Langue and Parole in Synchronic and Diachronic Perspective

Langue and Parole in Synchronic and Diachronic Perspective Related Elsevier Books KOERNER & ASHER Concise History of...

Author: C. Beedham

49 downloads 2792 Views 37MB 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

Langue and Parole in Synchronic and Diachronic Perspective

Related Elsevier Books KOERNER & ASHER

Concise History of the Language Sciences KOMATSU & WOLF Saussure's First Course of Lectures on General Linguistics (1907) KOMATSU & WOLF

Saussure's Second Course of Lectures on General Linguistics (1908-1909) KOMATSU & HARRIS

Saussure's Third Course of Lectures on General Linguistics (1910-1911) Related Elsevier Journals Language & Communication Editors: Roy Harris and Talbot Taylor Language Sciences Editor: Nigel Love Lingua Editors: John Anderson, Johan Rooryck and Neil Smith Sample copies available on request.

Cover drawing, a sketch of Ferdinand de Saussure against the St Andrews skyline, by Oleg Dobrovol'skij

Langue and Parole in Synchronic and Diachronic Perspective Selected Proceedings of the XXXIst Annual Meeting of the Societas Linguistica Europaea, St Andrews 1998

Edited by Christopher Beedham

1999 PERGAMON An imprint of Elsevier Science Amsterdam - Lausanne - New York - Oxford - Shannon - Singapore - Tokyo

ELSEVIER SCIENCE Ltd The Boulevard, Langford Lane Kidlington, Oxford 0X5 1GB, UK © 1999 Elsevier Science Ltd. All rights reserved. This work is protected under copyright by Elsevier Science, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may he made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier Science Rights & Permissions Department, PO Box 800, Oxford 0X5 1DX, UK; phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-mail: [email protected]. You may also contact Rights & Permissions directlythrough Elsevier's home page (http://www.elsevier.nl), selecting first 'Customer Support', then 'General Information', then 'Permissions Query Form'. In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (978) 7508400, fax: (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W 1P 0LP, UK; phone: (+44) 171 631 5555; fax: (+44) 171 631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier Science Rights & Permissions Department, at the mail, fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made.

First edition 1999 Library of Congress Cataloging in Publication Data A catalog record from the Library of Congress has been applied for. British Library Cataloguing in Publication Data A catalogue record from the British Library has been applied for.

ISBN: 0 08 043581 5 ©The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in The Netherlands.

To the memory of Sandor Hervey (1942-97), Reader in Linguistics at the University of St Andrews, and long-standing member of the SLE

From left to right: Dr Isabel Forbes, Prof. Dieter Kastovsky, Prof. Roland Harweg, Prof. Matti Rissanen, Prof. Konrad Koerner, Prof. R.H.Robins, Dr Christopher Beedham

CONTENTS Preface

vii

Introduction

1

I History of Linguistics 1 Three Saussures - One 'Structuralist' Avant la Lettre E.F.K. Koerner

19

2 Gegen eine herkommliche Interpretation von Saussures Langage, Langue und Parole - ein Merkzettel Ralph A. Hartmann

35

3 Eubulides as a 20th Century Semanticist Pieter A.M. Seuren

47

4 Key Dates in Twentieth Century Linguistics R.H. Robins

63

II Langue versus Parole 5 System Description or Systematic Prescription? Bozena Bednarikova

75

6 Completeness and Symmetricity of Paradigms Mati Hint

81

7 Langue und Parole - eine Neubestimmung unter den Gesichtspunkten von Referenz und Geltung Roland Harweg

89

8 Parole as an Individual Realisation of Langue Tadao Shimomiya

99

III Deixis 9 Aspect, Voice and Deixis in Russian Youri A. Poupynin

105

10 The Integration of the German Modals into the Paradigm of Verbal Moods Gabriele Diewald

119

vii

viii

Langue and Parole in Synchrony and Diachrony

IV Morphology 11 Noun Classifiers and Gender Classes - Another Look K. Connors

133

12 Was ist Stamm? Jozef Darski

149

V Semantics 13 One Size Does Not Fit All: A Semantic Analysis of 'small/large' vs. 'little/big' Yishai Tobin

169

14 Colour Categorization and Naming in French and Hungarian Isabel Forbes and Gabor Kiss

181

15 Associative and Semantic Word Fields in Bilinguals: The Case of Russian-Hebrew Bilingualism Larissa Naiditch

189

VI Phraseology 16 On the Cross-Linguistic Equivalence of Idioms Dmitrij Dobrovol 'skij

203

17 Towards a Description of English and French Phraseology Christopher Gledhill

221

VII Discourse Analysis 18 Register-Motivated Variation of Tense and Mood in German Final Clauses Introduced by damit Sven-Gunnar Andersson

241

19 Modals, Comparatives, and Negation Igor Boguslavsky

253

20 Langue and Parole in Speech Act Theories: Some Considerations and a Proposal Elisabetta Fava

263

21 Dinosaurs, Metaphors and Political Argument Andreas Musolff

283

22 Metatext-Sequenzen in deutsch-polnischen Gesprachen: Eine Fallstudie anhand konsekutiv gedolmetschter Texte Magdalena Jurewicz

295

Contents

ix

VIII Sociolinguistics 23 National Profiles of Language Perception: Instrumental vs Cultural-Value Conceptions Natalia Guermanova

307

24 Wieviel Verliert und Profitiert ein Wort auf seinem Weg von der Gebenden in die Aufnehmende Sprache Senta Setinc

321

IX Historical Linguistics 25 Early Collections of Private Documents: The Missing Link in the Diachronic Corpora? Radmila B. Sevic

337

26 Social Networks and Language Change in Middle English: The Challenge of Diachrony Alexander T. Bergs

349

27 On the History of Non-Finite Clauses in English and Other Languages Isabella Buniyatova

363

28 EinfluB der Aspekte auf die Umgestaltung der Futurparadigmen im Russischen Wladimir D. Klimonow

375

29 Possessivaussagen im Deutschen und die Auxiliarisierung von haben Michail L. Kotin

393

30 Dichotomische Zugriffe im Bereich der Aktionalitat des Westgermanischen (im Vergleich zum Slawischen) Michail L. Kotin

401

31 The Mystery of Consonantal Nominal Stem-Building Markers in Ancient Germanic Olga Ossipova

411

32 "Langue" and "Parole" in Proto-Indo-European Reconstructions Thomas V. Gamkrelidze

419

33 Linguistic Dynamics in a German Autobiography Fernande Krier

423

34 Accentual Reconstruction of the Proto-System of Mainland Japanese Dialects Akiko Matsumori

431

x


X Generative Grammar 35 Temporal Reasoning in Iterative and Habitual Contexts Florian Panitz

451

36 Doubling Clitics and Information Structure in Modern Bulgarian Ivanka P. Schick

469

37 The Categorial Status and Case Properties of Quantified Phrases in Slavic Nedzad Leko

489

Addresses of authors

509

Name Index

513

Subject Index

515

PREFACE For five warm and sunny days at the end of August 1998, 110 linguists from 25 countries in East and Western Europe and further afield met in St Andrews, Scotland, to discuss matters of language and linguistics. The occasion was the XXXIst Annual Meeting of the Societas Linguistica Europaea, and we offer here a selection of the papers presented. Four were plenary papers, viz. the papers by Koerner, Robins, Tobin, and Harweg, the rest were section papers (except one, which was given in a workshop). Local organisers of the conference were myself and Isabel Forbes. The organisation of any conference is a team effort, and many people contributed to the staging of this one. I would like to thank in particular, for their advice and assistance: Dieter Kastovsky, Wolfgang Dressier, and Christine Klein, at the SLE headquarters in Vienna; Marjory Brechin, at St Andrews' Conference and Group Services; Chris Gledhill, Ralph Hartmann, Natalie Braber, David Gascoigne, and Alison Aiton; Julian Crowe, for computing help; Wendy Anderson, for proof-reading; and Chris Pringle, at Elsevier, who attended the conference. I am grateful also for financial assistance to: The British Academy, The Russell Trust, The British Council, and SAILLS (St Andrews Institute of Language and Linguistic Studies). Enquiries about the Societas Linguistica Europaea may be made to: Prof. Dieter Kastovsky, Institut fur Sprachwissenschaft der Universitat Wien, Berggasse 11, A-1090 Wien, Austria; e-mail: [email protected]. CB, St Andrews, 17 June 1999

Xi

This page intentionally left blank

INTRODUCTION The pages of this volume contain 37 papers which were presented at the XXXIst Annual Meeting of the Societas Linguistica Europaea, at the University of St Andrews, Scotland, on 26-30 August 1998. The general theme of the conference was 'Langue and parole in synchronic and diachronic perspective', though it was open to contributors to offer a paper either under that heading or else on any topic of general linguistics. The theme was intended to be both broad, in order not to exclude potential contributors, and specific, in order to give contributors a peg on which to hang their ideas. The theme is, in fact, ambiguous, and although no attempt was made to unravel the ambiguity at the time of the call for papers, it is perhaps time to do so now. The reference to langue and parole in the theme could be an invitation either to reflect on the theoretical notions of langue and parole, or to take them as given theoretically and to use them in an actual analysis. Some contributors interpreted the theme in one way, others in the other way, and others still gave papers outwith the general theme. The selection of papers offered here reflects that broad range. The langue-parole dichotomy is as difficult and controversial today as it ever was. The distinction was made in Saussure's Cows de linguistique generale in order to say that in structuralism it is langue that constitutes the proper Untersuchungsobjekt of linguistics, not parole. And we are all supposed to be structuralists now. But it is pragmatics, discourse analysis, text linguistics, and the like, which are making the running in linguistics today, and these are all areas of parole! Yishai Tobin certainly hit the mark with the original title for his book which eventually appeared as Semiotics and Linguistics, the original title which the publishers did not want to use, and which was: The Saussurean Revolution that Didn't Take Place, as Prof. Tobin related at the start of his conference paper. There is no point in this Introduction in my attempting to summarise all 37 papers which appear here, not least because each paper begins with a short abstract anyway. What I will do instead is to take those papers which lie closest to my own areas of interest and expertise, and highlight what seem to me to be particularly interesting aspects of them. Konrad Koerner, in his paper Three Saussures - One "Structuralist" Avant la Lettre (19-34), locates the first ever published use of the famous phrase that a language is a 'systeme ou tout se tient'. It was first used in print by Antoine Meillet with reference to phonology, in a paper published in 1893 in a French sociology journal: Les divers elements phonetiques de chaque idiome forment un systeme ou tout se tient. Or 1'enfant, en apprenant a parler, s'assimile non une articulation isolee, mais l'ensemble du systeme. (Meillet 1893:318-19; quoted by Koerner, this volume p.26) Koerner points out that, although the phrase is usually - particularly in textbooks on linguistics - attributed to Saussure in the Cours, it does not in fact occur there at all. However, Koerner states that Meillet almost certainly drew the phrase from Saussure, probably from the courses which Saussure taught in Paris during the mid 1880s and which Meillet attended. Koerner also disputes the commonly held view that there were two Saussures, the historical linguist of the Memoire and the structuralist of the Cours. Rather, Saussure's work forms a coherent whole, in which the Memoire, a study of the phonological system of the Proto-Indo-European vowels, is a synchronic and hence structuralist piece. Koerner concludes his paper with an objection to

1

2


the fact that Roy Harris, in his translation of the Cours (Saussure 1983), translates the French term systeme with English 'structure', even though French structure does not occur anywhere in the Cours as a technical term or indeed in any of Saussure's writings; it is ironic, says Koerner, that 'structuralism' is the name by which Saussure's work has come to be known. Ralph Hartmann, in his paper Gegen eine herkommliche Interpretation von Saussures Langage, Langue und Parole - ein Merkzettel (35-45), says that it is widely accepted now by historiographers of linguistics that the famous final sentence of the Cours: la linguistique a pour unique et veritable objet la langue envisagee en elle-meme et pour elle-meme (Saussure 1972:317; quoted in German by Hartmann, this volume pp.35-6), was added by the editors of the Cours, and did not come from Saussure himself (cf. Saussure 1972:476-7). Hartmann goes on to argue that an examination of the published sources for the Cours, Viz. Godel 1957 and Saussure 1989-90, reveals that the notes as taken by his students indicate that Saussure did not distinguish between langue and langage as clearly as is done in the Cours, and that the editors of the Cours have sometimes used the term langage where the sources use langue. Hartmann questions the traditional conception of langue as being social, parole in contrast as being individual. The student notes, he claims, show that Saussure was uncertain on this point, as revealed in the following note by Albert Riedlinger: Dans la langue, il y a done toujours un double cote [sic] qui se correspond: elle est . ,. .. .. . Si on considere donc la sphere ou la langue vit, il y aura toujours la langue individuelle et la langue sociale. (Saussure 1989-90: 28, Riedlinger; quoted by Hartmann, this volume p.37) Against this background it is possible to see langue as encompassing not only a social aspect but also individual use and language as action, i.e. speech acts. This would lead to a great deal more commonality than has hitherto been suspected between Saussure and the later Wittgenstein (Wittgenstein's 'game theory' in the Philosophische Untersuchungen), as well as between Saussure and Searle - cf. Searle's assertion that: It still might seem that my approach is simply, in Saussurian terms, a study of "parole" rather than "langue". I am arguing, however, that an adequate study of speech acts is a study of langue. (Searle 1969:17; quoted in German translation by Hartmann, this volume pp.43-4) Youri Poupynin in his paper Aspect, Voice and Deixis in Russian (105-18) suggests that Russian aspect is a deictic category. He defines a deictic category as a category, at least one grammeme of which can denote the elements of the deictic centre, viz. the 'I - Here Now'. It is well recognised that tense is deictic. But there is a conception of the verbal system in Russian which sees it as having 5 tenses (not 3), inextricably linked to aspect: (i) the imperfective past; (ii) the perfective past; (iii) the imperfective present; (iv) the imperfective future; (v) the perfective future (Bondarko 1971). Since the Russian present tense exists in the imperfective variant only, then aspect, with its imperfective grammeme, participates in denoting the deictic centre, and consequently is deictic (Poupynin, this volume p. 106). Poupynin goes on to demonstrate the relevance of deixis to certain Aktionsarten in Russian. For example, it is well-known that imperfective verbs with the prefix pri- (e.g. prijti 'to come, arrive') cannot express processual meaning, which is one of the central meanings of

Introduction

3

imperfective aspect. The reason is a deictic one, viz. that the point of departure of spatial deixis (the 'Here') is strictly attached to the end of the action, whereas the interpretation of an action as a process requires its perception before it finishes, from other observation posts (Poupynin, this volume p. 109). The admission of an observer or perceiver can also help explain the semantics of ingressive verbs (e.g. zasumet' 'to begin to rustle' or 'to be rustling and to become available for perception by someone'), completive verbs, and the phase verbs nacinat' 'to begin' and koncat' 'to finish'. The reflexive phase verbs of the type nacinat'sja 'to begin', koncatsja 'to finish' can probably only mark real initial or final phases of an action where the subject is expressed by a deverbal noun (or a noun with the semantics of an action). If the subject is expressed by a concrete noun it is the perception by an observer of an initial or final phase that is meant by the phase verbs. To say Les nacinalsja/koncalsja 'the wood began/ended' is to say that the observation of the wood began or ended (Poupynin, this volume pp.111-12). Poupynin then applies the notion of observer/perceiver to the verbs of motion, which he calls 'verbs of displacement in space'. He distinguishes between process meaning and what he calls 'relative' meaning, i.e. the position of a perceiver relative to the action perceived. He believes that the various meanings of the verbs of motion, which are presented in the 4-volume Slovar' russkogo jasyka as separate lexical meanings, are actually a function of the interaction between lexis and grammar, specifically aspect. The sentence in 1 is an example of a verb of motion, podnimat'sja 'to go up, rise', with process meaning, sentence 2 shows the same verb with 'relative' meaning: (1) A Margarita v eto vremja uze podnimalas' stremitel'no vverx po lestnice ... 'and Margarita, meanwhile, was already going upstairs rapidly ... ' (M.Bulgakov) (2) Vdali sverkala 1'dami, podnimalas' iz-za rosci ostrym pikom gornaja versina. 'in the distance a lofty summit rose with its sharp peak behind a grove, and sparkled with ice' (A.N.Tolstoy) In 2 the summit does not literally rise up, instead the tense-aspect of the form podnimalas' pinpoints the position of a perceiver - the omniscient author as 'super-perceiver' together with the reader in his/her imagination - relative to the summit. Thus it can be seen that spatial deixis is crucial to the interpretation of the verbs of motion in Russian (Poupynin, this volume pp.113-15). Gabriele Diewald, in a paper entitled The Integration of the German Modals into the Paradigm of Verbal Moods (119-29), relates the 6 modal verbs of German (rnussen, konnen, etc.) in their epistemic use, e.g. as shown in 3: (3) Sie muB zuhause gewesen sein 'she must have been at home' to the verbal moods indicative, subjunctive I, and subjunctive II. She claims that both the modal verbs in their epistemic use and mood are deictic categories. The indicative 'expresses the deictic value of modal nearness, i.e. factuality' (Buhler's 'demonstratio ad oculos', or deixis proper) (Diewald, this volume, p.121). Subjunctive I indicates that the factuality judgement does not come from the present speaker, but from a reported speaker, 'i.e., from a displaced origo' (Buhler's imagination-oriented deixis) (Diewald, this volume, p. 122). On the face of it subjunctive II indicates nonfactuality, e.g. a speaker who utters 4:

4


(4) Sie stunde im Garten 'she would be standing in the garden' does not believe that the 'she' mentioned is standing in the garden. But there is more to subjunctive II than nonfactuality, since 4 is not exactly synonymous with 5: (5) Sie steht nicht im Garten 'she is not standing in the garden' Where the subjunctive II in 4 differs from the sentence negation in 5 is that subjunctive II alludes to an unfulfilled condition expressed somewhere in the surrounding text. Hence subjunctive II is an anaphoric/cataphoric expression, and qualifies as Buhler's third type of deixis, viz. the anaphoric use of deictic words (Diewald, this volume, pp. 122-3). In this respect subjunctive II resembles the pluperfect, which alludes to a reference point usually mentioned in the text, also in an anaphoric/cataphoric manner. The modal verbs of German in their grammaticalised (i.e. not fully lexical), epistemic use also express factuality values, e.g. as shown in 3 above or 6 below: (6) Ich kann mich getauscht haben 'I may have been mistaken' According to Diewald konnen and mussen express deixis proper, sollen and wollen are imagination-oriented deixis, whilst durfte and mag are anaphoric/cataphoric deixis. The recognition of the epistemic modals and mood as deictic items points up a parallel between them and tense (tense being already recognised as deictic): tense is basically inflectional, but with some periphrastic realizations; likewise mood is basically inflectional, but also has some periphrastic realizations, viz. the modal verbs in their epistemic use. Kathleen Connors, in a paper entitled Noun Classifiers and Gender Classes Another Look (133-48), ponders the difference between noun classifiers and noun genders. Noun classifiers are typically free morphemes, of a lexical nature, to be found predominantly in isolating languages. Noun genders, on the other hand, are realized by bound morphemes, of a morphological nature, and are to be found predominantly in agglutinative and inflectional languages. Classifier languages tend to have a large number, perhaps dozens, of classifiers (in keeping with their lexical nature), which display inter-speaker variation reminiscent of parole, whilst gender class languages tend to have only a small number of genders, between two and twenty (in keeping with their morphological nature), reminiscent more of langue. Classifiers are often generic or superordinate nouns, e.g. in Dyabugay the word for turkey, wawun, will typically occur together with the word mina, which means 'animal meat', in a noun phrase mina wawun lit. animal meat turkey 'turkey'. Although in general noun genders are semantically arbitrary, particularly for inanimate nouns, whereas classifiers are semantically motivated, classifiers are also to a certain extent arbitrary and unpredictable, and have to be learned. One encounters this in the classifiers that English has, for instance collective nouns, e.g. a pack of wolves but a pride of lions. Connors suggests that derivational and compositional nominalizing morphemes are classifiers (this volume, p. 135). An example is the English nominalizing suffix -ness, as in goodness, kindness: pre-school children generalise -ness to temperature adjectives to form the nouns *warmness, *hotness. Another example is the familiar diminutive suffix in many languages, e.g. in Canadian French the word jobinne from job, which classifies a job not only as small but also as of little economic worth. The semi-suffixes of German, e.g. -zeug as in Fahrzeug 'vehicle', or -tier as in Saugetier 'mammal', are superordinates in a manner reminiscent of classifiers. Such superordinates tend to be neuter in German, and the neuter

Introduction

5

gender itself can indicate superordinate, as in das Gewdchs 'plant', das Getrank 'beverage'. Often masculine and feminine indicate sub-terms of the superodinate, e.g.: das Steinwild 'ibex': der Steinbock 'male ibex', die SteingeiB 'female ibex' das Kraut 'plant': der Klee 'clover', die Klette 'burdock' (Connors, this volume pp. 136-7). Jozef Darski asks the question Was 1st Stamm? in the title of his paper (149-65) and provides a precise and detailed answer. The stem or root of a word is usually defined vaguely and indeed circularly as that part of a word which remains when the ending is removed. The stem of the verb in German used to be considered the imperative, until Bauer (1830:273) made it the infinitive with the ending removed. The problems that arise are illustrated by the present tense paradigm of the verb lacheln 'to smile', which is given in Engel (1996:418) as follows: (ich) lachl-e (du) lachle-st (sie) lachl-e (wir) lachel-n (ihr) lachl-e-t (sie) lachel-n (quoted by Darski, this volume p. 151) Assuming that the hyphen separates stem from ending, then we have here three different stems, viz. la'chl-, lachle- and Idchel-, which cannot be right. Darski claims that an adequate definition of and understanding of stem can only be reached by distinguishing between singular and plural, between the parts of speech (e.g. noun or verb stem), and for the verb between the tenses and the moods (present tense stem, past tense stem, indicative stem, subjunctive stem). Bein 'leg' has the singular stem Bein but the plural stem Beine. Machen 'to make' has the present indicative stem mach, but the present subjunctive stem mache. Darski suggests representing the stem ofraten 'to advise', which displays ablaut in the present tense (ich rate, du ratsf), as rAt, where A represents the change from a to a. In addition to the orthographic stem a more precise way of delimiting a stem is to do so phonologically. Geben 'to give' has the present indicative singular stem /'gIB/ and the present indicative plural stem /'ge:B/. /'gIB/ is realised as /'gi:p/ or /'ge:b/. /'ge:B/ is realised as /'ge:b/ or /'ge:p/. These stems can be conflated to a more general inflexional stem, which Darski represents as: /'gAB/ (Darski, this volume pp. 161-2). Yishai Tobin, in a paper entitled One Size Does Not Fit AH: A Semantic Analysis of "small/large" vs. "little/big" (169-80), takes the apparent synonyms little/small and big/large and uncovers the semantic difference between them. There are two reasons for suspecting that there is a semantic difference between them. One is common sense - why should English maintain both words if they mean exactly the same thing? The other is the structuralist tenet that there is no difference in form without a difference in meaning, which Tobin expresses by talking about the 'invariant meaning' of forms. So what is the different in meaning between the a sentences and the b sentences in 7 and 8? (7) a. I had a little brandy, b. I had a small brandy. (8) a. He has a big mouth, b. He has a large mouth. Why in 9 below does the writer first describe a car as small and then a few lines later as little?

6


(9) The next day Sholto picked up a small red car from somewhere and drove very fast from Rome. ... When he suddenly spun the wheel and sent the little car hurtling up a track, she squealed .... (Walker 1992:371, quoted by Tobin, this volume p.173) Why does the idiom in 10 occur with one member of the pair in question, but is ungrammatical with the other member of the pair? (The answer is not ossification in idioms.) (10) a. Who's afraid of the big bad wolf? b. ?Who's afraid of the large bad wolf? Finally, why is the protagonist of The Little Prince (Saint-Exupery 1943) first introduced as a small person, but then referred to throughout the text and most noticeably in the title as the little prince? The answer to all these questions has to do with what Tobin calls Semantic Integrality, by which he means the difference between perceiving something as discrete entities in discontinuous space on the one hand, or collectively as a set in continuous space, on the other. Little and big are unmarked for Semantic Integrality, i.e. they are neutral, whereas small and large are marked for Semantic Integrality, and express perception as a set in continuous space only (Tobin, this volume p. 170). Small and large express things as generic items, as a class of something, a prototype; whereas little and big may individualise an item as unique, they may focus upon it as a specific item, to be explored in greater detail. The analysis obtains greater generality, i.e. explanatory power, by virtue of the fact that Semantic Integrality has been used to explain other linguistic phenomena in English and Hebrew, including 'irregular' plurals, mass/count nouns, the synthetic vs. analytic systems of comparatives and superlatives, and the conditionals //vs. whether (Tobin, this volume pp. 170-1). Isabel Forbes and G.Kiss in their paper Colour Categorization and Naming in French and Hungarian (181-8) report on three tests which they carried out with native speakers of French and Hungarian on colour terms. Before 1969 most linguists held the Saussurean and Sapir-Whorf view that each language divides up the colour spectrum in its own (arbitrary) way, and furthermore that this formal-linguistic division determines the way in which people perceive colours. Since Berlin and Kay 1969, however, opinion has changed, and most linguists now believe that colour terms are not language specific and arbitrary but are universal, maintain Forbes and Kiss. In the first test just under 100 informants were asked to write down what in their opinion were the 'principal colour terms'. In the second test informants were shown a colour array containing 139 chips graded and numbered according to the Munsell system of colour notation, which was the system used by Berlin and Kay. The French and Hungarian speakers were consistent with each other in their recognition and naming of the colours, and consistent with the languages tested by Berlin and Kay. Thus the universality of the perception and naming of colours was confirmed (Forbes and Kiss, this volume pp. 181, 184). The third test was designed to discover how the French informants used brun and marron (both meaning 'brown'), and how the Hungarian speakers used piros and voros (both meaning 'red'). The informants were asked to list all the things that could be called brun, those that could be called marron, and similarly for the Hungarian terms. The results for French are given below in Figure 1:

Introduction

1

Figure 1 The distribution of brun and marron in informants' lists brun marron Category of object No. of mentions No. of mentions Hair 60 7 Eyes 20 45 Skin and other physical attributes 20 1 Persons (physical type) 12 0 Animals (including horses) 22 4 Trees and parts of trees 23 38 Clothes and material 41 23 Leather 14 22 Paint colour 11 18 Earth and Terra Cotta 9 4 (Wooden) furniture 15 8 Food 3 10 Tobacco 3 0 Miscellaneous 1 4 The table shows that the choice of word to describe a colour depends on the type of object which has the colour. Brun 'brown' is the favoured term, as against marron 'brown', to describe hair, skin colour, and physical type (Forbes and Kiss, this volume pp. 185-6). Similar results were found for the Hungarian terms for 'red' piros and voros. Dmitrij Dobrovol'skij asserts in his paper On the Cross-Linguistic Equivalence of Idioms (203-19) that it is a mistake to imagine that an idiom of L1 must always be translated into L2 using an idiom. It is far better not to use an idiom in L2 at all if the result is a functionally and stylistically better translation. Moreover, bilingual dictionaries of idioms are basically incapable of registering all L2-equivalents of a given LI-idiom for all possible contexts. Hence translators are best served by parallel textual corpora and other text-oriented tools. It is more important for the translator to know all the syntactic, semantic, and pragmatic properties of a given L1-idiom than to have a list of possible L2-equivalents, because the choice of equivalent always depends on the context. In this spirit Dobrovol'skij attempts to develop a typology of parameters for the cross-linguistic comparison of idioms, covering syntactics, semantics, and pragmatics. To take a semantic instance, many bilingual dictionaries consider the German idiom given in 11 to be equivalent to the Russian idiom given in 12: (11) German den Geist aufgeben "to give up the mind/spirit" 1. (obsolete) 'to die', and 2. (coll.) 'to stop working' (12) Russian ispustit' dux "to give up the mind/spirit" 'to die' A closer examination reveals, however, that whereas the Russian idiom can only mean 'to die', the German idiom is rare in the meaning 'to die', and is almost always used in the meaning 'to stop working', a fact which came out of a study of the textual corpora of the Institut fur deutsche Sprache in Mannheim. Therefore the idioms in 11 and 12 are not equivalent after all (Dobrovol'skij, this volume pp.212-13).

8


To take a pragmatic instance, on the face of it the German idiom in 13 meaning 'to be mad' could be translated into any idiom of a language which also means 'to be mad': (13) German von alien guten Geistern verlassen sein ,,to be left by all good spirits/genii" 'to be mad';«to have taken leave of one's senses But an investigation of the textual corpora of the Institut fur deutsche Sprache in Mannheim revealed that the idiom in 13 is restricted to hypothetical utterances only, for example 14: (14) Zugleich sprach sich der Finanzminister deutlich gegen Forderungen nach einem Sonderopfer der Bundesburger fur die DDR in Form von Abgabenerhohungen oder als Verzicht auf die Steuerreform aus. "Wir waren von alien guten Geistern verlassen, wenn uber den Umweg der Deutschlandpolitik sozialistische Vorstellungen eingebracht wurden", kommentierte Waigel dazu. (Mannheimer Morgen, 15.11.1989) [At the same time, the Minister of Finance clearly expressed himself against claims for a special sacrifice by the citizens of the Federal Republic for the GDR in the form of increased payments or abandoning the tax reform. ,,We would be left by all good spirits/genii if we indirectly brought in socialist ideas by way of the Germany policy", commented Waigel.] Since no Russian idiom meaning 'to be mad' could be found with the same illocutionary restriction, the inescapable conclusion is that Russian simply does not have an idiom fully equivalent in all respects to 13 (Dobrovol'skij, this volume p.217). Chris Gledhill, in his paper entitled Towards a Description of English and French Phraseology (221-37), puts forward an extended notion of collocation, based on Sinclair 1991. Usually linguists distinguish between free word combinations on the one hand, and collocations/idioms on the other. For Gledhill, however, every word is restricted in the syntactic patterns into which it can fit, and hence collocation extends to the whole lexico-syntax of a language - 'every syntagm is a collocation' (Gledhill, this volume p.226). The enabler of this extended notion of collocation is the computer, in corpus linguistics - the computer has done for linguistics what the telescope did for astronomy (Stubbs 1996, mentioned by Gledhill, this volume p.227). Furthermore, it is necessary to take into account register and genre to obtain a full understanding of collocations. Gledhill also brings in speech acts and rhetorical function, in distinguishing collocations from idioms. Idioms, he says, are marked rhetorically, whilst collocations are unmarked rhetorically. For example, dictums and proverbs often exhort or provide a metacomment, as in more haste less speed, meaning roughly 'take your time'. Moreover, corpus studies have shown that idioms are less widespread than previously thought. Moon 1987 found that out of 2265 idioms (e.g. out of the blue, foot the bill) examined in the 323 million word Bank of English (ex-Cobuild corpus, held at Birmingham University, England), only 135 occurred more than twice per million words, compared with, for example, 240 times per million words for the collocation of course (Gledhill, this volume, p.227). The implication is that collocation is more central to the 'idiom principle' than traditional idioms themselves. Collocations carry with them a 'semantic prosody'. For example, the English expression to set in typically has a negative subject, such as bad weather, gangrene. Consequently, any subject that appears with it becomes imbued with the pejorative semantics, e.g. The Tory Party had set in. It is one step from there to 'collocational frameworks' (Renouf

Introduction

9

and Sinclair 1991), which are short texts with the lexical words taken out, only the grammatical words remaining. Gledhill (this volume, p.228) gives the following example: (15) X the Xs with a X X X. X the X with X and X to X, and X each X in it. X the X, X with a X X or X and X over each X. X the Xs with Xs, X on Xly and X in X X until Xly Xed, and claims that a native or fluent speaker of English should be able to guess the genre, text-type, or rhetorical/communicative function of the text. He goes on to claim that collocational patterns vary according to register, and illustrates this claim with example texts of French and English scientific writing. Sven-Gunnar Andersson, in a paper entitled Register-Motivated Variation of Tense and Mood in German Final Clauses Introduced by damit (241-51), reports on a study which produced surprising results for the choice of tense and mood of the verb in final clauses introduced by damit in the spoken language. The type of sentence discussed in the literature on damit clauses is shown in 18 below: (18) Er schickte den Brief schon anderthalb Wochen fruher ab, damit er wirklich rechtzeitig ankam/ankomme/ankame. With regard to the written language the widely held view, as presented in Flamig 1964, is that the past indicative is neutral, contributing no special modal meaning; the present subjunctive is either used to mark the damit-clause as indirect speech or else has an adhortative function; whilst the past subjunctive is used either also to mark indirect speech or else has a counterfactual meaning. In an examination of damit clauses in written texts of the computerized COSMAS corpus in Mannheim the above analysis was confirmed (Andersson, this volume pp.242-5). But in a series of oral elicitation tests by far the most common choice made by informants was the present indicative. In sentence 18 above this would be the form ankommt (Andersson, this volume pp.245-7). Notice that the form ankommt is not given in 18 at all, i.e. the present indicative is nowhere discussed in the literature on damit clauses. Andersson interprets the preponderance of the present indicative in damit clauses with past context in the spoken language as an indication that such clauses are being recategorized as indirect speech (Andersson, this volume pp.247-8). Andreas Musolff, in his paper Dinosaurs, Metaphors and Political Argument (283-94), takes issue with George Lakoff's view that metaphors are fundamentally conceptual, not linguistic, in nature, and that the 'cognitive topology' or 'image-schema structure' of a metaphor are obligatorily present in every use of it in a given text, thus imposing limits on the possibility of free thought and inhibiting the creative use of metaphors (Musolff, this volume p.287). Instead, Musolff states: 'the political bias of metaphorical language use seems to me to be the function of the pragmatic context rather than an inherent feature of conceptual schemas' (this volume, p.289). The example he uses to illustrate his point is the metaphor of the dinosaur. If someone wishes to say about a person or an institution that they are in danger of becoming extinct, he can call them a dinosaur, as in the following example from the Times:

10

Langue and Parole in Synchrony and Diachrony (16) Conference dinosaurs [headline] Like the tail-thrashings of a dying dinosaur, Labour activists and trade unionists meet in Blackpool this week to protest at a new world for which many of them are feebly adapted. (...) Mr Blair must know that one of the main obstacles to taking tough decisions in a Labour government has been the party itself. By getting his "betrayal" in first, he hopes to win the arguments now rather than then. Labour activists understand that: which is why they see this week as their last chance to influence a possible Labour government. They also understand, however, that the apparent influence is illusionary. No wonder they sense that the current climate is inhospitable to their survival. (The Times 30 September 1997, quoted by Musolff, this volume p.286, italics A.M.)

