a t exhibited b y bald-í 'untilled, fallow, (fig.) idle', brav-io ' u n d a u n t e d ' , cabr-ío 'pertaining to g o a t s ' , corrent-io a n d manant- ' r u n n i n g (of w a t e r ) ' , P t g . doent-io 'ailing', estant-io ' s t a t i o n a r y , spiritless', obs. rad-io ' s t r a y , gone a s t r a y ' (from ERR ARE ; cf. P t g . errad-io ' v a g r a n t ' ) , P t g . laurad-io 'fit for tillage', regad-ío 'irrigated (land)', sequ-io ' d r y ' , tard-io ' l a t e ' , vac-ίο ' e m p t y ' — beside alt-ivo ' h a u g h t y ' . See Meyer-Lübke, Romanische Formenlehre (Leipzig, 1894), §497 ; Hanssen, Gramatica histórica, §285 (who isolates est-io, tard-io. a n d vac-ίο as t h e kernel); a n d m y own article in Lg, X V I I (1941), 99-118. F o r a full range of deposits o f * U M B R - Ī V U , esp. of its substantivized feminine, see FE W, X I V , 25 : V. Anz. umbria, O L o m b . ombria (AGI, X I I , 417), OGen. umbria (ibid., X I I I , 400), OVen. onbria (Mussafia, Beitrag...), OTrev. umbria, Servigliano umría, Surs. umbriva, Friul. umbrie, S p . umbria (beside Segov. ombría, which m a y be a compromise form). T h e coexistence, in W e s t e r n R a e t o R o m a n c e , of umbriva (Surselva) a n d sumbriva (Upper E n g a d i n e ) , which parallels so closely t h e rivalry between Sp. urn- a n d som-bria, m a k e s one wonder w h e t h e r s- was n o t originally welded on t o t h e UMBR- nucleus within this s u b - b r a n c h : *SUB-UMBRĪVUS '(growing. thriving, sheltered) in

468

ETYMOLOGY

appearance in post-medieval texts, but intrinsically presupposes an unbroken flow of mouth-to-mouth transmission; indirectly, its survival testifies to the fairly late consolidation and even later ultimate triumph of s(ol)ombra. (D) Can we pinpoint the cradle of umbr-ío, -í on the map of the Peninsula? Not with any degree of accuracy in the case of words so long, on account of their humility, shielded from direct observation ; but note that the use of  rather than  as a reflex of Lat. ŭ, esp. before a consonant cluster, points to the West rather than the Center; contrast Ptg. chumbo 'lead', curto 'short', funda 'sling (for hurling stones)', surdo 'deaf' with Sp. plonio, corto, honda, sordo.44 Incidentally, I am tempted to rescue from unfair oblivion, if only on a reduced scale, the connection between Sp. umbrai 'threshold' and UMBRA established by the pioneers;45 while it is correct that umbrai fundamentally reflects LĪMINĀRE 'pertaining to the doorway' (LĪMEN -INIS) and while it is true that deglutination of l- (mistaken for the definite article) and labialization of the unstressed vowel in the opening syllable marginally suffice to account for the shift i > u,46 I (lit. 'under') t h e s h a d e ' , spreading t h e n c e t o t h e n o u n , a n d only a t t h e concluding stage (Sp. asombrar) soldered to t h e verb. This coincidence c a p t u r e d t h e imagination of Ascoli, who in his "Saggi ladini", AGI, I (1873), 111, n. 3, p a r e n t h e t i c a l l y r e m a r k e d : "Cfr. nello spagn.: sombra, sombría, sombrío ; è notevole che il s iniziale riappaja nell'engadin. sumbriva = umbriva soprasilv.". Note t h a t t h e s e m a n t i c scope of ' u n d e r - ' s t r e n g t h e n s t h e c a n d i d a c y of t h e adjective for t h e posi tion of a leader w o r d ; the v e r b r a t h e r harmonizes with 'over-' (cf. E . overshadow, like overarch; G. überschatten). 44 These — carefully filtered — examples h a v e been culled from  v o n R e i n h a r d s t o e t t n e r , Grammatik der portugiesischen Sprache auf Grundlage des Lateinischen und der romanischen Sprachvergleichung (Strassburg a n d London, 1878), p p . 49f., a pioneer whose analysis is n o t in frequently distorted by erroneous etymologies (gume ' s h a r p edge' < ACŪ MEN, n o t — as he hypothesized — CULMEN), u n w a r r a n t e d a p p e a l t o " c u l t i s m o s " {culpa, culto, curso, duplo, justo, mundo, nulo, núpcias, rústico, tumulto), a n d o t h e r fiaws. N o t e t h a t t h e reverse distribution of vars., t h o u g h rare, can be d o c u m e n t e d : P t g . doce ' s w e e t ' : O S p . (predominantly) duçe < DULCE, Ptg. gósto ' t a s t e ' : S p . gusto < GUSTU, to say nothing of Arabisms (Ptg. enxofre : Sp. azufre 'brim stone' < Arabieized SULPHUR, P t g . enxoval :OSp. axuar 'bridal apparel, trousseau' < Ar. AŠŠUWĀR). I n some instances Spanish a n d P o r t u g u e s e jointly w i t h s t a n d t h e t e m p t a t i o n to shift Ŭ t o  : Sp. una, P t g . unha 'nail' < UNGULA, S p . sulco, P t g . surco 'furrow' < SULCU ' t r e n c h , ditch, t r a c k ' . N o t for n o t h i n g does R. Posner r a n k t h e reflexes of L a t . Ŭ as a choice e x a m p l e of the inapplicability of a n y rigidly n a r r o w conception of " s o u n d l a w " ; see her lively a n d entertaining review article, "Positivism in Historical Linguistics", RPh, X X , 321-331, a t 323f., w i t h illustra tions d r a w n , for a change, from F r e n c h (fn. 8). T h e W e s t e r n roots of Sp. umbrio are confirmed by (???al. umbrio, as attested by J . A. Saco Arce, Gramática gallega (Lugo, 1868), p . 2 9 1 . 45 One older a u t h o r i t y championing this view was t h e 11th ed. (1869) of t h e Madrid A c a d e m y ' s dictionary (p. 773, s . w . umbrai a n d umbralar ' t o lintel, place a n a r c h i t r a v e ' ) . 46 T h e discovery of such relatively u n c o n t a m i n a t e d medieval forms as limnar soon led t o t h e correct etymology ; cf. A. Zauner, Altspanisches Elementarbuch (Heidelberg, 1909), §74. T h e alterna tion of -al and -ar causes slight difficulty ; with t h e deglutination of l- t h e prime reason for favoring -ar — namely, consonant dissimilation — was eliminated a n d -al prevailed unopposed. F o r parallels see A n i t a K. L e v y ' s aforementioned perceptive m o n o g r a p h . T h e apheresis {l)umbral was briefly examined by R. J . Cuervo, " T e n t a t i v a s etimológicas", Rom, X I I (1888), 105-112, a t 108f., in t h e c o n t e x t of Sp. {l)amia 'shark', {l)atril, 'lectern', Cub. (l)ante-juela 'spangle' ( = Sp. lentejuela), Bog. {l)amedor 'licker, s y r u p ' a n d {l)imbo 'limb(o), h e m , edge' — t h e l a t t e r also twice represented in t h e Cancionero de Baena. Prior to t h e deglutination of /-, as G. Baist has convincingly shown ( " E t y m o l o g i e n " , ZRPh, V I I [1883], 115-12", a t 124, §28), lumbral ( = P t g . lomear) established a t e m p o r a r y c o n t a c t with LŪ MINĀ RE 'window'; t h e special denotation 'an opening, a light (in t h e building)' of LŪ MEN survives in Gal. sobrelume ' t h r e s h o l d ' = Sp. d-, l-intel. B y t h e t i m e of Golden Age literature t h e s t a n d a r d form was already umbral; cf. F r a y H . Paravicino, cited in R F E . X X X I I I (1949), 3 9 3 : " a l pie de t u s umbrales" (ed. J . M. Blecua).


469

should think that both formally and semantically an amalgam of *umbra and limn-ar, -al provides by far the more satisfying explanation. (E) It is unprofitable to start from the premise that the contamination of the readily identifiable umbr-ļombr- core with that less sharply profiled ingredient which in the end has yielded the s- increment (sol-, sal-, sel-, etc.) must neces sarily have occurred in — hierarchically — the very center of the UMBRA family : Not infrequently changes of this sort begin at the periphery, affecting at first an exposed derivative, then spreading gradually to other members. 47 Perhaps this is what Covarrubias meant when he rather awkwardly implicated SŌLIS in the crystallization of sombr-ero rather than of sombra. More important, it was, quite explicitly, Diez's position in inferring *SUBUMBRĀRE from OProv. sotzombrar — true to a pioneering comparativist's romantic stance which, rather indiscriminately, projected Old Provençal features (many local innova tions included) onto the level of Vulgar Latin or Proto-Romance and, in a similar vein, wilfully construed all manner of Sanskrit idiosyncrasies as mere survivals of Proto-ΙΕ usage. In a more critical mood, we may argue that occasional and sporadic combina tions of SUB- plus sundry members of the UMBRA family occurred at scattered points of the far-flung early Romance territory. AD-UMBRĀRE (and, before their extinction, OB-UMBRĀRE, etc.) may very well have given rise to SUB-UMBRĀRE,48 authenticated as a rare verb by Rönsch 48a and conceivably transmitted into Provençal through direct channels, except for the local substitution of fullbodied suBTUS- for leaner SUB-. The provincial Latinity of Spain may have followed a slightly different course ; the chances are that here, as in Western Raeto-Romance, *(SUB)UMBR-ĪVUS acted as the prime focus of a change which later engulfed the noun and ultimately dragged in the verbs sombr-(e)ar and a-sombr-ar.49 As regards the delicate situation in French, the adj. sombre could pass off as postverbal (cf. comble 'full, overflowing, heaped up to the top', from combler < CUMULĀRE 50 ), provided one grants that sombrer, semantically, repre sents but a wreckage of its originally wider scope ('to cast shade, gloom, dark ness' > 'to sink into darkness'?); alternatively, sombre might be pronounced a 47 See my note "Etymology and the Structure of Word Families", Word, X : 2-3 (1954), 265-274 ( = Linguistics Today, eds. A. Martinet and U. Weinreich). 48 Cf. the rise of *AP-, *EX- or *DIS-, and *SUP-PEDĀRE in Hispano-Romance, on the analogy of RE-PEDĀRE 'to trudge back', as pieced together in Section I of my Studies in the Reconstruction of Hispano-Latin Word Families, UCPL, XI (1954), 1-22, and 65-95; cf. D. Catalan's valuable elaborations and corrections in RFE, X X X I X (1955), 412-442. 48a To the two passages from archaic Bible translations adduced by that scholar in 1879 add the datum cited by W. Heraeus, "Beiträge zu den Tironischen Noten", A L L G , XII (1902), 49. 49 UMBRĀRE 'to shade, overshadow' has left numerous vestiges in Romance (REW3 §9048 lists Rum. umbrà, It. ombrare [whose semantic spectrum includes 'to shy' (intr.), 'take offense, be suspicious'; cf. Fr. ombrageux, Sp. a-sombr-ar — initially reflexive], Fr. ombrer, Prov. ombrar; also, indirectly, E. to take and give umbrage); but precisely in Hispano-Romance sombrar is repre sented quite marginally, while sombr-ear, more firmly rooted, is patently denominal and usually refers to the incipient stage of hirsuteness (cf. coll. E. 'five-o'clock-shadow'). 50 For details see the second installment of my article "Fuentes indigenas y exóticas de los sustantivos y adjetivos verbales en -e", RLiR, XXIV (1960), 201-253.

470

ETYMOLOGY

back-formation from *SUB-UMBR-ĪVU > sombrif. The ultimate decision here lies with seasoned experts in Gallo-Romance. (F) No systematic inquiry into the use of prepositions governing UMBRA seems to have been made in an effort to vindicate the inference of *SUBUMBRĪVU. Latin writers normally conjoined IN with UMBRĀ (cf. G. im Schatten, E. in the shade, Sp. en la sombra, etc.): vacua tonsoris in umbra 'in the cool barber's shop' (Horace, Ep. 17.50); ignaua Veneris cessamus in umbra (Ovid, Am. 2.18.3); Pompeia spatiabere cultus in umbra 'in the Pompeian portico' (Propertius, 4[5].8.75); studia in umbra educata 'in the closet' (Tacitus, A. 14.53). But we find traces of the phrase SUB UMBRĀ, esp. in figurative use — repeatedly in Livy : Sub umbra foederis aequi seruitutem pati (8.4.2); sub umbra auxilii vestri latere uolunt (32.21.31); sub umbra Romanae amicitiae latebant (34.9.10); occasionally in Justinian : morum uitia sub umbra eloquentiae primo latebant (5.2.7). A similar situation prevails in several other languages. Italian usage favors alVombra della legge 'under the shelter of the law', nato alVombra della cupola di San Pietro 'a Roman born and bred', stare alVombra 'to be in the shade', nell'ombra 'secretly, furtively', but recognizes sotto l'ombra dell'amicizia 'under the guise of friendship', reminiscent of Livy's construction. I n French, the norm is represented by dans l'ombre de la nuit, de la paix (Corneille), ...du secret (Racine ; but the same poet allowed for assise à l'ombre des forêts), ...du mystère (Voltaire), not to forget the Proustian title A l'ombre des jeunes filles ; but La Fontaine, a shade more free-wheeling and spontaneous, wrote in his Philémon et Baucis : Que des époux séjournent sous leur ombre ! R. ten' goes with v 'in': v teni (Pushkin), v cërnoj teni (Α. Ν. Tolstoy), the chief peculiarity of t h a t language being the use of a straight genitive for the source of the shadow only on the figurative level (ten' Nazona, Dania; proslogo, obid, somnen'ja) —otherwise with ot 'from': ot derev'jev 'from the trees', etc. But poetic usage admits a rival word, sen', lit. 'cover', fig. 'protection', which harmonizes with the prep. pod 'beneath' and has been used very much like SUB in connection with trees, woods, etc. ; cf. Lermontov's pod sen'ju zakona 'under the shelter of the law', which exactly matches Sp. amparo.

On balance, the pattern SUB UMBRĀ was an ever-present possibility through the telescoping into a single concept of the shade produced and the cover (roof, foliage) affording that shade. We do, incidentally, have proof of the residual adverbial use of Ast. solombra: M. J. Canellada, El bable de Cabranes (Madrid, 1944), p. 340, translates it 'bajo la sombra, a la sombra'. But if SUB-UMBR-Ā, -ĀRE, *-īvu were latently possible anywhere at the time of the crumbling Roman Empire — why did speakers avail themselves of this possibility in Spain and Portugal, in selected portions of France (the FrancoProvençal zone included), in Grisons ; but not in Catalonia, not in Italy, not in Rumania? This is, indeed, the crux of most, if not all, explanatory analyses in historical linguistics; fortunately, we are no longer entirely helpless in facing this kind of embarrassing question. The correct approach is to ask oneself what isolable feature could possibly have made sombra more attractive to speakers than ombra. Twelve years ago Corominas, avowedly inspired by the teaching of J. Jud, took one step in the right direction by surmising that in Old Spanish 51 In one of the least carefully heeded of his pronouncements Gilliéron himself insisted that the study of friction between near-homonyms remains to be undertaken ; see his (and M. Roques') Études de géographie linguistique d'après V "Atlas linguistique de la France" (Paris, 1912), p. 11 :


471

the irksome near-homonymy of ombre < HOMINE 'man', ombro < (H)UMERU 'shoulder', and *ombra/*umbra < UMBRA 'shade' — a homonymy lien to cognate languages — may have acted as a powerful deterrent to the continued use of the unadulterated form (DCE, IV, 2726).51 While Corominas' hypothesis is commendable, it does not exhaust the forces of causation worth canvassing ; the propensity toward consonant dissimilation, which we have here held respon sible for the shift SUBUMBRĀ(RE) > solombra(r), may have varied from language to language — an idea Alf Sommerfelt toyed with shortly before his death. Since a noun phrase is involved, the speakers' tendency to dissimilate consonants may in turn have been controlled by the character of the juncture between preposition and noun as well as by conditions of stress. Finally, the fact that major groups of speakers, confronted with the choice between ombr- and umbr-, opted for the latter (cf. umbrío) confirms the undesirability of the former. Interestingly, the retreat from homonymy lay either in the direction of  in preference to o, or in the addition of s-, but not in any combination of both resources; the theoretically conceivable var. sumbr- has nowhere crystallized, except where the shift  >  is clearly secondary.52 VI. Conclusion. Our modest exploratory tour has led us to examine jointly an unusual variety of the less than common coalescence of preposition (at a later stage, prefix) and noun. The variety is characterized by the fact that the newly emerging compound tends to replace the underlying primitive — either entirely, as in the cases of sombra and, on a severely local scale, of emberano, or in a particular semantic sector and on a specific stylistic level, as with poet. ensueño. The process is, I repeat, exceptional and presupposes a concatenation of un common circumstances. Sp. sueño suffers from a potential excess of polysemy, produced by a transparent homonymic collision. Verano, in its semantic relation to i(n)vierno, lies open to the inroads of lexical polarization, particularly on account of a whimsey : the accidental formal resemblance of the two words. In the closest vicinity to *om-, *um-bra, there presumably existed, from time immemorial, such derivatives as sombrío 'thriving, sheltered IN a shady spot, "La forte ressemblance peut jouer dans l'histoire d'un mot un rôle égal ou presque égal à celui de l'homonymie parfaite". Some relevant hints will be found in my twin inquiries, "Studies in Hispano-Latin Homonymics...", Lg, XXVIII (1952), 299-338, and "A Cluster of Four Homo phones in Ibero-Romance", HR, XXI (1953), 20-36, 120-134. Another elusive point, which even Gilliéron, with all his flair for complexities, did not suspect, but which K. Jaberg — by far the most balanced and theoretically oriented dialect geographer of his generation — once darkly alluded to, is the coincidence between lexical innovations and frontiers of "hazard-laden" sound shift : While it is true that in some cases the territorial correspondence is perfect (and it is this circumstance, dramatized by cartography, that made Gilliéron's discoveries half a century ago so irresistibly sensational), it is very likely that in most instances the "match" is only approximate ; see Aspects géographiques du langage (Paris, 1936), pp. 46f. In other words, the area marked by the collision of HOMIN-, (H)UMER-, and UMBR- need not represent more than the center of the zone in which sombra prevailed over pristine *ombr-¡umbr-. 52 Among remotely comparable instances of coalescence of so + NOUN one may cite sotámbaro (involving pre-Rom. *támaro) and dial. It. soitavina {ID, XII, 140) as interpreted by J. Hubschmid in BF, XII:2 (1951), 138f.

472

ETYMOLOGY

UNDER the foliage of a tree', sombrar 'to seek (or grant) shelter from scorching heat UNDER a protruding roof, a tree, etc.', a situation making the transfer of the s- to the primitive an ever-present latent possibility. So much for the justification of ORIGIN. The next question is the matter of SPREAD, and on this score we have witnessed a noteworthy scope of variation. In the cases under observation, at least, the agencies of polysemy and of lexical polarization turned out to be relatively weak forces : ensueno has never seriously endangered sueno 'dream', and emberano dislodged verano only in an isolated dialect. This discovery, among other considerations, made us skeptical of the allegedly exclusive rôle played by sol in transmuting its semantic opposite *ombra into s(ol)ombra; it is, however, conceivable, even plausible, that the influence of sol sufficed to make the dissimilatory mechanism alter awkward *sobombra into solombra, in preference to alternative lines of escape. (The inter play of several forces must at all times be kept in mind as the likeliest apparatus of causation in language change.) Threat of homonymy, on the other hand, we found to be a more powerful agency than mere polysemy ; whereas, to evoke a mood of drowsiness, romantic poets may have cherished sueno precisely on ac count of its dual reference to 'sleep' and to 'dream', it is hard to imagine any speaker or writer — other than one bent on eccentricity — who could profit from the lexemic overlap of 'man', 'shoulder', and 'shade'. The impending menace of such an overlap may, then, very well have provided, if not the initial spark, at least the actual motive force, that sent sombra on its way to ultimate victory. If we train ourselves to distinguish more sharply than did our predecessors between the origin and the eventual triumph of an innovation and to prepare for each process a separate analysis, we shall also stand a chance of recognizing far more neatly the many instances — ordinarily overlooked — of THWARTED MUTATIONS. The discovery of failures in the speakers' incessant attempts at innovation, and the ability to account for such failures no less cogently than for success, would doubtless rejuvenate the entire domain of genetic linguistics.

PRIMARY, SECONDARY, AND TERTIARY ETYMOLOGIES:

THE THREE LEXICAL KERNELS OF HISPANIC

SAÑA, ENSAÑAR,

SAÑUDO

I

The current trend of paying far closer attention to pre-1800 linguistics than was deemed possible or advisable only a few decades ago may exert a beneficiai effect also on etymology. I am not hinting here at a certain - more or less idle, if not downright morbid — antiquarian curiosity about what a few imaginative individuals so inclined (say, two or three or four centuries ago) may have thought about a given number of obscure word origins. Rather do I assume that, once a few precautionary measures have been taken, even our own front-line research in ety mology can occasionally, perhaps not infrequently, profit from consultation of those almost forgotten sources. The opinions of the pioneers of the 16th, 17th, and 18th centuries were still taken into account, after critical sifting, by the founders of comparative linguistics. But shortly after 1860, with intolerant Neogrammarians ante portas, scholars began to attach such crucial importance to phonetic accuracy or, at least, plausibility in etymological equations that the verdicts of the Renaissance and post-Renaissance trail-blazers, as a rule most vulnerable on the phonetic side, rapidly fell into disrepute, sometimes to the point of becoming unmentionable. This wholesale rejection of wellmeant and erudite, but scientifically seminaïve, pronouncements was, as we now begin to realize in retrospect, exaggerated and unwarranted. To be sure, the pre-1800 pioneers blundered about sound correspondences; our severe condemnation of their practices on that score need not be revised. They were guided, in their search of lexical ancestry and continuity, almost exclusively by rough similarity of form and by consideration of semantic affinity, and it is in this latter domain that a good deal of salvaging — not only for the sake of what German scholarship aptly calls Ehrenrettung — seems entirely legitimate. Even if those forerunners remain, from our own vantage point, mere amateur-semanticists, quite a few of them were undeniably endowed with a by no means negligible associa tive flair. They were also closer — in terms of temporal, spatial, and social proximity — to the mass of untutored speakers, the actual shapers of the language involved, and they were exposed, sometimes from childhood, to a rich store of folklore at its most effervescent, which often affords valuable etymological clues.

474

ETYMOLOGY

In precisely what situations can the verdicts of that vanguard be of help to us at this late date? Not in those cases, presumably a minority, in which a given word has, unexcitingly enough, developed without any interference from other lexical families, merely undergoing a predictable succession of sound changes. Such a state of affairs requires, on the part of the analyst, little more than competent handling of the apparatus of Neogrammatical linguistics. But suppose the parental Word A, at a certain point of its evolutionary curve, instead of advancing steadily in the fore seen direction toward its modern result A', began to be deflected from its original course through the influence of Word B. Certain features of A' are then bound to go back to  rather than to A. The Renaissance etymologist may not have recog nized the line connecting A with A', but as a — lexically sensitized and responsive — contemporary member of the speech community which was engaged in deflecting A from its straight course through secondary association with  — more often than not semantic, pictorial association — he would have been tempted to trace A' to B. By so doing he would, let us grant, have made an egregious historical error; however, by speaking up he would also have acted as an authoritative living witness - an irreplaceable witness - to a process which can otherwise be reconstructed only archeologically, through meticulous inspection of its results — that is, through hindsight. Let me provide one example, somewhat simplified and adjusted to our needs. Latin had a word for 'spit' or 'javelin', VERŪ; its diminutive VERŪCULU, literal ly 'small or short javelin', apparently also served as the designation of the bolt or latch (made of metal), cf. Fr. verrou and, even more clearly, the corresponding verb verrouiller 'to bolt'. The double r of verrou is best explained through assumption of contact with FERRUM 'iron', and Ptg. ferrolho 'bolt' shows, word-initially, an even higher degree of interpenetration. Old Spanish also favored forms like ve-, fe-rrojo, but mod. cerrojo, accompanied by cerrojillo — not only 'small bolt' but also 'wagtail, warbler' — and by cerrojazo 'loud noise of a bolt suddenly shut home', testify to further contamination by cenar 'to close. lock'. A Renaissance etymologist might not have recognized the blurred line linking the Romance forms to the Latin starting point; but if the primary etymology (VERŪ, VERŪCULU) would in all likelihood have eluded him, he could still have been a perfect, articulate observer of the agency of those encroachments which constitute a secondary and even a tertiary etymology — the impact of the local words for 'iron' or for 'to lock', say. His very error in groping for the ultimate source might thus constitute a piece of evidence, a priceless clue to an elusive concomitant, collateral influence. There are on record even more interesting cases, some of them of relevance to the literary scholar as well. In 1611, the Spanish pioneer etymologist S. de Covarrubias, in his Tesoro, associated the rustic name of a woman, Aldonça, with Lat. DULCE 'sweet'. This derivation gave point to Cervantes' famous word play on Aldonça and Dulcinea in his Quijote, which had recently been published and enjoyed an immense vogue. In 1940, the late Ramón Menéndez Pidal, quite correctly (or so it seemed in a narrow perspective), declared this derivation absurd and censured a noted Cervantes exegete for having uncritically acccepted it. Aldonça/Alduença, the dean of philologists stated, were offshoots of a Visigothic name, ALDEGUNDIA, paralleling OSp. Tedgüença, from THIUDEGUNDIA, and Ennegùença, from se-

PRIMARY, SECONDARY, AND TERTIARY ETYMOLOGIES

475

curely reconstructed *HINNEGUNDIA. Technically, Menéndez Pidal was right; but even by his private standard not entirely so. Only a few years later his own star pupil, Rafael Lapesa, demonstrated that once the Hispano-Latin name Dulce or, with hypercharacterization of gender, Dulcia (pronounced Dolça) had attained currency in the Peninsula, presumably in the 12th century, semantic association with Hispano-Gothic Aldonça and Ennegùença promptly ensued, as is attested by a hybrid variant, "Endulcia comitissa", culled from a Riojan document, A.D. 1200. So Aldonça carried with it two distinct legacies, Cervantes was justified - not only as an artist - in having recourse to the pun, the early 17th-century etymologist Covarrubias was not entirely in error when he cast about for a plausible explanation, and the mid-20th-century linguist Menéndez Pidal, for all his ingeniousness, was not quite justified in his angry repudiation. To a certain extent we are free to invoke multiple, equipollent etymologies, just as we have learned to appeal to multiple causation. Comparable observations, with slight adjustments, apply to many discrepancies between the opinions of 19th- and 20th-century etymologists, each of whom aspired to being entirely right and aimed at proving his opponents completely wrong. In some instances such neat polarization of "correct" vs. "erroneous" is, indeed, attain able and worth striving for; but in a surprisingly high percentage of cases each mod ern participant in the discussion of a complex etymological issue simply recognized, isolated, and, after a stubborn fight, often solved not the entire problem, but solely the one facet of the problem which happened to be of special appeal to him. In this context it would of course no longer be proper to speak of any one group of modern etymologists, the way we could afford to do in referring to Golden Age pioneers, as "witnesses" to a certain remote processes of language change, though it is perhaps admissible to credit today's native investigators and foreign investigators of a given language with different degrees or varieties of semantic intuition in matters etymological. Let me now select for somewhat more leisurely inspection one particularly intricate case history which may teach us not only how to disentangle the etymo logical strands (as a rule, forbiddingly enmeshed at first contact), but also how to rank and hierarchize them, i.e., how to assign to jointly responsible bases the labels of primary, secondary, and tertiary paternity. The specimen chosen is the Spanish and Portuguese favorite word, if not the only available one, for 'fury, rage', namely sana, which is flanked by the adj. sanudo 'furious' and by the reflexive verb ensanarse 'to fly into a rage, become angry'. Over the last century or so, this word-family (including some of its less prominent branches) has been genetically associated with three almost equally plausible lexical nuclei in the parent language. First, with SÄNUS 'sound, healthy' and its opposite, ĪNSĀNUS 'of unsound mind, mad. raving', flanked by the two verbs SÄNÖ, -ĀRE 'to heal' and ĪNSĀNIŌ, -ĪRE 'to be mad, rage, rave'. Second, with SANNA 'mocking grimace', flanked by SANNIŌ, -ŌNIS 'buffoon' (a derivative not spurned by Cicero). Third, with SANIĒS 'corrupted blood', which, like numerous 5th-declension words, actually boasted the variant SANIA (cf. FACIES 'face' beside Late Lat. FACIA, which underlies Fr. face, It. faccia', and DIĒS 'day' alongside *DIA, which is the immediate model of Sp. Ptg. día).

476

ETYMOLOGY II.

The majority opinion of present-day etymologists is formulated by J. Corominas, who, after dating OSp. sana ca. 1140 and its congener OPtg. sanha A.D. 1200, cautiously declares: Origen incierto; prob[ablemen]te de INSANIA 'locura furiosa'. Es verosimil que el verbo emanar, 1220-50, proceda del lat. v[ul]g[ar] INSANIARE 'enfurecer' (deriv[ado] de INSANIA), y que de emanar se extrajera después sana. Deriv. sanudo, 1251; ensanamiento, s[iglo] xIx. 1 The relation of sana to emanar, and of both these words to Lat. SĀNUS and ĪNSĀNUS, alongside the asymmetrically structured verbs SĀNŌ, -ĀRE and ĪNSĀNIŌ, -ĪRE, is indeed one important facet of the problem at hand, and even broadens out into the discussion of two homophonous prefixes: nominal IN- 'un-, non-' (reversing) vs. verbal IN- 'in-' (ingressive). But at least one equally arresting fact, namely the relation of the old adj. sanudo to the other members of the word family (specifically, against the background of the ambit of the suffix -udo) seems to have eluded the vigilance of Corominas. It is this inconspicuous detail, however, which may serve as a powerful wedge in rehabilitating a half-forgotten alternative to this conjecture - Jules Cornu's radically different approach to the genetic issue. 1) Breve diccionario etimológico (Madrid, 1961), p. 511a. Corominas' verdict involved a wide margin of departure from the pronouncement of the Academy Dictionary, 16th ed. (1936-39) — 17th ed. (1947). The Academicians, echoing J. Cornu's original hypothesis (see infra), traced sana to SANIĒS 'sangre corrompida, veneno' and attached to sana the adjectives in -oso and -udo plus the respective adverbs; but they linked ensan-ar, along with -ado and -amiento, to ĪNSĀNIA, placing desensanar and obs. asañar (tr. refl.), at least by implication, in the vicinity of ensanar, with the result that desensanar, defined 'hacer deponer la sana', yet decomposed into des + ensanar, must have left every inquisitive reader wondering whether SANIĒS or ĪNS NIA was the double-prefixed verb's ultimate source (see pp. 125a, 437b, 521c, 1138c, 1139a). Except for his omission of asañar, S. Gili Gaya, Vox: Diccìonario general ilustrado de la lengua espanola, rev. 2d ed. (Barcelona, 1953), pp. 588a, 69Ία, 1519a, repeated verbatim the Academy's self-contradictory judgments, adding here and there an original datum (e.g., the example "en sañarse en el vencido" and the synonymic equation ensanamiento =refinamiento). The chief rival of Corominas' two ventures {DCE, BDE), namely the Diccìonario etimológico espanol e hispánico by V. Garcia de Diego (Madrid, 1954), pp. 78b, 262a, 492b, traces in its opening part (Modern → Ancestral) a discernibly different picture of the situation. Drawing on noteworthy dialect material, thus far in part neglected, the author distinguishes between three homonyms: saña1 'hatred' < ĪNSĀNIA (to which he moors asañarse1 'to fly into a rage', ensañ-ar / -amiento, plus sañudo1 'furious'); saña2 'ugly gesture' < SANNA (with which he connects sañudo2 'grimacing'); and saña3 'dirt' <SANIĒS (from which a thread leads to asañarse2 'to be infected'). More persuasive than this apodictic pronouncement were some comments which Garcia de Diego made in Part II of his dictionary (Ancestral → Modern); for criticism see infra. The Academicians were so impressed that in the 18th ed. (1956) of Spain's official dictionary they swerved from their earlier path, crediting sana, along with the remarkable adver bial phrase a sanas 'sañudamente', to SANNA, while leaving ensanar and its immediate offshoots as satellites of ĪNSĀNIA, once again a split which marked, in comparison to the position of 1936 and of 1947, a mere reshuffling of data rather than any genuine progress. Though there are on record in Old Spanish a few -ar verbs newly derived from -udo adjectives (witness encornudar 'to make a cuckold' in the Fuero de Alcaraz, XI, 99; cf. J. R. Craddock, RPh, 24 [1970-71], 126), the Academy's desanudar (ed. 1956, p. 442b) has nothing to do with sanudo, being a mere variant of desanudar (based on nudo/nudo 'knot').

PRIMARY, SECONDARY, AND TERTIARY ETYMOLOGIES

477

For though Corominas' succinct statement comes close to reflecting the average opinion held by modern experts,2 it represents but the final distillate of much thinking and groping that has extended over more than three centuries and a half. As far back as Covarrubias' Tesoro (1611), one finds, s.v., a noteworthy dilemma between: (a) ĪNSĀNIA ("perdida la in, corno la perdió la palabra sandio") and (b) SANNA ("que vale 'ronquido' y 'bufido', que el que se ensana da muestra con estos acidentes senalados en las narizes, las quales se le hinchan y echan de si el ayre con violencia y sana"). That master-pioneer grouped, conceivably for the first time, sanudo and ensanarse with sana into a single, close-knit triad.3 The Madrid Academy merely echoed Covarrubias' split verdict a century and a quarter later,4 but documented the formations so far adduced, including the adv. sañudamente, rather fully and added to the record, for good measure, one occurrence of sañoso. R. Cabrera's posthumous dictionary contains no information on sana and its congeners, but is worth citing in as much as the author still championed the derivation of sandez 'inanity, simplicity' from SĀNITĀS and of sándio [sic] 'naïve, ingenuous, foolish' from ĪNSĀNIĒNS.5 It took F. Diez's acumen to show, in 1853, that the equivalence of Ptg. sandeu = Sp. sandio, paralleling as it does the correspondence Ptg. judeu = Sp.judio 'Jew', prompts us to turn our attention first to the elusive suffix. But even Diez had nothing new to offer on sana; so he repeated, on several occasions, Covarrubias' rough-hewn formula for the dilemma.6 Some other early workers operated with such bases as Ar. xana 'to hate' (actually, Šāna 'to disgrace'?) or It. sanna, zanna 'fang, tusk'; these conjectures were briefly mentioned by an unenthusiastic P. F. Monlau, who personally preferred to uphold the classic solution: (ĪN)SĀNIA.7 Against this backdrop of doldrums, a rival explanation suddenly proffered by a still very young Cornu — a conjecture forming part of his trail-blazing "Études sur le Poème du Cid"8 - acquires its full dramatic importance:

2) A good spokesman for his generation is G. B. Pellegrini, Grammatica storica spagnola (Bari, 1950), who discusses sana under apheresis (§40), under -nį- (§58), and apropos -IĒS → -IA(§111), then groups sanudo (§171.42) with barrigudo, barbudo, membrudo, and sesudo. Note the paleness of his definition of the suffix ("una qualità od un difetto") and the inaccuracy of the statement to the effect that this "caricaturing" -udo descends, genetically, from pastparticipial -udo, as in OSp. entendudo, tolludo. 3) S. de Covarrubias Orozco, Tesoro de la lengua castellana  espanola, ed. M. de Riquer (Barcelona, 1943). 4) "Diccionario de Autoridades", VI (1739), 45 ab. 5) Diccionario de etimologias de la lengua castellana, ed. J. P. Ayegui (Madrid, 1867), II, 609 f. 6) "Abgekürzt aus INSĀNIA, oder ist es SANNA Zähnefletschen?" Thus in the 1st ed. (Bonn, 1853); in the 3d (1869), the last revised by the author; and, most accessibly, in the 5th (1887), p. 485. Diez's originality in discovering an archaic variant sendío of sandio, in correcting the word's stress-pattern, and, less felicitously, in dragging in péndola < PENNULA, do not qualify for further discussion in this paper. 7) Diccionario etimológico de la lengua castellana (Buenos Aires, 1941); reprint of the Peninsular 1881 ed. The author omits SANNA 'mocking grimace' from his store of etymological possibilities. 8) Rom., 10 (1881), 75-99, at 81.

478

ETYMOLOGY

Diez EW IIb s.v. [i.e., EWRS3] veut tirer sana de INSĀNIA ou de SANNA. Mais ce dernier ne donnerait pas le portugais sanha et, si l'on admettait la première éty mologie, il faudrait rendre raison de la chute de la syllabe initiale. SANIĒS, en revan che, entré dans la première déclinaison, satisfait à la fois à la forme et au sens. Comparez la locution française: « se faire du mauvais sang ». Or, 'mauvais sang' au sens propre est précisément SANIĒS. Cornu's laconic suggestion was illuminating and elegant both in its positive and its negative facet. But while its negative message was before long accepted — SANNA disappeared from the discussion and nobody, in jettisoning it, bothered to raise the question as to whether Western sanha could not possibly have infiltrated, through diffusion, from the Center, where Lat. -NN- has yielded-ñ- — the positive ingredient of the solution met with a lukewarm reception. True, one finds non-committal hints of it in the posthumous 2d ed. of Monlau's dictionary (1881), revised by his son; in A. Scheler's supplement to posthumous reprints of Diez's dictionary; and in all three editions of G. Körting's ill-fated lexicographic venture. 9 But - as a startled Körting shrewdly remarked in 1901, then again in 1907 - Cornu himself, at least by implication, withdrew his hypothesis at the latest in 1888, when, in the context of his historical phonology of Portuguese, he set up, unwaveringly, the equation sanha < (ĪN)SĀNIA. In expanding his treatise almost twenty years later, he elaborated on this equation, which clashed to sharply with his own earliest view. 10 With the champion apparently convinced of his own error, it is small wonder that W. Meyer-Lübke, in the bold edifice of his fin-de-siècle comparative morpholo gy, chose not to worry about a clear-cut choice between SĂNIE 'corrupted blood' and (IN)SĀNIE 'rage', apropos of such isolated and outnumbered 5th-declension words as had collapsed with the swelling ranks of their 1 st-declension counterparts. 11 More was to come when Meyer-Lübke's dictionary appeared, straddling the prelude to and the conclusion of the First World War. Here ĪNSĀNIA, for the first time, received collateral support from the Surs. (= Obwald.) compound malsoña 'sickness'

9) Scheler, EWRS5, p . 777, adopts an attitude of strict impartiality: "Cornu . . . erhebt ernst liche Bedenken gegen beide hier vorgebrachten Etyma; nach ihm genügt lat. SANIES, *SANIA geworden, in jeder Beziehung; es heisst ja 'böses Blut' ". Körting, in his comments on INSĀNIA (LRW 1 [1891], §4326; LRW2 [l901],and LRW3 [1907], §5017), was cautious enough to aver: ". . . daraus vielleicht gekürzt sana". 10) "Die portugiesische Sprache," §102 ("Unterdrückung unbetonter Vokale") in G. Gröber, ed., Grundriss der romanischen Philologie, I (Strassburg, 1888), 744: ". . . sanha INSANIA, nicht SANIES", with a footnote reference to Late L. INSANIARE 'to infuriate' : "Oestrum genus tabani quod boves exagitat et insaniat" (Ambr. B, 31 sup.; cf. ALLG, I, 25). The analysis is reaffirmed in the revised 2d ed. of the Grundriss ("Verstummung unbetonter Vokale", p. 955), with an enriched corpus of parallels: barulhar ['to brawl, kick up a fuss'] Fr. cire, Lat. cēreu 'waxen' > Fr. cierge 'candle', Lat. circulu 'circle, circuit' > cercle, Gr.-Lat. cylindru 'cylinder, roller' > cylindre, etc. One might call this relationship a case of extra-strong phonetic change. Now examine a slightly different situation: If the ancestral words began with ca-, most French descendants will show a /š/ followed by one of several foreseeable vowels: campu 'field' > champ, cane 'dog' > chien, canale 'waterpipe' > chenal 'channel, passage, gutter', capillu 'hair' > cheveu, catena 'chain' > OFr. chaeine, mod. chaîne. But this time the learnèd words follow a separate course, witness cas 'case', catastrophe, cataclysme; similarly with Provençalisms, like caserne 'barracks'; with Italianisms, like cavale 'mare', cavalcade, cavalier; and with Hispanisms, like camarade, orig. 'roomful', later 'roommate', all of which diverge from the norm. Also, one can find doublets

506

ETYMOLOGY

— or should we modernistically call them 'minimal pairs'? — such as Champagne (the name of a province) alongside campagne 'flat country'. In short, we are still free to invoke a strong change, but no longer, in good conscience, any extra-strong change. Let us now cross the Pyrenees and scrutinize the situation in Spain, focusing our attention on the easily most celebrated sound change undergone by Latin in that former province of the Republic and the Empire, namely the tendency toward disappearance of ƒ- via h-, particularly in word-initial position; please note my cau tious use of the label 'tendency'. Certain familiar equations immediately come to mind: Lat. filiu 'son' yielded hifo via fijo, filu 'thread' became hilo via filo, fibra 'fiber, filament' emerged in the daughter language as hebra. But the number and caliber of exceptions is rather staggering, and each seems to belong, almost randomly, to one of numerous categories, some of which are only hazily delimited. Thus ƒremains if it is followed by r or /, as in frio 'cold', flor 'flower'; or if it is flanked by /w/, as in fuente 'fountain', fueron 'they were, became, went'; or if a learnèd lexical item is involved, as in faccion 'feature', familia 'family'; or if a Gallicism 'lato sensu' is at issue, cf. forja 'smithy' echoing Fr. forge; or if we are dealing with an Italianism, such as facha 'appearance, aspect, face, mien', or with a Catalanism, as applies to faena 'labor, toil' vis-à-vis native hacienda, from fac(i)enda 'things to be done'; or if consistent loss of/- would have entailed the rise of a pair of homonyms, cf. the fine differentiation of fiel 'faithful' (orig. disyllabic fiel) vs. hiel 'gall'; or if the disappearance of/- would have deprived the word of its last consonantal pillar, cf. feo 'ugly', from foedu, favored in an effort to escape from *(h)eo, etc. The socalled exceptions clearly outnumber the words which follow the rule, and the change deserves to be dubbed 'weak', since predictability is at a low point and instances of protracted wavering abound in Golden Age Spanish, e.g.,febrero beside hebrero 'February'. One would be faced with similarly disappointing results in discussing the conflicting outcomes of Lat. intervocalic p, t, k, and s in Italian: cf. potestà 'power' vs. podestà 'administrative head of a commune', potere 'to be able' vs. podere 'farm, estate', and so forth. I shall before long attempt to take up the prob lem of one possible reason, among others, for different degrees of regularity in sound changes. At this point let me remark that where that degree is conspicuously high, schematization is very opportune and adding one more etymological equation by way of illustration would amount to a luxury that can safely be skipped. Weak phonetic changes are quite a different story: Here every word history, with all its chronological implications, geographic intricacies, and semantic ramifications, is of vital importance. If we then agree to use the recommended scale ranging from extra-strong to ultra-weak, we find that it is at the weakness end rather than at the opposite pole that the actual need for intertwining phonological with the lexicoetymological analysis arises most acutely. Let us now cast one more glance at the polarization of strong vs. weak with regard to sound change, and reexamine the problem in terms of causation. Twenty or so years ago André Martinet produced quite a stir among students of explicative linguistics by championing the idea of chain reactions in sound changes. The idea is indeed seductive, but I am afraid that the advocate's performance has not been consistently flawless, and that a considerable part of the unevenness is traceable

ETYMOLOGY AND MODERN LINGUISTICS

507

to his notorious indifference to lexico-etymological problems. A quarter century ago Martinet, in his memorable article "Occlusives and affri cates with reference to some problems of Romance phonology" in volume V of Word, correctly sensed a certain connection between the simplification of geminate stops (dd, gg, tt, etc.), as in agger 'heap, mount, rampart'; voiced medial stops, as in ager land, territory, field'; and voiceless medial stops, as in acus 'needle, bodkin'. But he fell short of stating his opinion of the actual concatenation of events with maximum clarity. From the lexico-etymological vantage point, it is patent that the development of sounds such as -b-, -d-, -g-, which figure sparingly in word-medial position, produced chaotic results, especially in Hispano-Romance: pigritia laziness' yielded preguiça in Portuguese as against pereza in Spanish; integrare 'to make whole, renew, begin afresh' gave Spanish two verbs, with deep semantic differentiation: enterar 'to notify' and entregar 'to hand', etc.; also, we find in the older language nio beside nido for 'nest', cruo beside crudo for 'raw', from nīdu and crudu, respect ively. On the other hand, the shift from ρ to b, from t to d, and from  // to g is both fully predictable and copiously represented in terms of lexical frequency and of incidence alike. This regularity of t >d, etc., as against the striking irregulari ty of d to zero, is best understood on the assumption that the former involved a sharp-edged primary move and the latter merely a blunted secondary reverberation, weaker by definition, far less sharply profiled, incidentally, in Castile than in neigh boring Portugal. The fainter the reverberation, the more acute the need to examine each word biography under a microscope, including the best possible classification of a residue of etymological obscurities, as with OSp. deçir, Ptg. descer 'to descend', probably from Lat. discēdere 'to go or march away, swerve', confused with de scendere 'to climb down, sink, pierce' and the like. Against this background of the general relationship of lexicology to historical grammar, it might be appropriate to raise the question: Which etymological problems, amid the myriads of trivia (or items of local pertinence), can rank as truly interest ing? In this context, the qualifier 'interesting' is to be understood in the dignified sense with which students of philosophy and mathematics tend to endow it. And since the elements of 'relevance' and 'timeliness' enter by implication into this par ticular use of the adjective, one may go one step further and explicate the connection by reformulating the question thus: Which etymological problems deserve to be called singularly appealing, nay, gripping today? Keeping this scale of values in mind, I am inclined to divide the truly noteworthy etymological issues into two major categories. One group of postulated equations promises to stimulate our thinking about linguistic reconstruction: its presupposi tions, techniques, limits, goals, and the like. This group includes problems which no well-balanced linguist can afford to ignore, even if that linguist happens - as a matter of choice or of inborn limitation - to be utterly insensitive to history and culture, indeed, defiantly proud of his or her indifference. The other group demands of the practitioner not only dual curiosity but also finesse plus expertise: Though at no time divorced from his commitments toward linguistics proper, the etymolo gist, to achieve success in solving problems of this second type, must definitely have been blessed with intuition, or have developed a genuine flair for detecting elusive historical connections — elusive because the evidence available is, almost

508

ETYMOLOGY

by definition, doomed to remain lacunary. Let us leisurely examine at least one specimen of the former variety of problems in a test which evokes the atmosphere of a linguistic laboratory. Certain etymologies are pivotal because, if found to be tenable, they endow us with the long-soughtafter power to formulate various obscure phonological and grammatical relationships with enhanced precision. Take the Spanish adjective terco 'stubborn', viewed tor decades as a genetically insoluble case - as long as scholars were vainly searching for an archetypal qualifier beginning with a t-. But pause to consider the derivation of Sp. cuerdo 'prudent', another adjective describing a human character trait. It does not stem from any archetype beginning with a k- sound but was boldly ex tracted, through loss of prefix and other devices, from the verb acordar 'to recall, remember': A person who has a clear memory of earlier experiences is apt to profit from past disappointments by learning to proceed cautiously, prudently, with circumspection. If cuerdo was distilled from acordar, as every initiate agrees, could not terco have been similarly carved from some such verb as *atercar or entercar? Older Spanish texts do indeed show a few scattered traces of entercar 'to stiffen', but how is one to prove that the adjective was cut from the verb and not, inversely, the verb from the adjective, as in hundreds of comparable cases: cf. E. to brighten, sadden, sharpen, shorten, to cite just one neatly silhouetted pattern? And, even apart from this problem of derivational hierarchy, would the choice of the verb as the primary formation bring us any closer to the main goal of a satisfactory etymological solution? To what earlier form, in other words, would entercar lead its supporters? An escape hatch out of this impasse was detected when, through a chance discov ery, a rare sixteenth-century verb, enternegar, was identified in a writer who indulged in the use of a racy, rural lexicon. In a characteristic, unequivocal passage enternegar applied to the hardening, i.e., blighting, of grains of wheat as a result of frost. Enter negar and entercar thus turn out to be near-synonymous, but while entercar, as we recall, is genetically opaque, its counterpart enternegar is, felicitously enough, perfectly transparent: It echoes internecare 'to kill thoroughly', the very same verb that has left an isolated reflex in E. internecine (as a learnèd qualifier of strife or war). Internecāre was, obviously, a compound of the more familiar necāre 'to kill', with either an intensive (as in Spanish) or a reciprocal (as in English) connota tion attached to the prefix inter-. So enternegarse literally meant 'to become killed', through frostbite, and entercarse merely represents a syncopated variant with a slightly more 'advanced' meaning. Amusingly, the root word underlying the entire family, which none of the pioneers conjectured, began not with a t but with an n (necāre), a consonant which, through a whim of circumstances, in the end disap peared completely from terco. Incidentally, the Old Catalan form of the adjective, enterch, still shows the prefix en- intact, precariously maintaining its hold. The etymological derivation here presented in a nutshell contains very few cul turally significant insights - except perhaps the metaphoric use of agronomic terms for character traits, hardly unnatural in a rural country like Spain. By way of com pensation, it is a real  for the student of historical grammar. It illustrates the mechanism by which certain adjectives are extracted from verbs; documents the gradual jettisoning of prefixes as part of this process; shows how the -ter- seg-


509

ment, at the start an integral part of the prefix inter-, was eventually, through false separation, reinterpreted as the very kernel of the root morpheme; and demonstrates the protracted coexistence of syncopated and non-syncopated variants. Let me omit some further grammatico-phonological implications which might be of genuine concern only to the narrow specialist: the choice of the monophthong (terco) rather than the equally conceivable diphthong (*tierco); the preference given to -o over -e in Castilian (terco), though not in Galician, where terque prevails; the reduction of the medial cluster *-rnc- to -rc-; etc. In short, the equation established turns out to be a miniature treasure-trove of essential information on neglected but by no means unimportant phonological and grammatical processes. This is a telling example of the kind of etymology that no self-respecting linguist can, with impunity, indulge in shrugging off as 'irrelevant'. While it is hazardous to order the linguistic, even more so the grammatical, rele vance of etymological problems according to the form classes of the words involved, one is at liberty to state in probabilistic terms that the biographies of isolated nouns, for all their pungent anecdotal flavor and broadly cultural informativene ss, as a rule are less than very enlightening to the linguistic analyst; I am thinking of such merely 'entertaining' case histories as those of macadam and sandwich. Conversely, where close-knit groups are at issue, especially pronouns and prepositions, the mere sus picion of an etymological discovery is apt to set in motion a good deal of straight linguistic thinking. Within the category of nouns, the etymological probing of tightly organized subgroups (names of cardinal points, certain numerals, kinship terms, anatomic designations) holds out the highest promise of linguistic relevance. Verbs command heightened attention on account of their (typically) broader referential scope (a feature peculiar to adjectives too); also, in view of the heavier representa tion in their records of many side issues of unadulterated grammar (constructions involving voice, mood, aspect, person, etc.). If we now turn our attention for a few concluding minutes to the historian's, rather than the linguist's, favorite etymologies, I would urge you to consider, above all, those which cast unexpected shafts of light on facets of older culture insufficient ly laid bare either by archeology or by contemporary historiography. One characteristic area where etymological judgment, if properly exercised, takes us farther than any archeological or historiographic clue is the domain of intimacy, taken in its broadest sense, not necessarily prurient. Let us include in that domain a generous spectrum of diversified experiences, from religious mysti cism all the way down to erotic thrill, and reserve extra niches for superstition resulting in taboo, for sparks of humor, and for all sorts of informalities. The replace ment of OFr. suivre or suire 'mother-in-law' by belle-mère, for example, certainly smacks of a propitiatory attitude, which could otherwise be hypothesized, but not proven. The verbs Sp. arrebatar 'to snatch, carry off, captivate' and recamar 'to em broider with raised work', both of Arabic origin, fall into a familiar pattern of both extended and intense Hispano-Oriental symbiosis, further illustrating two of its facets already well-known: techniques of mobile warfare, on the one hand, and mastery of certain arts and crafts, on the other; but the Sp. verb halagar, orig.falagar 'to flatter, cajole, caress, distract, entertain' and its Ptg. cognate afagar, also of Arabic ancestry, seem to suggest a connection between East and West otherwise

510

ETYMOLOGY

hushed up in formal accounts, namely one in skills and talents that the seductive young Moorish women may have acquired or developed to perfection, in a harem, before their chance capture by the Christians. Fr. malotru 'ill-bread, uncouth', orig. 'born under an evil star', and OFr. dur-, mal-feü 'stricken by hard luck' (wanly echoed in the modern phrase feu la reine 'the late queen'), derived from such types as male- *astrūcu, male- *fātūcu, display unmistakable astrological overtones, etc. 4. A desideratum: zigzag movements from etymology to straight linguistics One defensible reason for the linguistic scientists' above-noted gradual disenchant ment with etymology can be stated in straightforward fashion: The farther etymo logical analysis proceeds, the greater the number of opaque and obscure cases is apt to become, with all sorts of unwelcome complications encumbering the worker's path in the establishment of each equation. Now the majority of 'pure etymologists' — those who aim to embark on a dictionary venture, for instance — cultivate a cer tain aloofness from grammatical preoccupations; consequently, as they forge ahead, leaping nimbly from one word history beset with unkowns to another, even less transparent, they tend to sweep these uncertainties under the rug (or to be satisfied with citing isolated parallels), instead of exposing the gaps to full view and volun teering to investigate them with the equipment of a well-stocked linguistic workshop. Let two examples illustrate this perilous situation. The French adv. vite 'fast, quickly', which was once also used in adjectival function, is ordinarily linked in less than convincing fashion - to the Latin family of videre 'to see' via its Vulgar Lat. past participle (cf. Sp. Ptg. It. vista View'), the implication being that something comparable to G. im Augenblick, Fr. en un clin d'œil may long ago have served as the point of departure. There is an excellent chance for a minority opinion to be intrinsically superior to that hypothesis: vite may indeed pertam to the family of vivere, 'to live', cf. Russ. Živo! 'quick!' (lit. 'alive!') and OProv. vìvatz 'quick(ly), rapid'. The ultimate decision must hinge on the results of a monographic inquiry into proto-Gallo-Romance past participles, which unfortunately is unavailable. A crosslinguistic study of the rival patterns of relevant imagery might also be helpful and tip the scales in the event of a 'tie', so, doubtless, could be the survey of the omission of the adverb marker in this particular semantic province (E. fast!, not *fastly!; Sp. !ligero!, not * !ligeramente!', etc.). The second specimen whets one's appetite even more. Sp. rincón '(inner) corner' (= G. Winkel) and Cat. racó have, on certain occasions, been bracketed together in a daring rapprochement and jointly traced to a Greek base meaning '(small) bay, inlet, corner' - a pictorially seductive, culturally engaging conjecture (cf. W. MeyerLübke's revised dictionary, §443a). But on the phonologico-grammatical side virtual ly everything remains to be done if this solution is to be vindicated. Urgently needed, for a start, are (a) an exhaustive monograph on the suffix and suffixoid -on; (b) another unabridged study of the insertion, or, alternatively (by way of hyperurbanism?), occasional omission, of a nasal before a given word's central consonantal pillar; (c) a third searching paper on patterns of vowel dissimilation in Romance (ironically, there exist several, and more that adequate, probings into consonant dissimilations, in contact and at a distance); and (d) an inquiry into the effects,


511

on the preceding unstressed vowel, of such dyadic medial clusters as nasal + occlusive. Without this minimum of four preliminary studies which, to be conducted persua sively, must rest on secure material, etymologically unimpeachable, and thus involve a good deal of spade-work, even drudgery, the more 'attractive' issue of rincón/racó cannot be seriously attacked. The cognitively - if possibly not experientially - most satisfying strategy, then, is for the etymologist to alternate word studies with the more sober and austere types of phonological and morphological soundings. The need for this con tinuous zigzag movement was well understood by the pioneers, but the technique was later neglected in favor of one-sided assaults — it now appears, to the severe detriment of all sectors of scholarship. It is time to conclude our analysis and to prepare a balance sheet. At the outset, I promised to answer certain moot questions as candidly as possible, and it is now my duty to attempt to redeem that pledge. Etymology is the diachronic ingredient par excellence of linguistics; no serious research in historical linguistics can be con ducted without ceaseless attention to it. But if historicism itself, as a result of a new Zeitgeist, is suddenly declared obsolete, then etymology will lose much of its raison d'être and its impact on imagination; the fact that it does not easily lend itself to stringent formalization is used — ex post facto, one suspects - to rationalize the emotionally nourished indifference to it of certain fashionable schools. Etymology pursued for its own sake and bereft of any significant link to the mainstream óf events in linguistics can indeed degenerate into an idle if genteel pastime, a kind of erudite parlor game, played to the tune of strident arguments between the few remaining participants. Linguistics without etymology may, by the same token, become barren, at least at the diachronic end, because the repeated use of the same perfectly transparent cross-temporal equations does not bring us any closer to the discoveries which await a generation of scholars capable of clearing away the underbrush of intricate or tenuous lexical relations. Etymology can be modemized if the potential yield of any problem newly raised and solved is measured not in con ventional terms of just routine filling of slots in a dictionary, but in terms of constant enrichment of underlying linguistic methodology. A reconciliation between spatiotemporal linguistics and etymology must occur because without it, either discipline is doomed to inanition.

THE INTERLOCKING OF ETYMOLOGY AND HISTORICAL GRAMMAR (EXEMPLIFIED WITH THE ANALYSIS OF SPANISH DESLEIR)

Not a few among the founders and early practitioners of historico-comparative linguistics developed a strategy which consisted in alternating large-scale experiments in diachronic grammar and etymology. A 19th-century scholar might, for instance, be tempted to prepare, by way of first step, a very sketchy comparative grammar, genetically slanted; such a skeletal outline would rest, as a rule, on relatively few lexical equations, each fairly simple and free from all sorts of disturbances or com plications. Since only a small portion of the total vocabulary of a typical language displays such welcome transparency, the next move was apt to be the compilation of an etymological dictionary, in which more intricate relationships between ap parently cognate forms could fittingly be mentioned or even discussed. These in creasingly sophisticated lexical analyses would allow a variety of inferences; some of these conclusions could — indeed should — have a retroactive bearing on the earlier crude formulations of phonological and morphological correspondences and be conducive to numerous corrections, qualifications, and additions. A gram matical edifice so refined might, in turn, serve as a stimulus for a new and vigorous combing through the thickets and swampy ground of the lexicon, as a prelude to a revised and generously expanded edition of the slender etymological dictionary previously compiled. This game or interplay could, theoretically, be carried on ad infinitum. Few 20th-century scholars have excelled at this kind of stimulating two-way exercise, and pathetically few among those born after 1925. Gradually, the gram marians began to form a coterie of their own, while the lexicologists, etymologists, and students of the onomasticon also seceded from the original team of manysided workers. What was the reason for this near-fatal estrangement? In small part, it was simply a matter of dividing labor and responsibilities more economically, given the steady growth of the available corpus of data and bibliographic tools. But this unavoidably increasing degree of specialization certainly does not exhaust the difficulties in mutual understanding and esteem that began to arise. A number of circumstances conspired to drive a wedge between grammar and etymology. Since most of these factors are familiar, I shall enumerate them very briefly. The late 19th century saw the rise and rapid development of a new instru ment of investigation, the linguistic atlas; although the cartographic method of projection per se shows no particular affinity for the lexicon, it so happened that dialect geography and lexicology produced a better blend than did dialect geography and grammar. The same circles that pushed the Sprachgeographie also promoted,

514

ETYMOLOGY

with equal or even greater zest, the joint study of the lexicon and of material civili zation, i.e., the celebrated Worter-und-Sachen approach. To be sure, the objects were not exclusively ploughs, harrows, flails, baskets, or fishing harpoons, but in cluded also anatomical designations, kinship terms, the names of the days of the week, of the months, of the seasons, and of certain festivals. Nevertheless, there was distressingly little in the material so captured for the grammarian to sink his teeth into. Lavish attention was being paid to diffusion, with special reference to lexical borrowing. This kind of data lent itself to particularly effective study in conjunction with trade, barter, commerce, raids, expeditions, wars, settlements, colonization, intermarriage, seasonal employment, i.e., ingredients of social, econom ic, political, and military history. Phonology and morphosyntax are far less subject to external pressures of this sort than is the lexicon. So far we have chronicled the events as if the initiative lay always with the ety mologists. This situation was, for a while, indisputably applicable to certain promi nent branches, e.g., to Romance studies during the first half of this century, but one must beware of hasty generalization. In other headquarters it was, on the con trary, the grammatical sub disciplines {lato sensu) that advanced with impressive speed, almost to the exclusion of etymology; and within the aggregate of these subdisciplines, phonology before long began to crowd out morphology, in sharp contrast to the state of affairs toward the start of the preceding century, when Rask, Bopp, and Grimm — independently, and yet in unison — had started to unravel the mysteries of Indo-European through a resolute attack on inflection. An even more suitable bridge between the lexical and the grammatical structures would have been the aggregate of affixal derivation and composition; but, through a caprice in the history of scholarship, these constituents of morphology suddenly became unfashionable and, as a result, failed to attract sufficient talent to sustain the mo mentum of steady progress. Thus, we have had a new and even a newer or newest phonology every decade or so, whereas there have been pitifully few stirrings in the morphological domain; and while it is true that phonology and etymology overlap in theory and practice and thus support each other, the psychological conditions of work in these two specialties have turned out to be radically different, almost to the point of irreconcilability. In sum, etymologists and grammarians, without going so far as to toil downright at cross-purposes, have of late acted as if they did not serious ly need each other's competence. This gradual drifting-apart has already produced grave consequences, which must be diagnosed, if a complete divorce is to be averted. Although we may hesitate to admit it, all too often an element of untidiness haunts our etymological equations and operations. In addition to the conspicuous sound correspondences, the celebrated (indeed, vaunted) "sound laws" of the Jung grammatiker, there exist hundreds of less common and less readily predictable recurrences - but recurrences all the same. These erratic features, reputedly of limited yield in grammatical inquiry, represent a certain mortgage on many of the less obvious etymologies and are often swept under the rug, in a deprecating or apologetic tone.