But consider now the following extract from a Guardian interview with the retired Conservative politician Julian Amery: (17) [Amery:] "My own background was Empire protectionist. I believed in the Empire as the basic system on which we must work and, after the War, in Europe as an extension of that empire. Of course, there's no one else left who thinks like that. We're all dinosaurs." (The Guardian 11 April 1992, quoted by Musolff, this volume p.289) Musolff claims that Amery's use here of the dinosaur metaphor in a self-mocking, nostalgic way deprives the metaphor of its negative import, and is actually a clever conversational move or argumentative gambit. The negative extinction part of the metaphor turns out to be optional (Musolff, this volume pp.289-90). Metaphors are not fixed and unchangeable in langue, but are open to communicative and pragmatic manipulation in parole. Natalia Guermanova, in a paper entitled National Profiles of Language Perception: Instrumental vs Cultural-Value Conceptions (307-19), discusses attitudes towards languages, particularly one's native language, and posits two kinds of attitude. A preoccupation of the Enlightenment with reason and logic led to an attitude which extolled clarity and precision in language. Guermanova calls this attitude the instrumental conception. The French and the English were particularly keen to emphasise these virtues in their own languages (Guermanova, this volume pp.313-14). But the Age of Reason gave way to Romanticism. Now people began to admire the expressive, poetic qualities of language. They also believed to find those qualities particularly in the language of ordinary folk, and grass-roots folk culture and local dialects suddenly acquired respect and even awe. Furthermore, languages became associated with nations and national psyche, for example in the writings of Herder and Humboldt. Guermanova calls this attitude the cultural-value conception. It is a conception held by the Russians towards their language, and by Americans towards American English. Americans consider their language to exhibit a democratic frontier spirit, a bold neglect of prescribed norms, in keeping with the American people, and are sympathetic towards slang (Guermanova, this volume pp.314-16). German, incidentally, occupies the middle ground between the instrumental and the cultural-value perceptions (Guermanova, this volume p.315). Language perception is important for various reasons, not least because it influences language theories and language teaching methodologies. After having summarised the views of several contributors to these Proceedings, it is now incumbent upon me, as the originator of the langue-parole theme of the conference, to

Introduction

11

give my own views on the langue-parole dichotomy.1 My own attitude is a rather traditional one, viz. langue is the system, parole is the substantiation of the system, and it is langue which constitutes the proper Untersuchungsobjekt of linguistics. I would perhaps, however, add one more element to this concoction, and that is the idea of progress in (core, theoretical) linguistics. The possibility of progress in linguistics is based on the idea that a 'descriptive' (i.e. not generative) grammatical analysis (including the analyses of traditional grammar), which is of course part of langue, is a hypothesis only, not a statement of obvious facts, and may be superceded by a better analysis, if one can be found.2 Many peculiarities and irregularities of language usually attributed to parole turn out to be artefacts of a particular langue analysis, they turn out to have been introduced into language artificially by the linguist-scientist in a bad analysis at the langue level. Parole is a fascinating object to study, and must be studied by (langue-oriented) linguists, but only because it contains the anomalies and self-contradictions which can lead us - if we recognise them as such and treat them properly - dialectically to a revision of and improvement in our rules and regularities of langue. Langue is the scientific part of language study, the ordered part, parole is the pre-scientific part, the chaotic part. Hartmann (1997:199) has called this approach 'naive scientism'. It is an interesting and well-chosen phrase from a parole point of view, but if Dr Hartmann will forgive me I will use it to illustrate the langue approach and to demonstrate that the criticism embedded in the phrase is ill-founded. Let us take the passive to exemplify the langue approach adumbrated above. The passive construction, consisting in English of be + V-ed, is traditionally derived from an underlying active, and is called a voice of the verb. This analysis attempts to capture the known regularities of the form be + V-ed, and is part of langue. Unfortunately there are several well-known irregularities and idiosyncrasies associated with this analysis of the passive. One of them is that most actually occurring active sentences in texts, when converted into the passive, produce grotesque curiosities such as Girls are liked by John (Svartvik 1966:2). Another is that, whereas the canonical grammarian's sentence to illustrate the passive is John was hit by Mary, with the dynamic verb to hit (which emphasises the idea that the subject 'receives the action of the verb'), a singular definite subject, and the agentive by-phrase realised, an examination of actual texts shows that the typical passive sentence is more like: This difficulty can be avoided in several ways, with a modal verb and passive infinitive, an abstract durative-type verb, an abstract noun as subject, the agentive by-phrase not realised, and an adverbial phrase present in the sentence (Quirk et al. 1985:167-8). A third problem is that most actually occurring passive sentences in texts - the figure is approx. four fifths - occur without the agentive by-phrase realised (Quirk et al. 1985:164-5), whereby the agent is usually non-recoverable, which contradicts the idea that they all came from an underlying active, where the presence of a subject would be obligatory. There

1

The remarks that follow apply to sentence grammar only, they do not apply to approaches which study chunks of language larger than the sentence, such as text grammar, corpus studies, theme-rheme analysis, speech act theory. 2 The term langue is therefore ambiguous. On the one hand it means the true and genuine system of language, which we pursue but never actually reach. On the other hand it means our attempts to capture that system, in the descriptions which we conceive and which we change and refine constantly in the light of scientific progress.

12


are many more anomalies associated with the voice analysis of the passive, but that will do for present purposes. Such anomalies are usually considered to be part of parole. Is it naive to expect linguists to be able to find an explanation within langue to the above-mentioned three anomalies? Is it expecting too much from linguistic science 'scientism' - to expect linguists to be able to find a structural explanation within langue to the anomalies? If one adopts the parole approach the answer will be 'Yes', and one will content oneself with a parole-style explanation, perhaps using theme-rheme organisation or some other device of text grammar. And such explanations will sound reasonably convincing. But the fact is, it is possible to explain the anomalies within langue. It is possible to find another analysis of be + V-ed which explains them, and that analysis is one which sees be + V-ed in its resemblance to the perfect, as an aspect of the type Auxiliary + Participle (Beedham 1982, 1998; see also Poupynin 1980, Schoorlemmer 1995). The actional passive has its own meaning, 'action + state' (i.e. actives and passives are not 'cognitively synonymous'), and like all Auxiliary + Participle aspects, the syntax (combinatorial possibilities) of the passive is determined by the lexical aspect of the verb and the compositional aspect of the sentence in which it occurs, i.e. the syntax of the passive is not determined by transitivity (on compositional aspect see Verkuyl 1972, 1993). The explanation for the above-mentioned three anomalies associated with the passive is therefore as follows. Firstly, the fact that the transformation of so many active sentences into the passive produces grotesque curiosities is simply one piece of evidence amongst many for the view that the grammatical voice analysis under which the grammarian derives the passive from an 'underlying' 'active' is incorrect. Under the aspect analysis in its application to English, be + V-ed is related to have + V-ed (the perfect) and to be + V-ing (the progressive), there is no transforming of actives into passives. Anyone who had resorted to a parole type explanation of odd passives such as theme-rheme analysis would have been barking up the wrong tree. The true explanation lies in a correction of the langue analysis. Secondly, Quirk et al.'s prototypical passive sentence This difficulty can be avoided in several ways looks very much like the kind of sentence found in discussions of compositional aspect. What is important is the type of subject NP (singular or plural, specific or non-specific), the lexical aspect of the verb, the presence of a modal verb and a non-finite form of the verb, the presence of an adverbial phrase. All these factors affect the compositional aspect of the sentence. And if the passive is an aspect of the type Auxiliary + Participle, these factors, because they influence compositional aspect, will inevitably affect passivizability (Beedham 1982:115-18). Once again, a parole-type or textual explanation was tempting, but it would have been wrong to go down that route. It has turned out that the problem, which manifested itself in parole, was an artefact of an incorrect grammatical analysis at the level of langue. The real and deeper explanation to the problem has been found in langue. Finally, the reason why corpus studies have found that four-fifths of passive sentences in texts (i.e. at the level of parole) occur without the agentive by-phrase is that it is wrong to derive passives from actives (at the level of langue) and thus to expect passives to occur with the agentive by-phrase realised. The passive is an aspect of the verb, not a voice of the verb, and has a derivation in English parallel to that of the perfect and the progressive. In that derivation the agentive by-phrase is a prepositional phrase which, like most prepositional phrases, is optional. Once again the solution to a problem which manifested itself in parole has been solved at the level of langue. It was not naive to expect to find a solution in langue, it

Introduction

13

was entirely realistic. To look for a solution to the passive in langue, not parole, is not scientism, it is science. Not that that is to say that all parole studies are inherently wrong. Quite the reverse, linguists must study parole. Text grammar, corpus studies, theme-rheme analysis, speech act theory, for example, are all interesting and fruitful areas of linguistics, and many of the arguments and data in Beedham 1982 come from parole.3 But the point is what you do with your parole findings. To the sentence-grammarian the study of parole is not an end in itself, but a means to an end. That end is the elucidation of langue. (The aspectual analysis of the passive is placed in the context of Saussurean structuralism in Beedham 1995:42-52, 66-86). The artificial and transitory nature of grammatical descriptions is emphasised by Mati Hint in his paper Completeness and Symmetricity of Paradigms (81-7), and also by Bozena Bednafikova in her paper System Description or Systematic Prescription? (75-80). Igor Boguslavsky, in his paper Modals, Comparatives, and Negation (253-62), demonstrates how a well-known ambiguity of comparatives results from the interaction between lexical semantics, communicative function, and conversational maxims. Elisabetta Fava, in her paper Langue and Parole in Speech Act Theories: Some Considerations and a Proposal (263-81), attempts to relate grammatical structure to specific speech acts, referring to English, Russian, Greek, and Italian. Magdalena Jurewicz, in her paper Metatext-Sequenzen in deutsch-polnischen Gesprachen: Eine Fallstudie anhand konsekutiv gedolmetschter Texte (295-304), using the example of German and Polish shows how the interpreter in consecutive interpreting does not simply relay information but contributes to the creation of the conversation as well. Radmila B. Sevic, in her paper Early Collections of Private Documents: The Missing Link in the Diachronic Corpora? (337-47), advocates using private letters and documents in diachronic studies in order to reach the nearest approximation to the spoken language of earlier periods. Alexander T. Bergs, in his paper Social Networks and Language Change in Middle English: The Challenge of Diachrony (349-62), attempts to apply the concept of Social Networks from contemporary sociolinguistics to a diachronic study of Middle English. Pieter Seuren, in his paper Eubulides as a 20th Century Semanticist (47-61), presents the neglected Ancient Greek philosopher Eubulides, bitter rival of Aristotle, as a precursor of 20th century logical semantics. And finally R.H.Robins, in his paper Key Dates in Twentieth Century Linguistics (63-72), looks back on linguistics in the 20th century and gives a personal view of the key dates, events, and landmarks. How can it be that we think we are all structuralists, and yet, as Tobin says, structuralism is the revolution that never happened? Hopefully the papers that follow will go some way towards providing an answer to that question. But the main point of these Proceedings, as was of the conference itself, is to provide a platform for discussion and debate in the dialectical process of the science of linguistics.

Christopher Beedham, Department of German, University of St Andrews, Scotland/UK

3

In terms of day to day methodology, the main difference between langue-oriented and parole-oriented linguists is that the former work with invented sentences and the grammaticaliry judgements of native speakers, whereas the latter work with texts, often large computerized corpora, and statistical frequency.

14


REFERENCES Bauer, Heinrich (1830). Vollstdndige Grammatik der neuhochdeutschen Sprache. Dritter Band. Reimer, Berlin. Beedham, Christopher (1982). The Passive Aspect in English, German and Russian. Narr, Tubingen. Beedham, Christopher (1995). German linguistics: An introduction. ludicium, Munchen. Beedham, Christopher (1998). The perfect passive participle in Russian: A review of Participial passive and aspect in Russian, by Maaike Schoorlemmer. Lingua, 105, 79-94. Berlin, B. and P. Kay (1969). Basic Color Terms: their Universality and Evolution. University of California Press. Reprinted 1991. Bondarko, Alexander V. (1971). Vid i vremja russkogo glagola (Znachenie i upotreblenie). Moskva. Engel, Ulrich (1996). Deutsche Grammatik, 3., korrigierte Auflage. Groos, Heidelberg. Flamig, W. (1964) Untersuchungen zum Finalsatz im Deutschen.(Synchronie und Diachronie). Sitzungsberichte der dt. Akademie der Wissenschaften zu Berlin. Klasse fur Sprachen, Literatur und Kunst. Jahrgang 1964, Nr. 5. Akademie-Verlag, Berlin. Godel, R. (1957). Les sources manuscrites du Cows de linguistique generale de F. de Saussure. Droz, Genf. Hartmann, Ralph (1997). Review of C.Beedham, German linguistics: An introduction (iudicium 1995). Informationen Deutsch als Fremdsprache 24, 198-200. Meillet, Antoine. (1893). Les lois du langage I: Les lois phonetiques. Revue Internationale de sociologie (Paris) 1.311-21. Moon, R. (1987). The analysis of meaning, In Sinclair, J. McH. (ed.) (1987) Looking up: an Account of the Collins COBUILD Project, 86-103. Collins ELT, London. Poupynin, Youri A. (1980). Funktsionirovanie vidov russkogo glagola v passivnyx konstruktsiyax (K probleme vzaimosvyazey kategorii vida i glagola) [The functioning of the aspects of the Russian verb in passive constructions (On the problem of the reciprocal links between the categories of aspect and the verb)]. Diss. kand. filol. nauk., Ph.D. dissertation, University of Leningrad. Quirk, Randolph, S.Greenbaum, G.Leech and J.Svartvik (1985). A Comprehensive Grammar of the English Language. Longman, London. Renouf, A. and J. McH. Sinclair (1991). Collocational frameworks in English, In K. Aijmer and B. Altenberg (eds.) (1991) English Corpus Linguistics, 128-144. Longman, London. Saussure, Ferdinand de (1879). Memoire sur le systeme primitif des voyelles dans les langues indo-europeennes. B. G. Teubner, Leipzig. Saussure, Ferdinand de (1916). Cours de linguistique generale. Ed. by Charles Bally & Albert Sechehaye, with the collaboration of Albert Riedlinger. Payot, Lausanne & Paris. Saussure, Ferdinand de (1972). Cours de linguistique generale. Edition critique preparee par Tullio de Mauro. Payot, Paris. Saussure, Ferdinand de (1983). Course in General Linguistics. Transl. by Roy Harris. Duckworth, London.

Introduction

15

Saussure, Ferdinand de (1989-90). Cows de linguistique generale. Edition critique par Rudolf Engler. 2 Bande. Harrassowitz, Wiesbaden. Schoorlemmer, Maaike (1995). Participial passive and aspect in Russian. OTS, Utrecht. Searle, John (1969). Speech acts: An essay in the philosophy of language. Cambridge University Press, Cambridge. Sinclair, J. McH. (1991). Corpus, Concordance, Collocation. Oxford University Press, Oxford. Slovar' russkogojazyka. V 4-x tomax. (1981-84). Moskva. Stubbs, M. (1996). Text and Corpus Analysis. Routledge, London. Svartvik, Jan (1966). On Voice in the English Verb. Mouton, The Hague. Tobin, Yishai. (1990). Semiotics and Linguistics. Longman, London. Verkuyl, Henk J. (1972). On the compositional nature of the aspects. Reidel, Dordrecht. Verkuyl, Henk J. (1993). A theory of aspectuality: The interaction between temporal and atemporal structure. Cambridge University Press, Cambridge. Wittgenstein, Ludwig (1980). Philosophische Untersuchungen (PU). Suhrkamp, Frankfurt am Main.

CORPUS Saint-Exupery, Antoine de. 1943. The Little Prince (translated from the French by Katherine Woods). Reynal & Hitchcock, New York. Walker, Elizabeth. 1992. Hallmark. Headline Book Publishers, London.


I HISTORY OF LINGUISTICS


1 THREE SAUSSURES — ONE 'STRUCTURALIST' AVANT LA LETTRE E. F. K. Koerner, University of Ottawa, Canada

Abstract This paper argues for the intrinsic unity of Saussure's linguistic thinking, from his enfantillage of 1872 to the posthumous Cours of 1916, against proposals made by Albert Sechehaye (1940) and Eric Buyssens (1942), who distinguished between three and six 'linguistiques saussuriennes', respectively, based on the Cours alone. In 1961 Buyssens revised his position following the publication of Godel's Sources manuscrites (1957), but the idea of at least 'two Saussures' (cf. Redard 1978), a Saussure of the Memoire (1878) and a Saussure of the Cours (1916), has largely been maintained in the annals of linguistic science. More recently, Harris (1994) suggested that we are talking of at least three Saussures of the Cours when using Saussure's name in order to refer to his general linguistics, a view that can hardly be maintained by a discerning scholar working with Engler's 'edition critique' (1968, 1974). The present paper focuses specifically on Saussure's concept of 'system', tracing its use throughout his thinking about language from 1872 till 1911. Harris' (1983) decision to translate Saussure's 'systeme' by 'structure' points to a serious misunderstanding of Saussure's legacy.

1. INTRODUCTORY REMARKS It was the fate of Ferdinand de Saussure (1857-1913) to be known for his work in historicalcomparative Indo-European linguistics during his lifetime but to become known almost exclusively for the posthumous compilation of his Geneva lectures during 1907-1911 on (at least in the manner they were presented by the editors and subsequently received by the next generations of readers) non-historical, general and 'synchronic' linguistics, after his death. During his lifetime Saussure was known, indeed famous, for at least two publications, the Memoire sur le systeme primitif des voyelles dans les langues indo-europeenms which he published after only six semesters of formal study at the Universities of Leipzig and Berlin in December 1878, at age twenty-one and fifteen months before submitting his doctoral thesis, and his 1894 and 1896 articles on the accentual system of Lithuanian, which appeared, respectively, in the years before and after his promotion from extraordinary to ordinary professor of Indo-European philology at the University of Geneva, and for which he has been credited with a 'Law' (see Collinge [1985:149-152], for details). This paper constitutes an

19

20


attempt to see Saussure's legacy as a lifetime commitment in search of system or systems with a view to demonstrating the essential wholeness of his work.

2. THREE OR MORE SAUSSURES? In 1940, at age 70, Albert Sechehaye (1870-1946), the person largely responsible for the manner in which Saussure's ideas were presented in the posthumous Cours de linguistique generale, first published in 1916 (slightly revised in 1922 and 1931 for the second and third editions and ever since frequently reprinted and translated into more than a dozen languages), proposed to distinguish between 'trois linguistiques saussuriennes', namely one he called 'linguistique de la parole pre-grammaticale', an enterprise which would get us back into pre19th-century language genesis speculations if taken seriously; a 'linguistique de la parole proprement dite', which would investigate spoken language as found in dialect study for instance, and a 'linguistique de la langue', which would approximate what Saussure had in mind as regards the central goal of linguistic research as it alone could properly investigate the system of language (Sechehaye 1940). However, two years later, when Eric Buyssens (b.1901) suggested that he could distinguish altogether 'six linguistiques de F. de Saussure' (Buyssens 1942), he was severely criticized by Sechehaye (1944). Buyssens had taken different passages from this 'vulgata' text to suggest that the three dichotomies advanced by Saussure — langue/parole, synchronie/diachronie, and linguistique externe/linguistique interne — could, misleadingly he admits, be seen as possibly leading to six manners in which to conduct linguistics. Instead, Buyssens proposes, quite in line with Saussure's own thinking (as Sechehaye [1944] concedes), a synchronic linguistics which studies the language system and its functioning in speech, and a diachronic linguistics that explains the history of language in terms of changes in the society which speaks it. In 1961 Buyssens revised his position following the publication of Robert Godel's (19021984) Sources manuscrites (1957), which corrects many of the seeming contradictions in Saussure's argument which had, more often than not, been introduced by the editors, as Engler's 'edition critique' has amply documented (Saussure 1967-1968). Yet the tradition to see Saussure in at least two terms, as a historical linguist (represented by his Memoire sur le systeme primitif des voyelles dans les langues indo-europeennes), on the one hand, and as a 'structuralist' (represented by the Cours), on the other, has continued in the annals of linguistic science (cf. Redard's 1978 account). In the present paper the attempt is made to illustrate that this fable convenue is false, since the Memoire of 1878, while dealing with an early stage of language — the reconstructed phonological system of the Proto-Indo-European vowels — is in fact a 'synchronic' study. Moreover, apart from the papers that led up to the Memoire, it is suggested that already in his enfantillage (Saussure's term in retrospect) of 1872, when he was less than fifteen years old, his 40-page "Essai pour reduire les mots du Grec, du Latin & de 1'Allemand a un petit nombre de racines" can in a way be seen as foreshadowing his later preoccupation with discovering or building on the presence of a system in language, which figures so prominently in the Memoire. It is Saussure's emphasis on 'systeme' (of which Mounin in 1968 reported 138 occurrences in the Cours) that rightly or wrongly made him 'the father of structuralism'. There is another aspect to the alleged 'multiplicity' of Saussure's teachings. In 1994, reviewing a monograph by a philosopher who appears to have discovered Saussure only

Three Saussures ?

21

recently as a subject of theoretical inquiry (Holdcroft 1991), Roy Harris ran through the exercise of asking how many Saussures there may be with regard to the Cours alone, namely (Harris 1994:184-185), as 1) Saussure, who "can be treated for convenience's sake, as the putative author of the Cours de linguistique generale, even though 'the real' Saussure did not write it", a kind of short-cut most of us use when discussing particular points made in the vulgata text; 2) his name could also be invoked as the "lecturer who gave any of the courses at Geneva during the years 1907-1911" but to whom certain views could be attributed only by reference to particular students' notes taken down during his lectures, and finally 3) a "third possibility", which would mean, in Harris's (1994:185) view to treat the name Saussure as referring to 'the real' Saussure, i.e., the theorist whose ideas about language seem to have evolved over a period of time, were probably never expressed fully in any course of lectures that he gave, and were undoubtedly still far from having reached a definitive form at the time of his death. The direct evidence of the views of this theorist (Saussure 3) is altogether more controversial, and certainly not identical with that available for establishing the views attributable to Saussure (1) or Saussure (2). However, this 'triplicity' is not the issue in the present paper. Besides, those who want to give Saussure the fairest reading possible will, before they attribute a particular position to him (= Saussure 3?), consult Engler's edition critique (Saussure 1967-1968, 1974). Saussure's general ideas about language and how to analyze them come out clearly enough in all three kinds of'Saussure' that Harris distinguishes. In the present paper I shall pursue the seemingly tortuous route that the idea central to Saussure's linguistic thinking throughout his life, namely, that of language as a system of mutually interdependent parts, has taken, not in Saussure's mind, as it appears to have been always clear to him, but in his teachings; for it is through both his published and spoken word that we have come to associate the concept with his name.

3. THE SEARCH FOR LANGUAGE AS A SYSTEM 'Ou TOUT SE TIENT' IN SAUSSURE More than a quarter-century ago, William Gamwell Moulton (b.1914), one of the very few direct students of Leonard Bloomfield (1887-1949) — who himself read Saussure's Cours with great interest (cf. Bloomfield 1923) — made the following statement at the outset of a review of four books devoted to Swiss German dialectology, an area in which he had distinguished himself: It is a tenet of 20th-century linguistics that language is systematic. This has been expressed variously in such statements as that language is 'un systeme ou tout se tient',1 that the grammar of a language is a 'system of rules', and that the use of language is 'rule-governed behavior'. (Moulton 1971:938-939) Hardly anyone would want to quibble with such an affirmation about the broad use of this idea today. In the present context, the footnote which Moulton appended to this characterization makes for more interesting reading, and we will see in what follows the quotation why: Although the famous phrase is customarily attributed to Ferdinand de Saussure, in a reasonably careful rereading of the Cours I was unable to find it. The earliest occurrence that I have been able to locate is in Maurice Grammont, Traite de phonetique (Paris, 1933), p.167. Did Meillet also say it? Was Saussure the first to say it? Can anyone help me?

22


To be sure, Moulton was neither the first nor the last to have been frustrated in his search for this turn of phrase in Saussure's Cours. Apart from the common observation one can make, namely that, with rare exceptions, good practitioners of the craft tend to be poor historians (if they take any active interest in the history of their discipline at all), it also seems that at least part of the reason for this frustration is due to the long-standing amnesia of American linguistics with regard to European scholarship, from what Thomas Sebeok once characterized as 'the smug thirties' to at least the 1960s, which included a lack of familiarity with or disdain for the work of Saussure, Meillet, Trubetzkoy, Hjelmslev and other European linguists (cf. Haugen 1951, for an early critique of this attitude). Indeed, following Bloomfield's opus magnum of 1933, we can also notice a considerable lack of interest among his followers in historical-comparative linguistics. (Note that those who were fairly well read in Indo-European linguistics — or Romance philology for that matter — and who kept the subject alive in the United States were more often than not, if not exclusively, refugees from Nazi persecution, not American born.1) Moulton's queries are even more interesting for other reasons, as he seems to have been the first scholar to publicly challenge the traditional ascription of the phrase "ou tout se tient" to Saussure. He also suspects that Meillet used it, and he refers to an author and a book which are both at least in some way relevant to the subject of this paper: like Antoine Meillet (18661936), Maurice Grammont (1866-1946) was a student of Saussure's in Paris during the 1880s, and the subject of his 1933 book, contrary to what the title suggests, was devoted to phonology, not phonetics. As regards the urgency with which Moulton expressed his request for help, one cannot but hope that some reader of his note came to his relief soon after this fourth 1971 issue of Language was published, probably at exactly the same moment that I defended my thesis on Saussure which was published fifteen months later, with only minor changes, a new preface and an index of names (Koerner 1973). In it Moulton might have found at least a partial answer to his questions, for it says — in an explanatory footnote to a passage from Meillet's "Le9on d'ouverture du Cours de Grammaire comparee au College de France" of 1906, in which the characterization of language as "un systeme ou tout se tient" figures quite prominently (Meillet 1921 [1906]: 16; Koerner 1973:231): This phrase defining language as "un systeme [...] ou tout se tient" was frequently employed by Meillet and can also be found in the writings of his pupils, notably in J. Vendryes'. Cf. Meillet's pamphlet, La Linguistique (Paris: Larousse, 1916[1915]), p.7: "Toute langue est un systeme rigoureusement agence ..., tout se tient dans le systeme d'une langue". It is interesting to note that Meillet uses this formulation in connection with Saussure's teaching at Paris (1881-91); Meillet then was apparently not aware of the fact that Saussure's lectures on general linguistics were about to be published. (Koerner 1973[1971]:240n.23) One reason for citing this note is also to defend myself against the accusation (e.g., by Szemerenyi 1980:160) that I had attributed this famous phrase to Saussure, a charge against which Brogyanyi (1983:143-144) defended me, referring in addition to other places in the same book of mine as well as to Koerner (1975 [1972]: 746, 797), a study with which Szemerenyi was also familiar (cf. his review in Phonetica 36.162-165 [1979]), in which similar 1 I am thinking in particular of Henry M. Hoenigswald (born in Breslau in 1915) among IndoEuropeanists, and of Yakov Malkiel (1914-1998), born in Kiev but educated in Berlin, and the Viennaborn Henry Kahane (1902-1992), both of whom were doctoral students of the Romance philologist Ernst Gamillscheg (1887-1971), but also of the Viennese Ernst Pulgram (b.1915), who did his second doctorate at Harvard under British-born Joshua Whatmough (1897-1964) in 1946.

Three Saussures?

23

observations had been made. Indeed, following the publication of Peeters' 1990 paper, which supplies further evidence of my subsequent research into the question (e.g., Koerner 1984, 1987), I think there is no need any more to publicly claim ownership of my original findings. (Perhaps I should add that my 1971 dissertation was devoted to the Cours and its impact on 20th-century linguistic thought, not to his Memoire; this interest of mine is more evident in Koerner 1987.) Still, as the record shows, this often-used tournure has received further historical investigation following Brogyanyi's 1983 paper (Toman 1987, Hewson 1990, Peeters 1990), in part also because the phrase — and the idea — is still widely attributed to Saussure in the critical literature, at times against Saussure and 'structuralism' generally (cf. Brogyanyi 1983:145-146). Even the Indo-Europeanist Oswald Szemerenyi (1913-1996), who was so well familiar with the essential tenets of Saussure's Memoire (Szemerenyi 1973), went so far as to regard the phrase as "absurd" (1980:160). While I do not think that the idea of language as a system of interrelated terms needs to be defended — I would find it hard to believe that there is any serious linguist around who would challenge such a view — it seems that a more detailed analysis of the primary sources concerning its early use is still desirable to set the record straight, and to permit future (careful) users of this famous dictum the benefit of an historical accounting of its evolution and proper application to the facts of language.

4. ATTRIBUTIONS OF 'Oil TOUT SE TlENT' TO SAUSSURE Brogyanyi (1983) made no mention of his one-time mentor's characterization of this phrase as 'absurd', but he showed (145-146), as I just noted, that some linguists used the quote to criticize (Saussurean) structuralism generally as not giving an adequate picture of living languages. Others, who ascribe it to Meillet, not Saussure, like Lepschy (1970: 34), remarked that when Meillet stated in the preface to his Introduction a l 'etude comparative des langues indo-europeennes that language was "un ensemble ou tout se tient" (Meillet 1903:x), "he was expressing a commonplace idea rather than an original one". The key term here (and the source of Lepschy's derision), it appears, was Meillet's use of the word ensemble rather than systeme2 in this passage, as the former has a seemingly lengthy tradition in French thought. Bogyanyi (p. 145) quotes the definition which Emile Littre (18031881) had given in his Dictionnaire de la langue francaise under the entry of 'systeme' (1869:2119),3 namely, as an "Ensemble de choses qui se tiennent", and which harks back, according to Mounin (1972:43), to the entry for the same term in the French Encyclopedic of 1765, which in turn is quoted by Walther von Wartburg (1888-1971) in his Franzosisches etymologisches Worterbuch as an "ensemble de choses qui se tiennent d'une maniere ou d'une autre, considerees sous Tangle de leurs relations" (1966:504). However, Lepschy (b.1935) did not go much further in his investigations and, as a result, missed Meillet's much more forceful definitions in his many other, earlier as well as later, writings, including the revealing passage

2 Even though the context shows that Meillet in fact used the term 'systeme linguistique indoeuropeen' just before the passage Lepschy refers to. 3 Erroneously, Toman (1987:404) misquotes Brogyanyi (1983) twice when substituting 'systeme' by 'structure'.

24


in the 1912 edition of Meillet's Introduction, which was retained in its many subsequent editions (see section 5.2 below).

4.1 Early attributions The earliest reference to Saussure as the originator of the phrase appears to have come, interestingly enough, from the pen of his one-time student at the University of Geneva and later co-editor of his lectures on general linguistics, Charles Bally (1865-1947). In the first edition of his influential Linguistique generale et linguistique francaise Bally noted (1932:9 = 1944:17): "Dans un systeme, tout se tient [...]: ce principe, proclame par Saussure, conserve pour nous toute sa valeur". This affirmation seems to have been the source of Nikolaj Sergeevic Trubetzkoy (1890-1938), who in his programmatic, if not propagandistic, article of 1933, "La phonologic actuelle", cites the phrase characterizing language as "un systeme ou tout se tient" altogether four times (Trubetzkoy 1933:241-244 = 1969:159-163), each time identifying Saussure as the originator and suggesting (without offering a particular locus — of course) that the phrase in question could be found in the Cours, from which he quoted elsewhere, for instance when referring to the following famous phrase from the Cours: "Les phonemes sont avant tout des entites oppositives, relatives et negatives" (Saussure 1931 [1916]: 164). It seems that on the authority of Bally and then possibly also on that of Trubetzkoy, it became fairly commonplace to attribute the phrase to Saussure. Whatever the (secondary or tertiary) source of this ascription, it seems that the phrase in question fitted well into the argument of those post-Saussurean linguists who took the Cours as a textbook of structuralism rather than as an original attempt at an overall general linguistics in which both synchrony and diachrony would find their proper place.

4.2 Later ascriptions By the 1960s, when linguistics had become a popular subject of university instruction, the expression of language as a 'systeme ou tout se tient' and its congeners had become widely used in the classroom and in the textbook literature. It was a handy phrase if one was to argue in favour of a structural approach to language. It would probably be useful to check all sorts of modern textbooks to make an accounting of more recent ascriptions of this 'tour de phrase'. These are more likely due to a kind of oral, classroom transmission than the result of original research as, for instance in Fishman's 'brief introduction' to sociolinguistics (1970:9), where he makes a general reference to the Cours, obviously without having read it. Given limitations of space, reference to just two more places may suffice where the phrase is credited to Saussure: in Jan Mulder's Martinet-inspired Foundations of Axiomatic Linguistics (1989:135) and an article by Jacob Mey on 'semiosis', where the author even goes so far as to supply a page reference to the Cours, affirming: Perhaps one of Saussure's most important (and at least most widely quoted) statements has been that in language, everything holds together: "dans le langage, tout se tient" (1966:72). The immediate corollary of this statement is that in language everything is possible, as long as one knows, and sticks to, the rules of the game, [...] (Mey 1992:229)

Three Saussures?

25

It is obvious that the author has no idea where this phrase has come from and in what context it was originally conceived. Still, Mey is largely correct in his belief that comparable ideas can be found in the text compiled by Bally and Sechehaye, but not where his reference leads us to, namely, to a discussion of phonetics ("Classification des sons d'apres leur articulation buccale").4

5. ON THE HEELS OF THE SOURCES OF 'Ou TOUT SE TIENT' It appears to me that most teachers of linguistics did not particularly care where the phrase had come from. Indeed, it did not really matter. It was a useful way to explain the nature of language structure, and this usually sufficed. We know of various other instances where concepts or a 'joli mot' (as Meillet called Baudouin de Courtenay's 1881 creation of 'morpheme' by analogy to 'phoneme') are ascribed to the wrong person.5

5.1 The circuitous history of 'ou tout se tient' As we have already noted, there appears to be a fairly long list of scholars to whom the phrase has been credited. To be sure, the idea of language being a system is not a 20th-century discovery. The organicist view of language which permeated 19th-century historicalcomparative linguistics implied that languages were open to systematic analysis (Koerner 1975). So when the French lexicographer Littre, an ardent follower of Auguste Comte's philosophy of science, by the way, defined 'systeme' as an "[ejnsemble de choses qui se tiennent" in 1869, he may be said to have done little else than formulate a generally-held view. As a result, it was not merely the context in which Littre's definition appeared which was probably much less promising and influential than Antoine Meillet's (1866-1936) Introduction a I'etude comparative des langues indo-europeennes, which first appeared in 1903, was translated into German soon thereafter, and had six more editions during the author's lifetime. It is probable that most readers of the book used later editions (notably the fifth dating from 1922), by which time the Cours de linguistique generate had appeared and begun to exert its influence. It would help explain why 20th-century linguists soon began to associate the phrase with Saussure and the Cours, and not Meillet and his Introduction — or Saussure and his Memoire, for that matter. However, as we shall see, Meillet used this now well-known phrase much earlier in his career, a whole decade earlier, at least where the printed record is concerned. This is significant, as we shall see later in the present discussion. In 1893, when the quote below appeared, Saussure had returned to Geneva less than two years earlier, and Meillet, then aged 27, was eager to establish himself as a scholar in his own right. It was no surprise that the 4

In his bibliography (p.238) Mey refers to the 1916 edition, and so one may speculate which edition — or translation? — he might have been referring to. Neither in the first French edition (whose pagination differs from all subsequent editions) nor in any later edition or, rather, uncorrected reprint is there anything said on page 72 about this matter; there is not even a mention of 'systeme' in either. 5 The frequent ascription today of the concept of 'markedness' to Chomsky, and not to Trubetzkoy, is just a case in point, and only historians of linguistics may take exception to such misleading claims. For a well-researched corrective to this 'now swollen notion', see Chvany (1996).