THE INTERLOCKING OF ETYMOLOGY

515

Let me supply a few examples. Two Latin verbs were collocated in a unique semantic opposition, APERIÖ -ĪRE 'to open' and its antonym OPERIÖ -ĪRE 'to cover, close'. Perhaps in conjunction with the general obsolescence of the prefix ob- (op-), to which the initial segment of OPERĪRE bore a certain resemblance, that verb was abandoned in provincial spoken Latin, and an intensive variant, in volving the genuine prefix com- (or, before vowels, co-), namely COOPERIÖ -IRE 'to cover, overwhelm', was substituted for the simple verb, witness Fr. couvrir (the source of E. cover), Sp. cubrir (originally cobrir), etc. This process is hardly spectacular in itself; after all, the Latin simple verb EDERE 'to eat', subject in certain regional dialects to the erosion of its lone consonantal pillar -d-, was replaced by phonically more resistant COMEDERE 'to eat up', hence Sp./Ptg. comer. Con sequently, the state of affairs in most Romance languages is by no means disquieting; cf. It. aprire 'to open' vs. coprire 'to cover'; Sp. abrir vs. cubrir, etc. But something strange happened in Gallo-Romance; there, judging from Fr. ouvrir rather than *avrir and from Occit. obrir rather than *abrir, the original Latin verb for 'covering' has become one for 'opening' i.e., its exact semantic opposite. Is this just an accident worthy of an anecdote, or should one seriously search for parallels and counterparts? Or take the family of E. marvel. The etymon could not be more transparent: n. pi. MIRABILIA 'strange, wondrous things or events', from substantivated MIRABILE, which is firmly moored to the family of the Latin deponential verb MĪRĀRĪ. Now MĪRĀRĪ has been transmitted into the medieval and modern languages in a fairly uneventful way, meaning here 'to wonder' and there 'to look with wonderment, to stare' and, ultimately just 'to look' (witness Sp. mirar). Its first vowel, a long i, ordinarily was not subject to change and has, in fact, been faithfully preserved in the verb itself and in its offshoots, cf. Fr. miroir, the immediate model of E. mirror. But MIRĀBILIA, although semantically close to the underlying verb and linked with it through a simple, common derivational mechanism, nevertheless drifted away, and speakers allowed its i to become e or a', hence It. meraviglia, Fr. merveille, Sp. maravilla, E. marvel — an unexplained split. My third example involves a preposition, namely the Romance equivalents of 'till, until'; they are all notoriously difficult, implicating either Lat. USQUE (as is true of Fr. jusque) or Lat. FINIS 'end' (as holds of It. fino a). But the most colorful substitutes for the Latin models made their appearance in the Iberian peninsula. There one finds a straight Arabism — the only one in this cultural sphere and semantic category — namely hasta, older fasta. Its etymon has long been known: Ar. hattā, and the older Spanish reflexes, fata, (h)ata, etc., seem normal enough; the discrepant vowel of Ptg. até can be explained away, since the older texts frequently display the sequence at(a) em. . . until (in), as far as'. The etymologist can even appease a startled cultural historian by remarking that the surveying of property in medieval Spain was often conducted by Moorish experts, an activity that lent itself to association with 'until', 'up to'. At that juncture, however, an unexpected complication arose: fata became fasta, then hasta, eventually (h)asta. We know practically nothing about the reasons for the intercalation of s before t. A minor sound law long overlooked? A blend with the semantic opposite desde 'from'? A moment of light contact, brief but rich in consequences, with a doomed word marked by a certain affinity of meaning, fascas 'almost' (unless fascas itself is a merger of USQUE and fata)?

516

ETYMOLOGY

The examples could obviously be multiplied. In many instances, strictly lexical blends will eventually emerge as the real culprits that have caused the aberrations. But in a small yet significant residue of cases, the deviations will turn out to have been anything but unique. Etymology can then serve as a driving wedge leading to the discovery of less diaphanous, less plentifully represented grammatical features — of traits by and large overlaid and barely visible in a few vestigial examples. Programmatically, then, we may start out from the supposition that the move ment of Wörter und Sachen - so brilliantly set in motion by Schuchardt — has by now run its course so far as methodological innovation and theoretical enrichment are concerned. We have developed to the virtual limit of its potentialities the tech nique of preparing monographs on the names of the lizard, of the grindstone, and of the pudding — equipped with a camera, with a kit for drawing and painting materials, with notebooks to be used in field interviews, and with an identification card guaranteeing access to the stacks of a fine research library; this know-how is ours - a privilege for us and our successors to enjoy - and we would act foolishly by refusing to apply it every once in a while to problems newly raised or to old problems left unsolved, as they come our way. It would be unrealistic, however, in light of our experiences over the last 30 years or so, to expect from such studies any further break-through for linguistics. The chief beneficiaries, henceforth, are likely to be the separate if allied inquiries into folk belief, material civilization, social anthropology, etc. Linguists need not, in a frivolous recoil from the phantom of déjà vu, relinquish the ground acquired at such heavy cost, but they have a right to reexamine periodically their own priorities. It is incumbent upon diachronic lexicologists to isolate from all the controversial etymological equations left fallow and even from those which at first glance seem smooth and satisfactory, the big and the small stumbling blocks that we have so far been tempted to shrug off as mere nuisances, and to regard them instead as so many urgent invitations, not so say imperatives, to revert - enthusiastically, rather than with a sour mien — to historical grammar, rejuvenating that discipline in the process. In the better-known, thoroughly-explored languages, analysts at present stand only a limited chance of coming across straight sound correspondences previously neglected; such new discoveries are marginally conceivable where the results of the agency of certain 'sound laws' are but dimly recognizable due to successive overlays. What we, conversely, may indeed succeed in laying bare is the long string of less tidy situations which, I submit, is not one whit less attractive to the unpre judiced observer than the clear-cut distribution, whose heightened appeal seems to me emotional, or at least esthetic, rather than truly intellectual. We are entitled to explore such issues as multiple causation and to inquire whether there exist hierarchies within the bundles (or alliances) of forces at work. We are eager to become acquainted with any thwarted mutations, brought to a halt as a result of internal conflicts or external pressures. We thrill at the prospect of learning to distinguish between sound-changes, pervasive or tendential, which were triggered by analogical reshuffling initially carried out within morphological structures. We know distressingly little, and ought to know a good deal more, about the inter twining of saltatory changes, which come close to representing 'universals', and


517

so-called regular changes; metathesis and haplology, in particular, have tradition ally been stepchildren seldom integrated with the remainder of sound changes undergone by the given language. And we are committed to paying closer attention — to certain recurrences not strictly grammatical, but recurrences all the same, e.g., all manner of hypercharacterizations (polarization, serialization, and the like). The direct stimulus for all such studies can emanate from a ruthlessly honest reexamina tion of partially adequate etymologies. If we are lucky and sufficiently imaginative, and if we persevere, we may even in the end devise a new unified theory of language change. I have selected for leisurely discussion a single example, which I trust qualifies for driving home my point. It constitutes a famous etymological crux — involving the Spanish verb desleir 'to dilute, diffuse, dissolve', which, I believe, indeed lends itself to definitive identification if only we agree to pay equal heed to its semantic kernel and to the numerous grammatical issues tangential to the nucleus. Before I embark on the vicissitudes of Sp. desleir and of the etymological pro nouncements on its extraction, let me make two preliminary remarks. First, the overwhelming majority of Spanish -ir verbs go back to Latin prototypes in -ere, -ere, or -īre; a very few have Germanic (i.e., Gothic) ancestors; practically none are of obscure substratai or of Arabic background; those newly derived from ad jectives or, in rare instances, from nouns are fairly transparent in their structure and have not, formally or semantically, drifted away from their primitives beyond easy recognition.1 Second, with just a handful of exceptions (such as OSp. deçir 'to descend', Sp. [refl.] engreir 'to become vain or conceited', OSp. troçir 'to pass by', Sp. zurcir beside Ptg. serzir 'to darn'),2 the -ir verbs, whichever difficulties their conjugational patterns may offer, have produced astonishingly little controversy on the etymological level — as a result, precisely, of their genetic translucency. Desleir, then, is entirely atypical. One reason for the peculiar problems raised by desleir is the fact that it seems to be an isolate. It may have distant congeners — clarifying that issue is, obviously, the very raison d'être of our inquiry —, but there are on record no obvious close cognates, not even in Portuguese, and the presumed source language, namely Latin, offers no immediately appealing prototype. 1) Among the surprisingly few exceptions one may mention asir 'to seize, grasp' which, counter to first impressions, turns out, upon closer inspection, to be unrelated to its French semantic counterpart saisir, itself Germanic and, in turn, the source of E. seize. Asir is a deriva tive of asa 'handle' (of jug, basket, etc.) from Lat. Ā(N)SA. In such a context there should nor mally have developed a post-nominal -ar verb, except for the circumstance that the space of such a verb asar had meanwhile been preempted by asar 'to roast', fig. 'to bother, annoy, pursue', based on the past participle a(r)su, from ARDĒRE 'to be on fire, burn, blaze'. 2) I cannot offer here more than just a few hints or clues. Much ink has been spilled on OSp. deçir, Ptg. descer; in the mid-'forties R. Menéndez Pidal, who had originally thought of DECI DERE 'to fall down, drop' as a suitable etymon (1908-11.618), reversed himself and championed as the base DISCĒDERE 'to go away, march off, scatter' (1944-46.1222f.), a step in the right direction except that the continued pressure of DESCENDERE 'to climb down, go down, sink' must be taken into account; incidentally, descend- and dēscid- alternated in Vulgar Latin. My own advocacy (1956.385-95) of TRĀDUCERE 'to lead across, bring over' as the ancestor of

518

ETYMOLOGY

The following etymological conjectures bearing on desleir have come to my attention:3 a. Some such lexical type as *DĪ- or *DIS-LIGĀRE 'to untie' - the opposite of LIGĀRE 'to fasten': Only the pioneer S. de Covarrubias Orozco toyed with this idea,4 while Cuervo and Körting briefly remembered it (see below). b. Class. Lat. DILUERE 'to wash away, break up, separate' (cf. E. dilute), which in the estimate of present-day Latinists is related to LAVĀRE 'to wash'.5 This conjecture was launched by a now forgotten Renaissance scholar, F. del Rosai

troçir might have been more effective had I more heavily stressed the link between that verb and deçir. (Note that they perished jointly - a clue to their interdependence?) J. Corominas' attempt to associate engreir(se) with CREDERE 'to believe' strikes me as very infelicitous (1955.285f.) why not start out from INGREDÎ 'to enter, undertake, begin, follow (footsteps)', with special attention to such stereotyped phrases as IN REM PUBLICAM INGREDI 'to enter politics' ( > 'to make oneself important, throw around one's weight'? Cf. Sp. embair 'to trick, deceive', from INVĀDERE 'to enter upon, undertake, assault, rush upon'. As regards the erratic relation of Sp. zurcir and Ptg. serzir to Lat. SARCĪRE  patch, fix, repair', the most thorough discus sion will be found in the recent paper by S. N. Dworkin (1973.26-36). 3) Certain reference works which ordinarily supply etymological information have, rather eloquently, for once kept silence, e.g., the Spanish Academy Dictionary, in its 15th (1925) and 16th (19[36]-39) editions. Other important sources of potential enlightenment have limited themselves to morphological comments; thus, F. Hanssen (1913.§196) observed that desleir coincided in certain conjugational features with engreir 'to make conceited', freir 'to fry', reir 'to laugh', and sonreir 'to smile' and referred the reader to the classic treatment in A. Bello's Gramatica de la lengua castellana, §547. Other scholars, while offering bold hypotheses of their own or digesting earlier explanations or engaging in both activities (such as R. J. Cuervo in 1893, see below), were careful to label desleir as a word of obscure descent. Because the only two derivatives so far identified, desleidura and desleimiento - both abstracts of old vintage - hardly throw any striking light on the prehistory of the word or on the configuration of its family, no one, it seems, has attempted to capitalize on them. Some authorities, e.g., W. Meyer-Lubke (1911-20.§2671), credit W. Forster (1882.108f.) with having taken active part in the discussion of desleir. This is inaccurate, since Forster scarcely went beyond untying the tricky tangle délai (the actual source of E. delay) and delayer 'to dilute' - in a note that gave rise to many heated discussions (G. Paris, A. Horning. E. Littré, G. Körting, Α. Scheler - especially in his 1887 Supplement to Diez's dictionary), which can be safely overlooked in the context of the present paper. For this reason I refrain from burdening my inventory with any such etymon as *DISLĀTĀRE or *DISLACĀRE. 4) Actually, recorded Latin lacked any such verb; the closest approximation to it on the formal side, DĒLIGĀRE, far from reversing the message of the primitive, tended to intensify it: 'to tie up, bind together, bind fast'. Cf. the relevant statement by Ernout and Meillet (195960.358). What motivated Covarrubias was the emergence, within Spanish, of an important verb desligar 'to untie, unbind, unravel, disentangle, exempt' beside desliar 'to unpack, unravel' (from lio 'bundle, package'). Here is Covarrubias' argument in his own wording: 'desatar alguna cosa y hazerla liquida, y vale tanto como desligar, de donde se dixo desleydo lo desatado, que se ha hecho potable', with a surprising hint of a powder or a solid substance dissolved in a liquid (1611.f.310r° , or 1943.459b). 5) DĪLUERE was absorbed through learned channels as diluir1 - a verb which suffered severely from collision with a homonym, diluir2 'to deceive', flanked in turn by a now forgotten satellite, dilusivo 'producing delusions'; the source seems to be DELUDERE. P. F. Monlau,


519

(whose dictionary to this day remains unpublished); the following century, it was revived by the Spanish Academy in the original, elaborate edition of its dictionary, and thence penetrated into R. Cabrera's posthumous vocabulary — a record of longevity in itself striking; even more curious is the fact that this derivation — ob viously untenable - still haunts the latest available, 19th, ed. of the Academy Dic tionary (1970.456c)!6  Though F. Diez, selective in his coverage, could easily have shirked the discus sion of desleir, he offered a provocative analysis of the Spanish verb, misplacing that item — unfortunately - s.v. It. dileguare (1853.123).7 His starting point was the re-discovery of the variant form deleido in one of Berceo's saint's Hves, a reading that has, incidentally, been confirmed in subsequent improved editions of the text.8 Though Diez was slightly inexplicit about his step-by-step advance, one gathers in his dictionary yet to be mentioned, had a single entry for these two semantically irreconcilable verbs. The original Academy Dictionary (1732.282b) documented düusivo as an equivalent of phantástico with a passage from Bachiller Alonso de la Torre's Vision deleytosa (or delec table). 6) For the amazing, uninterrupted vitality of this conjecture, based on a striking semantic resemblance, see the long string of Academy dictionaries from (1732.190b) to (1970.456c). Rosal's original verdict, traceable to ca. 1601, ran thus: "De DILUERE, lat., que significa deshacer  desatar enagua"; see the entry in S. Gili Gaya's composite vocabulary (1957.776c). R. Cabrera, as a member of the Royal Academy, may have had access to Rosal's manuscript, or may simply have consulted the Academy Dictionary in compiling his own Dìccìonarìo de etimologias. I am skipping here two obvious infelicities of 18th-century scholarship: first, J. de Siesso  Bolea's attempt (ca. 1720), reported by Gili Gaya (loc. cit.), to both produce a rapprochement between desleir and deslucirse 'to become tarnished, lose luster' and appeal to Fr. lueur 'light' as the ultimate source; and, second, P e M. Larramendi's alternative derivation from Basque desleyatu (based on leya 'cold'), which Monlau stülmentioned in the mid-19th century, but preferred not to dignify with any explicit refutation, while Diez traced Euskaric desleyatu and (l)eya to Romance models; see below. 7) Occasionally, Diez would hit upon the bizarre idea of listing a word with, or under, its assumed congeners even where he unequivocally rejected that assumption. The mere fact that so prestigious a scholar recorded desleir in the company of It. dileguare, Prov. deslegar, and Fr. délayer 'to liquify' channeled the thinking of his admirers into certain grooves and in the end actually misled them. Diez's analysis appears unchanged in the 3d ed. of his dictionary (1869.153), which afforded him the last chance for revising his opinion. Diez repudiated DËLIQUESCERE and DĪLUERE as suitable bases, implying that Sp. e could not possibly reflect either i or  [but what of reir RIDERE?]; also, he disavowed Larramendi's Basque solution, arguing that B. desleyatu and the seemingly underlying noun (l)eya 'cold' were, in point of fact, them selves of Romance parentage (from Prov. deslegar and Sp.yelo 'frost', respectively). Previously, in Vol. 2 of his comparative grammar, Diez had listed desleir among Spanish -ir verbs with a nuclear -e- traceable to ancestral i (1838.152), but had almost pointedly refrained from etymo logizing it, and to the end cultivated the same restraint in subsequent editions, however radically brought up date in other respects. 8) The two key passages of the Vida de Santo Domingo de Silos adduced by Diez read thus in J. D. Fitz-Gerald's 1904 edition (the editor cautiously omitted the word from the appended etymological glossary): "Avja de la grant cuyta los mienbros enflaquidos,/ las manos e los piedes de su siesto exidos,/ los ojos concobados, los braços deleidos" (quatr.540a-c; MS H: desleydos; in Fray Sebastian de Vergara's text [Madrid, 1736]: desleidos); and: "Todos dizian que esta era virtut cumplida,/ que sanò tan ayna cosa tan deleyda,/ ca tanto la contavan corno cosa transi-

520

ETYMOLOGY

that the alternation of the initial segments des- and de- suggested to him the wisdom of slicing off these segments as competing prefixes. Once this much was accom plished, Diez connected the remaining 'torso' leir with the Greek verb on the further assumption that the physical state of the protagonists of this particu lar episode in the given hagiographic text, namely a procession of paralyzed persons, provided a clue to the etymon. The conclusion that Diez drew from his observation has been unanimously rejected by his followers, starting with Monlau, Korting, and Cuervo, but the broader idea of closely watching the initial syllable caught his successors' attention.9 d. Still at the mid-century point, P. F. Monlau, who as a rule almost slavishly followed Diez, for once rebelled against him (and against Larramendi, a fanatic Bascophile, and Cabrera as well), arguing that DISLIQUÉRE 'to turn liquid', if not the corresponding inchoative in -ËSCERE, could perfectly well qualify as a base, on the strength of the equation of Lat. FRÏGERE 'to fry' > Sp. freír (1856, 1881; 1941.598Zb).10 e. After a period of relative doldrums,11 the controversy once more picked up momentum in the concluding decades of the past century. The first attempt at a bold innovation was made by a brilliant young scholar, R. Thurneysen, who, as a close and appreciative student of J. Cornu's, had received a thorough training in Romance and was just beginning to turn his attention to Celtic, a domain in which da" (590a-c). Actually, as Cuervo was to show in 1893, the word appears in yet another passage of the same text: "Aviénna deleyda los dolores cutianos" (589d; HV: desleída); quotation here emended after the 1904 edition. Incidentally, T. A. Sánchez already recognized deleido and desleído as mere variants and translated them accurately enough ('flojo, laxo, debilitado'), but abstained from supplying any etymon (1780.499, 501). Strikingly enough, Mark G. Littlefield's forthcoming paper "Riojan Provenience of Escorial Bible Manuscript I.j.8"(to appear in Romance Philology) contains evidence of wavering between de- and des-leir. On the other hand, the Alfonsine General Estoria includes, as Dana A. Nelson reminds me, the passage: "Desleyré de la faz de la tierra.. ." (1.26b39). 9) To the point of inducing some of them to seek the etymon in del- rather than in -le-. 10) Having direct access neither to the 1856 princeps of Monlau's dictionary, nor to the posthumous 1881 edition (seen through the press and slightly revised by the author's son), I must rely on the 1941 reprint issued in Argentina. It is not immediately clear whom Diez at tacked in repudiating DÊ-LÏQUÉSCERE and for whom Monlau thus took up the cudgels. 11) Note, e.g., the undoubtedly deliberate omission oí desleír by C. Michaelis de Vasconcelos from her elaborate inventory of Spanish doublets (1876.279-300), despite the obvious tempta tion to pair it off with diluir; she did include (des)liar ~ (des)ligar (285b, 291b). With scant enthusiasm, P. Forster tersely traced desleír to LIQUÉRE (1880.419), but altogether neglected to etymologize it in examining such verbs as switched from the -er to the -ir class (§417). D'Ovidio spoke of -LIQUERE (1884.87), which Meyer-Lübke later interpreted as *DISLIQUËRE. G. Kõrting, on three occasions, routinely listed the Spanish verb (which he consistently misspel led) s.v. *DISLIQUÃRE (1891.§2613; 1901 and 1907.§3011), only to declare that it was best understood as an offshoot either of *DISLEGIRE (from LEGERE 'to pick, select, read') or of LIGARE (which of course involved the infelicitous revival of the long-since-buried hypothe sis of Covarru bias). J. Alemany Bolufer, in his ill-fated historical grammar, supplied no clearcut etymology, but broadly hinted that desleír figured among the -ir verbs weaned away from the-er class (1903.129).


521

he was to excel in later life. His doctoral dissertation — which embodies a critique, from a Celticist's vantage, of Diez's etymologies, in the order of the entries in the master's comparative dictionary — contains an important pronouncement (1884.56) which, if one discounts Larramendi's dilettantish proposal, represents the first non-Latin conjecture for the descent of desleir. Thurneysen declared flatly: "Sp. desleir entspringt offenbar einem keltischen Stamme", then arrayed a number of words, including OIr. legaim 'to melt, dissolve' (speaking of snow), 'to dissolve' (speaking of corpses), Gael. NIr. leagu- 'to melt, liquify', Cymr. lleith, mod. llaith (from a radical lecto- or licto-) 'death', i.e., 'dissolution', also, adjectivally, 'moist, liquid', etc. 12 Thurneysen's book received an accolade, despite several demurrers, from the most authoritative critic then available anywhere, Hugo Schuchardt,13 and, equally important, just a few years later, Meyer-Lübke unhesitatingly endorsed the Celtic ancestry of Sp. desleir in one of the opening paragraphs of his tonesetting Romanische Lautlehre (1890.45). It was the sole Spanish verb so classified by Meyer-Lübke, consequently a real isolate; this anomaly apparently eluded the comparatist's own attention. f. The other significant innovation came from the Paris headquarters of the Colombian philologist R. J. Cuervo (1893.1095b-1096b). An unhurried, methodic worker, Cuervo assembled a valuable corpus of syntactico-semantically slanted attestations of Golden Age usage. Closer to home, he provided an aftermath of medieval illustrations, in addition to the already established preferences of Berceo: excerpts from 13th-century Bible translations, from D. Juan Manuel's treatise on falconry, from an old collection of lyric poems (the Cancionero de Baena) reflecting courtly taste, ca. 1400. Important details then came to light: The prefix des- was shown to have alternated with es-; also, in the translated passages vernacular (d)esleir corresponded to the Latin model text's DELERE 'to destroy'. Adding to these findings the two stark facts that, first, in Portuguese - a language so far badly neglected in the course of the entire debate — delir, echoing DELERE, was the actual counterpart of Sp. desleir, and that, second, direct descendants of DELERE had indisputably entrenched themselves in Occitan-Catalan-Valencian territory, Cuervo, imaginatively, established the chain: [*] des-delir [i.e., reinforcement of an intrinsi cally negative idea by a negative prefix, cf. Sp. desnudo alongside Ptg. nu 'naked'] > [metathesized] [*] desledir > desleir.14 He erred in gratuitously assuming that des- was prefixed to, rather than superimposed on, the segment de- of delir.

12) Thurneysen further recognized in Bret. leiz, (Vannes) leic'h a Continental prong of the Cymric word and declared the verbal derivative dadleithio the immediate source of desleir, adding for good measure to the credit of the Celtic base Fr. délayer, alayer, alier (from W. Förster's treasure trove). 13) See (1885.1104). Unfortunately for us, desleir received no special mention. 14) By way of afterthought, as it were, Cuervo remarked that LEGERE, which he ferreted out as an alternative proposal from Körting's pile of suggestions, might also qualify as a base, if one recalled esleir 'to choose' < Ē L I G E R E ; this analysis would make Sp. desleir ('to decom pose', etc.) the semantic opposite of Lat. COLLIGERE 'to conjoin'. Among other felicities, Cuervo showed how in the Ferrara Bible desleiste corresponds to Lat. allisisti (Is. 64).

522

ETYMOLOGY

g. The first two decades of the 20th century witnessed two major innovations. Menéndez Pidal, whether entirely on his own or after consulting earlier statements made by Körting and Cuervo, somehow surrendered to the magic of the Old Spanish verb for 'to choose', namely esleir, the predecessor (though not the direct ancestor) of modern learnèd elegir. If esleir, he argued, descended from Lat. ĒLIGERE via some such Vulgar Latin link as EXLEGERE, then desleir could be traced to no safer base than DILIGERE. Menéndez Pidal's argument that the Latin prefix di was bound to yield des- in Old Spanish was flawless; unfortunately, DILIGERE, in Latin, signified 'to single out, esteem, love, value', a range of meanings which clearly lacked the slightest contact, let alone overlap, with 'dissolving' or 'diluting'. Thus the conjecture, not devoid of grammatical finesse, remained singularly disappointing on the semantic side.15 h. Meyer-Lübke's early advocacy of Thurneysen's Celtic hypothesis has already been placed on record. In preparing the original edition of his Romance etymological dictionary, on the eve of World War I, Meyer-Lübke, after reconstructing the type *deslegīre (No. 2671), tagged it as an item of obscure provenience, conceivably Celtic; the definitive version of that dictionary, issued in the early 'thirties, was far more positive in this commitment.16 In the meantime, Meyer-Lübke was gradual ly becoming alert to a seemingly unrelated issue, namely the survival in Romance of Lat. DĒLĒRE 'to destroy'. By 1911, he was ready to admit that Prov. delir 'to melt' should be so explained (REW,1 No. 2533); twenty years later, after perus ing the crop of relevant studies by several younger scholars (O. J. Tallgren, M. de Montoliu, J. Jud), 17 he was willing to throw in for good measure Cat. dalirse 'to strive hard, be grieved' (= G. schmachten), but made an obvious slip in classifying newly-supplie d data from the Upper Garonne (= Aran) Valley, by crediting delir 'to melt' (speaking of snow) to DELERE, yet its transparent variant and synonym

15). See the later editions of the Manual de gramatica historica espanola (1918, 1925, 1940. §126); the statement was absent from the original edition (1904) of the textbook. While the author's remarks on the convergence of ancestral dī- and dē- in vernacular des- are unassailable (cf. DEVIARE > desviar 'to turn aside, deflect', DĒDIGNĀRĪ > desdenar 'to scorn'), the leap from DILIGERE 'escoger, preferir' to [*] DISLEGERE 'disgregar', with a reference to obsol. esleir 'to choose', remains baffling, to say the least. The author noted certain semantic discrepan cies, as when DĪLŪCĒRE meant 'to be clear, evident, obvious', while deslucir signifies 'to tar nish'. But then - one may ask - what is the reason for genetically linking deslucir to DĪLŪCĒRE, rather than analyzing deslucir as a neologism, arrived at from lucir? Cf. the above-cited analogous case of desligar. 16) The original entry contains a succinct negative assessment of several earlier hypotheses; lists Ptg. deslir [sic] as an alleged counterpart of desleir; and qualifies Thurneysen's alternative solution as a mere possibility. The revised entry refers to Cymr. dadleithio (unearthed by Thurneysen) and identifies J. Bruch (Brück is a patent misprint) as a further advocate of the Celtic conjecture. For Aran V. deli, Meyer-Lübke specifically rejects J. Corominas' champion ship of DĪLUERE (1925.66). 17) In his spirited review (1915-17.291) of Vol. 2 of the Butlleti de dialectologia catalana, Jud wondered, apropos of M. de Montoliu's elaborations on  J . Tallgren's "Glanures catalanes et hispano-romanes", whether Cat. dalirse should not be traced to Lat. DÈLËRE on the semantic analogy of G. sich verzehren, It. struggersi {di desiderio), Sp. deshacerse.


523

deli to Celto-Lat. *dislegīre. Taking that inconsistent decision obviously meant courting disaster.18 i. The immediately following decades produced no upheaval with respect to desleir itself, but led to renewed and deepened examination of the possible relics of DËLËRE, a propos of such genetically controversial words as the adjectives deleznable 'perishable, slippery, crumbly' and OSp. lezne 'smooth'. Some glottogeneticists leaned in the direction of DËLËRE, others favored Lat. LĒNIS 'smooth, gentle' as the ultimate source;19 the truth may lie in a compromise between these two 'extreme' solutions, because receding words, in the last phases of their eva nescence, are known to be particularly prone to all sorts of blends and disguises. j . The gradual emergence of DËLËRE as a serious contender was reflected in the verdicts returned by V. Garcia de Diego in the two sections of his etymological dictionary (1954.225b, 725). The author, conservatively enough, maintained that Sp. desleir, OSp. deleir, Ast. esleir, along with Piedm. deslué, all four go back to DĪLUERE, but admitted that in certain border-line cases, notably those of Cat. delir, Aran Valley deli, and Ast. delida 'diluida', it remained uncertain whether DĒLĒRE or DĪLUERE was actually involved. There is, of course, no need for us in the late 20th century to refute expressly, on the evidence of the stem vowel, the choice of DĪLUERE as the base of de-, (d)es-leir; however, the allowance the author made for a partial alternative is decidedly noteworthy.20 18) Rather characteristically, neither Jud in 1915, nor Meyer-Lubke sixteen years later, bothered to consult Cuervo's admirable 1893 assessment. 19) As a rank beginner (1944.57-65), I argued that deleznar 'to slip, slide' and the corres ponding adjectives in-able (also -adero, -adizo) might involve DĒLĒTIONE destruction', in which case obsol. lezne could be interpreted as a postverbal adjective. Corominas' critique of that note (1947.73-76) - he had become sensitized to the issue of dali, dalir since his first writings in the mid 'twenties - involved total rejection of my hypothesis, in favor of LĒNIS 'smooth', an alternative which, I admit, involved a more cogent explanation of the -e, but failed to account persuasively for the striking -zn- cluster. The survival of LĒNIS had, incidentally, been posited long before 1947 - e.g., briefly, by F. B. Navarro in 1879 and, more circumstantially, by A. G. Solalinde (1925), who discovered an Old Leonese var., desnalabre, best interpreted as a distortion of (isolated) des-len-able After taking cognizance of other reactions to my note, set in a very different key from the one that presided over Corominas' demurrer - see J. M. Piel (1949.285f.) and L. Spitzer (1945.298) - I offered a far more nuanced and compromising reconstruction of the sequence of events in the context of a derivational study concerned with postverbals (1960. 201-253, at 248-251). That summation also took into account two dictionary entries by Corominas which had followed upon his 1947 note (1955.120b-l21a; 1957.986). Perhaps I should have emphasized more heavily the fact that the advanced decay of the LĒNIS and the DĒLĒRE families best explains the contamination of their residues by other lexical units en dowed with equal or superior vitality - a type *licin- 'to glide' or the prototype of Ptg. lesma, OPtg. lezme 'snail'. 20) There are, as one would expect, numerous incongruities and inaccuracies in the author's statements. He does not bother to explain how the U of DĪLUERE (an early 17th-century conjecture, it will be remembered!) could possibly have cast off the nuclear e of desleir, although he is quick to pounce on those aiming at a rapprochement between Fr. delayer and Sp. desleir: "A pesar del parecido de forma y de sentido del fr. délayer y del esp. desleir, solo podria explicarse desleir como un galicismo [i.e., not as a congener], porque en la fonetica espanola es imposible la pérdida de  de dèliquare".

524

ETYMOLOGY

. In a rewarding and, in many respects, conciliatory entry of his monumental dictionary (1955, 143b-145a), J. Corominas cautiously favored deriving desleir from DĒLĒRE, without completely ruling out some of the rival hypotheses, except for Thurneysen's Celtic conjecture, despite the additional support that 'Einfall' had meanwhile received from J. Bruch.21 Even Covarrubias' quaint desligar was granted a marginal chance of plausibility, provided one replaced -ligar by -liar; the questionable derivation of Ptg. delir from DĪLUERE was likewise treated with kid gloves, on the strength of such equations as CO(N)S(U)ERE 'to sew' > Sp. Ptg. coser, CÖ(N)SP(U)ERE 'to spit, spew' > Sp. es + cupir, Ptg. cuspir.22 On balance, however, Corominas subscribed to Cuervo's advocacy of DĒLĒRE, which had meanwhile gained a lively supporter in Leo Spitzer, even though the younger scholar did not necessarily champion every argument advanced by those two predecessors.23 On the positive side of the ledger, let me mention Corominas' fine-meshed semantic analysis; using medieval texts, Renaissance dictionaries, and the older Judaeo-Spanish Bible translations, he made it irrefutably clear that the word, at the outset, was not referentially confined to liquids (Academy, 1732.190b: "separar  desatar las partes de algún cuerpo con algún liquor"), but rather contained, at the center of its spec trum, the allusion to 'weakening, softening', from which the other semantic nuances could then easily have flowed. Once the older semantic contour has been properly silhouetted, it is easy — Corominas contended — to establish a link to OPtg. delir,

21) See (1922.71), with a reference to Sp. derretirse, Fr. fondre. Corominas' three-pronged argument against the Celtic extraction is: (1) our newly-gained knowledge of the bonds linking Fr. délayer to Occ. deslegar, a connection which short-circuits any dependence on Celtic; (2) the emptiness of the prefix des- in this context; (3) the fact that verbs of proven Celtic ancestry were assigned to the -ar class in Spanish and Catalan, and to the corresponding -er class in French - nowhere to -ir. The first and the third points are well taken. 22) Corominas cites among the advocates of DĪLUERE J. Cornu (1906: §§117, 320) and O. Nobiling (1910.397), without making it clear that the conjecture goes back to a distinctly older period. Also, the parallelism between CO(N)S(U)ERE, CÖ(N)SP(U)ERE, and *DĪL(U)ERE that Corominas toys with, in the wake of those pioneers, hardly carries conviction, because CON- was, semantically, a rather pale suffix, while DĪ-, as shown precisely by its further morpho logical elaboration to des-, was endowed with vigor, a circumstance which, in a crisis, surely would have protected the nuclear syllable -lu-. The references to Cornu's and Nobiling's pro nouncements, incidentally, bear elaboration. In the two paragraphs of his 1888 grammar cited by Corominas, Cornu made no direct, binding reference to the filiation of Ptg. delir; in the 1906 revision, he was ambiguous enough to trace the verb, in a single context (§117), to DĪLUERE and DËLËRE; did he imply a blend of the two? Nobiling applauded Diez and Körting (1907) for rejecting DIS-LIQUÖ, while leaning himself toward DĪLUERE. His strength lay in isolating the segment -eir, which had emerged in lieu of *-uir [or, rather, *-oir], as the actual stumbling block, i.e., the crux was the replacement of the back vowel by a front vowel, a phenomenon which he viewed as a peculiar 'Lautwandel'. In citing, as a reverse process, possuir, he neglected to point out that OPtg. posseer > mod. possuir showed attraction by the (semantically kindred) rep resentatives of STATUERE and TRIBUERE. Nobiling's carelessness in matters of vowel quanti ty (e.g., his disregard of the length of i in DĪ- and the brevity of  in -LUERE) slights his analysis of Ptg. de- vs. Sp. de(s)- and of the assumed substitution of e for the back vowel. 23) Spitzer's incidental remark appeared in a footnote (1921.136) to a miscellany of etymo logical vignettes, which later provoked J. Bruch's deviant verdict (1923.71).


525

already examined in a magisterial study by C. Michaelis de Vasconcelos (1910.298),24 as well as to OProv. delir 'to ruin', mod. Occ. (a)deli 'to exhaust', which reappears in at least one variety of Aragonese (Venasque), Béarn, deli 'to melt' (in reference to snow or wax), and Cat. delirse 'to yearn, be consumed by desire'. This portion of Corominas' analysis seems to me above reproach. But I refuse to go along with his far-fetched explanation of the actual shape of the verb in the Old and Modern Spanish. DĒLĒRE, he maintained, produced delir in the West and in the East of the Peninsula, yet desleir in the Center. So far so good; yet at this point, he made the fatal mistake of guessing — counter to all probabilistic analyses — that desleir was the original, uncontaminated product of DĒLĒRE, traceable in a straight line to provincial Vulgar Latin, while delir represented a subsequent adjust ment: "Todo lleva a suponer que de las dos formas delir y desleir ésta es la más primitiva y no es verosimil que se deba a una ultracorrección" (144b). It will before long be my responsibility to demonstrate that, beyond the shadow of a doubt, the reverse sequence of events was true. 25 For the time being, if we attempt to tabulate the etymological preferences so far caught in our net (omitting a few obvious irrelevancies such as Fr. lueur, délayer, plus a Basque word family, and transferring from the footnotes a couple of data of only incidental pertinence), the following picture unfolds before our astonished eyes: 1. *DIS-LIGĀRE, semantically divorced from DĒ-LIGĀRE: Covarrubias (1611); briefly remembered by a few moderns, from Körting (1891) to Corominas (1955). 2. DÏLUERE: Rosai (1601), Spanish Academy (1732), Cabrera (before 1833), Nobiling (1910); balanced against DĒLĒRE by Cornu (1906) and Michaelis (1910), also, along a slightly different line, by Garcia de Diego (1954); treated without harshness by Corominas (1955); favored to this day by the Royal Acad emy (1970). 3. (Greek) (para)luein: Diez (1853, 1869); rejected by everyone else. 24) Etymologically, the author - under the influence of Cornu's 1906 statement, then fresh in her memory? - remained undecided as between DĒLĒRE and DĪLUERE. In part, one gathers, on the strength of the erratic survival of intervocalic -/-, in part in deference to the cultural context involved, C. Michaelis labeled Ptg. delir a semilearned or mock-learned word of gastronomic and pharmaceutic background. Important is her emendation of an oft-cited, visibly corrupted passage in the single extant MS of Don Juan Manuel's treatise on falconry: She corrected the MS's unparalleled desdeyr to better-integrated [*]deslyr, which, if independent ly confirmed, would indeed form the perfect link between Ptg. delir and Sp. desleir. The author clearly understood the encroachment of des- on the domain of de- and offered a parallel from Mestre Giraldo's text under scrutiny: the wavering between de- and des-folgar 'to breathe'. 25) Not unexpectedly, Corominas' entry (including the supplementary remark, 1957.987a) contains a heavy quota of infelicities, some of which testify to his partisanship. Thus, he remains tight-lipped about Menéndez Pidal's serious blunder, while expatiating on any slip made by his own arch-rival, Vicente Garcia de Diego, even on that etymologist's fairly irrelevant comment on OSp. deçir 'to descend'.

526

ETYMOLOGY

4. DIS-LIQUERE: Monlau (1856, 1881); LÏQUËRE: Förster (1880) and D'Ovidio (1884);DĒLĪQUĒSCERE: unidentified pioneer. 5. (Celtic) leg-/ leag-/ lleith: Thurneysen (1884), applauded by Schuchardt; *DIS-LEGÎRE: Meyer-Lübke (1890, 1911, . 1931), Bruch (1922). 6. DĒLĒRE: Cuervo (1893), with a reference to Ptg. delir. For the more lukewarm reactions of Cornu, Michaelis, and Garcia de Diego, see under DĪLUERE, above. Corominas (1955). 7. DĪ-, *DIS-LIGERE (from LEGERE): Menéndez Pidal (1918-41), after earlier hints by Körting and Cuervo. 8. Remnants of DĒLĒRE: Ptg. delir, Cat. dalir, Aran V. dali, Sp. deleznable (blend with LĒNIS?), etc.: Jud (1915-17), Malkiel (1944, 1960), Spitzer (1945). It certainly is difficult to imagine a higher degree of confusion. Rather than continuing in a critical, let alone polemic, vein, let me abruptly change the perspective at this point and provisionally admit that, with all the fresh and copious evidence from Portuguese, Old Spanish, and Catalan sources superadded, in over eighty years of continuous research, to Cuervo's original illustrations, his seminal idea of DĒLĒRE 'to destroy' being at the root of desleir 'to weaken, dilute' appears to have been essentially correct. The stumbling blocks that continue to obstruct our path are, obviously, neither of the geographic nor of the culturalsemantic order; they indisputably pertain to straight grammar, or rather to unjustly neglected nooks or corners of historical grammar. It is for this reason that the ety mological analysis of the chosen crux has been presented before this forum — to expose and, if possible, remedy the hazardous imbalance between lagging gram matical knowledge and fervid, uncontrolled etymological imagination. Let us take up the hidden difficulties one by one. First, DĒLĒRE displays as its opening segment dē-, not des-. Undeniably, there has occurred at the Romance level a tendential convergence of Lat. dē- 'down' and dis-, var. dì- 'asunder', usually in favor of the local descendant of dis-, which in Hispano-Romance happens to be des-.26 The chief reasons for this process were, on the one hand, certain formal-semantic affinities between dē- and dis-jdī-, starting with the identity of the first consonant phoneme and reinforced by the character istic merger of the vowels ē and i in most Romance languages; and, on the other hand, the less frequently cited fact that the majority of the traditional HispanoRomance prefixes — at first, no doubt coincidentally — end in -s, witness es-, as in esforçar to strengthen'; pos-, as in posponer 'to put aside, delay'; res-, as in resfriar 'to cool'; sos-, as in sostener 'to uphold', sostentar 'to sustain'; and tras-, as in tras passar 'to pierce, torment', trastornar 'to overturn'; so that des- fitted a ready-made 26) To be sure, one encounters scattered vestiges of atypical developments; thus, OSp. debuxar  draw, depict' (Ruiz) ended up by acquiring the ill-chosen prefix di- rather than des-; pseudo-learned Sp. disminuir conflicts shockingly with E. diminish, Fr. diminuer, etc. These few aberrant evolutionary lines hardly suffice to blur the total impression of the develop ment. (Cf. also the remark, above, on Sp. diluir 'to deceive' -ir), there remains the thornier question of the Castilian preference for -eir, as against Ptg. -ir and Cat.-Occ. -i(r). Upon the admission of qualified judges

29) As in pescuezo 'neck' and in pespuntar 'to backstitch', where vowel dissimilation (remi niscent of redondo 'round' < ROTUNDU) was operative.


529

— such as Nobiling and, later, Corominas — this is, in fact, the crux of the entire problem. And yet the ultimate source of the difficulty is not the inherent complexity of the given word biography, but the woefully meager share of attention scholars have been lending to the subtler filigree of historical grammar. Let us, once more, change the position of our camera and formulate our next query thus: Should, in light of the analysis so far proffered, the 'ideal' product of DĒLĒRE in Old Spanish, at the start, have been delir, as in the Peninsular lan guages spoken to the west and to the east of Castile, what could have prevented CastiHan speakers from accepting delir and prompted them to adopt, instead, a bizarre innovation, such as de-, subsequently des-leir? If our argument is to carry conviction, the feature we are searching for must come close to being distinctive, exclusive. I venture to think that a satisfactory answer can indeed be provided. The key to the problem is the chosen speech community's strong unwillingness, at the crucial juncture, to tolerate monophonemic verbal root morphemes. This attitude can be inferred from the jettisoning of Lat. EDERE 'to eat', which might have yielded such an unpalatable infinitive as *e-er or *e-ir; from the dissimilation, in the wake of the disappearance of that verb, of friir 'to fry', from FRÏG-ERE, -ERE, to freir (even though no immediate danger of shrinkage to an ultra-short radical was involved), as well as of frequently-attested riir 'to laugh', from RIDERE, to reir. Consider further the following cases: Lat. LEGERE 'to read' advanced to leer, which was not allowed to progress farther, to the expected 'logical' conclu sion of the process, namely *ler; neither was creer 'to believe', from CREDERE, compressed to *crer, and even obscene PĒDERE stopped at the stage peer, and the speakers resolutely shunned *per. (Alternatively, the West has had recourse to peidar, from the corresponding noun peido < PĒDITU 'crepitus ventris'.) Nothing dramatizes this peculiarity of the meseta dialects more effectively than the recital of the Portuguese counterparts: crer, 1er, rir, beside aberrant frigir. Small wonder the lusófonos have in the end also learned to live with por 'to put', ter 'to have', and vir 'to come' for older poer, teer, and vīir, namely Lat. PÖNERE, TENÉRE, VENIRE. At first glance Spanish offers two seemingly glaring exceptions from this rule. But, exceptions though they are, at second thought I hesitate to label them as 'glaring'. The local form of SEDERE 'to sit', after centuries of survival as seer, acquired, in addition to its traditional meaning ('to sit'), the incomparably more common meaning 'to be', and, as an existential verb, entered into an intimate para digmatic relationship with remnants of the Latin auxiliary ESSE 'to be', e.g., era 'I, he was'. In such forms of the motley paradigm — arrived at through suppletion — as so(y) "I am', somos 'we are', sodes > sois 'ye are', son 'they are', the radical was, from the start, s-; small wonder the inf. se-er became ser and the p. ptc. seydo followed suit, shrinking to sido. But note that the old seer haslingeredd on rather comfortably in the compounds, of which the best-known is poseer 'to own, possess' (of much smaller importance have been entreseer 'to be between' and soseer 'to underlie'). Again, the hypothesis here championed may be impugned with a reference

530

ETYMOLOGY

to ver 'to see', originally veer, and, once more, persuasive counterarguments can be cited. Among the compounds of veer at least one, proveer 'to furnish, provide', has parted company with veer, resisting the merger of the two consecutive e's unlike, it is true, entrever 'to glimpse, descry, guess', prever 'to foresee', and rever 'to revise, review'. But all five of these verbs, however divided in other respects, form their imperfect tenses in -eia, a segment which presupposes the bi-phonemic root-morpheme ve֊ rather than v-; the same full-bodied root-morpheme makes its appearance in the pres. subj. ve-a, which, incidentally, matches sea < Lat. SEDEAM, from ser. To put it differently, ser and ver are exceptions only in part, chiefly on the level of their infinitives. A genuine exception has been, at all times after the loss of its consonant pillar -d֊, oir,'to hear' (< AUDIRE); also, in the end, fuir, huir 'to flee' (< FUGERE) became one after the erosion of its initial consonant; to these two must be added ir 'to go' (< ĪRE). It is, consequently, more accurate to invoke an all-pervasive tendency in the -er class and no more than a very strong trend in the -ir class.30 How does this structural analysis affect our concrete etymological equation? Delir must have grated on the nerves of the early Spaniards, while delighting then Portuguese cousins, on account of the restriction of the stem to a single phoneme, namely the -l-, once speakers had begun to peel off de֊ as a prefix (and we know for sure that they did , in view of the successful substitution of des- for de-, beauti fully documented in the extant Berceo texts). If delir, at a given turning point, began to irritate the Proto -astilians, they surely lost no time in casting around for a solution, and the opportune change of riir to reír (also, from SURRÏDÉRE, of sonriir 'to smile' to sonreír), of friir to freír, of ÉLIGERE to esleír, etc., may very well have encouraged them to experiment with desleír. Fifth, and last, the one argument that, under all and any circumstances, had better be avoided in any effort to explain away the enigmatic -eir segment — both because the suggestion is wholly irrelevant and because it gratuitously opens up a box of Pandora - is the weak preterite of the Latin -ĒRE class. Unfortunately, Corominas, in his otherwise stimulating entry, fell into this trap. For one thing, while no one would deny the existence in folk Latin of such paradigms, in the -ĀRE and -ĪRE 30) This is, obviously, not the place for expatiating on the fate of ultra-short verbal root morphemes transmitted into Hispano-Romance. VĀDERE 'to make one's way' coalesced with ĪRE 'to go', though INVĀDERE 'to attack' has managed to eke out a precarious existence as embaír 'to trick, deceive'. A few verbs switched to the -ar class: esmaýr became (refl.) (djes֊ mayar 'to faint, swoon' and (refl.) so(n)rreýr, for a while, tended to yield ground to so(n)rrisar. There occurred two important blends: one between (DIS)CĒDERE 'to go away, depart, scatter' and DĒSCENDERE 'to go, come down' (cf. OSp. decir), the other between IMPEDIRE 'to entangle, hamper', -PEDĪRE 'to unfetter, extricate, get ready', and PETERE 'to strive for, demand' (cf. Ptg. despir 'to strip' vs. Sp. despedir 'to emit, ray out, discharge'; OSp. empeecer 'to harm'). Many potentially troublesome verbs disappeared altogether: AUDĒRE 'to dare' (which might have yielded *oer), FODERE 'to dig', CAEDERE 'to fell' (as against CADERE 'to fall', which did strike root), LAEDERE 'to knock, strike, fell', LUDERE 'to play', (impers.) TAEDĒRE 'to irk'; GAUDËRE 'to rejoice at', observable in OPtg. gouvir, Fr. jouir, etc., gave ground to a circumlocution (gozar).


531

classes, as amai, amasti, amāut, amammus, etc., or, for that matter, finii, finisti, finlut, fīnimmus, etc., it is highly dubious that vernacular speech developed and preserved for any length of time a comparable pattern for the precariously-entrench ed -ĒRE class except perhaps, under optimal circumstances, for the 3d sg., *deleut, witness the solid evidence of Portuguese and, vestigially, of Old Leonese beside a few scattered traces in Old Aragonese. The reconstructions Corominas devises (leaving them, to boot, unstarred): deleit, deleiste, deleerunt, deleisset, deleerit (1955. 144a), strike me, first of all, as utterly unrealistic; second, even if a scintilla of real ism attached to them, they would never have produced the expected results because, to select just two examples, deleerunt and deleerit would immediately have under gone contraction of the two successive homogeneous vowels. And how can one, with a sober mind, attempt to justify a process best placed somewhere in the 11th or 12th century with oblique inferences from doubtful epigraphic material traceable to the eclipse of Antiquity? To end on an optimistic note, a very recent attempt has indeed been made to cut a narrow path through the jungle of the Proto- and Old Spanish weak preterite (Malkiel, 1976.435-500); had the results of that finemeshed inquiry been known to Corominas twenty years ago, he might have steered clear of the trap. 31 I have presented here, in almost forbiddingly microscopic detail, a single ety mological problem, involving a Spanish verb, desleír, and a few cognates, provided the relationship between them can be demonstrated or made plausible in the first place. To put it differently, the issue on hand is multi-faceted. While it is hardly surprising that the pioneers a century or so ago, to say nothing of their clumsy Renaissance precursors, fumbled again and again, it is astounding that the solution here held to be correct, proffered by Cuervo in 1893, could not, for eighty-two long years, have been accepted without strong qualifications. And yet Cuervo's interpretation appears to us in retrospect, except for a sprinkling of trivial details, neatly-phrased and persuasively-argued. Why then, one may ask, the disquietingly long stretch of time needed for its vindication? The answer lies in the lag of thorough grammatical knowledge, or so much as sustained curiosity, behind the fertility of etymological minds. To prove Cuervo

31) Let me grant, in fairness to Corominas, that his misinterpretation of the weak preterite in Hispano-Latin represents but a grotesque extension of a grievous error already found, in reference to other conjugation classes, in Menéndez Pidaľs handbook of historical grammar (1904. §118; never weeded out from subsequent editions, despite the addition of much valuable embroidery). The North and Central European Romanists (Cornu, Baist, Meyer-Lübke, Zauner, Staaff, to name but a few), however sharply divided among themselves on that issue, in unison refused to adopt Menéndez Pidaľs projection of the medieval and dialectal Spanish paradigm onto the level of Vulgar Latin. Conversely, members of the Madrid school, unfortunately in cluding one scholar as influential as R. Lapesa, repeated the error — forgivable against the backdrop of 1904, but inexcusable in the 'fifties. As if to further complicate a situation already well-nigh hopelessly confused, an almost playful Corominas threw in as a final possibility ֊ which he immediately withdrew - a Plautine verb, DĪLĪDERE 'to break' (from LAEDERE 'to hurt').

532

ETYMOLOGY

definitely correct, it was mandatory for his followers to have accurate and copious (if not necessarily exhaustive) information on a variety of grammatical points of, if you wish, secondary, even tertiary importance — but of importance all the same. Either the etymologist's team-mate, or one of his students, or he himself should have bitten his lips, bided his time, and elucidated, beyond hazy generalities, such thorny questions as the mutual relation of the prefixes de(s)- and di(s)-; the norms or ten dencies (if there are any) behind 'false separation' of affix and root morpheme; the relation between verbal composition and conjugation class; the almost consistent avoidance of ultra-brief ('monophonemic') root morphemes in Spanish verbs (except for ir, oír, [h]uír), though by no means in their Portuguese counterparts; and, as the unexpected result of Corominas' extravaganza, certain Hispanic vicissitudes of the Latin weak preterite. To paraphrase this remark, what has slowed us down for over eighty years has been a cultural caprice (if you so prefer, a fad): the extreme popularity, within one tone-setting and, in certain respects, meritorious coterie, of 'cute' etymological puzzles and the concurrent boredom surrounding 'dull' grammatical inquiries. Our diet has not been properly balanced since the fin de siècle. Add to this impediment the unreasonable demands of the publishing business and of the Establishment in general. Society wants each of us to engage in a clearcut, neatly defined project. On the surface, the pledge to write an etymological dictionary ranging from A to Z or, alternatively, a close-knit historical grammar seems to make excellent sense. Conversely, to be at work on a set of disconnected etymological problems and to interrupt that investigation, again and again, at the risk of getting side-tracked while studying disjointed points of grammar, reeks of bad, inefficient strategy. But the dictates of society exerting this kind of pressure are at variance with a higher order of postulates: The demands of intellectual integrity. One cannot play the etymological game, fraught with hazards, frivolousness, and excitement, without pausing, at frequent intervals, for unhurried clearance of the grammatical underbrush; just as one cannot honestly examiné the more complex network of recurrent relationships, that is to say, of grammar, without allowing for a modicum of leisure in focusing on problems of individual word biographies, i.e., of etymology. Society can persuade us to present in separate volumes, for the convenience of future readers and library cataloguers, the discrete results of grammatical and lexical investigations. But while we toil in the incandescent light of cognition, let us at no time forget the constant interlocking of etymology and historical grammar.


533

BIBLIOGRAPHY Academia Española, Real (1732), Diccionario de la lengua castellana (= "Diccionario de Autoridades"), Vol. 3. , (1970), Diccionario de la lengua española, Revised 19th ed., Madrid, EspasaCalpe (printer). Alemany Bolufer, José (1903), Estudio elemental de gramática histórica de la lengua castellana, Revised 2d ed., Madrid, Tipografía de la RABM. Bello, Andrés (f 1865) (1887), Gramática de la lengua castellana, 14th ed., Madrid, López. Brüch, Joseph (1922), "Zu Spitzers kat[alanisch]-sp[anischen] Etymologien in der Biblioteca Archivi Romanici II:I", Miscellanea linguistica, dedicata a Hugo Schuchardt per il suo 80° anniversario (1922) 26-74, at 71, Biblioteca dell'Archivum Romanicum 2:3. Cabrera, Ramón (f 1833) (1837), Diccionario de etimologías de la lengua castellana, obra póstuma, ed. J. P. Ayegui, 2 vols., Madrid, Imprenta de Don Marcelino Calero. Cornu, Jules (1888), "Die portugiesische Sprache", Grundriss der romanischen Philologie, ed. G. Gröber, 1.715-803, Strassburg, Trübner. , (1906), "Die portugiesische Sprache", Gröber's Grundriss, rev. 2d ed., 1.916-1037. Corominas, Juan (also Corominas, Joan, etc.) (1925), "Etimologies araneses", Butlletí de dialectologia catalana 13.64-70. , (1947), "Spanish deleznarse", Word 3.73-76. , (1954-57), Diccionario crítico-etimológico de la lengua castellana, 4 vols., Bern, Francke, & Madrid, Gredos. Covarrubias Horozco, Sebastián de (1611), Tesoro de la lengua castellana o española, Madrid, printed by Լ. Sánchez; and ed. Martin de Riquer, Barcelona, S. A. Horta, 1943. Cuervo, Rufino José (1893), Diccionario de construcción y régimen de la lengua castellana, Vol. 2, Paris, Roger & Chernoviz; reprint, Bogotá, Instituto Caro y Cuervo, 1954. Diez, Friedrich (1838), Grammatik der romanischen Sprachen, Vol. 2: Wortbiegungs lehre, Wortbildungslehre, Bonn, Weber. , (1853), Etymologisches Wörterbuch der romanischen Sprachen, Bonn, Marcus. , (1869), Etymologisches Wörterbuch der romanischen Sprachen, revised 3d ed., Vol. 1, Bonn, Marcus. D'Ovidio, Fr[ancesco] (1884), "I riflessi romanzi di viginti triginta", Zeitschrift für romanische Philologie 8.82-105. Dworkin, Steven N. (1973), "Latin SARCĪRE, SERERE, SUERE, SURGERE in Hispano-Romance: A Study in Partial Homonymy, 'Weak' Sound Change, Lexical Contamination", Romance Philology 27.26-36. Fitz-Gerald, John D. (1904), ed., Gonzalo de Berceo, Vida de Santo Domingo de Silos, Paris, Bouillon. Förster, Paul (1880), Spanische Sprachlehre, Berlin, Weidmannsche Buchhandlung.

534

ETYMOLOGY

Foerster, Wendelin (1882), "Romanische Etymologien, Nos. 31-32: Délai, dilayer ά délayer, frz.", Zeitschrift für romanische Philologie 6.108-9. García de Diego, Vicente (1954), Diccionario etimològico español e hispánico, Madrid, Gredos. Gili Gaya, Samuel (1957), Tesoro lexicográfico, 1492-1726, Vol. 1:4, Madrid, Consejo superior de investigaciones científicas. Hanssen, Federico (1913), Gramática histórica de la lengua castellana, Halle, Niemeyer (Reprinted, Buenos Aires, El Ateneo, 1945). Horning, Adolf (1883), Zur Geschichte des lateinischen  vor e und і im Ro manischen, Halle, Niemeyer. Jud, Jakob (1915-17), Review of Butlletí de dialectologia catalana 2, Romania 44.291-4. , (1922), Review of G. Dottin, La langue gauloise, Archivum Romanicum 6.188-211. Korting, Gustav (1891), Lateinisch-romanisches Wörterbuch, Paderborn, Schöningh. , (1901), Lateinisch-romanisches Wörterbuch, revised 2d ed. , (1907), Lateinisch-romanisches Wörterbuch, revised 3d ed. (Reprint, New York, Stechert, 1923). Larramendi, Ρe M. de (1745), Diccionario trilingüe del castellano, bascuence y latín, 2 vols., San Sebastián, Riesgo y Montero. Malkiel, Yakov (1944),"Spanish deleznar 'to slide', [OSp.] lezne 'smooth, slippery'", HispanicReview 12.57-65. , (1960), "Fuentes indígenas y exóticas de los sustantivos y adjetivos verbales en -e (II)", Revue de linguistique romane 24.201-53, , (1976), "From Falling to Rising Diphthongs", Romance Philology 29.435500. Menéndez Pidal, Ramón (1904), Manual elemental de gramática histórìca española, Madrid, Suárez. , (1918), Manual de gramática histórica española, Revised 4th ed., Madrid, Suárez. , (1941), Manual de gramática histórica española, Revised 6th ed., Madrid, Espasa-Calpe. Meyer-Lübke, Wilhelm (1890), Grammatik der romanischen Sprachen, Vol. 1: Romanische Lautlehre, Leipzig, Fues (Reisland). , (1911-20),Romanisches etymologisches Wörterbuch, Heidelberg, Winter. , (1930-35), Romanisches etymologisches Wörterbuch, Revised 3d ed., Heidelberg, Winter. Michaelis, Carolina (1876), Studien zur romanischen Wortschöpfung, Leipzig, Brock haus. Michaelis de Vasconcelos, Carolina (1910), "Mestre Giraldo e os seus tratados de alveitaria e cetraria, 2: Estudos etimológicos", Revista Lusitana 13.222432. Monlau, Pedro Felipe (1856), Diccionario etimológico de la lengua castellana, pre cedido de unos rudimentos de etimología, Madrid, Rivadeneyra (printer). , (1881) Id., revised posthumous 2d ed., issued by J. Monlau, Madrid, Aribau (Reprint, Buenos Aires, El Ateneo, 1941). Montgomery, Thomas (1976), "Complementarity of stem-vowels in the Spanish


535

second and third conjugations", Romance Philology 29.281 -96. Nelson, Dana A. (1972), "The Domain of Old Spanish ֊er and ֊ir Verbs: A clue to the provenience of the Alexandre", Romance Philology 26.265-303. Nobiling, sk(1910), [Segment of] "Berichtungen und Zusätze zum portugiesischen Teil von Körtings LRW", Archiv für das Studium der neueren Sprachen und Literaturen 125.393-7. Piel, J. M. (1949), Review of Malkiel, "Spanish deleznar . . . ''Revista portuguesa de filologia 2.,285 -6. Rosal, Francisco del (ca. 1601), Origen y etymologia de todos los vocablos originales de la lengua castellana, Manuscript, Madrid, Bibl. Nac. 6.929 (18th-century copy). Sánchez, Tomás Antonio, ed. (1780), Colección de poesías castellanas anteriores al siglo XV, Vol. 2, Madrid, A. de Sancha. Scheler, August (1887), Supplement to 5th ed. of Diez, EWRS, Bonn, Marcus. Schuchardt, Hugo (1885), Review, Thurneysen, Keltoromanisches, Literaturblatt für germanische und romanische Philologie 6.110-14. Siesso y Bolea, José de (ca. 1720), [Diccionario español etymológico], Manuscript, Madrid, Bibl. Nac. 12.670. Spitzer, Leo (1921), "Lexikalisches aus dem Katalanischen und den übrigen iberoromanischen Sprachen", Bibl. dell' Archivum Romankum 2:1, Genève, Olschki. , (1922), "Zu Bruchs Bemerkungen Bibl. Arch. rom. II/, 26f.", Archivum Romanicum 6.494-504. , (1945), Review of Malkiel, "Spanish deleznar. .."Revista de Filología Hispánica 7.298. Thurneysen, Rudolf (1884), Keltoromanisches. Die keltischen Etymologieen im EWRS von Diez, Halle, Niemeyer. Togeby, Knud (1972), "L'apophonie des verbes espagnols et portugais en -ir", Ro mance Philology 26.256-64.

DISCUSSION The praise voiced by our discussants for Professor Malkiel's characteristic erudition and superb critical touch has my emphatic agreement. Regarding, however, the reserve voiced by some for the necessity of etymology to historical linguistics invites equally emphatic, if polite, disagreement. The Malkiel paper speaks for itself on the reciprocity of etymology and historical grammar. I would like to add that the two cannot be seen as separate entities, but only as complementary facets of a larger reality. All historical linguistic principles stand (or fall) on etymologies. Further, I would add, etymology has several crucial lessons for general synchronic theory. Firstly, successful etymologies establish that the currently fashionable dogma that underlying forms are stable is flatly in error, since these etymologies by definition have clarified what was unclear; successful etymologies reconstruct states of language that were lost and unknown. Secondly, unsuccessful etymologies, or cases where there is no known etymology, establish with equal conviction that

536

ETYMOLOGY

etymologically underlying forms have been lost, that synchronic rule-systems by no means 'mirror' diachronic processes. Even further, etymology shows that what is more or less stable through time is acoustic shapes, far more so than their underlying abstractions. What is transmitted and remembered is words, syntax far less. This casts strong doubt on the generativist notion of the sentence as prime, with words 'inserted' at a late point in the deriva tion of a sentence. Etymology, while not denying a proper place to syntax (of words), is one branch of linguistics that begins where the child begins, with words. Last of all, let us note that etymology has returned to Language (Number 51.3) with a bang, though it is not from the pen of a competent practitioner, such as Yakov Malkiel. J.P.Maher

To be blunt, what initially struck me in your paper, Professor Malkiel, was the simple affirmation that etymology could recover a productive relation to grammar. I suppose my reaction is only a further evidence of how lamentably wide the gulf between etymologists and grammarians has grown of late. The mere assertion that etymology has a future as a useful branch of linguistic research seems nowadays audacious. On reflecting, I recognized how advances in historical grammar could illumine and rectify etymologic solutions; but the inverse seemed far-fetched. Stated coldly: what did word-histories actually stand to contribute to our knowledge of grammars?1 What even are the main points of contiguity between etymology and the traditional edifice of grammar? Straightforward or transparent etymologic equations have, of course, formed the building blocks for historical phonology; but the relation of etymology to morphology seems less obvious. Therefore, I find a special virtue of your paper lies in pointing up the more significant areas of contact between etymology and morphology where the 'interlocking' of the two is recip rocal in less expected ways. I want first to address a few comments to the latter, specific portion of your paper and conclude by commenting briefly on one of your general observations. The much-vexed crux, desleír, is well-suited to exemplify the contact zones between morphology and etymology. Your analysis of its history is a model morphoetymological study for two reasons. On the one hand, you have drawn upon several known morphologic patterns, added by earlier etymologic research to our overall understanding of Castilian historical grammar, to account for some of the more troubling incongruities between desleír and its ancestor, DĒLĒRE. (E.g., the contin ued productivity of the semantically-kindred prefixes DË- and DÏ-/DIS- as the background for their interchange and thus for the folk reinterpretation and resegmentation of the initial syllable of *delir and, as a further consequence, for inter preting the second syllable as a root morpheme.) Thus morphology has clearly contributed to etymology; but, on the other hand, the test of reciprocity or inter locking is whether the reverse also obtains. In the case at hand, your specific etymo logic investigation appears to have acted as a catalyst for discovering what I consider a new broader insight into the potential "relation between verbal composition and

537


conjugation class", which in turn should be added to Hispano-Romance morphologic history. This is a principle you touch upon lightly. I would like to bear down on it more heavy-handedly. You observe that "in cases where the underlying Latin infinitive ended in -ĒRE, -ERE, or -ĪRE and where the resultant Romance infinitive displays a certain wavering (e.g., as between -er and -ir in Spanish, or as between -ere and -ire in Tuscan), it happens not infrequently that the simple verb ends in -er(e) and the compound verb in -ir(e), while the reverse distribution is virtually non-existent"— p. 527. Why should the derivative verbs have shifted from -er to -ir with greater alacrity than their simple bases? Even granting that -ar and -ir were the only productive or open conjugations and that the movement from -er to -ir was more frequent than the reverse, why should the prefixed variants have been more prone or more susceptible to innovation than their flanking root verbs? By way of premise, it must be recalled that certain processes of Romance derivation are plurimorphemic as against other, by far the more common, which involve a single affix. Verb-formation by prefix is of the former sort, and it operated in Old Spanish according to two set patterns: [prefix + root + -ar] or [prefix + root + -ir] and [prefix + root + -ecer].2 As more and more verbs were derived by these patterns, vestigial verbs of the -er group with an active prefix, e.g., OSp. abater, combater, co(n)fonder, escorrer, must have increasingly been sensed as anomalous. Against the background of close associations between the -er and the -ir conjugations throughout the paradigm and given their shifting alliances in the infinitive, the presence of an active prefix could have been a sufficient cue to trigger extension of the [prefix + root + -ir] pattern to prefixed verbs in -er, thereby integrating them into a more compact or unified structural pattern. E.g.,

COMPLĒRE → OSp. complir, cf. more vernacular (h)enchir ← IMPLĒRE CONTI(N)GERE → OSp. contir CONTUNDERE → OSp. contundir DESTRUERE → destruir, cf. con- instruir in Renaissance

DIVIDERE

→ dividir

DISCUTERE → discutir, cf. PERCUTERE → percudir SUCCUTERE → sacudir EXPRIMERE → es- ex-primir, cf. com- de- op- re- su-primir EXTINGUERE → extinguir

This particular pressure would not have come to bear upon the simple base verbs, and thus the schism observable between, say,


539

principle of least effort and for which one would therefore expect strong regularity, and yet which are nevertheless only sporadic in their yield.I have in mind, for exam ple, such normally broad-gauged phenomena as voicing of intervocalic consonants or generalization and unification of morphologic markers, which can be disrupted by the interplay of several social and stylistic registers within a single speech com munity. If we are anxious to open (historical) grammar to human reality — i.e., to acknowledge complexity and conflict in motives for innovation or for resistance to change, we must be willing to handle recurrences of all strengths, not merely those which have thus far qualified as grammatical rules. In short, modern linguists, sensi tized to sociolinguístic realities and to the plurality of rules, can clearly profit from the etymologist's perspective. You have shown here how an etymology which inter locks with a variety of morphologic patterns can be an excellent school for gram marians.

NOTES 1) The more so at a time when grammatical insights are valued in direct relation to their abstractness. After all, etymology seems to stand at the very opposite pole of a specificity ֊ or generality -scale from universal language principles. 2) This second pattern, enlarged by the inchoative interfix -sk-, need not concern us directly here, even though it was to usurp the role of simple -ir in deriving prefixed verbs outside the -ar class. 3) Such as occurred in our parallel (although independently conceived) treatments of the intertonic vowels in Hispano-Romance and Italian: "The Rise of the Nominal Augments in Romance. Graeco-Latin and Tuscan Clues to the Prehistory of Hispano-Romance", RPh 26. 306-334 (1972); and "Sedano, senero, prezzemolo, and the Intertonic Vowels in Tuscan", RPh 27.447461 (1974). 4) "Weak Phonetic Change, Spontaneous Sound Shift, Lexical Contamination", Lingua 11. 263-275 (1962), reprinted in Essays on Linguistic Themes, Berkeley-Los Angeles, 1968, 33-45. 5) "Diachronie Hypercharacterization in Romance", Archivum linguisticum 9.79-117 (1957) and 10.1-36 (1958).

Edward F.Tuttle University of California at Los Angeles

I would like to begin with Professor Malkieľs comments on the writing of ety mological dictionaries around the turn of the century. My regret in having looked through numerous etymological dictionaries for a variety of purposes is that many that have appeared never had a successor volume, if not by the original author, then by other students of the language, that would bring them up to date. I am thinking now of Meyer's etymological dictionary of Albanian and other such esoterica. Then I would like to turn very briefly to the point that was raised about the Wörter und Sachen approach and wonder whether in fact the approach is dead for linguists. Consider here, for example, von Wartburg's very large etymological

540

ETYMOLOGY

dictionary of French, where it seems the approach could almost be called Wörter und Sachen in terms of grammatical categories as well as substantive nouns and the like. Does this work not represent the possibility of integrating the older approach of studying lizzards and clouds with the newer work on grammatical particulars? Next I would wonder whether the kind of information that Professor Malkiel presents us with could not be expanded. Have there been any compilations of inflec tional and derivational paradigms for words like desleír? Consider the oft-cited set horror, horrible, horrid, and then terror, terrible, but no *terrid. Has any kind of comparative paradigmatic derivational study been done with the data at hand? Finally I would  to know what Professor Malkiel thinks about the feeling that the speakers of the time had about the grammatical meanings of some of these affixes, In other words, was there some sort of native-speaker awareness of the differences between de-, di-, and dis-? A. Richard Diebold, Jr. University of Arizona

Professor Malkiel has presented us with a brilliant exemplification of the idea expressed in the title of his paper, the interlocking of etymology and historical grammar. But I would like to propose, either for discussion or for future controversy, that the relationship between etymology and historical grammar is not entirely a relationship between equal partners. I believe that etymology needs historical grammar more than historical grammar needs etymology. In other words, it is possi ble to practice good historical grammar without deep involvement in etymology, but it is not possible to practice good etymology without a respectable knowledge of historical grammar. Next I would  to suggest a reason why the two branches of linguistic science are not currently practiced in conjunction, at least not since 1925, as Professor Malkiel has observed. I think էհե presentation itself constitutes an illustration of why this is not so. Sciences may be classified as cumulative or non-cumulative. Cumulative sciences are those where the next generation builds on the achievements of the previous one. Non-cumulative sciences would be those in which each genera tion has to make the discoveries for itself. There are several branches of linguistics that fall into these two types of science. I believe that historical linguistics, and etymology in particular, belong among the cumulative parts of linguistic science. Working in etymology, one has to master the vast scholarship of one's predecessors; and in order to stand on the shoulders of giants, one has to reach the level of those shoulders in the first place. And the reason why there are so few practitioners of etymology these days is, to me, the fact that there are very few linguists born after 1925 who can match Professor Malkiel in էհե depth of scholarship and erudition. Ilse Lehiste Ohio State University and Center for Advanced Study in the Behavioral Sciences, Stanford


541

Reply: I would like to reply selectively to these comments and questions. On the matter of the feeling of the native speakers concerning these affixes, we have no grammars of Medieval Spanish that would show us an awareness of these differences on a conscious, articulated level. But on a different, unarticulated level, the consciousness must have existed, as the case of desleír seems to show. On Professor Lehiste's comments, I believe that one can with very little etymo logical sophistication establish a skeleton historical grammar by using only the most transparent correspondences. The question is whether the traditional, transparent etymologies, usually kinship terms, are really characteristic, or whether they constitute a special, privileged domain, where interferences are particularly few. Admitting the merit of Professor Lehiste's argument, I would say: Almost equal partners, if not in fact equal.

NOTES

CONTACTS BETWEEN BLASPHĒMĀRE

AND

AESTIMĀRE

are agreed on the transmission, into Romance and into other "Western" languages, of an important Greek family centering around the verb BLASPHĒMEÎN 'to speak profanely', 'to speak ill, defame, blaspheme' and com prising also the abstract BLASPHĒMÍA 'profane language', 'evil-speaking, blasphemy' and the adj. BLASPHĒMOS (m.) 'speaking profanely', 'evil-speaking, slanderous'. It is usually argued that the channels of transmission were ecclesi astic, i.e., Church Greek and Church Latin; this assumption is certainly, by and large, correct so far as the diffusion of the learnèd forms is concerned: Sp. blasfem-ar, -ia, -o, and -ador; It. blasfem-r, -o, etc. Even here, however, caution is advisable, because the borrowing of a "cultismo" from one vernacular into the other remains an ever-present alternative, not easily detectable. Important clues in the reconstruction of such delicate threads are: accurate datings, the range of meanings and derivatives, the occurrence of the words at issue in widely-read or frequently-recited texts. Thus, E. to blasphème (14th c ) , blas phemy (13th ), blasphemous (15th c.) certainly sound like so many echoes of Old French usage,1 especially if one recalls that on the Continent blasphème (1190, Saint Bernard), blasphémer (1360, J. Le Fèvre), blasphémateur (1390, Ph. de Maizières), and blasphématoire (1516, P. Desrey) made their appearance at a distinctly earlier date. 2 One can thus speak of medieval France as a second ary focus. This impression of a delegated rôle of diffusion gains in strength once the analyst's attention turns to vernacular conduits, recognizable by the interplay of various sound changes. Thus, the principal representative of Gr.-Lat. BLAS PHĒMĀRE in Old French was blasmer, corresponding to OProv. blasmar. This verb not only underlies mod. blâmer, but has rayed out beyond the borders of France in various significant directions, frequently to the tune of semantic shifts and stylistic adjustments. It. biasim-o and ֊are are at present clearly recognized as Gallicisms; so are, in Spanish, obs. blasm-ar and blasmo, which, ALL AUTHORITIES

1

See .. Onions et al., The Oxford Dictionary of English Etymology (Oxford, 1966), p. 98b. A. Dauzat, J. Dubois, & H. Mitterand, Nouveau dictionnaire étymologique et historique (P., 1964), p. 91b. 2

544

ETYMOLOGY

if any additional proof were needed, opportunely emerge in romances of chivalry inspired by trans-Pyrenean models;3 E. blame 'to find fault with' is traceable to the 12th c , i.e., it preceded blaspheme (with which it failed to coincide semantically) by a conspicuously wide margin; while G. blam-ieren, orig. spelled -iren, is discernibly a latecomer, infiltrating as it did shortly after 1600 with the strong sense of 'vituperating' ('schmähen, beschimpfen') and becoming the equivalent of 'exposing to ridicule', refl. 'to make a poor showing, cut a ridiculous figure' only in the 19th c.4 Changes of focal points from Rome to Northern France were not yet clearly grasped by the pioneers. Diez, e.g., in 1853 placed It. biasim-o and ֊are, OSp. blasmo, Prov. blasme, Fr. blâme, all of them revolving around the concept of 'blame', genetically on the same level (EWRS1, pp. 53f.); he was not at all concerned — at least not on that occasion — with the strictly learnèd branch (blasfemar, etc.) and evinced no curiosity about older German, still less about medieval English, usage. Also, the founding father paid scant attention to the relation of nouns to verbs within the branch on which he focused attention, connecting biasimo, etc. with BLÁSPHĒMON (n.) and biasimare, etc. with BLASPHĒMEÎN. However, Diez was sufficiently alert to recognize a whole string of formations, stretching from Portugal to Rumania, in which ancestral -SPHwas replaced by -st-: It. biastemma, bestemmia; West. R.-Rom. ("churwälsch") blastemma՝, Prov. blastenh; OFr. blastenge; Rum. blëstëm, all of which he linked with BLASPHĒMÍA, implying that the corresponding verbs (e.g., It. biastemmare 'to blaspheme, curse') were offshoots of these nouns. Diez invoked a "selt[e]ne Vertretung des F [i.e., more accurately, -PH-] durch ť', a phrasing which on a later occasion (EWRS3, I [1869], 65) he replaced by "unorganische Vertretung", without further elaboration, burdening subsequent generations of scholars with the task of justifying the shift -SPH- > ֊st-. At the very end of his entry, 3 Blasmar: Cuento del Emperador Ottas, ed. J. Amador de los Ríos (Historia critica, V [1864]), Ch. xxiii: "Ca nunca ende seredes blasmada"; levantar blasmo 'to raise an accusation': Carlos Maynes, Ch. xii. I find blasmefar in a much later prose text, traceable via a lost Old Portuguese version to Middle English (Gower), namely Confisión del amante, ed. Knust, fol. 52r°. Brasfemar, -fernia, interestingly, occur in medieval Portuguese texts; A. de Morais Silva, Grande Dicionário ... 10 , II (1950), 607b, cites a passage from Inéditos de Alcobaça, III, 76 ("se por ventuira ouvira Deus a brasfemia daquelle principe"), while brasfemar in A. F. de Castilho's Camões (I, 108) is surely a deliberate archaism. Covarrubias altogether omitted blasmar and its offshoots. The Academicians, in 1726, intelli gently argued that obsol. blasmar could be either syncopated blasphemar or borrowed Fr. blasmer, and that obsol. blasmo might very well reflect Fr. blasme, documenting the verb with the [Tercera] crónica general and with Mena's Coplas, and the noun with J. Martínez de la Puente's epitome of the Cronica de Juan Segundo. The latest Academy Dictionary (1970), less liberal, is tight-lipped about the probability of the word's Old French extraction. The Diccionario histórico, II (M., 1936), 249b cites blasmar from Santillana (two passages) and blasmo from the same poet as well as from Cartagena and the Mem. de Fernando IV. 4 One encounters all the important dates and sources in W. Mitzka's revision of F. Kluge, EWDS18 (Berlin, 1960), p. 81a. I t was, of all experts, Kluge who, in his 1895 monograph on Ger man student slang (p. 64), reconstructed the series of facetious formations ("Ulkwörter") in -age (f.) which, having started out from Renommage, includes, as its principal support, Blamage, orig. 'vituperation', at present 'ridiculous, embarrassing situation; trap into which one falls through one's own fault'.

CONTACTS BETWEEN BLASPHEMARE AND AESTIMARE

545

he added to the record, as if on second thought, Sp.-Ptg. lástima 'Schimpfwort, Wehklage', alongside lastimar 'mißhandeln, beleidigen, zum Mitleid bewegen'. ('Beleidigen' in the modern sense = 'to offend' is an inaccurate rendition of 19th-c, as against Golden Age, lastimar, but the semantic configuration of the German family appealed to: Leid, Leids, (er)leiden, beleidigen, leider, leidtun, etc. unquestionably bears a certain resemblance to the contour of lastimar/ lástima). In the following years and decades Diez's basic idea that one should discri minate between several layers of transmission of BLASPHĒMĀRE gained ground; small wonder that the students of doublets and triplets (doppioni, Scheide wörter), particularly those concerned with cases involving semantic cleavage, included situations like Sp. lastimar ~ blasfemar ~ blasmar in their respective inventories.5 Certain facets of Diez's schema, however, became subject either to thorough elaboration or to radical revision. For one thing, the idea of two perfectly parallel strings, one nominal (attached to the adj. BLASPHĒMON), the other verbal (tied to Lat. BLASPHĒMĀRE), ceased to appear convincing. The discovery of countless instances of denominative verbs and postverbal nouns suggested the wisdom of subordinating one string to the other; also, the attachment of the Romance verbs to BLASPHĒMĀRE (and, through its instrumentality, to even older BLASPHĒMEIN) was particularly smooth. As a result, scholars clearly — and rightly — leaned toward declaring OFr. OProv. blasme and their cognates and satellites mere offshoots of blasm-er, -ar, etc., with no direct prototypes in the parent languages.6 This mode of thinking permeates, e.g., Meyer-Lübke's REW1. For another thing, the Latinization of the noun BLASPHÈMÍA began to loom as no less worthy of attention than the adjustment undergone by the correspond ing verb. The speakers had the choice between retaining the Hellenic stress pattern, especially if the primitive was also borrowed ("dual borrowing"), or 5 F. A. Coelho, "Formes divergentes de mots portugais", Rom., II (1873), 281-294, was not yet alert to this state of affairs. But  Michaelis [de Vasconcelos], Studien zur romanischen Wort schöpfung (L., 1876), ρ. 291α, pitted lastimar against blasfemar and blasmar. U. A. Canello, "Gli allotropi italiani", AGI, I I I (1878), 285-419, at 363f., used a semantic criterion in contrasting blasfemare, biastemare 'offendere a parole cose  idee sacre' with bias(i)mare, orig. blasmare 'rimproverare, disapprovare', orig. 'dir male'; compared the intercalation of -і- in biasimare to the treatment of χρίσμα > cresima; and posited, at the point of bifurcation of blasfemare and biaste mare [I take it that biasimare is a misprint here], the "intermediate form" [*] biasťmare. Note that 20th-c. Italian obligatorily geminates the m of bestemmiare, though not, of course, the / ŋ/ of the var. bestegnare. 6 This trend of thought appears in full view in the Dictionnaire général (ca. 1890) and should thus be credited to A. Darmesteter and A. Thomas. G. Lené's demurrer in Les substantifs postverbaux dans la langue française (diss. Uppsala, 1899), p. 27, s.v. blasme, is unconvincing: "Mais comme blasphemus existe en latin et qu'en français le substantif et le verbe apparaissent simul tanément, cette hypothèse est parfaitement inutile"; the author is oblivious of one crucial circum stance: Gr.-Lat. BLASPHĒMU did not function as a verbal abstract. Surprisingly, G. Paris and A. Tobler failed to chide Lené for this blunder in their respective balanced reviews, Rom., X X I X (1900), 440-445 and ASNS, CV (1900), 203-206; or else all three writers were too strongly influenced by §17 in Meyer-Lübke's GRS, I: Lautlehre (Լ., 1890). G. Gröber offered no comment in his "Substrate..." (1884-89).

546

ETYMOLOGY

changing -IA to -IA (under pressure by such preëxistent models as ASTÚTIA, INDUSTRIA, INERTIA, MODESTIA, MOLESTIA, SOLLERTIA) — the final, optional step then being the coinage of a corresponding -IUM (on the model of EXORDIUM, 7 INITIUM, PRINCÍPIUM, by way of "depluralization"). There also emerged the prospect of a split between -ÍA and -IA/-IUM, the former reserved for the un adulterated "cultismo", the latter hugging the ground of less inhibited folk speech. Even speakers of Latin endowed with a fine ear for the prosodic niceties of Greek could find some solace or justification for BLASPHĒMIUM in the authen tic Hellenism ENCOMIUM < ENKOMION 'hmn to a victor, formal expression of high praise, eulogy, panegyric', almost its semantic opposite. Eventually, the pressure of BLASPHĒMIUM produced in medieval Latin the coinage of another polar opposite, well-attested LAUDĒMIUM, which directly underlies Fr. louange 'praise' and, in fact, gave rise to a whole series of OFr. -ange abstracts, one member of which, namely, melange, orig. mesl- 'mixture' (A. Chartier, A.D. 1420; from mesler MISCULĀRE), happens to have lingered on to this day. 8 The protracted coexistence of the types (a) *blasmu, or an older stage in -ĕmu, and (b) ecclesiastically-inspired blasphemiu presumably affected the cor responding verb, at least in folk speech. It was easy for a by-form of the verb, in -MIARE, to spring into existence, so as to flank older -MARE, especially in view of the semantic proximity of CALUMNIĀRE 'to slander' (extracted from CALUMNIA), a verb exceedingly well preserved in medieval Romance and beyond (cf. E. challenge, via OFr. chal-ongier, -engier). It is this -IUM/-IĀRE branch that Diez was apparently bent on linking directly to BLASPHĒMIA, in view of It. -emmia, OProv. -enh, OFr. -enge (cf. SOMNIU 'dream' > songe and OFr. mençoigne ~ mençonge), etc. One point of which Diez was perfectly aware as early as 1853 and which has never since stopped plaguing scholars was the appearance of an unorganic medial t. The EWRS, I repeat, records such nominal forms as It. biastemma, bestemmia, Gris. blastemma, OProv. blastenh, OFr. blastenge, Rum. blëstëm; the corresponding verbs biastemmare, etc.; plus, separately and perhaps by way of afterthought, the Hispanic couple lástima ¡lastimar. In writings by a more mature Diez and by the early generations of his successors, the form of some of these presumed reflexes was modified (thus, the EWRS3 listed the Rumanian 7 W. Meyer[-Lübke], overriding Diez's verdict, did recognize in BLASPHĒMIUM the base of OProv. blastenh as early as his doctoral dissertation; see Die Schicksale des lateinischen Neutrums im Romanischen (Halle, 1883), p. 154. In the two editions of his dictionary he went farther, crediting OFr. blastenge alongside OProv. blastenh to BLASPHĒMIUM and It. bestemmia, flanked by a verb in -iare, to BLASPHĒMIA. On the rivalry of /nž/ and / ŋ / in Gallo-Romance, as in OFr. mençoigne ~ mençonge (later mensonge) 'lie', see BPh, VI, 121-172. 8 In §127 of his HGFS, II (Heidelberg, 1921), Meyer-Lübke recognized that Fr. vend-ange VINDĒMIA 'vintage' represented a blind alley, whereas blast-enge BLASPHĒMIA, despite its formal isolation, could, on the strength of its semantic load, have given rise to OFr. lou-ange (cf. Med. Lat. LAUDEMIA, OProv. lauzemia). He viewed, with G. Baist, los-enge 'flattery' as Germanic and so classified OFr. ha-enge, laid-ange, allowing for the influence of challenge on laidenge, but despairing of a cogent explanation for vid-ange, OFr. cost-ange, and mél-ange (plus the latter's satellite mollange). For criticism see Corominas, DCE, I I I [1956], 109b.


547

form as blestem, while S. Puşcariu's turn-of-the-century dictionary9 favored the spelling blastǎm as the entry, recorded the Daco-Rumanian vars. blestem, hlastam, and threw in for good measure Arom. and Megl. blâstim) ; moreover, the number of vernacular representatives displaying the erratic է increased by leaps and bounds, and the first attempts were made at causal explanation. In diagnosing the difficulty, the imagination of tone-setting scholars was perhaps too strongly haunted by the t; it might have been more advisable for them to think in terms of medial consonant clusters rather than of an isolated sound and to argue that certain varieties of Late Latin folk speech need not have, phonotactically, tolerated /sf/ which, at that time, should have been the normal rendition of Gr. -SPH-, a situation which could have left the door open for all sorts of surprises and emergency solutions. In principle, when etymologists are faced with a situation like Gr. -SPH- > Vulg. Lat. ֊ST-, the following options seem to be available to them: (a) They may argue that a blend with an authentically Latin word in -TIMARE could have occurred. The candidacy of AESTIMĀRE 'to appraise, value, esteem, hold' was advanced by E. G. Parodi; 1 0 Puşcariu, in 1905, took cognizance of this conjecture and W. Meyer-Lübke, a few years later, endorsed it unequivocally in the original edition of his comparative dictionary — without bothering to give credit to the proponent. (b) They may search, within the treasure trove of Greek dialects, for one or a few t h a t exhi bited a representative of the Class. BLASPHĒMEIN family with   "regularly" substituted for the troublesome -PH-. (c) They may hypothesize that BLASPHĒMEIN was absorbed early enough into Latin for the Gr. -PH- (at that stage still an aspirated  rather than a conventional graphy for /f/) to have undergone, through deaspiration, the expected change to  (as in PORPHYRA > PURPURA 'purple, deep-red [garment])', with the further probability of *BLASPÈM- being before long transmuted into *BLASTÈM- through dissimilation of bilabial occlusives in consecutive syllables. (d) A radically different line of thinking might have occurred to them: Syncopated BLASPHĒMĀRE, i.e., *blasmare, could easily, at least in certain dialects, have acquired the by-form *BLASTMARE, the way ASSULA 'splinter, chip, shaving' cast off a variant *astla (witness Sp. astilla) and the way PESSULUM 'bolt' (of a door) developed in Spanish into pestillo 'door latch' (dial. pecho, pesliu), via *pestlu. 11 A temporary wave hostile to syncopation — a kind of backlash — could then very plausibly have given rise to blastemare, just as it produced *astula, *pestulu — which in turn cast off -ella, -ellu vars. Interestingly, while OFr. blasmer allows of no direct inferences as to whether a է or an ƒ disappeared after the s and before the m, there is epigraphic evidence of blastemare on Gaulish soil. 12

Some of the spokesmen for the conjecture BLASPHĒMĀRE  AESTIMĀRE have already been identified. A variation was proposed by M. Grammont — in, of all places, his review of the 2d fascicle of the REW 1 —: only *BLASPĒMĀRE could 9 Etymologisches Wörterbuch der romanischen Sprachen, I: Lateinisches Element (Heidelberg, 1905), §205. 10 "Etimologie", Miscellanea nuziale Rossi-Teiss (Bergamo, 1897), pp. 335-354, at 340. 11 See my paper, "Studies in Hispano-Latin Homonymics: PESSULUS, PACTUS, PECTUS, DĒSPECTUS, SUSPECTUS, FISTULA in Ibero-Romance", Lang., XXVIII (1952), 299-338. 12 Cf. . Le Blant, Nouveau recueil d'inscriptions chrétiennes de la Gaule (P., 1892), p. 83. Gamillscheg, EWFS, s.v., cites J. Pirson, La langue des inscriptions latines de la Gaule (Bruxelles, 1901), p. 231.

548

ETYMOLOGY

have given rise to *BLASTIMÄRE, through dissimilation of the second out of three consecutive bilabials (b-p-m),13 an interesting pronouncement, coming as it does from an expert in dissimilatory processes, but one that still saddles us with the need to account for -PĒMĀRE in the first place.14 P. Fouché's idea of scuttling BLASPHĒMĀRE altogether and opting instead for a blend of LASSUS 'tired, weary, exhausted' and AESTIMĀRE was unfortunate — even though LASSUS is yet to cross our path. 15 In revising his dictionary in the early 'thirties, Meyer-Lubke tacitly dropped the reference to association with AESTIMĀRE, and Corominas, a quarter century later, visited the full measure of his scorn on this hypothesis.16 We shall revert to this issue and, indeed, attempt to salvage part of the wreck age. Meanwhile, let us remark that, in principle, a verb that means 'to blas pheme, curse, offend', etc. is, by virtue of that meaning, by no means unlikely to have been locally contaminated by other racy verbs and all sorts of "Kraft ausdrücke". To cite just one example: The coexistence, in colloquial Latin, of BLAT(T)ERARE, a phonosymbolically suggestive verb often recorded in glosses ('stulte loqui, stulte e-, ob-loqui, delirare, stupide et sine causa loqui') 17 could certainly have been a factor in tipping the scales in favor of BLASPHĒMĀRE and BLASPHĒMIA, to the detriment of native VITUPERĀRE and DĒTRACTIO (see CGL, VI, 145a). The projection of the change onto the plane of Greek often coincides with the advocacy of the dissimilatory explanation. The chief authority along this line was P. Kretschmer who, at the turn of the century, traced βλαστημβω to modern 13 See RLaR, LV (1912), 109. This pronouncement did not elude the attention of W. von Wartburg, FEW, fasc. 6 (1925) = Ι, 403ab, who hesitantly endorsed it, but whose own treatment of the entire problem was woefully inadequate. Paradoxically, the author was at his best in sifting older Italian dialectal data; he credited to Gallo-Romance provenience Lomb. braxemar and OGen. biassmar, analyzing as indi genous to Italy OPiedm. biastéma 'curse', Ven. biastemar, OLucc. biastimare, Sien. biastemmare. Far less satisfactory was his idea of declaring OProv. biastemar an Italianism, and OFr. blastemer a Provençalism; or of charging OProv. blastim, blastimia, in all likelihood, to Italian infiltration. Is it not simpler to dispense with so many borrowings and, instead, to reckon with a single semilearned transmission of *BLASTIMĀRE all over these territories? Among the older etymological dictionaries, A. Scheler's (1862, 18732) adopted Diez's view, while A. Brachet's and L. Clédaťs swept under the rug the main difficulty: the erratic appearance of 4-. E. Lommatzsch, a year or so before Wartburg, did an excellent job, in AFW, fasc. 7 (1924), of defining and documenting a number of pertinent formations ( = I, cols. 991-993, 995): blasfeme (., f.) 'blasphemy', blasfemëor, blasfemer (tr., intr.); blasm-able (= -ant), -e, -ement, -er {-ir is a hapax legomenon); blasteng-e, -eor, -ier1 (noun), -ier2 (verb). His fine-meshed syntactic and seman tic analysis (of major relevancy to the study of E. blame, too) discloses (a) that the blastenge branch overlapped with both blasfemer (blastengëor 'Gotteslästerer') and blasmer (blastenge 'Tadel, Schimpf, Schelte'); and (b) that blastengier involved both verbal abuse and physical harm, a point the biographer of Sp. lastimar should heed. 14 The renewed interest in consonant dissimilation has, of late, once more tended to place BLASPHĒMĀRE in this particular context. See, e.g., R. Posner, Consonantal Dissimilation in the Romance Languages (Publ. of the Philol. Soc, XIX; Oxford, 1961), pp. 55, 67, 75, in reference to Sp. lastimar and Sic. gastimari. The author invokes a probable influence of AESTIMĀRE. 15 "Études de philologie hispanique", RH, LXXVII (1929), 144f., in a section dealing with Menéndez Pidaľs Manual de gramática5. The word index to the definitive text of the Manual, of the year 1941, makes no mention either of lástima or of lastimar. 16 Meyer-Liibke, as we shall see, fell under the spell of Kretschmer's thinking; for many years the two scholars taught jointly at the University of Vienna. 17 See Corpus Glossariorum Latinorum, eds. G. Loewe & G. Goetz, VI (L., 1899), 145b.

CONTACTS BETWEEN BLASPHĒMARE AND AESTIMARE

549

Greek, especially the dialect of Lesbos, and, moreover, explicitly argued for consonant dissimilation.18 Other vestiges of relevant modern Greek usage were found — perhaps not coincidentally — in settlements outside Greece, e.g., in Calabria.19 For, counter to Meyer-Lübke, who, in his REW3 1155, drew the rash conclusion: "So wird es sich um griech[ische] Dissimilation handeln", at least some younger scholars have moved in the opposite direction; Coro minas, e.g., deems it more likely that modern Greek, on this score, reflects Vulgar Latin or Romance usage (DCE, III [1956], 406-42a, at 41b) — an opinion whose cogency it seems unnecessary to dispute.20 As regards the lexico-cultural osmosis between Greece and Rome, we have, for two millennia and a half, been so strongly conditioned to expect Hellenisms in Latin rather than Latinisms and Romanisms in Greek that it is hardly surprising to see a Meyer-Lübke, forty-odd years ago, trying to push even a minor event, such as the presumed consonant dissimilation, back into the Greek phase of the development. By the same token, after the research con ducted by Hasselrot, the Kahanes, and other younger experts, the picture that has begun to unfold is one of two-way traffic, a shift in direction which accounts for Corominas' inclination to reverse Meyer-Lübke's hierarchy, in regard to Calabria and to Lesbos. The single most hazardous supposition in Corominas' web of conjectures is his almost casual acceptance of [*]BLASPEMARE, with deaspiration of  [ph]. The argument would be unobjectionable had BLASPHĒMEIN been absorbed into Republican Latinity, 21 especially into archaic folk speech of which the Plautine plays have captured not a few priceless samples. However, a stratigrapher, stylistician, and connoisseur of A. Ernout's rank, as late as 1959, was explicit in crediting the loan to Church Latin — specifically to translations of the Old and the New Testament (DÉLL 4 , I, 72a).22 Are we, then, to assume that BLASPHĒMEIN infiltrated into Latin twice, oozing for the first time into folk speech, without leaving any trace in contemporary texts, early enough for the , still pronounced [ph], to have yielded  through deaspiration (hence *BLASPĒMARE, 18 Der heutige lesbische Dialekt, verglichen mit den übrigen nordgriechischen Mundarten (Academy Memoir; Wien, 1905), col. 176. 19 G. Rohlfs, Etymologisches Wtb. der unteritalienischen Gräzität (Halle; 1930), §336: vlastemmao, flastimoo. 20 In their lavishly detailed review article ("Greek in Southern Italy") of the expanded version, titled Lexicon Graecanicum Italiae Inferioris (Tübingen, 1964), of Rohlfs's EWUG, H. and R. Kahane have paid no attention to BLASPHĒMĀRE; but their elaboration on the Romance vicissi tudes of PHANTASMA and PHANTASÍA are directly relevant to the problem at hand. See RPh, XX, 401-438, at 435. 21 The classic discussion of this issue will be found in Meyer-Lübke's Romanische Lautlehre, §17, apropos PURPURA 'purple', APUA ( = άφνη 'sort of anchovy or sardine'), COLPUS 'blow', Apul. posperu (in lieu oí fosperiu) from φώσφορος. See also H. and R. Kahane's aforecited article. The case of COLPUS was later reexamined by Ernout. 22 Somewhat elusively, not to say cryptically, the authors of D É L L 4 remark: "... représentés dans les langues romanes dont les formes supposent blastimāre vec dissimilation de p{h), peutêtre sous l'influence de aestimare". Their acknowledged mentors, on the Romance side, were, in addition to Meyer-Lübke (whose wavering, not to say volteface, will be remembered), 0. Bloch and W. von Wartburg, DELF, a dictionary which, in its rev. 2d ed. (P., 1950), p. 71b, gave its readers the choice between two conjectures: consonant dissimilation or blend with AESTIMĀRE.

550

ETYMOLOGY

dissimilated to blastemare); then, for the second time, during the 3d and 4th centuries, i.e., at the height of Christianization, with the ph rendered by ƒ on the level of learnèd and semilearned transmission ? And are we, further, to conclude that the religious use ('to blaspheme, curse') did not in any way affect the early lay or pagan use ('to cuss, speak harshly, blame')? Some such unusual conca tenation of events — including a few that are purely speculative — would have to be postulated to do justice to the extraordinary spread and variety of -tforms (introducing quite a few not yet mentioned: Cat. blastomar (flastornar), Logud. frastimare, Campid. blastomai, brestomai, Friul. blestemá as against Engad. blastmer, etc. 23), unless one indeed espouses the hypothesis of a buffer consonant, of the type familiar from the biographies of ASSULA and PESSULUM; also, mutatis mutandis, of Ī(N)SULA 'island' > It. Ischia, etc. It will be remembered that, despite Meyer-Lübke's dramatic volteface and Corominas' violent protestations notwithstanding, not all etymologists hastened to wean away BLASPHĒMĀRE from AESTIMĀRE.24 The central thesis of the present note is to distinguish between (a) the mere POSSIBILITY of such an association at the Vulgar Latin stage from (b) its high PROBABILITY at the crest of the Old Spanish development. Three independent criteria militate in favor of the revision implied by the alternative: (1) As regards the range of its lexical meanings and, coincidentally, of its syntactic constructions, Sp. lastimar stands apart from the aggregate of its congeners (even Diez was aware of this discrepancy). Specifically, lastimar, in Golden Age Spanish, often, not to say predominantly, meant 'to offend' (or 'to hurt', physically) rather than 'to blame' or 'to curse, cuss'. 25 Now, 'offense' and 23 On some of these forms there exist special investigations, by C. Nigra ("Postille lessicali sarde", AGI, XV [1901], 481-493, at 487), . Salvioni, and others. The fl- variants are particularly striking and might, in certain places, be interpreted as substitutes for Byz.-Gr. vl-, a cluster normally alien to Romance, though perfectly acceptable to Slavic. But Nigra likens South. frastimái, Centr. frastimáre to the trajectory of frandigár < *BLANDICĀRE 'to soften, flatter, praise' and follows Körting in positing *blastemare. 24 Interestingly, while B. Migliorini and A. Duro, Prontuario etimologico4, s.vv. bestemmiare and biasimare, do not go beyond invoking the ancestry of Gr.-Lat. BLASPHĒMĀRE, G. Devoto, Avvia mento alla etimologia italiana: dizionario etimologico (Firenze, 1966), pp. 46b and 47a, traces bestemmiare to an amalgam of BLASPHĒMĀRE, BESTIA, and AESTIMĀRE, and biasimare, via OProv. blasmar, to BLAS(PHĒ)MĀRE. In fasc. 2 (1926) of his EWFS1, which of course appeared before the REW, . Gamillscheg posited — expressly on the authority of his teacher — a blend of BLAS PHĒMĀRE and AESTIMĀRE (p. 1126); forty years later, in fasc. 2 (1966) of his EWFS2, the author repeated the earlier statement (p. 117b), apparently oblivious or neglectful of Meyer-Lübke's reorientation, but added helpful data on the use of Salvianus (5th c.) and Fredegar (7th c ) , in part on the basis of M. Bambeck's new research. 25 The DCE, s.v., offers a particularly rich harvest of illustrations, from 17th- and 18th-c. sources: Aldrete, Cervantes, Covarrubias, Oudin, Góngora, Ruiz de Alarcón, Cienfuegos, Academy Dictionary; but its author is unable to explain (yet is not candid enough to admit this inability) why, between 1330 (earliest documentation: Juan Ruiz) and 1600 (peak of the Golden Age) this "etymological meaning" apparently cannot be documented. Here is how Corominas visualizes the semantic trajectory: "...en castellano y portugués se pasó de 'difamar' a 'ultrajar, agraviar' y de ahí por una especie de eufemismo 'herir físicamente'  por otra parte 'causar lástima'" (III, 40b). This last link in the author's reconstruction of the semantic chain is, paradoxically, the one represented in Juan Ruiz (1052b). In the Supplement (IV, 1037a), Corominas admits his uncertainty about Juan de Mena's use of lastimero: 'cruel' or 'pitiful' ?

CONTACTS BETWEEN BLASPHĒMĀRE AND AESTIMĀRE

551

'(token, expression of) esteem' do come very close to functioning as polarized concepts. If so, Corominas' main argument against the estimar conjecture, namely the alleged lack of such sufficiently neat contrast as might have justi fied mutual attraction and eventual interference, may conceivably hold water in Church Latin, but collapses under its own weight with respect to Peninsular use. (2) As regards chronology, observe that lastimar was absent from the abun dant and well-explored corpus of 13th-c. texts, in prose and verse. It appears for the first time, according to the DCE, in Juan Ruiz (ca. 1330-50), i.e., in a poet of indisputably clerical background and esthetic canons. Estimar, in this severely learnèd garb, made its appearance, on the authority of the same wellinformed source, in the late-medieval Latin-Spanish glossaries which have become available to us through A. Castro's edition and whose language, two years later, was subjected to L. Spitzer's close scrutiny. 26 Regardless of what we choose to think of the vernacular transmission of AEST-ĬMĀRE, UMĀRE (and by this I mean not only the vicissitudes of OSp. as-, os-mar, Ptg. esmar, but, probably more important, the gradually emerging possibility of linking tomar 'to take' with resegmented AES-TUMĀRE27), the gap between the earliest dates Interestingly, among the pioneer etymologists, the prevailing inclination was to start from the meaning 'to wound'; early suggestions of bases include Gr. TYMMA 'wound' and Lat. LAEDERE (S. de Covarrubias), while Alonso Sánchez de la Ballesta toyed with a link to STIMULĀRE in the sense of 'pricking'. Fray Martin Sarmiento, well-informed of these gropings, cautiously dubbed Gal. lastimar and lastimeiro words "de origen difícil" and, with his usual flair, correctly associated Sp. lisiar 'to wound' with LAED-/LAES-; see his Colección de voces y frases gallegas, ed. J. L. Pensado (Salamanca, 1970), p. 450. 26 See the Introduction to Castro's long-delayed and, in many ways, ill-fated edition of Glosarios latino-españoles de la Edad Media (M., 1936) and Spitzer's substantial review article in MLN, LIII (1938), 122-146. 27 The dominant form in Old Spanish was asmar 'to think (of), opine, guess, count [ = estimate the number of], suppose', found in innumerable medieval texts {Grail Fragments, ed. Pietsch, 253v°, 273r°, 285v°; Santa Catalina, 19v°; Barlán e Josaphá, 111v0, 113r°, 143r°, 144r°, 172r°; Ruiz, 196c, 806a (confused, through a scribal slip, with asomar); Confisión del amante, 213°; reconstructed, through a persuasive emendation, in v. 16 of Elena y María, by Menéndez Pidal: [as]mar; and traced by most etymologists to *AD-AESTIMĀRE (on the model of AD-MĪRĀRĪ?), transmitted through Provençal-Catalan (REW3 139, 246). If this is so, can we brush off as merely fortuitous the fact that blasmar 'to vituperate' and asmar 'to think (highly)' — the respective products of BLASPHĒMĀRE, -TĒMĀRE and (AD)AESTIMĀRE — should have both been borrowed by Old Spanish from trans-Pyrenean sources ? And is it further sheer coincidence that in medieval Gallo-Romance aesmer, esm-er/-ar should have prevailed, whereas in Old Spanish asmar was in disputably the dominant form ? Asmar clearly echoes blasmar and provides clinching proof, if such additional proof was at all needed, of the inextricable tangle of the Romance products of BLASPHĒMĀRE (here cast in an active rôle) and AESTIMĀRE (here, for once, reduced to a passive function). Asmar may have slightly outlived blasmar and seems, in its period of decline, to have adopted the sense of 'musing'; Covarrubias distinguishes between the two semantic hues, obs. 'pensar' and prospering 'quedarse un hombre suspenso y pensativo, traspuesto en la consideración de alguna cosa, y que por egual no respira...' (Marta Fisher suggests influence oí pasmar 'to astound, stun'); the Academy, in 1726, rules extinct (a) asmamiento (Fuero Real) and 'presumir, juzgar' (Coplas de Mingo Revulgo) and 'osar, intentar, emprender con esfuerzo' {El Conde Lucanor and Argote de Molina's glossary thereto), while sparing () asmadura 'discreción...' (J. López de Palacios Rubios, Esfuerzo bíblico). The purposeful attitude — not the spell of aimless meditation — is obviously closer to the semantic ambit of OFr. aesm-able, -ance, -e (m.; the opposite of blasme), -ement, also esm-e, -, -er, presented authoritatively in AFW, I, 166, and III, 1113-17; hence the plausibility of the connection with E. aim and of the conflation with Fr. aimer disclosed by J. Gilliéron and later reexamined by J. Orr. Note the brief lifespan of OLeon. aesmar {Fuero Juzgo; cf. DCE, II, 432α) and the use of esmo, in Portuguese, for 'estimate' rather than 'esteem'

552

ETYMOLOGY

for lastimar and estimar has dwindled to decades rather than centuries, a state of business which makes it highly plausible that the two verbs were absorbed at roughly the same time, toward the middle of the 14th century, and thus could indeed have influenced, and been dependent on, each other. (3) Finally, with respect to phonology, Corominas' tortuous arguments again fall short of convincing. The emergence of і instead of e < Ї in the intertonic syllable is characteristic, above all, of semilearned words, being ordinarily preceded by a long-drawn-out struggle between the two front vowels; thus LACRYMA (~ -UMA) 'tear' — a word abounding in religious implications, not unlike DULCÍS28 — yielded lágrema alongside lágrima; eventually the former gave way to the latter. 29 What is, conversely, remarkable about lastimar is the (Morais Silva, Dic, IV, 717b: Mendes Pinto; also esmar in the 19th-c. classic A queda de um anjo by Camilo [Castelo Branco]). OPtg. osmar, with a rare equivalent in Old Spanish, was shrewdly attributed by Schuchardt to a blend with husmar (cf. mod. husmear 'to scent, smell out'), an opinion rightly endorsed by Meyer-Lübke, except that the -m- may have imparted to the entire -sm- cluster a weakly labializ ing force which, allying itself with lexical contamination, could have produced the effect. Good examples of osmar are offered in M. Rodrigues Lapa's Vocabulário galego-português extraído ... das "Cantigas ďescarnho e de mal dizer" (ո. pl., 1970), p. 10a. On AES-TUMĀRE as a possible source of Sp. tomar see the appended Excursus. 28 The lexical stratum to which these words, and others genetically similar to them, happen to belong has never, to my knowledge, been subjected to stringent grammatical analysis. One is tempted to attribute to it certain fossilized nominatives, bequeathed by imparisyllabic formations in -TIÕ/-TIÕNE, e.g., OSp. bollicio (mod. bullicio 'bustle, rumble, wrangle', rebullicio 'loud uproar') and cansacio (mod. cansancio) = OPtg. cansaço 'fatigue, exhaustion', the former from BULLĪTIO 'seething', 'ebullition', the latter from QUASSĀTIÕ '(repeated) shaking, battering, shattering, smashing to pieces', i.e., 'break-down' (from QUASSĀRE, the intensive-iterative of QUATERE), contaminated by Graeco-Rom. (naut.) CAMPSĀRE, which underlies It. (s)cansare; add OPtg. inchaço 'swelling'INFLÃTIÕbeside OSp. finchazón. The loss of ծ- before -l-, in particular — a peculiarity un derlying Corominas' pointed comment on an earlier isolated finding by G. Bonfante, apropos ladilla 'crab-louse') — deserves to be studied in the context of the separate phonology of semicultismos. 29 Without denying the possibility of a link between lágrima and lástima, one cannot help insisting on its tenuousness. One piece of circumstantial evidence is the predominance of the plural of 'tear', lágrimas, either in stereotyped phrases such as verter lágrimas 'to shed tears' (San Millán, 310d; Duelo, 4b; Apolonio, 283c), derramar lágrimas 'id.' (Ruiz, 830d), tener lágrimas = mod. retener las lágrimas 'to control one's tears' (Apolonio, 160c, 175c), or in any other context (Apolonio, 326c; Ruiz, 636c, 741d, 792d, 842a); as a matter of fact, there is just a sprinkling of phrases compatible with the singular, e.g., no pudo echar lágrima 'no pudo hacer correr una sola lágrima' (Apolonio, 448d; . . Marden, in the Glossary attached to his edition [II, 128], refers the reader to Alexandre, P, 1358c). Nevertheless, few attempts were made to use lástimas '[manifesta tions, outbursts, spells of] pity', etc. — this despite such tempting precedents as celos 'jealousy', cariños 'words of love, show of affection', to say nothing of caricias 'caresses' and, optionally, of abstracts rooted in anatomic imagery: tener sorbido el seso ( = sorbidos los sesos) a 'to dominate, have unlimited influence on, be madly in love with, be deeply immersed in'; cf. Anita Katz Levy, "Plural Form Versus Singular Meaning in Hispano-Romance Nouns", RPh, XXVII, 13-25. The one exception is the familiar phrase listed by the Academy Dictionary: llorar lástimas. Incidentally, even though lágrima was the prevalent form in medieval literature (Libro de buen amor, S and G [including lágrimas de Moysén 'books of magic', 438d], Santa María Egipciaca, ed. Knust, 13r°, Guillelme, 38v°, 44v°, Confisión del amante, 56r°, 353v°, etc.), there existed alter natives. Thus, the unadulterated Latinism lácrima, traceable to Barlán e Josaphá, 180v°, alter nated in Lucas Fernández with lágrima; that Salamancan playwright used (l)lagrimoso, while El Cartuxano favored lacrimoso, a usage reminiscent of El Marqués de Santillana's lacrimable. See J. Lihani, El lenguaje de Lucas Fernández; estudio del dialecto sayagués (Bogotá, 1973), pp. 475, 478. The point is badly garbled in the DCE, III, 14b, which, on the other hand, helpfully docu ments lágrema (Santo Domingo, 544c) and lagremal 'lachrymal' = G. 'Tränensack' (Santa Oria, 201d; Pero Guillén de Segovia) beside lágrima (Milagros, 247c; the late 15th-c. vocabularies of Palencia and Nebrixa). The absence of *lástema does not exactly strengthen Corominas' conjec ture, though it might be explained away by the late date of the emergence of lástima.


553

speakers' steady preference for i, an, at first blush, unintegrated idiosyncrasy which begins to fall into shape once estimar is introduced into our equation. The erratic development bl- > l-, on which Corominas spills much ink, also gains in verisimilitude once we realize that some of its closest parallels, GLANDE 'acorn, acorn-like protuberance' > land(r)e 'small tumor (in glands of neck, armpit, groin)' and, in its wake, GLANDULA 'small acorn' > Ptg. lândoa 'crack, cleft in the trunk of an oak- or chestnut-tree', etc. seem to have pertained initially to the layer of folk-medical or anatomical terms closely associated by students of transmission with the semilearned stratum. 30 The few adjectival derivatives from lastimar (-ador 'hurtful, injurious', -ero [a] 'hurtful', (b) 'pitiful, doleful', ֊oso 'pitiful'), the dialectal variants (llastimin Sayagues), the representation of the word family in Portuguese and in certain easily identifiable loans to Modern Catalan both continental and insular (Barc. llàstima, for which purists substitute malaguanyat', Bal. llastimar 'to complain') and to Sicilian (lástima 'annoyance, bother'; see DCE) all pose interesting sideissues, which, despite the full spectrum of their implications, are unlikely to take us any closer to the solution of the nuclear problem. In contrast, the noun lástima 'pity, complaint', ringed by a number of sharply profiled idiomatic phrases (dar, hacer,  poner lástima 'to be pitiful, make one feel compassionate', estar hecho una lástima 'to be in a sorry sight', es lástima que... 'it is a pity that... ' [= Ptg. é pena que...], coll. llorar lástimas 'exagerarlas'), and used in exclama tions, almost on a par with interjections (¡lástima que...! 'it's too bad ...', ¡qué lástima! 'what a shame!' 'what a pity!'), certainly is relevant to the etymolo gical riddle. Is lástima a postverbal ? If so, it should carry the stress on the i, cf. estimar → estima, halagar→halago, recamar→recamo. (The handful of exceptions, such as suplicar ~ súplica, involve straight Latinisms.) Could the verb, in general, have preceded the noun, but in certain uses, as if by recoil, have been influenced by it ? One inclines to apply this analysis to lastimarse 'to complain about, feel sorry for'. To what extent did the emphatic exclamatory use affect, prosodically and in other respects, the form of the noun lástima ? The disappointments for the etymologist start with the discovery that textual critics are divided on the interpretation of the crucial quatrain, or rather stanza, 1052 in the Libro de buen amor. I n J. Ducamin's paleographic edition (1901), which confines exegesis to punctuation, the passage, based on MS S, reads: " T u con el estando aora de prima/ viste lo 30

The group of word-initial clusters VOICED STOP (or SPIRANT) + LATERAL in Latin and Ro

mance was, from the outset, weakened by the absence of *DL- and *VL- (among their voiceless counterpart the exclusion of *.TL- was partially compensated for by the availability of FL-). Given this general vulnerability of BL- and GL-, one is hardly surprised to discover that the occlu sive element was occasionally sloughed off. There are indeed a few cases of dissimilatory loss, occasionally in conjunction with folk-etymological interference (consider the case of GLOB-ՍԼԱ *-ELLU 'small ball [of yarn]' > Sp. ovillo lit. 'small egg', etc.; see REW 3 3791); there are further a few cases of transmission in a very rustic stratum: GLAREA 'gravel' > Ptg. leira, GLATTĪRE 'to yelp' > Sp. latir 'to throb'; GLĒBA 'clod, lump' > Ptg. leiva, Logud. lea (cf. GLĒBULA 'small clod' > Campid. leura); GLIRE 'dormouse' > Abruzz. lire, var. GLĒRE > Fr. loir, dimin. *GLĪRIÕNE > Sp. lirón, *GLĒR-IONE > Ptg. leirão, Gal. leirón;, *GLOBITSCELLU 'small ball [of yarn]' > OFr. gluicel, mod. luissel, etc. BL-appears to have been more resistant to erosion.

554

ETYMOLOGY

levando firiendo ¡que lastima! (G: 1. e feridas 1.),/ Pilatos judgando, escupenle en çima (G: e escupen lo),/ de su faz tam (G: tan) clara, del çielo resplandor". From this presentation H. . Richardson, three decades later, distilled the gloss 'pity' and classified the word as a noun. 3 1 The stanza is, unfortunately, absent from Maria Rosa Lida's annotated Selección (1941). Vol. III (1956) of Corominas' , at p. 40b, offers this ֊ metrically divergent — reading: "Tú con él estando,/ a ora de prima,/ vistelo levando/ firiendo que lastima;/ Pilatos judgando,/ escupenle encima/ de su faz tan clara,/ del cielo rresplandor", adding the transla tion: 'viste cómo se lo llevaban, golpeándole en forma que daba lástima'. In his critical edi tion (1967), however, the Barcelona philologist changed his mind, offering the quatrain, which was now allowed to enter into "De la passion de Nuestro Señor Jesuchristo":"Tú con él estando,/ a ora de prima/ viste que, levándol,/ ferido s'lastima;/ Pilatos juzgándol,/ escúpenle encima/ de su faz tan clara,/ del cielo resplandor". 3 2 This decision pushes Corominas into the uncomfortable position of having to admit, first, his inability to document the noun lástima before Sir Richard Percivale's vocabulary (1591): 'greefe, hurt'; 3 3 second, the word's prior existence in Portuguese — as early as 1540 (Mendes Pinto, Morais Cabral), even though Portugal assuredly was not the cradle of this lexical family; and, third and most embarrassing, his suspicion, based on accentual con siderations, t h a t the elusive lexical item is likely to be discernibly older. 34

It seems that almost all the loose pieces fall into place once we assume that *blastimar — with a b- inherently weak and amenable to change, including loss, at any light provocation — succumbed quickly to the spell of a group of semantically germane words of diversified provenience which, as has been in dependently shown,35 formed a kind of loose alliance or confederacy. This bundle of words included: lazrado 'unfortunate, wretched' (from laz-, laze-, lazd-rar, a polymorphous verb flaanked by lazer-ία, -io 'misery'), which was undoubtedly used in lamentations and imprecations; lastro 'trouble, affliction' (Ruiz, S, 1311a: "Saly desta lazeria, de coyta e de lastro"; T: lasco), cf. the corresponding verb lastrar or, in its Sayagués garb, llastrar 'to afflict': "Harto lo tengo llastrado/ y trabajado/ en passar vida tan triste" (Lucas Fernández, Cvi r°, after J. Lihani); 31 An Etymological Vocabulary to the "Libro de buen amor"... (New Haven & London, 1930), p. 133. 32 In a fn. the editor records two more interpretations: "Visto lo levando" (F. Hanssen) and "Viste lo levando a todos lastima" (J. Cejador y Frauca); also, in his running commentary (pp. 406b, 408a), admitting that the MSS have garbled the passage, he takes to task J. M. Aguado, H. H. Arnold, R. S. Boggs and his associates (Tentative...), and F. Lecoy, in addition to Ducamin and Richardson, for having misconstrued the stanza — but carefully avoids mentioning his own change of mind. He also offers a new alternative of his own: "Vístel que, levado, ferido, lastima", then hastens to withdraw it — an act of merely thinking aloud or of coquetry with emendatory conjectures ? 33 We are led to believe that lástima then, almost overnight, soared to great popularity, occurring as it did in Cervantes, Covarrubias, Oudin, etc. — all of them ca. 1600. Did there suddenly develop some fad ? 34 "Puede ser formación antigua, creada cuando todavía había verbos con presentes esdrújulos, o bien ser creación moderna y analógica de lágrima junto a lagrimar". But the type sinífica 'he, she, it signifies' here alluded to, peculiar to Berceo, not only antedates the appearance of the family of lastimar by fully a century, but, worse, was restricted to unadulterated Latinisms, especially those rich in ecclesiastic overtones. 35 See my paper "La familia léxica lazerar, laz(d)rar, lazeria: Estudios de paleontología lingüís tica", N R F H, VI (1952), 209-276, in general, and, in particular, the discussion of certain mar ginal variants and confusions such as la(z)drar, lastar, etc. (pp. 241-244); also, the hint (pp. 240f.) that the hapax legomenon llazdrar (alongside recurrent lazdrar) in the Libro de Apolonio may be due to contamination with llagar 'to wound'.


555

lastar 'to pay penalty' (Ruiz, G, 667a: "A las vegadas lastan justos por peca dores"); and, above all, lasso 'tired, weak, listless', used in romances of chivalry, in hagiographic accounts, and a favorite of Juan Ruiz's, 36 which undoubtedly lent itself to self-commiserating use no less than its presumable source, Fr. hélas!. Parodi's and, initially, Meyer-Lübke's margin of error, then, consisted in over playing the impact of AESTIMĀRE on BLASPHĒMĀRE, while an older but for once not wiser Meyer-Lübke's second error, aggravated by Corominas, involved the staunch denial of any such influence. The truth, unexcitingly, seems to lie in the middle between these two extreme positions. In certain contexts there appear to have occurred temporary rapprochements between the two verbs, with either the positive or the negative partner wielding the greater power. Thus, clearly, of the two or rather three Old French verbs, (a)esmer and blasmer, projected onto Spanish soil, it is the former that yielded pride of place to the latter. Under the impact of blasmar the two rival verbs aesmer and esmer (indubitably weakened by this split) assumed the form asmar, to the virtual exclusion of esmar, except in Portuguese. 37 Since both estimar and lastimar emerge at the surface in 14th-century texts, it seems legitimate to trace both to a clerical milieu and to argue the spread of -imar from learnèd estimar. Regrettably, the crucial step *blastimar > lastimar cannot be reconstructed through inspection of extant texts, but close semantic association with a cluster of words intimately enmeshed even before the arrival on the scene of the verb at issue, above all laz(e)rar, lastar, lastrar, lasso, all of them used in lamentations, would account for the fitness of last-, for the stress on the a, and even for assorted stylistic and phraseological peculiarities, which already caught the attention of Diez; if lágrima played any rôle at all in this process, it must have been a fairly small one. Above all, the meteoric rise of lastimar, lástima coincides with the dramatic decline of laz(e)rar, last(r)ar, lasso, etc.; on circumstantial evidence, then, lastimar, lástima inherited the doomed words' estate between the 15th and the 16th centuries and thus, to this day, Spanish represents them in an unbroken line of organic continuity. 36 Cf.: "Do non es tan seguida anda más floxa, lassa" (G, S, 523d; : massa, passa, traspassa); "non seas rebatado nin vagaroso, laso" (G, 550b;: passo, escasso, traspasso); "alegro-me con tristeza, lasa mas enamorada" (S, 855c; does MS G's reading lexa reveal the vernacular, indi genous counterpart, cf. the adv. lexos 'far away'?), apparently from LASSXJS 'weary, tired, ex hausted', contaminated by LAXUS 'wide, spacious, loose, lax, relaxed'. Found in a wide variety of medieval texts: Santa María Egipciaca, ed. Knust, 9v°, 11v°, 13v° [twice]; De una emperatriz..., ed. Mussafia, Ch. xiv (box lassa 'tired voice'); Carlos Maines, ed. Bonilla, Ch. xxxii; Emperador Ottas, ed. Amador de los Ríos, Chs. IV, XXVI, XL, XLI. On author's or narrator's Las! in Old French see Minette Grunmann, RPh, XXIX, 204. Old Portuguese also had, at a certain distance from the tangle at issue, laido 'wounded, hurt, offended' and the corresponding verb laidar, as distinct from lai 'ugly', although the same family of words borrowed from Gallo-Romance and, in the last analysis, a Frankish source seem to be everywhere involved (REW 3 , 4855); see M. Rodrigues Lapa, VocabulArio galego-português... (1970), p. 54. 37 Interestingly, OFr. aasme, disme '(ac)count' (G. Lené, Les substantifs postverbaux dans la langue française, p. 67) ~ esme 'id.' (Meyer-Lübke, HGFS, II, §109) failed to strike root in Old Spanish, even though blasmo was transplanted along with blasmar.

556

ETYMOLOGY

As for the cause of the measurably older intrusion of the է in BLASPH(Ē)MĀRE (cf. It. bestemmiare, etc.), shortly before the disintegration of Latin, the analyst retains several explicative options. Not the least persuasive of these is the argument that fluctuations of taste with regard to syncope prompted speakers using forms equipped with a buffer consonant (*blas-t-mare, not unlike OFr. cos-d-re 'to sew' from CÕ(N)S(U)ERE or OSp. laz-d-rar from LACERĀRE) to experi ment with more "relaxed" variants, such as *blast-imare or -emare. E X C U R S U S : T H E ETYMOLOGY OF H I S P .

TOMAR

Any careful inspection of the record of AESTIMĀRE, whichever its professed aim, is apt to lead to the reopening of one of the major issues of Romance etymology: the provenience of Sp.-Ptg. tomar 'to take'. Although the existence of a rival form AESTUMĀRE is well known (cf. MAXIMUS ~ MAXUMUS and the like a ), the Romance scholar's first reaction is to view the -IMĀRE variant as the base form and the counterpart in -UMĀRE as a mere alternative. We are encouraged in this belief by two facts: the greater productivity of t h e -IMĀRE branch within Latin (witness EXISTIMĀRE 'to appraise, value, estimate, judge', traceable to Plautus and surrounded by a number of satellites: -ĀTIÕ 'appraisal', -ĀTOR 'critic, judge', etc.); and, equally impressive, the well-nigh absolute prevalence of -IMĀRE in Latinisms since the Middle Ages: cf. E . esteem and estim-ate, -ation; Fr. estime beside estimer; It. stim-are, -a, -ativo, -atore, -abile, also disistima(re) and estim-are, -abile, -atore, -azione, -ativo, (subst.) -ativa beside estimo. However, the historically oriented Latinist does not see things at this angle a t all. I n preparing a dictionary, he tends to offer AESTUMO as his main entry and to list AESTIMO as a mere variant. b Again, the view of the semantic range of AESTIM-/AESTUM- is slightly blurred by our fore knowledge acquired through medieval and modern Western languages. The basic nuance was that of 'appraising, rating, valuing, estimating' (with the characteristic genitives MAGNI, PARVI) ; Hofmann starts from 'schätze den Geldwert einer Sache ab, taxiere'. Only then came such semantic hues as 'to take into account (or consideration), judge, hold' ( = Hofmann: 'würdige, ermesse'; AESTUMĀRE LITEM 'faire évaluation d'une valeur en litige' [Havet]) and, finally, 'to judge favorably, esteem, respect' (which to us moderns looms so dispropor tionately important). Certain details of the development, on the level of the literary langu age, betray the interference of Greek models. c a b

Cf. Ernout and Meillet, DÉLL (P., 1959-604), 13ab. The adjectival suffix -ŪMUS (FĪNITUMUS from FINIS, LĒGITUMUS from LEX, MARITUMUS from

MARE, cf. the noun AEDITUMUS) acquired its superlative meaning (as in MAXUMUS from MAG- and OPTUMUS from OPS) secondarily; the -IMUS vars. are the ones that survived into the modern langu age (cf. E. legitimate, maximize, optimal), while the Latinist hierarchically subordinates -IMUS to -UMUS. See L. Havet, "Mélanges latins", MSL, VI (1885), 11-39, at 18n2, who derived AUTUMĀRE from an adj. *AUTUMUS and even toyed with *AESTUMUS 'évalué en espèce' as the base of AESTUMĀRE; M. Niedermann, Précis de phonétique historique du latin (P., 19533), p. 24, who asses sed Quintilian's hint to the effect that і and  may be mere graphies competing for the rendition of /y/ and emphasized that MINIMUS alone is on record; R. G. Kent, The Sounds of Latin (Balti more, 19453), §§24, 32, 34III, who traced FACTL[L]UMUS, INFUMUS to inscriptions and also invoked an /y/; A. Ernout, Morphologie historique du latin (P., 19533), p. 77, who focused on OPT-IMUS/ -UMUS, PESSIMUS, PLŪRIMUS, and MINIMUS only within the framework of suppletion. c Thus, Cicero is reported to have minted AESTIMĀBILIS in an effort to emulate the Stoic philosophers' άξίαν €χων. AESTIM-IUM ~ -IA 'estimate' (actually pl. -IAE in Paulus ex Festo) have been tagged as "technical words". Though there is no lack of native Latin formations involving

-IUM or -IA (cf. DOMIN-IUM, IŪDIC-IUM; ASTŪT-IA, [ p l . ] ARGŪT-IAE, D Ī V I T - I A E ) , t h e c o e x i s t e n c e o f

Gr.-Lat. BLASPHĒMIA (and *-IUM) is, to say the least, noteworthy.