26


article was devoted to an expose of the main pillar of historical-comparative linguistic theory and practice since 1876, the question of the regularity of sound change as emphasized by the young scholars of the Leipzig circle around Georg Curtius (1820-1885), notably Karl Brugmann (1849-1919) and Hermann Osthoff (1847-1909), precisely during the time Saussure was a student there (1876-1878, 1879-1880). In his paper on "Les lois phonetiques", Meillet noted — obviously well aware of the idea, first enunciated by Felix Franke (1860-1886), of Artikulationsbasis or modern-day 'articulatory setting': Les divers elements phonetiques de chaque idiome forment un systeme oil tout se tient. Les personnes qui ont appris a prononcer une langue etrangere ont pu s'en rendre compte: ce n'est pas seulement parce qu'il prononce mal le th ou les consonnes finales que le Fran9ais est inintelligible en parlant Panglais, c'est que ni la position des levres, ni celle de la langue ne sont les memes pour parler les deux langues, et que pas une seule des voyelles n'est rigoureusement identique dans les deux. Or 1'enfant, en apprenant a parler, s'assimile non une articulation isolee, mais {'ensemble du systeme. (Meillet 1893:318-319; Koerner 1989:405; emphasis added: EFKK) Two years later we can read the following in an important and influential 215-page monograph of the phenomenon of 'dissimilation' found in Indo-European languages, by another former student of Saussure's during his tenure as Maitre de conferences at the Ecole des Hautes Etudes in Paris (1881-1891), Maurice Grammont, where the author holds up the following principle maintained elsewhere by historical linguists at the time: [...] si la dissimilation elle aussi obeit a des lois, tout se tient dans Pedifice, 1'ensemble est complet et il ne reste qu'a parfaire les details. (Grammont 1895:10) As Meillet's 1893 article appeared in a rather obscure place, and not in a regular linguistics journal, it is not surprising that Toman (1987:404), looking for pre-1903 locations of the phrase, expected to have found reasons to attrribute it to Grammont rather than to Meillet, of whom he only found a later (1899) attestation which, however, was much more detailed and so could have been taken as an elaboration on Meillet's part of his confrere's discovery. The latter, in a paper published in the prestigious Indogermanische Forschungen, on a fairly minute subject of Indo-European morphology, had given the following argument to explain a particular language change: [...] tous les mouvements qui concourent a la formation d'un phoneme etant solidaires, 1'alteration de 1'un d'entre eux a chance d'entrainer, soit immediatement, soit plus tard, 1'alteration d'une ou de plusieurs des autres. Du reste ce phoneme n'est pas isole dans la langue, il fait partie d'un systeme dont toutes les parties se tiennent et reagissent les unes sur les autres; [...]. (Meillet 1899:64; Toman 1987:403) Apart from the fact that Meillet's statement is quite distinct from Grammont's, we may be safe in assuming, also based on Meillet's earlier use of the idea, that both had drawn inspiration from the same source, even if later readers might have seen Meillet as its originator, a possibility Toman (1987:404-405) is willing to consider. Thus, it is interesting to note that Joseph Vendryes (1875-1960), one of Meillet's first students and not much later a close collaborator, should have revisited in a way Meillet's 1893 paper. In a programmatic paper on the 'sound laws', published in a Festschrift dedicated to Meillet (on the occasion of what appears to have been the 10th anniversary of his teaching at the Ecole des Hautes Etudes), he noted that the change of one phoneme presupposes the change of other phonemes in the evolution of a given language system ("evolution du systeme"): "L'alteration d'un phoneme suppose 1'alteration concomittante de plusieurs autres phonemes" (Vendryes 1902:116). Even

Three Saussures?

27

if the 'ou tout se tient' phrase itself does not appear in Vendryes' paper, the underlying concept is clearly in evidence.

5.2 Getting closer to the sources of 'ou tout se tient' Given the now widely accepted ascription of the concept and phrase to Meillet, it becomes of interest to the historian of linguistics to look for clues of their genesis in Meillet's mature work, in particular his influential Introduction a I 'etude comparative des langues indo-europeennes, published in a German translation in 1909, a considerable feat given Germany's preeminence in the field of comparative-historical linguistics at the time.6 What we could have noted thus far is that nowhere in his writings did Meillet claim ownership of the phrase in question; there is no hint anywhere that he thought he was expressing a novel idea. Where his Introduction is concerned, we should note right from the start that the book carries, on a separate page, the following dedication: A MON MAITRE

FERDINAND DE SAUSSURE A L'OCCASION DES VINGT-CINQUE ANS ECOULES DEPUIS LA PUBLICATION DU MEMOIRE SUR LE SYSTEME PRIMITIF DES VOYELLES DANS LES LANGUES INDO-EUROPEENNES (1878-1903)

As we shall see from what follows, this dedication is much more than a polite gesture of a student to his former teacher; indeed, Meillet was very conscious of Saussure's influence on his linguistic thinking as may be gathered from his obituary of the mattre. Referring to his own contribution to a volume of melanges dedicated to Saussure in 1908, the thirtieth anniversary of the publication of the Memoire, in which his former students were given an opportunity to indicate "leur dette vis-a-vis de leur maitre", Meillet spoke of his own indebtedness in the following terms: Pour ma part, il n'est guere de page que j'ai publiee sans avoir un remords de m'en attribuer seul le merite: la pensee de F. de Saussure etait si riche, que j'en suis reste tout penetre. Je n'oserai, dans ce que j'ai ecrit, faire le depart de ce que je lui dois; mais je suis sur que Penseignement de F. de Saussure est pour beaucoup dans ce que les juges bienveillants ont parfois pu trouver a y louer. (Meillet 1936[1913]:179) A possible hyperbole aside, we can be sure that Meillet meant what he said. Characterizing Saussure's teachings at the Ecole des Hautes-Etudes, Meillet (p. 178) points out that what the student received was of "une valeur generate, ils preparaient a travailler et formalent 1'esprit; ses formules et ses definitions se fixaient dans la memoire comme des guides et des modeles." This statement should be kept in mind, too. Together with the dedication of his Introduction, which was retained in all subsequent editions, Meillet's acknowledgement of the profound impact that Saussure had on his training as a linguist is, in my view, significant for the resolution of our puzzle concerning the original source of the well-known phrase concerning

6 In fact, Meillet's reputation had grown so much by then that the University of Berlin bestowed on him an honorary doctorate in 1910.

28


the systematic nature of language, of which we find so many expressions in Meillet's writings. The following additional locus may suffice to illustrate this. In his 1906 inaugural speech as successor to Michel Breal (1832-1915) in the Chair of Comparative Grammar at the College de France, Meillet chose to pronounce himself on "L'etat actuel des etudes de linguistique generate [sic]". There, speaking about "la realite d'une langue", he expatiates: Elle est linguistique: car une langue constitue un systeme complexe de moyen d'expression, systeme oil tout se tient et ou une innovation individuelle ne peut que difficilement trouver place si, provenant d'un pur caprice, elle n'est pas exactement adaptee a ce systeme, c'est-adire si elle n'est pas en harmonic avec les regies generates de la langue. (Meillet 1921 [1906]: 16; emphasis added: EFKK). Returning to Meillet's Introduction of 1903, we may read in his Avant-propos the following observation which, I believe, should lead us closer to the source of the idea: Comme toute autre langue, les differents parties du systeme linguistique indo-europeen forment un ensemble ou tout se tient et dont il importe avant tout a comprendre le rigoureux enchainement. (Meillet 1903:x = 81937:ix) Indeed, if a further piece of evidence was needed to demonstrate that Meillet had not merely comparative-historical Indo-European linguistics generally in mind but, specifically, Saussure's Memoire, when he spoke of a language as a system of interdependent terms, we may refer to the article "Aper9u du developpement de la grammaire comparee", which Meillet added to the third edition of his Introduction in 1912, barely a year before Saussure's death. There, after a detailed expose of the discoveries made by Saussure in his Memoire (Meillet 1937[1912]:473474), which established "d'une maniere definitive la theorie du vocalisme indo-europeen" (p.473), Meillet conies to the following judgment: [...] le Memoire apportait, par une innovation, un systeme coherent qui embrassait tous les faits, mettait a leur place les faits connus et en revelait une foule de nouveaux. Des lors il n'etait plus permis d'ignorer, a propos d'aucune question, que chaque langue forme un systeme ou tout se tient, et un plan general d'une pleine rigueur. (1937[1912]:474-475; emphasis added: EFKK) As early as 1915, a year before the first appearance of the essentials from Saussure's Geneva lectures on general linguistics, Meillet, in a presentation of the state of the art of French linguistics for the San Francisco World Exhibition, emphatically reiterates this formulation, and clearly in connection with Saussure's teachings in Paris, when affirming "Toute langue est un systeme rigoureusement articule [...], tout se tient dans le systeme d'une langue" (Meillet 1916[1915]:7; also quoted in Koerner 1973:240n.23).

6. THE PLACE OF 'SYSTEM' IN SAUSSURE'S LINGUISTIC THOUGHT7 In his review of the Cours de linguistique generale, Meillet had noted the following, which should be kept in mind when pursuing our quest for the source or sources of this well-known phrase:

7 Given time restrictions, this section lacks the kind of detail that I believe an article defond requires to settle the issue once for all, though I trust the drift of my argument is clear enough to make the point.

Three Saussures?

29

[...] la pensee de F. de Saussure s'etait fixee tres tot, on le sait. Les doctrines qu'il a enseignees explicitement dans ces cours de linguistique generate sont celles dont s'inspirait deja 1'enseignement de grammaire comparee qu'il a donne vingt ans plus tot a 1'Ecole des Hautes Etudes, et que j'ai re9u. Je les retrouve telles qu'il etait souvent possible de les deviner. (Meillet 1916:32; Mounin 1968:163) This observation is of importance to my argument as it suggests that Meillet picked up the idea and probably also the felicitous phrase during Saussure's courses in Paris during the mid1880s. It was during this time that Saussure was very much preoccupied with Indo-European linguistics, which was the subject of his courses, but also with defending the findings of his Memoire, which had not yet become widely accepted and had been, by some in Germany, even attacked. Saussure appears to have used the Memoire as a textbook; indeed, a second edition of the volume appeared in Paris in 1887. From Saussure's papers it is clear that he intended to add an answer to his critics (notably Osthoff), but was dissuaded by his colleagues from doing so (cf. Redard 1978:34). It is also important to note that Saussure referred to the Memoire as 'mon Systeme des Voyelles', the first paragraph of which should be quoted in full to illustrate the clear-headedness of this twenty-one-year old genius: Etudier les formes multiples sous lesquelles se manifeste ce qu'on appelle Va indo-europeen, tel est 1'objet immediat de cet opuscule [of 302 pages!]: le reste des voyelles ne sera pris en consideration qu'autant que les phenomenes relatifs a 1'a en fourniront 1'occasion. Mais si, arrive au bout du champ ainsi circonscrit, le tableau du vocalisme indo-europeen s'est modifie peu a peu sous nos yeux et que nous le voyions se grouper tout entier autour de 1'a, prendre vis-avis de lui une attitude nouvelle, il est clair qu'en fait c'est le systeme des voyelles dans son ensemble qui sera entre dans le rayon de notre observation et dont le nom doit etre inscrit a la premiere page. (Saussure 1879[1878]:1; emphasis added: EFKK) There can be no doubt that Saussure's entire argument is based on the idea of 'system' and the interrelationship between the parts. No surprise when in the final paragraph of his Memoire Saussure speaks of "structure considered en elle-meme" (p.283). Indeed, the idea of language as a system may be regarded as the fil conducteur of Saussure's entire work in matters linguistic. Already in his enfantillage (Saussure's term in retrospect) of 1872, when he was less than fifteen years old, the schoolboy wrote a 40-page "Essai pour reduire les mots du Grec, du Latin & de 1'Allemand a un petit nombre de racines" which can be seen as foreshadowing Saussure's later preoccupation with discovering or building on the assumption of the presence of 'system' in language.8 In the Cours, Mounin (1968:61) reports, the term systeme shows up 138 times, and if we add Saussure's use of 'organisme grammatical' and similar expressions which he employed as synonyms, we'd probably arrive at an occurrence of the concept on every second page.

7. CONCLUDING REMARKS In his 1990 paper, "Un systeme ou tout se tient: Origin and evolution of the idea", John Hewson quoted the following remark by the late Jean Stefanini (1917-1985): "L'idee que la

8 In fact, in his reminiscences of 1903 or thereabouts, Saussure (1960:17) reports: "La marotte linguistique me travaillait evidemment des cette epoque, car je n'eus pas plus tot appris quelques rudiments de grec a 1'ecole, que je me sentis mur pour esquisser un systeme general du langage" (emphasis in the original).

30


langue est un systeme a toujours ete consideree par toute 1'ecole de Paris, de Meillet a Benveniste, comme 1'apport fondamentale de Saussure" (Stefanini 1973:322), elaborating on this observation in the following manner (pp.787-788):

The notion of system is indeed fundamental to all of Saussure's work, both synchronic and diachronic; in fact, as Guillaume points out, it is fundamental to the whole comparativist enterprise (1986:10). The centrality of the notion of system in Saussure's work is to be found, in fact, as early as his 1879 Memoire sur les voyelles: in that work he proposed a system of three laryngeals entirely on regular correspondence in the Indo-European languages. While we may regard this statement as un peu rapide, it summarizes fairly adequately what I tried to expatiate on in this paper, namely, that there is a fairly straight line from Saussure's early comparativist work down to his last lecture on general linguistic theory in the summer of 1911. There are definitely not two Saussures, one of the Memoire, another of the Cours, and not only with regard to the notion of 'systeme' which is so central to his linguistic theory. That he struggled with it well into his series of lectures on general linguistics may be gathered from what he responded to one of his most assiduous students, Albert Riedlinger (1883-1978), who had asked him pointed questions about the nature of a synchronic approach to language on 19 January 1909: La langue est un systeme serre, et la theorie doit etre un systeme aussi serre que la langue. La est le point difficile, car ce n'est rien de poser a la suite 1'une de 1'autre des affirmations, des vues sur la langue; le tout est de les coordonner en un systeme. (Godel 1957:29) Indeed, what Wilhelm Streitberg (1864-1925), the Indo-Europeanist, said about Saussure's Memoire might well be applied to his entire work, maybe not always in its eventual execution but definitely in its intent. After having pointed out that in the linguistics of the later 19th century Saussure's work played a very special role, Streitberg (1966 [1915]: 100) notes that his fame lay not in individual discoveries but that seine wahre, seine einzigartige Bedeutung liegt in der systembildenden Kraft seines Geistes. Seine unvergleichliche Starke ist die Synthese; alle Einzelbeobachtungen sind ihm nur Bausteine zu dem planvoll gefugten Gebaude des Systems; er ruht und rastet nicht, bis sich alle Tatsachen, aus ihrer Vereinzelung erlost, zu einem harmonischen Granzen zusammenschlieBen. (101) [... his true, his unique importance is to be found in the power of his mind to form systems. His incomparable strength lies in synthesis; all individual observations are nothing but bricks for the well-thought-out edifice of the system; he does not rest until all facts, taken out of their isolation, are brought together into a harmonious whole.] In view of what has been said above, it must be seen as a sad irony that Saussure should have become known as 'the father of structuralism', given that the word 'structure' is never used as a technical term in any of Saussure's writings; it rarely occurs either in the Cours or the Memoire, which might be regarded as 'structuralist' in the sense of some modern schools of thought, I am not sure which. The irony is reinforced if one considers the fact that Saussure made every effort in his teachings to leave behind the natural-science legacy of 19th century linguistics and its organicist structuralism (Koerner 1973, 1995), and argued for a view of the field that would make linguistics the centre of socio-semiotic study. One cannot but deplore the fact therefore that Roy Harris, in an effort to replace the previous English translation of the Cours (Saussure 1959) by his own rendering of the vulgata text (Saussure 1983), should have

Three Saussures?

31

decided (without alerting the user of this translation to his choice) to replace the term systeme in the original text by 'structure', as if both Hjelmslev and Firth had not taught us in their writings of more than fifty years ago the essential difference between 'structure' and 'system', the first pertaining to syntagmatic ordering, the second to paradigmatic organization!

SELECT BIBLIOGRAPHY Bally, Charles (1932). Linguistique generate et linguistique francaise. E. Leroux, Paris (2nd ed., A. Francke, Berne, 1944.) Bloomfield, Leonard (1923). Review of Saussure (1916; 21922). Modern Language Journal 8.317-319. (Repr. in Cahiers Ferdinandde Saussure 21.133-135 [1964].) Brogyanyi, Bela (1983). A Few Remarks on the Origin of the Phrase 'ou tout se tient'. Historiographia Linguistica 10.143-147. Buyssens, Eric (1942). Les six linguistiques de F. de Saussure. Revue des Langues Vivantes 11.15-23, 46-55. (Separate publication, M. Didier, Brussels, 1942.) (1961). L'origine de la linguistique synchronique de Saussure. Cahiers Ferdinand de Saussure 18.17-33. Chvany, Catherine V. (1996) [1993]. The Evolution of the Concept of Markedness from the Prague Circle to Generative Grammar. Selected Essays by Catherine V. Chvany ed. by Olga T. Yokoyama & Emily Klenin, 234-241. Slavica Publishers, Columbus, Ohio. Collinge, N. E. (1985). The Laws of Indo-European. John Benjamins, Amsterdam & Philadelphia. Fishman, Joshua A. (1970). Sociolinguistics: A brief introduction. Newbury House, Rowley, Mass. Godel, Robert (1957). Les sources manuscrites du Cours de linguistique generale de F. de Saussure. Droz, GenevaGrammont, Maurice (1895). La dissimilation consonantique dans les langues indo-europeennes et dans les langues romanes. Imprimerie Darantiere, Dijon. (1933). Traite de phonetique. Delagrave, Paris. Guillaume, Gustave (1986) [1945/46]. Esquisse d'une grammaire de la langue francaise IV. Publiees sous la direction de Roch Valin, Walter Hirtle & Andre Joly. (= Lecons de linguistique, 7.) Presses de 1'Universite Laval, Quebec. Harris, Roy (1994). Criticizing Saussure. Semiotica 98:1/2.181-186. [Review of Holdcroft (1991).] Haugen, Einar (1951). Directions in Modern Linguistics. Language 27.211-222. Hewson, John (1990). Un systeme ou tout se tient: Origin and evolution of the idea. History and Historiography of Linguistics: Proceedings of the Fourth International Conference on the History of the Language Sciences (ICHoLS IV), Trier, 24-27 August 1987 ed. by Hans-Josef Niederehe & Konrad Koerner, vol.11, 787-794. John Benjamins, Amsterdam & Philadelphia. Holdcroft, David (1991). Saussure: Signs, systems and arbitrariness. Cambridge University Press, Cambridge. Koerner, E. F. K[onrad] (1973) [1971]. Ferdinand de Saussure: Origin and development of his linguistic thought in Western studies of language. A contribution to the history and theory of linguistics. Friedrich Vieweg & Sohn, Braunschweig.

32


(1975) [1972]. European Structuralism — early beginnings. Current Trends in Linguistics ed. by Thomas A. Sebeok, vol.XIII: Historiography of Linguistics, 717-827. Mouton, The Hague. (1984). French Influences on Saussure. Canadian Journal of Linguistics / Revue canadienne de Linguistique 29.20-41. (Repr. in Koerner 1988.67-88.) (1987). The Place of Saussure's Memoire in the Development of Historical Linguistics. Festschrift for Henry M. Hoenigswald ed. by George Cardona & Norman Zide, 201-217. GunterNarr, Tubingen. - (1988). Saussurean Studies /Etudes saussuriennes. Avant-propos de Rudolf Engler. Editions Slatkine, Geneva. (1989). Practising Linguistic Historiography. John Benjamins, Amsterdam & Philadelphia. - (1995). The Natural Science Model and Linguistic Theory in the 19th and 20th Centuries. Sapostavitelno ezikoznanie / Contrastive Linguistics 19:2.41-59. (Revised version published in Professing Linguistic Historiography by E.F.K. Koerner, 47-76. John Benjamins, Amsterdam & Philadelphia, 1995.) Lepschy, Giulio C. (1970). A Survey of Structural Linguistics. Faber & Faber, London. (2nd enl. ed., Andre Deutsch, London, 1982.) Littre, Emile (1869). Dictionnaire de la langue francaise. Vol.11, part 2: Q-Z. Hachette, Paris. Meillet, Antoine. 1893. Les lois du langage I: Les lois phonetiques. Revue Internationale de sociologie (Paris) 1.311-321. (1899). A propos du groupe -ns. Indogermanische Forschungen 10.61-70. (1903). Introduction a I'etude comparative des langues indo-europeennes. Hachette, Paris. (5th ed., 1922; 8th ed., with a foreword by Emile Benveniste, 1937. Repr., University of Alabama Press, University, Ala., 1964.) (1905). Comment les mots changent de sens. Annee sociologique 9.1-38. (Repr. in Meillet 1921:230-271.) (1906). L'etat actuel des etudes de linguistique generale: Lecon d'ouverture du Cours de Grammaire comparee au College de France lue le mardi 13 fevrier 1906. Revue des Idees (Paris) 3.296-308. (Repr. in Meillet 1921:1-18.) (1909). Einfuhrung in die vergleichende Grammatik der indogermanischen Sprachen. Authorized transl. by Wilhelm Printz. B. G. Teubner, Leipzig & Berlin. (1912). Apercu du developpement de la grammaire comparee. Appended to the third ed. of Meillet, Introduction a I'etude comparative des langues indo-europeennes, 439-473. Hachette, Paris. (= 453-483 in 8th ed. of 1937.) (1913). Ferdinand de Saussure. Annuaire de I'Ecole Pratique des Hautes Etudes (Section des sciences historiques et philologiques) 1913-1914, 115-125. (Repr. in Meillet 1936.174-183.) (1915). La Linguistique. La Science francaise: Ouvrage publie sous les auspices du Ministere de I'Instruction Publique a I'occasion de I'Exposition Universelle de San Francisco, vol.11, 117-124. Ministere de 1'Instruction Publique et des Beaux Arts, Paris. (Separate publication, with a portrait of Saussure, Larousse, Paris, 1916.) (1916). Review of F. de Saussure, Cours de linguistique generale (Payot, Lausanne & Paris, 1916). Bulletin de la Societe de Linguistique de Paris 20.32-36. (Repr. in Mounin 1968.161-168.)

Three Saussures?

33

(1921). Linguistique historique et linguistique generate. [Tome I.] H. Champion, Paris. (1936). Linguistique historique et linguistique generate. Tome II. C. Klincksieck, Paris. Mey, Jacob (1992). The Pragmatics of Semiosis. Current Advances in Semantic Theory ed. by Maxim Stamenov, 219-238. John Benjamins, Amsterdam & Philadelphia. Moulton, William G. (1971). Review of Rudolf Meyer, Zur Morphologie und Sprachgeographie des Schweizerdeutschen (Huber, Frauenfeld, 1967) [and three other monographs devoted to Swiss German dialectology]. Language 47:4.938-943. Mounin, Georges (1968). Saussure ou le structuraliste sans le savoir: Presentation, choix de textes, bibliographie. Editions Seghers, Paris. (2nd corrected ed., 1971.) — (1972). La linguistique du XXe siecle. Presses Universitaires de France, Paris. Mulder, Jan W. F. (1989). Foundations of Axiomatic Linguistics. Mouton de Gruyter, Berlin & New York. Peeters, Bert (1990). Encore une fois 'ou tout se tient'. Historiographia Linguistica 17:3.427-436. Redard, Georges (1978). Deux Saussure?. Cahiers Ferdinand de Saussure 32.27-41. Saussure, Ferdinand de (1879) [1878]. Memoire sur le systeme primitif des voyelles dans les langues indo-europeennes. B. G. Teubner, Leipzig. (Repr., Friedrich Vieweg, Paris, 1887.) (1916). Cours de linguistique generale. Ed. by Charles Bally & Albert Sechehaye, with the collaboration of Albert Riedlinger. Payot, Lausanne & Paris. (2nd ed., Payot, Paris, 1922; 3rd and last corrected ed., 1931.) (1959). Course in General Linguistics. Transl. by Wade Baskin. Philosophical Library, New York & London. (2nd ed., McGraw-Hill, New York-Toronto-London, 1966.) (1960) [c.1903]. Souvenirs de F. de Saussure de concernant sa jeunesse et ses etudes. Ed. by Robert Godel. Cahiers Ferdinand de Saussure 17.3-14. (1978) [1872]. Essai pour reduire les mots du Grec, du Latin & de 1'Allemand a un petit nombre de racines. Ed. from Ms. by Boyd H. Davis. Cahiers Ferdinand de Saussure 32.77-101. (1967-1968; 1974). Cours de linguistique generale. Edition critique par Rudolf Engler. 4 fasc. Otto Harrassowitz, Wiesbaden. (1983). Course in General Linguistics. Transl. by Roy Harris. Duckworth, London; Open Court, La Salle, 111., 1986. Sechehaye, Albert (1940). Les trois linguistiques saussuriennes. Vox Romanica 5.1-48. (1944). Review of Buyssens (1942). Cahiers Ferdinand de Saussure 4.65-69. Stefanini, Jean (1973). Sur la conception guillaumienne de 1'opposition langue/discours. Travaux de Linguistique et de Litterature 11.319-328. Streitberg, Wilhelm (1915) [1914]. Ferdinand de Saussure. Indogermanisches Jahrbuch 2.203213. (Repr. in Portraits of Linguists ed. by Thomas A. Sebeok, vol.1, 100-110. Indiana University Press, Bloomington, 1966.) Szemerenyi, Oswald (1973). La theorie des laryngales de Saussure a Kurytowicz et a Benveniste: Essai de reevaluation. Bulletin de la Societe de linguistique de Paris 68.1-25. (1980). About Unrewriting the History of Linguistics. Wege der Universalienforschung: Sprachwissenschaftliche Beitrdge zum 60. Geburtstag von Hansjakob Seiler ed. by Gunter Brettschneider & Christian Lehmann, 151-162. Gunter Narr, Tubingen.

34


Toman, Jindrich (1987). Not from 1903, not from Meillet: A final (?) remark on 'ou tout se tient'. Historiographia Linguistica 14:3.403-406. Trubetzkoy, Nikolaj S. (1933). La phonologie actuelle. Journal de Psychologie normale et pathologique 30.227-243. (Repr. in Essais sur le langage presentes par Jean-Claude Pariente, 143-164. Editions de Minuit, Paris, 1969.) Vendryes, Joseph (1902). Reflexions sur les lois phonetiques. Melanges linguistiques offerts a M. Antoine Meillet, 115-131. C. Klincksieck, Paris. (Repr. in J. Vendryes, Choix d'etudes linguistique et celtiques, 3-17. C. Klincksieck, Paris, 1952.) Wartburg, Walter von (1966). Franzosisches etymologisches Worterbuch. Vol.XII: Sk-S. Zbinden, Zurich.

2

GEGEN EINE HERKOMMLICHE INTERPRETATION VON SAUSSURES LANGAGE, LANGUE UND PAROLE - EIN MERKZETTEL Ralph A. Hartmann, Department of German, University of St. Andrews, UK

Abstract: In this paper the still common textbook-interpretation as far as the social dimension of language to be found within Saussure's triad of LANGAGE, LANGUE and PAROLE is challenged. It will be shown that LANGUE must not be interpreted as only social in its character as well as that PAROLE must not be seen as only individualistic in its character. A more careful consideration of the sources to Saussure's Cours de linguistique generale could even lead to an idea which proposes a more "pragmatistic" Saussure who might have had similar thoughts about language as for example Wittgenstein (in his Philosophical Investigations) and Searle (in his theory of speech acts).

Im vorliegenden Beitrag soll nochmals auf eine Schwierigkeit eingegangen werden, die mit der ("getreuen") Exegese des Cours de linguistique generale von Ferdinand de Saussure zu tun hat. Da Albert Sechehaye und Charles Bally als Herausgeber des Cours lediglich auf Vorlesungsmitschriften von einigen Saussure-Studenten zuriickgreifen konnten, ist es verstandlich, daB Aspekte des Buches durch subjektive Interpretationen der beiden "Rekonstrukteure" des Saussureschen Gedankenguts gekennzeichnet sind, daB somit keineswegs sichergestellt ist, ob Saussure den Cours, wie wir ihn heute kennen, fur eine Publikation unter seinem eigenen Namen zugelassen hatte (vgl. z.B. Buyssens 1952: 50). So spricht etwa Godel (1957: 98) davon, daB Bally und Sechehaye teilweise im Cours Fragen beantwortet hatten, die Saussure lediglich (in seinen Vorlesungen bzw. in den anderen Quellen) gestellt hatte. Nicht zuletzt deshalb ist eine "unmittelbare" Lekture des "Saussureschen Cours" nicht ausreichend und konnte unter Umstanden zu Fehleinschatzungen fuhren, wenn es darum geht, die vielfaltigen Facetten der Sprachtheorie des Genfers zu beleuchten. Ohnehin wird dieser Sprachtheorie eine herkommliche Interpretation (etwa in Lehrbuchern wie beispielsweise Bartschat 1996), die die LANGUE Saussures als ausschlieBlich sozialen sowie die PAROLE als uneingeschrankt individuellen Aspekt der Sprache wiedergibt, keinesfalls gerecht. Fur (den) Saussure (des Cours), wie ihn eine solche Tradition sieht, ist sowieso nur die LANGUE alleiniges Untersuchungsobjekt der Sprachwissenschaft:

35

36

Langue and Parole in Synchrony and Diachrony "[...] die Sprache [LANGUE, R.H.] an und fur sich selbst betrachtet ist der einzige wirkliche Gegenstand der Sprachwissenschaft." (Saussure 1967: 279)

Diese "Restriktion" des Untersuchungsgegenstandes der Sprachwissenschaft ausschliefilich auf die LANGUE, wie wir sie im SchluBsatz des Cours vorfmden, ist nun aber genau der Punkt, an dem wir am ehesten AnlaB haben, einen "nicht-authentischen" von einem "originalen / originaren" Saussure zu unterscheiden. Bei obigem Satz handelt es sich, wie in der Forschungsliteratur allgemein anerkannt ist (vgl. Godel 1957, Koerner 1973, Jager 1976), um einen "programmatisch bedeutsamen Eingriff (Scheerer 1980: 72) der Herausgeber des Cours in die tatsachliche Lehre Saussures. Ein derartig gravierender Editionsfehler bringt natiirlich beziiglich der Tradierung Saussureschen Gedankenguts auBerst negative Konsequenzen mit sich, die beispielsweise von De Mauro (1967: 452, Anm. 305) in dessen italienischer Ausgabe des Cours wohl am drastischsten skizziert wurden. Obwohl die Einschatzung De Mauros doch etwas ubertrieben erscheint, veranschaulicht sie dennoch, daB eine quellenkritische Lektiire des Cours von vornherein vermutlich viele MiBverstandnisse bzw. falschverstandene "Weiterfiihrungen", wie sie eben immer noch in aktuellen Lehrbuchern vermittelt werden (s. etwa den kritiklos iibernommenen SchluBsatz des Cours bei Bartschat 1996: 59), vermieden hatte. Im folgenden steht nun der Versuch im Vordergrund, anhand der "Quellen" des Cours1 (s. insbesondere Godel 1957, Saussure 1989-90) zu einer etwas "differenzierteren" Dreiteilung der Sprache zu gelangen, mit Hilfe derer sich einerseits die soziale Dimension der Sprache im Saussureschen Sinne verdeutlichen diirfte, durch die andererseits aber zusatzlich Ankniipfungspunkte zu sprachphilosophischen Sichtweisen ("spater" Wittgenstein, Searle) entstehen. Des weiteren wird damit ebenfalls das Problem des Zusammenhangs von LANGUE und PAROLE in der Saussureschen Sprachtheorie ein wenig erhellt, das bei der vorlaufigen "direkten" Lesart im Unklaren bleiben muB: Die ausschlieBliche Fokussierung auf die LANGUE als Gegenstand der Sprachwissenschaft (im Cours) scheint namlich die Frage eindeutig zu negieren, ob die PAROLE eventuell (beispielsweise als "Datengrundlage") in der LANGUE verankert sein konnte. So wird es sich zeigen, daB Saussure wohl keine so starke Trennlinie zwischen LANGUE und PAROLE gezogen hat, wie sie die "direkte" CoursInterpretation nahelegt. Saussure scheint nun, wie die Quellen es ausweisen, seine Vorlesung iiber die Grundfragen der allgemeinen Sprachwissenschaft in drei Abschnitte eingeteilt zu haben. Wie die Mitschriften der Studenten belegen, waren dies 1.) die Sprachen (les langues), 2.) die Sprache (la langue), 3.) Sprachfahigkeit und Sprach"ausubung" bei den Individuen (faculte et exercice du langage chez les individus). Allein schon diese generelle Aufteilung deutet gewissermaBen eine Divergenz zwischen dem Cours und Saussures tatsachlicher Auffassung an, zumal dort nicht explizit die Rede von den Einzelbegriffen LANGAGE und PAROLE ist. 1

Diese Quellen und Materialien sind in Rudolf Englers kritischer Edition des Cours (damit ist Saussure 1989-90 gemeint; s. Literaturverzeichnis) in sechs nebeneinander gestellten Spalten aufgeftihrt. Spatere Verweise auf diese Ausgabe beinhalten zusatzlich zu den Seitenzahlen den Namen desjenigen, aus dessen Mitschriften die betreffenden Passagen zitiert sind; fur die personlichen Notizen Saussures steht das Ktirzel "NS". Bsp.: "Saussure 1989-90: 40, Joseph" bezieht sich auf die Mitschrift von Francis Joseph auf S. 40 der kritischen Edition; "Saussure 1989-90: 436, NS" verweist auf die personlichen Notizen Saussures auf S. 436 der kritischen Edition; "Saussure 1989-90: 17, NS (A)" bezieht sich auf S. 17 des zweiten Bandes der kritischen Edition. Dieser zweite Band besteht aus einem Appendix (A) mit weiteren erganzenden personlichen Notizen Saussures.

Langage, Langue und Parole - quellengetreu gesehen

37

Zunachst scheint Saussure auch gar nicht zwischen LANGUE und LANGAGE zu trennen, wie Constantins Mitschrift nahelegt: "Sans separer immediatement les mots de langue et de langage, ou trouvons-nous le phenomena concret, complet, integral de la langue ou [Hervorhebung R.H.] du langage? C'est a dire, ou trouvons-nous 1'objet devant lequel nous avons a nous placer?" (Saussure 1989-90: 24, Constantin) Die Suche nach dem Untersuchungsgegenstand der Sprachwissenschaft beginnt also fur Saussure gewissermaBen mit einer Art Ratlosigkeit. Sprache, ob wir sie nun LANGUE oder LANGAGE nennen, 1st ein zu komplexes Phanomen, um ohne weiteres festlegen zu konnen, was bzw. welche dieser vielen Komplexitaten zu untersuchen ist/sind. Die Komplexitat der Sprache (LANGUE bzw. LANGAGE) erlaubt es demnach nicht, von einer "einheitlichen Sache" zu sprechen, die von verschiedenen Standpunkten aus betrachtet werden kann. Vielmehr sind es gemaB Saussure (1989-90: 26, NS) die verschiedenen Standpunkte, die die "Sache" erschaffen. Diese "Unsicherheit", wenn man es so nennen kann, setzt sich bei Saussure auch tatsachlich in der Einteilung der Sprache in verschiedene Bereiche fort. Wie oben schon angedeutet wurde, scheint Saussure zu Beginn nicht zwischen LANGUE und LANGAGE zu trennen. Auch die Herausgeber des Cours verwenden den Ausdruck LANGAGE, wenn in den entsprechenden Stellen der Quellen von LANGUE die Rede ist. So heiBt es beispielsweise im Cours einmal, daB die LANGAGE eine individuelle und eine soziale Seite habe (s. Saussure 1967: 10). Andererseits belegen die entsprechenden Quellen, daft Saussure in seinen Vorlesungen eindeutig die LANGUE primar durch zwei sich "korrespondierende" Seiten charakterisierte: "Dans la langue [Hervorhebung R.H.], il y a done toujours un double cote [sic] qui se correspond: elle est . .. ..—rp Si on considere done la sphere ou la langue vit, il y aura toujours la langue individuelle et la langue sociale." (Saussure 1989-90: 28, Riedlinger) Den Zusammenhang zwischen der sozialen und individuellen Seite der LANGUE erkennt Saussure nun darin, daB die Formen bzw. die Grammatik nur in einer "sozialen Weise existieren" konnen. Das bedeutet, daB sie nicht anders als in einer Sprachgemeinschaft, die sie bzw. ihren Gebrauch sanktioniert hat, existieren konnen. Hier ist also mit LANGUE das abstrakte Zeichensystem gemeint, das seine Geltung lediglich im Gebrauch, den die Sprachgemeinschaft davon macht, erhalten kann. Hingegen geht dazu im Gegensatz gemafi Saussure die Veranderung, also der Sprachwandel, eindeutig vom Individuum aus: Ein Individuum verwendet irgendwann einmal ein im Unterschied zum Zeichengebrauch der restlichen Sprachgemeinschaft "verandertes" sprachliches Zeichen (evtl. sogar ein "verandertes" Syntagma, das aus der Aneinanderreihung mehrerer einzelner sprachlicher Zeichen bestehen kann); im Laufe der Zeit finden sich weitere Individuen, die diesen veranderten Gebrauch eines Zeichens akzeptieren, bis der Wandel schliefilich von der gesamten Sprachgemeinschaft anerkannt bzw. sanktioniert ist. Somit besitzt die LANGUE als abstraktes System von sprachlichen Zeichen "soziale" Geltung in einer Sprachgemeinschaft, kann jedoch nur infolge eines "individuellen" AnstoBes Veranderung erfahren. Das Individuum besitzt aber nach Saussure nicht allein ein Potential, den AnstoB zum Wandel des Zeichensystems zu geben, sondern ebenfalls eine Fahigkeit, die er mit dem

38


Ausdruck Faculte de langage articule bezeichnet, eine Fahigkeit, die uns durch die Sprechorgane gegeben ist. Fur Saussure kann diese Fahigkeit jedoch nicht ohne das von "auBen" kommende, sanktionierte Zeichensystem der LANGUE ausgeiibt (oder: "aktiviert") werden. Den Zusammenhang zwischen LANGAGE, LANGUE und PAROLE exemplifiziert Saussure (1989-90: 41) anhand eines Schaubildes, welches nahelegt, daB der Phanomenbereich der LANGAGE einfach zu gleichen Teilen aus LANGUE und PAROLE besteht. Erlauternd zu diesem Schaubild spricht Saussure von einem "Instinkt", von einer "naturlichen Funktion" der LANGAGE. Zudem charakterisiert er die LANGUE als etwas Passives, das sich in der "Kollektivitat befmdet", als "sozialen Code", der quasi die LANGAGE "organisiert" und das notwendige "Werkzeug" fur die Ausiibung der Fahigkeit zu LANGAGE darstellt. Wenn dementsprechend die LANGUE eine soziale und individuelle Seite besitzt, so gilt das ebenso fur die LANGAGE, die ja sozusagen sowohl die LANGUE als auch die PAROLE umfafit. Wie sieht nun aber das Verhaltnis zwischen LANGUE und PAROLE aus, und woraus besteht die PAROLE im Unterschied zur LANGUE? Die einfachste Charakterisierung der PAROLE ware, daB sie mit den tatsachlichen Hor- und Sprechakten zu identifizieren ist, wie es die herkommliche Sichtweise vorschreibt. Doch ganz so einfach macht es sich Saussure nicht: Auch die PAROLE zeigt sich ihm als individuelles und gleichzeitig soziales Phanomen. Da die LANGUE ein soziales Produkt der Fahigkeit zu LANGAGE ist sowie die nachfolgende "Representation" der sprachlichen Fakten im Individuum bildet, einen LANGUE-"tresor", wie sich Saussure (1989-90: 383, Riedlinger) ausdriickt, muB man auch daran denken, daB das, was in die LANGUE eintritt, zuvor in der PAROLE, also sozusagen im konkret (Aus-) Gesprochenen, versucht wurde: "La langue n'est que la consecration de ce qui avait ete evoque par la parole." (Saussure 1989-90: 383, Riedlinger) Insofern, wenn man von einem "Datenlieferanten" einer Wissenschaft von der LANGUE sprechen wollte, ware eine solche Datengrundlage die PAROLE (vgl. hierzu auch Buyssens 1942: 17). Und naturlich besitzt die PAROLE ebenfalls eine soziale sowie individuelle Seite. So kommen die konkreten Sprachaufierungen fast ausschlieBlich im sozialen Kontakt zustande. Allerdings gehen sie individuell von der im (individuellen) Gehirn reprasentierten LANGUE aus. Das bedeutet: PAROLE ist die soziale Datengrundlage der LANGUE;2 die LANGUE hingegen liefert individuell wiederum die Grundlage fur neue sprachliche AuBerungen, bildet auf diese Weise eben auch die individuelle Basis der PAROLE, die das Individuum im konkreten SprechanlaB zum Ausdruck bringt. 2

Eine derartige Sichtweise bestatigt Saussure selbst beispielsweise in seiner Erorterung beziiglich der Schwierigkeiten, die bei der Abgrenzung bzw. "Segmentierung" sprachlicher Einheiten untereinander entstehen: "Pour faire cette delimitation [= Abgrenzung bzw. Segmentierung sprachlicher Einheiten; R.H.], nous pouvons accorder que la meilleure methode c'est de prendre la parole. La parole ne figure ici que comme un document de langue. En effet, les casiers existant a 1'interieur de notre cerveau, nous ne pouvons les explorer. Obliges d'employer un moyen exterieur dans la parole [Hervorhebung R.H.]." (Saussure 1989-90: 236, Constantin). Das impliziert, daB Ruckschliisse auf die LANGUE iiberhaupt nicht anders moglich sind als in Folge von Beobachtungen der PAROLE: Die PAROLE dokumentiert durch ihre (externen) Daten den (internen) Zustand der LANGUE, die so erst klassifiziert werden kann.


39

Somit erfahrt die (oberflachlich als solche erscheinende) Dreiteilung der Sprache in LANGAGE, LANGUE und PAROLE beim "authentischen" Saussure eine - man mochte sagen: dem Gegenstand adaquate - Differenzierung hinsichtlich der individuellen und sozialen Seite. Das betrifft insbesondere LANGUE und PAROLE, wobei unter Berucksichtigung des o.e. Schaubildes auch LANGAGE als die Einheit von LANGUE und PAROLE logischerweise diese "Doppelseitigkeit" aufweist: "[...] (le langage est social, il est vrai, mais pour nombre de faits, il est plus commode de le rencontrer dans 1'individu)." (Saussure 1989-90: 383, Riedlinger) Infolgedessen konnen wir generell festhalten, daB Saussure in seiner Charakterisierung der Sprache mit ihren in sich selbst dichotomischen Teilbereichen LANGAGE, LANGUE und PAROLE bin- und hergerissen ist, wenn es um die Frage geht, ob die individuelle oder soziale Komponente des immens komplexen Phanomens Sprache hervorgehoben werden soil. Eine direkte bzw. herkommliche Lesart des Cours mag nun den sozialen Aspekt betonen: Die LANGUE als eindeutig soziale Seite der Sprache wird hier zum Mittelpunkt linguistischer Untersuchung erkoren. Eine quellenkritische Lesart deutet hingegen an, daB sich Saussure die Sache nicht ganz so einfach gemacht hat. Man konnte sogar meinen, daB er sich in dieser Hinsicht sehr unsicher war. Auf jeden Fall bezieht sich die soziale Dimension der Sprache gemafi der oben dargestellten differenzierteren Dreiteilung der Sprache bei Saussure sowohl auf einen Systemcharakter als auch auf einen Handlungscharakter: Eine Sprachgemeinschaft macht Gebrauch von einem fur sie geltenden System von Zeichen (i.e. soziale und individuelle LANGUE); sie handelt gemaB den (unbewuBten) Regeln des Gebrauchs dieses geltenden Systems von Zeichen. Das System Sprache erfahrt mit der Zeit viele kleine Anderungen, die sich auf den Gebrauch einzelner sprachlicher Zeichen bzw. auf die (sprachlichen) Handlungen der diese Zeichen gebrauchenden Sprecher bezieht. Eine derartige Interpretation Saussurescher Gedanken weist betrachtliche Parallelen zu den spateren Ansichten des Sprachphilosophen Ludwig Wittgenstein (1889-1951) auf, die insbesondere in dessen posthum (1958) veroffentlichtem Werk Philosophische Untersuchungen zu fmden sind. Darin stellt Wittgenstein (1980: z.B. 19) den "Spiel"-Charakter der Sprache in den Vordergrund seiner Reflexionen. Ahnlich wie Saussure war sich auch Wittgenstein der Komplexitat des Phanomens Sprache voll bewuBt. Er faBte nun namlich diese Komplexitat von Sprache bzw. von mit Sprache unmittelbar in Beziehung stehenden Tatigkeiten/Handlungen als sogenannte "Sprachspiele" auf, in denen eben der Gebrauch der Worter - Wittgenstein wurde "Worte" sagen - eine entscheidende Rolle spielt, wie es einer der zentralen Satze in den Philosophischen Untersuchungen ausweist: "Die Bedeutung eines Wortes ist sein Gebrauch in der Sprache." (Wittgenstein 1980 (PU): 41) Wenn man nun Wittgensteins "Sprachspiel" und Saussures LANGUE und PAROLE betrachtet, erkennt man die enge Verwobenheit eines (strukturierten) Sprachsystems mit Formen des Verhaltens bzw. sprachlichen Handelns. Der Spielcharakter der Sprache, den Wittgenstein in seinen Philosophischen Untersuchungen zu verdeutlichen versucht, erhalt seine bekannteste Veranschaulichung in der

40


Analogiebildung zum Schachspiel. Das Interessante dabei 1st nun, daB diese Analogic ebenfalls bei Saussure erscheint. Auch John Searle verwendet sie im Rahmen der Erlauterung seiner Theorie der Sprechakte (s. Searle 1971), die mit Saussure und Wittgenstein in Beziehung gesetzt werden kann. Im von Bally und Sechehaye herausgegebenen Cours erscheint die Analogiebildung mit dem Schachspiel bzw. mit einer Partie Schach insgesamt viermal. Die Mitschriften der Studenten belegen nun auch, daB der "authentische" Saussure diese Analogiebildung in derselben Form in seinen Vorlesungen angefuhrt hat (vgl. Saussure 1989-90: 64, 195ff, 241f, 249). An der ersten Stelle, in der Saussure Sprache mit Schach vergleicht (Saussure 1967: 27), schreibt er dem Schachspiel zu, ebenso wie die Sprache eine Grammatik zu besitzen. Schachspiel und Sprache bestehen demnach aus einem System von Regeln bzw. Konventionen. Beziiglich der Analogic Sprache-Schachspiel weist er in einer umfangreicheren Passage (Saussure 1967: 104-106) dann u.a. darauf hin, daB man es sowohl beim Schachspiel als auch bei der Sprache mit einem Werte-System zu tun hat, das eigentlich aufgrund von Konventionen feststeht, das also prinzipiell unveranderlich ist. Entscheidend in bezug auf sprachliche Handlungen scheint der Begriff der Ubereinkunft zu sein, den Saussure verwendet: Die Mitglieder der sprachlichen Gemeinschaft haben sich sozusagen unbewuBt auf ein System von Werten geeinigt, das in einer speziellen Weise gebraucht werden muB, damit sich die einzelnen Mitglieder der Sprachgemeinschaft verstehen. Im Schachspiel entsprechen dem die Spielregeln, die im Gegensatz zur Sprache natiirlich bewuBt aufgestellt wurden. Ein veranderter Gebrauch einzelner Elemente hingegen kann auch das System der Werte wandeln, was Saussure dann nicht mit dem Schachspiel generell, sondern mit einer Partie Schach vergleicht (a.a.O). DemgemaB setzt ein einziger Schachzug nur eine einzige Figur in Bewegung; ebenso beziehen sich nach Saussure in der Sprache Veranderungen nur auf isolierte Elemente. Dieser Vergleich Saussures im Rahmen der Schachanalogie deutet nochmals die pragmatische Dimension des Wandels der Sprache an: Wenn ein Mitglied einer Sprachgemeinschaft ein sprachliches Zeichen irgendwann einmal verandert gebraucht, sei es in einer anderen lautlichen Form oder mit einer anderen Bedeutung, so kann das Auswirkungen auf das gesamte Sprachsystem (langue) mit sich bringen, die einfach nicht abzusehen sind. Ein veranderter Gebrauch bedeutet gleichzeitig eine Veranderung des Wertes eines sprachlichen Zeichens. Das impliziert jedoch, daB der der vorherigen (vor einem Wandel) bzw. der nachherigen (nach einem von der Sprachgemeinschaft sanktionierten Wandel) Ubereinkunft entsprechende Gebrauch eines sprachlichen Zeichens Teil der Regeln des Sprachsystems (langue) sind. Wenn Saussure (a.a.O) dann zusatzlich bemerkt, daB Veranderungen zwischen zwei Zustanden beim Sprechen (parole) keine Geltung ha'tten, dann sind eben nur die Veranderungen zwischen aufeinanderfolgenden Zustanden gemeint und nicht Veranderungen, die zeitlich sehr weit auseinanderliegen: Wenn der "Zustand" einer Partie Schach nach dem sechsten Zug beider Parteien und dann nach dem siebten Zug der zwei Gegner verglichen wird, so sticht lediglich dem Schachkenner sofort die Veranderung ins Auge; vergleicht man aber die Stellung der Figuren nach dem sechsten Zug und nach dem 27. Zug, so fallt die Veranderung auch dem ungeubten Schachspieler auf. Ebenso sieht es mit der Sprache aus, wie man sich leicht vorstellen kann. Der Vorgang der Veranderung in der Sprache verlauft langsam und in kleinen Schritten, fur die Verwender der Sprache eher unmerklich. Erst nach einer langen Zeitspanne wird die Veranderung iiberdeutlich. In gewissem Sinne erhalt die Differenzierung, die Saussure zwischen langue und parole vornimmt, durch eine solche Darstellung nochmals


41

Plausibilitat, da man ohne weiteres sagen konnte, daB die einzelne sprachliche Handlung mil ihrem (synchronen) Gebrauch sprachlicher Zeichen (also das Sprechen, die parole) durchaus losgelost ist von den Veranderungen, die im System (langue) im Laufe der Zeit feststellbar sind. Die einzelnen, zeitlich eng aufeinanderfolgenden sprachlichen Handlungen unterscheiden sich kaum voneinander, nur diejenigen, die zeitlich weit auseinanderliegen. Somit wirken die einzelnen sprachlichen Handlungen sich mit Fortlaufen der Zeit auf das gesamte System (langue} aus. Man konnte auch sagen, daB parole die konkrete sprachliche Handlung ausmacht, wahrend in der langue, im Sprachsystem, die Regeln reprasentiert sind, die jeweils eine allgemein akzeptierte sprachliche Handlung defmieren. Langue und parole interagieren somit in der Art und Weise, wie es oben schon erklart wurde; die sprachliche Handlung ist dementsprechend Bestandteil sowohl der parole als auch der langue. Die Analogic von Sprache und Schachspiel deckt gemaB Saussure (1967: 127) auch generelle Problematiken auf, die beiden "Systemen" gemeinsam sind: Wenn wir Schach spielen wollen, so miissen wir die Figuren ("Spieleinheiten") und deren Funktion ganz genau kennen. Doch dies allein genugt noch nicht; wir sollten zusatzlich auch noch iiber die Moglichkeiten des "Zusammenspiels" der separaten Bestandteile (und deren Funktion) wahrend der konkreten Partie Bescheid wissen, urn ordentlich spielen zu konnen. Genau das ist es, was Saussure mit der problematischen Abgrenzung der einzelnen Elemente meint: Im Spiel selbst konnen wir zwar noch die Figuren sowie ihre Funktion (bzw. ihren Wert) gegeneinander abgrenzen; aber im Gesamtzusammenhang des Spiels ist es ein Miteinander der Elemente, das nicht mehr im einzelnen abgegrenzt werden kann. Ebenso bei der Sprache: Jeder an der Sprache Teilhabende muB standig uber die Bestandteile (die sprachlichen Zeichen) und deren (regelhaften) Gebrauch Bescheid wissen - die Einheiten scheinen klar abgrenzbar zu sein. Im tatsachlichen Gebrauch der Sprache spielen jedoch so viele Dinge zusammen, daB eine Abgrenzung nicht mehr moglich ist (gemaB Saussure). Diese Sichtweise Saussures mag erneut seinen "Respekt" gegenuber dem Faktum Sprache bzw. vor dessen Komplexitat - damit ist nicht nur die langage, sondern nun auch explizit die langue gemeint - veranschaulichen. Mit diesem Problem der Abgrenzung der sprachlichen Einheiten untereinander kommt ebenfalls die soziale Dimension der Sprache wieder zum Vorschein: Das Sprachsystem, seine einzelnen Bestandteile werden von einer Sprachgemeinschaft so verwendet, daB sie nicht mehr als Elemente erkennbar sind (d.h. eindeutig voneinander abgrenzbar), sondern lediglich als ein funktionierendes ganzheitliches System, dem die Sprachgemeinschaft Geltung verleiht. Die Stellen, an denen Saussure im Cours das Schachspiel als Vergleich zur Sprache heranzieht, konzentrieren sich auf den Aspekt der Veranderung. Man kann nun nicht gerade behaupten, daB dieser Gesichtspunkt im Mittelpunkt der Uberlegungen des (spaten) Wittgenstein stand (vgl. etwa Harris 1993: 223). Dennoch bestehen auffallige Parallelen zwischen Saussure und Wittgenstein bezuglich ihrer Sprachauffassung, wenn ihre jeweiligen Erlauterungen der Schachspielmetapher herangezogen werden. So finden wir beispielsweise innerhalb der Verwendung der Schachspielanalogie auch bei Wittgenstein (1980: 127) im Hinblick auf die soziale Dimension der Sprache den Begriff einer Institution. Wittgenstein (a.a.O) betont, daB Sprache kein (ausschlieBlich) individuumsbasiertes Phanomen sein kann. Das bringt allein schon die Redeweise von der Regelbefolgung mit sich. Man konnte gar keine Regeln feststellen, wenn gewisse (sprachliche oder "spielerische") Handlungen nur ein einziges Mai aufgetreten waren. Der Regelbegriff verlangt von dem Phanomenbereich, auf den wir ihn anwenden, den Charakter einer Institution, wie es Wittgenstein ausdriickt, oder, urn es ein wenig abzuschwachen: den Charakter einer

42


"Gepflogenheit". Vielleicht erlautert der zuletzt genannte Ausdruck den sozialen Aspekt sogar noch besser: Eine Gepflogenheit defmiert sich dadurch, daB sie von vielen Menschen oftmals und in der gleichen Form (d.h. regelhaft, regelmaBig, regelgerecht) ausgefiihrt wird. Eine Gepflogenheit ist per defmitionem "sozial". Fiir Wittgenstein sind nun sowohl das Schachspiel als auch die Sprache soziale Gepflogenheiten, soziale und regelhaft, regelmaBig, regelgerecht ablaufende Handlungen von Individuen, die, indem sie gerade Sprache verstehen, wissen, wie sie die (sozialen) Regeln des sprachlichen Handelns anwenden miissen: Sie beherrschen die Technik der Anwendung der Regeln zum Gebrauch der sprachlichen Elemente. Sprache ist demzufolge fur Wittgenstein auf ahnliche Art und Weise eine soziale Institution wie fur Saussure. Wittgenstein ist das Regelsystem des Gebrauchs sprachlicher Zeichen wichtig; er betrachtet die Sprache aus dem Blickwinkel des Pragmatikers. Wenn er jedoch schreibt, daB Sprache keinesfalls ein "unraumliches und unzeitliches Phanomen" sei (s. Wittgenstein 1980: 78), so stellt er sich eindeutig gegen einen (platonistischen) Realismus im Shine von Katz (1985) so wie gegen seine eigene, noch im Tractatus logico-philosophicus (Wittgenstein 1982) vertretene Position, die mit derjenigen der Logischen Empiristen (z.B. Russell) zu identifizieren ist, wonach - grob gesagt - Sprache als rein abstraktes Objekt zu behandeln ist, das keinerlei raumliche bzw. zeitliche Dimension aufweist. Somit demonstriert Wittgenstein eine weitgehende Ubereinstimmung mit Saussure: Er erkennt mehrere Dimensionen des Faktums Sprache an, auch wenn er nicht an alien interessiert ist.3 Wittgensteins spate Sprachphilosophie insbesondere der Philosophischen Untersuchungen ubte groBen EinfluB auf die angelsachsische Philosophic der Nachkriegszeit aus. Man konnte die Formierung der Schule der "ordinary language philosophy" in Oxford bedeutende Vertreter waren beispielsweise Geach, Ryle, Strawson und Grice - als sprachphilosophisches Legal Wittgensteins bezeichnen, das vor allem deshalb von Bedeutung ist, weil aus ihm die Theorie der Sprechakte hervorging, die eben maBgeblich durch die Wittgensteinsche Spielanalogie inspiriert wurde. AuBerdem zahlt Wittgenstein in den Philosophischen Untersuchungen schon das auf, was spater als Sprechakte bezeichnet werden sollte (s. Wittgenstein 1980: 28-29). Er nimmt im Rahmen dieser Aufzahlung Bezug auf seine "friihe" Philosophic des Tractatus logico-philosophicus und bekennt sozusagen die Defektivitat dieser im Hinblick auf die dort fehlende Beriicksichtigung dessen, was eigentlich vor sich geht, wenn Menschen (auf "nattirliche", "gewohnliche" Weise) miteinander sprechen. Worauf Wittgenstein anspielt, ist die Tatsache, daB man mit Hilfe der (kalkulisierten, begrifflich prazisen) Kunst-Sprache der Logik in dieser Welt nicht die Moglichkeit hatte, das auszudriicken, was die Sprecher einer naturlichen, gewohnlichen Sprache mit ihrer Handlungsweise vermogen zu tun. Genau dieser Gesichtspunkt stand, wie der Name schon bezeugt, im Mittelpunkt der Betrachtungen der ordinary language philosophy ("Philosophic der Alltagssprache" bzw. "Philosophic der normalen Sprache" - vgl. Savigny 1980) und ganz speziell der Sprechakttheorie, die sich aus dieser Philosophic der normalen Sprache entwickelte und deren bedeutendste Vertreter John L. Austin (1963) als Griinder sowie John Searle (1971) sind. 3

Eine ausfuhrlichere Analyse der Wittgensteinschen Verwendung der Schachanalogie im Zusammenhang mit seiner Sprachauffassung ist bei Hartmann (1998: 46-57) zu finden, wo die Parallelen zwischen Wittgenstein und Saussure durch Heranziehung einschlagiger Zitate noch klarer werden.


43

Searle defmiert nun den Oberbegriff des Sprechaktes mit einer Unterteilung in drei Arten von Akten: 1.) Den sozusagen "physikalischen" Teil bezeichnet er als Aufierungsakt, d.h., wenn ich einen Satz ausspreche, auBere ich WQrter (bzw. Morpheme und Satze) - ich vollziehe somit einen AuBerungsakt; 2.) den zweiten Aspekt nennt Searle dann den propositionalen Akt, was soviel bedeutet, daB ein Sprecher auf etwas Bezug nimmt, er referiert auf bestimmte Objekte und pradiziert etwas auf sie - ich vollziehe also einen propositionalen Akt, wenn in meinem geauBerten Satz die Eigenschaften Referenz und Pradikation gegeben sind; 3.) Searle geht dann von einem illokutionaren Akt aus, fur den er die ublichen Beispiele (Behaupten, Fragen, Befehlen, Versprechen, etc.) anfuhrt. Spa'ter erganzte Searle diese Trias mit Austins perlokutionarem Akt, unter dem die Wirkung verstanden wird, den illokutionare Akte (sozusagen als Sprechabsichten) unmittelbar auf die Angesprochenen ausuben. Zu solchen perlokutionaren Akten zahlt Searle beispielsweise uberreden, uberzeugen, erschrecken, alarmieren, dazubringen, etwas zu tun. Diese kurze Erlauterung dessen, was Searle unter dem Oberbegriff Sprechakt versteht, war notwendig, urn die These und deren Erklarung, die er in Searle (1971) auBert, im Zusammenhang mit der sozialen Dimension von Sprache sowie den Bezug dieser Dimension zur Schachanalogie zu verstehen. Searle geht davon aus, "daB illokutionare Akte im allgemeinen im Medium der Sprache und damit auf Grund bestimmter Regeln vollzogen werden und tatsachlich nicht vollzogen werden konnten, wenn die Sprache nicht die Moglichkeit ihres Vollzuges erlaubte." (Searle 1971: 62) Das bedeutet also, daB die Sprache (als System konstitutiver Regeln, also die Saussuresche langue) das Medium bereitstellt, welches illokutionare Akte iiberhaupt erst moglich macht. In verschiedenen Sprachen bestehen dann verschiedene (unbewuBte) Konventionen dariiber, wie diese illokutionaren Akte (in der jeweiligen Einzelsprache) zu realisieren sind. Searle verwendet haufig die Begriffe Konvention, Regel und Akt. Um diese Begriffe tiefgriindiger zu klaren, benutzt er die Schachanalogie (Searle 1971: 62-63). Die Parallele zwischen Sprache und Schach ist gemaB Searle nun folgende: Im Schachspiel ist es eine Konvention, welche a'uBere Form der Konig hat; es muB allerdings die Konigsfigur in irgendeiner Form vorhanden sein, damit die zugrundeliegende Regel "Der Konig darf sich bei einem Zug nur auf ein seinem jetzigen Feld direkt benachbartes Feld begeben" realisiert werden kann. Das gilt auch fur eine Schachpartie, die lediglich in der Phantasie zweier Gegner gespielt wird; beide haben eine Vorstellung von der Figur, die einer Konvention entspricht. Die Konvention tiber die aufiere Form der Konigsfigur ist somit notwendig fur die Realisierung der zugrundeliegenden Regel, damit der "Akt" des speziellen Schachzugs iiberhaupt erst erfolgen kann. Searle sieht es nun so, daB dasselbe auch fur die Sprache gilt. Es gibt ein zugrundeliegendes konstitutives sprachliches Regelsystem, das nicht realisiert werden ko'nnte, wenn es keine Konventionen iiber den Vollzug sprachlicher Verhaltensweisen ga'be. Das hat als Konsequenz fur Searles illokutionare Akte, daB diese selbst, also beispielsweise befehlen, warnen, etc., Bestandteil eines zugrundeliegenden Regelsystems sind, ihre Realisierung jedoch abhangt von Konventionen uber die Art und Weise ihres Vollzugs. In diesem Sinne ist einerseits Searles These zu verstehen, daB Sprache sprechen eine regelgeleitete Form des Verhaltens sei, sowie andererseits seine Bemerkung:

44

Langue and Parole in Synchrony and Diachrony "Nun konnte es [...] scheinen, als ob der Gegenstand tneiner Untersuchung - in Saussures Terminologie - »parole« im Gegensatz zu »langue« ware. Ich behaupte jedoch, dafi es sich bei der adaquaten Untersuchung von Sprechakten um eine Untersuchung der langue handelt." (Searle 1971:32)

Die Sprechakte selbst sind gemafi Searle Bestandteil des zugrundeliegenden sprachlichen Regelsystems, also der langue, wie sie der Saussure des Cours versteht. Die Konventionen iiber die Art und Weise ihrer Ausfuhrung gehoren jedoch nicht in dieses System und scheinen darauf auch keinen EinfluB zu haben. Generell birgt diese Einschatzung demnach die Schwierigkeit der Trennung zwischen Konventionen und Regeln, zumal die Frage aufkommen mag, ob die von Searle angesprochenen Konventionen nicht Teil der Regeln sind. Sowohl in der Schachanalogie als auch in der Sprache konnte man doch eher von (kulturellen) Spezifikationen der Regeln sprechen, anstatt eine Trennung von zugrundeliegenden Regeln und Konventionen iiber die Art und Weise ihrer Ausfuhrung vorzunehmen. Uberhaupt wird auch nicht klar, welchen Zweck diese Trennung haben soil, wenn Searle davon ausgeht, dafi eine Sprache sprechen eine regelgeleitete Form des Verhaltens ist. An welcher Stelle greifen hier die Konventionen? Sind somit die Konventionen nicht eigentlich Bestandteile der Regeln, so daB ich sagen kann: Eine Sprache sprechen heifit, sich gemafl Regeln, die spezielle Konventionen beinhalten, zu verhalten? Die Rolle der Sprachgemeinschaft, somit die soziale Dimension der Sprache ware hier deutlicher: Die Sprachgemeinschaft ist fur die Festlegung des Regelsystems sozusagen mitverantwortlich, indem sie sich in einer bestimmten Weise verhalt, diese Verhaltensweise aber mit der Zeit in kleinen Schritten verandert, bis schliefilich das Regelsystem auch Wandlungen unterliegt Insofern stellt das gesamte Regelsystem - um in den Worten Saussures zu sprechen - eine soziale Konvention dar.

Literaturverzeichnis Austin, John L. (1963). How to do things with words. Clarendon, Oxford. Bartschat, Brigitte (1996). Methoden der Sprachwissenschaft. Von Herrmann Paul bis Noam Chomsky. Erich Schmitt Verlag, Berlin. Buyssens, E. (1942). Les six linguistiques de F. de Saussure. Revue des langues vivantes, 8, 1523. Buyssens, E. (1952). Dogme ou libre examen. Cahiers Ferdinand de Saussure, 10, 47-50. De Mauro, Tullio (1967). Notizie biografiche e critiche su F. de Saussure; Note. In: Corso di linguistica generate (Saussure, Ferdinand de), 285-452, Laterza, Bari. Godel, R. (1957). Les sources manuscrites du Cours de linguistique generate de F. de Saussure. Droz, Genf. Harris, Roy (1988). Language, Saussure and Wittgenstein. How to play games with words. Routledge, London/New York. Harris, Roy (1993). Saussure, Wittgenstein and la regie dujeu. In: Linguistics and Philosophy. The controversial interface (Harre, Rom/Harris, Roy, Hrsg.), 219-231, Pergamon Press, Oxford. Hartmann, Ralph A. (1998). Grundlagenprobleme der Sprachwissenschaft. Kritische Analyse und Abwdgung der allgemeinen Ansichten uber Sprache von Saussure, Chomsky undPiaget. Hartung-Gorre Verlag, Konstanz.


45

Holdcroft, David (1991). Saussure: Sign, systems and arbitrariness. Cambridge University Press. Jager, Ludwig (1975). Zu einer historischen Rekonstruktion der authentischen Sprach-Idee Ferdinand de Saussures. Dissertation. Eigenverlag, Dusseldorf. Jager, Ludwig (1976). F. de Saussures historisch-hermeneutische Idee der Sprache. Bin Pladoyer fur die Rekonstruktion des Saussureschen Denkens in seiner authentischen Gestalt. Linguistik undDidaktik, 27, 210-244. Katz, Jerrold J., Hrsg (1985). The Philosophy of Linguistics. Oxford University Press. Koerner, E.F. Konrad (1973). Ferdinand de Saussure: Origin and development of his linguistic thought in western studies of language. Vieweg, Braunschweig. Koerner, E.F. Konrad (1975). European Structuralism: early beginnings. In: Current trends in linguistics (Seboek, T.A., Hrsg.), Vol. XIII, 717-827, Mouton de Gruyter, The Hague. Saussure, Ferdinand de (1967). Grundfragen der allgemeinen Sprachwissenschaft, 2. Auflage. De Gruyter, Berlin. Saussure, Ferdinand de (1989-90). Cours de linguistique generate. Edition critique par Rudolf Engler. 2 Bande. Harrassowitz, Wiesbaden. Savigny, E.v. (1980). Die Philosophic der normalen Sprache. Eine kritische Einfuhrung in die ôrdinary language philosophy«, 2. Auflage. Suhrkamp, Frankfurt am Main. Scheerer, T.M. (1980). Ferdinand de Saussure. Rezeption und Kritik. Wissenschaftliche Buchgesellschaft, Darmstadt. Searle, John R. (1968). Austin on locutionary and illocutionary acts. Philosophical Review, 77, 405-424. Searle, John R. (1971). Sprechakte. Ein sprachphilosophischer Essay. Suhrkamp, Frankfurt am Main. Wittgenstein, Ludwig (1980). Philosophische Untersuchungen (PU). Suhrkamp, Frankfurt am Main. Wittgenstein, Ludwig (1982). Tractatus logico-philosophicus. Logisch philosophische Abhandlung. Suhrkamp, Frankfurt am Main. Wittgenstein, Ludwig (1984). Philosophische Grammatik. Suhrkamp, Frankfurt am Main.