557

The fact t h a t commercial value was the original frame of reference explains why the ancients so easily read AES -RIS (Ո.) 'copper, bronze, money, payment, reward, wages' into AESTUMĀRE. Paulus ex Festo was quite explicit on this score: "Aestimata poena ab antiquis ab aere dicta est, qui earn aestimaverunt aere, ovem decussis, bovem centussis, hoc est decern vel centum assibus" (23,1). Interestingly, at least one noted modern linguist, L. Havet, etymologized AESTUMÕ as a denominative verb geared to *AIS TEMOS 'he who cuts bronze', citing in support the juridical formula per aes et libram expendere atque aestimare.d Critics have held against this conjecture one damaging circumstance: *TEM- 'to cut' is a lexical morpheme alien to Latin (see DÉLL4, I, 13b). Granting the validity of the objection, even a secondary association of AES with AESTUMĀRE would clearly have stimulated speakers to grant a certain autonomy to *-TUMĀRE and to endow it with a meaning. Could this seg ment *-TUMĀRE, assumed to have cut loose from AESTUMĀRE, possibly be the long-sought source of Sp. Ptg. tomar 'to take' ? This is, clearly, not the place to review the record of earlier etymological conjectures proposed for tomar — a knot difficult to disentangle —, especially since a miniature study of this kind has just been undertaken elsewhere. e Suffice it to state that, half a century ago, a scholar as experienced, as endowed with intuition, and, at the same time, as realistic as Jakob J u d recognized no better way out of the maze of blind alleys (Goth. *TÔMJAN, etc.) t h a n a return to AUTUMĀRE 'to assert, affirm, say', f a hypothesis originally launched by P. Rajna (1919) and, thirty years after the interlude of J u d ' s trespassing on Hispanic territory, upheld by Corominas. g AUTUMĀRE and AESTUMĀRE as rival candidates for the d Ernout is a bit inaccurate in citing p. 18n2 of Havet's article, a passage where actually an alternative explanation was ventilated; the correct page reference should have been 23. J. B. Hofmann, in his revision of A. Walde's LEW (see I 3 , 20f.), also refers to F. de Saussure, Recueil des publications scientifiques (Genève, 1922), p. 591, and supports the second rather than the first of Havet's alternative conjectures, throwing in for good measure phraseological evidence: the idiom per aes et libram beside Cicero's expendere atque aestimãre. Hofmann also reviews — with brief pungent comments — some older explanations of AESTUMĀRE, no longer tenable. As regards Saussure's position, note that the passage at issue occurs in his paper "Sur les composés latins du type agricola" and is quite brief: "Un mot *aes-tuma 'le coupe-bronze' a été supposé par Louis Havet... pour expliquer aestumare: il s'accorderait au mieux avec les formes disyllabiques grecques de la racine en question". From a lengthy footnote appended to this remark one gathers: "...ľétymologie d'aestumare y était doublée d'une analyse ď*aes-tuma lui-même, de laquelle il ressort que l'auteur posait -tuma comme identique à l'élément disyllabique radical de τέμα-χος etc.,... C'est en deux lignes, on le voit, tout l'essentiel de la théorie que nous présentons nousmême, en l'appliquant à l'ensemble des mots comme agricola...". e See my paper "Deux catégories d'étymologies 'intéressantes'", to appear in RLiR. I now regret not having put to use on that occasion A. Ernout's magisterial note on "Autumāre" in Latomus, I (1937), 75-79. The initial meaning of this verb was 'to say'; it was so used by Apuleius, Tertullian, and Arnobius. The meaning 'to believe' is clearly secondary; it made its appearance once in St. Jerome, while Cassianus and Claudius Mamertinus were divided in their preferences; the new meaning became exclusive in Ammianus, Prudentius, St. Augustine, Sidonius Apollinaris (three attestations), and Gregory of Tours. The Carolingian glossaries (absorbed into the CGL) were split in their preferences. Ernout shows convincingly that a single passage in an in fluential treatise (Quintilian, Inst. or., VIII,3,24: "Reor, tolerabile; autumo, tragicum"), whose message did not point in any innovative direction ('Reor est acceptable, autumo est du vocabu laire tragique'), led to an unforeseeable rapprochement between the two verbs paired off by the great educator. Ernout views autumõ entirely as a lexical isolate, of unknown provenience, and bereft of any Romance progeny; he attributes its success in Late Latin (and beyond) to artificial revival at the hand of archaizers. Crucially important for our purpose is the anecdotal fact that aestumõ and autumõ were associated in an "étymologie fantaisiste" launched by Nigidius Figulus and reported by a dissenting Aulus Gellius and by Macrobius. f As is well known, Meyer-Lubke, REW3, 975, s. v. (onom.) u() 'Fall', 'Plumps', rejected the Rajna-Jud hypothesis categorically: "ist wenig wahrscheinlich". However, there is not one scintilla of expressivity, phonosymbolism, or iconicity in the actual medieval use of tomar. g , IV [1957], 490b-494b. As regards details, Corominas distinguishes between Rajna's and Jud's formulations, preferring the latter version, and cites such back-formations as (IN)COLUMIS

and (IM)BECILLUS, (RE)CUPERĀRE and (COM)BUSTULĀRE, (CO)OPERĪRE and (DES)TINĀRE, and falsely

558

ETYMOLOGY

niche to be occupied by the actual progenitor oí tomar have this in common: that, to qualify, they each must undergo severe apheresis. But while the mutilation of AESTUMÃEE makes sense to a certain extent against the background of Antiquity's half-serious, half-playful rapprochement of AES and AESTUMĀRE, the surgery to be performed on AUTUMĀRE — a verb of doubtful durability, to boot — is somehow lacking in plausibility. Since AESTUMĀRE, at an advanced stage of its semantic growth ('to judge, hold'), and AUTUMĀRE ('to affirm, say') became in any event near-synonyms and were, upon occasion, confused by copyists, it is not inaccurate to reconcile the thesis here proffered with the Rajna-Jud conjecture by averring t h a t the coexistence, as close neighbors, of AES-TUMĀRE, subject to folk-etymolo gical segmentation, and AUTUMĀRE against the backdrop of marginal prefix variant AU(cf. AUEERRE 'to bear, take, or snatch away', alongside EERRE 'to carry'; AUFUGERE [tr.] 'to escape, flee from', [intr.] 'to run away', alongside FUGERE 'to flee'h) was apt to contribute to the gradual autonomy of an independent lexical morpheme *TUMĀRE. The process here hypothecated is, admittedly, rare, but hardly as outlandish as Corominas' bizarre and yet not entirely absurd derivation of Sp. tez 'complexion' from *atez 'smoothness' (cf. OSp. apteza 'perfection, skill', OCat. apte(s)a 'adroitness, robustness'). i If our argument holds water, then tomar, counter to the more familiar sequence of events, must initially have had a fairly broad, general meaning, which to this day survives in tornar a bien 'to take the right way' (i.e., in the right spirit), tornar a mal 'to take offense at', tomar por 'to take for' (in the sense of 'misjudging'); only step by step did tomar become the fullfledged equivalent of Fr. prendre which, descending as it does from PRAE-HENDERE, from the start meant 'to snatch, seize, arrest (or take prisoner)', exemplifying the more common evolutionary line from the concrete toward the abstract. In the Iberian peninsula, it is apparently the rise of tomar t h a t blocked the spontaneous development of prender, congeal ing it semantically close to its starting point. If this analysis is correct, then we should distinguish between three deposits of AESTUMĀRE/-IMĀRE in Hispano-Romance: (a) AES/TUMĀRE, presumably in intermittent contact with AU/TUMĀRE; (b) asmar, imported from beyond the Pyrenees and slightly influenced by blasmar (and eventually by pasmar ?); () learnèd estimar, absorbed ca. 1400.

segmented (re)hez as a source οf so-(h)ez (p. 493a) — interesting examples indeed, except that they all involve the omission of common prefixes (as in cuerdo beside a-, re-cordar, or in prieto beside apretar), whereas the au- of AUTUMĀRE could not qualify as more than a marginal prefix variant. By substituting AESTUMĀRE for the Rajna-Jud-Corominas conjecture, through appeal to Paulus ex Festo's derivation from AES, we offer a pattern of resegmentation for which actual witnesses, i.e., bearers of prima facie evidence, can be summoned. h The evidence of AUCEPS 'fowler', fig. 'eavesdropper', AUCUPIUM 'fowling, trap', pl. 'quibbling', and of AUCUP-ĀRE/-ĀRĪ 'to lie in wait for, watch for, chase, catch' is not to be thrown out just be cause au- < AVI- suggests blurred composition rather than neat derivation. AU-CEPS was con trastable with PRĪN-CEPS, AU-CUPIUM With PRĪN-CIPIUM, AU-CUPĀRE with OC-CUPĀRE (all three stem allomorphs: --, -CIP-, and -CUP- happen to be related to CAP-IÕ/-ERE). Even this minor reinforcement scarcely suffices to bolster the status of AU-, (Although, for the purposes of Latin, one is tempted to classify AU- as a variant — preferred before F. of Ā-, AB-, ABS-, its Ι.-Ε. source is, actually, different and it has cognates in Old Irish, Old Prussian, and Old Slavic, see DÉLL 4 , p.2b.) 1 See Corominas, "Problemas del diccionario etimológico" (II), RPh, I, 79-104, at 100-103; and DCE, IV, 437a-439a.

SUPPLEMENT The following comments and stray data represent an attempt to bring some of the preceding pieces up to date, without any radical recasting of the original wording. The stress has been principally placed on later studies of my own in which some of the approaches that can be gleaned from the papers included were allowed to under go further revision and refinements. Here and there incidental references have been made to critical reactions to the items as initially published, or to germane studies undertaken by others, or to more elaborate analyses of relevant details than those that were available to me at the original time of writing. As a further aid, several forthcoming publications have been included in this survey, which, however, has not been conceived as something truly exhaustive, since not a few discussions (perhaps even the most fruitful ones) happen to be still in a state of flux. Bibliographic items of only indirect relevancy have sometimes been cited with laconic brevity, especially those that can be identified with relative ease with the help of standard bibliographies.

"Linguistics as a Genetic Science" Embodying as this paper does the text of a formal address, it is, for once, lacking in documentation; in cross-references to already available or forthcoming studies; in bibliographic references; and in any other routine ingredient of the typical scholarly apparatus. For the one exception from this rule, the article alluded to in n. 1, more accurate information can be provided at present: To Honor Roman Jakobson: Essays on the Occasion of his Seventieth Birthday (The Hague & Paris: Mouton, 1967), II, 1228-1246. The concept of multiple causation pervading both pieces was further refined in an article not yet foreseen in the mid 'sixties: "Multi-Condi tioned Sound Change and the Impact of Morphology on Phonology", Lang., LII (1976), 757-778, plus a note (forthcoming in HR): "Interplay of Sounds and Forms in the Shaping of Three Old Spanish Medial Consonant Clusters". The analysis offered under Problem No. 1 (pp. 35-38) rests squarely on a number of individual word biographies which cannot all be laboriously pieced together in էհե context. Suffice it to refer to a few sample studies: On OSp. rezio vs. Ptg. rijo 'stiff see S.N. Dworkin, "Therapeutic Reactions to Excessive Phonetic Erosion: The Descendants of rigidus in Hispano- and Luso-Romance", RPh, XXVIII (197475), 462-472. On Sp. sandio 'foolish', orig. sandio, beside Ptg. (m.) sandeu, (f.) sandia, see p. 481 of my own article, "From Falling to Rising Diphthongs: The Case of Old Spanish io < éu", RPh, XXXIX:4 (1976), 435-500. Regarding amplio 'wide' it might have been stated that the starting point appears to have been the n. amplius 'more, longer, further, besides' rather than the m.f. -ior; and that amplius, in Clas sical Latin already, competed with plūs, magis, and potius. For Problem No. 2 (pp. 38f.) the collateral sources of information, from among

560

SUPPLEMENT

my own — preceding or later — pieces, are: "Estudios de léxico pastoril: piara y manada", BH, LIII (1951), 41-80; "Ernst G. Wahlgren - et les perspectives d'une réhabilitation de la morphologie", StN, XLIX (1977), 69-85 ֊ with a circumstan tial discussion of the Swedish scholar's controversial monograph: Un problème de phonétique romane: Le développement D > R (1929) and of reactions to it; plus the necrological essay on Tomás Navarro Tomás in RPh, XXXIV, Special Issue (February 1981), pp. *98-*115, at *109. For Problem No. 3, involving hypercharacterization (pp. 39f.), I could fall back on a major study of much earlier vintage: "Diachronic Hypercharacterization in Romance", ArL, IX (1957), 79-113, X (1958), 1-36, best consulted in its expanded Italian version which we owe to the skill of Ruggero Stefanini; see my miscellany Linguìstica generale, filologia romanza, etimologia (Bologna: il Mulino, 1970), pp. 170-239. A new piece has now been grafted onto the 1957-58 article: "Hyper characterization of Pronominal Gender in Romance", to appear in 1982 in: Language, Meaning, and Style: Essays in Memory of Stephen Ullmann (issued by the Depart ment of French, University of Leeds). Issues raised in Problem No. 4 (pp. 40-42), a propos consonant dissimilation in Romance, have since emerged in countless etymological vignettes, but no new bold synthesis, on the order of R. Posner's and K. Togeby's rival experiments, has made its appearance. The side-glance at binomials was made possible by my earlier "Studies in Irreversible Binomials", Lingua, VIII (1959), 113-160; cf. the elaboration by D.L. Bolinger, "Binomials and Pitch Accent", ibid., XI (1962), 3444. The elusive matter of secondary causation enters into the gamut of questions broached in a lengthy inquiry of mine, "Alternatives to the Classic Dichotomy Family Tree/Wave Theory? The Romance Evidence", which is to form part of a new miscellany (provisionally titled Language Change) now being launched by Irmengard Rauch; the publisher is Indiana University Press.

"History and Histories of Linguistics" This composite review article, coming in the wake of a more impressionistic appre ciation (1964) of studies in the history of linguistics by John T. Watermann, L. Kukenheim, and Iorgu lordan/Werner Bahner, reflects a certain characteristic mood of the 'sixties. From the late 19th century on one detects isolated attempts at spadework in the history of linguistics. Twenty years or so ago, something of a trend, not to say infatuation with this side issue, developed — perhaps indepen dently in a variety of countries, including Sweden, England, France, Italy, Yugo slavia, the Soviet Union, Argentina, and of course the United States. It is arguable that there suddenly prevailed the feeling that linguistics, in part as a result of its strangely zigzagging advances, was moving away from its age-old moorings to lüstory at such a giddying speed that it was imperative to recapture its past and the earlier phases of its subsequent evolution before it was too late, i.e., while its essential identity was still recognizable. As a consequence of this latent anxiety which, if one views things in retrospect, appears to have gripped large groups of professionals and laymen, there arose a certain need to fill the above-mentioned vacuum as expedi-

SUPPLEMENT

561

tiously as possible; in response to that acutely felt hunger for information, a whole crop of rather uneven books was speedily — almost hastily — produced (not always, one is chagrined to add, by the best-qualified researchers). The purpose of the article here reprinted, dating from 1969 and meant to mirror that volatile Zeitgeist, was threefold: to lay down certain norms conceivably worthy of acceptance or, at least, of discussion by those engaged in this sort of inquiries; to assess a number of relevant books characteristic of those vintage years, books then still new; and to offer, by way of supplement, a working bibliography, which it would have been pointless to bring up to date, since the mood hinted at has mean while become part of an irretrievable past (for details see below). A companion piece published just nine months later under the title "Linguistics (Including its History) and the Humanities: Two New Approaches to a Fluid Relationship" dealt with Karl D. Uitti's Linguistics and Literary Theory, elegantly printed by Princeton University Press, and with Alberto Vàrvaro's Storia, problemi e metodi della lingui stica romanza (Napoli: Liguori, 1968): Though ostensibly, judging from the title, that textbook was slanted in an entirely different direction, its strong secondary emphasis on the succession of schools of thought made it eligible for inclusion in this group, whose core comprised Maurice Leroy's, Milka Ivić's, Bertil Malmberg's, and Giulio . Lepschy's well-known guides, to which must be added Thomas A. Sebeok's Portraits of Linguists — actually a string of reprinted necrologies. Part IV of the original text, which occupied pp. 566-572 of the piece as it appear ed in Vol. XXII of the journal Romance Philology, was from the pen of my talented junior colleague Margaret Langdon; it has here been reluctantly omitted, in an effort to preserve a modicum of homogeneity. Its generally fine quality and, in particular, its verve are not, needless to say, at issue. By concentrating on the image of New World linguistic scholarship, as it emerged from the recent portrayal by — chiefly European — historians of linguistics, my co-author hit upon a very original focus. It is a pity, let me add parenthetically, that certain Europeans conversant at first hand with the North American scene (e.g., the late Alf Sommerfelt, a distinguished Norwegian), should never have bothered to offer their accounts of this segment of the United States' intellectual growth; also, that an appraisal as circumstantial as André Martinet's of Charles Fillmore's grammar should have become available at a relatively late point along the time axis. Margaret Langdon's survey invites com parison with a book such as the long-delayed (1979) collection — edited by Henry M. Hoenigswald — of papers read at the Third Golden Anniversary Symposium (New York, 1974) of the Linguistic Society of America, a miscellany which contains, inter alia, the Dutchman EM. Uhlenbeck's rather critical piece ("Linguistics in America 1924-1974: A Detached View"), plus D. Terence Langendoen's spirited discussion of it (pp. 121-159, with rejoinder and surrejoinder included). My own contribution to that venture ("Aspirations, Organization, Achievement", pp. 107119) was throughout in a fairly low key. Now that the history of linguistics has ceased to be a fad, research along that line is carried on methodically and without unseemly precipitancy (perhaps un leashed by Noam Chomsky's highly controversial book on Cartesian Linguistics). Shorter high-level contributions appear these days in the newly-launched quarterly, Historiographia Linguistica, in which I had the good fortune of seeing my own note,

562

SUPPLEMENT

"Between Heymann Steinthal and Adolf Tobler: Georg Cohn in Turn-of-the-Century Berlin" appear in 1978 (V: 3.237-251). I have not had any opporunity yet of presenting any chunk of research signifi cantly transcending "History and Histories of Linguistics" on the side of methodo logy. However, various ideas sketched out in it have been further developed in three kindred genres of critical research: (a) extended book reviews concerned with these issues, (b) necrological essays, as well as, at rarer intervals, (c) pen-portraits of lin guistic scholars of a more distant past. (a) My interest in Diez as the co-founder of comparative Romance linguistics has led me to review in commensurate detail the transactions of the 1965 Diez Colloquium in Bonn; see Lang., LIV (1978), 426432. Two collateral studies of Diez frommy pen - one distinctly earlier, the other slightly later — are: (1) "Friedrich Diez's Debt to pre1800 Linguistics" in the miscellany promoted by Dell Hymes, Studies in the History of Linguistics: Traditions and Paradigms (Bloomington & London: Indiana Univ. Press, 1974), pp. 315-330; and (2) "Friedrich Diez and the Birth Pangs of Romance Linguistics", embodying a lecture which appeared in a Supplement {Friedrich Diez Centennial Lectures) to the journal Romance Philology, XXX:2 (Nov. 1976), 2-15, with a commentary by Edward F. Tuttle attached to it. Pre-Diezian linguistics is highlighted in two circumstantial book reviews: one bearing on the slightly bizarre 1978 Liège monograph by Daniel Droixhe: La linguistique et Vappel de ľhistoire (1600-1800): Rationalisme et révolutions positivistes (see Lang., LVI:2 [1980], 427431): the other focusing on an influential book of the Napoleonic era: Jens Lüdtke, Die romanischen Sprachen im "Mithridates" von Adelung und Vater; Studie und Text (see Krat, XXIV [1979-80], 117-126). (b) It might be pedantic to specify exhaustively the string of necrological essays (published for the most part in Romance Philology) stretching from the fairly short piece on Karl Jaberg (XII:3, 258-261) all the way to those, longer and, by a wide margin, fuller of factual details, on Émile Benvemste (XXXIV, 160-194) and Tomás Navarro Tomás (XXXIV, *98-*115). Some other linguistic figures so portrayed have been, in alphabetical order, Vittorio Bertoldi, Bernard Bloch, Ernst Gamillscheg, Georges Gougenheim, Bengt Hasselrot, Helmut Hatzfeld, Hayward Keniston, Jerzy Kurylowicz, Ramón Menéndez Pidal, Bruno Migliorini, Manfred Sandmann, Arnald Steiger, Knud Togeby, Max L. Wagner, Uriel Weinreich, Edwin . Williams, and Viktor Žirmunskij. The most ambitious flashback, preceded by a thumbnail sketch (RPh, XXVI:2 [1972], 337-341), has been to the era of Elise Richter in Vienna, in connection with the belated salvaging of some of the finest among her shorter and middle-sized articles. See the Introduction (pp. 9-12) and the Running Commentary (pp. 555582) to: E.R., Kleinere Schriften zur allgemeinen und romanischen Sprachwissen schaft (Innsbruck, 1977), as well as the critical reactions by H. Christmann, H. Kahane, and E. Pulgram, already available; as well as those by others, announc ed as forthcoming. Shorter pieces of this sort have been written by myself, over the years, on Adelung, Grimm, Hofmannsthal, Miklosich, Pott, the Schlegel brothers, Schuchardt, and Vising. Finally, it seems defensible to mention here three "Forschungsberichte" from my pen, despite their being more informational than truly analytical. One of these

SUPPLEMENT

563

appeared in Vol. IV: Ibero-American and Caribbean Linguistics (The Hague & Paris: Mouton, 1968) of Current Trends in Linguistics (ed. T.A. Sebeok): "Hispanic Philology" (158-228); the other two in Vol. IX of the same series: Linguistics in Western Europe (1972): "General Diachronic Linguistics" (82-118) and "Compara tive Romance Linguistics" (835-925). The first of these three chapters, or sections, was eventually equipped with a 34-page Supplement and various indices, and issued as a soft-cover book under a different title: Linguistics and Philology in Spanish America: A Survey (1925-1970), Ianua Linguarum, Series Minor, XCVII (The Hague & Pam: Mouton, 1972); it elicited various readily identifiable critical res ponses. If in all the items so far enumerated the accent has been preminently on the accomplishments of an individual, on the achievements of a school of thought, on the record of a sub-discipline geographically or chronologically delimited, and the like, it is obviously possible to start out, conversely, from a preëxistent concept (say, that of a phoneme) and to trace the successive or competing definitions offered for it across a given time span. This is the rival perspective I have aimed to furnish in my major article, originally the text of a Collitz Lecture (1980), "Drift, Slope, and Slant: Background of, and Variations upon, a Sapirian Theme" (the revised version is slated to appear in Lang., LVII [1981], 535-570). A much shorter com panion piece, centering about such key terms as "independent parallel development", "typological convergence", "multiple borrowing", "periodicity", "drag", is to appear independently in Romance Philology briefly thereafter.

"Range of Variation as a Clue to Dating" Absolute chronology, attainable through historico-philological techniques of investigation, has everywhere at all times been viewed, by scholars and scientists alike, as one of the principal goals of reconstructive inquiry. Where absolute dates cannot be directly established through analysis of archival data, attempts have been made to infer events, phases of growth, and the like from circumstantial evidence. One of the avowed goals of diachronic linguistics, e.g., has been to propose cogent sequences of shifts: Even if we cannot accurately date the processes X, Y, and Z, the mere demonstration of the fact that X must have preceded both Y and Z, and that Y must have followed upon X, while preceding Z, has been hailed as a major step forward. Appended to W. Meyer-Lübke's classic Historische Grammatik der französischen Sprache one finds, e.g., in Supplement II to Part I, a tentative temporal (but not strictly chronological) sequence of 19 sound changes, with liberal allowance for interverted, or otherwise alternative, patterns of successions of modifications (see 2d and 3d eds., 1913, pp. 266f.). From here Elise Richter derived her more ambitious — possibly too ambitious? — model for a complete presentation of Vulgar Latin and Proto-French phonology in terms of such temporal concatenations. Among the critics of her magnum opus, or rather of its torso, the Chronologische Phonetik des Französischen bis zum Ende des 8. Jahrhunderts (1934), E. Gamillscheg and H. Lausberg deserve to be singled out. For no good reason, except the puncturing of inflated hopes, this stimulating approach seems to have meanwhile been compie-

564

SUPPLEMENT

tely abandoned in Romance quarters. Menéndez Pidal's skepticism about such neat sequences (a doubt voiced in his Orígenes) is well known. The assumption of certain constants in the gradual attrition of core vocabularies has prompted various American, and a few European, anthropological linguists to operate with glottochronological and lexicostatistic techniques, the aim being this time the determination of the approximate moment when certain language families disintegrated into individual daughter languages. The exaggerated claims and ex pectations of this school of thought were exploded by severe criticism of skeptics, which included the argument reductio ad absurdum. Certain ingredients of this approach can presumably be salvaged by the small residue of practitioners still left to this once promising movement. My own article of the year 1968, illustrating a still different way of reasoning, was meant to be the first of a series of probings, but the pieces reserved for followup operations have, unfortunately, been left as mere drafts. I started out from the premise that in a speech community otherwise known for vigorously leveling the paradigms of verbs, the occasional clustering of variants in just one small part of the given total paradigm may have distinctive temporal implications: The unevenness is best explained on the assumption that the speech community involved simply has not had sufficient time to assert the power of its analogical machinery, which ordinarily wipes out — as has been known from independent observation — any such manifestations of exuberance, in deference to an intrinsic ideal of paradigmatic leanness. If that much is granted, there arises the probability of relative recency of events which have led to such unopposed luxuriant growth. Thus, it is striking that in Hispano-Romance the preterite of traer/'trazer 'to bring' (from ancestral trahere 'to carry, drag') should show such erratic diversity as Ptg. trouxe (influenced in its actual pronunciation by disse < dīx-ī -it 'said') beside OSp. traxe/trase and troxe/troše/ surviving at present as traje and dial. truje, res pectively. To these better-known forms must be added a phalanx of regional or extinct variants, e.g., treixe, trasque, tro(u)gue, alongside trouge, as well as trouve, and completely regularized traí, arrived at chiefly under pressure from the com pound retraí. The implication is that most of these variants, far from being traceable to colloquial Provincial Latin prototypes, are readily explained as due to the applic ation of analogical pressure, probably at a fairly late date: Thus, trouve echoes (h)ouve < habuit 'he had', trougue calls to mind jougue < iacuit 'he lay', trasque seems to be modeled on nasque 'he was born' (a modification of nāscit-ur 'he is being born'), etc. This analysis of flux seems to mesh well with the new crop of sociolinguists' steady concern with linguistic variation, which presumably goes back to the standardsetting article — by Uriel Weinreich, William Labov, and Marvin I. Herzog — "Em pirical Foundations for a Theory of Language Change", included in: Directions for Historìcal Linguistics: A Symposium, eds. WP. Lehmann and Y.M. (Austin & Lon don: Univ. of Texas Press, 1968 [actual date of oral delivery: 1967]), pp. 95-195. Among further experiments of my own geared to the exploration of such clues to dating as the erratic clustering of variants in a verbal paradigm is an (unfinished) inquiry into the syncopated future tense in Old Spanish.

SUPPLEMENT

565

"Factors in the Unity of ROMANIA " This unpretentious note embodies a half-hour invitational talk before the (by now dissolved) Romance language-and-literature section of the Modern Language Association of America, a meeting presided over by Aldo D. Scaglione - with Anna Balakian and the late Helmut A. Hatzfeld acting as the other members of the panel and representing the literary side of a complex and involved relationship. Several points touched upon in this causerie barely ten years or so ago would today invite rather vigorous rephrasing. Thus, a propos the split of Romance into a Western and an Eastern wing, it is no longer certain whether Wartburg's classification championed in the early 'thirties still ranks as "Establishmentarian", with so many and such important features of Hispano-Romance (resistance to voicing of inter vocalic surds, widespread preservation of parental ĭ and, expecially, of  as і and  rather than e and o, respectively, and the like) militating for an entirely different split. The distinctly longer span of the linguistic than of the literary "Romance era" brings with it that, unlike students of, say, Old French or Old Provençal or Old Spanish literature, who are essentially unreformed medievalists, practitioners of Romance linguistics lean more and more toward closer self-identification with students of Classical Antiquity.

"Review Article: Lexical Borrowings in the Romance Languages " The phenomenon of inter-Romance borrowing was already familiar to pioneers of the 17th and 18th centuries (G. Ménage in France and T.A. Sánchez in Spain, to limit myself to two names), but the full measure of its extent dawned upon linguists not before the end of the 19th; to realize the magnitude of this leap it suffices to compare a few specimens drawn from the last revised comparative ety mological dictionary by F. Diez with equivalents in both versions (1911-20, 193035) of W. Meyer-Lübke's counterpart. Most of the necessary spadework has been done — historically-slanted dictionaries apart — in the form of etymological notes and articles of varying size and degree of sophistication. There also arose, before long, a special genre of monographic investigation (embodying not infrequently a plodding doctoral thesis) bearing on, say, Gallicisms in Portuguese; not all such dissertations have seen the light of day (e.g., D.J. Pratola's Berkeley venture on Italianisms in Portuguese has not, hence must be consulted in situ). The traditional center of this research activity has been Central Europe, and the vehicle of communication has, as a rule, been German — even though J.H. Terlingen's piece (Amsterdam, 1943) dealing with Italianisms in Spanish absorbed until the start of the 17th century, commendably enough, was written in Spanish. English, until recently, was rarely appealed to as a medium of exposition for this sort of study, and the output was meager in yet another sense: De Forest's Yale disser tation, directed by Henry R. Lang, on early Gallicisms in Spanish, rather typically, was so slender that it could easily be fitted into an extended journal article (1918). To be sure, Fraser MacKenzie, a widely-traveled New Zealander, by 1939 presented

566

SUPPLEMENT

a bidirectional two-volume study on migratory words (Wanderwörter); but its two focal points involved one Romance and one non-Romance culture, namely French and English; see Frank R. Hamlin's 1980 necrology in RPh, XXXIV: 1,139-141. .. Hope's two-volume monograph (1971) on Italianisms in French and Galli cisms in Italian — the outgrowth of a, largely, self-directed British dissertation — came as a major surprise, not least on account of the big patch of ground (800 years) it aimed to cover — and fully deserved a review article. Its title, Lexical Borrowing in the Romance Languages, was infelicitous, though, because Hope seldom, if ever, transcended the chosen domain for narrow-gauged concrete demonstrations — so that any perspectives on Portuguese, Spanish, Catalan, und Rumanian remained virtually absent from his purview. As if by compensation, he did take into account, at least in the concluding section of his book, the involvement of general linguistics, familiarizing himself — probably at the height of the final revision of his monograph — with the writings of, above all, Einar Haugen and Uriel Weinreich in this country, all of which circumstances, I felt, justified the rather unusual length of my own reaction in a journai serving the needs, primarily, of generalists. It might be otiose to enumerate here my own, rather scattered, studies focusing on, or touching upon, sundry processes of lexical borrowing involving one, two, or more Romance languages. Suffice it to mention two ֊ mutually complementary — papers following Arabic hatta 'until' on its way into Ptg. até, OSp. (f)a(s)ta, mod. Msta, on account of the various vicissitudes attending upon that peculiar borrowing process: (a) "Español antiguo des(de), ha(s)ta, fazia y fascas", Homenaje a Julio Caro Baroja (Madrid, 1978), pp. 711-733, and (b) "Studies in the Diachronic Differ entiation of Near-Homophones", Lang, LV (1979), 1-36. Also, in my typological studies of lexicography, I busied myself with the various facets of the problem that both objectively-tuned dictionaries of borrowings and militantly-orchestrated vocabularies of (rejected) foreignisms tend to pose, with appropriate bibliographic hints: (a) "Distinctive Features in Lexicography: A Typo logical Approach to Dictionaries Exemplified with Spanish", RPh, XII: 4 (1959), 366-399, and XIII: 2 (1959), 111-155, at 114f., 121f.; (b) "A Typological Classi fication of Dictionaries on the Basis of Distinctive Features", in: Problems in Lexico graphy, edd. F.W. Householder and S. Saporta, UAL, XXXVIII:2:4.3-24 (2d ed., 1967), absorbed into my Essays on Linguistic Themes (Oxford, 1968), pp. 257279, and reviewed by Karl D. Uitti in RPh, XVI:4 (1963), 417419; plus (c) Ety mological Dictionaries: A Tentative Typology (Chicago University Press, 1976), pp. 29-32. Finally, some of the research I have editorially supervised or academically di rected is concerned with the borrowing of affixes, particularly Anita Katz Levy's monograph-length article, "Contrastive Development [s] in Hispano-Romance of Borrowed Gallo-Romance Suffixes", RPh, XVIII:4 (1965), 399429; XX:3 (1967), 296-320; and Suzanne Fleischman's doctoral dissertation, Cultural and Linguistic Factors in Word Formation: An Integrated Approach to the Development of the Suffix "-age", University of California Publications in Linguistics, LXXXVI (1977); see Curtis Blaylock's penetrating review in RPh, XXXIV: 1 (1980), 100-104. A final synthesis of all these gropings remains to be provided.

SUPPLEMENT

567

"Gender, Sex, and Size, as Reflected in the Romance Languages " Like some of the other items included in this collection, this piece was originally devised for oral delivery — as an invitational lecture at the University of Michigan Romance Linguistics Symposium. The lecture was repeated, under a slightly dif ferent title, at the University of Pennsylvania soon thereafter. Although I eventually was given a chance (and availed myself of it) to add a — chronologically arranged — list of bibliographic references (pp. 172-175) as well as a string of footnotes, the basic format, and in particular the length and the tone of the whole venture were determined by its prime purpose. Not included in the definitive version were the questions directed at the speaker from the floor — including some by Ernst Pulgram, who presided over the session — and the brief clarifications supplied, at the Ann Arbor affair; the same holds for the Philadelphia sequel. This essay shows a limited affinity with other writings of my own that are con cerned with the linguistic correlates of sex and with the morpho-semantic category of gender, including: (a) "Ancient Hispanic vera(s) and mentira(s): A Study in Lexical Polarization", RPh, VI (1952-53), 121-172; (b) "Diachronic Hypercharacterization in Romance", ArL, IX (1957), 79-113; X (1958), 1-36; (c) "The Luso-Hispanic Descendants of põtiõ; a Study in Lexical Proliferation", in Hispanic Studies in Honour of I. González-Llubera, ed. F. Pierce (Oxford, 1959), pp. 193-210; and (d) "Hypercharacterization of the Pronominal Gender in Romance", forthcoming in: Language, Meaning, and Style: Essays in Memory of Stephen Ullmann. The short piece cryptically alluded to at the bottom of p. 175 can now be safely identified as my contribution to Studia Gratulatoria Robert A. Hall (Madrid, 1979), pp. 191196: "The Analysis of Lexical Doublets". There are two major cross-linguistic domains that interlock with the narrower Romance field here principally investigated (gender attached to names of tools, containers, and the like): (a) the explicit marking of names of female agents, by affixation (Fr. directrice, Sp. directora, E. governess) or composition (R. zenscina -vrac, E. woman doctor), often in rivalry with names of wives of male agents or title-holders (L. rēgina, R. tsaritsa, E. duchess); and (b) the capricious marking of female gender (e.g., by appropriate endings) in hypocoristics, as when, in the inti macy of a family circle, a little girl otherwise known, in Russian, as Asja or Vasja emerges, half-playfully, as As] or, even more striking, Vas, respectively; cf. Fr. Louison, Marion. On both aggregates of problems there have gradually accumulated whole bodies of significant writings. On the former, e.g., the exploration by Kathleen Connors (directed by me at Berkeley), "Studies in Feminine Agentives in Selected European Languages", RPh, XXIV:4 (1971), 573-598, retains its full usefulness, and there was later superadded an even more fine-meshed inspection of the Old French terrain in a solid Supplement volume to the ZRPh; on the latter, there exists a rapidly-growing corpus of research in child language. At proof, let me cite R. Stefanini's article, AGI, LXV (1980), 41-73. If the discussion launched by Wartburg, Hassehot, Toole-Kahane, etc., is to pick up momentum once more in the future, it can be energized only through combi nation of these two reservoirs of strength: (a) continued analysis of the consequences of the decline of the Latin neuter gender at the Romance stage; (b) independent

568

SUPPLEMENT

harnessing of the flow of data supplied by general linguistics. In the past, these two potential sources of information, as a rule, were not clearly distinguished. "The Social Matrix of Palaeo-Romance Postverbal Nouns" My lengthy paper "Between Monogenesis and Polygenesis", presented at the Third International Conference on Historical Linguistics (Hamburg, August 1977) and due to be included in a transactions volume now in press (Amsterdam: Benja mins), concerns itself with the absorption of several (reconstructed) Germanic verbs into Provincial Latin (including *wardon 'to observe', *warnjan 'to warn', *warôn 'to watch, preserve', *warjan 'to ward off), besides discussing the source of Fr. gagner (orig. gaaignier), Ptg.ganhar, It. guadagnare 'to win, gain' (in contrast to Sp. ganar); to that extent, it supplements the section of the older article included here (pp. 204-207), with respect to It. guida, Sp. guia, OFr. guie; also Fr. garde, Sp. guard(i)a; Hisp.-Ar. atalaya, etc. More immediately relevant, all told, is the article by Joanne Martin Baldonado, "Affixation and Gender Desinence in the Old Spanish Postverbal Nouns", which, entering into a complexly architectured piece by J.M.., Andrew S. Allen, and myself: "New Studies in Romance Parasynthetic Derivation" (forthcoming in the August 1981 issue of RPh: "Henry R. Lang Memorial - Frank M. Chambers Testi monial", Vol. XXXV:1), pushes forward by a sizable margin the predictability of the selection of (m.) -o as against (f.) -a. The author accomplishes this by breaking down the total of verbs qualifying for this process into several neatly contoured groups, each of which is apt to develop a separate pattern: (1) From (actually or seemingly) simple verbs, such as buscar 'to seek, search', one can expect the abstract busca; (2) from verbs with "strengthened" suffixes (like -ear, -antar, -entar) one can, by the same token, foresee derivatives in ֊o, witness crebanto, rodeo ֊ pelea 'fight' being the one prominent exception; (3) from prefixed verbs, including bor rowings, deverbatives, and parasyntheta, a masculine abstract in -o will, speaking probabilistically, be preferred (abondo); (4) the same holds for compounds of prefixed verbs (des-a-fuero); etc. This promising, richly documented analysis forms part of the author's gradually maturing Berkeley doctoral dissertation, which transcends by a wide margin the distinctly older research on postverbals conducted in the Old French and the Old Italian fields. "Each Word Has a History of its Own" The circumstances under which this essay arose, as a contribution to a 1964 symposium, are pointed out in the starred footnote relegated to the bottom of p. 226. It happens to be one of my very few writings traceable, in the last analysis, to classroom experience, since L. Bloomfield's Language, some sections of which are here subjected to minute analysis, has been used, over the years, as a textbook, first in my graduate course at Berkeley on Old Spanish and later in my upper-division introductory course to historical linguistics.

SUPPLEMENT

569

It is fairly clear, to those familiar with his style, that Bloomfield himself did not feel very comfortable with his tortured presentation, of what essentially should have gone under some such title as "Etymology", under rather bizarre, not to say evasive, chapter headings, such as "Fluctuation in the Frequency of Forms" (392-403) and "Intimate Borrowing" (461-475). From the mid 'fifties to the early 'eighties I have, again and again, drawn readers' attention to the centrality of the issue of "individual lexical growth", which is what the object of etymology has at this point at long last come to be (over against its earlier definition as "identification of word origins") within the context of rapidly changing linguistic (specifically, grammatical) theory. Some of the stepping stones for էհե particular line of curiosity have so far been: (a) "The Place of Etymology in Linguistic Research", in BHS, XXX (1954), 78-90, originally an Indiana Univer sity lecture; (b) "Etymology and the Structure of Word Families", Word, X(1954), 265-274, a journal volume also issued as a book venture: Linguistics Today, edd. A. Martinet and U. Weinreich (N.Y.C.); () "Etymology and Historical Grammar", RPh, VIII:3 (1955), 187-208 ֊ also available in Italian as a chapter of Linguistica generale, filologia romanza, etimologia (Bologna, 1970), pp. 132-145 (a consolida tion of two Indiana University lectures delivered in 1953); (d) "Weak Phonetic Change, Spontaneous Sound Shift, Lexical Contamination", Lingua, XI (1962), 263-275 (=Studia Gratulatoria Dedicated to Albert Willem de Groot), absorbed into my Essays on Linguistic Themes (Oxford, 1968), pp. 33-45, and included, in Italian garb, in Tristano Bolelli's reader, Linguistica generale, strutturalismo, linguistica storica . . . (Pisa: Nistri-Lischi, 1971) pp. 247-260; (e) "Etymology and General Linguistics", Word, XVIII (1962), 198-219 (=Linguistic Essays on the Occasion of the Ninth International Congress of Linguists [in Cambridge, MA], also issued as a separate book), included in both Essays on Linguistic Themes (1968), pp. 175198, and Linguìstica generale, . . . (1970), pp. 39-66; (f) "Primary, Secondary, and Tertiary Etymologies: The Three Lexical Kernels of Hispanic saña, ensañar, sañudo", HR, XLII (1974), 1-32; (g) "Deux catégories d'étymologies 'intéressantes'," RLiR, XXXIX (1975), 255-295; (h) "Etymology and Modern Linguistics", Lingua, XXXVI (1975), 101-120 (originally a 1974 lecture delivered at Leiden University; also at North Carolina [Chapel Hill], Virginia [Charlottesville], Jerusalem [Givath Ram], Copenhagen, and Uppsala); (i) "Lexis and Grammar — Necrological Essay on Emile Benveniste (1902-76)", RPh, XXXIV:2 (1980), 160-194 (expanded and polished version of a lecture given at Princeton on October 19, 1977, and repeated at Berkeley, under the sponsorship of the Study Group in Historical Linguistics, on March 8, 1978); (j) "Etymology as a Challenge to Phonology: The Case of Ro mance Linguistics", Lautgeschichte und Etymologie: Akten der VI. Fachtagung der Indogermanischen Gesellschaft, Wien, Sept. 1978 (Wiesbaden: L. Reichert Verlag, 1980), pp. 260-286. There have been several critical reactions to "Each Word . . . ", ranging from Rebecca Posner's in "Thirty Years On", pp. 445f. (appended to the 2d ed. [Oxford: Blackwell, 1970] of I. Iordan and J. Orr, An Introduction to Romance Linguistics: Its Schools and Scholars) all the way to William Labov's in the latest (June 1981) issue of Language: "Resolving the Neogrammarian Controversy" (LVE, 267-308). See also J.R. Craddock, S.N. Dworkin, and C. Poghirc, "Romance Etymology",

570

SUPPLEMENT

in R. Posner and J.N. Green, edd., Trends in Romance Linguistics and Etymology, I (The Hague, etc.: Mouton, 1980), 181, 220f.

"Multi-Conditioned Sound Change and the Impact of Morphology on Phonology " This is, in essence, a leaner, more nuanced, and better balanced presentation — eight years later — of ideas offered "in the raw", while they were still jelling, at a symposium sponsored by The University of Texas in April 1966 — a colloquium programmatically concerned with some of the central theoretical topics of diachronic linguistics. The experiment led to a book venture, with a paper-back edition eventual ly allowed to flank the original hard-cover edition: Directions for Historical Lin guistics: A Symposium, edd. W .P. Lehmann and YJVl. (Austin & London, 1968), in which my authorial share (pp. 21-64) was the slightly revised text of the earlier conference paper: "The Inflectional Paradigm as an Occasional Determinant of Sound Change". There exists an Italian version of the entire book, thanks to Ruggero Stefanini: Nuove tendenze della linguistica storica (Bologna: il Mulino, 1977), with a Preface by Luigi Heilmann, a version in which my own piece (pp. 33-75) appears with two post-scripts (dated 1971-72 and 1976). Midway stops between the roughhewn 1966-68 item and the 1976 article in Language here reprinted have been several scattered items of varying length and weight, including one contribution couched in French: "Le nivellement morphologique comme point de départ d'une loi phonétique: La monophtongaison occasionnelle de ie et ue en ancien espagnol", in Mélanges . . . Jean Frappier (Genève: Droz, 1970), pp. 701-735; and "Morpho logical Analogy as a Stimulus for Sound Change", Lingua e stile, IV (1969), 305327, essentially a balance sheet of earlier findings, with a minimum of revision. During the early years of experimentation along this line (1966-70), І incline to assume that, in the case of certain sound changes — to be sure, "regular", but atypical or otherwise difficult to account for in strictly phonological diachronic terms, the virtually exclusive motivation for the shift might have to be sought in a peculiar constellation of morphological conditions; in the last analysis, i.e., in a sort of "analogical" pressure. The first two instances of such suspected concatena tion of events were (a) the change of Lat. -lģ-, -nģ-, and -rģ- (i.e., of clusters which raise no puzzling problems in most Romance languages, including near-by Portu guese as well as territorially more distant French and Italian) into OSp. -Iz-, -nz(beside -nn- /ñ/), and, especially, -rz-, as in gingiva 'gum' > OSp. enzia; and (b) the reduction of the OSp. rising diphthong ie /je/ to /i/ in certain environments, e.g. in the vicinity of /λ/, /1/, /r/, /R/, /p/, /s/, and, a fortiori, of any combination of these phonemes, accompanied by the later and discernibly rarer — also, typologically, less than strictly parallel ֊ reduction of ue /we/ to /e/, as in OSp. priessa 'haste' > mod. prisa and, by the same token, in fruente 'forehead' > frente. In both situations the number of primitives involved was fairly limited, as against the pro fusion of instances of -ietto > -illo, and the ratio of "exceptions" from an except ional treatment! embarrassingly high (e.g., fiesta 'festival, feasť). For the *-lģ- > -lzchange and its analogues (a case history reconstructed most painstakingly in the

SUPPLEMENT

571

Texas Symposium paper), it was argued that the morphophonemic alteration pre vailing in the paradigms of certain key verbs (digo, dizes, . .. dizen, diga) may have offered a model suitable as a starting point. For the reduction of ie to і (and, with the aforementioned delay and diminution of incidence, of ue to e) it was arguable that the joint pressure of the hypocoristic suffixes -ico, -ito, -in(o), presided over by the high front vowel, could have prompted speakers to prefer -ilio to older and etymologically "correct" -ietto < Lat. -ellu, particularly since the pioneering role of -i(e)llo words had been independently diagnosed by Menéndez Pidal, through use of a microscopic lens trained on datable charters. (This point was developed most fully in my contribution to the Frappier Testimonial.) In the early 'seventies my attention was drawn to situations where analogical pressure was, probably, best characterized as a concomitant rather than prime mover of a conspicuous sound change of limited lexical penetration. Thus, the hazard of highly atypical syncopated future and conditional tenses in verbs such as It. chiedere (from quaerere), conquidere (from conquirere), fiedere (from ferire՝) could, at a certain juncture, be averted through substitution of (analogically arrived-at) -dere for "regular" -rere, and the intermediate situation of prolonged wavering might very well have stimulated dissimilatory processes (r . . . r > d . .. r and even r . . . d) outside the realm of verbs; see my "Etiological Studies in Romance Dia chronic Phonology", ALH, XIY:2 (1973 [-74]), 201-242, at 214-225. Similarly, the tendential change in Middle French of -r- to -s-, as residually preserved in chaire ( < cathedra, cf. E. chair) to chaise, except in highly specialized semantic niches, may very well be a mere reverberation of the collision of two derivational suffixes: (a) -eux, -euse, from -õsu, -õsa; and (b) -eu(r), anal, -euse, from -ore; see "Deux frontières entre la phonologie et la morphologie en diachronie", in Langages, No. 32, Dec. 1973: Le changement linguistique, pp. 79-87. With this liberalized view of an interaction, or enmeshment, of forces I reverted to the field of Old Spanish problems left fallow for a while, except that there had meanwhile appeared several critical reactions, especially to the ground-breaking 1966-68 study ֊ some positive, others reserved, yet others partly or wholly negative. Cf. J.R. Craddock's appraisal, noted for its astomshing lack of empathy, in Lang., XLVI (1970), 688-695; and, among more temperate responses, those by M.B. Fon tanella de Weinberg, Fil., XV (1972), 318-327, at 321-324; Giulio Lepschy, JL, VI (1970), 136f.; R. Posner, RPh, XXIII:2 (1965), 143-153, at 146f., 149-151; O. Szemerényi, General Linguistics, X:2 (1970), 121-132, at 123f.; . Togeby, Revue romane, V:2 (1970), 257-260;K.D. Uitti, RF, LXXXI (1969), 7-9. The new way of thinking, which crystallized in the mid and late 'seventies, stresses the interplay of factors, not all of which need be morphological. "Multi-conditioned Sound Change . . . ", which concentrates on the issue "rising diphthong" vs. "secon dary monophthong", innovated by invoking the late-medieval speakers' wavering between certain sets of preterital endings: -iemos ~ imos, -ies-te(s) ~ -iste(s) (with the -i- forms destined to win out) as one of the reasons why, say, aviespa < vespa (+apicula 'little bee') tended to become avispa; and, in the process, identifies the hitherto misunderstood reason as to why -s-, -sp-, and, in its wake, -st- so strongly favor monophthongization (on the conflicts in the ranks of the OSp. preterit see HR, XXVI [1959 = Joseph E. Gillet Memorial Volume], 435-481). Another

572

SUPPLEMENT

conditioning factor turned out to be the encroachment of metaphonic on diphthong izing verbs, as when sigo Ί follow', sirvo 'I serve' (echoing pido 'I ask') dislodged archaic siego and siervo, from seguir and servir, respectively. (On this long-drawnout process see Lang., XLII [1966], 430472), while the noun siervo 'serf, slave' remained intact. Then again, suffix rivalry in the ranks of -uero and -ero, -ueño and -eño derivatives accelerated the switch from, say, afruenta to afrenta 'offense'; occasional substitution of -iembre for ֊imbre, as in urdi(e)mbre 'warp', fig. 'scheme', now emerges as an instance of hypercorrection; see RPh, XXI: 1 (1967), 41. A comparable measure of smoothing-over, applied to the rise of ֊lz- from -lģ-, etc., marks a paper scheduled to appear in HR, "Interplay of Sounds and Forms in the Shaping of Three Old Spanish Medial Consonant Clusters". The leitmotiv of that article, in rather pronounced contrast to that of its 1966-68 predecessor, is the thought that -Iz-, -nz-, -rz- actually preexisted in Old Spanish, as so many outcomes of syncopated -l'ć-, -n'ć-, and -r'č- (cf. OSp. anzuelo 'allurement', lit. 'little hook', enzina 'holm oak, evergreen oak', manzera 'handle of the plough', manzilla 'blot, blemish', etc.) before being apt to qualify for their secondary role as substitutes for *-lz-, etc. In this context it turned out to be advantageous to realize that inde pendent proof of the speakers' of Proto-Spanish relative aversion to the /ž/ phoneme, in any environment, had previously been proffered. See my note: "In Search of Penultimate Causes of Language Change: Studies in the Avoidance of /ž/ in ProtoSpanish", Current Studies in Romance Linguistics, edd. Marta Luján and Fritz Hensey (Washington: Georgetown University Press, 1976), pp. 27-36. Clearly, where any irreducibly phonological nature of processes which in super ficial view seem to involve little more than ordinary sound changes is expressly denied extra caution must be exercised in regard to etymological equations underlying all genetic relationships invoked; hence the wisdom of observing under a microscope all known or suspected instances of ue > e (see pp. 260-273 of the abovecited 1980 paper, "Etymology as a Challenge to Phonology"). Then again, under such risky circumstances heightened attention to the minutest details of the sur rounding phonological processes remains mandatory. Even if one recognizes the agency of identifiable morphological (analogical) factors behind the specific config uration of rising diphthongs in late Old Spanish (ie > i, ue > e), it is nevertheless true that these transmutations enter into a much wider ensemble of broader changes, predominantly phonological, which tended to weaken or dilute the — once very neatly profiled — status of these diphthongs across the board in late medieval Spanish; see the following mutually complementary studies of very recent vintage: "The Fluctuating Intensity of a 'Sound Law': Some Vicissitudes of Latin e and o in Spanish", RPh, XXXIV: 1 (1980), 48-63; and "The Abandonment of the Root Diphthong in the Paradigms of Certain Spanish Verbs", Incontri linguistici, V (1979 [-80]: Miscellanea Vittore Pisani), 123-138. In these most advanced outposts atten tion has been drawn to such side-issues as the dialectal reduction of ue to u, a close parallel indeed to ie > z'; and a morphophonemic explanation has been proposed for the eventual rejection, by the vast majority of speakers, of such nascent para digms as aprebo ~ aprobar 'to approve' or remilgo~*remelgar (refl.) 'to be prim and finicky' — in favor of remilgo~remilgar — for the sake of morphophonemic economy.

SUPPLEMENT

573

A promising side-line of this tapestry of investigations stands apart by making appeal to the instrumentality of phonosymbolism, in conjunction with affixation; conceivably the recently proposed tag "morphosymbolism" would here be more appropriate than the traditional label "phonosymbolism" (see my discussion of these tags in "From Phonosymbolism to Morphosymbolism", Fourth LACUS Forum [Columbia, S.C. 1978]). The particular problem at hand, which has for a century — since the days of A. Horning — plagued Romanists is the emergence in a whole cluster of Old Spanish derivational suffixes of a voiced affricate z where one would, on straight phonological evidence, have expected  ҫ (its voiceless counterpart) to make its appearance instead. (Portuguese, in most of these cases, unexcitingly enough, has at all times displayed  c.) We have, then, before us a chain of unex plained voicings restricted, grosso modo, to suffixal contexts, such as -azo < -ãceu beside -izo < -īceu, -eza < -itia, verbs not only in -izar < -idiāre but also in -uzar < -ūtiāre, such as aguzar 'to sharpen', and the like. Hypocoristic and pejorative (comic) suffixes are significantly excluded from this special development. There appears to exist no strong alternative to the assumption — made after decades of rather futile debates and controversies about the allegedly self-contra dictory rules governing the rise of Old Spanish sibilants ֊ that, at a crucial juncture, the Proto-Spanish community adopted the semantic marking — through voicing of the affricated sibilant acting as the pillar or anchor of the entire affix — of "noble suffixes" (relational, abstract, purely verbal), so as to set them off from the larger number of "cruder suffixes". For the advocacy of this idea, see my piece, "Deri vational Trar parency as an Occasional Co-Determinant of Sound Change: A New Causal Ingredient in the Distribution of -c- and -z- in Ancient Hispano-Romance", RPh, XXV:1 (1971), 1-52. Peripheral to the main thrust of the ideas here expounded has been the sporadic exploration of more limited issues in which they intervene. The tendency to change syllable- (including word-) final s to /θ/ spelled z is familiar from such instances as bizcocho, biznieto, lápiz; this trend may go back to the replacement of pres. ind. -esco, pres. subj. -esca by -ezco, -ezca in inchoative verbs, on the understanding that -ezco involves a blend of, or a compromise between, -esco and -ece(s), etc., thus exemplifying Schuchardt's classical formula "innere Formenmischung". See my note, "The Case of Old Spanish /sk/ Changing to /θk/", RPh, XXIII:2 (1969), 188-200. — The record of a stunted (or thwarted) sound change comes up for discussion in a slightly later note, "Quelques avatars romans d'un zoonyme et d'un ornithonyme latins", Etudes . . . Félix Lecoy (Paris: Champion, 1973), pp. 377384: In such meagerly represented developments the role of each lexical carrier of the given shift, plus the formal and/or semantic connection between such carriers, acquire extra weight. — "Old Spanish maraviella 'Marvel', Late Old Spanish Sierta 'Syrtis'," RPh, XXXIII:4 (1980), 509f., is a comment on two instances of over extension of -ie- at the cost of -i-, and thus supplements the earlier remarks on urdi(e)mbre and the like. - "Points of Abutment of Morphology on Phonology: The Case of Archaic Spanish esti(e)do 'stood'," in RPh, XXXIV:2 (1980), 206209, involves a cautious suggestion that the move from OSp. 3d sg. estiedo 'stood' < stětit (the predecessor of analogical estovo, later estuvo) to estido, under pressure from (metaphonized) 1st sg. estide < stětī, could have had a lubricating effect on

574

SUPPLEMENT

OSp. sieglo 'world' > siglo 'century', or on the extraction from OSp. (res)crieço 'crack, chink' of mod. quicio 'pivot hole, door jamb'. No appraisal or dissection of the ensemble of these studies has so far come to my attention; but Ana María Barrenechea came close to offering one, with her usual finesse, at least for the early crop in RPh, XXV:3 (1972), 325f., while Carlos-Pelegrin Otero, in his reaction, acted as the spokesman for orthodox generative-trans formational dogmatism: Evolución y revolución en romance: minima introducción a la fonología (Barcelona: Seix Barral, 1971), pp. 146-152. As Roger Wright has remarked in his pungent BHS review of the revised 8th edition (Madrid: Gredos, 1980) of R. Lapesa's authoritative Historia de la lengua española, one regrets that scholar's reluctance (passim) to bring to bear his range of experience and erudition on front-line debates. Conversely, one of the most sensitive and constructive apprai sals that have caught my attention has just come from Julius Purczinsky, "Romance Historical Phonology", in R. Posner and J.N. Green, edd. Trends in Romance Lin guistics and Philology, I (The Hague, etc.: Mouton, 1980), 99-101. In self-criticism, I am tempted to observe that the emphasis henceforth might well be placed more heavily on the analogical rather than on the morphological ingredient of the chief conditioning factor. The main reason for this qualification is that one is, conceivably, apt to encounter isolated contexts in which straight phonological analogy was operative. One such state of affairs, which is still under scrutiny, may be tentatively described thus. It is notoriously difficult to account for Classical ě's and ŏ's resistance to diphthongization in Old Castilian (though neither in Old Leonese nor in Old Aragonese) before certain consonants either palatal, or on their way to becoming palatal, in Provincial Latin or Proto-Spanish; but paradoxically, not before all such consonants. Thus, spĕculu 'mirror' yielded espejo; from cordŏliu 'sorrow at heart, grief one arrives at cordojo; despolio 'I rob, plunder' underlies despojo, etc. The main diffi culty consists in this: A following palatal consonant, far from blocking diphthongization in Old Provençal, made it optionally possible, while in an archaic layer of the Old French lexicon one stumbles across filiations such as tĕrtiu 'third' > tierç (mod. tiers). How can this pattern of growth be made compatible with local refusal, across the Pyrenees, of say, folia 'leaves' to allow a ue diphthong to sprout (witness OSp. foja, mod. hoja 'leaf)? One way not yet tried out to circumnavigate the obstacle would be to start out from words such as régula 'straight piece of wood, rule(r)' vis-à-vis rĕgō -ĕre 'to keep straight, rule, govern', or teguk 'roof tile' vis-à-vis tĕgō -ěre 'to cover'. In a backwoods variety of Provincial Latin, like the one that plausibly underlies Old Spanish, the speakers' feeling for a long vowel preceding -ula and, by extension, -ulu (note that both regula and tegula, as lexical units, have been preserved unimpaired in Hispano-Romance) could have spilled over to include ŏculu 'eye', henceforth pro nounced *oculu, and, I repeat, spëculu, altered into *speculu, hence OSp. ojo 'eye' and espejo, respectively, via inferred */oλu/, *espeλu/, alongside foja, via /foλa/. Cuscŏliu 'scarlet berry of the holm oak', cordoliu, and despoliō chimed in (OSp. coscojo, cordojo, despojo). It now becomes understandable why /ñ/, for once parting company with /λ/, failed to have any blocking effect on diphthongization, witness luene 'far away' < longe (where luengo 'long' < longu could but needed not have

SUPPLEMENT

575

reinforced the diphthong), as well as sueño1 'sleep' < somnu and sueño2 'dream' < somniu. Note that past authorities (F. Hanssen, R. Menéndez Pidal) were unable to explain the anomaly. Any objection to this reconstructed chain of events, to the effect that /λ/ might before long have been followed by /ñ/ in lengthening the preceding vowel and thus, in the long run, blocking the diphthongizations, can be parried with the remark that this primary /λ/ of Hispano-Romance rapidly yielded ground to /ž/ — hence ojo /ožo/, etc. — and that /ž/ obviously could not have served as a behavioral model for /n/. Independently, the p. ptc. tēctu 'covered' could in this milieu have induced speakers to switch from lectu 'bed' to *lēctu, hence Sp. techo 'roof, lecho.

"Multiple Versus Simple Causation in Linguistic Change" This paper — its size predetermined by the configuration of the testimonial volume into which it had to be fitted — has several point of contact with two other articles absorbed into the present miscellany: (a) the presidential address, "Linguis tics as a Genetic Science" (pp. 2345), which pertains approximately to the same period, and (b) the distinctly later "Multi-Conditioned Sound Change and the Im pact of Morphology on Phonology" (pp. 229-250), a situation which will save us the need to repeat here certain matters already expounded above. At the same time, the core section of this paper, dealing with Spanish adjectives in Հօ, also antici pates certain ideas which were to bear fruit as many as fifteen years or so later, and thus invites comments from an angle not yet readily foreseen in the mid 'sixties. One cannot invoke a legitimate derivational suffix in discussing the word-final segment (m. sg.) Հօ of numerous Spanish adjectives, because they constitute, or at least give the impression of being, primitives. What loomed particularly important in 1967 was the fact that the words containing Հօ were only in a minority of cases inherited from transparent prototypes (say, Lat. -ǐdu), as is true of lacio, limpio, lucio, dial. nidio, rancio, recio, dial, rocio, sabio, sucio, tibio, turbio. Most -io adjec tives that clustered around this kernel owed this characteristic segment to wholly different antecedents: learned transmission of ancestral -iu (espurio, eximio, medio, nimio, obvio); confusion of the adjective itself with its equally widespread satel lite, the abstract (as when soberbio, orig. sobervio, reminds the analyst more of supěrbia 'pride' than of supěrbu 'proud'); contamination by the adjectival verb (as when agro 'sour', under the impact of agriar 'to sour', developed into agrio), or by the given adjective's, or corresponding adverb's, comparative degree (witness the use of amplio, suggested by the adv. amplius, in lieu of expected *amplo), etc. The constant interplay of these coefficients, directed toward the spread of a favorite type, was meant to illustrate multiple causation, even though each separate instance also lent itself to an individual explanation, invariably less effective through pro portionately weaker integration with the mainstream of events. Given Roman Jakobson's championship of the ingredient of teleology in language history, this slant of a study seemed to vindicate its inclusion in a series of volumes dedicated to him. As suggested above in passing, the ranks of Հօ adjectives highly characteristic

576

SUPPLEMENT

of Old, and even more so of contemporary, Spanish, pose yet another problem, namely the fitness of certain phonemically (including syllabico-accentually) charac terized forms to perform specific services — defined in part grammatically, in part semantically. My earliest encounter with this issue was in conjunction (a) with Spanish adjectives exhibiting certain nuclear vowels (especially e and o); (b) with the reiteration of the syllable-initial consonant; (c) with the given word's concur rent confinement to two or, at most, three syllables; and (d) with the stress falling on the penultimate, as in Sp. lelo or memo 'stupid, dull'; see "Fuentes indígenas y exóticas de los sustantivos y adjetivos verbales en -e", II, RLiR, XXIV (1960), 201-253. From there I advanced, past the contribution to the Jakobson Festschrift here under scrutiny, to a critique of R. Menéndez Pidal's celebrated article (1905) on so-called unstressed suffixes in Hispano-Romance, as in casc-ara'rind', murciégalo (later murciélago) 'bat', Ht. 'blind mouse', arguing that the intercalary segments -ar-, -al-, etc. were not, strictly, suffixes at all, but instead augments or excrescences enabling speakers to arrive at a favorite (sometimes semantically motivated) syllabico-accentual pattern; see "The Rise of the Nominal Augments in Romance: Graeco-Latin and Tuscan Clues to the Prehistory of Hispano-Romance", RPh, XXVI2 (1972), 306-324. With these three separate forays into uncharted territory to fall back upon, I have tried to apply the same approach to a variety of other languages (sailing in the wake of F. de Saussure, so far as Latin adjectives with nuclear -a- denotative of physical defects and sicknesses are concerned); see my forthcoming major piece, "Semantically Marked Root-Morphemes as a Problem of Diachronic Morphology", in which the Hon's share of attention is given to English colloquial (in part racy) adjectives of the type flimsy, nasty, sassy, snazzy, tizzy, which I interpret as an innovative type, despite undeniable links to older -y adjec tives traceable to Germanic (= G. -ig) or Anglo-Norman (as in petty, puny). This analysis, which regards such words as indivisible wholes, clashes with H. Marchand's more conservative approach; see his book, The Categories and Types of Present֊ Day English Word-Formation . . . [Wiesbaden: Harrassowitz, 1967]). The application of some facets of such thinking to individual word biographies prevails in some of my latest narrow-meshed inquiries, e.g., "Morpho-Semantic Conditioning of Spanish Diphthongization; the Case of teso ~ tieso: Implications of their Partial Overlap" (to appear in RPh) and a piece pitting Archaic Sp. liedo 'cheerful' against Late Old Spanish ledo, a borrowing from Galian-Portuguese; see La Corànica, IX:2 (1981), 95-106. Cf. also "Dos voces hispanoamericanas: zonzo y secante . . . " , Simposio internacional de lengua y literaturas hispánicas (Bahía Blanca, 1981),229-235. Thomas Montgomery has of late been ploughing over an adjoining field by won dering about the relation of nuclear vowels to vowel-dominated conjugation classes in Spanish - and providing some memorable answers. (See, e.g., "Complementarity of Stem-Vowels in the Spanish Second and Third Conjugations", RPh, XXIX:3 [1976], 281-296.) Spanish -io adjectives occupy the focus of attention in Steven N. Dworkin's article, "Derivational Transparency and Sound Change; the Two-Pronged Growth of -idu in Hispano-Romance", RPh, XXXL4 (1978), 605-617. The phantoms of Menéndez Pidaľs "sufijos átonos" haunt the Studia Hispanica

SUPPLEMENT

577

in Honorem R. Lapem (Madrid: Gredos, 1974), in a longer study jointly fathered by Charles B. Faulhaber, John K. Walsh, and Steven N. Dworkin in collaboration with myself — with some assistance from Raimundo Lida as regards stylistic polish ("El nucleo del problema etimologico de pícaro ~ picardía: en torno al proceso del préstamo doble", II, 307-342), and in a shorter separate piece from the pen of Jerry R. Craddock, "Las categorías derivacio nales de los sufijos átonos: picaro, páparo • y afines", III (1975), 219-231.

"The Five Sources of Epenthetic /j/ in Western Hispano-Romance" As its subtitle indicates, this study deals overtly — much as do several others of the series of papers here assembled — with pluricausal development, a view that presupposes a modernist approach to linguistics, based on concern with etiology. On the other hand, its dedication to the memory of R. Menéndez Pidal, whose death all but coincided with its completion, is witness to the author's firm belief in the far-going reconcilability of the traditional (fact-gathering) and the experimental (interpretive) styles of analysis. Their reconciliation need not involve anything as hopelessly flat as a lame compromise; there surely exist more imaginative responses to the challenge, or, to put it differently, escapes from the predicament, whose discovery is imperative. For a pen portrait of Menéndez Pidal, not only as dialectologist and historical linguist, see my necrological essay in RPh, XXIII:4 (1970), 371-411: "Era omme esencial.. . ". The present article admittedly makes difficult reading for the non-Hispanist, for two separate reasons: first, the majority of forms cited, being dialectal, are bound to strike the non-specialist as unfamiliar; and second, some of the lexical and etymological tangles involved are notoriously "tricky" - suffice it to mention Sp. columpiar 'to rock' {columpio 'swing') as a source of many an etymologist's severe headaches — almost foreseeably so, given the genetic opaqueness of numerous words endowed with such a meaning. My personal curiosity about the Asturian dialect goes back, I might add, to my review of María Josefa Canellada's doctoral thesis (El bable de Cabrones, Madrid, 1946) in Lang., XXIII (1947), 60-66, and may have reached its crest in the mono graph-length study, Patterns of Derivational Affixation in the Cabraniego Dialect of East-Central Asturìan, which was concluded and made its appearance at approxi mately the same time as the article here commented upon, in the "University of California Publications in Linguistics" series, LXIV (1970), and has since provoked a lively discussion. This would tend to place my circumstantial review of L. Rodrí guez-Castellano's book, La variedad dialectal del Alto Aller, in Lang., XXX (1954), 128-153, fairly close to the mid point of the entire span of commitment. Menéndez Pidal's involvement in Asturo-Leonese, which started at the threshold of this century, reached a high degree of intensity by 1906, when his pioneering synthesis, El dialecto leonés — the outgrowth of a university course — appeared in a journal (expanded book-length edition, by Carmen Bobes, in 1962). The corpus of relevant studies includes the author's magisterial review, in the Revue de dialec tologie romane, of E. Staaff s splendid edition and elucidation of several collections

578

SUPPLEMENT

of Old Leonese documents: Étude sur l'ancien dialecte léonais d'après les chartes du XIIIe siècle (Uppsala, 1909). The 1906 monograph is usually selected as the starting point for inquiries into the epenthetic / j / highly characteristic of the Penin sula's entire Northwest: dial, muriu 'wall' in lieu of Cast. muro, dial. quiciás 'perhaps' corresponding to Cast. quizá(s). What undoubtedly attracted me to the investigation of the five isolable sources of this / j / was the prospect of demonstrating once more how often the first gambit in a subtle phonological play is made under the auspices of morphology, as it were. Thus, as one sifts such Ast.-Leon. forms as alabancia 'praise', comparancia 'compari son', and folgancia 'ease, leisure', whose counterparts in Old Spanish ended in -onza, one cannot help remembering the fact that learned -ancia and vernacular -anca (reinforced by the influx of certain Gallicisms and Provençalisms) were pitted in keen competition in medieval Spanish (particularly striking is the eventual triumph of the Gothico-Latin hybrid ganancia 'gain' over its rival, still obliquely observable in OPtg. gaança, mod. gança). As a matter of fact, in an adjacent field learnèd -encia almost completely crowded out "popular" -iença, preserved only in relics (e.g., OSp. simiença 'seed') — witness hem-, him-encia 'ardor, vehemence' mitad; against this background, the peculiar remedial action speakers took in the case of fee-, feadad 'ugliness', by expanding it to fealdad and by, later, minting frialdad 'cool ness' from frío to circumvent any need to use *friedad (see my earlier study in University of California Publications in Linguistics, I:5 [1945], 189-214) acquires a much-needed additional dimension. ֊ Finally, "Proliferation of Variants as a CoDeterminant of Sound Change? The Case of Old Spanish c < DJ" (pp. 391-395) has points of contact, methodologically, with "Range of Variation as a Clue to Dating" (1968) — included in the present collection of articles - and, topically, with the piece, now in press, "A Hidden Morphological Factor Behind Instances of Erratic Distribution of ҫ and z in Old Spanish?", to appear in RPh, XXXV:1 (Aug. 1981).

"One Characteristic Derivational Suffix of Literary Italian: ~(T)AGGINE" The triad of Latin derivational suffixes -ÄGÖ, -ĪGŌ, and -UGŌ poses a host of pro blems in that language and, to an even larger extent, in the daughter languages. I was aware of only one small fraction of those problems and their ramifications in addressing that cluster of morphemes for the first time: "The Suffix -āgō in Astur []

SUPPLEMENT

583

-Leonese-Galician Dialects", Lang., XIX (1943), 256-258. By the mid 'seventies, when the piece here reprinted was being written, some progress had been achieved on the Hispanic side, see "Some Late-Twentieth-Century Options . . . ", BHS, LII (1975), 1-11, esp. 4-5; "Pre-Classical French une (~ un) image 'likeness, statue', Old Portuguese urn (~uma) viage(m) 'journey': A Study of Parallelism in Reverse", RPh, XXVIII: 1 (1974), 20-27 (with numerous further bibliographic clues in fn. 15); "In Search of 'Penultimate' Causes of Language Change: Studies in the Avoid ance of /ž/ in Proto-Spanish", Current Studies in Romance Linguistics, edd. Marta Luján and Fritz Hensey (Washington, D.C., 1976), pp. 27-36; "Infinitive Endings, Conjugation Classes, Nominal Derivational Suffixes, and Vocalic Gamuts in Roman ce" (in press); and "Las dos fuentes del sufijo luso-hispánico -én: -ĀGINE y -ĒDINE" (to appear in Vol. I: Estudios lingüísticos of the Homenaje a Manuel Alvar, now in process of crystallization). The last-mentioned item became possible after the breakthrough achieved by Martha E. Schaffer, "Portuguese -idao, Spanish -(ed)umbre, and Their Romance Cognates", to appear in RPh, XXXV:1 (August 1981). The paper here offered is an invitational contribution to the Miscellanea Bruno Migliorini, a scholar whose growth and accomplishments I had the privilege of analysing in a critical necrology published elsewhere {RPh, XXIX:3 [1976], 398408). Let me remark, in self-criticism, that insufficient stress has here been laid by me on the semi-learnèd status of the chosen suffix; in purely rural transmission -āgine tended to yield -ana or -aña in both Luso-Hispanic and Italian dialect speech, what through hypercharacterization of the feminine gender, what through confusion with the substantivated descendants of -āneu, -ānea (the latter either f. sg. or n. pl.). For some hints to this effect see "The Etymology of Spanish maraña", BH, L (1948), 147-171; "The Etymology of Spanish calano", PhQ, XXVII (1948), 112122; "The Romance Progeny of Latin pedaneus", AGI, XXXVI (1951), 49-74; and "Nuevas aportaciones para el estudio del sufijo -uno", NRFH, XIII (1959), 241-290. The paper "Medieval Roots of the Spanish Derivational Model sabid-or ~sabid uría" incidentally referred to in fn. 3 has meanwhile appeared in RPh, XXXIII:1 (1979),102-116. "(

The Double Affixation in Old French gens-es-or, bel-ez-or, Old Provençal bel-az-or "

Certain details adduced in this note have been set straight in a short but incisive critique by F. Lecoy, see Rom., XCV (1974), 462. The piece can, one hopes, be fruitfully consulted in conjunction with two far more sweeping inroads on terra incognita: "Los interfijos hispánicos; problema de lingüística histórica y estructural", in: Estructuralismo e historia; miscelánea-home naje a André Martinet, II (La Laguna, 1958), 107-199, and "Studies in Irrevers ible Binomials", Lingua, VIII (1959), 113-160. Numerous details have been successfully probed by various scholars on more recent occasions. To pick but one telling example, the study of OProv. viatz, OFr.

584

SUPPLEMENT

viaz 'lively, quick(ly)' has been pushed forward by Keith E. Karlsson both in his note "A Midway Report on the Etymology of Fr. vite 'quickly', 'quick!', OFr. viste 'clever, agile, quick'," in RPh, XXXI:2 (1978), 243-254, and in his belated, if searching review (ibid., XXXIII:3 [1980], 420426), of Lars Lindvall's weighty dissertation, "Sempres", "lues", "tost", "viste", et leurs synonymes.. .(Göteborg, 1971). What remains to be ascertained, however, is why these uses of the comparative and absolute-superlative degrees of Latin adjectives and adverbs have lingered on, in Northern and Southern France, for centuries after their extinction elsewhere, including Hispano-Romance. What is urgently needed, after the irrefutable identifi cation of Gaul as the part of the Empire from which most innovations rayed out, is the inventory of archaisms which, counter to expectation, have survived, or for a while eked out a modest existence, to the North of the Pyrenees, long after having been elsewhere swept off into oblivion.

"The Rise of the Nominal Augments in Romance" Menéndez Pidal's strength as historical linguist did not He in the felicity of his adoption or coinage of technical terms. Thus, he sided with those precursors or contemporaries who, ambiguously enough, accepted "inflexión" for 'metaphony', 'Umlaut'; and he endorsed a confusing use of the label Vocal tematica' (see my "Editorial Comment" to this effect in RPh, XXXII:2 [1979], 333-335). Something similar happened when, in his otherwise brilliant contribution to the 1905 A. Mussa Testimonial (Bausteine zur romanischen Philologie), the Madrid scholar described certain semantically elusive suffixoids as "sufijos átonos", a label that, disappoint ingly, has attached to them ever since, not only — as is understandable — down to the latest writings of the Madrid School (e.g., R. Lapesa, Historia de la lengua es pañola, rev. 8th ed. [Madrid: Gredos, 1980], p. 634), but also, as is less condonable, in those of relative outsiders. On the strength of theoretical analysis, the term "suf fix" is not applicable to segments performing no clear-cut grammatical (functional) service and devoid of any neatly-silhouetted meaning. The elements involved ought to be somehow segregated from the contingent of genuine suffixes, through use of some such unequivocal tag as "augment" or "excrescence". Another point of major concern to me was the readiness of numerous scholars, from the mid 'forties to the late 'sixties, to assign to a Mediterranean or nonde script substratum what could not readily be traced to Latin, or Germanic, or Arabic. Without in the least questioning the ever-present possibility of substratum influence, in a cultural climate of bi- or pluri-lingualism, I nevertheless became inured to caution, especially after reading and reviewing in painstaking detail certain books by Menén dez Pidal himself and by an even more ardent and radical "substratist", Johannes Hubschmid; see my critical experiments along this axis in Speculum, XXIX (1954), 128-153; and inLang., XXXVIII (1962), 149-185, and XLVII (1971), 465-487. New vigor could be instilled into the study of such "adjuncts" (on second thought, an even better designation than "augments") as -go, -lo, -mo, -no, -ro and their feminine counterparts by bracketing it with inquiries into Romance proparoxytones.

SUPPLEMENT

585

An unbiased inspection of the record discloses that among Italian nouns stressed on the antepenultimate, many are either straight Hellenisms or pertain, referentially, to domains of real life strongly affected by Greek influence (horticulture, phar macology, navigation, and the like). As a vogue or fashion, the predilection for voci sdrucciole and voces esdrújulas in the two Peninsulas could very well have start ed in this sector of the lexicon. The leveling or the hardening of ֊go, -ga, etc. into attachable syllables serving the purpose of lending a more pleasing prosodic contour to words appealing to the speaker's imagination are further developments idiosyn cratic to Hispano-Romance. They seem innovative and attributable to the, generally, sweeping power that certain structural features have at all times tended to exhibit in Spanish, although some side-connection to pristine Mediterranean languages cannot be categorically ruled out. The many-faceted problem of the esdrujulismo deserves to be re-opened; what Amado Alonso had to say on "cambios de acento" more than a half-century ago was, at best, satisfactory or good for its own time; but the missing social dimension, the relation to the various registers, must be built in; and point-by-point compari son with Portuguese and Italian, in lexical items such as academia, anécdota, atmós fera, farmacia, océano, prototipo is unavoidable, in synchronic and diachronic perspective. My own insistence on the heavy share of Arabisms in the diffusion of this syllabico-accentual pattern (see my review of the revised 2d ed. of Lapesa's Historia de la lengua española in RPh, VI·l [1952], 52-63) seems to me justified, in retros pect, even though it left the author himself only half-convinced, judging from his lukewarm reaction in the 1980 revision of his book. Particularly heartening has been the repeated enthusiastic response by María Rosa Menocal. As regards the initial stimulus conceivably provided by Hellenisms in Antiquity, I have had the satisfaction of seeing that conjecture anything but roundly denounced by such cautious experts as Henry and Renée Kahane; for their authoritative position see, in particular, "Paideia — a Linguistic Subcode", in: Wege zur Universalienforschung (Tübingen: Narr, 1980), pp. 509-520, at 510.

"Identification of Origin and Justification of Spread in Etymological Analysis " When a certain style — temporarily very successful — of historical linguistics asserted itself in the late 19th century, any etymology newly-proposed had to be proved defensible on the side of the shifting sound systems (hypothesis of the ab solute "regularity of sound change"), with some less structured attention reserved for morphological and semantic criteria. Occasionally, scholars would identify the social milieu in which the innovation is likely to have arisen (e.g., among sailors or soldiers). Conversely, little testing of the probable reasons for the slow or rapid spread of the newly-minted item was deemed necessary. A century later, no etymological proposal is complete unless its advocate includes in his plea for the acceptance of the hypothesis an independent set of circumstances that must have favored the adoption of what could, initially, have been an experi-

586

SUPPLEMENT

mental coinage. To the examples cited in the 1969 paper let me add this one, extracted from an ongoing research project. Lat. culmus signified 'stalk, stem, straw' and would thus, on semantic grounds, be a nearly perfect prototype for Sp. cuelmo 'candlewood' beside techumbre de cuelmo 'roof of thatch', with which it also shares the gender and several phonic features. Lat. , however, does not normally yield ue in Castilian and so, for this reason alone, one highly qualified scholar (J. Corominas) declared it necessary to abandon an almost obvious equation in favor of some non-Latin fantasy. A year later another expert (V. García de Diege) tried to rehabilitate the cûlmus hypothesis by averring that in certain dialects to the west and the east of Castile it is not unusual for ancestral  to be represented by ue. Even if this is so, the explanation proffered is still, at best, partial, since one is left wondering why a Leonese or an Aragonese form, in this particular semantic sector, should have invaded Castile. But if one adds, to García de Diego's argument, that the ideal, or regular, reflex of culmus, namely *colmo, could very well have been avoided given the threat of its collision with colmo 'height, limit', (adj.) 'overflowing', from ances tral cumulu (cf. colmar = Fr. combler 'to fill to overflowing' < cumulāre), to say no thing of colm-illo 'eye-tooth, tusk' (Ht. 'little column'), from an inferred masculine counterpart of columella, or of colmena 'beehive' (of unknown origin), then the speakers' avoidance of obnoxious polysemy through appeal to an unambiguous dialect form, namely cuelmo, becomes truly understood; perhaps I should have written "of some speakers", since the latest reference works (e.g., E.B. Williams' excellent bilingual dictionary) also record colmo 'thatch, thatched roof, with Coro minas' invocation of Celt. kólmos 'straw' thus collapsing under its own weight. The spread, or itinerary, of a newly-borrowed or newly-invented word is some times best determined through reconstruction of another lexical item which the newcomer must have dislodged before striking root itself. Thus, the many con jectures centering about Sp. hasta and Ptg. até 'until' having been loaned by Arabic (a unique instance in the ranks of prepositions) fall short of conviction unless one includes, among the antecedents, some educated guess about the temporary survival of usque 'id.' in Hispano-Romance — with full attention to the ever-present possibili ty of conflation. The brief statement (pp. 468ff.) on the alternation of  and o, as rival products of ä, in Hispano-Romance (Sp. plomo/ Ptg. chumbo 'lead', Sp. honda/Ptg. funda 'sling', and the like) has since become the starting point of several studies now in preparation or already in press; their upshot may well be an attempt at re-defining the position of the bundle of Hispano-Romance dialects within the total edifice of Romance - in part, on the strength of the varying outcomes of Lat. . If it is true that Sp. sombra 'shade' represents a sort of compromise between solombra < sub iliā umbrā, lit. 'under [i.e., in] the shade' and ombra/*umbra (the latter inferred from the adj. umbrío), a conflation presumably achieved under the influence of the word's semantic opposite sol 'sun', then sombra/'sol, already en visioned as moving in tandem by S. de Covarrubias (later supported by Corominas), would beautifully illustrate the agency of lexical polarization; see the concluding segment of the last paper in this collection.

SUPPLEMENT

587

"Primary, Secondary, and Tertiary Etymologies " The quintessential message of this paper is that, perhaps more frequently than had been thought possible, the etymologist, instead of deciding on one of two or three solutions so far proposed, will be best-advised to opt for the assumption of a conflation or coalescence of two originally independent lexical families (in the given case [a] insānia 'madness' from īnsānus 'of unsound mind, mad, raving'; [b] sanna 'mocking grimace', 'face contorted or convulsed with rage'; [c] saniēs, var. sania 'corrupted blood' conceived of as a correlate of 'angry mood'. One close parallel to this analysis will be found in the earlier paper, "En torno a la etimología y evolución de cansar, canso, cansa(n)cio", NRFH, IX (1955), 225276, which preaches the wisdom of accepting as a starting point, in isolation, neither quassāre 'to shake vigorously', nor Gr.-Lat. campsāre 'to circumnavigate, shun' — a dilemma pervading almost all earlier discussions — but a blend of the two, against the background of general fluctuation between -ns- and -s(s)-, whereas It. (s)cansare 'to avoid, evade, escape from', (refl.) 'withdraw' is indeed a faithful replica of camp sãre, without further interference. For the theoretical foundation of such mergers of conjectures turn to my earlier piece, "The Uniqueness and Complexity of Etymological Solutions", Lingua, V (1956), 225-252; absorbed into my Essays on Linguistic Themes (1968), pp. 229250; translated into Italian and, in the process, expanded in: Linguistica generale, filologia romanza, etimologia (1970), pp. 99-131. In regard to the luxuriant growth in Romance of the Latin suffix -ütus (pp. 483492), which has been a boon for those concerned with measuring the share of saniēs in the crystallization of saña/en-sañar/sañudo, there is a point in mentioning the roughly contemporary companion piece, "Ancien français faü, feü, malostru; à la recherche de -ŪCUS, suffixe latin et paléo-roman rare de la mauvaise fortune", in: Mélanges . . . Paul Imbs = TLL, XI: 1 (Strasbourg, 1973), 177-189, except that the record of the same Latin derivational suffix here serves to aid me in rejecting the reconstruction *fãtutus, while there it helped me to vindicate *saniūtus.

"Etymology and Modern Linguistics'' In contrast to the more theoretical essays on etymology (like the following piece in this collection, say), this paper — tilted from its conception so as to qualify for presentation as a public lecture — aims at plunging the audience or a reader right into the "atmosphere" of a present-day assembly of linguists, or of persons passively interested in linguistics (whether a colloquium or a classroom), from which, as is very well known to all habitués, concern with etymology as the most "anec dotal" and "volatile" of all linguistic disciplines has, to all intents and purposes, been banned. Hence the need for overdramatizing the issue. In Romance quarters. where etymology, not so long ago, ranked as an overridingly important discipline, its widespread abandonment has knocked the very foundation from under almost all organized diachronic research. The purpose of this talk, delivered here and there in an almost indifferent, not

588

SUPPLEMENT

to say hostile, academic climate, was to isolate two situations in which systematic neglect of etymology had threatened to become downright hazardous. While it is true that some minor, unintegrated etymological discovery is unlikely to revolu tionize science, there do exist, at least, two categories of etymological "trouvailles" which can be of truly pivotal importance: The one sheds light on historical contexts that cannot be otherwise illumined (through archaeological spadework, for instance); the other catches glimpses of zigzag in internal language growth that might slip through the wider meshes of historical grammar. There is no implied recommendation here that anything else be constantly neglected, downgraded, or omitted to make room for etymology: The goal is rather one of a perfect equilibrium, as suggested indepen dently, and with different chunks of supporting evidence, in "Lexis and Grammar: Necrological Essay on Émile Benveniste", Rph, XXXIV:2 (1980), 160-194. One side-effect of papers of this kind has been to encourage and promote origi nal etymological research among younger scholars, starting with my own advanced and former students, provided they have felt an urge to engage in explorations so slanted. Steven N. Dworkin, Keith E. Karlsson, Consuelo López-Morillas, and David A. Pharies are some of the names that come to mind in this context; for speedy identification of fine specimens of their work see several recent volumes of Romance Philology. Some of the issues briefly alluded to in this paper have been presented elsewhere in distinctly more elaborate form. Thus, apropos the velar insert in OSp. fago (mod. hago), yago, etc. one finds a more copiously annotated presentation of the issues in: "New Problems in Romance Interfixation: The Velar Insert in the Present Tense . . . ", RPh, XXVII:3 (1974), 304-355. The discovery of internecāre 'to kill (through hardening frost)' used in reference to plants, as the ultimate source of Sp. terco, dial. tierco, Gal. terque 'stubborn' was announced and vindicated in PMLÂ, LXIV (1949), 570-84; etc. It is heartening to see that, in response to such appeals for heightened attention to etymological probing, the array of Vol. I of Trends in Romance Linguistics and Philology, edd. R. Posner and J.N. Green (The Hague, etc.: Mouton, 1980) should, remarkably enough, have allowed for the inclusion of a separate section, "Romance Etymology: A State of the Art" (pp. 191-240). Deliberately, no concrete samples of my own etymologizing have been included in the present volume, except for the excursus on the provenience of Hisp. tomar 'to take' appended to the last item (pp. 556-558), and for the extraction of Sp. desleír 'to dilute' from dēlēre, plus its implications (pp. 513-541).

"

The Interlocking of Etymology and Historical Grammar"

This paper was originally written for, and presented at, the Second International Conference on Historical Linguistics, held at the University of Arizona (Tucson) in January 1976. It was included in the Proceedings of the Conference later that same year, my only regret being the unfortunate typographic garb in which it appear ed in that context. I am pleased that, in addition to the not insubstantial footnotes,

SUPPLEMENT

589

the piece also contains three comments made from the floor (by Edward F. Tuttle, A. Richard Diebold, Jr., and Ilse Lehiste, plus my short reply to these constructive elaborations). A well-established and entirely defensible tradition of Romance scholarship demands that a well-rounded-out etymological essay contain not only a forcefully presented new proposal (or solution), but also a critical - and, ideally, exhaustive — account of what earlier generations of linguists have contributed to the attack on the chosen problem, plus a statement of what has not yet been achieved and thus remains to be done, by way of vindication of the new conjecture. By this fine tradition I have at all times tried to abide as best I could, over a period of more than thirty-five long years. See such word-biographical studies from my pen — representa tive of different periods of my career — as "Old French soutif 'Solitary'," Modern Language Quarterly, III (1942), 621-646; "The Etymology of Portuguese iguana", Lang, XX (1944), 108-130; "Relics of mergus, mergulus, and mucrō in IberoRomance", American Journal of Philology, LXVII (1946), 151-167; "Three His panic Word Studies . .. ", UCPL, I:7 (1947), 227-296; "Cundir: Historia de una palabra y de un problema etimológico", Homenaje a Rodolfo Oroz, Boletín del Instituto de Filología de Chile, VIII (1954-55), 247-264; "Sobre el núcleo etimoló gico de esp. ant. desman(d)ar, desman(o): lat. DĒ-, DĪ-MĀNĀRE", Homenaje a María Rosa Lida de Malkiel, I, in: Filología, VIII (1962[-64]), 185-211. In some etymological articles either "l'historique du problème" just alluded to, or the advocacy of the new hypothesis may contain a lesson or message for the student of general diachronic linguistics, which it seems appropriate to salvage from the huge pile of raw data and of references to secondary sources of information relevant only to the highly specialized reader. When such an atypical situation obtains, it is hardly a disservice to organized scholarship to alert the generalist, or the experts tilling adjoining patches of ground, of the methodologically pertinent nu cleus of the given problem by announcing it right from the start, namely in the title. Hence the wisdom (one hopes) of such novel titles as "Polysémie, homonymie et dérivation verbale en paléo-roman: Autour de la reconstruction de *traginãre, *tragicāre, *traxināre", in Estudis Romànics, XIII(1968[-70]): Miscel·lania Pompeu Fabra), 1-12, "Phonological Irregularity vs. Lexical Complexity in Diachron ic Projection: The Etymological Substructure of Luso - Hispanic abarcar t o clasp, embrace, contain', in: Issues in Linguistics: Papers in Honor of Henry and Renée Kahane (Urbana, etc.: University of Illinois Press, 1973), pp. 606-635; to which one can now add "The Interlocking of Etymology and Historical Grammar, Exemplified with the Analysis of Spanish desleír", here offered in improved shape. All in all eight entirely different interpretations of the enigma have been offered over the years, and it is interesting to watch how the analysis here held to be correct, namely the assumption of descent form dēlēre 'to destroy', was for the first time postulated by R J . Cuervo (1893) and initially failed to arouse much enthusiasm in any quarters or to win support even from a trail-blazer of the caliber of Menéndez Pidal. What concealed so long the access to the correct base was the rank-and-file scholars' inadequate knowledge of minor points of historical grammar, such as the rivalry of the prefixes de- and des- or the relation of -eer to -eir and -uir infinitives, despite

590

SUPPLEMENT

such displays of latent possibilities as the change, in Portuguese, of posseer into possuir. All of which intensifies the urgency of the message that could have been peeled off from my necrological essay on Émile Benveniste (RPh, XXXIV:2 [Nov. 1980], 160-194): the need for an improved balance of lexis and grammar.

"Contrasts Between blasphēmāre and aestimāre" A stroke of luck, forty years ago, led me to investigate (in slightly superficial fashion, I am tempted to add in sober retrospect) certain influences in form (e.g., as regards the choice of the stressed vowel) and in grammatical behavior (e.g., as regards gender) that the stronger of two semantic opposites is apt to exert on its weaker partner — with "strength" and "weakness" being, in each instance, cultural ly determined ("Lexical Polarization in Romance", Lang., XXVII [1951], 485518). The article failed to arouse more than a conventional amount of critical atten tion (there appeared brief appraisals by G. Rohlfs and S. Ullmann), but it opened before myself almost unlimited possibilities, doubly valuable where the chance to eliminate etymological unknowns thus began to bloom. One such experiment, in volving 'hand' and 'foot', prompted me to write "Estudios de léxico pastoril: piara y manada", BH, LIII (1951), 41-80. The study concluding the present collection also comes under this rubric, except that it deals with  pair of verbs rather than of nouns, namely 'praising' (Ht. 'esteeming, estimating') vs. 'cursing'. It is, of course, equally worth-while when applied to other forces operating in language history; some such alliances and interferences come to the fore in a fairly recent paper, transcending Romance and composed in a slightly different key: "Semantic Universals, Lexical Polarization, Taboo: The Romance Domain of 'Left' and 'Right' Revisited", in: Festschrift für Oswald Szemerényi on the Occasion of his 65th Birth day (Amsterdam: Benjamins, 1979), pp. 507-527. Then again, it seems legitimate to subsume lexical polarization under the broader spectrum of hypercharacterization; to this effect, see my piece of later vintage than the pioneering article, namely "Diachronic Hypercharacterization in Romance", ArL, IX (1957), 79-113;X (1958), 1-36, at 103-106, which lets lexical polarization run neck and neck with "lexical serialization" (106-112) and contrasts them both with "grammatical hypercharac terization" (97-100), while allowing for a narrow border-zone between grammat ical and lexical hypercharacterization (112f.). The sporadic transmutation of Gr.-Lat. blasphēmāre into *(b)lastimare has long been something of a riddle as long as it has, again and again, been focused upon in isolation. Certain baffling features, including the wholesale replacement of the -ēmāre segment by -titare, lose much of their outlandishness once aestimāre is recognized as the prime stimulant, by virtue of its opposite meaning. Other dis quieting features remain to be explained; thus, the word-initial consonant cluster BL- is a notorious crux, as is its steady companion GL-; the loss of b- and g- in this context is, to say the least, a strong possibility (witness glattīre 'to yelp, yap' > Sp. latir 'to beat, throb'). In short, we have before us the manifestation of a "weak sound change", as tentatively defined in Lingua, XI (1962), 263-273. The muchneeded counterpart of the earlier inquiry into the vicissitudes of CL-, FL-, PL· in

SUPPLEMENT

591

Hispano-Romance ("The Interlocking . . . ", ArL, XV [1963], 144-173, XVI [1964], 1-33) remains to be written. The excursus on the provenience of Hisp. tomar involves a chance discovery. While scanning the record of aestĭmare in the context just adumbrated, I came across the doublet — once widely in use — aestümāre, reminiscent of the dyads maxïmus~maxumus, and the like; and, having previously been alerted to the myste ry surrounding the rise of tomar, I recognized the possibility of a folk-etymological division of aes/tumāre, with aes standing for 'copper, bronze', 'money'. Since aes fell into desuetude at a fairly early date in spoken Latin, this etymology, if accepted, would strenghten one's belief in the basic archaicity of Hispano-Romance, to which other peculiarities (e.g., the widespread preservation of u as u rather than o, a feature that no longer can be swept under the rug, as it consistently has been for over a century) would also tend to point. Any future return, on my part, to lexical polarization — with a view to arriving at some broader synthesis ֊ would have to take into account my single most elabo rate effort in that direction: "Ancient Hispanic vera(s) and mentirai(s): A Study in Lexical Polarization", RPh, VI (1952-53), 121-172.

'Post-script" To avoid overextending these afterthoughts, I have aimed at keeping the number of cross-references to a modicum; it could easily have been doubled. For the same reason, only a small selection of references to all sorts of digests, epitomes, and brief reviews (some of them distinctly valuable, and all much appreciated) has been provided. Those in need of exhaustive information will find their efforts richly rewarded if they scan the relevant volumes of such serials as the United Kingdoms's The Year's Work in Modern Language Studies and Colombia's Boletín del Instituto Caro y Cuervo; as well as certain pertinent surveys, e.g., Vol. I: Romance Compara tive and Historical Linguistics (1980) of Posner and Green's Trends in Romance Linguistics and Philology and a few contributions to the larger and older Current Trends in Linguistics series, including a major one by Dell Hymes. At proof, I can further report the appearance of the following long-delayed piece of wide-range criticism: Otto Gsell, "Das sprachwissenschaftliche Œuvre Yakov Malkiels. Ein Forschungsbericht (1958-1978)", Ibero-Romania, XIII (Tübingen: Niemeyer, 1981), 1-29. Further appraisals — as a rule, too flattering — can be picked from prefaces and introductions to book ventures in which I happen to be represented, and from routine reviews of such volumes in the journals, also. At the request of friends, I have embarked on the unhurried preparation of an auto-bibliography, which is unlikely to make its appearance before 1983. That piece, if it ever springs into existence, should be less parsimonious about such bits and snippets of collateral information. By way of compensation, it may turn out to be more austere and stringent in the few value judgments that it is apt to contain. Let me finish with an apology for any possible inadvertent overlap of this Supple ment (1981) with the Introduction, of much earlier vintage.

INDEX OF NAMES

The family names have been alphabetized, and the first names abbreviated, according to the English norms, so that Spanish ch-, e.g., has been wedged in between ce- and ci·; German Philip figures as P. rather than Ph. ; and French Frédéric as F. rather than Fr. In Spanish names the first component of the family name, and in Portuguese names its last component, have been foregrounded; thus, Menéndez Pidal, R., contrasts with Viana, A. dos R. Gonçálvez (or Gonçalves). For women, the maiden name has been emphasized wherever advisable, thus: Michaelis de Vas concelos, . Names of editors of texts and of co-authors of books have been listed selectively; a single entry has been provided where two members of the same family have worked together (partnership or subsequent revision), Cyrillic script has been transliterated. Diacritic marks, for the most part, do not count at all, except that ä has been equated with ae, etc., which is the standard procedure in American library card catalogues; and that š- and ş- have been treated like sh-.

Abraham, R.D. 116 Acevedo  Huelves, B. 275f., 279, 287, 291,295,482 Adams, E.L. 182, 193, 205, 489 Adelung, J.C. 562 Adolf, H. 61 Aebischer, P. 438 Ageno,F.(B.)190,355,360 Aguado, J.M. 554 Ahlsson, L.E. 206 Alarcos Llorach, E. 79, 233, 248, 388f. Alcalá, P. de 36 Aldrete, . 550 Alemany Bolufer, J. 194, 196, 433, 461,465,483,520,533 Alessio, G. 360, 408, 433 Alighieri, D. 70 Alinei, M.L. 79 Allen, A.S. 568 Allen, C.G. 335 Allen, W.S. 218 Alonso, A. 59f., 60, 67, 69, 71, 97, 108, 115, 118, 132, 163, 258f., 293,348,360,443,456,585

Alonso, D. 59, 89, 349, 433, 437 Alonso, M. 455 Alonso Garrote, S. 122, 270, 464f., 483f. Alsdorf-Bollée, A. 139, 149, 183, 431,445 Alvar, M. 110 Álvarez, G. 270 Amador de los Ríos, J. 391, 544 Ampère, A.M. 28 Ampère, J.J. 307, 317 Appel,C.97,414f.,456 Arbois de Jubainville, H. d' 375 Arens, H. 83 Argote de Molina, G. 551 Aristotle 70 Arlía,C. 191 Armistead, S.G. 87 Armstrong, E.C. 343 Arnauld, A. 70 Arnold, H.H. 554 Arriaga, E. de 294 Artigas, M. 463 Areiller, R. 460

594

INDEX OF NAMES

Ascoli, G.I. 14, 28f., 31f., 59, 67, 209, 261, 323-30, 332, 339, 34045, 347, 354, 360, 374, 454, 458f., 468, 580 Asensio, E. 349 Asín Palacios, M. 101 Austerlitz, R. 87 Ayegui, J.P. 460, 477 Badia Margarit, A.M. 97, 431 Bahner, W. 83, 560 ,. 180 Baist, G. 112, 118, 122, 233, 238, 323, 334, 349, 359, 382, 433, 468, 489, 531 Balakian, Α. 565 Baldonado, LM. 568 Bally, . 68, 148, 196,438 Bambeck,M. 550 Barbier, P. (fils) 23, 139, 446 Barnett, F.J. 466 Barrenechea, A.M. 231, 378, 574 Bàrtoli, M. 67f., 71, 73, 362, 375 Bataillon, M. 171,175,350 Battaglia, S. 3lf. Battisti, . 32, 408 Baudouin de Courtenay, J. 75 Baumann, H.H. 445 Baumgartner, E. 87 Becker, P.A. 73 Beeler, M.S. 87 Behrens, D. 341,438 Beinhauer, W. 444 Belardi, W. 80 Belić, A. 74 Bello, A. 67, 380, 518, 533 Belloni-Filippi, F. 160, 174 Benedict, R. 59 Benfey, R. 49, 64 Bénichou, P. 253 Benveniste, É. 25, 60, 64, 68, 70-72, 82, 138, 149, 205, 378f., 503, 562, 590 Bergh, Â. 77 Bernhard, . 166, 168-70 Bertoldi, V. 23, 32, 71, 139,433,435,

562 Bertoni, G. 71, 73, 433 Betz,W. 138 Bezzola,R.R. 139,149 Blaylock,  87, 107f., 113, 434, 566 Blecua, J.M. 468 Bloch, B. 63, 77, 562 Bloch, O, 213, 325, 331, 415, 461, 549 Blomgren, Â.L. 436 Bloomfield, L. , 49, 53f., 63, 71, 73-75, 78, 81, 89f., 138, 144, 149, 218, 220f., 224, 266, 326, 444, 502, 568 Bloomfield, M. 391 Bly, P.A. 333 Boas, F. 63, 165 Bobes, . 269, 577 Bodmer, F. 220 Böhl de Fsber, J.N. 464 Böhtlingk, O. von 68 Boggs, R.S. 554 Bolelli, T. 78, 83, 569 Boléo, M. de Paiva 168, 282 Bolinger, D.L. 560 Bolza, G.B. 25 Bonaparte, L.-L. 29 Bonet, J.P. 62 Bonfante, G. 113, 190, 246, 552 Bopp, F. 26, 28f., 56, 65, 67f., 70, 514 Borao, J. 337 Bork,H.D. 195 Bottiglioni, G. 436 Bourciez, É. 99 Bourciez, J. 92 Brachet, Α. 157, 173, 324, 340, 548 Brattö, . 77, 170 Braun, G. 362 Bréal, M. 27f., 69f., 70, 72, 138, 157, 180,183,206,324,374 Bremer, . 88 Breslin, M.S. 87, 145, 149, 446 Brøndai, R. 330 Brøndal, V. 26, 71, 74f. Bross, D. 59

INDEX OF NAMES

Bruckner, W. 197 Brüch, J. 102, 206, 331, 466, 522, 524, 526, 533 Bruerton, . 88 Brugmann, . 28, 40, 55, 65, 70f., 160, 164, 173, 198, 265 Bruneau, C. 343 Brunot, F. 139, 196, 309, 317, 343 Bruppacher, HJP. 457 Brutaüs,J.F. 110, 114 Butler, J.L. 246, 248, 378, 444 Byskov, J. 266 Carera, Α. 350 Cabrere, R. 460f., 464f., 477, 519f., 525, 533 Caix, N. 340, 362, 441 Calcano, J. 106 Calderón Escalada, J. 465 Calepino, A. 68, 70 Callaghan, CA. 87 Canellada, M.J. 270-2, 275, 277-9, 284, 287, 289, 291, 294, 420, 458, 470, 577 Canello, U.A. 157, 324, 374, 545 Cannon, G.H. 61 Capidan, T. 99 Carlsson, R.L. 89 Carnoy,A. 139 Caro Baroja, J. 14 Carreño, Α. 106 Carter, H.H. 106, 116, 350, 360 Carvalho, J.G.C. Herculano de 140, 149 Castro, A. 115, 120, 124, 209, 256, 335,349,356,391,551 Catalán, D. 67, 87, 102, 110, 113, 115, 233, 248, 289f., 458, 469 Cate, J.P. ten 115, 464 Cejador  Frauca, J. 554 Chantraine, P. 433 Charaudeau, P. 445 Chatterji, S.. 61 Chen, M.Y. 235, 248 Chézy, A.L. de 67 Chiappini, F. 455

595

Chomsky, N.A. 57f., 71, 79, 147, 359 Christmann, H. 562 Cihac,A.de l87 Cintra, L.F. Lindley 436 Cioranescu, A. 187 Orot, G. 110 Clédat, L. 548 Coelho, F.A. 96, 116, 157, 173, 324, 350,436f., 439, 545 Cœrdoux, D.L. 62, 70 Cohen, M. 71 Cohen, G. 490, 562 Collin, C.S.R. 139, 183 Collinder, . 76 CoUinson, W.E. 138 Comte, A. 23, 28 Conev,B. 187 Connors, . 145, 149, 167, 175, 507 Cooch, A. 483 Cornu, J. 96, 108, 114, 238, 289, 293, 342, 433, 463, 476f,. 478f., 483, 485, 520, 534-6, 533 Corominas, J. 19, 101f., 120, 203, 233, 237, 248, 253f., 256-8, 278, 283, 291, 294, 303f., 317, 332, 334f, 351, 355, 357, 359f., 377, 384, 429, 435, 452f., 461f., 465, 471, 476f., 479, 495, 518, 422-6, 529-33, 546, 548f., 550f., 553-5, 557f., 586 Correas, G. 465 Coseriu, E. 71, 88, 148, 378, 431, 434 Cotarelo, E. 105,120 Covarrubias Orozco, S. de 19, 459f., 561, 465, 474, 477, 479, 494, 518, 524f., 533, 544, 550, 554, 586. Craddock, J.R. 87, 175, 231, 233, 248, 290, 292, 295, 420f., 431, 433, 435, 445, 476, 490, 569, 571, 577 Crespo, J.S. 336 Crews, C.M. 113,464 Croce, B. 71,73 Cuervo, R J . 10, 19, 59, 67f., 257, 293, 348, 380, 396, 461, 468, 483, 518, 520-4, 526, 531, 533, 589

596

INDEX OF NAMES

Curtius, E.R. 49, 61 Cuveiro Pinol, J. 354 Damas-Hinard, J.J.S .Α. 353 Dambiertmont, J.L. 466 Damourette, J. 156, 168, 172 Dardel, R. de 92, 112, 115, 118f. Darmesteter, A. 11, 14, 28, 33, 190, 222, 308, 310, 317, 323f., 339-45, 545, 580 Darwin, C.R. 26, 69 Daupiás, J. Guimarães 436 Dauzat, Α. 77, 139, 167, 169f., 174, 182, 213, 325, 331, 345, 349, 353, 491,543 Debenedetti, S. 67 De Felice, E. 325 DeForest, J.B. 149,435,565 de Gorog,R.(P.)139 de Groot, A.W. 9, 18 Delbrück, . 49, 65, 68, 70, 83 Delcorno, . 373 Delius, N. 96 Dembowski, P.F. 87 Deroy, Լ. 138 Devoto, G. 68, 71, 73, 138, 304, 317, 330,355,480,550 Deyermond, A.D. 333 Diebold, A.R. (Jr.) 539, 589 Diez, F. 28f., 59, 61, 96, 179-83, 189, 191, 195, 197f., 206, 255, 257, 301, 306f., 323f., 335-7, 353, 362, 374, 422, 424, 426, 456, 460ff., 464f., 477-80, 519-21, 524f., 533, 544-6, 555, 562, 565 Diez Castañón, M.C. 108 Dihigo, J.M. 106 Doroszewski, W. 71 D'Ovidio, F. 533 Dozy, R. 257 Droixhe, D. 562 Dubois, J. 182, 196, 213, 305, 325, 331,491,543 Ducamin, J. 212, 358, 553f. Dudas, . 231

DumézÜ, G. 71 Durkheim, É. 71, 75, 77 Duro, Α. 142, 182, 304, 354, 361, 364-6, 480, 550 Dworkin, S.N. 13, 212, 326, 347, 497, 518,533,559, 569, 576f. Dyer, N.J. 311,317 Echave-Sustaeta, J. de 83 Edgerton,F.61f., 391 Egerod, S. 87 Eguílaz y Yanguas, L. de 257 Elcock,W.D. 137,149 Emeneau, M.B. 63, 501 Engler, R. 71 Ernout, A. 96, 180, 192f., 195, 197, 200f., 209, 254, 258, 287, 313, 317,336,346,351,354,359, 362, 364-5, 411, 429, 432, 436, 440, 442, 462, 480, 482, 518, 549, 556-8 Espinosa, A.M. 97, 108, 258 Estienne, H. 306-8,310,312 Falk, P. 302,318 Fanfanı, P. 191 Faulhaber, C.B. 577 Fay, P.B. 87, 151 Fernández y Fernández, M. 482 Fernández Gómez, . 455, 486 Fernández-Guerra y Orbe, Α. 115f. Figueiredo, . de 350 Fillmore, C.J. 561 Finck, F.N. 72 Firth, J.R. 77, 138, 147, 149 Fisher, M. 551 Fitz-Gerald, J.D. 519, 533 Flechia, G. 324 Fleischman, S. 171,175, 566 Förster, P. 96, 118, 461, 520, 526, 533 Foerster, W. 323f., 343, 518, 521, 534 Fontanella de Weinberg, M.B. 231, 571 Ford, J.D.M. 96, 103, 110, 120, 122, 303,318,334,349,351,461 Fortunatov, F.F. 65, 74

INDEX OF NAMES

Fouché, P. 548 Foulché-Delbosc, R. 464 Frei, . 71 Fréret, N. 62 Frings, T. 77 Frisk, H. 75 Fuchs, A. 95f. Gabelentz, G. von der 68, 75 Gaibrois de Ballesteros, M. 120 Galmés de Fuentes, A. 289f. Gamillscheg, E. 64, 130, 139, 142, 149, 161, 174, 193, 205, 213, 224, 258, 293, 302, 318, 325, 331, 413, 415, 439,491,547,550, 562f. Garcia, E.C. 87, 142f., 149, 171, 175 García Blanco, M. 168 García de Diego, V. 19, 118, 122, 257, 259, 336f., 348, 350, 353, 359, 388, 433, 476, 495, 523, 525f., 534, 586 García-Lomas, G.A. 107, 122, 259, 271, 294 García Rey, V. 270, 274, 284, 351, 465 Garda-Soriano, J. 121, 456 Gardette, P. 437 Gardiner, A.H. 77 Gaskins, L J . 323 Gaspar Remiro, M. 464 Gassner, Α. 96, 11lf., 114, 117f., 120 Gauchat, L. 40, 268 Gauthiot, R. 61 Găzdaru, D. 59 Geckeier, Η. 148f. Geijer,P.A. 190 Gellius, Α. 69 Georges, E.S. 139, 149, 183, 431 Gesner, . 70 Gessner, E. 95, 270, 293, 350, 464 Giacomini, . 67 Giese, W. 139 Gili (y) Gaya, S. 476, 519, 534 Gillet, LE. 59, 115, 349, 355, 461, 463, 465 Gilliéron, J. 33, 41, 67, 70, 77, 98, 123, 125, 159, 167, 170, 173, 185, 218-21, 223f., 230, 234, 266f., 307, 428,445,470f., 551

597

Glinz, H. 77 Godei, R. 71 Goetz, G. 548 Goidànich, P.G. 67 González-Campuzano, J. 107 González llé, F. 290, 412, 445, 490 Gooch,A.483 Gorosch,M. 108 Gorra, E. 96, 466 Gougenheim, G. 166, 174, 466, 562 Gräfenberg, S. 120 Grammont, M. 40f., 68, 70, 265-8, 345, 373, 375, 377f., 466, 547, 581 Grandgent, CH. 97, 367-8, 371-3, 379 Graur, Α. 83, 99 Greenberg, J.H. 130, 141, 145, 149, 156, 172, 187, 297, 312, 316, 318 Greimas, A J . 147,149 Grégoire, A. 70 Greive, Α. 421 Griera, A. 113 Grimm, J. 12, 28f., 64, 70, 160, 165, 180,190,206,514,562 Grisch, M. 455 Gröber, G. 49, 61, 129, 179, 188, 190, 329, 336, 460, 478, 545 Grunmann, M. 555 Gsell,0.591 Gudschinsky, S.C. 88 Guillaume, G. 59, 77 Guiraud, P. 72, 79, 148 Gulsoy,J.87,105, 121, 349 Gyarmathi, S. 65 Gysling, F. 454 Haas, M.R. 501 Hahn, E.A. 63 Haisland, N. 63 Hakamies, R. 436 Hall, Robert A. (Jr.) 80, 89f., 98, 190, 340 Hamlin, F.R. 421, 566 Hammersfahr, A. 414 Hansen, MX. 323 Hanssen, F. 95f., 101, 105, 107, 110, 114, 116, 118, 120-2, 124, 162, 173,

598

INDEX OF NAMES

189, 194, 209, 238, 244f, 248, 280, 283, 289, 332, 353, 380, 382, 384, 422, 433, 444, 461, 467, 483, 519, 534, 554, 575 Harris, J.W. 236, 248, 484 Harris, Z.S. 60,71 Hartmann, P. 80 Hasselrot, . 20, 77, 142, 144, 149, 160, 164-71, 174f., 229, 248, 412, 549, 562, 567 Hatzfeld, H.A. 562, 565 Haugen, E. 138, 145, 150, 566 Haust, J. 58 Haverkate, W.H. 445 Havet,L.481,556f. Hegel, G.W.F. 69 Heilmann, Լ. 77 Heizer, R.F. 89 Hendriksen, . 75 Henríquez Ureña, P. 59, 80, 106, 110, 118 Henshaw, H.W. 59 Heraeus, W. 469 Herasimchuk, E. 231 Herbillon, J. 170 Hermann, E. 221 Herodotus 70 Herzog, E. 481 Herzog, M.I. 564 Hills, E.C. 106 Hjelmslev, L. 63, 68, 71, 75, 148, 268, 502 Hock, H.H. 232 Hockett, CF. 90 Höfler, M. 195 Hoenigswald, H.M. 49, 466, 561 Hofmann, J.B. 556f. Hofmannsthal, H. von 562 Hoijer, H. 60, 90, 326 Hollyman, K.J. 139 Holmes, U.T. (Jr.) 203 Hope,.. 11,137-51,566 Horning, A. 336, 343, 414, 518, 534 Householder, F.W. 141,150 Hrozný, . 69 Huber,J.117f., 121,463

Hubschmid, J. 259, 421, 426, 431, 435-7, 440, 471, 584 Huidobro, E. de 107 Humboldt W. von 26-28, 58, 61, 67f., 70, 148 Hymes, D. (H.) 57, 59, 64-6, 87f., 217, 591 Ineichen, P. 160 Iordan, I. 78, 83, 217, 328, 569 Ipsen, G. 148 Isidore (-rus) of Seville 68 Ivić,M. 50,74,82, 561 Jaberg, . 64f., 101, 123, 159, 169, 173,218,412,471,562 Jacobsen, W.H. (Jr.) 233, 248 Jänicke, . 196 Jakobson, J. 164 Jakobson, R. 30, 60, 63, 65, 68, 72, 74, 502f., 575 Janer, F. 111 Janner, H. 433 Jasanoff, J.H. 246, 248 Jespersen, . 23, 29, 63, 70, 76, 173, 221,344 Jones,W.(Sir)61f.,67,69f. Joos, M. 64 Joret, . 481 Joukovsky-Micha, F. 212 Jud, J. 59, 335, 351, 433, 470, 522f., 526, 534, 557. Jungemann, F.H. 434, 445 Kahane, H. (R.) & R.T. 87, 100, 142, 144, 150, 160, 168-70, 174, 423, 549, 562, 567, 585 Kany, C.E. 335 Karlsson, K.E. 579, 584, 588 Keller, Α. 96 Keller, H.E. 329 Keller, J. 100f. Keniston, H. 562 Kent, R.G. 556 King, E.L. 455 Kiparsky, P. 232 Kirchhoff, A. 68

INDEX OF NAMES

Kjellman, . 185 Klare, J. 195 Kloeke, G.G. 220 Klose, F. 209 Kluge, F. 344, 500, 544 Knapp, W.I. 461 Knust, H. 339, 357, 544 Koberstein, G. 101, 111 Kock, Α. 76 Koenig, W. 330 Körting, G. 306f., 460, 478, 518, 520, 522, 524-6, 534 Křepinský, M. 257, 259 Kretschmer, P. 548 Kroeber, A.L. 58-60, 65f., 89, 138 Kronasser, H. 72, 138 Krüger, F. 118,270,433,465 Kruszewski, M. 75 Kuersteiner, A.F. 120, 242, 358 Kuhn, A. 101,169 Kuhn, T.S. 57,224 Kukenheim, Լ. 83, 560 Kuryłowicz, J. 69, 71, 75, 88, 562 Kvavik, K.H. 143, 150, 171, 175 Kypriotis, C. 323 Labov,W.231,564,569 Lafuente  Alcántara, E. 122 La Grasserie, R. de 168 Lamano  Beneite, J. de 122f., 359, 461,465 Lang, H.R. 106, 565 Langdon, M. 10,49,561 Langendoen, D.T. 561 Lapa, M. Rodrigues 331, 350, 352, 436, 552,555 Lapesa, R. 96, 100, 237, 248, 259, 335, 417, 434, 476, 534, 574, 588, 584f. Larramendi, M. (de) 519-21, 534 Lathuillère, R. 460 Laudiert, F. 122 Lausberg, H. 60, 212, 238, 275, 290, 412,437,490,563 Lazar, M. 359 Le Blant, E. 547

599

Lecoy,F.327,445, 554, 583 Le Gentil, P. 434 Lehiste, I. 77, 87, 540f., 589 Lehmann, W.P. 83, 98, 90, 218, 233, 248, 564 Leibniz, G.W. 70 Lené, G. 189-91, 193, 197, 204, 206, 545 Lenz, R. 69f. Lepschy, G.C. 78-82, 231, 561 Leroy, M. 50, 68-71, 73f., 82 Leskien, A. 55, 67f., 70 Levy, A.K. 138, 150, 165, 175, 183, 212, 357, 431, 467f., 552, 566 Levy, E. 72 Lévy-Bruhl, Լ. 77 Lévy-Strauss, . 77 Lewis, .. 180 Lida, R. 87, 455, 577 Lida de Malkiel, M.R. 59, 105, 111, 121, 207, 212, 333, 349, 352, 444, 582 Lihani, J. 552, 554 Lindqvist, 77 Lindsay, W.M. 198 Lindvall, Լ. 584 Littlefìeld, M.G. 520 littré, É. 23, 28, 33, 307-10, 312, 318, 324,518 Livet, C.-L. 306,318 Livingston, C.H. 351, 355 Llatas,V. 121,349 Llorente Maldonado de Guevara, A. 270 Lloyd, P.M. 87 Loewe, G. 548 Lohmann, J. 72 Lombard, Α. 98-100 Lommatzsch, E. 413-5, 548, 551 López-Morillas, . 328, 330, 333, 337f., 350, 589 Lozinski, G. 190 Lüdtke, J. 562 Luria, M.A. 465 Luther, A.M. 139 Lutta, C.M. 454 Lyons, J. 138,150

600

INDEX OF NAMES

Ma 231 Mackenzie, F. 140, 150, 565 Mätzner, Α. 307, 318 Magne, Α. 93, 104, 116f. Maher, J.P. 535f. Mahn, K.A.F. 422 Malmberg, . 50, 76, 78, 82f., 140, 561 Malone, K. 63, 68 Mandelbaum, D.G. 59f. Mańczak, W. 379 Marchand, H. 576 Marden, C.C. 114, 120, 311, 318, 358f., 552 Marouzeau, J. 31, 138, 166f., 174, 345 Marr, N. Ja. 69, 74 Marroquim, M. 123 Martineau, R. 67 Martinet, A. 70, 74, 80, 138, 148, 233, 249, 506f., 561 Marty, A. 75 Mathesius, V. 69 Matoré,G.72, 148, 150 Maurer, T. 132 McCartney, J.M. 123 Mead, M. 59 Megiser, J. 70 Meier, H. 121,421,433,435 Meiklejohn, M.F.M. 139 Meillet, A. 40, 61, 63f., 68, 70, 72f., 77,96,130,138,145,173,187, 192f., 195, 197, 200f., 209, 254, 266, 287, 336, 346, 351, 354, 359, 362, 364f., 375, 377f., 411, 429, 432, 436, 440, 442, 480, 482, 518, 556-8 Meinhof,  165 Melander, J. 229, 249 Melchiori, G.B. 455 Meilor, G. 92 Ménage, G. 565 Menéndez Garcia, M. 274 Menéndez Pidal, G. 89 Menéndez Pidal, R. 10, 16, 19, 65, 68, 77, 100-3, 110, 112f., 118, 120-2, 124f., 147, 190-2, 196, 199, 209-11, 213, 217, 224, 232f., 235-9, 247, 249, 253, 255, 259, 269-75, 377f.,

283f., 286-9, 291, 293, 295, 332, 335, 337, 342, 349, 353, 355-9, 371, 382-5, 389f., 393, 420, 425-31, 433f., 437, 441, 443f., 446, 451, 458, 461, 464, 473, 475, 479, 481, 517,521,525f., 528, 531,534, 548,551,562,564,571,575-7, 584, 589 Menocal, M.R. 585 Meringer, R. 30 Merk, G. 195 Merlo,  191 Meyer, G. 539 Meyer-Lübke, W. 10, 28f., 32, 49, 65, 73, 88, 96-8, 102, 117f., 131, 139, 145, 150, 159f., 173f., 179, 182-92, 195-7, 200, 202f, 206, 209, 222, 233, 236, 245, 249, 253f., 256f., 266, 270, 287, 292, 300, 306, 308, 310, 312f., 318, 323, 327, 329, 333, 335-8, 342f., 346, 350-5, 359, 362, 371, 376, 379, 382, 387f., 41lf., 414, 416, 420, 422f., 429, 432f., 437-9, 446, 454f., 457, 460f., 467, 469, 478-83, 489-92, 510,518,521-3,531,534,545-8, 549-52, 555, 557, 563, 565 Michaelis de Vasconcelos, . 103, 116, 118, 121, 157, 245, 249, 324, 357, 420-5, 461, 483, 485, 520, 525f., 534, 545 Michaëlsson, . 77, 170, 230, 249 MigHorini, . 142, 150, 182, 190, 196f., 200, 206, 304, 315, 318, 354f., 362, 364-6, 455, 480, 550, 562, 582f. Mignot, X. 466 Miklosich, F. 187,562 Miliarder, G. 101 Mistral, F. 466 Mitterand, H. 182, 213, 325, 331, 491, 543 Mitzka, W. 344, 544 Moignet, G. 345 Moldenhauer, G. 335, 350, 355, 357 Moll, F. de . 97 Møller, . 266

INDEX OF NAMES

Mondéjar, J. 466 Monfrin,J.413 Mongelli, G. 400 Monlau, P.F. & J. 461, 477f., 518-20, 534 Montes Giraldo, J.J. 445 Montgomery, T. 87, 105f., 124, 231, 240, 249, 360, 527, 534, 576, 580 Monti, P. 454 Montoliu, M. de 159, 173, 521 Morel-Fatio, A. 96, 101, 113, 270, 289, 293, 349, 422, 464 Morley, S.G. 88 Mounin, G. 83 Mourin, L. 445 Müller, M. 70, 75 Múgica, P. de 107, 122,270 Muñoz y Romero, T. 120 Munthe, Â.W. 107, 118, 121, 157, 262, 420, 422, 433, 465, 489 Mussafia, A. 114, 324, 335, 362, 454 Nascentes, A. 123 Nauton, L. 466 Navarro, F.B. 523 Navarro (Tomás), T. 38, 62, 230, 236, 249, 258, 262f., 376, 430, 560, 562 Nebrixa, E.A. de 68, 110, 257, 552 Neka Martínez, J. 270, 465 Nelson, D.A. 354, 360, 520, 527, 535, 580 Neto, .. de M. Lino 351, 438 Neto, S. Silva 351 Neumann, F. 336, 343 Neuvonen, E.K. 102, 137, 150, 304, 318,352,439 Nicholson, G.G. 412 Nida,E.A.81,138 Niedermann, M. 40, 197, 255, 266, 440, 556 Nigra,  550 Nilsson-Ehle, H. 310-318 Nobiling, O. 524f., 529, 535 Nogueira, R. de Sá 351 Noreen, Α. 76

601

Northup, G.T. 58 Nunes, J.J. 108, 114-6, 350-2, 435, 437,439f.,461 Núfíez, Η. 465 Nykl, A.R. 338 Nyrop, . 68, 97, 192f., 306, 308, 310,312f.,318,343f.,490 Oakley, K.P. 89 O'Bryan, M. 232, 249 Oelschläger, V.R.B. 100-2 Oliver Asín, J. 192 Onions, .. 543 Onís, F. de 115, 120,122,124 Orr, J. 67, 148, 217, 223, 328, 351, 551 Osthoff, H. 55,70 Otero, .. 231, 574 Oudin, . 500, 554 Pagliaro, A. 80 Palencia, Α. de 552 Panini 69 Paris, G. 11, 96, 179, 185, 190, 192, 204, 206, 266, 308, 375, 378, 518, 545-8 Parker, E. 80 Parodi, E.G. 67, 362, 547, 555 Patriarchi, G. 455 Paul, H. 5, 27, 55, 70, 500 Paz y Melia, A. 271 Pedersen, H. 26, 76, 83, 217, 335 Pellegrini, G.B. 118, 477 Pel(l)etier, J. 306 Pensado, J.L. 117,121,551 Percival(e), R. 554 Pereira, M.P. da Silva 433 Perry, T.A. 456 Peter, H. 148, 150 Pfìster,M. 581 Pharies, D.A. 588 Philipon, É. 433 Pichon, É. 156, 168, 172, 196, 304, 319 Piel, J.M. 139, 150, 158, 195, 308, 319, 437, 440f., 490, 523,535 Pietrangeli, Α. 168, 175 Pietsch, . 58, 334, 350, 352, 356