3

ElJBULIDES AS A 20TH CENTURY SfiMANTICIST

Pieter A.M. Seuren, Nijmegen University, The Netherlands

Abstract It is the purpose of the present paper to highlight the figure of Eubulides, a relatively unknown Greek philosopher who lived ± 405-330 BC and taught at Megara, not far from Athens. He is mainly known for his four paradoxes (the Liar, the Sorites, the Electra, and the Horns), and for the mutual animosity between him and his younger contemporary Aristotle. The Megarian school of philosophy was one of the main sources of the great Stoic tradition in ancient philosophy. What has never been made explicit in the literature is the importance of the four paradoxes for the study of meaning in natural language: they summarize the whole research programme of 20th century semantics, including the problems of vague predicates (Sorites), intensional contexts (Electra), and presuppositions (Horns). It is a surprising and highly significant fact that a contemporary of Aristotle already spotted the main weaknesses of the Aristotelian paradigm.

0. INTRODUCTION Eubulides was a Greek philosopher in the fourth century BC, and is the almost forgotten author of four so-called 'paradoxes', the Liar, Electra, Sorites and the Horns. Although these paradoxes pose extremely serious problems for Aristotle's theory of truth, and consequently for modern semantics and logic, their significance has been underrated through the ages, probably because no answers were found. They were often depicted as flippant cracks, unworthy of serious attention, but also sometimes as embarrassing riddles. During the 20th century these four problems have come up again. Indeed, one may say that they define, more or less, the whole research programme of 20th century semantics. Yet the lineage to Antiquity has been acknowledged only for the Liar and Sorites. Electra and the Horns have been rediscovered (mainly by Frege), and there appears to be no awareness that they were part of the package of arguments put together by Eubulides. It now seems time, therefore, for a radical reappraisal and a rehabilitation of Eubulides as one of the fathers of 20th century semantics.

1. THE LIFE AND TIMES OF EUBULIDES Who was Eubulides? In actual fact, very little is known about his life and works.1 His exact date of birth is not known, but as an educated guess it may be placed towards the end of the 1

The following reconstruction of Eubulides' biography implies a correction of Seuren (1998:426). 47

48


fifth century, perhaps around 405 BC. We know that he was born in the Greek colony Miletus, in Asia Minor which is now known as Turkey. We know that as a young man he migrated to Greece, and that he took over a small school of philosophy in Megara, not far from Athens, in 380. According to the Oxford Classical Dictionary (s.v. Eubulides), 'he is said to have taught the Athenian rhetor Demosthenes dialectic [i.e. logic P.S.] and rhetoric', an important detail against the background of his well-known quarrels with Aristotle: Demosthenes was Athens' political leader in the struggle against the Macedonian king Philip II, and thus politically opposed to the Macedonian philosopher Artistotle, who taught in Athens from 335-323 BC, under the protection of Macedonian rule. It is not known exactly when he died, but one may surmise that it will have occurred some time around 330 BC. Eubulides is mainly known for his four 'paradoxes', till recently generally considered pointless or even flippant. Kneale & Kneale (1962:114-5), however, suspect that 'he must surely have been trying to illustrate some theses of Megarian philosophy', and that 'the Megarian study of the paradoxes was a serious affair and not mere perversity.' In the context of modern semantics, we now understand that Eubulides was criticising Aristotle, in particular the basic premisses of Aristotelian truth theory and logic, viz. truth as correspondence and the principles of Contradiction and of Bivalence. It is on account of his paradoxes that we feel a total reappraisal of Eubulides is called for. Let us first lend some colour to the historical setting in which we encounter Eubulides. His city of birth, Miletus, was one of the cities founded by the Greeks for the purpose of maritime trade. Like almost all Greek foundations on the coasts of Asia Minor, and also on the Sicilian and Southern Italian coasts, it had become rich and powerful well before the cities in metropolitan Greece reached their heyday. One does not have to be a Marxist to surmise that the early rise of philosophy in the Greek colonies of Asia Minor, Sicily and Southern Italy was one of the side effects of the economic well-being of these cities. The sixth century philosophers Thales, Anaximander and Anaximenes were from Miletus. Heraclitus, who lived from ± 535 till 480, was from Ephesus, also in Asia Minor. Pythagoras, though born in Samos around 580, taught at Croton in Southern Italy. The city of Elea, also in Southern Italy, had an important school of philosophy during the sixth and fifth centuries. Empedocles (483-423) was from Agrigentum in Sicily and taught philosophy there. It is therefore not surprising to see a young man from Miletus turn to philosophy during the early years of the fourth century. Meanwhile, philosophy was beginning to develop in Greece proper, but in a very different vein. Whereas the earlier philosophers in the colonies had concentrated on questions of time, infinity and numbers, of the unity of physical matter, and the forces in the physical universe, the philosophers who began to appear in metropolitan Greece after about 450 took a wholly different, much more practical, direction. Most of them originated from the colonies, but they practised in Greece proper as teachers on questions of political power, morality and truth, with a strong admixture of rhetoric. Again, one does not have to be a Marxist to see a connection between this change in focus and the rise of the first democracies in Greece proper (but not in the colonies, which all stayed under totalitarian rule). Since the early years of the fifth century, many Greek cities had assumed a form of government where decisions were taken in the people's assembly, consisting of free male adults of a certain economic standing. Clearly, in such a setting, it was important for ambitious youngsters to be able to sway public opinion in the political gatherings of their city state. To satisfy this demand, a number of men began to advertise themselves as teachers of rhetoric, political theory and whatever was available in the way of relevant knowledge in those days. These teachers, who were called 'Sophists' (literally 'experts'), travelled from city state to city state, offering their teaching for good money.

Eubulides as a 20th Century Semanticist

49

Not too much is known about the Sophists, as none of their writings have survived. All we know about them is based on secondary sources, i.e. quotations and comments by later authors, who vary in their judgments on the quality and moral standing of these itinerant teachers. The careful piecing together of all the evidence by modern scholars has resulted in a mixed overall picture. Whereas some of the Sophists, apparently, deserved a reputation of being a trifle too opportunistic, others stand out as intellectuals of high academic and moral standing. When the best of them taught their young charges how to convince a public assembly by means of clever rhetoric, they also went into the question of the morality of doing so. One important question, obviously, was: may one tell the assembly falsehoods in order to gain power and influence? The answer was a guarded 'No'. But it was recognised, at the same time, that the question of what constitutes truth and falsity could not be answered satisfactorily. In the absence of any serious truth theory, the best they could do was make a democratic appeal to majority opinion: if the majority believes something to be true, then we may take it to be true (not unlike what is commonly seen in modern science). Similar questions arose as to norms of justice. Given the lack of absolute definitions of 'right' and 'wrong', each proposal to do with restrictions on personal freedom or with inflicting damage or punishment had to be judged on grounds of well-informed public opinion and a well-developed civic sense of justice and injustice. A good politician, therefore, had to be well-informed and share his knowledge with the citizens in the assembly, in order to come to a consensus on the best possible solution, given the circumstances. This was what the best of Sophist teaching amounted to. It is easy to see what Sophist teaching of less high standing would have been like. One man in Athens, however, felt unhappy with this type of reasoning. This was the famous Socrates (469-399), an eccentric who went along the stalls of traders and craftsmen in the market, interrogating them about the philosophical foundations of their trade (and not making himself popular by doing so). Especially towards the end of his life, Socrates gathered a circle of mostly upper class young men around him, inspiring them with his questions and his analyses. One of these was Plato (429-347), whose family belonged to the old Athenian aristocracy. Unlike Socrates, who never wrote, Plato was a gifted and prolific writer. His philosophical dialogues, always figuring Socrates as a protagonist, occasionally reach the highest peaks of literary excellence. The story of Socrates and his circle of admirers, including Plato, is a classic in our history books. It has been told and retold endless times, usually in admiring and often even romanticising tones. What was, however, usually either left out or definitely understressed, until Popper (1945) and Stone (1988), was the political side of things. Socrates, who was himself of what we would now call lower middle class origin, held ideas that were not typical of his class. He liked to associate with the antidemocratic, conservative upper class, which was one of the factors that united him with his circle of upper-class, conservative young followers. Stone, in particular, shows in minute detail the basically totalitarian attitude of Socrates, who favoured benevolent but absolute power on the part of one leader or a small group of leaders. Popper, half a century ago, made the world fully aware of Plato's political ideas: Plato actively propagated ruthless dictatorship, with a secret police, a ban on literature and travelling, a stifling of public debate, and all the other grim features of dictatorial regimes that we have come to know so well during the past century. Although Popper was scorned, and Stone largely ignored, by the community of ancient historians, their important contributions to our way of looking at this episode in history have gradually been accepted and integrated, and it is now no longer anathema to mention these more controversial aspects of early Athenian philosophy.

50


It is important to know this, because for a brief period of fourteen months, between 404 and 403 BC, the democratic system of government in Athens was overthrown by a group of dictators, the so-called 'Thirty Tyrants'. The leader of the group was Critias, who belonged to Socrates' circle and was a first cousin of Plato's mother. Critias and his associates were responsible for over a thousand political murders and other atrocities committed during the year their regime lasted. Democracy was restored in the summer of 403, but the wounds remained. It is Stone's merit to have shown that the trial of Socrates, in 399, was part of the process, on the part of the democrats, of coming clean with the past without causing too much social damage. Socrates was convicted nominally on charges of 'corruption of youth', but in fact he was a scapegoat for the antidemocratic movement that had led to the tyranny of the Thirty. Socrates drank the hemlock and Plato fled Athens with a group of friends to live in exile. Not until 387 did Plato return to his native city, where he became highly respected and taught philosophy in his Academy till his death in 347. Among Plato's students in the Academy, in fact the most prominent among them, was a young man called Aristotle. Aristotle was not from Greece but from Macedonia, a kingdom just north of Greece with a rough and uncivilised tradition but now rapidly assimilating Greek culture along with the Greek language. Aristotle was born in Stagira, not far from the capital Pella, in 384 BC. His father had been the personal physician and a close friend of the then king of Macedonia, Amyntas II. An exceptionally gifted boy to begin with, Aristotle joined Plato's Academy at the age of 17, to remain there till Plato's death in 347. Having been passed over as Plato's successor, he left Athens to live, first, in Asia Minor and then in Lesbos. In 343-342 Amyntas' son and successor, Philip II of Macedonia, invited him to come and teach his son Alexander, then 14 years old. This he did for two years, taking Alexander and a group of his friends to a mountain farm and teaching them all there was to know in those days. During the 23 years of his reign, Philip had one great ambition, the conquest of Greece. Yet, despite many campaigns and battles, he never succeeded completely. Strong political opposition, especially on the part of the Athenians, and endless revolts kept frustrating his efforts. His main opponent in Athens was Demosthenes (384-322), who managed to inspire the Athenians, through his rhetorical skills, to keep resisting Macedonian domination. In 336 BC Philip II was murdered and Alexander succeeded him, barely twenty years old. Already an army commander of unusual skill, he first completed the task his father had set himself but had failed to bring to an end, the conquest of Greece, hi less than a year Alexander gained control of the whole of Greece, having set the total destruction of Thebes, ally of Athens, as a terrible example. After that, as one knows, he crossed into Asia and became master of the entire Near East, getting as far as the Indus. In April 323, at the age of 33, he died in Babylon, one of the capitals of the Persian empire which he had conquered. By then he had become a legend if ever there was one. In Asia he was widely worshipped as a god. In Greece one was simply awed and baffled. He entered history as Alexander the Great, the source of a rich mythology through the remainder of Antiquity and the Middle Ages. Meanwhile, Aristotle returned to Athens in 335, a year after his former pupil Alexander had established Macedonian rule there. Aristotle founded his own school of philosophy, the Lyceum, called after an ancient grove dedicated to Apollo Lyceios north-east of Athens (the site has recently been discovered and is now being excavated). There he taught till 323, when news of Alexander's death reached Athens. No longer certain of Macedonian protection, he left Athens overnight and sought refuge in Chalcis, just north of Athens, where a Macedonian garrison was stationed. One year later, in 322, he died of an intestinal disease.


51

Now back to Eubulides. No doubt attracted by the fame of Socrates and Plato, Eubulides migrated to Athens as a young man, and in or around 380 he took over a small but respectable school of philosophy that had been founded by a philosopher called Euclides (not the mathematician!) in the town of Megara, some thirty miles west of Athens. This Euclides is reported to have provided hospitality to Plato and his friends as they fled from Athens in 399, after Socrates' death. As was said at the beginning, tradition has it that Eubulides taught logic and rhetoric to the Athenian Demosthenes, the great opponent of Philip of Macedonia. The evidence is scanty, but it suggests that the well-known personal animosity between Aristotle and Eubulides was at least partly also rooted in political differences. It is hardly thinkable that anti-Macedonian feelings had suddenly subsided after Alexander's forced entry in Athens. It is known, moreover, that Aristotle could only teach in Athens as long as he was protected by Macedonian power The conclusion seems justified, therefore, that there must have been a strong anti-Macedonian faction in Athens and elsewhere in Greece throughout the period in question. And we may add that Eubulides was no doubt an ardent member of that faction. In any case, no sympathy was lost between him and Aristotle.

2. THE ARISTOTELIAN THEORY OF TRUTH But it is time now to turn to the real issues. One of the recurrent themes in Plato's dialogues, and probably also in Socrates' teachings, is the question of the nature of truth and falsity, broached by the Sophists. One of the things that worried Socrates and his followers was the fact that the Sophists presupposed an absolute notion of truth, even if adequate knowledge of what is true and false is often not attainable. If that is so, they reasoned, we must try and find out what truth, taken in and by itself, amounts to. Thus began the quest for the nature of truth and falsity. The theme recurs in many of Plato's dialogues, but it is treated most systematically in his dialogue The Sophist, written in his later years. Here, however, no clear and final conclusion is reached. Yet Plato makes some important statements, which have become integrated into the philosophical tradition. In Sophist (263 BC), we witness a little exchange between the Eleatic Stranger (ES) and the young Athenian Theaetetus (Th): ES: And the true sentence says the things about you the way they are. Th.: What else? ES: And the false one says things that differ from the way they are. Th.: Yes. ES: That is, it says things that are not, as if they were. Th.: More or less, yes. Here Plato states that truth consists in correspondence between what is said and what is the case, while falsity consists in presenting a state of affairs that differs from what is the case. This notion of correspondence was further elaborated by Aristotle. After all the equivocations and inconclusive debates of the preceding century, Aristotle wants to be done with this question, and get on with things, such as the development of a proper truth calculus or logic. Early on in his Metaphysics, at 1011^26, he curtly defines truth as follows (translation by Warrington (Aristotle 1956:142)): We begin by defining truth and falsehood. Falsehood consists in saying of that which is that it is not, or of that which is not that it is. Truth consists in saying of that which is that it is, or of that which is not that it is not.

52


We see that, remarkably, Aristotle does more than repeat Plato's correspondence criterion. He introduces a new element, the definition of not as a simple truth-functional inverter of truth values: a toggle between true and false. This has momentous consequences. We must have a closer look at the issues involved, since this is where Eubulides attacked him, as we shall see presently. In his Posterior Analytics (77a 10-24), Aristotle presents, with great emphasis and insistence, two axioms ('no proof can be provided') for his truth theory, which are also applied in his logic. The first axiom is the Principle of Contradiction or PoC, which says that nothing can be the case and not be the case at the sime time. It follows that no sentence can be true and false at the same time. The second axiom is the Principle ofBivalence or PoB, or the Principle of the Excluded Third or PET. This says that propositions expressed in sentences fall into two (mutually exclusive) classes: the true ones and the false ones. A proposition is the mental act of assigning a property (accidens) to an entity (suppositum). This mental act can be expressed as a well-formed sentence in a language, whose subject term refers to the entity or suppositum in question, and whose predicate expresses the property assigned to the suppositum. Sentences that express a proposition, says Aristotle, have no choice but to be either true or false, with nothing in between and nothing outside these two values. Aristotle places particular emphasis on his thesis that the opposition between truth and falsity is not gradual but absolute. In many other cases of opposite pairs there is a gradual transition from one extreme to the other, as with good and bad, or light and dark. But for him, true and false do not belong to this class of gradual opposites, but to the class of absolute opposites, like dead and alive (for entities that have biological life), or even and odd (for numbers). This is his Principle of the Excluded Middle or PEM. It applies absolutely, but for one reservation. In De Interpretatione 19a30 Aristotle discusses the problem of the sea battle which is going to take place, and he then makes a possible exception for statements about future events, which would be exempted from PET. Otherwise, however, PET applies absolutely. Although PEM and PET are often confused, even in respectable writings, one can easily see that PEM is only a subaxiom of the more general PET. The other subaxiom, which we may call the Binarity Principle or BP, is Aristotle's thesis that there are exactly two truth values, and not three or even more. This yields the following rendering of Aristotle's truth theory, applied in his logic: Axioms: I Principle of Contradiction (PoC) II Principle ofBivalence (PoB) or Principle of the Excluded Third (PET) Subaxioms: a. Principle of the Excluded Middle (PEM) b. Binarity Principle (BP) But there is more. If truth consists in correspondence between what is said and what is actually the case, that is, if truth is found in the linguistic expression of a proposition that mentally assigns to an entity a property which the entity really has, in the actual world, then it should make no difference for the truth or falsity of an uttered sentence what expression is used to refer to the entity in question, or what expression is used to denote the property in question. What matters is only that the referring expression does indeed successfully refer to the entity intended, and that the predicate expression does indeed denote the property which the speaker has in mind. In other words, it is a direct consequence of Aristotle's notion of truth as correspondence that referring terms or predicate expressions may be substituted for each other salva veritate (i.e. without affecting the truth value) provided the terms refer to the same entity and the predicate expressions denote the same property.


53

This consequence was known to Aristotle but he did not subject it to closer scrutiny. It was not until Leibniz (1646-1716) that it was actually formulated as a principle. Leibniz gives a variety of wordings, the best known of which is probably (Gerhardt 1890:228): Eadem sunt quorum unum potest substitui alteri salva veritate. or 'If two terms can stand in for each other salva veritate, their reference is identical'. This has since become known as the Principle ofSubstitutivity, fundamental in modern semantics.

3. THE EUBULIDEAN PARADOXES 3.1 The Paradoxes and Aristotle But what has all this to do with Eubulides? The answer is simple. Eubulides is mainly known, in what little there is about him in the literature, for his so-called paradoxes. The Greek author Diogenes Laertius (early 3rd century AD), an invaluable source of information about ancient philosophy, writes (ii,108): Among the successors of Euclides was also Eubulides, who raised many questions of logic: the liar, the unnoticed man, the Electra, the hooded man, the heap, the horned man, and the bald man. In Kneale & Kneale (1962:114) we read: From the explanations given by various writers of later antiquity it appears that some of the seven paradoxes specifically attributed to Eubulides were merely variants of others and that the list can probably be reduced to the four following items: (1) The Liar. 'A man says that he is lying. Is what he says true or false?' (2) The Hooded Man, the Unnoticed Man, or the Electra. 'You say you know your brother. But that man who came in just now with his head covered is your brother, and you did not know him.' (3) The bald man, or the Heap. 'Would you say that a man was bald if he had only one hair? Yes. Would you say that a man was bald if he had only two hairs? Yes. Would you ..., etc. Then where do you draw the line?' (4) The Horned Man. 'What you have not lost you still have. But you have not lost your horns. So you still have horns.' This is explosive stuff for Aristotle's theory of truth. According to Kneale & Kneale (1962: 228), Aristotle himself was aware of the paradoxes (e.g. De Sophisticis Elenchis 180b2-7), and he was certainly sharp enough to have seen the deadly power of the arguments. But he declined to answer. This is remarkable, because he did reply to other criticisms directed at him by the Megarians, for example on modalities, or on the distinction between acting and undergoing. In those cases his comment is invariably that the Megarian point of view is 'silly' (Greek dtopori). But he never replied to the paradoxes put forth by Eubulides. One may well surmise, therefore, that Aristotle simply had no reply.

3.2 The Liar Paradox The Liar paradox is, of course, very well known nowadays. Its simplest form is in the sentence What I am saying now is false. Anyone saying this gets into trouble, for if what is said is true, it is at the same time false, and if it is false, then it is at the same time true. It looks very much as

54


if this paradox undermines both Aristotle's Principle of Contradiction and his Principle of Bivalence, since it appears to lead to the conclusion that sentences such as the one just given can be both true and false at the same time, which means that there would, after all, be a third possibility not catered for in Aristotle's theory of truth. Clearly, no-one will seriously propose this as a reasonable alternative to the Aristotelian theory, but if that theory is to be saved, then a satisfactory answer must be given to this paradox. In Antiquity, the Liar paradox was well-known, but nothing much was done about it. Cicero was well aware of the Liar paradox, and of the others as well. In Academicae Quaestiones (ii,96) he formulates the paradox, but attributes it wrongly to Chrysippus, a later Stoic philosopher: If you say that you lie and you speak the truth, you lie. But then you say that you lie and you speak the truth, so you lie. This question was raised by Chrysippus, but not solved by him. In his De Divinatione (ii,l l) Cicero mentions the Liar paradox along with the Sorites ('heap'): Can anything of what is dealt with by logicians or physicists be solved by soothsaying? Whether there is one world or many, what the first beginnings were of the world, from which everything originated: those are questions that come within the physicists' competence. And how to solve the Liar, also called by the Greek name of Pseudomenon, or the Sorites (which one may call the Acervalis ['heap'] in Latin if one finds that necessary, but it isn't necessary because the very word philosophia and many other Greek words, such as Sorites, have gained currency in Latin): those are questions for the logicians, not the soothsayers. And many other references to the Liar and the other paradoxes are found in the literature. A weak but famous reflex is found in Paul's Epistle to Titus (I, 12-13), where Paul, cursing the Cretans, writes, obviously unaware of the problem: One of their own prophets said: 'Cretans always lie, the wicked beasts and lazy bellies', and this witness is true. The 'prophet' referred to is a Cretan poet, Epimenides (6th century BC), which is why the paradox is also often called the Epimenidan Paradox. Strictly speaking, however, this is not paradoxical. It is sufficient to assume that the Cretan 'prophet' Epimenides simply said a falsehood if he said that all Cretans always lie. The problem arises only if one makes him speak the truth, as Paul does. In that case, there is indeed a problem, because then Epimenides allegedly spoke the truth and at the same time, being a Cretan himself, must have lied. Perhaps due to this passage in Paul's letter to Titus, the Liar paradox became very popular in medieval philosophy from the 12th century onward. The term used for the paradoxes, especially the Liar, was insolubilia ('unsolvables'). Kneale and Kneale (1962:228) write: [S]o far little work has been done on the surviving texts, and we cannot even say yet when or how the study of insolubilia began.... It may be ... that the [Liar] paradox was discovered afresh in the twelfth century. In any case medieval logicians were not satisfied with simple versions such as Ego dicofalsum, but invented complicated variants like 'Socrates says "What Plato says is false", and Plato says "What Socrates says is true", and neither says anything else. Is what Socrates says true or false?' It was realized that the trouble arose from the attempt to produce a certain sort of self-reference, and an insolubile was defined as a 'propositio habens super se reflexionem suae falsitatis aut se non esse veram totaliter vel partialiter illativa.' But there was no single agreed doctrine about the bearing of such paradoxes on the theory of meaning and truth. On the contrary, various medieval logicians suggested various ways of dealing with the trouble. Kneale & Kneale then mention an anonymous manuscript (Bibliotheque Nationale, Cod. Lat. 16617) probably dating from the early 14th century, where three possible solutions to the Liar are discussed. One solution is called restrictio and consists in a simple prohibition of


55

token reflexivity (self-reference) in the use of language. Another solution proposed is that of cassatio or nullity of meaning. This implies that sentences or texts that suffer from the Liar paradox are uninterpretable due to the impossibility of identifying an entity (suppositum) for the subject term in a Liar sentence like This very sentence is false. The third solution, cryptically called 'secundum quid et simpliciter', is (Kneale & Kneale, ib.), 'an unsatisfactory suggestion extracted from the passage of Aristotle mentioned above' [i.e. De Sophisticis Elenchis 180b2-7]. The solution proposed by Russell and Tarski in our century is a modern form of restrictio, and it is now considered the standard solution to the Liar paradox. Yet, although it seems to work satisfactorily in logical languages, which can be subjected to restrictions like a prohibition to mix object language and metalanguage, in natural language, as Kneale & Kneale rightly observe (1962:228), this would exclude all sorts of harmless forms of self-reference, such as What I am now saying is a sentence of English. In fact, one may add, natural language use is replete with perfectly natural and harmless mixings of object language and metalanguage, which, of course, nobody is going to forbid. A simple example is: (1) Jones has lived in Dnjepropetrovsk for many years, but he can still not pronounce it. where // refers not to the city of Dnjepropetrovsk but to its name. It would seem, therefore, that the Kneales are right in rejecting this restrictio for the theory of natural language meaning. Cassatio has hardly been explored in modern times, largely due to the dominant position of Russellian logic, which claims to have solved the Liar paradox definitely, completely disregarding the consequences of its solution for natural language. Yet we can easily pursue or reconstruct the cassatio argument. In a sentence like This very sentence is false the term this very sentence must refer to a proposition expressed, since the predicate false requires propositions to apply to. We thus need a proposition to serve as the suppositum of the predicate false. But all there is to be found is the very same proposition expressed, which is then again said to be false, and so on ad infmitum. There is thus no proposition to be found of which it can be said that it fails to satisfy the satisfaction conditions of the predicate false.2 The third solution mentioned in the medieval manuscript, secundum quid et simpliciter, does not appear to deserve further scrutiny.

3.3 The Electra Paradox Now to the second paradox, which we shall call the Electra, though it is also known as 'the Unnoticed Man' and 'the Hooded Man'. The mythical story, generally known, of course, in Eubulides' and Aristotle's day, goes as follows. While Agamemnon, king of Mycenes, was away to fight in the Trojan war, his wife Clytaemnestra had set up house with another man. Obviously, when Agamemnon was due back, there would be a problem. So when Agamemnon came home Clytaemnestra let him have a warm bath, during which she chopped his head off with a sword. That would have been the end of the affair, had there not been children, in particular a son, Orestes, who now had the holy duty to avenge his father. However, in order to do that, he would have to kill his mother, which would be a heinous crime. In order to sort out his moral dilemma, Orestes went to stay with an uncle for a while. At the end of that period he has made up his mind and has decided that the right thing to do, after all, is to kill his mother. So This argument is proposed as the correct solution for natural language in Seuren (1987).

56


he returns to Mycenes, but, sensibly fearing to receive a treatment similar to his father's, he disguises himself as a beggar so as not to be recognised. He then knocks at the gate and is let in. His sister Electra has him shown to the kitchen and given some soup. At this point in the story, Eubulides steps in and asks: 'Is the sentence Electra knows that her brother Orestes is in the kitchen true or false?' It should be true, since she knows that the beggar is in the kitchen, and the beggar is identical with Orestes. Therefore, given the Principle of Substitutivity mentioned above, Electra should also know that Orestes is in the kitchen. Yet she obviously does not know that, because if she did she might well have raised the alarm. We see here that the Principle of Substitutivity apparently does not always apply, and the question is why not. Aristotle may have seen that this is a very serious threat to his theory of truth, but there is no record of any reply. Nor is there any record through the ages of a proper analysis of the Electra paradox in the context of Aristotelian truth theory, until Frege (1892), who rediscovered the problem but was apparently unaware of its history in the Electra paradox. Frege found, first, that substitution of coreferring terms in identity statements complicates Aristotelian truth theory. In the true identity statement: (2) The morning star is the evening star. the terms the morning star and the evening star both refer to the same entity, the planet Venus. Here, substitution of one term for the other, as in: (3) The morning star is the morning star. will not change the truth value of the sentence, but it adds the peculiarity that the sentence then becomes necessarily true, i.e. true in all possible situations, given the metaphysical axiom that everything is identical to itself. In contrast, (2) is a contingent statement, reporting a discovery that was made some time in the distant past (probably by the Greeks). His solution consisted in an application of the old distinction between intension and extension not just to predicates, for which it was invented in Medieval philosophy, but also to definite terms. Frege now stipulates that the extension (or reference) of a term is the entity referred to, while the intension (or sense) of a term is 'the way in which it is presented' ('die Art des Gegebenseins', p. 26), or, in other words, the procedure by which the listener arrives at the entity referred to, given the term in question.3 Now, Frege says, the terms the morning star and the evening star have the same extension, but different intensions. This explains that (2) is a contingent statement, since when two different paths are followed it may come as a surprise that they lead to the same point, while (3), where the same path is followed twice and thus the same point is reached twice, is a necessary truth. Our interest, however, lies in Frege's second finding. Like Eubulides, he found that Substitutivity actually fails in what he called 'oblique' ('ungerade') contexts. That is, when a sentence or clause is embedded (usually as a that-clause) under a predicate that assigns a property to thoughts as entities, then substitution of coreferring terms may well lead to different truth values. Frege gives precious few examples, but his meaning is clear enough. Consider: (4)a. Jones believes that there is life on the morning star, b. Jones believes that there is life on the evening star. If Jones does not believe that the morning star is identical with the evening star, (4a) may well be true and (4b) false, or vice versa. The parallel with the Electra paradox is obvious:

3

Cp. Dummett (1973:281): 'In Frege's theory, the sense of a name consists of the means we have provided for determining an actual (existent) object as its referent.'


57

(5)a. Electra knows that the beggar is in the kitchen. b. Electra knows that her brother Orestes is in the kitchen. Since Electra does not know that the beggar is identical with her brother Orestes, (5a) is true but (5b) is false (in the story). It was clear to Frege, and one may hope also to Eubulides and to Aristotle, that this failure of substitutivity in what we now call intensional contexts poses a serious threat to the Aristotelian theory of truth, which entails substitutivity. An answer is therefore absolutely necessary. Frege's answer consisted in extending the extension-intension distinction to sentences and clauses as well. For him, the extension of a sentence is its truth value, while its intension is the underlying thought expressed in the sentence. For him the following two sentences: (6)a. There is life on the morning star, b. There is life on the evening star. have the same extension (truth value) but different intensions (underlying thoughts). However, when a sentence is embedded under a predicate assigning a property to thoughts, such as believe or know, the embedded clause is an argument term of the predicate and refers to a thought, not to a truth value. Now it follows that the 7/zotf-clauses in (4a) and (4b), or (5a) and (5b), refer to different thoughts, which means that substitution of one that-claase for another does not guarantee the preservation of the truth value of the whole sentence. It follows that substitutivity of terms which in fact corefer must fail in intensional contexts, which involve a subject that may not be aware of the fact that the two terms corefer. Frege's solution appealed to modern logicians. Yet when they began to develop what is known as 'formal semantics', around 1950, the notion 'thought' was considered suspect and unmanageable in a formal context. It was for that reason that Frege's solution to his second problem, the failure of substitutivity in intensional contexts, was reformulated in terms of a model-theoretic calculus involving possible worlds. We shall not go into the details of the theory of formal semantics, but it will be clear that Frege's second problem, which is, in fact, the Electra paradox, has been crucial in its development. And since, rightly or wrongly, formal semantics has dominated semantic studies since the early 1970s, we may say that the Electra paradox is at the very centre of modern semantic theory. Yet there is not a single mention of this fact in the literature. All references are to Frege, and Eubulides has been totally forgotten.

3.4 The Sorites The third paradox, generally known under the name of Sorites (from Greek soros 'heap'), but also known as 'the Bald Man', is a direct attack on the Aristotelian Principle of the Excluded Middle (PEM). It says, in effect, that the opposition between truth and falsity is like that between light and darkness: there are infinitely many intermediate values between the two extremes. This is embarrassing for the Aristotelian truth theory, since if truth consists in correspondence between what is said (or thought) on the one hand and that which is the case on the other, then, clearly, there is correspondence or there is not. There is no way in between. Nowadays, however, many philosophers and linguists feel that a different look at what constitutes truth easily undermines this argument. If truth is seen not as correspondence but rather as the result of the satisfaction of the conditions set by the predicate of the sentence (proposition) in question by the entity or entities referred to by the term(s), then PEM is no longer a necessary consequence of the truth theory. For the satisfaction conditions of predicates may

58


themselves leave room for partial fulfilment. The predicate dark, for example, is fully satisfied when there is no light at all, but it is partially satisfied when there is a little light, but not enough to justify the predicate light for the entity referred to by the subject term. Thus, one may say that a sentence like This room is dark is partially true of a room where there is just a little light, or that This man is bald is partially true of a man with just a few hairs on his head. Predicates like dark or light are called 'vague' predicates, and it will be clear that natural languages contain large masses of them. A number of theories have been developed during the 20th century, both in logic and in semantics, to account for this kind of vagueness. Best known are the logical approaches. The Polish logician Lukasiewicz devised a three-valued logic, where the third value is to be considered intermediate between truth and falsity. In "Lukasiewicz (1930) it is shown that the intermediate value ('Vz') can be extended to encapsulate all denumerably infinite intermediate values, so that this logic may be taken, in a certain sense (not intended by-fcukasiewicz himself, who applied his logic to modalities) to account for a denumerable infinity of intermediate values that may arise with vague predicates. The same logic, as far as negation, conjunction and disjunction are concerned (but with a different implication), is used in Kleene (1952). The 'fuzzy' logic presented in Zadeh (1975) is more radical, in that it is based on a 'fuzzy' set theory and implies a non-denumerable infinity of intermediate values (see Haack (1978:165) for further comment). Again, we shall leave the technical details aside, but it is clear that the Sorites has had an important impact on modern semantic and logical studies. In this case, the name 'Sorites' has not been forgotten. In fact, it is a standard reference in the vagueness literature. Yet hardly any author seems to realize that the Sorites was first formulated by Eubulides as a contemporary attack on Aristotle's truth theory, and in the context of other similar attacks.