602

INDEX OF NAMES

і,.. 81 Pires de Lima, A.C. 351 Pirson, J. 547 Pisani, V. 71, 138, 378 Plato 70 Plattner, P. 306, 309 Poghirc, . 569 Politzer, R.L. 92 Pollak, H. 80 Polt,J.H.R.83,456 Ponce, P. (Fray) 62 Pope, M.K. 413 Popescu-Marin, M. 445 Porzig,W. 148 Porzio, S. 420 Posner, R. 41, 87, 92, 147, 150, 231, 266f., 283, 376f., 380, 466, 468, 548,560,569,571,581,588 Post, A.C. 97, 106, 110 Postel, G. 70 Postgate, J.P. 27 Pott, A.F. 29, 265, 562 Pottier,B. 168,437 Powell, J.W. 59 Prati, A. 190 Pratola,D.J. 137, 150, 565 Pratt, O. de 157, 173 Puhvel, J. 416 Pulgram, E. 562 Purczinsky, J. 574 Puşcariu, S. 187, 209, 304, 319, 547 Rajna, P. 557f. Ramírez de Carrion, M. 62 Rask, R. 26, 28, 63, 65, 68, 70, 76, 514 Rato y Hevia, A. de 107, 121, 272, 275-7, 279, 287, 289, 291f., 294f., 359, 465 Reid, T.B.W. 413 Reinhardstoettner, C. von 96, 116, 436, 468 Renan, E. 69 Renson, J. 329 Rey, Α. 122,358 Reyes, Α. 59 Reynolds, . 400

Rheinfelder, Η. 80, 92, 132, 334 Ribezzo, F, 433 Rice, C.C. 203 Richards, LA. 138 Richardson, H.B. 211, 241, 249, 358, 485, 554 Richelet, P. 454 Richter, E. 61, 88, 103, 185, 301, 562f. Riegler, R. 23 Rigutini, G./Bulle, . 162 Ringenson, K. 76f., 457 Rippere, . 330 Riquer, M. de 460 Robins, R.H. 57, 83 Rochet, . 231f.,250 Rodinson, N. 335 Rodríguez-Castellano, L. 270, 465, 577 Rönsch, H. 182, 202, 460, 469 Rohlfs, G. 60, 98, 102, 168, 170, 194f., 341, 368, 370, 375, 380, 400, 421, 435, 437f., 549,590 Roques, M. 123,470 Rosal, F. del 19, 525, 535 Roselblat, Á. 67, 97, 108, 118, 122, 250, 360 Rosetti, Α. 99, 466 Rosset, T. 306 Rothwell, W. 466 Roudü, J. 388 Rousseau, J.-J. 69, 71 Rozwadowski, J. 163, 173 Ruhlen, M. 232 Rydberg, G. 343 Saco Arce, J.A. 468 Sajnovics, P. 65 Sala, M. 232, 250 Salvioni,  67, 266, 376 Sánchez, T.A. 96, 353, 459, 520, 535 Sánchez de la Ballesta, Α. 551 Sandfeld, . 146, 151, 266 Sapir, E. 5, 26, 58-60, 66, 68, 71-3, 75,82,130,145,151,187,297, 316, 501f. Saporta, S. 141, 150 Sarmiento, (Fray) M. 551

INDEX OF NAMES

Saroïhandy, J. 293, 351 Sassetti, F. 68, 70 Saussure, F. de 27, 30, 55, 60f., 64f., 68-71, 73-7, 80f., 138, 198, 221, 359, 375, 500f., 503, 557 Scaliger, J.J. 69f. Schach, P. 138 Schaff, A. 79 Schaffer, M.E. 583 Schalk, F. 456 Scheler, Α. 336, 340, 460, 478, 518, 535, 548 Scheludko, D. 309f., 312, 319 Schick, CM. 79 Schlegel, F. 67, 70 Schleicher, A. 26, 67, 70, 72 Schmid, Η. 237, 250 Schmidt, J. 29, 68, 158, 173, 217, 375 Schneider, M. 163f., 168, 174 Schön, I. 186 Schorta, A. 455 Schuchardt, H. 28, 30, 40f., 58f., 65, 70, 73, 75, 96, 122, 217, 219, 224, 238, 266f., 323f., 330, 420, 438, 516,520,526,535 Schulten, A. 433 Schultz-Gora, O. 97, 332f., 340, 411, 456 Schulze, W. 436 Schwan, E. 336, 341, 438 Sebeok, T.A. 50, 58, 60, 65, 83 Sechehaye, A. 75 Segre, . 67 Seidenspinner de Núñez, D. 234, 250 Sequeira, G. de Matos 433 Serrano y Sanz, M. 101 Sevilla, Α. 121 Šaxmatov, A.A. 65 Sandru, D. 99 Ščur,G.S. 148, 151 Short, C. 180 Siertsema, . 148, 151 Siesso  Bolea, J. de 519, 535 Silva, A. de Morais 544, 552 Silveira, J. da 434 Sjöstedt, G. 77

603

Skeat, W.W. 141,157,324 Sklar, R. 57 Solà-Solé, J.M. 234, 250 Solalinde, A.G. 100,523 Sommer, F. 198 Sommerfelt, Α. 61, 63, 71, 76, 266, 471 Spargo, J.W. 83, 217 Spaulding, R.K. 120 Spengler, . 24 Sperber, Η. 138 Spitzer, L. 59, 61, 63f., 73, 106, 165, 167-9, 174, 190, 195, 230, 250, 306, 309f., 312, 314f., 319, 323, 333, 339, 436, 491, 523f., 526, 535, 551 Staaff, E. 115, 118, 190, 206, 230, 238, 250,289,531 Steene, A. 138 Stefanini, R. 87, 148, 151, 231, 444 Steiger, A. 102, 138, 151, 294, 357 Steinthal, H. 49, 61, 67f., 75 Stender-Petersen, A. 221 Stephany, U. 65 Storm, J. 340, 362, 420, 429 Streitberg, W.49f., 61, 70 Sturtevant, D.C. 59, 63 Sturtevant, E.H. 63,71 Subak, J. 456 Suchier, H. 323 Svedelius, C. 76 Szemerényi, O. 231 Tagliavini, C. 80, 83, 454 Tallgren [-Tuulio], OJ. 522 Teriingen, J.H. 137, 149, 151 Terracini, . (). 66, 71, 433 Tesnière, L. 77 Thomas, A. 266, 545 Thomsen, V. 26, 83 Thurneysen, R. 375, 466, 520-2, 526, 535 Tiktin, H. 99, 187 Tilander, G. 76, 105, 311, 319, 350, 437 Tiraboschi, A. 455 Tobler, A. 11,49, 170, 179, 190f.,

604

INDEX OF NAMES

306-10,312,315,319,413-5, 480,490,545,551 Todd, H.A. 461 Togeby, . 20, 41, 87, 186, 231, 240, 250, 266-8, 354, 375, 378, 466, 527,535,581 Tognelli, J. 79 Toledo, O. de 122 Tollemache, F. 190f. Tommaseo, N./ Bellini, . 162 Toole-Kahane, R. 120, 160-4, 168, 170, 174 Torre, A. de la 121, 294 Torres Fornés, . 121, 293, 351 Torres Rioseco, A. 59 Tovar, Α. 435 Toynbee, A. 24 Trager, G.L. 77 Trier, J. 72, 148, 151 Trombetti, A. 433 Trubetzkqj, Ν. 29, 63, 65, 68, 74, 77,81 Tuchel, H.G. 466 Tuttle, E.F. 15, 333, 536-9, 589 Tynjanov, Ju. 57 Ubieto,A. 351 Uitti, K.D. 87, 109, 231 Ullmann, S. 72, 79, 139, 145, 148, 151,160,590 Umphrey,G.W. 113,293 Unamuno, M. de 429, 461 Ungarelli, G. 455 Vaganay, H. 306,310 Valdés, J. de 110, 118f. Valesio, P. 378f. Valkhoff, M. 137, 146, 151 Valmar, marqués de (L.A. de Cueto) 119,352 Varón Vallejo, E. 464 Vàrvaro,A.83,297,319 Vasconcelos, J. de 422 Vasconcelos, J. Leite de 103, 106, 108, 117f., 121,422,426,439, 463

Vaugelas, C.F. de 70, 306 Velleman, Z. 455 Vendryes, J. 61, 63f., 68, 498, 501 Veres d'Ocón, E. 353, 463 Vergara, (Fray) S. de 519 Verner, . 35, 70, 76, 251, 342, 382384, 389 Viana, A. dos R. Gonçálvez 157f., 173,423,433,464,466 Vico, G. 68, 70 Vidal de Battini, B.E. 351 Vidos, B.E. 137,142,151 Viëtor, W. 70 Vigón,B.295,334,420,439 Vihman, M.M. 274 Viterbo, J. de S.R. de 360 Voegelin, C F . 66 Vogt, H. 138, 145, 151 Vollmöller, K. 122 Voretzsch,K. 341 Voßler, K. 55, 64, 71, 73, 77, 140, 151 Wackernagel, J. 165 Wagner, M.L. 98, 101, 122, 194, 196, 253,349,351f.,438,464 Wagner, R.-L. 345 Wahlgren, E.G. 38f., 229-32, 247, 250, 262f., 376, 430 Wald, L. 83 Walpole, R.N. 89 Wang, W. S-Y. 235, 248, 250 Wartburg, W. von 31, 71, 131f., 146, 148,151,159,161,168,173,175, 195, 213, 292, 325, 329, 333, 336, 343, 350, 415, 436, 441, 454, 461, 467,491,539,548f. Waterman, J.T. 70, 75, 83 Wedgwood, H. 460 Weinreich, U. 79, 82, 138, 141, 145, 151,503 Weisgerber, L. 77, 148 Wellek, R. 63 Wells, H.G. 24 Westermann, D. 76 Wexler,P.J. 148, 151

INDEX OF NAMES

Whitney, W.D. 26f., 29f., 33, 67, 499f. Whorf, B.L. 58, 82 Wiese, . 213, 337, 368, 375, 400 Wiggers,J. 104 Wilbur, R. 232 Williams, E.B. 115, 117, 121,276, 285,492 Willis, R.S. (Jr.) 114, 120, 357, 360, 464 Wind,B.H. 137, 151 Windisch, E. 138 Woelfflin, E. von 377 Wolf, HJ. 183

Woodbridge, B.M. (Jr.) 88 Worthington, M.G. 87 Wrede, F. 77 Wuest, Α. 483 Wundt, M. 70, 75 Zamora Vicente, Α. 59, 270, 276 Zauner, Α. 108, 115, 118,238,433, 444,468,531 Zeitlin, M.A. 233, 242, 250, 358, 486 Zeuner, F.E. 88 Zsirai, M. 65 Zvegincev, V.A. 83

605

SELECTIVE INDEX OF KEY TERMS abandonment (of forms, ideas) 379, 504, 572, 587 aberrancy 9, 11, 30, 33f., 59, 102, 199, 208f., 230, 232, 247, 392, 446, 457, 516,526,529 ablative case 307, 358 abnormality 30, 306, 362, 371, 419, 436 abolishment of sound laws 219 abrasion 342 absence (of a feature, or form) 404, 552f. absolute (fact, degree, chronology) 90, 342, 563, 584 absorption (into the stem, etc.) 286, 307, 348 abstract (adjectival, verbal) 179-81, 183, 185f., 189, 192-7, 200, 204-7, 209, 214, 256, 259, 272, 274, 278, 306, 308-12, 316, 334, 349, 368, 383, 386, 389, 394, 399, 401, 403-6, 412, 442, 453f., 457, 462, 482, 485, 518, 543, 545f., 552, 573, 575, 578f. abstraction, level of 6, 57f., 280, 500, 536, 539 abundance 432 academic (tone, traditions, discourse) 52, 78, 90, 268, 588 acceleration (of a trend) 103, 208, 336, 438,447,481 accent mark 306 accentology 14, 343, 498 accent shift 36f., 99, 257, 259, 261, 272, 344, 366 accentual structure 17, 112, 199, 208, 253f., 259, 335, 369f., 399, 411, 419, 423, 433, 443, 458, 466, 554, 576 acceptance, ready 113 accessibility (of area, source) 109, 301

accident 95, 204, 222, 261, 317, 348, 357, 362, 372, 412, 459, 466, 515, 538 accretion 426, 467 accumulation of variants 11, 94 accuracy 261, 305, 308, 342, 437, 468, 473, 499, 523 accusative with infinitive 133 acephalous form 334, 352 "acoustic equivalence" 102, 263, 430 acoustics 78, 277, 288, 390, 536 action noun 171, 179, 198, 206, 246, 399, 406, 409 active (partner, verb) 199, 458 actor (vs agent) 205 adaption (of loans, etc.) 38, 115, 261 ad hoc conference 60 adjective 95, 137, 156, 185, 192, 251f., 254, 256, 258f., 263, 271, 279, 305, 313f., 358, 372, 400f., 403, 411, 428, 457, 462, 476, 481, 484, 486, 489, 494, 553, 556, 575f. adjoining (or adjacent) dialect area 105, 111, 118, 158, 210, 224, 281, 333, 543, 578 "adjunct" 584 adjustment (primary, secondary) 9, 91, 93f., 205, 229, 305, 356, 364, 428, 435, 438, 443, 474f., 525, 543, 545 administrative unit 53 admiration, signaling of 489 admixture 32, 119 adoption 347, 454 adornment 214, 427 adstratum 31f. advance of knowledge 223, 225, 229, 426,519,436,560,572 advanced (stage, variant) 115, 120, 241, 243, 298, 304, 306, 359, 383, 387, 417,455,474,490,495,558

608

SELECTIVE INDEX OF KEY TERMS

advances and retreats 26, 31, 34, 52 advantage (yielded by a change) 192, 214, 261, 286, 311, 337, 380f.; (to a conjecture) 201, 299, 309f.,419 adventitious (factor or result) 254f., 289, 380 adverb(ial) 103, 105, 201, 293, 302, 305, 309, 311-6, 411f., 470, 510, 575, 579 adverbializer 415 adverbial comparative 413, 415 adverbial phrase 315, 372, 476 adversative 327, 355, 404, 481 affectivity 425, 432 affinity 35, 65, 106f., 112, 146f., 189, 201, 234-6, 239, 244, 251, 253, 276, 285, 288, 294, 303, 328, 348, 363, 366f., 371, 378, 381, 395, 404, 407, 442,473,515,526,567 affixation 5, 7f., 34, 39, 189, 193, 198, 262, 264, 315, 400, 433, 447, 532, 537, 566, 568, 573 , double 15, 411, 413, 476 affricate 15, 392-4, 407, 440, 573 age, measurement of 88 Age-and-Area hypothesis 33, 73, 221 agency (of forces) 94, 125, 131, 297, 300, 314f., 347, 367, 377, 462, 467, 472, 474, 505, 516, 572, 578 agent of transmission 81 agentive 137, 145, 167, 179, 182, 192f., 196-8, 204f., 206, 335, 349, 370, 412,446,479,486,567 agglutination 41, 267 aggrandizement 172, 489 aggregate of forces, 241, 268 agnosticism 308, 371 agronomic words 508 ahistoricism 499 "ailing word" 17, 103,451,462 allegory 170 allegro form 392 alliance of forces 6, 38f., 100, 104, 184, 233, 235f., 262f., 282, 299, 314, 317, 352f., 405, 407, 458, 516, 554,

570 alliance of languages (Sprachbund) or dialects 130, 289, 413 allo-graph,-morph,-phone 38, 202, 262, 390, 527 allotrope 374 almost-learnèd transmission 41 alphabetic order 305 alternation (of forces, vowels) 112, 119, 121, 124f., 235, 239, 241f., 245, 262, 272, 297, 405, 463, 468, 485, 504, 511, 513, 517, 520f., 527, 552, 571 alternative (to an idea or a process) 15, 102,107,114, 119,139,142,162, 172, 180, 182, 189, 198, 201f., 222, 230f., 241, 251, 258, 268, 283, 286, 302f., 312, 316f., 326, 328, 332f., 335-7, 339, 345, 354, 356, 359, 364, 373f., 378, 389, 406, 408, 426f., 437f., 443, 445, 463, 467, 469, 476, 480, 491, 519, 521, 523f., 525, 543, 550, 554, 557, 560, 563, 573 amalgam (=merger) 100, 268, 278, 280, 287, 331, 374, 417, 428, 469, 471, 494, 503, 550 amateurs 24, 118, 167, 473 ambigeneric 40, 158, 163, 264f. ambiguity 38, 96, 106, 115, 145, 166, 222, 261, 297, 309f., 312, 337, 385, 457, 479, 498, 524, 584, 586 ambit, semantic 203, 454, 476, 551 ambivalence 479 amenability (to change, observation, etc.) 236, 301, 326 American linguistics 81 amorous parlance 167 amorphousness 77, 261 amplitude (lexical) 430 amusement 17 anachronism 63, 141, 500 anagram 501 analogy 39, 90-3, 99, 104, 107, 117, 119, 124, 197, 206, 219, 221, 229f., 241, 246, 255, 262f., 265, 274, 284, 290, 298, 304, 314, 328, 357, 364,


370, 375f., 379-81, 415, 469, 488, 492, 500, 564, 570-2, 574 analysis 5f., 23, 316, 324, 331, 341, 344, 357, 400-7, 427, 445, 468, 470, 472,478,501,530,548,585 analytic biography 63 "analytical" grammar 130 anaptyxis 422, 424 anatomical (imagery, term) 169, 197, 384, 392, 481, 483-7, 489f., 491-3, 514, 552f. ancestral culture (language, form, family) 56, 94, 104, 131, 157, 187, 192, 201, 210f., 299, 316, 324, 336, 341, 431, 473, 479, 504f., 523 ancient science 165 ancillary (vowel, status, operation) 29, 338, 502 anecdote 317, 504f., 557, 587 Anglophone 141 animal name 161, 169f., 264, 402, 443, 484 animate being 155, 165, 169 animization 169f., 172 anisomorphism 311 annotations 80 anomaly 9, 165, 256, 316, 537 anonymity (of copyists) 123 antagonism (among linguists) 64, 73 antecedents, historical 44, 182, 575, 586 — of phonemics 69, 460 antedating 427, 554 antepenult 288, 419, 434, 441, 444, 585 anteriority 422, 452 anthology (of linguistics) 50, 64 anthropological linguistics 53, 66, 88, 564 anthropology (esp. social) 89, 147, 516 anthropomorphism 486 anthroponymy 27, 77, 271, 426, 431 anticipation 278, 292, 330f., 346f., 458 antihiatic function 284f., 352, 376, 427, 430, 467 anti-homonymy 456

609

antinomies in linguistic theory 71, 75 antiphilological bias 187 antiquarianism 464, 473 Antiquity 49, 56, 68, 360, 390, 420, 441, 454, 483, 531, 558, 565, 585 antisubstratism 436 antonym 421, 460, 515 apathy (in reactions) 91 apheresis 90, 323f., 328f., 332, 335-7, 339, 341, 345, 347, 355, 453, 468, 477, 558 apocope 90, 185, 238, 312, 580 apocryphal writings 71 apophony 5, 180,286,354 apparatus 80 apparent incongruity 220 appeal (of novelty, variants, etc.) 246, 259, 289 appearance, first 393, 400, 461 appendage 420, 442 appliances, names of 433 apprenticeship and training 4, 9 approximation (in dating, analysis) 91,125,132,144,471,564 aprioristic assumptions 72 Arabic veneer 192, 203, 446, 468 arbitrariness (of judgment, sign) 71, 76, 79,156,160,479,499 arbitration of disagreement 422 archaic (imagery, phraseology) 33, 147, 192,208 archaic(ity) 89, 94f., 106, 234, 253, 274, 293, 309, 331, 334, 359, 377, 390, 412, 414, 416, 427, 429, 437, 455, 464, 469, 477, 482f., 557, 584,591 Archaizers vs. Innovators 57, 557 archeological approach 89, 189, 474, 509, 588 archetype 115,197,429,508 architecture (of grammar) 113, 204, 261 archival research 50, 53, 63, 295 areal (distribution, linguistics, pattern) 56, 87f., 219, 221, 329 argot 167, 185

610


argument (clinching, cogent) 386, 388, 390,427,466 aristocratic speech 32 army milieu 201 arrhizotony 277, 459 article (definite) 130, 467f. articulation 143, 211, 244, 317, 341, 344,381,390,428 articulatory phonetics 77, 236, 277 artificial (word, language) 29, 134, 142 artisans' tools, names of 165, 193f., 199, 204f. artistry, verbal 88, 147, 333 ascendancy of a language 133 ascending diphthong 389 Ascoli's Law 209, 323f., 327, 340-3, 349, 354, 458, 580 aspect 104, 580 aspiration 547, 549 assessment of evidence 94 assibilation 289, 411, 427 assimilation 38, 261, 263, 335f., 341, 371,379,438 assistance (from a concomitant) 390 association 77, 112f., 159f., 165-7, 186, 190, 192, 202, 209, 213, 229f., 234, 241, 262, 265, 302, 330, 332f., 340, 351, 358, 371, 374, 378, 408, 415, 422, 442, 453, 467, 473-5, 492, 494, 497, 515, 527, 537, 548, 551, 557 associative interference 36, 38, 117, 462 asymmetry 38, 149, 233, 247, 261, 476,481 asyndeton 315 athletics and sports, domain of 12, 24, 185 atlas, dialect 141, 159, 161, 164, 168, 218,220,503 atmosphere, social 200 atomism 4, 141, 189, 194, 495 atomicity 326, 340, 384, 419-21, 426f., 431, 433f., 441-3, 576f. atrophy (of a device) 130, 338 attestation in texts 201, 211, 303, 458, 475, 483 attraction 112, 243, 253, 300, 307,

348, 357, 378, 407, 443, 453, 461, 407, 470, 504, 511, 551 attrition, rate of 14, 33, 64, 87, 125, 304, 564 atypicality 179, 198, 304, 383, 385, 406, 517, 526, 570f., 581 auditive phonetics 381, 390 argument 16, 247, 272, 278, 361, 377, 419-21, 423, 428, 433, 440-3, 445-7, 576,581,584 argumentative 40, 143f., 156f., 160, 162-70, 265, 407 austerity (of grammar) 124 authenticity of records 108, 114, 117, 144, 168, 181, 187, 359f., 336, 350, 359, 377, 383, 438, 453, 460, 482, 527, 546f. authority, philological 44, 118, 306, 312, 316, 382, 419, 436, 455, 463, 468,474,482,504,551,575 author's references 88 autobiographies of linguists 58 autochthony 259, 289, 428-30, 442 autodidactic concern with linguistics 68 automatic reaction 290 autonomy (of sounds, of a discipline, of a milieu)44, 75, 78, 90, 97, 185, 222, 260, 265, 298, 312, 315, 347f., 494, 557f. auxiliary verb 379, 529 availability (of channels, resources) 143, 156, 247, 294, 326, 337, 393, 404, 421,463,553 avant-garde 54 average (rate, speaker) 87, 144, 168 avoidance (of forms, of homonymy, of polysemy) 112, 219, 222, 240, 243, 311, 315f., 336f., 386, 390, 504, 572, 583, 586 awareness, degrees o f . . . 146, 200, 202 axiom 218, 267 axis (of time, space) 354, 528 babbling (of infants) 69 back-information 185, 203, 328, 375 background, backdrop 179, 187f., 192,

SELECTIVE INDEX OF KEY TERMS 195, 208, 220-2, 236, 274, 283, 290, 311, 316, 324, 334, 338, 347, 358, 367, 386, 403, 430f., 435, 452, 551 "backlash", lexical 143, 185, 547 back vowel 96, 111, 114, 243f., 294, 328, 332, 338f., 342 balance 163, 198, 217, 223, 231, 328, 405; ֊sheet 172, 221, 316 base 157, 187, 260, 283, 291, 327, 331, 360, 363, 392 behavior, patterns of 370, 373, 409 belated (= delayed) change 186 belles lettres, acquaintance with 5 benefit, fringe 239, 307, 309f., 312 bias 72, 76, 187 bibliography 10, 63, 74, 80, 108, 139, 167, 282f., 302, 330, 376, 381, 433, 436 bicephalous (word-family) 243 bicuspid 403 bidirectional 39, 100, 132, 137, 263, 353, 370 bifocal approach 19 bifurcation 43, 76, 98, 202, 253, 286, 301, 324f., 336, 344, 379, 440, 457 bilabials 12, 36, 234, 237, 281f., 333, 337,365,420,547f. bilinearity 100, 102 bilingual(ism) 37, 42, 66, 143, 139-42, 144, 162, 207, 261, 297, 302, 316, 342 binarism 170, 172 binomial, (ir)reversible 42, 164, 316, 416 biography (of a scholar) 59, 63; - (of a word) 222, 229, 241f, 266f., 304, 333,346,386,435,451 biological (approach, model) 33, 73, 159, 166 biphonemic 338 bird name (= ornithonym) 346 birth of language 70 bizarre forms 96, 114, 198, 222, 225, 234,361,370 blend, lexical 17, 36, 41, 267, 277, 294, 333f., 346f., 352, 372, 375-7, 384,

611

411, 415, 417, 426, 429, 436, 438, 547f. blind alley 77, 381; - force 271; - law 222,314 blockage 121, 200, 206, 222, 237, 256, 258, 306, 339, 352, 361, 366, 378, 383 block (of words) 400-3, 405, 412 bloom, forms in full 408 blossoming (of forms), late 192 blurring 189, 198, 258, 264, 306, 328, 331,353,380,383,390,442 boldness of change 333 bond, semantic 294 book review, extended 60, 80 bookish words 252 border 292; ֊ dialect 282; ֊ line 271, 314, 403, 442; — zone (between disciplines) 229 borrowing, lexical 11, 15,41, 132, 137, 140f., 144-6, 160, 187-9, 192, 194, 202, 208, 210-4, 234, 256, 260f., 265, 274, 298, 300, 302f., 326, 344, 353, 364, 374, 377, 392, 415, 436,451,543,545 botanic frame of reference 266, 429 bothersome feature 214 bound form 390, 436 boundary (of areas, morphemes) 97, 107,224,355,376,456 bracketing (of features) 261, 314, 359 branch (fig.) 189, 293, 316, 335, 354, 417,422 breadth, broadening 8, 182, 299, 391 break (sharp) 111 break-down, break-up 131, 252, 308, 400, 408; break-through (in evolu tion, methodology) 39, 56, 191, 263, 426, 446 breeding-place of change 245 brevity 184, 199 bricklayers' jargon 271 bridge(-building) (fig.) 171, 186, 218, 302,351,426,452 bucolic strain 432 buffer 94, 350, 380, 438, 550

612


buffoonery 434 bundle (of dialects, forces, lines) 122, 168,261,326 by-form 203, 239, 291, 313, 333, 408, 445, 453, 460, 546; by-product 246, 265 calendar unit, name of 142, 386 calque 146 camouflage 9, 146 canon (of taste) 314, 456, 551 canonic form 144, 355 caprice 124, 142, 219, 261, 265, 372, 394 caravan route 23 cardinal points, names of 230 cane (linguistique, sémantique) 71 cartography 56, 123, 218, 224 case endings 298; — system 206 case history 143, 255, 332f., 335 cataclysmic (impact) 387 catalogue 193 catalyst 237, 380, 440 category (of change, words, etc.) 155, 185, 197, 251, 271, 277, 297, 299, 363, 420, 445 causal (connection, explanation, force) 37,170-2,247,307,314,316 causation (simple, complex, multiple, primary, secondary, major, minor) 9f.,23f.,35,37,42,50,90,251, 255, 260f., 267, 269, 283, 289, 295, 297, 299, 302, 305, 308, 317, 326, 332, 336, 338, 347, 361, 366, 371, 378,381,386,434,462,547 cause-and-effect 35, 270, 428 centennial segment 142 center of gravity 199; — of the lexicon 266; semantic — 409; — of a word family 17, 346 central dialect 354; - pillar 268; - problem 104; - vowel 92, 332, 407, 428 chain (concatenation) of causes, events 34,69,223,301,315,326,377, 415, 428;-reaction 317

challenge 307 chamber (fig.) 292 chance 214, 258, 312, 323, 354, 407 change of affix 327; — within continui ty 57; — of focus 544; — of gender 142, 164; - of guard 57; - in lan guage 217, 297, 300, 362; - of meaning 145; - of sound 230, 372, 384, 543; -of stress 344 channel (of transmission) 132, 157, 247, 276, 310, 376, 385, 400, 440, 543 characteristic (sounds, forms) 96, 102, l l l , 143f., 155f., 168, 182, 191, 200, 205, 207, 238, 268, 281, 293, 300, 303, 313, 329f., 344f., 358, 386f., 406f., 409, 424, 427 characterization 28,139, 327, 365, 377 characters 115, 120, 232 charters 427 chauvinism 167 checked syllable 342, 370 checkered record 313 child language 338 chivalry, expression of 133 choice (speakers') 239, 268, 284, 286, 293, 312, 314, 323, 345, 370, 467 chorus of opinions 435 chromatic (verb) 327 chronicle, style of 324 chronology 68, 76, 90, 93f., 104, 118, 139, 169-71, 206, 212, 233, 241, 245, 247, 264, 301, 304, 312, 332, 343, 391, 400, 408, 424, 427, 436, 551 chronometry 87 circle of linguists 70 circularity 371, 385 circum-jacency 334; — locution 309; - maritime 111 ; - vention of a tradition 102, 113, 123 circumstance 160, 162, 167, 170f., 190, 204, 207, 209f., 214, 222, 251, 277, 282f., 288, 297f., 314, 327, 359, 371, 383f, 392f., 399, 408, 432, 439, 442, 462

SELECTIVE INDEX OF KEY TERMS circus jargon 208 civilization, material 156 clash (of homonyms, etc.) 18, 164, 185, 199, 202, 206, 211, 218, 221, 251, 262, 312, 315, 326, 352, 367, 376, 421,443,451 classicism 58, 70, 147, 307 classification (of forms, languages) 42, 72, 107, 144, 164, 193, 204, 268, 372, 414 classifier 119 clause 378 clear-cut (case, fact, pattern) 261, 284, 311,323,401 cleavage (two-way, three-way) 88, 113, 123, 221, 303, 325, 327, 329, 332, 355, 373, 373, 378, 383, 385, 405, 467,545 climate (academic, social) 5, 149, 161, 166,268,381 clinching (argument, effect, factor) 239, 254, 271, 310, 326, 336, 340, 344,380,388,429,431,447 clipping 11, 179, 185, 189, 200-2, 208 close-knit (group, relations, structure) 229, 246, 264, 305, 352, 383, 387, 389,404,411,451 clue 156, 182, 302, 342, 333, 335, 543 clumsiness (verbal) 313, 352f. cluster (of consonants, variants) 11, 94 101,144, 193, 211, 229, 236f., 240, 271, 275, 280f., 300, 303f., 324, 328, 335, 338, 346, 359, 362, 381f., 392, 394f., 437-40, 445, 547 coalescence 18, 206, 239, 254, 289, 302,335,451,458 coarseness, suggestion of 408 co-conditionment 299 coda 355, 443, 445, 447 co-determination 100, 194, 265, 391f., 428 codifier 456 coefficient, -ency 247, 326, 367 coexistence 91, 95, 99, 125, 200f., 209f., 213, 237, 242, 244f., 257,

613

260,262, 272, 277, 282, 291, 302f., 307, 311, 324, 338f., 347, 362, 365, 370, 380f., 383-5, 390, 414, 421, 425, 430, 438f., 546 cogency 171, 189f., 202, 270, 301, 308, 427 cognates (and cognate languages) 11, 15, 38, 92, 97, 125, 129, 134, 212, 268, 285, 292, 297, 303, 324, 346, 354, 391, 411f., 434, 441, 459 cognition 225, 268 cohesion, coherent whole 231, 352, 377 coinage (tentative) 93, 106, 204, 208 coincidence 44, 167, 199, 214, 220, 247, 264f., 288, 300, 303, 310, 341, 344, 353, 372, 415, 433, 442, 544, 548-50 collapse 140, 298 collateral (issue, pressure, source) 38, 172, 224, 257, 261, 335, 411f. collation 358 collection (of forms, writings) 60, 116, 160,185,203,218,260,297 collective (= mass noun) 162-4, 169, 197, 208 collision 106, 253,216,456 collocation 198, 342, 361 colloquialism 130, 167, 183, 185, 190, 201, 245, 345, 348, 368, 405, 415, 421,429,437 colonial languages 432, 434 coloring 115, 413, 443;color name 392 combinations 125, 274, 292, 304f., 307, 346, 407, 457 command word 12, 218, 415 comment(ary) 118, 157, 167, 190, 194, 306, 324, 329, 430 commingling (of dialects) 219 commiseration, expression of 408 common denominator 6, 36, 210, 247, 304,338,346,369,371,405; — descent, stock 131 ; — sense 335, 340; — nouns 429, 432; - stage 186f., — word 162; commonness 391,459 communication 73, 91, 132, 288

614


community of speakers 90f., 95, 106, 112, 117, 130, 146, 166, 213f., 231, 238, 265, 286, 294, 297f., 345, 362, 378, 387, 392, 394, 406 companion (force, form, study) 38, 117,230,261,343 comparability, comparison 52, 139, 172, 180 comparat(iv)ism 49, 62f., 70, 76, 87, 116,129,139,157,187,189,218, 306, 337, 342, 361, 381, 389, 391, 460 comparative degree 16, 37, 101, 260, 331, 411-3;415f.;- syntax 165 compatibility 35, 218f., 252, 260, 297, 312,327,372,411,425,447 compensation 170 competence 359 competition 97, 110, 121, 123, 190, 205f., 229, 238,247,289,311, 324, 328, 338, 350, 373f., 407, 445 complementation, -arity 35, 37, 252, 260, 420 complex (n.) 283; complexity (of cause, process) 222, 251, 261, 268, 282,299,301,354,443 complication 98, 104, 125, 268, 284, 286, 290, 313, 335, 358, 380, 382f., 389, 403, 407, 439 component 213, 241, 297, 301, 304f., 315f.,338,352 composite (book, review, suffix, text) 80,115,244,451 composition (vs. derivation) 5, 14f., 17, 34, 130, 195, 199, 315, 368, 443f., 451 component (affix, tense, verb) 92, 104f., 130, 190, 199, 210, 245, 331, 334, 374, 379, 389f., 401, 425 compression 115, 242, 247, 355, 367, 379, 384, 390, 430 compromise 147, 267, 287, 295, 303, 311, 346f., 349, 359, 366, 383, 414, 438 computation of rates 89 concatenation (of events, meanings) 28,

87, 171, 185, 214, 279, 301f., 315, 328,372,425,446,458,550 concept(ualization) 74, 132, 212, 314, 317,337,551 concluding segment 425 concomitancy 54, 106, 110, 113, 123, 145, 156, 167, 212, 223f., 251, 254, 259, 264, 280, 282, 286, 299, 310, 317, 353, 365, 367, 371, 380, 383, 390, 429, 431, 438, 440, 447, 459 concrescence 458 concrete (test, meaning) 9, 102, 194f., 267,275,427 concretization 193, 246 concurrency 201, 222, 299, 360, 363f., 374, 408, 441, 452 condensation 315 conditional (clause, tense) 103, 230, 357,371,378f. conditioned (development) 443 conditioning factors 13, 244, 247, 270, 303, 316, 340, 342, 417, 442, 447 conditions, local 202, 295, 329, 363, 374, 381, 401, 428, 435; condition 'state' 400 conduit (= channel of transmission) 104,112,254 configuration 97, 103, 123, 144, 155, 157, 163, 165, 199, 208, 223, 225, 240, 245, 297, 303, 329, 340f., 350, 381,443,545 conflation of currents 55, 207, 244, 270, 300, 377, 390, 423 conflict 14, 90, 95, 106, 108, 116, 124, 144, 230, 234, 239f., 242, 245f., 282, 286, 312f., 317, 323, 330, 342f., 345, 358, 373, 407, 435 confluence 17, 310, 325f. confrontation 116, 122, 124, 171, 283, 389,421 confusion 95, 267, 277, 282, 285, 311, 346f., 349, 358, 381, 457 congealment 131, 217, 305, 315, 412 congener 156, 191, 255, 258, 265, 283, 305,339,352,386,550 conglomeration (of variants) 94

SELECTIVE INDEX OF KEY TERMS congress (colloquium) of linguists 3, 58, 81 conjecture 24, 72, 103, 146, 169, 184, 187, 191, 203, 205, 212, 225, 240, 255, 257, 272, 290, 298f., 303, 312, 324, 334f., 360, 377, 422, 430,547,549,551 conjugation 19, 92, 95, 109, 184, 191f., 200, 204, 230, 234f., 269, 283, 288, 292, 354, 357, 361-71, 377, 379f., 444 conjunction (-ture) (of forces; as a form class) 13, 134, 267, 282, 295, 298, 308, 315,404,433,440,444,470 connection (genetic, overt or hidden, simple or dual) 9, 134, 271, 408, 415,446 connoisseurship (of languages, of verbal arts) 62, 87, 417, 549 connotation 255, 331, 415, 432, 439 connubium, linguistic 130, 208, 343 consciousness 123, 146, 251, 346 consensus 202 consequence 310 conservatism 4, 90, 95, 99, 102, 115, 125, 148, 202, 218, 241, 246, 272, 306, 316, 344, 370, 374, 394, 417 consistency 97, 100, 112, 201, 210, 214, 264f., 275, 337, 358, 381, 392, 428 consolidation 119, 259, 336, 468 consonant 235, 237, 240, 262, 280f., 286, 289, 326, 329, 342, 345, 357, 362f., 382, 393, 402, 409, 441, 446; - pillar 231 ; - ization 389 conspicuity 299, 303, 316, 338, 346, 381,406 constants 372 constellation (of circumstances) 3f., 210,381 constituent 251, 300, 305, 308, 400 constraint 112, 170 construct, artificial 142 construction, syntactic 133, 311, 358, 411,470,550 contact 90, 133, 208, 211, 262, 266,

615

301, 304, 342, 346, 351, 365, 371, 468 contagion 113, 310,323 containers, names of 137, 156, 159f., 162,166,169,208,363,443 contamination 9, 38, 115, 200, 213, 219, 256f., 261f., 282f., 287, 291, 325,351,371,468f. contemporaneousness 240; contempora ry 433 contempt, expression of 408 contending forces 263 content, semantic 383 context 39, 116, 125, 140, 167, 172, 197, 202, 205, 211, 213, 263, 269, 282, 293, 297, 299f., 3024, 308, 331, 339, 344f., 367, 380, 395, 400, 407,413,415,429 contiguity 90, 180, 235, 286, 289, 371, 415 contingent 309, 317 continuity 72, 147, 172, 226, 242, 315, 400, 408, 473 contour 90, 123, 185, 188, 199,243, 257, 264, 298, 307, 323, 344, 380, 384, 386, 400, 419, 429, 438, 545 contraction 254, 335, 341 contradiction 43, 100, 221, 233, 242, 308,242,308,327,391,428 contrast 90, 105, 107, 170, 263f., 275, 288, 293, 303, 317, 325, 330, 334, 361, 366, 368, 381, 383, 385, 420, 442, 446; - ive analysis 80 contributing factor 37, 131, 246, 268, 282,301,313,316,366 control 76, 87, 95, 210, 213, 219, 247, 261,275,281,328 controversy 160, 167-9, 221, 225, 22931, 238, 242, 251, 299, 305, 316, 344, 358, 366, 376, 378, 413, 425 convention (form of research, judgment, social approval, sources) 10, 64, 164, 179,212,220,231,405 convergence 38, 131, 145, 261, 274, 277, 285, 299f., 312, 325, 367, 370, 437, 439, 444

616


conversion (to novelties) 6 coöccurrence 236f., 306, 391, 404, 407 coördination 317 copyists 108, 123, 311, 354, 358f. core (syllable, vocabulary) 41, 192, 327, 335,405,427,442,469 corpus (of data, variants) 5, 90, 119, 131, 163, 204, 309, 324, 353, 357, 391,400,408,431 "correctness" 123, 246 correlation 87, 125, 156, 209, 284, 291, 303 correspondence (of sound changes, etc.) 219, 229, 241, 283f., 471, 543, 545f. ; (exchange of letters) 426, corroboration 98, 107, 427 "corruption" 32 cosmopolitanism 76, 82 counterargument 203; — balance 113, 198, 213; - current 185; - example 194; ֊ force 366; - intuitive 165; — movement 94; — proposal 426; — theory 266; — thrust 309; — view 139, 426 counterpart 114, 157f., 163, 180, 183, 186, 203, 234, 252, 258, 277f., 282, 285, 288, 299, 301f., 343, 354, 358-60, 367, 369, 406, 425 countertone (-stress) 14, 233, 243, 341f., 344, 368, 370,382,389 couple 546 craving (for novelty, pleasure) 54, 447 creation 141; creativity 33 credibility 187,261,290, 354 Creole studies 29 crest (of a movement) 125, 444, 550 crise de conscience 4, 8 criteria 19, 72, 75, 89, 164, 342, 400, 550 critic 433 critical portrayal 67; — stage 105; — verdict 69; — years 75 critique 76, 88, 108, 162, 164, 171, 230f., 262f., 266f., 327, 334, 340, 342f., 353, 375, 378 cross (as in breeding) 118, 267, 437,

441 cross-connection, -cultural, -current, -dialectal, -fertilization, -lexical, -spatial, -temporal 34f., 40, 102, 145, 199, 206, 208, 210, 265, 284, 340, 346, 381, 385, 426, 452; at - pur poses 409 crumbling 131, 261, 282, 338, 417 crux (of a question) 314, 331, 382, 385, 470 crystallization (fig.) 212, 268, 294, 313, 327, 329, 367, 380, 408, 412, 415, 442,446,469,471 cubs, names of 165 culinary terms 146 cultismo 123, 334, 428, 444, 467, 543, 546 cultural 333, 378, 392, 428, 439, 549; broadly — 209; — area or zone 11, 53; — factor 302; — history 57; — loan 147; — prestige 15; — status 442 culture (and cultural history) 73, 78, 88, 130, 147, 155, 163-5,, 180, 187, 212, 214, 223, 226, 226, 273, 288, 369, 372, 433 cumulative force 125 cumbersomeness (verbal) 314-17 currency (of use) 188, 379 current (fig.) 77, 424 curriculum, linguistic 5 custodial jargon, 185, 205-7 cyclical theory 24, 55 cynegetic (lexicon) 102 dactylic coda 76 danger (= hazard, risk) 313 "dangling" vowel 209, 328, 332, 334, 338,356 Darwinism 26 data gathering 5, 24, 27, 145, 147,1624, 168, 274, 311, 376, 436, 433f. dating (direct, indirect) 11, 87-9, 125, 290, 308, 426, 455, 543 daughter language 130f., 133, 156, 179, 204, 209, 324, 330


de-adjectival (verb) 279; de-affrication 302; de-aspiration 547, 549; decom position 291, 294, 331 ; dediphthongization 336; deëmphasis 400; deglutination 41, 267, 468; depalatalization 100f., 387; depluralization 207, 546; deverbal 192, 278 "deadwooď'91,447 deaf-mutes, teaching of 62 debate 167, 266, 268, 306, 376, 421, 433 debilitation 203 débris (lexical) 104,208 decay 169, 285, 304, 326, 344, 360 decision, decisive 309, 311, 348, 357, 370f. declension 170, 183, 199, 206, 264, 298,344,366,371,417,425 decline 117, 131, 207, 415 décor, verbal 417 defect and default (physical) 400, 409 defectivity, -eness 48, 264 definability, neat 342 definite article 130 definition 147, 171, 287, 301, 332, 340, 405 deflection 92f., 102, 108, 189, 225, 256f., 351, 424, 429, 441, 443 degree of... (freedom, etc.) 100, 111, 258, 261, 274, 311, 359, 374, 440 delay 58, 290, 439 deliberateness 146, 345, 366, 428, 430, 434 delight, speakers' 312 delineation 221, 265, 298, 391 demarcation, line of 428 demonstrative 105, 130, 230 dendronym 343, 436 denomin-al, -ative (augmentative, verb) 171,183,279,334,469,545 denotation 406, 468 density (of a network) 224 dental (consonant, stop) 100, 211, 237, 393f.,407,411 departure, point of 285, 414 dependence (mutual, etc.) 191, 261,

617

311,552 deponential (verb) 184 deposit 31 f., 120,429 depth (of force, etc.) 34, 247 derivation(al) 7, 9, 17, 45, 139, 144, 158, 169f., 171, 180, 183f., 186, 188, 194, 201, 205f., 229f., 234, 240, 245f., 251, 255, 258, 261f., 272, 300, 325, 328, 333, 356, 369, 371, 381, 387, 389, 406, 430, 434f. derivative (word,language) 130, 171, 181, 186, 195, 199, 201, 203, 210, 231, 274, 290, 307-9, 340, 350, 354, 366, 375, 383, 401, 408f., 413, 416, 443, 543, 555; derivative-in-reverse 184,209 derogation 156, 167 descendant 93f., 101, 192, 194, 210, 252, 272-5, 277, 292, 312, 315, 327, 334, 339, 345, 353, 368, 373, 375, 385, 387, 407, 412, 429, 440f. descent (line of) 131, 141, 202, 211, 279,297,351,374,407,424 description 6, 31, 53, 71, 140, 163, 260, 267f., 270f., 443f. design (of a language) 143, 188f., 199, 208,226,302,315 designation 265, 274, 447 desirability 329 desuetude 193, 201, 279, 354 detachability 121, 339, 412, 432, 447 details (fine, tell-tale) 88, 186, 189f., 194, 199, 238, 266f., 279, 282, 289, 300, 316, 329, 343, 412, 416, 424 détaülisme 24, 56 detection 218, 434 deterioration 101, 251 determinant of change 9, 12, 23, 35, 39, 42-4, 251, 261, 263, 301, 316, 430 determinism 57, 222 deterrent 471 detour 200, 252 détresse 326 development 160, 163, 183, 253, 258, 265, 279, 294, 300f, 304f, 308, 328, 342f., 354, 360, 362, 383, 394,

618


439f. deviation 168, 393 device 96, 104, 124, 158, 166, 184, 417,443,508 diachron-y, -ism -ist 6, 63, 71, 141f., 252,297,331,498,503 diagnosis 18, 92f., 242, 354, 547 dialect 131, 317, 367; - atlas, map 141, 164, 168; - boundary 56, 288; — cluster 28; — geography 30, 56, 58,68,123,161,218,471,503; — glossary 168; — level 120; — mix ture 219; — source 96; — speech 191, 269, 459; - variant 553; - zone 43, 220, 425; city - 442; social - 208 dialect(al) coloring, nuance 101, 323; — evidence 231; — form 195; — record 233 dialectology 27, 29, 70f., 77, 107, 295, 358,423 dialogue (scholarly) 14, 52, 166 diametric (contrast) 161, 211, 353 dichotomy 76, 98, 104, 149, 172, 205 dictionary 8, 45, 141, 301, 323, 327, 350, 547 didacticism, 72, 74, 110, 314 differen-ce, - tiation 14, 43, 76, 88, 90, 106, 143, 157, 163, 166, 199, 201f., 219, 231, 238, 254, 265f., 275, 282, 297f., 313, 334, 344, 367, 370, 384, 387f., 406, 441, 446, 452, 457f., 506f. diffraction 125 diffusion (propagation, spread) 11, 42, 58, 109f., 132, 179, 187, 235, 240, 279, 286, 310, 315, 344, 366, 368, 428, 436, 439f., 446f., 451, 498, 505, 543 digest 60, 343 digraph 122 dilapidation (verbal) 417 dilemma 73, 89, 108, 116, 192, 299, 332,335 dilution 327 dimension 172, 189 diminutives 40, 143, 156-9, 1624, 168-

70, 184f., 195,212,232f.,235, 242f., 264f., 366, 412, 421, 425, 428, 432, 435f., 442, 444, 446, 463 dihthong(ization) 5, 92, 140, 211, 229, 232, 235f., 238f., 2414, 247, 256, 258, 264, 280f., 286, 288-90, 311f., 328, 337, 339, 343, 3824, 386-9, 425, 459 direct reflex 98 direction (of change) 117, 133, 163, 182, 184, 217, 228, 235, 253, 263, 267, 272, 286, 300, 314, 325, 330, 338, 343f., 348, 354, 357, 360, 362f., 370, 427, 447, 466, 470f., 497, 501 dis-advantage 192, 212; - agreement 262; - ambiguation 309; -cord 299; ֊ couragement 501 f., 506, 508; - engagement 505; — entanglement 301; ֊ integration 109, 206, 298, 352, 464; - parateness 265, 310; - parity 204, 295, 406; - pensability 267, 292; — proportion 146; — quisition 306; — ruption 224; — section 328, 405 disappearance 103, 117, 165, 267, 280f., 298, 300, 385, 417, 463, 506, 508 discipleship 64 discipline, autonomy of 53 discourse, level of 116, 209, 278 discovery 6, 269, 283, 328, 340, 342, 377, 434, 459, 499, 508 discrepancy (of results, etc.) 44, 89, 123, 181, 186, 189, 255, 262, 323, 366, 424 discreteness (of factors, forms) 17, 131, 310, 360 discrimination 168, 182, 265f., 325, 403, 430 discursiveness 268 discussion 10f., 179, 189, 194, 196, 251, 270, 276, 297, 301, 305, 309, 332, 343, 372, 376f., 419f., 433 disguise 245, 279, 337 dislodgment 103, 166, 239, 242, 272, 288,307,313,348,379,441


disposal 91 dispute 171, 299, 307, 310, 421 dissimilation (of consonants, vowels, etc.) 10, 15, 38, 41, 213, 219, 234, 236, 241, 265f., 274, 289, 294, 332, 334, 337, 363-5, 369, 372-7, 379, 381, 387, 390f., 408, 423, 428, 440f., 447, 464, 466 distance 110, 192, 237, 265, 337f., 342, 364 distant past 91, 368, 371, 373 distillation 258 distinction 115, 229, 312, 354, 377, 406, 456 distinctive (feature, mark) 183, 199, 235, 260, 268, 300, 303, 444 distortion 217, 227, 294, 333f., 339, 386,414,434,436,460 distribution 74, 102, 105, 211, 219, 221,235,245,282,288,315, 328-30, 337, 344, 353, 375, 382, 392,412,463 disturbance 262, 268, 298, 301 disyllabicity 198, 200, 314 divergence 106, 129, 131, 146, 220, 274, 298, 429, 437, 505 diversi-fícation, -ty 34, 75, 88, 162, 276,281,304,437 dividing line 132, 138, 285; division (prime) 164, 204 divorce (fig.) 205, 285, 376, 384, 416, 497, 507 doctrine 76, 223, 251, 366, 498; doctrinaire 434 document(ation) 53, 101, 108, 110, 112, 115f., 119, 142, 161,166, 168-70, 187,201,203,211,213, 223, 229, 246, 270, 287, 290, 301, 308, 310, 331, 359, 361, 373, 383, 389,431 dogma(tism) 72, 218, 268, 501 doldrums (fig.) 189 domestic animals, names of 264, 402, 406, 425, 432 domestication 265; domesticity 170 dominance (of trend, etc.) 114, 217, 219, 256, 262, 268, 315, 327, 334

619

donor language 38, 261 doomed (feature, form) 234, 246, 384 dormancy (= latency) 214, 257 dosage 280 doublet, duplication 44, 90f., 110, 115, 143, 157, 166, 193, 210, 266, 290, 324, 344, 350, 366-8, 372, 374, 381, 384, 393, 408, 411f., 437, 440, 545 doubt, margin of 500 downgrading 299, 315 drift (of events, etc.) 34, 130, 145, 187, 207, 212, 257, 283, 297, 307, 312, 328, 350, 352, 357, 380, 390; driving force 240, 261 dual borrowing 206, 210; - develop ment 327; - effect 108; - reference 472; duality 454 dyadic array 236, 312, 422 dynam-ics, -ism 94, 147, 203, 217, 221, 236, 260, 309f., 498 eagerness, speakers' 312, 438 ease of pronunciation 370 eccentricity 472 ecclesiastic 112, 132, 213, 333, 386, 439, 443, 543, 546, 549 echo(ing) 117, 186, 199, 210, 233, 307, 309, 328, 346f., 364, 543 eclipse (= partial or total disappearance) 99, 372 economy 37, 95, 144, 201, 260, 292, 314,391,447 edifice (= structure) 8, 142, 188, 345, 391 educated guess (in linguistics) 200 education(al) 124, 209, 310, 345 effect 161, 201, 266, 316, 371 efficacy 456; efficiency 76 Einfall 119, 462 Einreihung ( = serialization) 447 elaboration 108, 114, 160, 163, 168, 189, 200, 223, 234, 236, 245f., 257, 262, 265, 269, 287, 291, 294, 3068, 314, 340, 349, 361, 415, 422, 427, 430,447,461,469,503, 544f.

620


elasticity 217, 225 elative degree (in grading) 412 elegance 9, 42, 52, 265, 268, 316, 332, 386, 503 element 305, 308, 326, 328, 499, 507 elevation (fig.) 121 elicitation of data 5, 110 elimination (of a harmful or superfluous element) 214, 236, 267, 274, 286, 374, 387 elitism 146, 298 elusiveness 235, 269, 299, 317, 458, 507 embedment 241 embroidery 185, 343 embryonic stage 184, 191, 303, 310 emendation 111, 114, 349f. emergence 124, 214, 255, 257, 265, 275, 279, 291, 298, 303, 317, 367, 382, 453 emergency (= détresse) 125, 210, 547 -ernic (vs. -etic) 32 emotional stress (scholar's, speaker's) 55,167 emphasis (linguistic, literary) 5, 264, 308f., 328, 345, 431, 499, 553 empiricism 147 "empty" morph 1-6, 445, 447 enclave 334, 465 encoding 268 encroachment 101, 109, 118, 133, 209 209, 241, 297, 306f., 326, 366, 425 encyclopaedic perspective 129 end, loose 498 endangered (form, sound) 231, 332, 334, 338-40, 356, 360, 394, 407 endemic 100 ending (coda, suffix) 105, 202, 206, 298 endowment (wth a faculty) 213, 275, 342, 405f., 426, 461, 473 enforced emigration 66 engagingness 467 enlightenment 499 enmeshment 117, 257, 293, 301, 505 enrichment 112, 190,378

ensemble (ofcircumstances,features) 34,42,88,125,243,268,282 enthusiasm 76, 80,196, 225, 299, 324, 461, 502 entrenchment 101, 110, 120, 203, 211, 217, 256, 286, 313, 379f., 464 entry (in a dictionary) 142, 157, 188, 544, 547; (point of penetration) 160 environment 102, 201, 212, 219, 237, 337, 348, 363, 365f., 372, 413, 441, 466 epenthesis 13, 259f., 269-71, 276, 279, 282-6, 288, 290, 295, 312, 324, 329f., 338f., 345, 349, 356, 376, 423f., 427, 430 ephemer-al, -ous (existence, form, solution) 17, 100, 112, 258, 306, 309,372,406,415,432 epicene nouns 165, 277 epigraphy 187, 197,201,547 episode 106,236,301,316 epistolarnim 59 equality, near-equality 42, 340 equation 133, 142, 200, 225, 253f., 262, 277, 290, 297, 423, 429, 457, 459, 506, 553 equidistance 54, 377; equilibrium 503 equivalent 142, 279, 283, 292, 294, 297, 303, 316, 406, 430, 457, 544 erosion 44, 203, 213, 245, 254, 283, 302, 309, 313, 315, 325f., 341, 370, 379,390,400,428,433,439 erratic (course of events, etc.) 109, 121, 196, 198, 208f., 240, 247, 256, 264, 305, 307, 327, 332, 345, 350, 357,391,436,547,553 error 185, 187,304,340,342 erudition (verbal) 255, 309, 353, 368, 393 escape (route, etc.) 112, 119, 125,258, 332,374,388,467,506,508 esdrujulismo 419,426, 443f. essentials 342 esthetics 25, 73f., 121, 134, 285, 314f., 457,551


estrangement 133, 143 ethnic matrix 52, 134 ethnolinguistics 75, 82; ethnonym 424 etiology 14, 24, 35, 45, 251, 268, 297, 336,361,441 etymological controversy 19; — crux 101; — dictionary 65, 503 etymology 7-9, 12, 23, 27, 29f., 41, 52, 57,70,89,118, 121,13942,146, 157, 194, 202, 217, 219, 222, 224f., 230, 242, 251, 253, 257, 260f., 285, 289, 292f., 298, 301f., 307, 324f., 330f., 333, 337, 357, 365, 376f., 384, 388, 394, 403, 411, 415, 423-5, 426f., 429, 435f., 451, 462, 468, 473,497-511,553 etymon 497 euphony 210, 423 evaporation of overtone 432 evasion 258, 365, 379, 438 events (course, chain of) 159f., 163, 166, 171, 185, 197, 200, 223, 251, 260, 265, 268, 298, 301f., 304f., 326, 355, 357, 370, 379, 384-6, 420, 439, 549f. eviction 326, 365, 453 evidence for (amount, shape of) 108, 270, 272, 308, 310f., 326, 359, 391,408,414,427,432,460 evolution(ism) 24-6, 30f., 33, 76, 181, 184, 209, 251, 266, 281, 300-2, 312, 355, 393, 424, 440 evolutionary curve 241; — sciences 10, 23 exaggeration 75, 78, 191, 230, 392, 473 excavation 433 exception 164, 198, 205, 209, 218, 232, 241, 247, 289f., 313, 328, 361, 289, 424, 506 excerpts 142, 190, 210, 309 excess 202, 263, 303, 326 exchange, lexical 146 exclamation 270, 553 exclusion, -iveness 205, 209, 233, 251, 260, 263, 268, 271, 289, 328, 370

621

excretion 16 excusable (error) 187 exegesis 70, 234, 242, 279, 553 exemplification 10, 265 exhaustiveness 342, 345 exoticism 37, 68f., 109, 189, 257, 265, 357, 385, 424, 431f., 444 expansion 163, 261, 270, 307, 309, 312, 343, 352, 363, 368, 405, 408, 416,423,446 expectation 195, 202, 218f., 229, 241, 252, 255, 262, 268, 277, 283, 285, 287, 301, 310, 333, 346, 370, 380, 384, 388, 424, 444 experience 31, 73, 251, 370, 499, 502 experiment (-al, -ation) 7, 9, 27, 55, 74, 88, 105f., 124, 130, 144, 148, 166, 172, 185, 187, 210, 240, 258, 294, 316,334,348,419,438,451 expert(ise) (unique, dual, or triple) 5, 10,77,142,225,299,310,312, 326, 499, 503, 507 explanation, explicative 143, 167, 172, 179, 183, 186, 195, 197, 202, 204, 208-10, 214, 230, 232-4, 240, 243, 247, 251, 260, 265, 267, 269, 282f., 290, 295, 299, 301-3, 307, 313, 316, 324f., 329, 343, 351, 357, 371, 381, 387, 394, 424, 445, 451, 469, 499, 506, 547 explicitness 40, 166f., 265, 445, 498 exploitation (of prejudice, resources) 52,113,146,214,299 exploration 279, 251, 262, 293, 300, 324, 330 expository style 26, 314 exposure (lexical, etc.) 276, 339, 469, 473 expressi-on, -vity 41, 69, 73, 166, 219, 267, 293, 304, 420, 425 extension (of scope, etc.) 109, 112f., 145, 186, 193, 201, 206, 219f., 244, 259f., 271, 284, 298, 307, 309, 312, 346, 356,364,441f. external (factor, pressure, spread) 235, 251,261,279,371

622


extinction 325, 436 extraction 180, 193, 197, 200, 272, 277, 291, 316, 331, 352, 358, 368, 388, 413, 426, 498, 508, 546 extra-frequency 232; — linguistic cause 52, 172, 298; - strength 506 extrapolation 82, 106, 110, 125, 308, 312,358,414,435,460,505 extreme (case) 279; -ism (= radicalism) 4,41 extrication 456 exuberance (of growth) 400, 407, 419, 437, 443 eye-opener 508

fabric, total 297 face value 342 facet 430; negative — 371 ; semantic — 193, 206 facetiousness 167, 185, 212, 306, 401, 406, 443 facile (explanation) 266 fact (raw, historical) 393, 395; factual 51 faction (of linguists) 7, 141 ; (of speakers) 394 factor (cultural, principal, concomitant, positive, negative) 14, 132, 134, 171, 205, 222, 234, 239, 246f., 251, 255, 261, 281f., 289, 295, 297, 303, 309, 313f., 316f., 326, 332, 366, 371, 377f., 380, 399, 416, 453, 467 fad, fashion, 52, 64, 73, 78, 134, 157, 238f., 312,316,324,497f., 514 falling diphthong 37, 244; — into place 385,391,413,419 false (analogy) 400; — (regression) 42, 222, 230, 268, 347, 355; - (resti tution) 345; — (restoration) 332; - (segmentation) 223; — (separation) 509 'fame', 'glory', 'honor' (expressions for) 212, 214 familiar (= informal) 334; 415; (= wellknown) 366, 400, 413; familiarity

with 201 family, of languages 29, 88, 156, 247, 316, 324, 368; ֊ of suffixes 244, 434; - of words 199, 203, 207, 222, 267, 330, 352, 356, 373, 392, 510; ֊ background 392; - tree projection 217; familial (pattern) 431 fantasy 170; fancy 425 far-fetched (argument) 186; - flung (domain) 356 fascination (with linguistic form or theory) 184, 297 fatal (coincidence) 432 favorable (conditions, result) 210, 213f., 230, 239f., 298, 344, 381, 393, 399; favorite (idea, student) 262, 340, 367 favoring (of a word or feature) 234, 237, 241, 260f., 279, 286, 336, 387, 389, 425, 439, 459, 506 feature (salient, tell-tale) 88, 108, 130, 144, 184, 206, 269, 325, 387, 392, 452; (distinctive) 190, 235, 295 "feed-back" of information 24, 105 "feeling" I {Sprachgefühl) 282, 314f. Feldtheorie 148f., 151 felicity (of formulation) 137, 146, 180, 184 fellow-member (of a set) 399 female 150, 166, 170, 264; female agentive 145; feminine 142-4, 150, 155-9, 162-4, 166f., 169, 183f., 186, 189, 193f., 199-201, 204-7, 209-12, 265, 276f., 288, 305, 313f., 392f., 419, 425, 454, 465, 467; feminize 276 Fernassimilation 371 fertility 437 Fest- & Gedenkschrift 59 festivals, names of 514 feuilleton style 308 fiction 465 fidelity (to a tradition) 273, 284 field-interview, -notes, -trips, -work(er) 5,30,81,99, 110,157,164,221, 224, 295, 423, 426, 455, 465; lexical


-409 figurative 264, 292 filiation (of languages) 217 ; (of texts) 114 filling of slots 511 filtering 55, 92, 96, 193, 225, 261, 414, 468 final outcome 263; — phase, stage 413, 438; - segment 378; - syllable 340; - vowel 238, 269, 452; - word segment (= coda) 10, 156 findings) 342, 377 fine-grained (study) 386; - meshed (sifting) 137, 295, 406; finite (verb) 351 finesse 507 firmness (of articulation) 351 ; (fig.) 515 fit (adj., ., v.) 181, 201-3, 206, 266, 279, 338, 347 - fix 32; "fixed" phrase 355 flair 218, 262, 335, 385, 507 flamboyant (formulation, terminology) 33, 224, 435 "flanking" 100, 160, 193, 197, 213, 241, 243, 254, 271, 278f., 307, 324, 334, 336, 347, 350, 354, 370, 379, 391,403,411,452,455 "flatness" (= lack of depth) 333 flattery 335 flaw (in an analysis) 187, 230, 261, 339, 375f., 426, 468, 506 fleeting (contact) 458 flexibility 206, 225, 299 floating (through linguistic space) 310 flow 305, 342 flowering (fig.) 122, 148, 191, 294, 505 fluctuation (of form) 10, 27, 36-8, 89, 95, 148, 170, 221, 232, 260, 263, 273,313,347 fluidity (of usage) 130, 262, 279, 282, 284,305,338,351,381,442 flux 45, 185, 236, 269, 351, 467, 498 focus, 96, 263, 328f., 372, 469 foil (one language to another) 361 folk belief 333; ֊ etymology 41, 56, 202, 222-4, 234, 267, 330, 333, 337,