3.5 The Paradox of the Horns We now come to the last of the four paradoxes. It consists of the following logical riddle: What is wrong with the following argument? Major: What you haven't lost you still have. Minor: You have not lost your horns. Ergo: You still have your horns. Besides the somewhat scurrilous aspect of the example, which would have irritated Aristotle (never known for his sense of humour), there is the logical problem: if the argument is correct, then everyone could be rightly branded as either a cuckold or an ex-cuckold. We now know that this riddle shows up the presupposition problem, which has not found a generally accepted solution in either logic or semantics (or pragmatics). The problem is that presuppositions are normally preserved under negation, but that it is often possible to deny the presuppositions as well. This presupposition-cancelling negation is subject to certain conditions, for example that the negation has to be a separate word (not a bound morpheme), must be used emphatically and (for English) constructed with the finite verb form. There are thus (at least) two different possible uses of the negation, in English as in Greek, with differing truthconditional properties. The sentence You have lost your horns carries the presupposition that the addressee once had horns. This presupposition derives from the satisfaction conditions (meaning) of the predicate have lost, which requires that what has been lost must once have been possessed (just


59

like what has been forgotten must once have been known, or whoever is back, must once have been away). If this sentence is made negative in the normal, presupposition-preserving way, we get You haven't lost your horns, which still presupposes that the addressee once had horns. The sentence presented in the Major of the argument, What you haven't lost you still have, expresses the general statement that in all cases, if one hasn't lost Y, one still has Y. This general statement is based on the presupposition that goes with sentences of the type X has lost Y, and on the interpretation of the negation n 't as presupposition-preserving. Again, in the sentence presented in the Minor of the argument, the word not is used ambiguously, since it is not clear whether this is the presupposition-preserving or the presupposition-cancelling (use of the) negation. So the argument is valid (and the addressee is a cuckold) only if both negations, in the Major and the Minor, are interpreted as presupposition-preserving. But if not in the Minor is taken as the presupposition-cancelling not, then the argument is not valid, and the addressee need not worry. This is the answer that Aristotle could have given but didn't, probably because he was, like Strawson (1950), unaware of the possibility of presupposition-cancelling not. Not much is known about the presupposition problem in Medieval philosophy (see Seuren 1998:429). Two authors have so far been traced who deal with presuppositions, which they call propositions praeiacentes ('propositions that lie before'). They are Peter of Spain (13th century), and the Englishman Walter Burleigh (±1275-after 1344). They focus on the presupposition that comes with (the Latin equivalent of) only (the 'exclusivae'), as in (la), and those that are induced by aspectual verbs like stop, as in (7b): (7)a. Only the children laughed, (presupp.: the children laughed) b. Jones has stopped smoking. (presupp.: Jones has once smoked) To what extent the presupposition problem was a topic in Medieval philosophy is not known. But it does seem as if the link with the Horns paradox had already been lost.4 In modern times it is not, again, until Frege (1892) that the problem is rediscovered, and in a different shape and context. Frege does not deal with the presupposition problem specifically, but only mentions it in passing (1892:31-2), in a purely philosophical (metaphysical) context: Idealists and skeptics may already have objected: 'You simply talk [in the sentence 'The moon is smaller than the earth'] about the moon as if it were an object. But how do you know that the definite description 'the moon' has a referent at all? How do you know that anything has a referent?' My answer to this is that it is not our intention to speak of the mental representation of the moon, and that we likewise do not content ourselves with just the sense (meaning) when we speak of the moon. But we presuppose a referent. ... Now we may be wrong in assuming this presupposition, and such errors have indeed occurred in the past. But the question of whether we are always mistaken can remain unanswered here. It is sufficient to point to our intention when we speak or think, to justify our talk about the referent of a sign: there is always the reservation 'in case it really exists', [italics mine, P.S.] We see here that Frege is not concerned with presuppositions like those induced by only or by specific predicates like stop or have lost, but with presuppositions of existence, which he links up with the definite article the. Technically, Frege's problem is that if there is no referent (extension) for the (subject) term in question, there is no truth value (extension) for the sentence,

4

An echo of Walter Burleigh's text on presuppositions (without attribution) is found in Geach (1950), which has, however, provoked no reaction at all in the modern presupposition literature (cp. Seuren 1998:431).

60


since then there is no entity available to test if it satisfies the conditions set by the predicate. In this respect, Frege's account harbours the threat of a violation of Aristotelian bivalence. The point was taken up again in Russell (1905), where Russell presents his famous Theory of [definite] Descriptions. In order to save Aristotelian truth theory and logic from this threat, Russell proposes that the logical analysis of sentences should never contain a definite determiner (article), and that all definite descriptions of the form the so-and-so should be dissolved into an existential quantifier and a few prepositional functions. Thus, the sentence: (8) The present king of France is bald, should be assigned the logical form (9a) or, in technical notation, (9b): (9)a. There is an x, such that x is king of France and x is bald, and for all y, if y is king of France, y is identical with x b. 3x [KoF(x) A Bald(x) A Vy [KoF(y) -> y = x]] Now the negation of this sentence: (10) The present king of France is not bald. does not imply that there is a king of France: (1 l)a. Not [there is an x, such that x is king of France and x is bald, and for all y, if y is king of France, y is identical with x] b. -6x [KoF(x) A Bald(x) A Vy [KoF(y) -» y = x]] Unfortunately, however, says Russell, speakers of natural languages, for reasons best known to themselves, normally interpret (10) not as (11), as they should (and also do in exceptional cases), but as (12), with the negation covering only 'x is bald' or 'Bald(x)': (12)a. There is an x, such that x is king of France and not[x is bald], and for all y, if y is king of France, y is identical with x] b. 3x [KoF(x) A -[Bald(x)] A Vy [KoF(y) -> y = x]] In this interpretation, the existence of the king of France is entailed. But it is not the logical negation of (9). Thus, by the brutal destruction of grammatical structure and the invention of a logical analysis that does not fit the grammatical facts at all, Russell tried to save Aristotelian truth theory and the Aristotelian axiom of bivalence (PET). Russell's rough handling of natural language was criticised by the Oxford philosopher Strawson (Strawson 1950 and later publications). Strawson, restricting himself to existential presuppositions, felt that the Aristotelian PET should be given up in favour of a so-called gapped bivalent system with the values 'true', 'false', and 'undefined'. The value 'undefined' results when one or more presuppositions of a sentence are false. For him, the negation is per se presupposition-preserving, which gives the following truth-table for gapped bivalent not (~): A

~A

T F U

F T U

('A' stands for any proposition, 'T' for 'true', 'F' for 'false, and 'U' for 'undefined.)

A great deal has been said and written about presuppositions and presupposition-cancelling since Strawson put forward his proposal. Later authors have extended Strawson's notion of presupposition to one that encompasses other types of presupposition as well, such as those that are induced by lexical satisfaction conditions, or words like only and even, etc. Various schools


61

of thought have sprung up, in logic, semantics and pragmatics, all differing in their views of how presuppositions are best accounted for. It cannot be our purpose here to try and adjudicate upon this complex question, but it is clear that we have here a major development in the semantics and pragmatics of natural language. And here, too, we see that Eubulides had already got hold of the issue, and in a way that was critical of Aristotle.

4. CONCLUSION The overall conclusion is that, in Aristotle's own day, Eubulides formulated four heavy arguments against Aristotle's theory of truth and therefore his strictly bivalent logic. These paradoxes or riddles were beyond the powers of philosophers and logicians for many centuries, and were consequently ridiculed by many, including Aristotle himself, although some did see their significance. Yet the historical awareness of their author and of the original context in which they were conceived was generally lost. It was only in the 20th century, in the context of the mathematisation of logic and of a renewed interest in natural language semantics, that the paradoxes of Eubulides were rediscovered and recognised for what they are: heavy broadsides on the Aristotelian paradigm. As was said at the outset, the four paradoxes more or less define the entire research programme of modern semantic studies. It is the purpose of this attempt at historical reconstruction to broaden the perspective of those who take an interest in the history and other backgrounds of the semantics, logic and pragmatics of natural language.

REFERENCES Aristotle (1956) Metaphysics. Edited & translated by John Warrington. Dent & Sons, London. Dummett, M. (1973) Frege. Philosophy of Language. Duckworth, London. Frege, G. (1892) Ueber Sinn und Bedeutung. Zeitschrift fur Philosophic und philosophische Kritik 100:25-50. Geach, P.T. (1950) Russell's Theory of Descriptions. Analysis 10:84-88. Gerhardt, C.I. (1890) Die philosophischen Schriften von Gottfried Wilhelm Leibniz. Vol. VII. Olms Verlagsbuchhandlung, Hildesheim. Haack, S. (1978) Philosophy of Logics. Cambridge University Press, Cambridge. Kleene, S.C. (1952) Introduction to Metamathematics. North-Holland, Amsterdam. Kneale, W. and M. Kneale (1962) The Development of Logic. Clarendon Press, Oxford. •fcukasiewicz, J. (1930) Philosophische Bemerkungen zu mehrwertigen Systemen des Aussagenkalkuls. Comptes rendus des seances de la Societe des Sciences et des Lettres de Varsovie, Cl. iii, vol. 23:51-77. (English translation: 'Philosophical remarks on many-valued systems of prepositional logic' inMcCall 1967:40-65.) McCall, S. (ed.) (1967) Polish Logic: 1920-1939. Clarendon Press, Oxford. Popper, Karl R. (1945) The Open Society and its Enemies. London: Routledge & Kegan Paul. Russell, B. (1905) On denoting. Mind 14:479-493. Seuren, P.A.M. (1987) Les paradoxes et le langage. Logique et Analyse 30 (No.l20):365-383. Seuren, P.A.M. (1998) Western Linguistics. An Historical Introduction. Blackwell, Oxford. Stone, I.F. (1988) The Trial of Socrates. Jonathan Cape, London. Strawson, P.P. (1950) On referring. Mind 59:320-344. Zadeh, L.A. (1975) Fuzzy logic and approximate reasoning. Synthese 30:407-428.


4

KEY DATES IN TWENTIETH CENTURY LINGUISTICS R. H. Robins, Professor Emeritus, University of London, England

Abstract Very shortly historians of linguistics in the twentyfirst century will have to examine the structure of the twentieth century's work and thinking in linguistics as a general science. In this examination certain dates are likely to emerge as landmarks or key dates, rather like the four 'breakthroughs' of Hockett (1965). This paper considers the concept of historical key dates, and then, passing through the main persons, schools, and doctrines, suggests some possible dates for consideration in the twentyfirst century.

This paper is bound to be in part a personal one, based on circumstances, experiences, and memories. I am a child of this century. In so far as calendrical labels are useful in historical thought, many would say that in political and social history the twentieth century began in 1918, the previous century with many of its stable institutions having been blown to bits in 1914. I was born in 1921 and I am still alive in the middle of 1998. I was educated in the Classics, and for reasons partly connected with the course of the second world war I became a university teacher of linguistics in 1948. One feature of linguistic studies in the last quarter of the twentieth century, and a most welcome feature, has been the growth of interest, scholarly research, and teaching in the history of the subject. During and since the decade 1965 - 1975 several textbooks and more advanced publications have appeared, and a specialised periodical Historiographia Linguistica (1974). Compare the state of things today with Hockett's statement in the preface to his 1958 Course in modern linguistics that he had omitted all mention of the history of the subject as not appropriate to an introductory book (1958: vii-viii). If we are going to study any aspect of history, political, social, artistic, literary, or scientific, some referential structure must be imposed in advance, and this inevitably involves temporal sequence, as history is about the past in its relations with the present. We are encouraged today to be empathetic, to put ourselves in the position and mentality of particular groups in particular periods, say agricultural labourers in Chaucer's day in Britain, to cite the first chapter of the famous English social history of G. M. Trevelyan (1942). But to do this some temporal identification of the period in relation to the past and present situation of those involved must be clearly envisaged; clearly it did not start at Chaucer's birth and cease

63

64 Langue and Parole in Synchrony and Diachrony immediately at his death. How can we best do this sort of periodization in our branch of historical thinking? The simplest and easiest to envisage is the purely annalistic approach. Events and participants are linked together in particular years, irrespective of any other associations: for example in 1848 King Louis Philippe's regime in France was overthrown, General Zachary Taylor was elected President of the United States, and Thackeray's Vanity Fair was first published. Annals are useful as references and indispensable in checking details, but do not themselves throw much light on the course or courses of history. In the history of humankind centuries are no less arbitrary divisions than are single years. Yet contingently they do sometimes coincide roughly with major changes in thought, social relations, centres of power, and economic conditions. In his recent (1996) book Europe: a history Norman Davies entitles the last three chapters covering the last two centuries, 'Dynamo' (1815 - 1914), 'Tenebrae' (1914 - 1945), and 'Divisa et indivisa' (1945 - 1991), matching, not all that roughly, the nineteenth century and the two halves of the twentieth. This century has not long to last, and the expression 'Present day linguistics' or 'contemporary linguistics' with reference to twentieth century work will soon be, if it is not already, inadmissible. Some historians of the subject have already played safe: Holger Pederson made the first title of his English translation (1931) Linguistic science in the nineteenth century, and in 1991 Malmberg entitled his coverage 'From the Sumerians to de Saussure'. Certainly temporal distance lends a coherent, albeit sometimes a falsely coherent view, to our subject, or indeed any subject. This was effectively brought to my attention some years back in setting questions in a paper on Linguistic Theory as an external examiner. I was told that Bloomfield was now considered as belonging to the history of linguistics rather than to current theory. At the outset of my professional life Bloomfield's Language (1933) was one of the then few essential textbooks. Of course we must envisage historical changes and developments in the field of ideas and ways of thinking as essentially gradual processes, but the historian of linguistics in the twenty first century, however he or she may tackle the subject, will want to look for a few key dates, be they lectures, particular publications, or the beginning or the end of distinctive schools, to form some sort of time frame within which to recount and analyse the course of linguistics in general or some aspect of linguistics. In his 1964 Presidential Address to the Linguistic Society of America, Hockett (1965) made a selection of four key dates covering roughly the past two hundred years of linguistics at that time. The Address was not primarily about history; it was entitled 'Sound change'; but the subject is reviewed within his four temporal divisions. These were: 1786, the year of Sir William Jones's Third Address to the Asiatick Society; 1975, the year of Karl Verner's paper 'Eine Ausnahme der ersten Lautverschiebung', 1916, the first edition of de Saussure's posthumous Cours de linguistique generate, and 1957, the first publication of Chomsky's Syntactic structures. Hockett refers to all these four dates as 'breakthroughs'. This is fair enough: linguistics was never, and could never have been the same since the events referred to. But the term does nevertheless suggest rather a Whig historical approach, whereby each change is for the better. In British history in general the first paragraph of Macaulay's History of England (1871: 1) is a splendid example of Whig history, 'How from the auspicious union of order and freedom sprang a prosperity of which the annals of human affairs had furnished no example'. Bloomfield makes the same sort of assertion, though in a more negative way, at the beginning of his book Language (1935: 2): 'It is only within the last century or so that language has been

Key Dates in Twentieth Century Linguistics

65

studied in a scientific way'. Perhaps the term 'key dates' in the history of linguistics may be less question-begging and more appropriate. What is meant by a key date, or a turning point? Various interpretations are possible. One such is a retrospective view taken by later writers and generally accepted, but not embodying the intention of the original author. Sir William Jones's famous paragraph taken out of context from the rest of the Third Discourse is a case in point. The date 1786 is a date everyone in linguistics knows; it is like 1066 in British history or 1776 in the history of America. But it was not so intended at the time, to set a guideline for the later comparative and historical linguistics, as is so often assumed, especially by those who have not taken the trouble to read the rest of the Discourse or any of the other Discourses that Jones delivered. Jones's ideas on comparative linguistics were far from the later Junggrammatische Richtung; the battle between structure and etymology as criterial evidence had yet to be fought and decided. Jones's 1786 Discourse, like the others, was intended to encourage his colleagues in official positions in the East India Company to learn Sanskrit and other features of Indian culture in order to improve their competence in the conduct of legal cases in that country. And in fact this is how the lecture was first perceived in Europe. Franz Schlegel and his brother August both spoke of their debt to him in their seeking to foster Sanskrit studies in Germany, and among the first publications directly arising from Jones's lectures were grammars of Sanskrit (Wilkins 1808; Carey 1806). Hockett's choice of 1875, the date of Verner's paper, was actually chosen, half whimsically (1965: 185), as lying exactly half way between 1786 and 1964, the year of his Presidency of the Linguistic Society of America; but there was a cluster of publications around that period: the formulation of the 'Law of palatals' independently by several scholars between 1874 and 1878 (Davies 1998: 243), Leskien's Slavic and Germanic inflections, Osthoff s 'Das Verbum in der Nominalkomposition' (1878), and the joint manifesto of Osthoff and Brugmann (also in 1878) which with others set the seal on the Junggrammatiker Ausnahmslosigkeit der Lautgesetze for more than a generation. The controversies aroused by the Osthoff-Brugmann manifesto served to refine the doctrine but not to displace it from the centrality of this dominant aspect of linguistic science. Sweet was later to protest at this German imperialism, as he saw it (1885: vii): "swarms of young programme-mongers turned out every year by the German universities". But unlike Jones's paragraph, the Osthoff-Brugmann manifesto of the 1870s was explicitly set out by its authors as a challenge: if comparative-historical linguistics is worth doing, do it our way with us. The beginning of the full Neogrammarian programme was a key date, both effectively and intentionally. The last two of Hockett's 'breakthroughs' belong to this century. The first of them is the first publication of de Saussure's posthumous Cours de linguistique generale, the assertion of the equal status or even the primacy of synchronic, descriptive linguistics as against diachronic historical linguistics, the darling of the nineteenth century, and the structural theory of synchronic linguistics (strangely, it was left to de Saussure's disciples in the Prague School to apply his structuralism to historical linguistics (cp. Jakobson 1931: 247-267; Trubetzkoy, 'Notes autobiographiques' in Trubetzkoy 1949: xix). One aspect of Prague linguistics deserves especial attention in an historical study of linguistics in the twentieth century, the concept of distinctive features, and this is principally associated with two Prague scholars, Trubetzkoy and Roman Jakobson. In his work and in his life dates (1896 - 1982) Jakobson can be seen to bridge the dividing line between the

66 Langue and Parole in Synchrony and Diachrony Bloomfieldian half and the Chomskyan half of the century, located at 1957. Distinctive features arose from the application of de Saussure's theory of linguistic 'formes' to phonemic theory (cp. Vachek 1966: 18-22). As early as 1936 Jakobson was applying it to morphological analysis; and with his crossing of the Atlantic during the war he began a fruitful alliance with contemporary phonetic research at the Haskins telephonic research station. Distinctive features made themselves central in structuralist phonology, and, passing through the doctrinal divide of 1957, distinctive feature theory provided the main organization of phonological rules in the generative grammar of the 1960s and 1970s. Its syntactic analogue was taken into the area of grammatical analysis. Things change, but they also persist. Trubetzkoy's features were articulatory, Jakobson's were acoustic, but Chomsky and Halle in their descriptive and historical account of the Sound pattern of English (1968) reverted to articulatory features, though these were not the same as Trubetzkoy's had been half a century before. Hockett's fourth and last 'breakthrough' was, of course, Chomsky's first publication of his short Syntactic structures (1957). This last 'breakthrough' he called 'exact accountability', referring to just one of its dominant demands (Hockett 1965: 197). The two publications cited by Hockett are somewhat different as landmarks. It was after de Saussure's death that a few of his students were so impressed with the importance of his lectures that they decided to make them public from their own and other people's notes; the Cours was subsequently translated into all major European languages. Chomsky's Syntactic structures embodies his own challenge, repeated in a number of subsequent writings (1957: 6): 'We shall find that a certain very simple communication-theoretic model of language and a more powerful model that incorporates a large part of what is now generally known as "immediate constituent analysis" cannot properly serve the purposes of grammatical description'. Thereafter linguistics has never been the same again. Certain publications though brief but challenging, and later overtaken by subsequent writings, still serve as the initial markers of a new school or a new trend or a new school later to develop as a dominant component of an intellectual culture. In an earlier paper (1985) I ventured to compare the position of Ayer's Language, truth, and logic (1936) as the initial assertion of full-scale logical positivism with that of Chomsky's Syntactic structures at the start of generative grammar. Hockett identifies his fourth 'breakthrough' as 'exact accountability'. This passes over the introduction or reintroduction of formal transformations into syntax to mark Katz and Postal's assertion (1964) that a linguistic description must, at least ideally, set out all that is comprehended in a native speaker's command of his own language. Now more than thirty years since Hockett's Presidential Address, how do his two twentieth century key dates stand up to criticism? Perceptions have changed over the last thirtythree years, and though Hockett's dates and their historical significance cannot be ignored, they need some reinvestigation. It would be impertinent of me here at this time to try to lay down one or more key dates in the present century as the framework for linguistic historians in the century ahead of us. I am in any case at or near the close of my own professional life, and younger and wiser scholars than I will have to work out the development and the texture of linguistic thought in this century. All that I am trying to do here, after a brief survey of the concept of key dates in historiography, is to draw attention to points that may invite historical attention. One must be sensible and context-sensitive. At one extreme, a reductio ad absurdum, of course, every public lecture, every book, article, and review could signify or be made to


67

signify a change in contemporary science or the reinforcement of the status quo. This might be likened to the 'butterfly's wing' theory of physics. But historically prominent and productive change points are fewer; their number and their criteria for recognition must vary according to the scale of the historical record and the aspect of the subject favoured for close attention. Both of Hockett's key names, de Saussure and Chomsky, declared themselves to be making a necessary change in theory and practice in the general interests of their subject. De Saussure clearly did this in his lectures; he would not otherwise have been resurrected in the original edition of the Cours de linguistique generate (1916) nor in the many editorial revisions by later scholars. In periodization lives are important, but not readily segmentalizable as such. Books and articles are, if they indicate a marked, even though a gradual, change. In philosophy Ayer's Language, truth and logic (1936) has been claimed as an initial mark of logical positivism, at least among English-speaking philosophers. In his British Academy Memoir on Ayer, Quinton ranks this book as (1997:261) lying somewhere near Descartes's Meditations and Berkeley's Principles in the history of philosophy. He goes on: 'In the sixty years since it was published no philosophical book has combined its style, economy, and capacity to incite.' Similar words could have been applied to the Cours and to Syntactic structures. In his early review of this latter book, Lees says as much (1957: 375-407), stressing the revolutionary aspect of Chomsky's linguistic thinking, despite Chomsky's own insistence at the time on his adherence to traditional European grammar, reacting rather against the immediately preceding American structuralism. Though de Saussure must be recognized as the father of structural linguistics, particular aspects of structuralism, especially in America, were directly derived from Bloomfield's Language (1933), whatever his own views on the founding of schools may have been. In his recent Grammatical theory in America Matthews makes 1933 his major starting point. More recently some people have preferred to contrast structural linguistics and linguistic structuralism (cp. Bloomfield's review of the Cours (1923)). In the 1950s there coalesced a number of accepted principles, the reluctance of linguists to say anything much about meanings, the biuniqueness requirement that phonemic analysis must be wholly prior to and independent of morphology, syntax, and semantics, and the insistence on strictly publicly observable data in any linguistic generalization. For a time and among some even the notion of semantic contrast in determining phonemic distinctiveness was barred, and we remember from the early 1950s the controversy in Language on this between Bloch (1953) and the more pragmatic Kenneth Pike (1947; 1952). These principles were set out in Trager and Smith's little book, Outline of English structure (1951), and this was considered so important that copies were issued free of charge to all members of the Linguistic Society of America at the time. In the same year 1951 the same principles were expounded by Harris, though the book had been largely written in the late 1940s. Interestingly, at the end of the preface thanks are expressed to a certain 'N. Chomsky' for help with the proofs. Also in 1951 I recall a personal conversation with the now veteran and most distinguished Professor Murray Emeneau. He said on Harris's Methods: 'I guess this is where we have come to the end of the line'. He was not, I think, saying that it was about to pass away; he was himself in his descriptive work, as in his wartime study of Vietnamese, a structuralist. What he was saying, and what others at the time were saying, was that further theorizing was not what was needed, but the steady application of established theory to the analysis of more languages. This was the view expressed in several textbooks of the time.

68 Langue and Parole in Synchrony and Diachrony We might say, then, that in the eyes of others the year 1951 marked a key date in the completion of structuralist theorizing, apart from the inevitable details. In two other specific fields it is interesting to note that in the early 1820s it was considered that inland water transportation had at last achieved its goal in the operation of locks and canals, just before the railways were to take over, and in the 1860s sailors maintained that the 'Tea clippers' were the perfection of wind-powered ships, on the eve of the supremacy of steam navigation. But what we do know now was that six years later in 1957 and after, American linguistics and worldwide linguistics was to meet a great new theoretical challenge. In view of the contrast of theories facing linguistics at this time it might be said that 1951 and 1957 were both key dates, the one marking the fulfilment of structuralism, particularly in America, the other standing at the initiation of a now continuing volteface in theoretical linguistics. The intervening years were a kind of theoretical calm and a time for reflection. Noticeably several general introductory textbooks were written during these years and published in or soon after them, all based on and expounding post-Bloomfieldian structuralism (Harris 1951; Hill 1958; Hockett 1958; Gleason 1961). We have seen the convenience of locating the date of first publications of a particular book or article as the starting point of a new scientific direction, where this can be historically justified. Hockett's first date 1786 is obvious, though it needs interpretation. He chose 1875 (Verner's Law) as his date for the ascendancy of the Neogrammarians, and 1916 as the start of structural linguistics. His exact choice of these later dates, was, as he admits, chosen partly in play for a bit of 'numerology'; but what matters is the near coincidence of various movements of thought, the so-called 'conspiracy theory' of historical change. During the later 1950s and after there have been a number of attempts at a revision of the dominant structuralist theory, and in a more detailed survey than this can be, their key dates could be identified. I am referring to such movements as tagmemics, functional grammar, stratificational theory, J. R. Firth's contextual semantics and prosodic phonology (cp. further Robins 1989: 149-99, 307-20) and socio-linguistic studies such as those undertaken by Labov (1966), and Bernstein (1971). But the rise and progress of what was first called transformational-generative grammar, later to drop the term transformation, was the most profound. Several major changes were demanded and set out. Hockett chose as his title for the period after 1957 as 'Exact accountability'; the demand, at least ideally, that in a linguistic description and the theory underlying it everything involved in the language must be made explicit; in the words of Katz and Postal: 'A linguistic description of a natural language is an attempt to reveal the nature of a fluent speaker's mastery of that language'. Nothing must just be left to 'commonsense'. But other changes conspired to construct the new theory. 1965, the publication date of Chomsky's Aspects of the theory of syntax, marks two major changes, the reversal of the place of phonetics in the linguistic hierarchy and the refocusing of attention on linguistic meaning. So far from the structuralists' insistence on the prior role of phonetic and phonemic analysis, P(honological) rules now converted prior syntactic rule outputs to transcribed or uttered sentences. And meaning, in the guise of 'semantic interpretation' with much of the traditional apparatus of subject, location, object, qualifier, and the like returned to their accustomed place in linguistics, mostly under the guise of'logical form'. But things were not allowed to rest there. Famously, 'Interpretive semantics' was challenged by 'Generative semantics'. In histories of linguistics primarily focused on the


69

second half of this century, this dispute figures prominently in both Newmeyer's Linguistic theory in America (1980) and the much more recent Western linguistics (1998) by Seuren. In fact Seuren ends his survey with an 'Epilogue' based on the conclusion of it. Their judgements are very different; Newmeyer uses the term 'Linguistic wars', and writes of the deserved collapse of generative semantics, but Seuren regrets that linguistics has been left in a depressing state, due largely to Chomsky's 'petty, cynical, and unprofessional' conduct. Time and future historians will settle the dates and the significance of this within the ongoing context of generative grammar. But above all we must recognize one mode of thinking as marking the most significant difference between the Bloomfieldian generation, key-dated to 1933, and the Chomskyan and post-Chomskyan generation, taken as a broad coalition with all its later variations, Generalized Phrase Structure Grammar, Lexical Functional Grammar Optimalism, Minimalism, and whatever the future may hold. This is the quite radical change in philosophic or perhaps better, scientific outlook. This twentieth century has been a century of scientific thinking, just as was the seventeenth century before it. And in no subject more than in linguistics has the case for scientific status been pressed. This was begun in the later nineteenth century with Schleicher and the Neogrammarians, but it came to its first head with Bloomfield, and its second with Chomsky. In earlier days literary studies were not far removed from the study of language; the ambiguities between the continental and the British use of philology make this apparent. But literary scholarship finds little place in either the Bloomfieldian linguists or among the now many varieties of generative grammarians. Both schools emphasize the importance of treating linguistics as an exact science, but there the contrast begins. Throughout Bloomfield's writings it is clear that science means strictly empirical science. This is made even more clear in a less well known but vitally important piece of writing, Bloomfield's encyclopaedia article (1939), Linguistic aspects of science. He wrote: 'science shall deal only with events that are accessible in their time and place to any and all observers or only with events that are placed in coordinates of time and space' (1939: 13). This is the key text, and all Bloomfield's linguistics flows from it, his antipsychologism, his extreme physicalism in dealing with internal bodily feelings such as hunger and thirst without potentially further reference to potentially observable phenomena ('some of her muscles were contracting' ... 'her tongue and her throat were dry', 1935: 23), and his reluctance to go further into meanings than strict behaviourism would sanction. His understanding of linguistic structure was different from de Saussure's (cp. his review, 1923); structure now concerned itself with the arrangement of phonemes and morphemes within longer sequences, giving rise to the term distributionalism, (Harris 1951: v, 5). Bloomfield's stance was heroic and challenging, for all the justified criticism of later writers. The contrast between Bloomfield and those influenced by him, and Chomsky and those influenced by him is nowhere clearer than in their opposed attitudes towards universal or general grammar, on the classical philosophical distinction of induction and deduction. In Bloomfield's stated opinion general or universal grammar can only be approached after a great deal more information has been made available than we have today. 'The only useful generalizations about language are inductive generalizations' (1935: 20). For Chomksy, writing as a self-asserted rationalist in linguistics (1966: 72-73), universals are not the long awaited product of patient induction, but hypotheses, on a single language if necessary, to be tested deductively for possible falsification and consequent revision. With a number of roughly coincident strands of theory and method coming together to form the central constitution of

70 Langue and Parole in Synchrony and Diachrony generative grammar, various dates propose themselves for the consideration of future historians. Falsification as the test of scientific significance was the assertion of Popper in his Die Logik der Forschung (1935), and Matthews (1993) asks whether the appearance of the English translation in 1959 may have been a significant date in the early development of generative theory in the English-speaking world. The practical results of these two stances have also been antithetical. The Bloomfieldian generation achieved descriptions of a vast array of languages hitherto scarcely known, especially among the Indian languages of the American continents, greatly increasing our available data base of the languages of the world. The generative grammarians have greatly deepened the analysis of the better known European languages, on which they have mainly concentrated, most of all modern English, with grammatical rules covering generalizations either not made before or assigned to stylistic factors. It is through the work of these two American linguistic scholars, Bloomfield and Chomsky, that in the course of the twentieth century America has become the centre of linguistic science, for so long the property of continental Europe. This paper has been intended primarily to consider the notion of key dates or turning points in relation to twentieth-century linguistics, and it is certainly not for me to prescribe any such years for future historiographers of linguistics. But perhaps we might, in a general coverage of what has been thought and written, confirm 1916, the year of the first publication of de Saussure's Cours, and follow it by the setting up of the Prague Linguistic Circle in 1926 (Vachek 1966: 26), the culmination of its work in 1939, when Trubetzkoy's Grundzuge der Phonologie first appeared; in America 1933 marks the date of the first American edition of Bloomfield's Language, initiating the rise of what became structuralism at its high point in 1951. Since 1957 many changes have occurred in linguistic theory and practice, but it may stand as the year when linguistics as a whole turned in a new direction taking almost every aspect of the subject with it to the end of the century and of the second millennium.

REFERENCES Ayer, Alfred J. (1936). Language, truth and logic. Gollancz, London. Bernstein, B. (1971). Class, code, and control. Routledge, London. Bloch, Bernard (1953). Contrast. Language 29, 59-61. Bloomfield, Leonard (1923). Review of de Saussure, Cours de linguistique generate. Modern Language Journal 8, 317-9. Bloomfield, Leonard (1933). Language, first British edition 1935, Allen and Unwin, London. Bloomfield, Leonard (1939). Linguistic aspects of science. International Encyclopaedia of Unified Science 1.4. Chicago University Press, Chicago. Carey, William (1806). Grammar of the Sungskrit language. Mission Press, Serampore. Chomsky, Noam (1957). Syntactic structures. Mouton, The Hague. Chomsky, Noam (1965). Aspects of the theory of syntax. MIT Press, Cambridge, Mass. Chomsky, Noam (1966). Cartesian linguistics. Harper and Row, New York. Chomsky, Noam and Morris Halle (1968). The sound pattern of English. Harper and Row, New York. Davies, Anna Morpurgo (1998). History of linguistics 4. Longman, London. Davies, Norman (1996). Europe. Oxford University Press, Oxford.


71

De Saussure, Ferdinand (1916). Cours de linguistique generate. Payot, Paris. Gleason, Henry A. (1961). An introduction to descriptive linguistics. Holt, Reinhart and Winston, New York. Harris, Zellig S. (1951). Methods in structural linguistics. Chicago University Press, Chicago. Hill, Archibald (1958). Introduction to linguistic structures. Harcourt Brace, New York. Hockett, Charles F. (1958). A course in modern linguistics. Macmillan, New York. Hockett, Charles F. (1965). Sound change. Language 41, 185-204. Jakobson, Roman (1931). Prinzipien der historischen Phonologic. TCLP 4, 247-267. Jones, Sir William (1786). Works (ed. Anna Maria Jones) 1, Discourses and other papers. Robinson and Evans, London. Katz, Jerrold and Paul Postal (1964). An integrated theory of linguistic descriptions. MIT Press, Cambridge Mass. Labov, William (1966). The social stratification of English in New York City. Center for Applied Linguistics, Washington. Lees, R. B. (1957). Review of Chomsky, Syntactic structures. Language 33, 375-407. Leskien, August (1876). Die Declination im Slavisch-Litauischen und Germanischen. Hirzel, Leipzig. Macaulay, Thomas (1971). History of England 1. Longman, London. Malmberg, Bertil (1991). Histoire de la linguistique de Sumer a Saussure. Presses Universitaire de France, Paris. Matthews, Peter H. (1993). Grammatical theory in the United States of America. Cambridge University Press, Cambridge. Meillet, Antoine (1903). Introduction a I'etude comparative des langues indo-europeennes. Hachette, Paris. Newmeyer, Frederick (1980). Linguistic theory in America. Academic Press, New York. Osthoff, Hermann (1878). Das Verbum in der Nominalkomposition. Costenoble, Jena. Osthoff, Hermann and Karl Brugmann (1878). Morphologische Untersuchungen 1. Hirzel, Leipzig. Pederen, Holger (1931). Linguistic science in the nineteenth century. Harvard University Press, Cambridge Mass. Pike, Kenneth L. (1947). Grammatical prerequisites in phonemic analysis. Word3, 155-172. Pike, Kenneth L. (1952). More on grammatical prerequisites. Word 8, 106-121. Popper, Karl (1935). Logik der Forschung, English translation 1959, The Logic of scientific discovery. Hutchinson. London. Quinton, Anthony (1997). Alfred Jules Ayer. Proceedings of the British Academy 94, 255-284. Oxford University Press, Oxford. Robins, Robert H. (1985). Linguistics in 1984: retrospect and prospect. Grevatt and Grevatt, Ne wcastle-upon-Tyne. Robins, Robert H. (1989). General linguistics: an introductory survey (fourth edition). Longman, London. Seuren, Pieter A. (1998). Western linguistics. Blackwell, Oxford. Sweet, Henry (1885). Oldest English texts. Clarendon Press, Oxford. Trager, George and Henry L. Smith (1951). Outline of English structure. Battenburg Press, Norman, Oklahoma. Trevelyan, George M. (1942). English social history. Longman, London. Trubetzkoy, Nikolaj (1939). Grundziige der Phonologic. TCLP 1.

72 Langue and Parole in Synchrony and Diachrony Trubetzkoy, Nikolaj (1949). Principes der phonologic (French translation with additional material by J. Cantineau), Klincksieck, Paris. Vachek, Josef (1966). The linguistic school of Prague. Indiana University Press, Bloomington. Verner, Karl (1875). Eine Ausnahme der ersten Lautverschiebung. Zeitschrift fur vergleichende Sprachforschung23, 97-130. Wilkins, Sir Charles (1808). A grammar of the Sanskrita language. Black, Parry and Kingsbury, London.

II LANGUE VERSUS PAROLE


5

SYSTEM DESCRIPTION PRESCRIPTION?

OR

SYSTEMATIC

Bozena Bednafikovd, Palacky University, Olomouc, Czech Republic

Abstract The paper does not solve the dilemma of whether there is or is not a dichotomy langue/parole. It simply takes into account the system and the system description, as the major reference grammar of modern Czech does. What is innovative in the second volume of this thesaurus is the shift from the surface to the underlying level in grasping the structure of word forms. The conception of the stem suffix reveals the parallelism and isomorphy between synthetical and analytical verbal forms, it explains some examples of homonymy, etc. On the other hand, the mechanical and systematic application of the newly revealed deep structure rules to the description of the adjectival forms prescribes to the system of modern Czech an unnatural regularity and an incomprehensible uniformity, disturbing in consequence the system itself.

Even though this paper has been classed by the organizers of the conference under the topic Langue versus Parole, even though it actually derives its inspiration from the main topic of the conference, it neither attempts nor would dare to solve the dilemma of whether there is or is not a dichotomy langue/parole. The dichotomy in question, originally the first truth of de Saussure's system of general linguistics and which is, to be precise, a trichotomy, that is langage - langue - parole, has been undergoing a long and complicated development. One of the crucial points seems to be what is actually understood by langue. Is it a code (a possibility prior to individual speech acts)? Or a construct (the product of scientific, here linguistic classification)? In other words, is it ascribed an ontological or gnoseological (noetical) status? Unlike some other papers presented at the conference this paper does not try to answer the question of whether there is or is not a system in a language. The aim of this paper is rather modest. It simply rests upon the gnoseological approach to langue and identifies with the system the structure which is to be more or less approximatively revealed, grasped and described. The resultant structure, a quasi-structure, could/should represent an approximate model of a language.

75

76


It is a procedure which is claimed to have been followed by the major reference grammar of modern Czech, the so called ,,academy grammar" - see Mluvnice cestiny I, II, III (1986a, 1986b, 1988). The so called ,,academy grammar" of Czech is composed as an academy description of the contemporary state of the literary/standard language. It is a major grammar work, in many features pioneering, written on a highly theoretical level regarding both the Czech and worldwide trend of treating so-called communicative grammars. This ,,thesaurus" of Czech grammar has also influenced in principle the universities' concept of teaching grammar and linguistics in general. I have been working on a project called The Reader of the Academy Grammar of Czech, which in the form of a terminological dictionary and educational texts will enable students to read about, to understand and to grasp both the basic principles and the basic roots of the above mentioned language description. What is at the centre of my attention, at least in the first stage of the project, is the formal morphology, especially the construction of the word forms and morphemic analysis of the word forms. (Note - the second volume of the thesaurus of the Czech language is devoted to morphology, both functional and formal.) And so it is in this paper. What is innovative here with the morphological system description is the shift from the surface to the underlying level in understanding and grasping the structure of the word forms. This can be seen perfectly in the verbal inflection description. Let us start with the verbal inflection, as the stem suffix concept seen below has been most consistently and integrally applied here. The structure of the verbal forms is based upon a consistent stem principle. The central concept/term in this kind of morphological description is the stem suffix/STEM FORMANT. What in this concept is emphasized with the stem suffix is its connective function, i. e. it works as a link between the morphological basis of the form and the grammatical morpheme(s). Associated with the aforesaid statements about the stem suffix appear to be the following three explanatory remarks: 1) This sort of understanding of the stem suffix goes back to Trubetzkoy's work - see Trubetzkoy (1934) - and was also adopted for example by the Czech morphologist Komarek - see Komarek (1978). 2) In its connective function the stem suffix has even been given a specific name, i. e. CONNECTEME. 3) The stem suffix/connecteme is not a "fully valuable" morpheme, as it bears neither lexicological nor morphological or even derivational meaning. It is defined functionally (by its connective function) and ranks with the so-called submorphemes. Therefore, in the academy grammar, two stems are identified: present stem and past stem. A thorough description of the verbal inflectional system cannot be presented in this paper. For illustrative purposes the present tense paradigm of the conjugational type NEST follows, serving as data for further evaluative commentary:

System Description or Systematic Prescription?

77

Traditional Grammars Nest Sg.

1st p. nes-u 2nd p. nes-es 3rd p. nes-e

The present stem is identical with the morphological basis, i. e. there is no stem suffix (-u, -es, e are the endings). The description of the verbal inflection in traditional grammars identifies the stem suffix just in the 2nd and 3rd verbal classes (there are five of them as far as the present stem is concerned). Hence the system of the personal morphemes (endings) is inevitably rich, numerous and highly variant and differentiated.

Academy Grammar of Czech Nest Sg.

1st p. nes-0-u 2nd p. nes-e-s 3rd p. nes-e-0

The form of the 3rd person singular (indicative, present tense) is identical with the present stem, i.e. there is the zero morpheme ending (the segment -e- being classified as the stem suffix). It follows that the system of the personal morphemes (endings) is considerably simpler and more consistent and homogeneous: Nest Sg.

1st p. -u/-i,-m 2nd p. -s 3rd p. -0

P1.

-me -te -ou/-i

What was added to this unified repertoire of personal endings are special rules defining the alternations of the stem suffixes in various verbal forms. In turn, this conception of the stem suffix revealed the parallelism and isomorphy between the synthetical and analytical verbal forms in Czech - see Komarek (1987): Present tense (indicative) Sg1st p. nes-0-u 2nd p. nes-e-s 3rd p. nes-e-0

Past tense (indicative) nesljsem nesljsi nesl 0

Conditional nesl by-ch neslby-s nesl by-0

The presupposed zero personal ending in the 3rd person present tense indicative corresponds to the zero personal morpheme in analytical verbal forms as the past tense indicative and conditional.

78


The crucial point of the conception, the stem suffix, and corresponding interpretation of the 3rd person sg verbal form also helps to explain some examples of surface level homonymy, i.e. a completely isolated sort of homonymy in the verbal system (from the synchronic point of view unexplainable): Compare the conjugational type TRPET: On trpi Sg. 3rd p. trp-i-0

Oni trpi P1. 3rd p. trp-0-i

The interpretation (and explanation) is as follows: There is the zero ending in the 3rd person singular as it is regular with other conjugational types. The stem suffix is fully realized as it is common before either zero or consonantal endings. On the contrary, the 3rd person plural has the vowel ending -i, i.e. the same ending which appears, for example, in the conjugational type DELAT (oni del-aj-i) or in the bookish variant of the 3rd person plural in the type MAZAT (oni maz-0-ou/i). According to the newly adopted stem suffix alternation rules the stem suffix has zero realization if a vowel ending follows. To sum up, the formal morphology description has been shifted to a more abstract level, preferring the deep structure of the verbal forms over the surface structure. On the other hand, the mechanical application of these newly revealed deep structure rules to the description of other parts of speech may be ,,sold at a loss". For it appears the adjectival inflection description in the academy Grammar of Czech.

The Adjectival Inflection The adjectival inflection differs widely from the substantive one but it displays some agreement with the verbal inflection. The differential feature here is that there is/is not the stem suffix. While most of the substantive declinational types do not have any stem suffix (from the synchronic point of view) the construction of the verbal forms is based upon the stem principle. So it is true about adjectival inflection, which, being based upon the stem principle, seems to be closer to the verbal inflection. The so called hard adjectives (of the type MLADY) have two complementary variants of the stem suffix (/-e-/, /-i-/), which are bound with the grammatical opposition sg. vs. pi. - see Komarek(1972): Mlady (masc.) Sg. Norn, mlad-0-i Gen. mlad-e-ho Dat. mlad-e-mu Ace. mlad-e-ho Loc. mlad-e-m Instr. mlad-0-im

PL mlad'-0-i mlad-i-ch mlad-i-m mlad-0-e mlad-i-ch mlad-i-mi

The soft adjectives (of the type JARNI), on the other hand, have only the one variant of the stem suffix - /-i-/:

System Description or Systematic Prescription?

Jarni (masc,) Sg. Nom. jar-n-0-i Gen. jar-n-i-ho Dat. jar-n-i-mu Acc. jar-n-i-ho Loc. jar-n-i-m Instr. jar-n-0-im

79

P1. jar-n-0-i jar-n-i-ch jar-n-i-m jar-n-0-i jar-n-i-ch jar-n-i-mi

For explanation of the stem suffix alternations it is necessary to take into account the phonological characteristics of the adjectival endings. It is obvious that the stem suffix has a positive (that is phonemic) realization before consonantal endings, and a negative (i.e. zero) realization before vocalic endings. This conception of the stem suffix again enables one to grasp some examples of the surface structure homonymy (see loc. and instr. of the type JARNI, note: The extension of the plural -i- to singular in the type MLADY would be abnormal and unmotivated, it would not correspond to the pronominal inflection). Even if the same mechanism of dissimilation has been applied here as in the verbal inflection the academy grammar adopted a different mechanical and non-systemic rule: Whatever may follow the morphological basis of the adjective has the status of the stem suffix. Hence the rule violated the deep structure interpretation of the surface homonymy - see loc. sg. jar-n-i-m, instr. sg. jar-n-i-m. But more seriously, it shifts the Czech morphological system towards the sphere of the ,,introflectional" languages: a) The opposition of grammatical number is expressed inside the word, not with the help of the root alternation, but with the help of the stem suffix: Nom. sg. mlad-a-0

Nom. pi. mlad-e-0

b) The gender differences are expressed again in the stem suffix: Nom. sg. masc. mlad-i-0

Nom. sg. fern. mlad-a-0

Nom. sg. neutr. mlad-e-0

c) Even the case forms are expressed here with zero endings (!), which in a language like Czech, a language with very rich inflection (the richest of all the Slavonic languages), sounds quite absurd: Mlada Sg. Nom. Gen. Dat. Ace. Loc. Instr.

mlad-a-0 mlad-e-0 mlad-e-0 mlad-ou-0 mlad-e-0 mlad-ou-0

80


On the face of the survey presented above it is quite absurd and unacceptable. This application not only deprives the system of the grammatical endings, but also establishes a heterogeneous set of stem affixes whose alternations are inexplicable, materially not derivable one from another.

Summary and Conclusion The mechanical, highly systematic yet not systemic application of the stem suffix conception to the description of the adjectival forms prescribes to the system of modern Czech an unnatural regularity and simplicity and an incomprehensible uniformity, disturbing in its consequences the system itself.

REFERENCES Komarek, M. (1972). Stavba tvaru adjektivni a pronominalni flexe v spisovne cestine. Slovo a slovesnost, 33, 139n. Komarek, M. (1978). Prispevky k ceske morfologii. AUPO, Prague. Komarek, M. (1987). Ke dvema koncepcim stavbyjednoduchych slovesnych tvaru v cestine. Studia Bohemica IV, 105n. AUPO, Prague. Mluvnice cestiny I, II, III. (1986a, 1986b, 1988). Academia, Prague. Trubetzkoy, N. S. (1934). Das morphologische System der russischen Sprache. Travaux Linguistiques de Prague 5, 14n., Prague.

6

COMPLETENESS PARADIGMS

AND SYMMETRICITY

or

Mati Hint, Tallinn Pedagogical University, Estonia

Abstract The completeness and symmetricity of morphological paradigms may be the result of theorizing in grammars of literary languages. In extreme cases new forms (e.g., noun cases, verb forms) may be added to paradigms as the result of artificial form creation by grammarians. On the other hand, some morphological forms which are extinct or inactive in speech may continue to live in literary language. The formation of paradigms in written grammars depends upon the intuition of grammarians; especially the interpretation of analytic verb forms and collocations with auxiliaries may be fabricated. Obviously these phenomena may be interpreted with the help of a Saussurean understanding of differences between speech and language: the postulated (artificial) forms of literary language may turn into grammatical facts of spontaneous speech during subsequent generations, while some other forms in well-formed paradigms may have but marginal (or zero) importance in speech. Therefore, the concept of language has ambiguous meaning for literary languages.

It is easy to demonstrate how sound changes in the history of language break the order and symmetry in paradigms, which usually (in grammar books) are considered to be regular and symmetrical. Phono tactically impossible or homonymous grammatical forms, obscure stem inflections, and fusional processes in morphology are the usual outcome of sound changes. Sound changes are, of course, connected with speech events. It remains disputable how they are represented on the more abstract level of langue. In the generative framework of the theory of language it is possible to consider relatively recent sound changes as superficial rules of speech production which do not influence the underlying system of language - until the restructuring of the underlying system has taken place. But eventually a chain of sound changes inevitably leads to less systematic paradigms with gaps, irregularities, and asymmetry. Speech ruins the systematic nature and completeness of paradigms. Yet the paradigms in languages, especially in literary languages and their grammars, are interpreted as systematic. How is this achieved? What are the mechanisms of curing the injured paradigms, or creating a full set of forms in a paradigm? The most common and most thoroughly investigated mechanism for curing paradigms is spontaneous analogy. But analogy cannot explain everything.

81

82


In the following I will present two examples in order to demonstrate that sometimes the speaker, or grammarian, is free to choose between different models of morphological forms. It may be accidental which of them develops into a paradigm or paradigmatic form in language.

1.1ST PERSON PL NEGATIVE IMPERATIVE IN ESTONIAN There is a rather peculiar imperative form for the 1 st person pi both in Finnish and Estonian. In Finnish grammar the imperative with the meaning 'let us' has the form with the ending -kaamme or -kaamme (vowel harmony), where -mme is the regular personal ending (in the literary language) and -kaa or -had is the formant of the imperative mood: menkddmme, menkdmme (men/kdd/mme, men/kd/mme) 'let's go'. In the Estonian literary language the corresponding form is mingem (min/ge/m), but this form is perceived as highly pathetic, almost artificial. In dialectal speech it is practically absent, and the form is not used in neutral literary speech, but it is a regular translation counterpart, for example, in the translations of the Bible (since 1739) and in different translations of the New Testament both in Northern and Southern Estonian. The morphological form of interest is the negative form of imperative 1st person pi, which is formed with a special negation auxiliary. In Finnish this form is dlkddmme menko, where quite characteristically for Baltic-Finnic languages the personal ending -mme is represented in the auxiliary and not repeated in the main verb, but the formants of imperative mood -kdd(-kaa) and -ko(-ko) are repeated in both the auxiliary and the main verbs. In the older Estonian literary language this form was different from its Finnish counterpart. John 19: 24 (Let us not tear it, but cast lots for it...) Finnish: Alkddmme (dl/kdd/mme) leikatko (leikat/ko)... Estonian Bible 1739, and the North Estonian New Testament 1825, 1865 - Arge (ar/ge/0/) kiskugem (kisku/ge/m/) sedda mitte lohki, vaidheitkem (heit/ke/m) liisko... South-Estonian New Testament 1872: Arge loikagem (loika/ge/m) sedda mitte katski, enge heitkem (heit/ke/m) liisko... In the early Estonian-language Estonian grammars of the late 19th and early 20th century we find only congruent forms: drgem kiskugem, where the suffixes of imperative mood and personal ending are repeated in both auxiliary and in main verb form (dr/ge/m kisku/ge/m). This is deviant from both the Finnish grammar and the earlier usage in literary Estonian texts. As the form is not commonly used, it is difficult to make any conclusive statement about the origin of this modification, it is as obscure as the origin of non-congruent forms in the earlier editions of the Bible. In colloquial modern Estonian, instead of the imperative the indicative mood forms are used: kisu/me, negative dr/me kisu/me, and in the speech of the younger generation kisu/me and dr/me kisu (with a zero personal ending in the main verb). There has to be some reason why both in Finnish and Estonian the regular 1st person pi imperative is not used - in colloquial Finnish instead of tehkddmme se! 'let's do it!' the passive form (without the imperative marker) is used: tehdddn se! Thus in the 1st person pi the suffix of the imperative is avoided in both languages, maybe for semantic reasons.

Completeness and Symmetricity of Paradigms

83

Should we try to categorize these forms, we will have the following chart for the negative imperative 1st p. pl: Imperative marker Auxiliary Main verb Finnish al/kaa/mme leikat/ko Earlier Bible Estonian ar/ge kisku/ge/m (ar/ge loika/ge/m) Modern Estonian grammar ar/ge/m loiga/ke/m Colloquial Estonian usage ar/me loika/me Colloq. Est. of young people ar/me loika

Personal ending Auxiliary Main verb

+

+

+

-

+

+

-

+

+

+

+

+

-

-

+

+

-

-

+

-

Or otherwise: Auxiliary Imperat. Pers.ending Finnish al/kaa/mme leikat/ko Earlier Bible Estonian ar/ge kisku/ge/m (ar/ge loika/ge/m) Modern Estonian grammar ar/ge/m loiga/ke/m Colloquial Estonian usage ar/me loika/me Colloq. Est. of young people ar/me loika

Main verb Imperat. Pers.ending

+

+

+

-

+

-

+

+

+

+

+

+

-

+

-

+

-

+

-

-

In cases like this it is very difficult to put all these things in a theoretically good order to say how the underlying paradigmatic relations may lead to modifications in one or another direction. It is equally difficult to say how speech events, usage, creates such paradigmatic forms, as the frequency of these forms is low. It is obvious that forms for the negative imperative exist in both langue and parole, but their interrelations are simple only on a rather general and under specified level. Probably the underlying language system provides only a possibility of combining certain grammatical meanings, and there may be several noncontradictory and nonambiguous means of realizing those combinations, i.e., different forms with the same meaning may be produced. It is inappropriate to explain such different forms as resulting from analogy. This example shows how a language may choose more or less redundant forms in order to express a certain combination of grammatical meanings.

84


Very often languages with rich morphology have parallel inflectional forms, that is, the same combination of grammatical meanings may be expressed by more than one resulting form. This is true of Estonian and Finnish morphology: for example, aasta/te/st = aasta/i/st ' year, pl., elative case (Est.)', kartano/i/tte/n, kartano/i/de/n, kartano/je/n 'manor house, pl. gen. with double plural suffixes in the two former forms (Finn.)'. Some more complex cases of parallel forms have been introduced into the Estonian literary language after the 1920s: 'I would have come', ole/ksi/n tul/nud (auxiliary to be + conditional mood + 1st person sg + to come + past participle), obtained a parallel form tul/nu/ksi/n (to come + past participle + conditional mood + 1st person sg). The dialectal basis of such forms is very weak, in the literary language they are perceived as artificial. The language system in the meaning of langue seems to operate with rather abstract units whose main content is meaning. Grammar on this level provides rules for the combination of these meanings. Equally unclear is the question of analytical paradigms with very complex grammatical meanings, such as for example the perfect forms of the negative imperative. In the grammars of Estonian of the beginning of this century this paradigm is absent, but Finnish grammars and the Estonian grammars of the 1930s present forms like the Finnish 3rd person al/koon ol/ko leikan/nut and the Estonian ar/gu ol/gu loiga/nud. It remains a question, to what extent such forms may be considered as belonging to an established morphological paradigm of the language. Again, it seems more appropriate to think that in the underlying system there exist grammatical meanings, whereas in speech there occur concrete needs to combine these meanings, and different ways of expression may evolve (this is self-evident on the sentence level). How these possibilities are reflected in grammar books of literary languages largely depends on grammarians.

2. THE EVOLVING 'DESCRIPTIVE PRESENT' IN ESTONIAN No modern Estonian grammar has noticed the impetuous evolving of new analytical verbal forms parallel to the most common verbal paradigm - the present indicative. In Estonian we should have but one synthetic paradigm for the present tense, fully unmarked in the active indicative, except for personal endings (which are dropped in negative forms): Mina vddri/n usaldust Sina vaari/d usaldust Tema vddri/b usaldust Meie vaari/me usaldust Teie vaari/te usaldust Nemad vaari/vad usaldust

'I deserve trust' 'You ... 'He/She ... 'We ... 'You ... ' They ...

In modern newspaper language, side by side with these most common forms a new and excrescent analytic expression is used, although there is practically no semantic need for new forms with the auxiliary to be in the present and the main verb in the present participle form. The participle agrees with the person in number:

Completeness and Symmetricity of Paradigms Mina ole/n usaldust vddri/v Sina ole/d usaldust vddri/v Tema on usaldust vddri/v Meie ole/me usaldust vddri/va/d Teie ole/te usaldust vddri/va/d Nemad on usaldust vddri/va/d

85

'I am deserving trust' 'You are deserving ... 'He/She is deserving ... 'We are deserving ... 'You are deserving ... 'They are deserving ...

The 3rd person singular is the most common form used in such an analytic fashion. This form is the basis for the expansion of analytic forms. It should be acknowledged that there are some prerequisites for the use of the participle form in the present tense. First, there are traditional present participle forms (both active and passive) which are used as common adjectives, often as predicatives: meeldiv 'pleasant, "likable"', elav 'lively', huvitav 'interesting', haarav 'exciting', voluv 'charming', kusitav 'questionable', ravitav 'curable (passive)', soovitav 'desirable (passive)', etc, which have been used as adjectives in connection with the third person in traditional grammar, as well as with other persons (very much like in Germanic languages): Ma ei ole siin soovitav Sa oled voluv See haigus on ravitav

Tm not wanted here' 'You are charming' 'This disease is curable'

Only in rare cases may these participles be replaced by finite verb forms but the result is not stylistically on a par with participles: Sa oled voluv 'You are charming' or Sa volud (mind, meid) 'You charm (me, us)' Secondly, there are phraseological compounds with present participles as their second components: rahulolev '(self)complacent', valjapaistev 'outstanding', paljukonelev 'expressive (much-saying)', eemaletoukav 'odious, "off-pushing"', kasutoov 'profitable (bringing profit)', etc. These participles are mostly used as adjectives, syntactically as attributes: See romaan on tdhelepanuvddriv saavutus. 'This novel is an achievement deserving attention (attention-deserving achievement).' But when one wanted to drop the head word achievement, in traditional grammar it was more appropriate and in most cases more neutral to use a finite present tense form: See romaan vddrib tdhelepanu 'This novel deserves attention.' Not any more nowadays: in the language of the media one is very likely to see See romaan on tdhelepanuvddriv 'This novel is attentiondeserving'. In these cases the present participles function as adjectives, not as components of analytical verb forms. Anyway, the formal model for using present participles as components of analytical verb forms was created due to this traditional usage of participles as adjectives. Given (presumably) the impact of English, this model has begun to expand, being used with more and more verbs, and gradually filling more and more patches in the large field of verbal forms.

86


Present tense: Mina ole/n kirjanduse poole kaldu/v 'I am inclining towards literature' (instead of Mina kaldu/n kirjanduse poole 'I incline towards literature'). Aktsiate hind on langev (pro lange/b) 'The price of shares is falling (falls)'. Present tense negative: Minevik ei ole (auxiliary) asjassepuutu/v (ei puutu asjasse) 'The past has nothing to do with these things'. Past tense: Javlinski oli (past tense of auxiliary) vigu tunnistav (tunnista/s vigu) 'Yavlinski was "mistakes-admitting"'. Artikkel oli huvipakkuv (participle) 'The article was interest-affording' (instead of Artikkel pakku/s huvi 'The article afforded interest') Perfect tense with past participle: Koloriit on minu toodes kogu aeg domineeriv ol/nud (on domimerinud) 'Colour has always been dominating in my works'. Perfect + comparative of participle: Seekord ma olen olnud (perfect tense of auxiliary) otseutlevam (comparative of participle) 'This time I have been more outspoken ("more directly-saying")' (instead of Seekord olen ma delnud otsemini 'This time I have said it more directly'). Negative quotative (modus obliquus): Ma ei ole/vat usaldust vaariv (Ma ei vadri/vat ualdust) T was said not to be deserving trust' In questions: Kellele see on (auxiliary) kasutoov (participle) 'For whom is it profit-bringing?' (instead of Kellele see too/b kasu? 'Whom does it bring profit to?'). Comparison: See voitlus oli kauakest/va/m (comparative of present participle of presumed verb kaua kestma 'to last long') ja sojalist kuulsust too/va/m 'This struggle was more long-lasting and more military-fame-bringing', instead of kesti/s (past tense of main verb) kauem (comparative of adverb) and toi (past) rohkem sojalist kuulsust 'brought more military fame'. In derivatives: OMine on tornide valjapaistvus (abstract noun from present participle) 'Important is the visibility ("visible-beingness") of towers'. The question is, why is the participle expanding? From the viewpoint of traditional stylistics the analytic expressions with the participle are cumbersome and not so easy to analyse. The strategy of analysis has to be changed in order to make these constructions productive. Usually analogy does not create entirely new paradigms with a new grammatical meaning. Therefore, analogy is not a good explanation for this development. Probably the IndoEuropean languages (Russian, English, German) with their more excessive use of the present participle have influenced the emergence of corresponding formal means in the Estonian media today. The question why these new means of expressing the 'descriptive present' are unfolding must remain without a definite answer. A tentative answer is again that as new formal possibilities evolve there is a chance that they will be used. The common inclination of bureaucratic style to prefer nominal constructions obviously fosters this kind of shift. The more these analytic forms are used the stronger their tendency to develop into full-scale

Completeness and Symmetricity of Paradigms

87

paradigms, either with the same meaning as synthetic present forms, or with a split in the meaning after due time (with the participial descriptive present tending to express a more permanent state). On the abstract level oflangue the main role belongs to semantics, but in realizing semantic intentions parole is in ever-changing movement, in any language, at any time. Sources Different editions of the Estonian Bible and the New Testament; common grammars of Estonian and Finnish; media use of Estonian in the 1980s-90s.


7

LANGUE UND PAROLE - EINE NEUBESTIMMUNG UNTER DEN GESICHTSPUNKTEN VON REFERENZ UND GELTUNG RolandHarweg, Ruhr-Universitat Bochum, Germany

Zusammenfassung Der Artikel macht den Vorschlag, Langue-Einheiten (zu denen ich iibrigens nicht nur Worter, sondern auch Satze und Texte rechne) als sprachliche Einheiten ohne (aktuelle) Referenz und (adressatenbezogen-pragmatische) Geltung, Parole-Einheiten hingegen als sprachliche Einheiten mit (aktueller) Referenz und (adressatenbezogen-pragmatischer) Geltung zu definieren. Die Kriterien der Referenz und der Geltung zeitigen bei vielen Einheiten die gleichen, bei einigen aber auch unterschiedliche Ergebnisse der Zuordnung. Im letzteren Fall entscheide ich mich fur das Geltungskriterium. Die Langue-ParoleUnterscheidung wird auch auf das Phanomen der Metasprache iibertragen, und das Langue-Konzept wird (was sich aus den verwendeten Kriterien ergibt) auf den Bereich der Performanz ausgedehnt.

ZUSAMMENFALL DES REFERENZ- UND DES GELTUNGSKRITERIUMS Eine alltagliche Situation Stellen wir uns einmal folgende Situation vor. Ein junger Deutscher, der des Franzosischen nicht machtig ist, will einer jungen Franzosin, die des Deutschen nicht machtig ist, eine Liebeserklarung machen. Er fragt seinen Freund, der sowohl des Deutschen als auch des Franzosischen machtig ist: "Du, was heiBt eigentlich auf Franzosisch Ich Hebe dich?", und der Freund antwortet ihm: "Je t'aime". Der junge Deutsche geht daraufhin zu der jungen Franzosin und sagt - man ist versucht zu sagen: ebenfalls -: "Je t'aime". In dieser Episode konimt zweimal die AuBerung Je t'aime vor, aber in unterschiedlicher Funktion: einmal als bloBes Ubersetzungsaquivalent eines bestimmten deutschen Satzes und einmal als eine regelrechte AuBerung, als eine aktuelle Liebeserklarung. Der Unterschied zwischen den beiden Funktionen ist, schon rein intuitiv, gewaltig, und er laBt sich fassen unter Rekurs auf die Begriffe der Referenz und der Geltung, genauer: unter Rekurs auf die Begriffspaare 'Referenzhaltigkeit vs. Referenzlosigkeit' und 'Geltungshaltigkeit vs. Geltungslosigkeit'. Als bloBes Ubersetzungsaquivalent ist die AuBerung referenz- und geltungs 1 o s, als aktuelle Liebeserklarung hingegen ist sie referenz- und geltungs h a 11 i g. In

89

90


dem Ubersetzungsaquivalent bezieht sich das Pronomen je nicht auf den Sprecher, das Pronomen te nicht auf den Adressaten und die Verbform aime auch nicht auf einen aktuellen Seelenzustand, in dem sich der erstere gegeniiber dem letzteren befande, ja die Ausdriicke beziehen sich auf niemanden und auf niemandes Seelenzustand und haben in diesem Sinne keine Referenz. Und indem sie und der Satz keine Referenz haben, hat der Satz auch keine Geltung. Das einzige, was er, semantisch gesehen, hat, ist Bedeutung, Referenz und Geltung miissen ihm erst noch verliehen werden. Im Rahmen unserer Episode allerdings ist dies bereits geschehen, geschehen durch die Verwendung dieses Satzes als aktuelle Liebeserklarung. In dieser bezeichnen die Pronomina namlich tatsachlich den Sprecher und die Angeredete und die Verbform tatsachlich einen Seelenzustand des ersteren in bezug auf die letztere. Die Ausdriicke haben Referenz, und die Aufierung hat Geltung.

Langue, Parole, Metalangue, Metaparole Was aber hat die beschriebene Situation mit Langue und Parole zu tun? Ich glaube: sehr viel; derm fur mich sind die Antwort "Je t'aime" auf die Frage "Was heiBt eigentlich auf Franzosisch Ich Hebe dich?" und der Ausdruck Ich Hebe dich in dieser Frage selber LangueEinheiten, die erstere eine Einheit der franzosischen, der letztere eine Einheit der deutschen Langue, und eine Parole-Einheit ist nur die aktuelle Liebeserklarung.1 Aber - so konnte man einwenden - ist nicht der Unterschied zwischen der aktuellen Liebeserklarung und den beiden zuvor genannten Ausdriicken eher ein Unterschied zwischen Sprache und Metasprache? Darauf wiirde ich antworten: Ja, das ist er a u c h , aber aulierdem ist er ein Unterschied zwischen Parole und Langue, und das heilit letztlich, er ist ein Unterschied zwischen Parole oder genauer: primarer Parole2 und Metalangue. Der Ausdruck Ich Hebe dich in der Frage "Was heifit auf Franzosisch Ich Hebe dich?" und die Antwort "Je t'aime" sind, obwohl dies im Falle der Antwort nicht auf Anhieb deutlich sein mag, metasprachlich und, da sie zugleich Langue-Einheiten sind, Einheiten der Meta-Langue; die aktuelle Liebeserklarung Je t 'aime aber ist primarsprachlich und, da sie zugleich eine ParoleEinheit ist, eine Einheit der primaren Parole. Primare L a n g u e -Einheiten sind demgegenuber jene Satze der deutschen bzw. der franzosischen Sprache, die durch den Ausdruck Ich Hebe dich in der metasprachlichen Frage und den Ausdruck Je t'aime in der Antwort darauf b e z e i c h n e t werden und genauso lauten wie ihre Zitate, Zitate in einem etwas erweiterten Sinne des Wortes. Eine Einheit d e r M e t a - P a r o l e schlieBlich liegt dann vor, wenn die junge Franzosin, die die Liebeserklarung erhalten hat, spater zu einer Freundin sagt, ein junger Deutscher habe zu ihr gesagt: "Je t'aime", also bei einem Zitat im engeren und eigentlichen Sinne; denn dieses Zitat ist genau so referenz- und geltungshaltig wie die OriginalauBerung ihres deutschen Verehrers. Auf die Opposition von Langue und Parole speziell im Zusammenhang mit dem Phanomen der tfbersetzung bin ich etwas genauer in Harweg (1984) eingegangen. Statt von primarer Parole und primarer Langue habe ich in Harweg (1980), noch der traditionellen Objektsprachterminologie folgend, aber bereits damals wider besseres Wissen handelnd, von Objektparole bzw. Objektlangue gesprochen. Es empfiehlt sich jedoch, diese Begriffe und - ganz allgemein - den Begriff Objektsprache nur fur Einheiten zu verwenden, die tatsachlich O b j e k t e der Metaparole bzw. der Metalangue oder allgemein: der Metasprache sind. Vgl. dazu Harweg (1981) und Harweg (1982).

Langue und Parole: Referenz und Geltung

91

Kompetenzlangue und Performanzlangue Bin neues Langue-Parole-Konzept - aber wie verhalt sich dieses neue Langue-Parole-Konzept zum herkommlichen Langue-Parole-Konzept? Erinnern wir uns, daB das herkommliche Langue-Parole-Konzept im wesentlichen durch Merkmaloppositionen wie 'abstrakt' vs. 'konkret', 'sozial' vs. 'individuell', 'potentiell' vs. 'aktualisiert' oder 'virtuell' vs. 'realisiert' charakterisiert und die Langue dabei als abstrakt, sozial, potentiell und virtuell, die Parole hingegen als konkret, individuell, aktualisiert und realisiert bestimmt worden ist,3 so scheinen meine Stufe der primaren Langue und meine Stufen der primaren Parole und der Metaparole mil der traditionellen Konzeption durchaus ubereinzustimmen. Etwas anders verhalt es sich jedoch mit meiner Stufe der Meta-Langue; derm die Zitate primarer Langue-Einheiten, die ja die Meta-Langue bilden, sind zumindest nicht abstrakt, sondern konkret. Ob sie potentiell oder aktualisiert oder virtuell oder realisiert sind, ist vielleicht nicht so eindeutig zu sagen, aber das zeigt nur, dafi es sinnvoll ist, diese Kriterien durch das Referenz- und das Geltungskriterium zu ersetzen. Es gibt indes auch p r i m are Langue-Einheiten, die statt abstrakt konkret sind. Solche Einheiten liegen z.B. vor, wenn jemand auf die Frage Was ist (statt wie heifii) die franzosische Entsprechung von 'Ich Hebe dich'? antwortet Je t'aime oder wenn jemand, in Ausfuhrung der Aufforderung Bilden Sie einen deutschen Satz mit dem Verbum 'lieben'/, sagt Ich Hebe dich. In diesem Fall sind die Ausdrucke Je t'aime und Ich Hebe dich keine Zitate, also keine Meta-Langue-Einheiten, sondern bloBe Langue-Einheiten. Das gleiche ist der Fall, wenn in Sprachfuhrern einander entsprechende Satze zweier Sprachen paarweise untereinander oder einander gegenuber stehen oder wenn in Konversationsgrammatiken ganze einander entsprechende Gesprache einander gegeniibergestellt sind. Auch Sammlungen von Musterbriefen in sogenannten Briefstellern sind solche bloBen Langue-Einheiten. All diese Einheiten sind, obwohl sie als Einheiten der Langue zu interpretieren sind, offensichtlich konkrete Spracherscheinungen oder mit einem Begriff der generativen Grammatik gesprochen: Einheiten der Performanz. Sie sind Einheiten einer Schicht von Langue, die den - widerspriichlich erscheinenden - Namen Performanzlangue verdient.4 Das bedeutet, daB das referenz- und geltungstheoretische Konzept der Langue nicht zusammenfallt mit dem Konzept der Kompetenz, sondern in zwei Schichten zerfallt: die Schicht der Kompetenzlangue und die Schicht der Performanzlangue. Kompetenzlangue ist, referenz- und geltungstheoretisch gesehen, referenz- und geltungsloses Wissen iiber eine Sprache, Performanzlangue hingegen referenz- und geltungslose Realisierung dieses Wissens - eine Realisierung, wie sie sich auf der Wortebene z.B. in unseren Worterbuchern zeigt.