623

356, 368, 429 497, 500, 502; - lore 134, 170, 503; ֊lyric 351; - science 424 -speech 203, 208, 212, 234, 423; "folksy" (tone) 212, 427, 430 followers 220, 223, 306, 341, 431, 460, 466 "footnote" = (account of) minor event 317 force 210, 237, 240f., 251, 263, 268, 297f., 307, 340, 356, 467; forcible (change) 273, 286 foreground of events 163; forerunner 100, 191, 308, 334; foresight 214, 223 foreignism 132; foreign-sounding 273 foreseeable (course) 407 form vs. meaning 106, 500; form class 509 formal (analysis) 426; (aspects, criteria) 89, 138, 140;-(caique) 1 4 6 ; (proximity) 211 ; -(resemblance) 192, 310;-(side) 213. -(variation) 184; formal (vs. informal) 392, 510; formalization 511, 513 formation (= lexical item) 205, 207, 252, 260f., 288, 295, 308-10, 329, 344, 350, 409, 411, 413, 420, 423f., 428,431,433,454 formative (period, stage, years) 75, 141, 184,387,409 Formenmischung 36, 41 formula 206, 222, 268, 340, 416f., 453, 458, 502; formulation 130, 134, 148, 157, 170, 172, 180, 265, 295, 301, 308, 317, 324, 344, 431, 446, 460,462,497,508 fortuity 125,259,328,431 fossilization 32, 89, 99, 105, 196, 309, 401 foundation, theoretical 498f. fragment (of patterns) 379; -ary record 53,68, 123; -ation 131, 385 frame (work), cultural 167, 270, 345; — of reference 333 free choice 284; - syllable 144, 334,

624


340, 370; - variation 45; - -wheel ing 223; - will (speaker's) 222; freedom (from) 380; freely available 361 frequency 89, 113, 122, 146, 182, 199, 221, 232, 237, 266, 298, 325, 331, 377-9, 404, 507; frequentative 192 fringe (benefit, from a shift) 103, 257, 265, 307; ֊ (of a function) 162 front line research 18, 140f., 232; front position 452; front vowel 324, 332, 338f., 314f., 366, 382, 424; frontal attack 157; frontier of knowledge 224, 232; frontier line 288; fronting 385 fruitfulness of approach 217 ; fruit-trees, designation of 184 full-bodied (ending) 292, 469; ֊ -fledged (verb) 404; - retreat 421 function(alism) 4, 69, 73, 78, 158, 162f.,165f., 183, 193, 199, 204, 210, 229, 254, 277, 284, 299, 309, 328, 399, 401, 406, 413, 417, 430, 447,452f., 510 fundamentals 75, 93 fusion 286, 351 future tense 103f., 357, 367, 369, 371, 378-80; futurism 75 "fuzziness" 306 gains (and losses) 74, 298 galaxy (of derivatives) 459 Gallicism 15, 102, 137, 139, 142, 146, 212, 268, 274, 311, 335, 368f., 383, 425, 506 gambit (fig.) 99, 203, 205, 341, 403, 425,498 gamut (of meanings, prefixes, vowels) 144, 201, 223, 341, 367, 427, 434, 462 gang, jargon of a 392 gap (semantic, etc.) 105, 108, 129, 149, 203, 218, 245, 293, 324, 330, 395, 401,452,460,510 gates, opening of (fig.) 213 gaudiness (of image) 333

Gelehrtenkorrespondenz 59 gemination 355, 359f., 463, 507 gender 10f., 39f., 100, 130, 140, 142f., 155f., 160, 164f., 167, 169, 172, 191, 193, 196f., 200, 202, 204f., 210, 223, 231, 234, 263-5, 260, 313, 354-6, 453, 457; gender switch 143, 149,156,159,163,169,204, 210,276f.,443 genealogy of manuscripts 111 ; — of variants 323 general diachronic linguistics 8; — gram mar 75; — linguistics 78f., 138, 155, 196, 217; generalist 500 generalization 27, 106, 108, 232, 240, 243, 257, 286, 342, 355, 370, 384, 387,441f.,445f., 514 generation (of scholars) 14, 57, 82, 131, 134, 225, 239, 314, 323, 342, 345, 422, 511 generative grammar 76, 78f.; — pho nology 301 ; — semantics 147 genetic 348, 380, 424, 513, 517; - analysis (of language) 8, 30; dilemma 192; — explanation 179, 269; - influence 180; - issue 190; - link 204, 236; - relationship 187, 307, 312, 363; - reconstruction 409; - research 295; — source 446; genesis 'rise' 394; genetics 25 genitive (case) 452 genre (of inquiry) 51, 72, 81, 141, 157, 323, 391 genteelness 511 genuineness 260, 334, 445, 515 geo-chronology 89; - graphy 267; - graphic (area, center, context, direction, distance, matrix, name, spread) 52, 103f., 110f., 144, 168, 219, 276, 303, 354, 498; - metric design 226 germane 113, 247, 268, 366, 393f. gerund 113, 122 Gestalt 147 gesture 304 ghetto language 385


ghost-word 141, 327, 464 gladiators' speech 200 gloss 102, 180, 209, 279, 301, 339, 349, 400, 412, 420, 463; -ematics 63, 148;-ary 164, 168, 519 glottalized consonants 130 glottoarcheology 433; — chronology 76, 82, 88f.; — dynamics 25; — esthetics 74; - geography 74, 159, 282; — history 9; glottologia 31, 78; glotto-palaeontology 182 goal (of language) 447; (of an opera tion) 142, 314, 508; goal-orientation 300 Golden Age trauma 51 good taste (as a norm of language change) 314 gorgia toscana 344 gradient 103; -ing (fine) 103, 130, 206, 412 graduality 94, 131, 141, 204, 232, 245, 251, 257, 298, 309, 331, 353, 370, 405, 408, 412, 421, 469, 504 grafting (fig.) 98, 122, 145f., 275 grain (of truth) 323 grammar 142, 223; comparative — 186, 381; historical - 8, 12, 19, 27, 88, 217, 241, 297, 301, 332, 362, 505; grammarian 110, 169, 197, 503, 505, 514 grammatical austerity 125; — category 43; - device 124; - design 190; — dilemma 180; — feature 211; — mechanism 109; — morpheme 144; — organization 95; — preference 88; — process 7; — resistance 439; — signal 286; — status 458; — tool 427 ; grammaticization 166, 315 grand strategy 56 grapheme 229; graphy 234; graph 114, 211, 234, 285, 347, 358f., 429 graphic 289, 338, 352, 441 gratification of urge 447 gratuitousness 130, 204, 314, 349 gravity, center of 199, 243 ; gravitation (toward) 206, 243, 358, 399, 442

625

"gray zone" (fig.) 224, 298 Grenzgebiet 5 groove (fig.) 106, 148, 193, 519 groping(s) 73, 78, 106, 123, 231, 263, 323, 353 grotesquerie 141, 167, 170 ground, solid 395; -lessness (of change) 257; -work 308; to break - 308, 460; to hold - 399, 516; to yield -341,352,358,425 group, of ingredients 262; - of people, speakers 185, 392; of suffixes 406; - of words 164, 167, 204, 206f., 378, 388, 402, 411, 424, 431, 509; grouping of data 144, 147 grove, designation of 447 growth 171, 209, 223, 285, 292, 306, 334, 360, 381, 386, 400, 408, 436f., 456,513 Grundriß = specialized encyclopedia 49, 129 guarantee 356 guess (anyone's, educated, fair) 200, 356, 465 guidance (bibliographic, philosophical) 163f., 193, 206, 209, 497; guided (vs. blind) force 251 gulf (fig.) 467 gustatory association 422 gymnasts' jargon 200 habit 304, 354, 392 hagiography 112, 129, 132, 520 half-baked (research) 309; -hearted (advocacy) 424 halo of connotations 255 halt, to bring to a 516 hapax legomenon 122, 413f. haphazard 198, 445 (cf. fortuity); happenings 305 haplology 193, 209, 246, 367, 405, 466,517 harbinger (of some change) 263 hard core 442, 451, 505 hardening 131, 440 harmon-ize, -y 242, 286, 288, 298,

626


503f. harness, terms pertaining to the 198 hazard (of homonymy, etc.) 8, 143, 199, 213f., 332, 338f., 342, 354, 376, 380, 385, 394, 432, 437, 462, 509 hazy (contour) 107, 166, 221, 225, 288, 400, 405, 506 head(word) 17, 199,243 healthy (fig.) 165,327 heavy (accent, emphasis, stress) 184, 197, 311, 340, 343, 379, 421,431; (cluster) 241; (commitment) 461 (diphthong) 211 ; (liability) 499; (representation) 509; (share) 212; heavily-stressed 240; heavily-struc tured 195, 314f. hegemony 71, 133 Hellenism 15f., 277, 366f., 371, 442f., 444, 446; Neo-Hellenism 137; Hellenization 420 helter-skelter (of approaches) 147 hereditary (feature, form) 25, 424 heresy (intellectual) 69, 73 heritage 132, 277 hermeticism 77 hero worship (among linguists) 70; heroic siyle 416 hesitancy 156, 329, 380, 414 heterogeneity 131, 141, 163, 261, 263,301,361,426 hiatus 104, 342, 383, 463 hibernation (fig.) 502 hidden (factor, feature, value) 6, 9, 212, 323, 420, 427, 436 hierarchy (of factors) 14, 43, 35, 129, 132, 144, 171, 188, 199, 202, 224, 230, 286, 301, 304, 310, 312f., 316, 353, 403, 438, 469, 503, 508, 516 high degree 311, 340; — level (inquiry, etc.) 251, 280; - rank(ing) 316; - tide (fig.) 207; height (=peak) of a process 246 hilarity 444 hindrance 390

hint 282, 323, 328, 330, 333, 376, 426, 430,517 Hispanism 133, 300, 325, 423, 505 Hispano-Arabism 151, 424, 428 historical accident 362, 459; — ap proach 76, 139; — axis 498; - linguistics 10, 179, 295; - moment, situation 261; — perspective 408; — reason 285; — research 268 historicism 31, 306, 499, 502; historico-comparative approach 9, 72, 116, 129; historico-cultural realm 149 historiography 110, 120, 132, 509 historique du probleme 168, 191 history, cultural 272; — general 50, 163, 222, 503; - intellectual 50; - social 148; ֊ of a language 7f., 24; - of linguistics 10, 49f., 80-3; - of litera ture 129; - of scholarship 339; — of science 172 hobby 503 "hold-out" (= stubbornly resisting) 313 homely (situation) 221 homo-geneity 131, 141, 289; -nymy 17, 56,97, 106,145,160,199,214, 218f., 221, 253, 306f., 309-12, 326, 345-7, 352, 374, 431f., 451, 454, 456, 505f., 518; -phony 146, 201, 235, 312, 354, 417; -(o)rganicity 94 to hone (a word segment into . . . ) 292 horizon, narrow 441 horror aequi 266 hospitality (fig.) 291, 465 human-s 156, 169, 193, 205, 222; -ities 11, 72, 502; -ists, -ism 110, 129, 124,139,147,503 humbleness (of a register), humility 245 humor 56, 198, 231, 275, 407, 427, 435, 509 hushing-up 510 hybrid(ism) 166, 192, 206, 254, 431 hydronymy 102, 164, 170, 298, 432 hyper-characterization 7, 10, 39, 90, 100, 150, 222, 231, 234, 258, 260, 263f., 269, 276, 290, 350, 364-7, 432, 467, 517; - correction 246,

SELECTIVE INDEX OF KEY TERMS 370, 373, 304; - sensitivity 310; — trophy 163, 307, 316; - urbamsm 41,267,394,510 hypocharacterization 8, 39 hypocoristics 143, 156, 159, 161, 167, 229,231,247,402,432,435 hypothesis 108, 117, 117, 143, 147, 173, 179, 182f., 191,193,202, 213, 230f., 251, 261, 268f., 289, 295, 299, 302, 324, 342-4, 359f., 370, 384, 387f., 415, 423, 426, 430f., 437, 447, 451, 460, 462, 466,505,509f.,518 hypothetical (= inferred, reconstructed) base 93, 101,187 iceberg (fig.) 72 ichthyonym 429, 463 iconography 433 idea, ideal 212, 223, 323, 500, 519; "ideal" (expectation, form) 36, 301, 314, 350, 382; ideals 82, 221 ; ideal ism (philosophical) 55, 268; idealiza tion 64, 217, 238; idée fixe 191; ideology 3, 71 identification 8, 72, 93, 117, 188, 197f., 235, 238, 246, 261, 266f., 269f., 297, 331, 351, 355, 357, 403,405,463,465,469,517; identity 133, 199, 212, 245, 303, 363f., 414 417, 423f., 432, 438, 445,451,469,508 idiolect 108, 358; idiom 141; idiomaticity 74, 349; idiosyncrasy 71, 88, 234, 276, 343, 345, 358, 393, 399, 412,455 ill-assorted 309; -defined (force) 237, 290, 498; -fated 520; -placed (sound) 388;-prepared 325 illative (prefix) 18, 327, 355 illiteracy 306 illumination, mutual 93 illusion, illusoriness 230, 267, 505 illustration (pictorial, verbal) 29, 162f., 182, 188, 193, 221, 229, 234, 295, 329, 336f., 341, 345, 377, 404, 415,

627

425,447,497,510,521 image(ry) 52, 169, 224, 226, 268, 283, 299,500,504,510 imagination 27, 73, 77, 79, 87, 99, 117, 157, 195,198,214,261,333,351, 372,378,451,502,511,517 imitation 93, 133, 184, 187, 196, 202, 213, 273, 306, 370, 373, 384, 402, 413f.,432,435, 467 immediacy 109, 119, 148, 163, 200, 285, 332, 339, 402, 416, 515, 521 immigrants' (milieu, speech) 32, 140 immobilization 241, 259 immunization 499 impact 90f., 121, 124, 133, 137, 148, 159, 212, 224, 229, 234, 236, 243, 247, 298, 308, 316, 342, 370, 387, 431,499,511 impairment 298, 302, 452 impasse 123, 231, 374, 508 imperative 104f., 106, 191; (fig.) 516 imperfect tense 106, 122, 284, 369, 466 impetus 208, 352 implication (chronological, etc.) 91, 96, 104, 137, 181, 220, 223, 233, 272, 277, 327f., 341-4, 276, 408, 424, 428f., 431, 433, 440f., 452, 461, 497, 500f., 506, 509f., 515, 519 importance (of event, factor, ingredient) 233, 302, 305, 316, 324, 373, 379, 385, 465, 469 import(ed) (material, word) 112, 242, 287, 356, 439 impoverishment, phonetic 457 impression (first) 93, 383, 288, 337, 342, 351, 368f., 405, 436, 501f., 517; impressionistic (statement) 307, 406; imprint 377 improvement (of language) 23, 251, 268,461 impulse 342; impulsive (speech) 117 inaccuracy 191, 211, 424 inadequacy 9, 148, 264, 329, 343, 367, 445 inanimacy 155, 159, 165, 169f.

628


inchoative 331, 520 incidence (obligatory, optional) 39, 41, 162, 203, 254, 257, 328, 507;inci dent 359; incidental (borrowing, concern, emphasis) 40, 132, 251, 264, 403, 421, 430, 436, 461, 519 incipient (phase) 169, 298, 465, 469 incomparability 146 incompatibility 74, 219, 279, 284, 289, 297,312,327 incompleteness 500 inconclusiveness 277, 378, 419 incongruity 179, 220f. inconsistency 76, 282, 333 incorporation (= absorption) 310 increasing (complexity) 222, 355, 360, 513; increment 279, 427f., 443, 447, 469 incrustation 198, 353 incubation (of fashions, ideas) 75, 93, 145, 185 independence (of conclusion, control, knowledge, mind, observation, survival) 73, 75, 87, 89f., 103, 130, 209-11, 222, 231f., 263, 275, 284, 291, 294, 297, 300, 380, 385, 391, 393, 408, 420, 430, 444, 453, 459, 464 indeterminacy 13, 143, 222, 316, 422 index 499f. indifference 96, 100, 137, 158, 502, 507,511 indigenous (channel, form, stratum, words) 147, 157, 202, 209, 256, 334, 428, 430, 432, 439, 439, 442; indigenismo 444 indirect(ness) 286, 339, 382, 468; - dating 87; - reflexes 129 indiscriminacy 164, 469 individual case 401 ; - form 93, 96, 107, 117, 167, 219, 361;-growth 217, 255; — history, background 30, 36, 196; — language (within a family) 204, 266; - person 198, 222, 298, 406; - preference 297; - word 194, 219, 229, 499, 505; -ism, -ity 71, 73,

161 ; to individuate 300, 368, 428 indivisibility 452 Indo-European (language, linguistics) 71, 76, 155, 158, 265, 501, 503, 514 inelegance 314 inertia 131 inexplicitness 519 inextricability 117 infelicity 91, 192, 196, 246, 310, 358, 361,374,461,519f. inference (right or wrong) 13, 113, 122, 124,131,141,159,187,192,203, 264, 267, 291, 315, 323, 331, 340, 342, 359, 395, 417, 437, 460,469, 513 inferiority, linguistic correlates of 452 infiltration (lexical, etc.) 11, 44, 121, 133, 137, 207, 254, 260, 275, 294, 346 infinitive, infiniteness 104-6, 113, 1179,133,180,217,231,262,278, 345, 353, 369. 375f., 378, 380, 384, 403, 405 infixation 7, 331, 445, 458 inflection 34, 90, 94, 96, 104, 107, 125, 186, 206, 229f., 240, 264, 271f., 290, 292, 298, 352, 380, 391 influence 71, 74, 81, 80, 96, 99, 106, 121, 132f., 137, 144, 158, 164, 182, 185, 192, 205f., 210, 213, 230, 233, 236, 239, 241, 247, 260, 267, 271, 277, 293, 298, 307, 309, 311, 315, 326, 331, 335, 339, 344, 346, 368, 376, 394, 423, 425, 429, 432, 434f, 439, 443f, 453, 461, 500f., 504f. influx 133, 144, 203, 212, 221, 272, 366,371,387,444 informality 509 information 82, 88, 96, 107, 156f., 164, 182, 194, 280, 284, 315, 324, 334, 341, 353, 376, 407, 424, 427, 435, 455, 502, 509; ֊theory 75 infrequency 189, 327, 334 ingeniousness 185, 339 ingredient (= element) of a situation 38,

SELECTIVE INDEX OF KEY TERMS 93, 123, 143, 156f., 165, 219, 224, 232, 237, 247, 254, 261f., 168, 292, 307, 312, 325, 344, 362, 371, 381f., 384, 390, 431, 434, 462, 469, 503, 511, 514; ingressive 327, 355 inherency 212, 337 inheritance (fig.) 131, 285, 339, 348, 423f. inhibition 333 initial (phase, stage; consonant, seg ment, syllable, vowel) 55, 91, 93, 95, 109, 198f., 203, 247, 298, 312, 325f., 338, 341f., 348f., 354, 357, 362, 366, 380f., 447, 459, 520; initiative 362, 514; miriate 508 injection 378, 419 inner (diffusion) 447; -(sanctum) 72; in(ner)most chamber (fig.) 292; innere Struktur, Sprachform 67 innovative (methods, theories, words) 6, 10, 40, 138, 141, 272, 303, 326, 344, 348, 351, 367, 411, 416, 424, 442, 469; innovation 87, 90, 97, 103,107,111,145,157,163,179, 183, 193, 196, 199, 202, 209, 212, 221, 246, 265, 275, 290, 314, 326, 333, 363, 375, 409, 429f., 469, 503, 516,520f. inorganicity 259 inroad, lexical 212 insecurity 282, 385 insensitiveness 507 insert(ion) 269, 276, 284f., 290, 292, 330, 345, 376, 430, 438, 441, 504, 520 insight(fulness) 148, 374, 508; inspec tion 90, 92, 157, 179, 295, 329, 342, 348 insoluble 508 instability 466 institution (social) 73, 265 instrument 193, 205; -al phonetics 76; instrumentality 234, 258, 331, 431; instruments (musical), names o f . . . 435 intactness 209, 278, 281, 286, 303, 508

629

integration (of details, parts, results) 34, 72, 143, 146, 361, 419, 447, 503 intellectual (appeal, climate, experience, refinement) 51, 59, 147, 149, 161, 499,516 intelligibility 74 intensi-ty, -fication 117, 182, 208, 214, 278, 287, 295, 344, 347, 421, 444, 466,509,515 intentionality 146, 435 inter-action 119, 131, 144, 196, 209, 223, 232, 288, 298, 366,445, 501; -calary (element, sound), -calate 104, 269, 272, 277, 282, 284-6, 290, 348, 356, 422, 438, 445, 465, 467, 515; -change 210, 331, 348, 439; -connection 122; -cultural (marriage) 66; -dependence 301, 312, 518; -dialectal 189, 283; -ference 104, 117, 140, 147, 172, 212f., 229f., 238, 240, 253, 262, 264, 297, 311, 327, 331, 333, 365, 371, 389, 394, 435, 438, 462; -fixation 7, 15f., 231, 262, 290, 292, 369, 405, 412, 419, 430, 445, 447, 467, 497; -jacency 244; -jection 293; -locking (patterns) 9, 19, 124, 204, 211, 238, 299, 513; -lude 74, 301; -marriage 514;-mediary 252;-mediate (pattern, stage, step, zone) 19, 146, 170, 238, 291, 293, 302, 341, 357, 406, 415, 460; -mittency 120, 223, 419; -necine (conflict) 169; -penetration 90; -play (of causes, forces, forms) 39,42,54,72,91,93,104,117, 165,172,187,195,213,219, 223-5, 236, 242, 261-3, 269, 273, 277, 279, 283, 294, 298, 302, 312, 315,332,354,361,366,385, 404, 419, 422, 440, 445f., 513; -position 289; -pretation 161, 168, 170, 197, 222, 285, 304, 312, 214, 330f., 337, 343f., 346, 355, 361, 373, 382, 413, 435, 437, 504f.; -rela tion 384, 443; — Romanic 369; -rupted sequence 372; -section 274,

630


405, 409, 442; -tonic (vowels) 75, 94, 240, 333, 340, 357, 361, 363, 367f., 370f., 383, 385, 428, 444; -twining 339, 504, 506, 521 ; -vention 103, 263; -version 167, 304; -viewing (techniques) 50, 503; -vocalic 210, 274, 280, 376, 3924, 466, 506; -weaving 212 internal 251 ; - analogy 112;- conflict 516; - criteria 142; - diffusion 149, 179, 240, 279; ֊ evidence 257, 349, 391; ֊ force 297, 299; ֊ profilera tion (spread) 90, 233; — reconstruc tion 76, 87f., 187; - segmentation 148 interval 180, 261, 264, 302, 310,421, 427 intimacy 59, 61, 104, 138, 211, 221, 509 intralexical (spread) 310 intransitive 193 intricacy 6, 42, 240, 266, 269, 299, 304, 335, 356, 404, 419, 443, 506, 513 intrinsic (superiority, value) 110, 125, 171,468,510,521 introspection 49, 171 intrusion 99, 132, 146, 234, 255, 262, 264, 274, 289f., 303, 327f., 331, 348, 350, 384, 389, 432, 442, 444 intuition 29, 79, 360, 507 invader (word) 40; invasion 334 invariability 259 inventory 39, 58, 93, 96, 105, 113, 119, 130, 156, 162, 162, 164, 172, 238, 242, 263, 272, 286, 297, 300, 310, 359, 374, 392, 409, 460, 465, 518 involuntary 406 involvement vs. detachment (analyst's, cultural) 25, 298, 460, 503 iridescence, semantic 432 irksome (feature) 258 irony 180, 211, 266, 274, 360, 510 irreconcilability 122, 305, 336, 514, 519 irregularity 94, 221, 285, 306, 379,

392, 507 irrelevancy 98, 111, 238, 305, 509 irremediability 499 irretrievable (loss) 68, 355 irreversibility 109, 327, 406, 428 irritant 93, 95, 406 islet (fig., of speech) 125, 164,454 isogloss 219, 282, 431, 444 isolation (of a form, a factor) 38f., 41, 93, 95, 103f., 109, 144, 146, 184-6, 198f., 208, 219, 222, 229f., 246., 256, 260f., 267-9, 293f., 299-301, 306f., 309, 312f., 326, 334, 339, 355, 358, 364, 366f., 371, 376, 378, 380, 391, 394, 401, 406, 412, 414, 420, 435, 439f., 445, 452, 454, 462, 467, 500, 505, 508, 510, 526; isolate 517, 521 isomorphism 97 issue, controversial 238, 310, 323, 327, 353,434,466,507,517 Italiamsm 137, 139, 145, 150,182, 254, 300, 505f. item, lexical 155, 260, 271, 308, 422, 426, 429, 432, 443 iterative (verb) 182, 289, 287, 347, 392, 452 itinerary 131,225,283 ivory tower (fig.) 81 jargon (soldiers', etc.) 200, 208, 340 jeopardy 273, 333 "jettison" (fig.) 335, 428, 508 jocularity 425 joint (agency, consideration, effort, evidence, momentum) 101,103, 263, 268, 399, 467f., 510,518 jolt (fig.) 94, 117 journalistic (style) 305 juggling (with excess of sth.) 329 juncture (fig.) 90, 131, 156, 159, 286, 298,302,339,433,499,515 Junggrammatiker (= Neogrammarians) 217f. juridical (words) 282, 386; jurisdiction (ofalaw)251


justification 208, 240, 247, 264, 301, 306, 312, 328, 336, 351, 355, 429f., 451,504f. juxtaposition 419, 433 kernel, lexical 17, 159, 254, 339; — semantic 407, 517; ֊ of a group 192, 217, 414, 467; ֊ of inquiry 104, 360; ֊ of a root morpheme 509 key-concept 72, 314; ֊ feature 430; — issue 431 ; — passage 269, 519; — problem 148; — question 138; — role 462; — section 87; — study 435; — tense 97; — word (= leader word) 44, 208, 422f. kinship (term) 331, 338, 509, 514; — (between languages) 432; kindred (words) 500 knot (fig.) 301, 304 knowledge 224; know-how 516 Kulturpolitik 130; -sprache 147, 151 label 119, 134, 156, 159, 164, 189, 192, 204, 258, 265f., 269, 305, 422, 441 labial 255, 281f., 286, 294, 363, 367; -ization 332f., 428, 468; labio dentals 235, 237, 243, 382f.; -velars 12,41,294 laboratory (case, expertise jargon) 17, 73, 232, 264, 508 laconicity 208, 340 lacuna 7, 24, 38, 56, 123, 142, 204, 231, 245, 261, 308, 329, 395, 406, 409, 499f. lag 147, 211, 241 "lame" (argument, conjecture) 230, 306, 384 landmark (fig.) 339; landscape, terms of 443 language growth 103; — historian 104, 132; - history 27, 297, 305; - and literature (disciplines) 4; — specificity 235; languages in contact 138, 371 langue (vs. parole) 71 lapidary formula 324, 340, 411

631

lapse (of time) 225, 269; lapsus calami 267 large-scale 327, 436, 513 latency 42, 113, 146, 214, 268, 299, 303, 309, 312, 317, 369, 378, 390, 421,434,444,463 lateness, late date (of appearance, etc.) 192, 198, 233, 257, 260, 290, 294, 336,412,427,433,436 lateral area 103; - consonant 367, 440; - evidence 460; laterality 235, 452; latitude (log.) 156, (sem.) 453 Latinizaton, Latinisms 102, 134, 146, 274, 283, 290, 311, 313, 347, 349f., 365f., 384, 401, 414, 431, 435 lavish-ing (of attention) 433; -ness (of detail) 316, 426 "law" (of sound change) 35, 222, 268, 323f., 327, 340, 345 laxity 44, 69, 201, 225, 394, 434 layer, lexical 98, 253, 325, 353, 388, 428, 432; - , social 394; - , temporal 1 8 8 ; - , textual 115, 147 lay(man's) curiosity 69, 131, 147, 224, 499 leader word 193, 199, 206, 309, 468 league of sciences 73 "leak" (fig.) 100 leaning (on) 430; - (toward) 239, 242, 342,369,416 "leap" 205, 231, 347, 366, 467 learned borrowing 146; — transmission 19, 36f., 94.157, 198,219,231, 252, 255f., 259, 268f., 272f., 282, 288, 290, 307, 309f., 334, 350, 353f., 368, 374, 383, 385, 389f., 393, 400f, 412, 419, 424, 428, 434-6, 441, 444, 452, 454, 467, 505f.,508, 518 learning (process), a linguist's 4, 346 lecture(s) 266, 421, 433, 497, 502 legacy (fig.) 147, 409, 428; legal domain, texts) 114, 196, 201, 210; legislation 132; legitimacy (of doubts) 327 legends (as a clue to semantic shifts) 333

632


legionaries' jargon 200 leisurely (consultation, discussion, pace) 137,295,422,499 length (of articulation) 211, 306, 360; — (of a period of time) 142; — (of a word) 143, 198, 308, 310, 313f., 355, 409; lengthy 301 ; lengthening 184,306,356,360,365,409 lento form 392 level, dialectal 114, 120, 124; - , etymo logical 517; - , social 167, 219; -, stylistic 455; —, temporal 115, 193, 235, 369; - of discourse 116, 172, 185, 201, 209, 212, 232; level ing (force, process) 13, 34, 36, 44, 91, 94, 99, 112, 125, 219, 223, 235, 239f., 256, 274, 277f., 284-6, 370, 380,412,504f. lexical (blend) 334, 372, 377, 405, 497, 516; - (carrier) 340; - (conflict) 230; — (contamination) 213; — (deri vation) 206; - (element) 308; - (ex change) 1 4 6 ; - (family) 119;-(field) 409; - (item = unit) 144, 222, 242, 326, 393, 506; - (loss) 339; — (material) 433; — (morpheme = lexeme) 302; — partner (302); — (polarization) 167, 208, 213f., 230, 254, 303, 316, 348, 459f., 462; — (replacement) 379; — (resistance) 439; ֊ (slant) 298; - (stock) 283; — (system) 2 8 8 ; - (vicissitudes) 265; lexicalization 166, 194, 411 lexicography, -logy 8f., 30, 33, 110, 139,141,146,150,162,191,197, 300, 307, 451, 455, 498, 503, 507; lexicotactics 457 lexico-derivational (problem) 163; — grammatical 422; — morphology 243; — semantic (process) 7, 140 lexicon, lexis 133f., 140, 148, 183, 190, 266f., 325, 327, 356, 369, 420, 454,513 liberal (representation) 294; liberation 199 library records (as a source of informa

tion) 60 life-span (also fig.) 65, 222 light (= illumination) 430; light-weight (words) 198 limit, chronological 212; — stretchable 498; — ultimate 426; — on possi bilities 516; — on scope 251 ; -ation 259f.,361,444,507 line, direct 312; —, evolutionary 161, 437; -, straight 202, 225, 434; - of descent 134, 374; — of transmission 460; hneal descendant 253, 297; linear analysis 331; linearity 71 lingering (on) 199, 203, 209, 274, 315, 394,412,452 linguistic atlas 56, 513; — context 276; - geography 76f., 88, 161, 259; - map 123; narrowly - 133 linguistics 72, 76f.; advanced — 503; descriptive — 24; diachronic - 26, 53, 87, 503; evolutionary - 24f.; explicative - 40, 506; general — 434, 446f.} 501, 503; genetic - 23, 25, 27; historical - 23-6, 297, 316, 511 ; historico-comparative — 513; Indo-European — 26; modern — 498; old-style - 23, 81 ; pure - 110, 503 link (hierarchical, redeeming, semantic), linkage 112f., 166f., 197, 199, 211, 217, 236, 238, 245, 255, 257, 288, 301-3, 315, 341, 345, 351, 353, 377, 380, 387, 405, 407f., 415, 462, 465,502,511 liquid (consonant) 439 literacy 276; literal(istic) 309, 508; literary factor 134; — history 87f., 272; - language 15, 94, 114, 121f., 183, 305, 314, 372, 455; - period 241 ; - texts 211 ; - usage 110; literate (speaker) 460; literature, fine 73 liturgy 132,210 live (= living), being 161, 180, 205, 406; - meeting 499; ֊ speech 270, 295, 465; liveliness (of growth, of speech) 219,254,283,310,435


load (semantic) 144, 208, 315, 378 loan-blend, -shift, -word 131, 138-40, 148, 150, 300; -homonym 146; — polysemy, -synonym 146; - translation, — creation 145f. local (choice, conditions, feature, product, scale, scene, usage) 99, 108, 202, 269, 290f., 293, 372, 426, 453, 465; locale 193, 312; localization 93, 107, 121, 293, 295, 372, 426, 453, 465; locative case 462; locus 147, 300 logico-grammatical structure 148; — mathematical approach 73 long-accepted 268; — active 244; -con cealed (data) 498; -distance perspec tive 226; -drawn-out 210, 299, 313, 349, 425; - hidden 164; -influenced 500; -misunderstood 392; -over looked 232; -range (dating, effect) 89, 161; -term (research) 204; longevity 519 loophole (in reconstruction) 304 loose (ends) 498; to cut - (from.. .) 294, 304, 352; looseness (of connec tion, group, structure) 95, 268, 275, 299, 419, 431, 434, 454; loosening 131,416 lopsided (analysis) 431 loss (of construction, form-sound) 94f., 211, 302, 332, 334, 336, 354, 356, 374, 376, 382, 385-7, 389, 391, 436, 466; — (of face) 499; — (of perspec tive) 137 low (degree) 261 ; lowering (of vowel) 302 lukewarm (fig.: response) 426 "lumping together" 368, 434 Lusism l37,423 luxuriance 285, 456; luxury 506 lyrical (style) 416 machine (translation) 75; -ry 197, 205 macrocosm 217 maelstrom (of events) 98 magic 457

633

magnet (fig.) 104, 328; -ic (personality) 78 magnum opus 140; magnitude = caliber 195, 298 main course of events 99; -road (fig.) 77; -stream 200, 301f., 371, 386, 439, 511 major (branch, figure, language) 96, 155,323;-ity 132, 134, 143, 161f., 213, 243, 384, 388, 394, 406, 424, 439, 441f., 510 malaise 312 male chauvinism (implications in lan guage) 167 mandatory 278, 327, 457, 479, 505 maneuver 93, 309; - ability 25 manifestation (at the surface) 117, 316 manners, vocabulary of 133 manu-facture (fig.) 162; -script 118, 501 many-faceted 230; -pronged 118, 306, 338;-sided 513 map (cartology) 77, 98, 123, 164, 168, 189,220,257,288 margin (liberal, narrow) 118, 171,289, 298, 442f., 445, 482; -al (area, representation) 92, 94, 103, 108, 187, 194, 239, 245, 252, 276, 305, 307f., 330, 335, 359, 487, 489, 493, 500, 502f. mark-er 99, 105, 109, 292, 315; -ing 93, 159, 164, 172, 234, 260, 263, 265, 276,281,286 martial sports' jargon 202, 204, 208 masculine 143, 155-9, 162-6, 171, 181, 183f., 188f., 191,193f.,201f., 204-6, 208-10, 213, 263, 276-8, 305, 438, 457 mask(ing) 256 mass-noun 160, 163f., 272, 389, 399, 432, 447;-iveness 498 master-mind 159;-piece 168, 426; -pioneer 477; -ship 88, 130 match(ing) 105, 164, 256, 258, 278, 291,370,393,417,467,495 material 74, 107, 115, 123, 158, 348;

634


- civilization 23, 139, 156, 297, 316,423 mathematical (approach) 73, 75, 78; -ization 223 matrix (cultural, social) 63, 94, 179, 185,200,204,208,245 maturity 164, 168 maverick (individual) 298 maxim 219; -urn 231, 242, 258, 260, 263, 507 maze 462 meagerness 168, 191, 194, 201, 203, 265 meandering (line) 225, 455 meaning 27, 43, 104-6, 160, 192, 194, 200, 210, 274f., 349, 351, 406, 414, 416, 420, 422, 427, 432, 434, 457, 483f., 487, 500, 508;-ful 108, 222 means 265, 456 measure 161 ; -ability 240; -ment (of strength, etc.) 18,90,99 mechanism 71, 73, 83, 95, 97, 109, 124, 156, 166f., 179,283f.,349, 380,401,464,472,508,513 medi-al (position) 236, 302, 335, 389, 486, 507, 509, 511;-ation 143, 197, 200, 224, 235, 349, 452, 486; -ator 169;-um 137 medical (terminology) 33, 165, 347, 453 medieval (culture, dialect, form, record) 68,92,96, 105,108,113,234, 236, 252, 256, 258, 272f., 302, 309, 311, 349, 353, 357, 380, 385, 391,458f.,464, 493;-ist 112 meeting 498f. melioration 43, 143 membership (in cluster, family, group, set) 106, 162, 237, 240, 277, 279, 302, 315, 335, 352, 355, 389, 469, 476, 494 memoirs; memoiristic essays 59, 67; memory 508 menace 378, 472 mental 490, 494; -ism 71, 75, 82;

-ity 297;map 123 merger 125, 157, 161, 183, 191, 217, 231, 244, 247, 263, 286, 299, 303, 325, 351, 385f., 399, 457, 483, 494, 515 message 41, 90, 108, 156, 194, 325, 359, 416, 452, 463, 489 meta-language 141 ; -morphosis 223; -phony 5, 13, 96, 108f., 236, 241, 254, 257, 288, 349, 357, 366, 459; -phore (in linguistics) 32, 71, 92, 218f., 224f., (in speech) 124; -thesis 37-9, 94, 112, 219, 234, 240, 256, 258, 260f., 263, 270, 281, 283, 285f., 288, 291f, 334f., 351, 355, 365, 379f., 425, 438 meter 36, 257, 457; metrics 52, 88 method 72, 76, 83, 87, 330, 500f.; -ic 104, 309; -ology 8, 10, 26, 49, 74, 88, 138-40, 225, 232f., 327, 511 meticulosity (of research) 24, 80, 229, 311,428,435,463,474 metropolitan (speech) 124 micro-scopic (inspection) 131, 139, 184, 226, 231, 295, 329, 342, 348, 286, 406, 424, 485, 507; -toponymy 432 mid-century 195, 272; -point 348, 500; Middle Ages 70, 483; middle ground 105; - of the road 4, 76 (cf. medieval) migration 337, 369 miHeu 119, 140, 179, 184f., 192,200f., 203 milit-ancy 140, 499; -ary parlance 12, 185, 200, 203, 205-8, 258; -ating 306 min-iaturized (projection) 70, 150, 161f., 172, 509; -imal (pair, suffix, vowel) 192, 341f., 384, 392, 495, 506;-imum 138, 198, 264, 270, 326, 446; -or detail, jolt) 94, 125; (sound change) 34, 271 ; -ority 109, 288, 316, 363, 388, 406, 424, 474, 484f., 510;-uteness 24, 162, 168 minting (fig.) 141, 182, 203, 210

SELECTIVE INDEX OF KEY TERMS mirror(ing) 140 mis-alliance 438; -conception 316; -leading 97, 160, 172, 384;-nomer 24; -placement 114; -print 327; -reading 120; -spelling 111, 114; -taking 114, 191, 468; -under standing 119 missing (a point) 195 mission 171;-ary 81 "mixed" ("strong"  "weak") 111, 430; mixture (of dialects) 90, 218 mobile (design) 302 mock-diminutive 212; -Latin 502; -learned 245, 255; -participle 404; -respectful 192 mode (of thinking) 297; -al 305 model 73, 93, 105, 109, 113, 117, 119, 170, 179, 181, 183-5, 189, 191f., 198, 206-8, 210, 212, 232, 235, 240, 258-61, 278, 284, 290, 306, 309f., 311, 327, 333, 348, 357-9, 364, 367, 375, 379f., 384, 392, 399, 413, 416, 428,441,475,479,515 moderation 76 modern (dialects) 94, 95, 124; -(lan guages) 24, 155, 243, 252; -(linguistics) 68; -ism, -ity 4, 6, 74, 76f., 79f., 141, 233, 477, 497f., 500, 503, 506;-ization l l 0 , 120 mod-icum 270;-ification 150, 162, 170, 186, 256, 261, 293; -ulation 88 molecule (grammatical, lexical) 182, 199,208 momentum 140, 158, 268, 289, 355, 401,438,447,514 mono-dialectal 290; -geneste 130; -graphic (research) 27, 73, 101, 137, 139, 159, 164, 229, 262, 270, 345, 361, 454, 498, 510; -lingualism 42, 162, 268; -lithic (book) 81, 162, 270; -nuclear 300; -phthong(ization) 92, 229, 231, 233, 235-40, 242, 247, 280f., 285, 311f., 324 241, 343, 350, 362, 382, 389f. ; -poly 98; -syllabic 302, 344; -tony 97, 139 monumentality 495, 498

635

mood (gramm.) 509; -y 502 mooring 179, 515 moral (qualities) 486 moratorium, temporary 27 moribund 118,388,417 morph(eme) 45. 421, 438, 490; -boundary 97, 355; lexical or root - 302; -ic 196, 202; -ological (con straint, distress, interference, shift) 9, 12f., 15,34, 109, 123, 130, 232, 371; (design, level, tendency, variant) 92, 117, 208, 219, 326, 353; -ology 8, 12, 95, 125, 143, 229f., 236, 244, 246f., 263, 2724, 283, 286, 298, 305, 328, 338, 346, 381, 400, 412, 427f., 434, 438, 451, 481, 504, 511, 514f.; -ophonemics 36, 234, 241-3, 256, 390, 504f.; -osemantics 94, 96 mortgage (fig.) 514 mosaic (fig.) 34, 87, 108, 503 motion, to set in 117, 125, 129, 254, 262, 270; motive 266; -ation 199, 338, 346, 370, 378, 438, 467, 472, 604 motley (group, picture) 99, 138, 196, 207 mo(u)ld (fig.) 134, 197, 277, 338, 355, 394, 426 move (= gambit, leap) 99, 141, 205,313, 360, 498, 507, 513;-ment(of thought) 74f., 159, 171,419,510f. multi(ple) causation (= pluricausality) 9f., 12, 15, 41f., 103, 260f., 269, 283, 295, 299-301, 305, 307f., 317, 326, 332, 347, 362, 366, 462, 475; -commitment 66՛, -conditioning 220, 247, 250; -farious 265; -formity 91; -licity 100, 123;-pronged 93, 125, 157, 257; -tiered 494; -tude 125, 286 municipal (document) 108, 115 Murmelvokal 341 musical (notation, terms) 70, 194 mutation 41, 266, 472 muteness 211, 230, 305 mutilation 96, 210 mutual (attraction) 100, (illumination)

636


93, (reinforcement) 43; -ity 193, 253, 260f, 268, 286, 289, 301, 312, 428 myst-eriousness 143, 500, 514; -icism 509 myth-ical 132; -ology 27, 71, 208, 498

ing) 336f.; -ism 140, 195, 495 neglect 203 neighbor 113, 348;-ing 118, 288, 376, 394,451 Neogrammárians 26, 29f., 70-2, 75f., 217-21, 266, 327f., 336, 340, 473; — Hellenism 137; — Latinism 134; naïveté 155,202,328,497 — linguists 71, 73f., 77; -logism 138, name, first 271 ; —, personal 159; name = 185, 278, 307f., 488;-phyte 81 designation 162, 165, 169 network (of references, relationships) narrative (prose) 76, 116, 124 77,90,156,224,380,444 narrow (scope) 7f., 327; -ly (linguistic) neuro-muscular (skill) 91 133, 451;-(ing) down 264, 489f. neuter 155, 157f., 160, 165,263,411, nasals(s) 293, 327, 331, 338, 348, 3'55, 413, 416, 454, 457; neutral (vowel, 359, 363, 367, 441, 458, 485, 510; etc.) 181, 341, 390;-ity 237;-iza-ization 140, 213; - insert (epen tion (of contrast) 17, 162, 166, 169, thesis) 328, 330, 339, 345-7, 349f. 209, 286, 409 nascent (form, language) 197, 201 newer (style) 141 ; newly-developed nationalist bias = chauvinism) 69, 74; 209, 425;-emerging 297 - culture 63, 88; -language 134; niche (fig.) 109, 160, 182, 192, 293, — style (of linguistics) 78; — team 71 345, 381 native (descent, speaker) 141, 171, 194, nickname 191, 374 207, 212, 287, 331, 370, 384, 439, nomenclature 134, 165; nominal 158; 451 -augment 361, 419f., 433, 441, 443; nature, fact of 159; -al (course, develop -zation 180, 191f., 254, 385, 392; ment) 109, 257; (philosophy) 169; nominative 411 -alness 166 nonce 401 nautical (term) 230, 287, 336, 347, non-committal 251, 358, 371, 478, 505; 371; navigation 23 -compliance 241 ; -etymological 285; near (= quasi-) atrophy 130; -concor -existent (language) 35; -finite 122; dance 486; -fatal 513; -homonymy 8, -linguistic (factor) 131 ; -obligatory 14, 106, 201, 307, 310-2, 345;-inter 214, 290; -phonological (change of changeable 331 ; -Latinism 401 ; sounds) 219, 304 -opposite 414; -parallel(ism) 180, norm 44, 93, 103, 124f., 263, 301, 328, 330, 332, 339;-symmetry 239; 364f., 394, 505; -al 99, 107, 109f., -synonymy 8, 36, 100, 188,196, 119, 158, 165, 247, 255f., 258, 199,254,256,283,291,395-7, 261,302,357,439,444,453; 367,406,481,493 -ative (grammar) 68 neatness 82, 105, 130, 132, 134, 156, nostalgia, linguist's 56 198, 243, 356, 258, 263, 272, 286f., not-arial (jargon) 213; -ation 301 301, 359, 380f., 412, 427, 438, noun (= substantive), nominal 105, 463, 481 156f., 162, 171, 179f., 184, 188, necrologies (as a source of information) 194, 197f, 200, 203, 212, 234, 60, 63, 67 236, 239, 243, 254, 258, 260, 271, need 239, 266, 339, 395 278f., 285, 290, 293, 311, 356, 372, negative (fact, feature) 168, 405; (mean405, 451f, 457, 467, 469, 471, 484,


509;-phrase 471 novelty (= innovation) 90, 94, 103, 206, 289, 326, 487 nuancing 72, 80, 100, 141, 160, 162f., 166, 168, 204, 212, 229, 236, 266, 307, 323, 337, 412, 462, 492, 498 nucle-ar (cluster, diphthong, vowel) 92, 108, 114, 180,303, 384;-us (of speech or speakers, statement, text, words) 32, 44, 76, 94, 101, 111, 259f., 344, 407, 428, 442, 447, 452, 475,479,483,491 "nuisance" 504 numbering) 114, 211, 238, 305, 344, 487, 506; numerals 389 nursery (words) 425 obedience (fig.) 119 obfuscation 325, 342 object 155, 158f., 162-5, 181;-case 416; -ion 245; -ionable 109, 100, 315; -ive (ո.) 498; -ivity 64, 159 obligatoriness 192, 206, 263, 275, 314f.,326 oblique (case) 170; -ness (of clue, representation, view) 99, 123, 294, 461 obliteration 258, 453 oblivion 328, 468, 501 obscurity (of dialect) 92, 236, 245, 431, 435,473,504, 507f., 510f. observation, 5, 72, 78, 80, 89, 95, 99, 109,118,122,124,155,159,171, 184, 186, 239, 242, 245, 248, 258, 297, 301, 308, 314f., 326, 344f., 383, 452, 463, 468, 474 obsession 72 obsol-escence, -eteness 9, 92, 96, 99, 106, 139, 167, 198, 205, 234, 241, 252, 254, 271, 309, 325f., 365, 368, 511,515 obstacle 119,203,240,309 obstruction (of the view) 137, 348, 380, 385; obstruent 130 obvi-ation 256, 439; -ousness 112, 119, 459

637

occasion, -al 97, 122, 141, 160, 255, 469f., occup-ancy (of a niche) 160, 182;-ation 180, 198 occurrence 303, 477 oddity 294, 451 'odor' 275 off-shoot 102, 188, 243, 272, 275, 294, 329f., 339, 350, 393, 446, 453, 463, 465, 474, 479, 493, 515; -spring (fig.) 187, 201, 209, 212, 485, 494 offender 314 old(er)-style (scholarship) 77, 141, 170, 278, 498; -er (school of thought) 98; (languages) 155; -er, -est (forms) 184,447,467 omission 158, 181, 303, 359, 510 one-sidedness 191, 262, 495; "one world" (climate) 58 onomast-ics, -icon 52, 77, 170, 237, 266, 325, 332, 426-8, 431f., 434, 513 onomatopoeia 42, 77, 195, 233, 267 onset 213, 343 onslaught (= irruption) 199, 239 oozing 117 opa-city, -queness 36, 182, 195, 237, 275,329,425,493,508,510 open (letter, genre) 60; -mindedness 6, 27, 30; -ing (segment, syllable) 302, 326,331,335,338,462,483 operation 73, 82, 93, 104, 117f., 138, 140, 165, 168f., 187, 197, 204, 224, 236, 240, 254, 261, 291, 316, 328, 331, 348f., 353, 403, 451, 460, 477, 498,503,514 opinion 305, 323, 473 opportun-ism 4; -ity 214, 258, 506 opposit-e (direction, end) 92, 107, 139, 211f., 214, 255, 258, 260, 267, 276, 298, 302, 314, 331, 343, 362, 386, 458, 472, 475, 487, 502; -ion (to change) 77, 167, 241, 283, 340; (polar) 292, 352, 376f., 380 optim-al (circumstances) 90, 214; -ism 25,74-6;-istic 91

638


option 286, 294, 325f., 471, 498; -simplification 218; -spilling 241, -akty 214, 263, 291, 302, 338 308; -statement 33; -taking 113, 278, orbit (lexical) 18, 265, 483, 495 292; -tone 72, 112, 167, 192, 213, orchards, names of 447 274, 278, 348, 385, 408, 427, 432, orchestration 265, 268 435,453,489 order-ing 509; -liness 193 overtness 164,297 ordinance, style of 115f. organ-icity 26, 67, 109,201; -ization padding 287 80,95,158,162,204,206,226, paganism 386 413,509 pair (of words) 95, 157f., 160, 162, Orient-al (philology) 28, 36; -ation 3f., 164, 181f., 199, 210, 272, 307, 346, 83,225,252 416, 181; -ing off 105, 120, 243, origin (of speech, etc.) 27, 192, 257; -al 261,332,412,461,480 (vs. copy) 115; -ality 58, 65, 75, 83, palatal 290, 360; -ity 235f., 285, 387; 96, 157, 161, 164f., 195, 198,217f., -ization 140, 195 224, 230, 343, 426, 451, 469, 472-4, paleness (fig.) 140,162, 189 489, 501f.,-ally 316 paleo-botany, -ntology 32f., 87, 182; ornithonymy 23, 223, 234 -graphy 49; -Mediterranean (lan ortho-doxy (scientific) 73, 79, 134, guages) 16 141, 217, 225; -graphy (= spelling) palette (of expressions) 293 116,125,334 pan-chronic 378; -demie 419; -topic oscillation 89 378 osmosis (cultural, lexical) 11, 90, 132, panorama (of linguistics) 79, 130, 483 255 paper (used for records) 88 ostensible 278 paradigm 9, 1 If., 43, 77, 82, 91f., 95, "ousting" (= dislodgment) 246 99, 103, 106, 108, 113, 117-9, 122, out-come 93, 99, 107, 110, 210, 263, 158, 170, 172, 186, 192, 199,212f., 267f., 273, 302, 328, 332, 506; 231, 235f., 239-41, 243, 264, 269, -disstance 184; -growth 182, 232, 277, 283-6, 290, 326, 338, 347, 357, 254, 399, 458; -landkhness 111, 333; 371, 373, 376, 375-80, 384, 395, -line 80, 247; -number 184, 247, 415, 444, 459;-atic 109, 366, 369, 478, 506; -post 94; -rank 354, 373, 399, 504 447, 486; -set, from the 421, 485; paradox 73, 113, 160, 168, 179f., 182, -sider 157; -spoken 125 ; -weighing 267, 310, 331, 343, 345, 350, 361, 90 366,372,393,414,498 over-abundance 133;-estimation 199; parallel(ism) 73, 101-3, 119, 133, 143, -flow 431 ; -hearing 110, 306; -in 167, 170, 180, 199f., 207, 209, 214, dulgence 298; -interpretation 193; 231, 243, 256, 258f., 264, 273f., -lap 99, 103, 106, 118, 123, 155, 282, 285, 290, 299f., 304, 308f., 165, 201, 210, 310, 312, 316, 372f., 312, 316, 325, 329, 345, 347, 349, 503, 514; -lay(ing) 13, 30, 121, 192, 356-9, 363, 366, 370, 373, 380f., 212, 232, 311, 413, 423; -looking 384f., 395, 412, 417, 431f., 442, 116, 123, 189, 237, 259f., 331, 335, 446f., 457, 474, 479, 481, 493, 498, 463, 472, 483, 486, 502; -reaction 510 274, 338, 371 ; -riding 214, 266, 337; -running 121, 442; -seas 120f., 124; paraphrase 314

SELECTIVE INDEX OF KEY TERMS parasitic (sound) 229, 269 para-stratum 32; - synthesis 328, 403; -taxis 411 parchment (used to date records) 88 parent (language) 28, 125, 131, 156, 161, 246, 313, 326f., 343, 381, 474f., 505; -age 206, 352, 495 parenthetic (remark) 81 parl-ance 165, 167, 274, 302, 447, 504; -or game (fíg.) 157, 511 parochialism 24, 191, 274 parole 71,76 parsimoniousness 11, 181, 252, 260 partial-ity 18, 222, 427; - cause 301 particip-ant 234, 266, 270, 317, 421, 475, 511;-ation 80, 132,155,499; -atory (process) 300; -ial adverb 30810, 315; -le 179, 183, 303, 388, 399, 415,417,437, 510; (absolute) 133 particulars (of change) 6, 25, 35, 170, 251,260,276,372,436 partisanship 69, 268, 298 partner, language as a 137, 163f., 166, 238, 241, 300, 344;-ship 117, 303, 407, 458, 494, 497 passage 114-6, 120, 124, 335, 357f., 360, 457, 487, 499, 508 passive (partner, voice) 297, 458, 461 past (distant, recent) 91 ; -participle 95, 113, 188, 229, 231, 306f., 314, 350f.;-subjunctive 239; -tense 95, 99 pastime 511 patch (in a dialect area) 123 paternity (fíg.) 475 path (= line, road), straight 225, 308, 355,479,510 path-ology (grammatical, lexical) 33, 185, 223, 307;-etic 498 patois 123,482 patriotic (overtness) 64 pattern 73f., 87f., 93, 95, 99, 103, 109, 112, 122, 125, 139f., 160, 166, 171, 179, 185, 187f., 195, 211, 217, 219, 223, 225, 232, 234, 243, 286, 288, 290, 298, 311, 316, 328, 332, 338,

639

348, 350, 359, 361, 371f., 379, 385, 393, 412, 416, 425, 440, 443, 445, 452f., 463, 493, 508-10 paucity 404 peak = summit (fíg.) 161, 169, 238, 309, 324, 335, 349, 466, 493, 501 peculiarity (= idiosyncrasy) 88, 104, 110, 123,158,161, 163, 188, 195, 211,344,428,470 ped-agogy 110, 308; -antry 315, 335 pedestal (fíg.) 141 peers, languages as 134; - , scholars as 141,501 pejorative (form, meaning) 43, 407, 432 pend-ulum, swing of the 217; -ing 263; -ant 417 penetration 292f., 303, 460 pen-portraits (of linguists) 59-66, 217 penult 419f., 426, 441 perceptiveness 257 perfect-um 346 ; -ion 510 perform-ance 80, 109, 155, 168, 170, 506;-er 205 period 335, 429, 493;-piece 141;-ical (= journal) 81 ; -ization 70, 427 peripher-y (of lexicon, movement) 37, 41, 77, 245, 266, 369, 447, 469, 487, 489;-al (form, zone) 108, 131, 195, 230, 306, 329, 495 periphrasis 352 permissiveness 45 perpetuation 209, 253, 260, 277, 289 perplexity 143 persiflage 167 person 105, 123, 443, 453, 463, 509; -al (bias, pronoun, style, suffix) 69, 73, 105, 344; -ality 81, 88, 502; -ifìcation 165, 169, 172,181 perspective 50, 80, 102, 113, 137, 170, 184, 226, 229, 232, 236f., 247, 252, 329, 339, 342, 348, 376, 382, 385, 403, 445, 465, 474, 480, 498 persuasion 163, 206, 497, 503 pertinence 205, 259 pervasiveness 184, 244, 431

640


pessimism 501 petrifact 32, 355, 409, 412, 416 phase,-ic 56, 115, 131, 197,210,290, 298, 359, 432f., 447 phenomenon 131, 168, 206, 219, 256, 263, 265, 267, 269f., 301, 330, 345, 371, 419f., 426, 438, 459, 505 philolog-y, -ist 5, 24, 49, 53f., 66f., 73, 76, 87, 92f., 96, 107, 118, 139f., 179, 184, 191, 234, 245, 258, 335, 385, 390, 414, 420, 433, 459, 464, 474, 429, 498, 500, 503; -al lin guistics 68 philosoph-y 70, 73, 78, 163, 169, 222, 5 0 7 ; - of history 23 ;-ical (key) 297 phoneme 76, 92f., 143, 199, 211, 235 -7,262,268,350,392,483; phonemics (vs. phonetics) 31, 34, 41,96,265f.,286,385 phon-etic (change, conditions, corre spondence, erosion, law, note, pattern, trend, transcription) 9, 18, 94, 109, 164, 219, 232, 235f., 240, 272, 286, 297, 301, 304, 309, 316, 327, 340, 352, 428, 439, 473, 505; -etics 52, 70, 72, 74, 77f.; -etician 230; -ic 88, 102, 317, 378, 457, 483 phono-graph record 81 ; -logical (analysis, change, criterion, pressure, relation) 39,70f., 132f., 143,221,231-3, 243, 247, 256, 258, 263, 271, 314, 341 ; -logy 8f., 70, 74, 77-9, 90, 93, 109, 119, 124, 191, 211, 225, 229f., 232f., 276, 297f., 328, 346, 353, 357, 392, 437, 439, 451, 504, 509; -stylistics 310;-symbolism 170, 172, 195, 236, 316, 427, 432; -tactics 247, 294, 348, 374, 408f., 437 phrase 130, 212, 302, 314, 334, 354f.; -ology 133, 140, 315, 333, 422, 457, 459;-ing 97, 141,494,500 physi-ology 78, 165,444,481-5,493f.; -cal (defects) 400; -ognomy (fig.) 414 physique 434, 486 phytonym 101, 170, 180, 255, 355,

385, 429, 432, 443, 487 pictorial (phrasing) 293, 297, 303, 437, 474,505,510 "piecing-together" (fig.) 187, 334, 435, 505; piece of evidence 474 pilgrimage (as a factor in language growth) 335 pillar (fig.) 71, 103, 231, 268, 303, 423f.,436,443,506 pinnacle 191 "pinpointing" 164, 168 pioneering) 11, 70, 76, 93, 96, 104, 116, 121f., 130, 157, 165, 168, 179f., 208, 218, 223, 225, 232, 265, 269, 299, 306, 308, 323, 330, 340, 353, 374, 377, 389, 391, 459, 464, 468, 473f., 477, 494, 508, 511 piquancy (fig.) 109, 161, 211, 310 pitch 130 pivotal (role) 230, 500, 508 'place' 243, 463; place-name (= toponym) 428f. plane, plat-eau (fig.) 125, 197, 272, 337, 346, 417; -form (fig.) 218, 498 plasticity (of expression) 298 plausibility 93, 141, 181, 234, 247, 260f., 267, 285f., 290, 312, 325, 350, 356, 385, 436, 444, 472f., 425, 478 play, coming into 294; -fulness 56, 130, 425;-ing down 116, 297 pleasing (= flattering) form, pleasure 17, 423,445,447 plur-al 99, 156-8, 160, 163, 169, 186, 238, 293, 329, 453,457; -alization 348; -ality 160, 225, 239, 300; -inuclear 300 poet-ics 52;-ic (discourse) 120, 194, 336, 453, 455f., 458, 463, 467, 470f.; -ry 133, 155, 167, 244, 252 point (of departure) 232; (of intersec tion) 494 polar (opposite) 292, 302, 359; -ity 71 ; -ization 239, 303f., 312, 316, 334, 351,413,475,500,502,506;

SELECTIVE INDEX OF KEY TERMS lexical - 7f., 19, 167, 200, 208, 2114, 230, 254, 348, 354, 363, 380, 384, 390, 458f., 462, 471f., 493 polemic(s) 73, 169, 182, 224, 359, 423, 435 politeness 497 polit-icization 76; -ics 81 poly-genesis 130; -graphs (as linguists) 63;-morphism 11,97, 377;-semy 56, 199, 202, 219, 222, 337, 355, 451, 456, 459, 471f.; -syllable 213, 342, 344,443 popular (notion) 169; -ity 184, 366; -ization 75; — (= vernacular) 351 portrayal 406 posit-ing 115, 187, 260f., 263, 265, 291, 327, 349, 437, 460, 483; -ion 'location, point of view' 101, 117, 130, 160,168, 239, 251, 263, 276, 342, 345, 355f., 382, 384, 441f., 447, 462,466,486,493, 503, 506f.;-paper 60 positive (vs. negative) 42, 336; -ism 23, 28,55,141,461,478,495 possessives 95 possibility 124, 143,170, 246, 254, 267, 345, 462,472,494 (cf. poten tiality) posteriority 452 post-consonantal 346; -dental 381 ; -humous (recognition) 58; -medieval 275, 312,468; -nominal (verb) 195; -participial 182, 188, 399; -ponement 382; -position 130; -script (as a source) 59; -structural ism 503; -tonic (vowel) 288, 343, 384, 419f., 424, 426,437,441; -verbal (abstract, noun) 11, 101, 143, 167, 179, 184f., 188f., 191-5, 197, 199f., 203f., 206-9, 212f., 235, 259, 271, 278,300, 315, 337, 373, 388, 399, 425,469; (agentive) 204 postulate 217, 381, 504, 507 poten- (fig.) 104, 117, 328; -tial 405, 456, 511; -tiality 214, 236, 239, 327, 357 (cf. possibility)

641

pottery, terms of 423 "pract-ical" (linguistics) 66; -ice 77, 299, 473, 495,498; -icing (linguist) 73; -itioner 10, 87, 138, 141, 217, 229, 455,460,498, 507 pragmatism 299 precarious (balance) 332, 355, 508 preced-ence 171, 220, 264, 308, 366, 494; -ent(s) 239, 358, 367, 467; -ing 131, 162, 280f., 285, 346, 350, 401,408,413 precellence 210 precinct (fig.) 171 precision 108, 508 pre-cursor 76, 137; -decessor 234, 251, 472,491 pre-determination 217, 239, 498 predicament, scholar's 141, 456 predict-ion, -ability (of change) 23, 37, 88,97,125,136,194,197,199, 219, 223, 235, 240, 247, 260f., 271, 277, 284, 295, 305, 330, 357, 401,436,474,505,507 pre-dilection, -ference (grammatical, lexical) 88, 95, 110, 113, 116f., 129, 234, 239f., 161,184,210f., 258, 267, 270, 273, 297, 305, 314, 329, 334-6, 358, 370, 388, 394, 400, 424, 428, 444, 477, 481, 483, 493, 509 predisposition 73, 304 predominance 219, 237f., 254, 294, 315,350,391,430,466,492 predorsal 236 preëminence 119, 133, 305, 335, 445 preëmption 102, 196, 245 preëxistence 106, 157, 183, 202, 310, 328, 348, 355f., 366, 372, 383, 426, 453 preface (as a source) 59 prefix 19, 130, 140, 198, 207f., 262, 267, 304, 325, 327ff., 331, 336, 345, 348, 353, 355, 400, 451f., 454, 459, 462,476, 479-81, 493f., 508f.; - change 329; -oid 329, 334, 338, 387

642


prehistory 23f., 89, 191, 195, 378, 419, 433, 444, 458 prejudice 76, 129 preliminar-ies 93, 511; -y (step) 258, 263, 306, 325, 348 prelinguistics 23, 62 preliterary 285, 493 prelude (fig.) 91, 235, 247, 306 premise 222, 290, 469 pre-noun 452 preoccupation 324 prephilology 23 preponderance 199, 207, 282, 384 preposition 8, 292f., 298, 302, 377, 417, 451f., 470f., 509; -al (phrase) 3024 prerequisite 183 prerogative 444 prescientific (term) 23 pres-ent (tense) 98, 104, 106, 109, 112, 119, 191f., 346;-participle 313; -subjunctive 237, 270, 272, 285f.; -entation 161, 443, 479; -ence 348 preservation 97, 100, 102, 109, 111, 113,116,131,143,198,255,280, 285, 288, 306, 343, 355, 363-5, 368, 385, 412, 419, 426, 428, 441, 448, 456f., 490 pressure (social, systemic) 77, 106, 108, 110,121,124,143,167,170,186, 189, 196, 210, 212, 236, 240, 261-3, 270, 273f., 289, 302, 329, 331, 336, 348, 351, 356, 359, 371, 375, 381, 415,440,453,459,505 prestige 112, 132, 141, 213, 221, 239, 371, 447, 460, 499 presumpt-ion 107, 122, 160, 247, 323, 493;-uousness 328 presupposition 73, 103, 111, 187, 255, 261, 291, 300, 323, 327, 343, 345, 359,468,479,484,498,507 preten-se 133; -tiousness 198 preterit(e) 11f., 93-8, 103-5, 107-13, 115f., 119, 124, 229, 233, 238, 240, 247,285,352,417 pretext 326

pretonic 233, 235, 273, 281, 290, 309, 332, 340f., 345, 368, 382-4, 388, 390 prevalence 99, 101, 109, 120f., 184, 195, 214, 238, 240, 257, 267f., 274, 276, 279, 281, 289f., 313, 343, 412, 441,490,497 prevelar 446 prevention 266, 274, 290 pre-verb 452 preview (of a trend) 79 prima facie (resemblance) 9; prim- 325; -ary (cause, source) 234, 255, 289, 294, 299, 369, 400; -e (classi fier, mover, reason) 114, 118f., 256, 367, 370, 382, 384, 421, 439, 469, 507f.; -itive (vs. derivative) 45, 143, 158, 162-4, 170f., 182, 184, 192, 194, 196, 200, 208, 211, 235, 255, 348, 355, 358, 387,400f., 416, 425, 430, 437, 447, 452f., 462, 471, 480, 484-7, 489-91, 493, 498;-itivism 336 primeval 494 principle, general 103, 138f., 162, 170, 219,266,268,494 print (as a clue to dating) 88; -er's proof (as a source) 60 priority 164, 172, 217, 298, 483 prism (fig.) 72, 116, 186, 212, 242, 331,344,463 pristine (language, meaning) 77, 98, 200, 428, 432, 434 privacy 486 privilege 125, 342, 444; -d (form) 191, 357, 503 probabil-ity 180, 260, 344, 370;-istic (statement) 37, 156, 261, 429, 509 probing 251, 420, 503, 510 problem 170, 179, 336f., 348, 435, 443, 451 ; -atic 116, 182, 195, 266, 353,360,419,421,453,475 procedure 107, 119, 425; process 87, 131,133,140,143,162,165,183, 185, 189, 192, 211, 213f., 233, 239, 242, 244, 246, 254, 256, 258, 260f, 263, 265, 270, 272-4, 276, 279, 286,


290, 298, 300, 310, 338, 345, 355f., 359, 378, 381f., 394, 419, 422, 428, 438, 446f., 457, 466, 474, 494, 505 proclivity 133, 185, 371, 434 product 107, 232, 290, 351, 385, 429, 458, 467; -ion 256, 440, 481, 494; -ive (model) 35, 38, 434, 438, 447, 493; -ivity, levels of 65, 143, 199, 207, 230, 234, 245, 257, 259-61, 289,300,408,421 profession (of faith) 73 profiling 114, 179, 219, 298, 302, 353, 359, 469, 507 profusion 125,491 progeny 254, 350, 393, 428, 494 prognosis 75 programmatic (question, tract, treatise) 29,55,60,141,170,297,503 progress (in linguistics) 26, 30, 74, 76, 101, 188, 191, 225, 251, 274, 284, 366, 426; -ion (lineal) 70, 76, 314; -ive 417 project 141, 431, 498;-ion 95, 115, 130, 164, 192f., 197, 218, 235, 292, 312,337,462,498 prolegomena (to history of linguistics) 50 proliferation 90, 95, 189, 323, 362, 390f.,425,482 prominence 120, 158f., 237, 256, 300, 305,326,443,475 promise 509 prong (fig.) 115, 119, 184, 344, 464, 494 pronoun 157, 345, 417, 509 pronomi nal pattern 232 pro-nouncement 75, 96, 169, 197, 254, 266, 307, 377f., 414, 429, 433, 460f.,473;-nunciation 112, 121, 124, 303, 306f., 325, 344, 358, 370, 386f., 390, 394, 445, 463 proof 140, 247, 460, 470 propag-anda 80; -ation 234, 270, 285 proparoxytone 343, 362, 365-7, 370f., 376,419,428,439,447 propelling 110

643

propensity 184, 278, 343, 471, 487 proper name 498 propiti-atory (term) 43, 509; -ousness 183 "proportion" 184, 201, 277, 279, 294, 370 propos-al 186, 192; -ition 222, 451 propulsion 253 pros-e 116, 120;-ody 14, 199,211, 306, 323, 342f., 502 prospect 77, 210, 300, 407, 445, 497; -ing 79 prosthetic 326, 338f. protagonist('s name, speech) 88, 207 protean 444, 502 protection 109, 115, 214, 331, 342, 360, 428 prototype 41, 70, 92f., 103-5, 112, 119, 181, 187, 192, 208f., 232, 246, 253, 258, 260, 267, 273, 301, 326, 332, 348, 353, 359, 367, 369, 377, 379, 411, 414, 456, 481, 504f. protraction 95, 109, 143, 155, 210, 257, 261, 264, 270, 298, 306, 313, 327, 343, 354, 384, 408, 419, 437, 447, 494, 509 Provençalism (= Occitanism) 137, 212,300,311,335 provenience 102, 132, 206, 289, 299, 303, 309, 327, 337, 422, 493 proverb 118 provinc- (of knowledge) 266, 510; -ial (milieu, speech) 101, 125, 130, 186, 188, 203, 208, 324, 343, 387, 444, 447, 454, 467; -ialism 264, 412 provisional 171, 267 provocation 160, 179, 224, 276, 327 proxim-ate (past) 98; -ity (= vicinity) 90,102,106,131,211,244,252, 274,367,378,389,473,491 prudence 508 prurience 509 pruning (fig.) 335, 461 pseudo-diphthong 385; -problem 299 psychiatry 494 psycho-analyste 316; -linguistics 74f.,

644


83; -logical (implication) 70, 72, 78; -logism 75, 161; -logizing 167; -logy (of speech) 33, 165, 168f., 266, 394, 497 pugilists' jargon 200 pun 475 "pure" linguistics 53, 66, 81, 110, 503; phonetics 94; shape 493 ; -itanism (in linguistics) 82; -ism 192, 225, 257, 274 purpose 72, 268, 356, 380; -fulness 314 purview 246, 460 puzzle 99, 161, 225, 231, 236, 246, 347,373,391 quadripartition 163 quaintness 257 qualif-ication 89, 131, 155, 233, 240, 259, 326, 375;-ier 155, 159, 162, 164, 222, 374; -ying (adjective) 313; -ty 406, 442 quanti-fication 38, 261 ; -tative (gram mar) 78, 222; -ty (of vowel, etc.) 181,287,314,359 quasi-compound 379; -homonymy 309; -participle 312; -prefix 452; -prepo sition 303 quest (for) 299; -ion 119, 125; -ing 298 raciness 167, 191, 255, 299, 335, 432, 479 racing (toward) 113 racism 69 radical 104, 107, 111, 121, 184,243, 269, 344, 437; -changing 240, 243; -initial 356; -ism 3f., 79, 81, 191, 335, 386; -final 93, 98, 109, 253, 282; -stressed (= rhizotonic) 37, 92, 182f., 206, 260, 277, 405; -ization 222 radi-o(active-carbon method) 89; -ation (= raying out) 132,335 raising 280f. rallying-point 199, 254 ramification 99, 155, 158, 186, 259, 270,345,359,467

rampant (development) 117 randomness 40, 131, 166, 187f., 195, 251, 260, 265, 305, 357, 382, 408, 436 range (of experience, meanings, varia tion) 87, 89, 100, 125, 184, 221, 251, 263, 336, 354, 384, 384, 406, 441,455 rank(ing) 121, 130, 134, 158, 194, 205, 207, 254, 265, 273, 285, 307, 316, 432f.,475,478 rapid (vs. gradual) 109 rapport 77 rapprochement 133, 217, 244, 300, 304, 446, 480 rarity 162, 172, 180, 192, 258, 336, 357,456,469,481 rash (fig.) = eruption 92, 125, 291 rashness (of conclusion) 259, 434 rate (of change) 89, 305; -ing 73; -io 212 rational-ism 23; -ization 74, 223 raw (data, facts) 24, 102, 133, 137, 393 reaction 79f, 90f., 139, 195, 223, 273, 312, 339, 362, 379, 382, 433, 444; cham- 93 reading 110, 116, 118, 124, 140, 191, 242,358,413 re-adjustment 104; re-admission 257 real (life, linguistics) 51, 159, 222, 480; -ism 95, 167, 197, 233, 263, 266, 268, 442, 462; -istic 44, 290; -ization 123 realm (= domain) of knowledge 224 re-apportionment 45; re-appraisal 78, 262 rear-guard (operation, in research) 18 reason (= explanation) 119 re-assessment 274 rebel 229; -Hon (intellectual = revolt) 57,72,79,125,439 rebound 350 rebuttal 169 recalcitrance 208, 357 recantation 6


recapitulation 119 receiv-ing (language) 261 ; -er (= bene ficiary) 334; recipient 279; recep tion (of a theory) 478 "recent (trends)", analysis of 50, 78; recency 94, 110, 123, 125,291,356, 426 recession (of a vogue) 82, 109, 206, 247, 304 reciprocity 43, 229, 346, 427 recogni-tion (of discovery) 58, 286, 343, 427, 472; -zability 271, 352 recoil 102, 115, 221, 235, 264, 286, 292, 332, 338, 351, 378, 381f., 388 recommendation 117, 314 recomposition 189, 200, 253, 356, 389, 411,481 reconcil-ability, -iation 103, 122, 219f., 251,256,285,312,411 reconsideration 479 reconstruction (of cases, events) 6, 8, 19,26,34, 53,76, 87f., 99, 180, 185-7,189,194,200,210,223, 234, 259f., 301, 327, 333, 353, 360, 376, 387, 392, 408f., 413, 435f., 440, 471f., 482; internal 88 record (., v.) 53, 91, 94, 97, 99, 101, 106f., 118, 122, 157, 159, 184, 203, 213, 222, 229f., 233, 257f., 286, 313, 347f., 351, 356f, 414, 425, 440, 454, 462, 464, 479, 482 recurrence 199, 205, 210, 266, 302, 305, 342, 443; recourse 303, 305, 459 redeeming (element, feature) 112, 197 re-definition 314 reduction 92, 94f., 108, 233, 237, 239, 242, 245, 300, 302, 327, 341, 347, 355, 393, 408, 468 redundancy 41, 267, 298 reduplication 5, 94, 96f. reexamination 384, 479 refashioning 210 reference, frame of 73, 258, 265, 454, 472; — straight vs. oblique 88; -work

645

139;-ent 159;-ential(axis) 112, 180, 191,331 refinement 133, 218, 225, 240, 272f., 301,348,383,385f. refle-x, -ction 95, 98, 102f., 111, 124, 129f., 133, 157, 160, 172, 211, 230, 252, 272, 274, 277, 280, 283f., 286, 288, 293,303, 306, 311, 333f., 336f., 341, 349f., 353, 376, 382, 384, 388f., 394, 412, 419, 440, 454, 468, 477, 481 ; -xivity 234, 345, 356 reform 268 refract-ion 172, 283, 419; -ory 452 refutation 160, 169,253 régimes, ideological 6 regional(ism) 38, 91, 99, 103, 1224, 158f., 166, 188, 219, 245, 260, 275, 278, 282, 288, 293, 330, 334-7, 400,415,425f.,465 regis-ter 202, 245, 345; -trarfs records, as a source) 60 regression (incl. false) 41, 96, 179, 185, 208, 213, 222, 266, 268, 347, 355, 428 regul-arity (of sound change or corre spondence) 9, 12, 34, 38, 76, 92f., 103, 112f., 119, 209, 219, 222, 224, 240-2, 262f., 266, 271, 283, 297, 301f., 304, 307,314,316,372, 376,384,416,428,445,456; -arism 219; -ation 283 rehabilitation (of pioneers) 28, 58, 212, 299, 420 re-inforcement 133, 180, 259, 283, 288, 349, 384, 390, 428, 437, 444 re-interpretation 193, 233, 272, 355, 384, 481 re-introduction 260 rejection 90, 95, 109, 169, 206, 243, 307, 309, 314, 336, 338, 379, 382, 393, 473 rejuvenation (of a dicipline) 3, 82, 120, 141,472 relat-ed 108; -ive (chronology) 11, 241, 302; -ivity 87, 225, 237, 420; -ion

646


(ship) 139f., 164, 166, 171, 182, 186, 188f., 194, 199, 201, 209, 240, 263, 272f., 298, 305, 308, 311, 350, 361, 364, 372f., 385, 424, 435, 441, 458, 480 relaxation 392 release 290 relegation 108,252 relevancy 18, 104, 155, 179, 195, 242, 254, 259, 279, 333; relief, bold 137, 211, 304, 306, 312, 356f., 359 reliability 156 relic(s) 221, 273, 339, 352, 412, 467; relic form 379 religi-on, -ous (sphere) 112,134, 180, 333 reluctance 124, 131, 252, 285, 392 remark 308, 334, 343, 349, 351, 459f. remed-y, -ial (action) 91, 192, 210, 298, 309, 312,326, 338, 347, 387, 447 remnant(s), remainder 94, 96, 104, 106f., 129, 239, 290, 303, 335, 375, 381,444 remodeling 111, 286, 298f., 365, 442 removed, one step 180, 341; remoteness 434, 475 rendition 303 renewal 358 re-orientation 76, 79, 141 re-organization 338 repetition 307, 372, 432 replacement 17, 56, 95, 120, 125, 131, 143, 184, 192, 201, 207, 253, 264, 268, 273, 290, 303, 309, 315, 326, 359, 364, 379, 381, 423, 436, 457 replemshment 225 replica 294, 456 report 117 repository 275, 333 represent-ation, -ativeness 97, 131, 181, 186-8, 204f., 210, 237f., 252, 263, 305,335,354,392,454,480 repudiation 192 repulsion 106 reputation 460 rescuing 467

research institutes 3 re-segmenting 245 resemblance 9, 102, 104, 106, 133, 188, 192f., 199, 213, 264, 299, 303, 307, 310, 313, 331, 335, 348, 364, 368, 395,466,471 reservoir (= store) (fig.) 225, 427 reshaping 117, 222, 293, 384, 461 reshuffling 264, 302, 304 residu-, -urn 24, 30, 32, 138, 141, 231, 240, 252, 263, 328, 412, 440; -al 104, 109f., 203, 205, 219, 255, 274, 302f., 331, 441, 456, 460, 470, 481 resistance (to attrition, sound change) 14, 39, 44, 94, 107, 109, 166, 206, 244, 263, 274, 291, 313, 333, 344, 352, 390, 399, 417, 438f., 451, 480 resource(s) 94, 112, 117, 408, 419, 471 response 214, 310, 344, 466, 474; -ibility 268, 471 rest 225; -lessness (linguistic) 106, 117, 393 restatement (as a technique) 214, 251 restitution 345, 347 restoration 44, 109f., 213, 238, 273, 302, 332, 369, 389 restr-aint 168, 299, 429, 482; -iction 205, 240, 260, 303, 309, 346, 417 restructuring 99, 104, 254, 260, 335, 339, 376, 380 result 169, 171f., 240, 280, 298, 324, 421,434,438,463,474 retardation 96, 107,428 retention 103, 241, 335, 363, 384, 431 retreat 108, 184, 199, 225, 253, 315, 339,379,471 retrenchment 26, 30, 60, 73 retrieval 129, 327,433 retro-action 463; -spect 49, 51, 56, 79, 186, 231, 300, 306, 317, 327f., 385, 473 return (to earlier states or positions) 58,105,133,243,290 reverberation 71, 94,98, 133, 186, 242, 260, 269 revers- (condition, direction), -al 43,


80, 165f., 182, 184, 205, 212, 224, 230, 252, 254, 263, 289, 330, 370, 481 ; -ible 203 ; -ive (prefix) 18 revision 169, 179, 308, 333, 431 revival (of cults, theories; obsolete forms) 36, 58, 82, 257, 328 revolution (vs. continuity) 57, 217, 224, 226; -ary (change, idea) 54, 199 reward, speaker's or scholar's 157,435, 479 rewrite 339 rhetoric 307, 447 rhizotonic (= radical-stressed) 180, 207, 235, 259, 326 rhotacism 38, 262 rhyme 36, 106, 111f., 114, 120f., 167, 211,242,257,358,414 rhythm 294, 367 rig-or (excessive) 72, 94, 434; -idity 122, 134, 206, 225, 266, 289, 315, 401 ringing (= sound) 439 rings-of-a-tree simile 33 ripple (fig.) 90f., 183 rise (= genesis) 17, 95, 113, 117, 164, 179, 192, 195, 200, 206, 208, 210, 212, 231, 286, 293, 326, 405, 419, 439, 446, 469; -ing (diphthong) 140, 232, 241, 249, 303, 339, 345, 382, 387, 390 risk 209, 246, 295, 362, 437 ritualistic) 446 rival(ry)90, 101, 110f., 114, 116f., 119f., 122f., 125, 200, 204, 209-11, 233, 235, 237-9, 257, 259, 264, 267, 275f., 282f., 286, 298, 312, 315f., 330, 332, 335, 337f., 349f., 355-8, 362, 366, 371, 373, 384, 389f., 393, 405, 407, 415, 428, 451, 454, 470, 477 role, playing a 103, 109, 143, 179, 182, 199, 346, 382, 389, 413, 420, 426, 440, 466 Romance philology 49 "romantic" (phase, pose) 179, 469, 472 root (fig.) 106; to strike - 119, 283;

647

— consonant 357; —final402; -morpheme 156, 348, 357f., 365, 376, 378, 380, 384, 413, 436, 440, 445 roster 455, 461 rough(-hewn) 214, 466; -ness (of transi tion) 201, 285 rout-e 258, 368;-ine 161 rubric 454 rudiment 187 rule 143, 194, 283, 288-90, 327, 445; -of thumb 89 rural (character, life) 123, 245, 290, 350, 386, 432, 482; rusticity 32, 110, 115,118,120-2,131,185,268, 272-4, 290, 386, 388, 392, 474, 482 S impurum 326 "sacrifice" (fig.) 398; (rit.) 446 safety 156 sailors'jargon 165 salience 106, 110,329,366 saltatory (change) 38, 40, 219, 262, 266, 372 salvaging 197, 437, 473 sample 120, 129, 266, 310, 349, 424 sanction 124,210,344 sanitary (conditions) 264 satellite (grammatical, lexical) 92, 124, 161, 164, 192, 199, 203, 246, 281, 292, 331, 334, 348f., 352f., 373, 376, 380, 383, 404f., 412, 440, 453, 483 satisfaction 469, 478, 482 saturation 223 scaffold, consonantal 209; -ing 212 scale 66, 129, 167, 203, 221, 231, 240, 247, 251, 268, 284, 288, 298, 308, 333,388,437,443,453 scantness (of records) 108 scattering 92, 108, 131, 185, 192, 197, 203, 208, 270, 282, 310f., 329, 349, 356,385,413,436 scene (literary, local) 117, 121, 290 schem-a -e (= model) 71, 89, 163,166, 172, 184, 186, 223, 235, 238, 240,

648


279, 330, 368f., 430; -tization 73, 89 schol-arly (discourse) 141;-arship, linguistic or literary 88, 433; -asticism 307; school (of thought) 3, 5, l l , 51, 57,70f.,74,222,297 Schwärmerei 56 scien-ce, -tifie 72f., 78, 222, 224, 260, 274, 309, (jargon), 419, 473; -tism 340 scope 91, 113, 123, 134, 191, 203, 212, 246, 251, 254, 260f., 264, 266, 270, 298, 316, 405, 431, 453, 469, 472 scrap (of evidence) 108,435 scribal (idiosyncrasy, lapsus) 41, 100, 122, 213, 229, 241, 267, 311, 331, 358 script (as a clue to dating) 88, 299, 306 scrupulosity 168 scrutiny 91, 312, 427 seasons, names of 196 seclusion 185 secondary (adverb, association, cause, channel, complication, form, litera ture, meaning, process, source, stimulus) 10, 98, 132, 137, 157, 183, 192, 195f., 233f., 254f., 276, 278f., 290, 294, 299, 301, 307, 334, 356, 364, 369, 380, 381-3, 407, 417, 429, 431, 438, 473f. sect-ion (of territory) 102, 465; -or (of the lexicon) 18, 133, 170, 264, 286, 347, 353, 357, 386, 394, 442, 471 segment (of time, trajectory) 77, 87, 91,99,101,129,139,142,184, 188, 191, 198f., 209, 212, 235, 244, 252-4, 259f., 273, 286, 290, 298, 300, 302, 313, 315, 325f., 328, 331, 334, 340, 342f., 345, 348, 355-7, 359, 364f., 367, 370f., 377f., 387, 423-5, 431, 436, 442f., 447, 462, 480; -al (phoneme) 99; -ation (of knowledge) 4, 156, 223, 412, 416, 445 sediment (= deposit) 32 "seeping" (= percolation) 313 segregation 346

selec-tion 72, 77, 105, 125, 130, 239, 316, 354, 382, 408; -tivity 6, 96, 275, 344, 361 self-confidence 204; -confinement 137; -consciousness 291 ; -contradiction 141, 242, 373; -correcting 142; -critidsm 168;-discipline 172; -enrichment 133; -imposed 186, 341; -identification 72; -sufficiency 90 semantic(s) 27, 52, 70, 72, 77, 79, 95, 97f., 100-2, 104, 112,119,140, 157f., 161, 167, 182f., 187f., 191-3, 198f., 201, 203-7, 210-4, 222, 240, 245, 247, 253-5, 257f., 268, 271, 275, 277, 293f, 298, 302f., 309, 327, 338, 334, 337f., 347f., 351-5, 359, 364, 367, 370, 373, 378, 383f., 389, 394, 400, 406, 409, 412, 432f., 435, 439, 444, 447, 456, 460f., 469, 471f., 479;-(aspect) 138;(bridge) 42; (change) 219; (convergence) 17; (criteria) 142; (differentiation) 90f.; (estrangement) 143; (gap) 171, 330; (opposite) 19; (spectrum) 45; (wave theory) 28; semiotics 77 semi-consonant 270, 280, 285f., 291, 382, 390, 393f., -facetious 185; -latent 214; -learned (transmission) 15, 122, 124, 195, 252, 268, 274, 288, 307, 332, 339, 346, 376f., 385, 428, 439; -Romance 134; -technical (literature) 67 seminar style 81 Semitism 302 sens-ation 298; -ationalism 33, 76, 159; -itivity (to form) 104, 157, 353, 404; -itization 474; -ory 482; -uality 167 sentimental (bias) 24, 56; -ity 405 sentry-duty jargon 207 separ-ation, -ateness 107, 155, 240, 244, 256f., 261, 264, 266, 271f., 282, 303f., 311, 340, 351, 456, 481f. sequen-ce (of events, sounds) 23, 34f., 87, 106, 109, 125, 130f., 171, 210, 230, 270, 280, 289, 292, 298, 301f., 304, 316, 325, 328, 331, 349,


353, 356, 366, 369, 372, 382, 399f., 405, 417, 458f., 464, 466f.; -tial precedence 42 seri-es 183, 192, 196, 199-201, 205, 222, 238, 259, 261, 273, 277, 282, 285, 288, 308, 325, 348f., 373, 386, 400,413,424,428,434,462; -alization 8, 447 seriousness (of purpose) 230 sermo plebeius 208 service, pressing into 162, 384, 466 set (of circumstances, conditions, data, forms) 97, 239, 243, 259, 284, 312, 328,389,413,426,434,446; -(t)ing, social 185 settlement 44 severity 429 sex (in language) 11, 40, 155f., 165, 264f.; -uality 167, 172, 425; -ualization l56, 165 shade (of meaning) 210 shape 156, 199, 423, 434; -lessness 73; -r 473 sharing (responsibility) 106, 211, 213 sharp-ness (of break, contour, focus) 96, 111,123, 131, 298, 329;-edged 438; -ening 447 shell, grammatical 90 "shedding" 388 shelter 185 shift (= sound change, analogical) 117, 119, 125, 229, 239, 247, 258, 263, 267, 273, 288, 300, 306, 308, 336, 344, 348, 356, 363, 366f., 381, 442, 457 short-circuiting 194; -coming 188; -cut 335, 340; -distance 338; -ening 344; -hand 62; -lived 114, 254, 309 shred (fig.) (of information) 157 shrinkage 97, 166, 224, 417; shriveling 98 shunting off 112 sibilants 96, 231, 342, 395 sickness, names of 400 side (pos., neg.) 240; -alley or line (of

649

development, cognition) 77, 122, 200; -connection 133; -effect 185, 427; -glance 441; -issue 159, 172, 191, 201, 239, 259, 295, 300, 326, 329, 338, 353, 385, 445; -line 274, 347, 351, 392;-position 452;-re mark 327;-track 433, 463 sifting (the record) 125, 142, 168, 232, 238, 354, 473 sigmatic (tense formation) 94, 97-9, 104,115 sign 71, 76f., 108 (= symptom); -al 286, 334; -ificance 481; -ifier 71 silence (eloquent) 96, 299 silhouetting 198, 206, 213, 223, 283, 306, 438 similarity 96, 180, 191f., 208, 347, 380, 437, 460, 473 simpl-e vs. complex (causation) 9, 13, 300; — vs. compound (verb) 105, 354; -icity 95, 261, 332, 401; -ification 76, 95, 98, 117,286 simultaneity 122, 170, 202, 258, 268, 281,289 sing-le (factor) 300, 452; to ֊ out 117; -ness 170, 280 (of phoneme); -ular (number) 156-8; -ization 197 sister language 331, 399, 411f., 417, 428 situation (= state of affairs) 93, 95, 98, 102, 106, 117, 123, 130, 142f., 156, 165,172,261,286,301,312,315, 328, 331, 336, 338f., 344f., 362, 365, 367, 371, 373, 388f., 451, 469f., 474 size, linguistic, correlates of 11, 155f., 158-60, 164, 172 "skeletal" (documentation, outline) 12, 32,213,238,280,464 skepticism 266, 472 sketch (of a language) 29, 122; -iness 168 skewing 131 slackening (of discussion) 196 slang 185 slant(ing) 140, 199, 225, 251, 272,

650


298, 301 slicing up 142; slice (of knowledge) 443,451,464 slip(page) 122, 290, 522; slipping 'erring' 342, 356, 358f. slot 340 sloughing off 209, 298, 302, 387, 493 slowness 232, 247 smallness (of size) 158, 163 smooth-ing out or over 185, 423; -ness (of communications, pronunciation) 104, 143, 158, 254, 283, 298, 315, 331,391,450,493,527 snugness (of fit) 104, 480 social context 184; -dialect 219, 235, 386; -group 53; -institution 40, 265, 268; -matrix (of a language) 11, 52; -proximity 473; -science(s) 72, 89; -set 185; -stratification 91 society 73, 81, 394; -centered 73 socio-cultural background 24, 274, 276, 288; economy 392; -educational 110, 124, 209, 272, 400, 442; -linguistics 15,71,75,82,139,167,231,348, 392, 428; -logy 53, 71, 73f., 77; -spatial (habitat) 17; -spatiotemporal (edifice) 5 soldiers'jargon 201 solemnity 383 solid (block) 277, 328, 356, 412, 500; (= 'unbroken') 465; -ity (of know ledge) 437, 458 solution 100, 105, 114, 117, 124, 211, 239, 265,286,327, 334, 338, 393, 494,503,510,530 sophistication 75, 82, 97, 140, 161, 172, 183, 191, 220, 230, 267, 290, 299, 310, 417, 504 sound change (= shift) 34, 40, 76, 93, 103, 107, 132, 218f., 224, 229f., 232, 245-7, 261-3, 302, 361, 366, 372, 381, 391, 435, 443; -feature 230;-"law" 41, 119, 219-23, 265f., 268, 302; -sequence 283, 384; -sym bolism 156; - system 381 source (of change, information) 98,

133, 139, 164, 224, 245, 272, 282, 285, 299, 302, 206, 308f., 329, 334, 348, 357f., 361, 454, 464, 470, 473, 490, 495, 497, 523; -book 83; - word 423 space, lexical 245; -semantic 79; spatiotemporal (context, linguistics) 3, 11, 33,55,125,172, 191,211,343, 393, 413f., 452, 511 ; space vs. time 354 spadework 232, 328, 385, 511 span (of time) 219 "spark" (fig.) 93, 472, 489 (of humor), 509 sparseness, verbal 200, 203; sparing (= parsimonious) 257 speci-al(ized) (knowledge) 87; -ization 5, 7, 98, 337, 514; -alist 125, 509; -fication 88, 109, 489; -ficity 113, 168, 179, 232, 310, 443; -fies 235, 269, 271, 341; -men 120, 167, 186, 260, 295, 328, 376, 392, 409, 487, 490,508,510 spectrum (of varying opinions, variants) 96, 103, 137, 180, 200, 218, 276, 280, 300, 339, 406, 488, 509, 524 speculation 96, 167, 169, 235, 302 speech 73, 254, 260, 270, 295 (live). 442, 500; -area 221 ; -community 9, 11,15, 17, 90f., 95, 106, 112, 125, 230, 166, 210, 214, 231, 298, 312, 345, 351, 372, 387, 394, 474, 493, 529; -habit 108, 354; -therapy 62; speaker 143, 159, 203, 214, 251, 257, 263, 265, 272, 283f., 286, 290, 316, 344, 346, 348, 356, 473, 497, 504; spoken language 131 ; spokes man 110, 274, 455 speed 95, 113; -y (action) 91 spell (= attraction) 121, 457; -ing (as a clue to dating, etc.) 88, 110, 115, 121, 141, 180, 202, 205, 235, 250, 385, 463, 500 sphere (cultural, juridical, religious) 201,255,494,515 spiral-shaped (progress) 55


splicing 264 split 43, 87, 101, 141, 192, 243, 276, 329, 335, 337, 343, 350, 353f., 357, 382, 453, 477 sponsorship 350 spontaneity (of change) 9, 122, 125, 130, 141, 210, 219, 273, 341, 348, 372,419,437,497 sporadicity 9, 12, 38, 41, 92, 122, 186, 213,232,241-3,247,262,265, 267, 289, 313, 315, 324, 326, 330, 348, 350, 356, 363, 372, 380, 384, 445, 459, 469 spottiness 203, 246 Sprachbund 130 spread (of forms, ideas, techniques) 58,97, 113, 119, 133f., 164, 169, 179, 196, 235, 239, 243, 260, 277, 279, 286, 293, 302, 310, 312, 326, 329, 344, 364f., 371, 412, 416, 431 441, 444, 447, 451, 469, 472, 487, 504 "sprinkling" (fig.) 105, 120, 139, 194, 271 Sprossvokal 367 sprouting (fig.) 119, 259, 291, 304, 336, 338, 409 spuriousness (of entry) 142, 191,358f., 464 spurt 462 "squeezing (in)" (dialect area, etc.) 123, 426, 503 stabilization (of usage) 45, 223 stage (= phase) 109, 125, 160, 163, 217, 240, 247, 254, 293, 301, 303f., 310, 313, 327, 345, 377f., 385, 387, 391,393,460 stagnancy 33, 493 stamp (fig.) 161 stampede (fig.) 239 stance 469 standard (form, language) 93, 109, 121, 137, 164, 170, 233, 258, 284, 315, 344f., 380, 383, 387, 426, 475, 486, 493 stand-ing 112; -point 504; -still 166

651

starring point) 86, 108, 159, 163, 182, 192, 195-7, 208, 233, 246, 264f., 291, 294, 335, 348f., 357, 361, 364, 366, 368f., 374f., 483, 392, 400, 413f., 424, 428, 435, 441, 443, 453,457,463,466,469, 473, 474, 480, 529 stat-e 109;-(of affairs) 108, 156, 172, 181, 337, 354, 387, 391, 419, 504, 515 ; -able 317 ; ement 349 ; -ic (view) 217,498;-istics 72,75,78, 168, 17If., 258, 493; -ure (of a scholar figure) 97; -us (= rank) 109, 197-9, 265, 291, 313, 326, 339, 348, 359, 377, 385, 420, 432, 434, 441f., 458, 480, 485,494; statu(s) quo 241 ; in statu nascendi 263 steadiness 474, 494, 513f. stem (= radical) 198, 272, 286, 495; -variant 286; -vowel 231, 281, 283, 292, 366, 523, 528 step (fig.) 183, 236, 280, 293, 303f., 370, 393, 425, 470, 498; - by - 519 stereotype 167, 192, 197, 311, 314, 334,341,354,387,457,500 sterility 299 stif-fen 289;-le 290 stiltedness 198 stimul-us, -ation 102f., 143, 203, 208, 211, 235, 247, 268, 294, 312, 326, 334, 342, 347, 351, 353, 378, 415, 428, 492, 502f., 513,517,527 stirrings 514 stock 88, 131, 187, 258, 262, 276, 279, 283, 311, 405, 431, 442, 489, 493 stop (= occlusive) 211, 388, 507; -ping 529 straddling (fig.) 119, 138, 238, 344, 370,405,428,500,504 straight (loss) 94; (course, line [= descent] ) 202, 210, 226, 275, 288, 300, 429, 434, 453, 474, 493, 509f., 515f., 525;-forward 510 straggler (fig.) 454 strain (= line of descent) 45, 114, 208, 309,337,357

652


strand (fig.) 275, 475 strangeness 452 strategy (of research) 25, 71, 73, 138, 140, 224, 267, 308, 389, 407, 511 strat-um 189, 334; -ification 31,91, 343, 408; -igraphy 31, 87, 89, 292, 332, 490 stray (= random, scattered) (data) 106, 139, 163, 195, 274, 334, 376, 465 streak 420, 447 streamlining (fig.) 95 stress 140, 143, 162, 184, 233, 235, 240, 284, 311, 315, 340-3,367,369, 371, 379, 462, 471; -pattern 211, 363, 375, 443; -shift 5, 253, 272, 334, 344, 365, 384f., 442; -ed vowel 170,232,288f.,489 stretching (limits, theory) 119, 205, 244 strict (limits) 93, 426 string (of suffixes, etc.) 208, 266f., 301, 304,315,413,516 striving 286, 303 "strong" (past tense, past ptc.) 94, 96, 98,105, 111,113, 124, 417;-er (partner) 461 ; -hold (of a form) 101f., strength 106, 160, 332, 344, 352, 495, 499, 507; -ening 369 strophic pattern (as a clue to dating) 88 structur-e,-ing 91, 95, 117, 130, 139, 230, 233, 246f., 251, 265, 267f., 278, 280, 299f., 302, 315, 328, 349, 356, 419, 424, 428, 443, 453, 459, 514, 516, 530;-al 15, 92, 97, 125, 179, 195, 493; -alism 4, 31, 41, 68, 71, 74-9, 82, 140, 218, 225, 499, 503 struggle 505 stubborness (fig.) 246, 338, 475 stumbling block 209, 526 styl-e 73, 133, 139, 299, 316, 383, 385, 445, 498; -ist 116; -istic (differentia tion, polish) 69, 91, 120, 167, 184, 187, 202, 317, 400, 416, 455, 471; -istics 57, 72f., 75, 299, 309f., 314,

344, 348, 426f.; -ized (discourse) 42, 53, 447 sub-brand 427; -category 401; -class 204, 206, 311, 421; -conscience 480; -dialect 42, 270, 282, 284, 295, 349, 362, 371, 388; -division 114, 139, 280, 400;-family 156; -group(ing) 187, 252, 275, 308, 373, 424; -jacent 187f., 193, 404, 463; -ject (case) 411 ; -junctive 133, 285; -limation 212; -liminal 123, 317, 445;-merged 192, 221, 446; -mor phemic 196; -ordination 199, 317; -plot 492; -sequent 280, 492; -series 110; -sidiary (function) 271, 307; -soil (fig.) 111; -stance 494, 496; -standard 122, 167, 392; -stantiate 195; -stantivated (infinitive) 335, 348f.; -stitution 40, 90, 120, 131, 183f., 192,201,211,213,231, 236f., 350, 352, 354, 359, 364f., 368, 375f., 390, 415, 419, 423, 425, 439, 497, 504f., 515, 530; -stratum 16, 28, 31f., 37, 42,44, 172, 261, 299, 344, 363, 420, 427, 434, 444, 447, 469, 517; -sumption 116,195,247,261,263,336, 351,354,371,376; -system 382; -title 137, 160; -traction 268; -type 119;-variant 118 subtleness 379, 388, 437,445, 462, 529 success 207, 258, 267, 269, 445, 481; -ion, close 108, 122, 205, 301f., 316, 391, 404, 472, 499, 503, 516, 530; -or 182, 202, 208, 323, 336, 520 suddenness 199 suffix, derivational 15, 18, 105, 138-40, 142f., 156, 159, 161f., 164, 166, 170, 172, 201, 205, 231, 233f., 239, 244, 247, 253, 259f., 262, 265, 269, 272, 276, 286f., 300, 309, 311, 314f, 330f, 371, 381, 383, 389, 400, 413, 420, 422, 427, 432f., 444, 477, 487, 496, 510, 527; -chain 11, 15; -change 41, 291, 365;-oid


162,241,402,445,447,510 suggest-ion 90, 155f., 195, 273, 292, 323, 479, 490, 503, 509; -iveness 169, 241, 257, 268, 297, 315, 338, 416, 480, 452, 487f., 502, 505 suitability 429, 514 summit (fig.) 501 super-addition 526; -ficiality 158, 251, 381, 446, 500; -fluity 91; -imposition 92, 106, 130, 289, 442, 521; -iority 110, 375, 452, 456; -lative 417; -seding 246;-stition 425, 436, 509; -stock 298;-stratum 31, 76, 210, 299, 344; -venient (complication, factor) 104, 134, 255, 294, 348 supplanting 350, 354 suppletion 16,411, 529 support 203, 255-7, 261, 276, 334f., 443, 487; -er 461, 478f., 514, 524; -ing (argument, factor, role, vowel) 89,193,332,431 supposition 207, 290, 391, 516 supremacy 73, 119, 124, 311 surface (., v.) 171, 183, 245, 254, 500; -causation 338 surgery (fig.) 315 surmis-e, -al 261, 306 surprise 112, 115, 118, 162, 285, 435, 447, 482, 484, 492 surrender 270 surrounding (issues) 104, 332, 452, 462; -s222 survey 79f., 83, 96, 106, 122, 139, 141, 168, 170, 197, 206f., 231, 238, 260, 266, 270, 288, 373, 414, 453, 462,489,492,503,510,515 survival (of the fittest) 41, 98, 102, 198, 203, 209, 238, 260, 266, 291, 301, 334, 340, 343, 355, 399, 412f., 428, 439, 451, 453, 468f., 481f., 491, 504, 522, 529 susceptibility 235 suspension (of judgment) 479 suspicion 140, 303f., 308, 487, 492 sustained (attention) 189, 503 swarabakhti 423

653

sweep 185; -ing 196, 206, 231, 246, 442,446,457,514 "swelling" 307 "switch" 120, 132, 162, 164, 170, 193, 203f., 210, 239, 313, 316, 330, 338, 348f., 354, 380, 394, 413, 445, 490 syllab-ic (division, structure, syncope) 17, 94, 104, 208, 320, 339; -le 94, 101, 199, 211, 214, 234f., 284, 290, 300, 311, 326, 332, 334f., 338, 340f., 344, 361, 367, 371, 382, 384, 423, 433, 438f., 462, 468, 478; -le-final, -initial 234, 300 sym-biosis 12, 180, 192, 206, 343f., 417, 500; -bolic (logic) 75; -bolization 225, 306; -metry 44, 108, 142, 163, 208, 247, 273, 380; -pathetic (growth) 12; -pathy (intellectual) 169; -posium 81f., 499; -ptom (of decay, growth) 82, 112, 207, 263, 291,391,439 syn-chrony 9, 71, 88, 92, 140, 203, 282, 331, 348f., 501 ; -cope 93f., 101, 103f., 211, 213f., 234, 253f., 334f., 353, 357, 363, 367, 369, 371, 378-80, 394, 445, 509;-onymy 56,72,124,164, 194,258,291, 293f., 302, 331, 346, 349, 369, 391, 432, 486, 493, 508; -opsis 300; -tagm (= phrase) 77, 170, 172, 299, 311, 344, 366, 399, 461;-tax 75, 130, 133, 185, 193, 298, 315, 367, 380, 383, 417, 451, 498, 503, 521; -thesis 79, 100, 163, 220, 295, 376 system(atization) 28, 34, 105, 122, 288, 298, 300, 328, 352, 366, 381, 435; -atic 229, 414, 470; -ic 442; Systemzwang 415 taboo 56, 425, 436, 451, 509 tabulation 525 tacit (assumption, omission) 51, 134, 302,381,462 tactics (of research) 73, 224, 308, 461 tag (= label) 72, 156, 165, 498, 522

654


tail (segment) 442 taint 120,309 talent 96, 514 tampering 241 tangential (issue) 517 tangle (fig.) 202, 260, 338, 494 taste, canon of 5, 134, 137, 251, 314, 498 taxonomy 253 teach-er 314; -ing (of linguistics) 57, 500 team 108, 199, 513; -work 3, 139 techn-ical (advice, formulation, term, vocabulary) 78, 88, 102, 134, 192, 209, 254, 266, 302, 367, 377, 405, 433, 475, 432; -ique (of research) 7,72,87,90,141,187,218,221, 224,316,500,516 teology 15, 33, 141, 221, 251, 300 telescoping 181, 300, 470, 490 tell-tale feature 88, 206 temperament, speaker's or researcher's 4, 484f. tempo 241 ; -ral (axis, distribution, gap, inference, plateau, slant) 70, 87, 117, 129, 139, 316, 330, 376, 426, 457, 473; -rary 119, 198, 225, 243, 317, 335,381,417,463 temptation 121, 159, 316, 328, 336, 344, 460, 474, 480 tenability 203, 508 tenden-cy 117, 163, 184, 194, 207, 221, 264, 273, 289, 300, 303, 312, 328, 338, 344, 349, 351, 367, 369, 383, 390, 419, 464, 466, 472, 506, 530; -tiality 254, 278, 288, 328, 330, 332, 338, 344, 347, 350, 355, 364, 378, 423, 447, 452, 504, 516, 526 tense 103f., 106, 116, 119, 124, 239, 367 tension 125, 219 tentativeness 93f., 130, 157, 171, 185, 231,233,242,308,460 tenuousness 480 teratology (= study of monstrosity) 33 termin-al (point, suffix) 241, 414, 443;

-ation 300; -ology 74, 172, 221, 274 terr- {incognita) (fig.) 241 ; -ain 432, 443;-itory 87, 112, 117, 123, 219-21, (fig.) 309, 329, 344, 463, 469 tertiary 281, 473-5, 491 test 87, 142; -ing ground 93; -tube 184 testimony 90, 131, 140, 429; -ial 140 text(s) 5, 110, 112-5, 120-2, 125, 129, 284, 311, 345, 349, 351, 353f., 391, 458, 460, 508, 515, 524; -book 77, 81, 90; -transmission 197; -ual (critic ism) 70, 139, 301, 390, 415 "thaumaturgy" (fig.) 223 theatrical 81; theater, jargon of the 192 thematic (vowel) 240, 245, 369 theological G'argon) 132, 309 theor-ist (vs. historian) 65, 141, 498, 503; -y (grammatical, linguistic) 6, 10, 56f., 74, 76, 83, 87, 91, 101, 110, 130, 139f., 166, 170,217,266, 298, 327f., 420, 434, 451, 471, 498, 500,513f.,516;-izing 229 therapeutic(s) (glottal) 33, 91, 223, 307, 326, 328, 428 thesis 234 thickening (fig.) 393, 422 thinness (of disguise, spread, veneer) 92, 134,203,263,327 thoroughness 137, 142, 168, 198 thought 77, 298 thread (fig.) 211, 269f., 300, 304, 346f. threat 109, 119, 133,354,472 three-way (= tripartite) (pattern, split) 88,435 thrift (fig.) 117; thrive 119, 467 thrill (speaker's) 509 thrust, main (fig. ) 133, 317, 355, 488 thwarting 42, 283, 299f., 369, 472, 516 tide, high (fig.) 196, 207, 439, 444 tidiness 108, 261, 264, 270, 285, 384, 436,438,495,516 tie (= bond, link) 192, 244, 258, 284, 333,357,374,383,416 tight-ening (of resources) 94, 419; -knit 379; -ness (of organization,

SELECTIVE INDEX OF KEY TERMS structure) 6, 11, 304, 323, 342, 413, 509 time-axis 76, 87, 502; -depth 50; -honored 87, 141; -immemorial, from 315; -lapse 89; -perspective 361; vs. space 354 timidity, scholarly 141 title and subtitle 10, 137, 192, 431 togetherness 282 token 172 toler-ability 202; -ance 124, 212, 225, 256, 258, 274f., 288, 333, 454, 529 tone, lexical 43; - (of a book) 139; -setting 110, 158, 182, 222, 238, 272, 362, 407, 503; tonicity 421 էօօե, names of 130, 159f., 165, 180, 193-6, 198f., 204, 243, 258, 427, 463,513 topic 129, 161, 191; -al coverage 69; -in formation 88; toponymy 77, 165, 221, 230, 236, 253, 271, 280f., 289, 329, 374, 387, 427-30, 432, 434, 438, 444 torso (fig.) 305 'tortured' (rule) 246 totality 380, 386, 391, 513 touch, finishing 180 toys, names of 435; "toying" 487, 518 trace (= vestige), -able, -ing 106, 112, 122, 133, 166, 182, 203, 211, 234, 241, 243, 273, 301, 303, 309, 313, 315, 323, 329f., 344, 346, 349, 353f., 356, 358, 374, 380, 383, 385, 388, 393, 409, 423, 426f., 431f., 446, 453, 457, 460, 474, 480-2, 494, 499f., 510, 522, 525 trade (jargon) 302 tradition 95, 98, 103, 111, 113, 115, 124, 162, 192, 202, 224, 229, 242, 263, 267, 300, 313, 354, 392, 420, 463, 499, 502, 504f., 517; -alism (in research) 10f., 77, 81, 117, 122, 141, 367, 481, 495; -alist 284 traffic 109, 370 trail (left behind) 298; -blazer 76, 93, 179, 187, 199, 234, 268f., 292, 364,

655

473, 477; -ing 171, 225, 409,490, 503 training (ground) 7, 73, 520 trait, character 274, 406, 484,493 trajectory (fig.) 129, 201,211,222, 225, 253, 267, 292, 304, 330, 332, 347, 389, 453 trans-action 380; -cend 244, 276, 323, 344, 442, 502; -cription 30, 164, 311, 428; -fer 119,157, 206, 209, 211, 262f., 307, 336, 338, 370, 381, 384, 425, 474, 482; -form 106, 290, 493; -formation 262, 264, 283; TG (grammar) 78; -generational (con trast) 63; -ition 31, 157, 197, 201, 205, 264, 286, 309, 332, 341, 351, 373, 399;-itive 293, 356; -lation 70, 74, 81, 120, 308f., 315, 357, 524; -literation 82; -lucency 194; -mission 53, 100, 269, 272, 275-7, 283, 339, 347, 353, 355, 368, 370, 374, 376, 383, 385, 392, 400, 424, 427, 440, 460, 468,495; -mitter 283; -muta tion 94, 101, 103, 198, 219, 222, 230, 232, 288, 293, 297, 301, 333, 353, 423, 446f., 457, 459, 472, 481 ; -parency 9, 13, 90, 92, 104, 108, 131, 163, 189, 201, 206, 242, 253, 325, 349, 357, 379, 394, 411, 413, 416, 428, 442, 451, 471, 479, 484, 486,491,508,543,515,517; -planting 214; -portation 499; trespassing 306, 309 trap (fig.) 530 travel (abroad) 66 treat-ise 76, 313; -ment 99, 139, 246, 255, 266f., 270f., 374f., 381, 388, 428, 442, 446, 500 tree of knowledge (fig.) 323 "tremor" (= jolt) 94, 186 trend 90, 95, 108, 112, 125, 185,219, 254, 259, 264, 268, 276, 284, 288, 314, 327, 337, 339, 344, 362, 367, 382, 387, 402,442, 447, 457, 464, 473, 530 tri-ad(ic) 117, 192, 200, 231, 236, 243,

656


302, 312, 246, 477; -angular (re lationships) 184;-dimensionality 162; -nuclear (lexical family) 17; -phasic 235; -plets 91, 157; -ptych 306; -syllabicity 198, 211, 213, 257, 383, 413, 443, 457, 463; -via(lity) 97, 99, 133, 272, 278, 301, 350, 364, 372, 379, 430, 433, 451, 48,5, 502, 507, 527. trial (balloon) 166 tributary 357 Triebkraft 297 "triggering" 366, 516 trill 381 trimming 192 triptych 370, 376 triteness 293 triumph (fig.) 110, 184, 189, 446, 451, 468, 472, 502 "trouble(some)" 392, 498 troubadours' (language, vocabulary) 132,211 trumpcard (fig.) 160, 181 truncation 192, 388,427 truth absolute 225 turn-of-the-century (speech, fashion, research) 191, 218; -ing point 530 twig (fig.) 350 twin (developments, models) 188, 399; -formations 199, 416, 465 twist (of circumstances) 163, 357, 420; -ing 300 two-case (declension) 344, 417; -stage (effect) 108; -step (development) 232;-way (split) 88 typ-e 117, 184, 352, 510, 518;-icality 76, 105, 120, 134, 143, 155, 162, 274, 276, 278, 290, 315, 330, 345, 356, 406, 425, 439, 451, 453, 462, 483, 504, 513; -ification 351; -ography 68f., 82; -ology 8, 28, 72, 88, 123, 153, 180, 185, 202, 206, 247, 268,304,316 ultimate (cause, source, etc.) 239, 257, 263, 333, 335, 348, 469, 472, 474, 479, 494; ultra-brief 325; -fashion

able 141; -modern 504; -short 529f.; -weak 506 un-adulterated 471 ; -altered 373; -analyzable 329, 400; -attached 339; -available 83, 99, 180; -avoidable 180, 286; -broken 379, 468; -charac teristic 183, 259, 369; -common 120f., 471; -convincing 160, 253; -decided 354; -defined 133, 333; -deniable 112, 184, 224;-desirable 329, 362, 378, 471 ; -developed 328; -differentiated 91, 95, 453; equal 399; -equivocal 103, 139, 142, 258, 357, 479; -eventful 92, 125, 222; -expected 325, 358; -explained 331; -explored 353; -familiar 351 ; -identi fied 526; -known 133, 200, 510; -marked 363, 370, 441 ; -mentioned 332; -motivated 348; -natural 384; -paralleled 109, 122, 189, 205, 256, 353;-predictable 92, 140, 193; -protected 336, 338; -proven 119, 169; -qualified 343, 382; -realistic 358, 444; -related 338, 349, 352, 479, 495; -stressed 251, 258, 269, 277, 326, 328, 332, 339, 343, 377, 383, 419-22, 428, 468 ; -tutored 258, 283, 290, 384, 400, 444, 473; -tying 360; -usual 327, 331, 452; -wanted 362, 388; -welcome 387, 445; -willing 529 unanimity 118, 219, 343, 520 under-developed 166, 251, 328; -go 254, 498, 504; -ground (culture) 60; -lying 210, 246, 269, 305, 307, 30911, 313, 353, 356, 358, 390, 400, 404, 471, 475, 491, 505, 508, 511; -pinning 125, 261 ; -scoring 344; -standable 244, 246, 395, 438, 440, 444, 451, 527;-standing 513 undulatory (movement) 90 uni-causality 42, 268; -directionality 39, 137, 363; -fication 462, 517 ; -formity 97, 401, 464; -linearity 301 ; -queness (of change) 25, 37, 34f., 54, 141, 170, 205, 235, 251, 297, 304, 393,


442, 485, 498, 515f.; -t 157, 350, 501 ; (phoneme) 300; - lexical 326; -ty 131, 133f. univers-als (of change) 25, 27, 34f., 54, 141, 170, 205, 235, 251, 297, 304, 372, 462, 516; -ity (study) 3-6 up-grading 116; -heaval 90, 219, 223; -surge 433;-to-date 384 urban (speech) 32, 131, 392 usage 98, 109-11,120,122, 161,166, 188, 192, 196, 208, 210, 212f., 262, 264, 279, 282, 284, 305, 309, 315, 335, 371, 455, 469, 486, 498; use 257, 260, 452f., 486, 511 ushering in (fig.) 281 utterance 100,445 vacillation 109, 143, 273, 338, 350f. vacuum 192,304 vag-ary 73, 456; -ueness 222, 265 val-idity 103, 328; -ue 435, 502, 507 vanguard 474 vantage (point) (fig.) 77, 168, 187, 345, 426, 473, 507 vari-ant 8, 11, 17,89,91-7,100-2, 111, 114, 118-22, 124f., 164, 184, 186f., 211, 233, 238f., 256f., 263, 268, 270, 273, 275f., 285, 292, 303-5, 309, 311f., 323f., 327,332,334, 336f., 352-4, 357f., 365f., 368, 375, 377, 380, 384, 389-91, 405, 440, 447, 452, 463, 471, 475, 480, 509, 515, 519, 522; -ation 10, 87, 91, 99, 116, 123, 125, 164, 166, 184, 186, 194, 204, 238, 246, 271, 294, 312, 327, 348, 407, 414, 462, 467, 471f.; -ety 113, 131, 140,210,300,315, 324, 338f., 343-5, 348, 352, 441, 453, 464, 471, 475, 498; -egated 208; varying 261, 269, 287, 447, 458,465 veering away 140, 211, 460 vegetation (= phytonyms) 482 vehicle (fig.) 81, 294, 456, 499 velar 104, 234, 237, 284f., 363f., 366f., 388, 425, 441 ; - insert 504; -ity

657

199 veneer (fig,) 134, 192, 203, 346, 352 venture 498 verb 103, 105, 179, 183, 188, 196, 192, 194, 198, 202, 212, 234, 236, 243, 271, 277-9, 290, 335, 354, 359, 367, 372, 384, 405, 508; -al (ab stract) 199, 214, 275, 308-10, 316; -(artistry, arts) 87f., 192f., 196; -(diversity) 88; ֊ (family) 346; -(noun) 452; - (stem) 425, 432 verdict 168, 197, 306, 375, 378, 473, 477,450,501 vernacular (transmission) 36, 94, 98, 102, 104, 123f., 129, 132, 157, 170, 209, 213, 219, 230, 245, 255f., 260, 269, 272f., 288, 293, 307, 309, 313, 327, 332, 346, 350, 353, 385, 390, 393, 401, 412, 421, 427, 452, 457, 467, 480f., 521, 528 Verschleppung 256 vers-ion (of a text) 120, 486; -atility 501f. verve 224 vestig-e 104, 106, 113, 160, 168, 185, 189, 235, 238, 274, 282, 312, 330, 334, 337, 465, 490; -ial (evidence) 32, 97, 124, 303, 326, 347, 349, 365f.,387,432, 516 veterinary (terminology) 165 viability 17,239,320 vibration 211 vicinity 17, 113, 188, 235, 237, 241, 387,393,447,467,471 vicissitudes 104, 137, 158, 209, 217, 240, 256f., 265, 269, 302, 304, 316, 323, 328, 339, 376, 381, 435, 458, 517 victory) 110, 192, 199, 239, 263, 282; -ious(form) 104 vignette, lexical 217, 223, 255, 361 vigor 266, 273 vindication 121, 140, 470, 483, 510 vintage (fig.) 485 violent (conflict) 95 virtual 299

658


virtuoso (performance) 493 vista 140 vitality 112, 200, 259, 356, 415, 455, 494, 497 voc-abulary 132, 139; -alic (gamut) 245f., 367, 481; -(notation) 70; -alism 386, 438; -ational (jargon) 208 vogue (of fashion) 88, 94, 264, 370, 379, 474 voice (gramm.) 509 voicing 231, 234, 237, 239, 363, 383, 392f., 507; voicelessness 363, 392 volatility 432 volteface 76, 218, 372, 431, 499 volume (= size) 117 voluntaristic 222 vowel (quantity, resistance to erosion, structure) 43, 94, 170, 181, 211, 234, 240, 306, 315,325, 334, 342f., 362f., 367f., 370, 382, 462, 515; — (assimilation) 341; - (corre spondence) 342; - (dissimilation) 213, 332, 337, 364, 374, 510; back, front, intertonic 339f.; —, dangling 328, 332, 334, 338; - , word-initial, -medial 339, 349; - , neutral 367 vulnerability 119, 125, 187, 199, 267, 332, 338f., 341, 360, 387, 473, 499 waiver 465 Wanderwort 133 war-cry 384; -fare 509 warding off (fig.) 327 warning 142 "washing ashore" 436 waste 185 water-mark (as a clue to dating) 88; -shed (fig.) 502 wave (fig.) 90, 444; -theory 29, 70, 217,221 wavering 95, 108, 116, 122, 138, 143, 186, 191, 201, 210, 231, 242, 260, 270, 288f„ 293, 304, 306, 311, 327, 334, 338, 348, 350f., 354, 357, 366, 368, 383, 385, 387, 389, 410,

424, 446, 478, 506, 528 "weak" (phonetic change) 9, 95, 97, 113, 119, 122, 124, 506; (variety) 300; - (verb, preterite) 105, 111, 229; -(inference) 203; -ening 298, 302, 399; -er (partner) 458; -ly stressed 233, 326, 342, 370, 419, 442; -ness 90, 237, 263, 275, 439, 446, 472 wealth (of material) 142, 424 "wear and tear" (fig.) 451 web (fig.) 238, 360 wedg-ing in, driving a -e (fig.) 214, 257, 269, 282, 288, 292, 324, 445, 476, 513, 516; opening-224 "weeding out" (fig.) 327 week-days, names of 385 weigh-ing 261, 462; -t 199, 239, 434; -ting 162, 254 welcome 493 welding on (fig.) 207, 452, 458 well-attested 183; -informed 232; -rounded 183 welter (of variants) 124 whelps and cubs, names of 165 whim 347, 498; — of circumstances 384, 456, 508; -sey 471 ; -sicality 414 wholeness 404 wide-ranging 317; -spread 244, 247, 263, 278f., 289, 304, 324, 328, 344, 348, 350f., 377, 384f., 393, 419, 438, 463, 480; -ly representative 326 "wild leap" 94 wing (of an edifice) 298 winner (in rivalry) 184, 197 wisdom = advisability 179, 219, 222, 377, 422 wit 161 withdrawal 166, 257, 268, 478, 485 withering 247, 315 witness 236, 270, 274, 356, 474f. women lingmsts 61 ; woman's name 394 word 89, 222, 243; -biography 12, 217, 229, 346, 366, 386; -biend 417;

SELECTIVE INDEX OF KEY TERMS -boundary 404; -family 89, 336, 427, 475f.; -final (segment [= coda], vowel) 16, 108, 117, 143, 172, 186, 286, 298, 300, 315, 341, 423f., 431; -formation 168; -history 435; -initial 258, 300, 338, 341-4, 356f., 369, 382, 386f., 359, 459, 505f.; -medial 258, 300, 342, 389, 394, 424; -of-mouth 269, 325; -play (= pun) 474; -study 501, 411 ; Wörter und Sachen 156, 219, 514, 516; Wort forschung 220 world-view 40, 265; -liness 498 would-be (educated) 310 wreckage (fig.) 469

659

writ-er 143, 203, 316, 348; -ing 298; -ten (language) 133 yardstick 260, 400, 488 yield (n.) 168, 469, 514 young, names of the 165 zero 159,172, 189,315 zest(fulness) 133, 140, 266, 433, 488 zigzag (movement, process) 25, 94, 143, 166,225,510f. zone 108, 111, 113, 119, 169f.,220, 224, 229, 276, 298 (gray), 330, 425, 464, 470 zoonym(y) 10, 40, 101, 161, 264, 295, 356, 435

From Particular to General Linguistics: Selected Essays, 1965-1978 (Studies in Language Companion Series, V. 3)

Language in Context: Selected Essays

From Linguistic Areas to Areal Linguistics (Studies in Language Companion Series, Volume 90)

Language in Context: Selected Essays

Studies on German-Language Islands (Studies in Language Companion Series)

Essays on Nominal Determination: From morphology to discourse management (Studies in Language Companion Series)

An Introduction to Corpus Linguistics (Studies in Language and Linguistics)

Linguistic Diversity And Language Theories (Studies in Language Companion Series)

Linguistic Diversity And Language Theories (Studies in Language Companion Series)

Language Complexity: Typology, contact, change (Studies in Language Companion Series)

Course in General Linguistics

i> in Slavic (Studies in Language Companion Series)

Grammatical Relations in Change (Studies in Language Companion Series)

Course In General Linguistics

Linguistics (Oxford Introduction to Language Study Series)

General linguistics

A Functional Grammar of Gooniyandi (Studies in Language Companion Series)

Cross-Linguistic Semantics (Studies in Language Companion Series)

General Linguistics

Latin Linguistics and Linguistic Theory (Studies in Language Companion Series, 12)

Companion To Semiotics and Linguistics

Studies in Evidentiality: v. 54 (Typological Studies in Language)

The Linguistics Enterprise: From knowledge of language to knowledge in linguistics (Linguistik Aktuell Linguistics Today)

Self to Self: Selected Essays

From Special To General

Studies In French Applied Linguistics (Language Learning and Language Teaching)

Selected Essays

Self to Self: Selected Essays

Culture, Language and Personality: Selected Essays

Self to Self: Selected Essays

The Irish Language in Ireland: From Goídel to Globalisation (Routledge Studies in Linguistics)

From Particular to General Linguistics: Selected Essays, 1965-1978 (Studies in Language Companion Series, V. 3)

Language in Context: Selected Essays

From Linguistic Areas to Areal Linguistics (Studies in Language Companion Series, Volume 90)

Language in Context: Selected Essays

Studies on German-Language Islands (Studies in Language Companion Series)

Essays on Nominal Determination: From morphology to discourse management (Studies in Language Companion Series)

An Introduction to Corpus Linguistics (Studies in Language and Linguistics)

Linguistic Diversity And Language Theories (Studies in Language Companion Series)

Linguistic Diversity And Language Theories (Studies in Language Companion Series)

Language Complexity: Typology, contact, change (Studies in Language Companion Series)

Course in General Linguistics

i> in Slavic (Studies in Language Companion Series)

Grammatical Relations in Change (Studies in Language Companion Series)

Course In General Linguistics

Linguistics (Oxford Introduction to Language Study Series)

General linguistics

A Functional Grammar of Gooniyandi (Studies in Language Companion Series)

Cross-Linguistic Semantics (Studies in Language Companion Series)

General Linguistics

Latin Linguistics and Linguistic Theory (Studies in Language Companion Series, 12)

Companion To Semiotics and Linguistics

Studies in Evidentiality: v. 54 (Typological Studies in Language)

The Linguistics Enterprise: From knowledge of language to knowledge in linguistics (Linguistik Aktuell Linguistics Today)

Self to Self: Selected Essays

From Special To General

Studies In French Applied Linguistics (Language Learning and Language Teaching)

Selected Essays

Self to Self: Selected Essays

Culture, Language and Personality: Selected Essays

Self to Self: Selected Essays

The Irish Language in Ireland: From Goídel to Globalisation (Routledge Studies in Linguistics)

Recommend Documents