3 4

Vgl. dazu z.B. Coseriu (1962), der auch einen Uberblick iiber die Langue-Parole-Diskussion bis 1952 gibt, und Heger (1969, 147f.). Die Performanzlangue ist in gewisser Weise eine Schicht von Langue, die zwischen der Langue im uberkommenen Sinne, der Kompetenzlangue, und der Parole liegt. Auch andere Forscher haben bekanntlich solche Zwischenschichten etabliert, so z.B. Biihler (1965, 48ff.) die zwischen den Polen Sprechhandlung und Sprachgebilde angesiedelten Schichten Sprechakt und Sprachwerk, Hjelmslev (1959) die zwischen Sprachbau (Schema) und Akt lokalisierten Schichten der Norm und des Sprachgebrauchs, Buyssens (1943) die zwischen Parole und Langue situierte Schicht des Discours, Coseriu (1962) die zwischen System (sistema) und Parole (habla) angesiedelte Schicht der Norm (norma) und Heger (1969, 156ff.) die zwischen Parole und Langue lokalisierte Schicht einer sogenannten Zparole. Aber all diese Schichten sind Schichten ganz anderer Art und Begriindung.

92


Texte als Langue-Einheiten Die Wortebene 1st bekanntlich die Ebene in der Hierarchic sprachlicher Einheiten, iiber die das traditionelle, d.h. das von Saussure gepragte und von vielen anderen ubernommene LangueKonzept, von Ausnahmen abgesehen, nicht hinausgeht. Schon Satze gehorten for Saussure (1955, 172) und seine Nachfolger5 nur der Parole an - eine Konzeption, die sich dam allerdings im Rahmen der generativen Grammatik geandert hat; derm die Satze, die die generative Grammatik erzeugen will, sind ja keine Parole-, sondern Langue-Einheiten, und erzeugt sie ein einzelner Grammatiker, so sind sie Einheiten der Performanzlangue. In m e i n e m referenz- und geltungstheoretischen Konzept von Langue und Parole aber konnen schlieBlich, wie die Gesprache in Konversationsgrammatiken und die Musterbriefe in Briefstellern zeigen, nicht nur Satze, sondern sogar Texte als Bestandteile der Langue figurieren - zumindest als Bestandteile der Performanzlangue.6 Performanzlangue-Texte manifestieren Erweiterungen des traditionellen Langue-Konzeptes in zwei Richtungen: einmal in der Richtung von Abstraktheit zu Konkretheit, von Kompetenz zu Performanz und zum andern in der Richtung von Wortern iiber Satze zu Texten. Wenn Referenz- und Geltungslosigkeit das entscheidende Merkmal von Langue-Einheiten sind, sind dann, so konnte man weiter fragen, nicht auch f i k t i o n a l e Texte LangueEinheiten? Meine Antwort ist: Ja. Auch die sogenannten fiktionalen Texte sind, referenz- und geltungstheoretisch gesehen, Langue-Einheiten, auch sie sind Performanzlangue-Texte. Nur die meisten fiktionalen Texte erschopfen sich nicht in ihrem Langue-Text-Sein, die meisten fiktionalen Texte haben zugleich eine - allerdings fiktive - Parole-Schicht, ja sie sind letztlich nur deretwegen iiberhaupt da. hi dieser Parole-Schicht aber haben sie - jeweils fiktive Referenz und Geltung (vgl. Harweg 1979 und Tschauder 1989). Referenz- und geltungslos und damit Langue-Einheiten sind sie als Schopfungen ihrer nichtfiktiven Autoren, der Dichter, referenz- und geltungs h a 11 i g und damit Parole-Einheiten sind sie als Produkte der fiktiven Erzahler (in die sich die Dichter, wenn sie erzahlen, versetzen) oder als Auflerungen der fiktiven Personen der Handlung eines Dramas. Es gibt jedoch auch fiktionale Texte o h n e eine solche zweite Schicht, also fiktionale Texte, die nur die Perforrnanzlangue-Schicht, nicht aber auch die Parole-Schicht kennen. Solche fiktionalen Texte finden sich meines Erachtens haufig in der Gattung der Lyrik. Ein Beispiel fur einen solchen Text ist etwa das liedhafte Goethesche Gedicht:

5

6

Man vgl. nur Gardiner (1960, 88) und Benveniste (1966, 128-131). Start von Parole spricht Gardiner allerdings von "speech" und Benveniste von "discours". Die communis opinio vertritt die Ansicht, dafi Texte nur zur Parole gehorten. Aber die communis opinio griindet den Unterschied zwischen Langue und Parole entweder nicht auf die Kriterien der Referenz und der Geltung, oder aber sie bezieht, wenn sie, wie z.B. Sgall (1984, 124 und 128), doch auf solche Fragen eingeht und, wie Sgall, fur Texte fordert, dafi sie referenzhaltig seien, Texte vom Typ der Gesprache in Konversationsgrammatiken und der Musterbriefe in Briefstellern nicht in die Betrachtung mit ein. Ein Autor, der Texte zur Langue oder genauer: zu einer besonderen (aber wohl eher als eine Art von Kompetenz- denn als eine Art von Performanzlangue zu begreifenden) Form von Langue rechnet (er spricht von Texten "in potentia" oder "in abstracto"), ist Danes (1977, 157). Auch dessen Landsmann Palek (1968, 3 Iff. und 127ff.) hat den Vorschlag gemacht, Texte der Langue zuzurechnen.


93

Uber alien Gipfeln ist Ruh, in alien Wipfeln spiirest du kaum einen Hauch. Die Vogelein sckweigen im Walde. Wane nur, balde Ruhest du auch. Das Gedicht bezieht sich auf keine konkrete Situation und hat keine konkrete Geltung. Es ist ein Gebilde ohne konkrete Botschaft, es ist Poesie im etymologischen Sinne des Wortes. Sein Pronomen du z.B. ist im Grunde nicht referenzhaltiger als dasselbe Pronomen in einem Sprachfuhrer- oder Grammatikersatz.

ElNE

NACHTRAGLICHE

SPEZIFIZIERUNG

DES REFERENZ-

UND DES

GELTUNGSBEGRIFFS Einmal angelangt auf der Textebene, konnen wir indes nicht umhin, eine Spezifizierung des Referenzbegriffes vorzunehmen, eine Spezifizierung, die sich zwar in vielen Fallen auch auf der Satzebene schon als notig erweist, deren Notwendigkeit auf der Textebene jedoch geradezu in die Augen springt; denn mogen Langue-Texte auch, wie wir gesagt haben, keine Referenz aufweisen, so pflegen sie, vor allem wenn sie nicht gerade nur aus einem Satz bestehen, doch das Phanomen der K o r e f e r e n z aufzuweisen, sind im Grunde sogar entscheidend durch dieses bestimmt. Koreferenz aber prasupponiert auch Referenz. Doch wie konnen LangueTexte einerseits referenzlos sein und andererseits doch Referenz aufweisen? Nur so, dafl der Begriff von Referenz, in dessen Sinne sie referenzlos sind, ein anderer ist als der, in dessen Sinne sie referenzhaltig sind. Ware es nicht tautologisch, wurde ich die Referenz der Koreferenz einfach als Langue-Referenz spezifizieren, so aber ist es besser, sie als virtuelle Referenz zu charakterisieren. In einem Grammatiker-Satz wie Er wascht sich sind die Pronomina zwar einerseits durchaus referenzlos, andererseits aber koreferent und deshalb auch referenzhaltig. Diese ihre Referenz aber ist keine aktuelle, sondern eine virtuelle. Wir mussen unsere referenztheoretische Ausgangscharakterisierung der Langue-Einheiten also spezifizieren und sagen, sie seien referenz 1 o s im Sinne a k t u e 11 e r Referenz. Verstehen wir den Referenzbegriff in diesem Sinne, also im Sinne von aktueller Referenz, so sind das Referenzkriterium und das Geltungskriterium in unseren bisherigen Beispielen nicht auseinandergefallen, ihre Ergebnisse vielmehr stets dieselben gewesen, und so konnte sich die Frage erheben, ob dies wohl auf der ganzen Linie so sei und die beiden Kriterien dann vielleicht sogar synonym seien. Nun, synonym sind sie nicht, bestenfalls sind sie teilsynonym, sind dies dann, wenn man auf jenen Geltungsbegriff rekurriert - denn ich unterscheide auch zwei verschiedene Geltungsbegriffe -, der charakteristisch ist fur einen ganz bestimmten Texttyp, namlich den der Gesetze; denn wenn ein Gesetz gilt oder - wie man iiblicherweise sagt - in Kraft ist, dann haben seine Satze auch Referenz, Referenz in dem Sinne, daB die Normen, die es aussagt, existieren; gilt es aber nicht, dann existieren auch die Normen, die seine Satze aussagen, nicht, und es hat in diesem Sinne keine Referenz. Die Referenzen oder

94


genauer: die Korrelate, die diese Geltung impliziert, sind allerdings von ganz besonderer Art, namlich durch die Geltung allererst geschaffene (vgl. dazu Harweg 1988). Diese wirklichkeitsschaffend-semantische Art von Geltung ist allerdings nicht d i e Art von Geltung, die ich als Kriterium fur die Unterscheidung von Langue und Parole bemuhe. Diese kennzeichne ich statt als wirklichkeitsschaffend-semantisch als a d r e s s a t e n b e z o g e n p r a g m a t i s c h , und mit dem Begriff der Referenz, genauer: der aktuellen Referenz ist diese Form von Geltung nicht nur nicht teilsynonym, sondern iiberhaupt nicht synonym. Zwar liefern die adressatenbezogen-pragmatische Geltung und die aktuelle Referenz, in ihrer Anwendung auf Spracheinheiten, in vielen Fallen, wie wir gesehen haben, die gleichen Ergebnisse, aber es gibt auch Falle, in denen sie es nicht tun, und diesen wenden wir uns nunmehr zu.

DER AUSEINANDERFALL DER BEIDEN - SPEZIFIZIERTEN - KRITERIEN Zwei Texttypen, in denen die beiden spezifizierten Kriterien, das spezifizierte Referenz- und das spezifizierte Geltungskriterium, zu unterschiedlichen Ergebnissen fuhren, sind Lugen und Scherze. Lugen und Scherze sind Auflerungen, die im Sinne meines weiten, d.h. auch - anders als etwa bei Searle (1970, 26ff.) - auf Pradikate angewendeten Referenzbegriffes teilreferent sind. In normalen Liigen wie Karl ist abgereist oder Scherzen, sagen wir: Aprilscherzen wie Du sollst Heinrich besuchen haben zwar die Ausdrucke Karl, du und Heinrich, nicht aber die Ausdriicke ist abgereist oder sollst besuchen Referenz, und waren die Scherze Scherze im Stil von Radio Erewan, so ware es genau umgekehrt: Dann hatten die Verbalformen, nicht aber die Ausdrucke Karl, du und Heinrich Referenz. Die mangelnde Teilreferenz dieser Aufierungen aber laBt dieselben in gewisser Weise auch als ganze referenzlos sein. Doch unbeschadet dieser Referenzlosigkeit sind sie, im adressatenbezogen-pragmatischen Sinne, nicht g e l t u n g s l o s , sondern vielmehr geltungs h a 11 i g. Sie sind geltungshaltig insofern, als der Sprecher mochte, daB der Adressat, wenigstens fur eine gewisse Zeit, g 1 a u b t, sie seien referenzhaltig, und er in bestimmten Fallen, wie z.B. bei Meineiden, auch fur sie verantwortlich gemacht werden kann. Einen den Liigen und Scherzen genau entgegengesetzten Texttyp bilden, unter dem in Rede stehenden Gesichtspunkt, bestimmte G r a m m a t i k e r s a t z e , namlich Grammatikersatze generischen oder definit partikularen Inhalts, also Grammatikersa'tze wie Der Mensch ist sterblich bzw. Paris ist die Hauptstadt Frankreichs. Sie weisen, und zwar als ganze, eine jeweils aktuelle Referenz auf, aber sie haben, als Grammatikersatze, die sie sind, keine adressatenspezifisch-pragmatische Geltung. Der Grammatiker will ihre Referenz, wenigstens primar, niemandem mirteilen. Die Satze dieses Typus sind, wie die Beispiele andeuten, normalerweise Satze mit generisch verwendeten Gemeinnamen oder referentiell eindeutigen Eigennamen. Allerdings linden sich auch Satze oder haufiger vielleicht sogar Texte dieses - durch Referenzhaltigkeit einerseits und Geltungslosigkeit andererseits defmierten - Typus, die wie Deiktika aussehende P r o n o m i n a enthalten, also Ausdrucke, die normalerweise einer aktuellen Referenz entbehren, in diesem Falle jedoch, ausnahmsweise, iiber eine solche verfugen. Es handelt sich um allgemeine, um "kommunale" G e b e t e , Gebete wie z.B. das Vaterunser. Sowohl der durch das Pronomen wir bezeichnete kommunale Sprecher als auch der durch das Pronomen du bezeichnete individuelle Adressat sind darin, fur die Glaubigen, eindeutig bezeichnet. Der Text ist, in der Glaubenswelt der Glaubigen, referenzhaltig, aber solange er nicht aktuell gebetet


95

wird, nicht geltungshaltig. Seine adressatenbezogen-pragmatische Geltung erlangt er immer erst, wenn er aktuell gebetet wird. Liigen und Scherze auf der einen Seite und generische und defmit partikulare Grarnmatikersatze sowie nicht aktuell gebetete kommunale Gebete auf der anderen Seite sind sprachliche Einheiten, in bezug auf die, wie wir gesehen haben, die Langue-Parole-Kriterien der Referenz und der Geltung auseinanderfallen. Wir mussen uns also, im Hinblick auf die Langue- oder Parole-Zuweisung dieser Einheiten, entscheiden, welchem der beiden Kriterien wir den Vorzug geben wollen: dem Referenz- oder dem Geltungskriterium. Geben wir dem Referenzkriterium, genauer: dem Kriterium der aktuellen Referenz, den Vorzug, sagen also, daB Langue-Einheiten referenzlose Einheiten und Parole-Einheiten referenzhaltige Einheiten seien, so sind Liigen und Scherze Langue-Einheiten und Grammatikersatze vom Typus Der Mensch ist sterblich oder Paris ist die Hauptstadt Frankreichs oder das nicht aktuell gebetete Vaterunser Parole-Einheiten. Geben wir demgegeniiber dem Geltungskriterium, genauer: dem Kriterium der adressatenbezogenpragmatischen Geltung den Vorzug, sagen wir also, daB Langue-Einheiten geltungslose Einheiten und Parole-Einheiten geltungshaltige Einheiten seien, so ist es genau umgekehrt: Liigen und Scherze sind dann Parole-Einheiten und jene Grammatikersatze und das nicht aktuell gebetete Vaterunser Langue-Einheiten.

IM ZWEIFELSFALL DAS - SPEZIFIZIERTE - GELTUNGSKRITERIUM Sehen wir einmal von den Liigen und Scherzen ab, so habe ich mich, in friiheren Veroffentlichungen, bereits fur beide Alternativen ausgesprochen, einmal fur das Referenzkriterium (in Harweg 1992) und einmal fur das Geltungskriterium (z.B. in Harweg 1980). Allerdings habe ich, als ich mich fur die erstere Alternative, das Referenzkriterium, entschied, die Grammatikersatze vom Typus Der Mensch ist sterblich oder Paris ist die Hauptstadt Frankreichs und das nicht aktuell gebetete Vaterunser nicht als Parole schlechthin, sondern als eine bestimmte Schicht von Parole, namlich als Kompetenzparole, bezeichnet (Harweg 1992, 191f.) und diese einer Performanzparole gegeniibergestellt. Aus heutiger Sicht scheint mir, daB diese Losung nicht geeignet ist, zwischen den besagten Grammatikersatzen und denselben Satzen als normal verwendeten Satzen zu unterscheiden; denn beide sind Performanzprodukte. Der Kompetenzschicht gehoren die Satze namlich nur so lange an, wie sie nicht realisiert sind, gleichgiiltig ob der sie Realisierende ein Grammatiker oder ein normaler Sterblicher ist. Ich entscheide mich also heute, im Hinblick auf die in Rede stehenden Einheiten, fur das adressatenbezogen-pragmatische Geltungskriterium. Das bedeutet, daB die referenzhaltigen Grammatikersatze und das nicht aktuell gebetete Vaterunser Langue-Einheiten sind und als solche sowohl Einheiten der Kompetenzlangue als auch solche der Performanzlangue sein konnen, ersteres, wenn sie nur als Wissen existent sind, und letzteres, wenn sie, in einer Grammatik bzw. in einem Gebetbuch, in nicht geltungshaltiger Form realisiert sind. Erst normal verwendet, und das heiBt im Falle des Vaterunsers: aktuell gebetet, sind sie dann Einheiten der Parole. Meine jetzige Entscheidung fur das Geltungskriterium erfahrt durch die jetzige Mitberucksichtigung von Liigen und Scherzen noch eine gewichtige Unterstutzung; denn die Interpretation der Liigen und der Scherze als Parole-Einheiten (die diese Entscheidung

96


impliziert) 1st, wie mir scheint, intuitiv viel uberzeugender als ihre - von der Entscheidung fur das Referenzkriterium implizierte - Interpretation als Langue-Einheiten. Es ist namlich, wie ich meine, uberzeugend, zu sagen, dafi es Lugen und Scherze nur in der Parole, nicht aber in der Langue gebe. Was es in der Langue einer Sprache allenfalls geben mag, sind referentiell f a 1 s c h e Satze, also Satze wie Paris ist die Hauptstadt Englands, und was es, nach meiner Konzeption, aber entgegen allgemein verbreiteter Auffassung, in der Langue einer Sprache auch gibt, sind falsch k o n s t r u i e r t e , also g r a m m a t i s c h falsche Satze. Es sind dies z.B. die von Grammatikern zu Demonstrationszwecken konstruierten und mit einem Sternchen versehenen u n g r a m m a t i s c h e n Satze, und wenn diese nach meinem Kriterium ebenfalls der Langue angehoren, so deshalb, weil auch sie geltungslos sind. Aber die Langue, der sie angehoren, ist naturlich nicht die normale, sie ist, wenn man so will, eine n e g a t i v e Langue negativ im Gegensatz zu der positiven Langue, in der die grammatisch korrekten LangueEinheiten beheimatet sind. Bildlich gesprochen, ist diese negative Langue gewissermaBen die Unterwelt der betreffenden Langue. Oder: Ist die Parole gleichsam die Erde und die normale Langue der Himmel einer Sprache, so ist jene negative Langue sozusagen ihre Holle.

LlTERATUR Benveniste, Emile (1966). Les niveaux de 1'analyse linguistique. In: Problemes de linguistique generale, pp. 119-131. Gallimard, Paris. Biihler, Karl (1965). Sprachtheorie. Fischer, Stuttgart. Buyssens, Eric (1943). Les langages et le discours. Office de publicite, Bruxelles. Coseriu, Eugenic (1962). Sistema, norma y habla. In: Teoria de lenguaje y linguistica general, pp. 11-113. Gredos, Madrid. Danes, Frantisek (1977). Zum Status der Textebene. In: Probleme der Textgrammatik II (F. Danes und D. Viehweger, eds.), pp. 153-158. Akademie-Verlag, Berlin. Gardiner, Sir Alan (1960). The Theory of Speech and Language. Clarendon Press, Oxford. Harweg, Roland (1979). Inhaltsentwurf, Erzahlung, Inhaltswiedergabe. Zum fiktionstheoretischen Doppelstatus fiktionaler Erzahlungen. In: Grundfragen der Textwissenschaft (W. Frier und G. Labroisse, eds.), pp. 111-130. Rodopi, Amsterdam. Harweg, Roland (1980). Metasprache und Objektsprache als Langue und Parole. ALASH (olim ALH) 30. 247-273. Harweg, Roland (1981). Verwendung und Erwahnung und die Unterscheidung von Objekt- und Metasprache. ZPSK (olim Z.Phon.) 34. 285-292. Harweg, Roland (1982). Reelle und imaginare Stufen in der Meta- und Objektsprachenhierarchie. Orbis3l. 5-29. Harweg, Roland (1984). I testi come unita di parole e di langue. In: Linguistica testuale (L. Coveri, ed.), pp. 5-18. Bulzoni, Roma. Harweg, Roland (1988). Der Weltbezug von Rechtstexten, insbesondere Gesetzen - verglichen mit dem der fiktionalen ErzShltexte. Archivum luridicum Cracoviense 21. 21-35. Harweg, Roland (1992). Communicative and Non-Communicative Language. In: Signs of Humanity. L'homme et ses signes (M. Balat und J. Deledalle-Rhodes, eds.), Vol.1, pp. 187-194. Mouton de Gruyter, Berlin. Heger, Klaus (1969). Die Semantik und die Dichotomic von Langue und Parole. ZRPh 85. 144215.


97

Hjelmslev, Louis (1959). Langue et Parole. In: Essais linguistiques.(= Travaux du Cercle linguistique de Copenhague 12), pp. 69-81. Nordisk Sprog- og Kulturforlag, Copenhague. Palek, Bohumil (1968). Cross-Reference. A Study from Hyper-Syntax. Universita Karlova, Praha. Saussure, Ferdinand de (1955). Cours de linguistique generate. Payot, Paris. Searle, John R. (1970). Speech Acts. An Essay in the Philosophy of Language. University Press, Cambridge. Sgall, Petr (1984). Remarks on Text, Language and Communication. In: Contributions to Functional Syntax, Semantics, and Language Comprehension (P. Sgall, ed.). Academia, Praha. Tschauder, Gerhard (1989). Textverbindungen. Ansdtze zu einer Makrotextologie, auch unter Berucksichtigungfiktionaler Texte. Brockmeyer, Bochum.


8

PAROLE AS AN INDIVIDUAL REALISATION OF LANGUE Tadao Shimomiya, Gakushuin University, Tokyo

Abstract Parole is an individual realisation of langue. The paper applies this principle to some linguistic and extra-linguistic phenomena. Following a suggestion by Jakobson and Bogatyrev (1929), extra-linguistic phenomena like the relation between the Finnish national epic Kalevala and its version by Elias Lonnrot (1835, 1849) can be compared to langue and parole, or the proverb 'A great city is a great solitude' and its comment by (e.g.) Erasmus, or its illustration by C.Yanagida can be compared to langue and parole. The Holy Island (which produced the Lindisfarne gospels) off Berwick-upon-Tweed as the langue is reproduced in pictures or by artists as facts of parole.

The present paper is not 'gegen eine herkommliche Interpretation' (see the paper by Hartmann, this volume), but rather a faithful 'textbook' interpretation of langue and parole and its further application to some linguistic and extra-linguistic phenomena. The dichotomy langue/parole is defined as system/members, total/partial, social/individual, constant/occasional, latent (possible)/apparent (real). In the following, / /are examples of langue, [ ] are those of parole. A phoneme covers several sounds. The phoneme Id for example is realised as [ei] [ei] [63] [e4]...[en]. Only after [en] begins the open Id. This can be schematized as /phonemes/ vs. [individual sounds]. Figure 1 owes its idea to Milewski (1973:49), which despite its modest title abounds in suggestions in many respects.

Fig. l(cf. Milewski 1973)

99

100 Langue and Parole in Synchrony and Diachrony Contextual (situational) meaning is a matter of parole. The lexical meaning /town/ as in I live in a town is contextually (situationally) [Bonn],[Stratford], [Berwick], or /the river/ as in / am going to the river refers to [the Rhine], [the Avon], [the Tweed], /television/ meant [black-and-white] in the 1960s, but [color] in the 1990s. In the transition period which television is meant is a matter of parole. In the same way /fire/ differs contextually in [You are fired] and [There is a fire].

Japanese onomatopoeia distinguishes between peaceful sleep 'suya suya' and noisy (or sound) sleep 'guu guu', while English has only one /z-z-z/ on the level oflangue, though it may have [z-z-z] or [Z-Z-Z] or even larger [Z-Z-Z] on the level of parole. In lexicology langue can include the structurally possible words /mermaid, merman, merking, merqueen, merprince, merprincess/. The word mermaid is immortalized by Hans Christian Andersen in his 'The Little Mermaid' (1837), and merman in Danish ballads ('Agnete og havmanden') or Matthew Arnold's 'The Forsaken Merman' (1849). But the Oxford English Dictionary (2nd ed. 1989, 20 vols. with 400,000 words) gives only [mermaid] and [merman]. German uses the term /Nordistik/, but the structurally possible Sudistik does not exist. /Europaistik/ is known, but /Asiatistik/ is not (for Asienkunde). /Orientalistik/ exists, but so far not /Okzidentalistik/. E.Coseriu (1979:276) gives Italian giocare 'to play', giocherellare 'to idle away', which exist in the dictionary, and their derivatives (structurally possible, but not registered in the dictionary) rigiocherellare, rigiocherellamento, rigiocherellatore, rigiocherellamentista, rigiocherellamentistico, rigiocherellamentisticamente, rigiocherellamenticizzare. 'La parole est une realite de la langue, une realite individuelle' (H.Kobayashi 1937:340): Hideo Kobayashi, a Japanese linguist, was the world's first translator of Saussure's Cours de linguistique generale (Tokyo 1928). Langue is a deep structure phenomenon, while parole is one of its possible surface manifestations in the soundsequential (vocal, verbal) setting (Decsy 1988). Langue is the train scheduled to leave Geneva for Paris every day at 2:00 p.m. as listed in the timetable of the railroad company, while parole is the actual train (it may consist of different railroad coaches every day which leave the Geneva railroad station and need not be physically the same coaches every day) (Decsy 1988:22). The process from the provisional state of parole to the established state of langue is seen in the case of analogy, as Saussure explains why German ich was became ich war: a few individuals began to say ich war by analogy. It was only a fact of parole and not a fact of langue as long as there were only a few individuals who did it (Komatsu & Harris 1993:119). Following a suggestion by Jakobson and Bogatyrev (1929), who compared folklore and its literary production to langue and parole, I wish to give similar extralinguistic parallels:

Parole as an Individual Realisation ofLangue

101

if one were to take the Finnish national epic Kalevala as the langue, then its literary work, the 1835 or 1849 edition by Elias Lonnrot, or English translation by W.F.Kirby (1907), Japanese translation by Kakutan Morimoto (1937) or by Tamotsu Koizumi (1976) may be called examples of the parole, because the works by Lonnrot, Kirby, Morimoto and Koizumi are individual interpretations (or realisations) of the folklore tale Kalevala. The proverb /A great city is a great solitude/, known from the ancient Greek days, is compared to langue, while its commentary (beginning with Erasmus) and illustrations, as shown in Figure 2 (illustration by Chifuyu Yanagida), are facts of parole.

/Holy Island/ (which produced the Lindisfarne Gospels, raided by Vikings in 793 AD) off Berwick-upon-Tweed as langue is reproduced in [picture 1, picture 2, picture 3,...] or by [artist 1, artist 2, artist 3,...] as facts of parole. The science /phonetics/ as langue is elaborated by [Sweet], [Jespersen], [Passy], [Victor] as facts of parole.

REFERENCES Coseriu, Eugenio (1979). Japanese translation of Lezioni di linguistica generale (Boringhieri, Torino, 1973). Sanshusha, Tokyo. Decsy, Gyula (1988) A Select List of Language Universals (Bibliotheca Nostratica, ed. G.Decsy, Vol.8). Eurolingua, Bloomington. Jakobson, Roman and P.Bogatyrev (1929). Die Folklore als eine besondere Form des Schaffens. In : Donum Natalicium Schrijnen, Nijmegen-Utrecht, N.V.Dekker & Van de Vegt, 900-913; Selected Writings of Roman Jakobson, IV, 1966, 1-15. Kobayashi, H.(1937). Gengo to buntai (Language and style). Sanseido, Tokyo. Komatsu, E. and R.Harris (ed. and transl.) (1993). Troisieme Cours de linguistique generale (1910-11) de Ferdinand de Saussure. Pergamon Press, Oxford. Milewski, Tadeusz (1973). Introduction to the Study of Language. Mouton, The Hague.

102 Langue and Parole in Synchrony and Diachrony Shimomiya, T.(1994). Taschenworterbuch der geflugelten Worte in deutscher und anderen europaischen Sprachen. Dogakusha, Tokyo. Sweet, H.(1998). A Handbook of Phonetics (Oxford 1877). Ed. by Kenzo Kihara, Sanseido, Tokyo.1

1 This book reproduces H.Sweet's own copy, with marginal notes by Sweet and J.Storm. Sanki Ichikawa (1886-1970, Professor of English philology at the Imperial University of Tokyo), while studying at Oxford, acquired a large part of the libraries of Henry Sweet and W. W.Skeat (both died in 1912). When Ichikawa retired from Tokyo University in 1946, he gave most of his library to the University. Sweet's Handbook of Phonetics passed into the hands of Prof. Kihara through two successive disciples of Ichikawa.

III DEIXIS


9

ASPECT, VOICE AND DEIXIS IN RUSSIAN Youn'A. Poupynin, Institute of Linguistic Research, Russian Academy of Sciences, St. Petersburg, Russia

Abstract Aspect and voice are usually regarded as non-deictic. The paper argues that both categories have some clear deictic features in Russian. Aspect is a deictic category, since the perfective marks the absence of coincidence of an action with the deictic centre and the imperfective does not mark it. Voice is deictic, since the active (in comparison to the passive) is a form more adapted to denote a situation in the deictic centre "I - Here - Now". Aspect has a number of interrelations with deixis in the area of aspectual lexicogrammatical classes (Aktionsarten) as well.

1. Deictic relevance of aspect and voice as grammatical categories The notion of deixis is usually derived from Btihler's concept (cf. his "Zeigfeld" (1934), cf. also some more recent publications on the deictic theme as follows: (Apresjan, 1986; Clark, 1973; Here and There, 1982; Leiss, 1992; Paducheva, 1996)). Grammatical categories can be treated as deictic if they inform as to how their content is connected with speech parameters (primarily with temporality, relation to participants of a speech situation and location in space) (cf. Levinson, 1996). Traditionally, mood, tense, and person are considered deictic categories (or shifters, according to Jakobson, 1971). Aspect, as a rule, is regarded as non-deictic (Comrie, 1985) (cf. also the argument by Apresjan (1995) who objects to this opinion in the sphere of the perfective aspect), and as to voice, its non-deicticness seems to be obvious. My paper's aim is to prove that both categories, i.e. aspect (including perfective and imperfective) and voice, have some clear deictic features in Russian. As is well-known, in the theory of deixis a crucial role is played by the so-called "deictic centre", i.e. by three parameters "I - Here - Now". Practically every utterance implies a deictic centre. Cf., for example: (1) Vcera my ezdili v Peter gof 'Yesterday we drove to Peterhof, 105

106 Langue and Parole in Synchrony and Diachrony where the other deictic situation ("We - There - Yesterday") is explicated, but the deictic centre "I - Here - Now" is implied, indicating a point of departure on the axes of time, person, and space. It is also possible to point out some cases where the deictic centre itself is explicated, cf.: (2) Sejcas my edem vPetergof 'Now we are driving to Peterhof, if we assume that it is said by a driver to his friend after turning left or right (this utterance can also have a meaning of future in another situation). The ability of a category to participate in expression of the complex "I - Here Now" appears to be an essential argument in favour of its deicticness. This feature allows us to define a deictic category in a more formal way. In my view, a category is deictic if it (I mean, at least one of its grammemes) can denote the elements of the deictic centre. Therefore, mood, tense and person are deictic, since their grammemes of indicative mood, present tense, and 1st person participate in the explication of an action which is included in the deictic centre (cf. (2)). In this case, it seems to be a great misconception to consider aspect as a non-deictic category, because in Russian no tense form exists separately from aspect. There is an account which distinguishes between five, not three tenses (i.e. temporal-aspectual forms) in Russian, taking into account the interaction of both tense and aspect (cf. Bondarko 1971). The tenses are 1) the imperfective past, 2) the perfective past, 3) the imperfective present, 4) the imperfective future, 5) the perfective future. Since the Russian present tense exists in the imperfective variant only, it is possible to say that aspect, with its imperfective grammeme, participates in denoting the deictic centre and consequently is deictic. The perfective can denote an action which can come very close to that centre (in perfective past) but, nevertheless, does not coincide with it. For example: (3) Natasa tol'ko cto us la 'Natasha has just gone (away)', where the result of the action is simultaneous with the moment of speech ("true present time", according to Comrie (1985)), whereas the action itself only precedes that moment. Thus, we can give the following deictic definition of the category of aspect: the perfective marks the absence of coincidence of an action with the deictic centre and the imperfective does not mark it (i.e. the action denoted by imperfective can coincide with the deictic centre, in principle). There are a number of interrelations with deixis in the area of aspectual lexico-grammatical classes (Aktionsarten) as well, which are discussed in Section 2.

Aspect, Voice andDeixis in Russian 107 As to Russian voice, let us first discuss some examples. If somebody tries to denote the whole deictic centre by means of passive, he will probably fail. It seems to be possible, cf.: (4) Rycagperedvigaetsja v polozenie "nol"'... Nazimaetsja krasnaja knopka... The handle is being moved to the position "Zero"... The red button is being pressed...' But if we assume that this situation is demonstrated on a screen, in the foreground we can see both the handle and the button moving, and, obviously, a speaker's hands perform these actions (but, most probably, not a speaker him/herself). In other words, we have only the parameters "Now - Here", whilst the parameter "I" is out of the screen and not explicated in the utterance. Cf. also the following example: (5) Sejcas namirazrabatyvaetsja novaja model'etogo samoljota 'now the new model of this aeroplane is being elaborated by us' Such constructions are rather rare in Russian. Even if such a construction is used, it does not have the meaning of "true present time", since the speaker usually makes a pause in working before talking about his/her work in such a way. In general, this construction can be used in indirect speech acts, cf.: (6) Etotzurnalcitaetsja mnoju! This magazine is being read by me!' in a situation where someone tries to take a magazine which the speaker is reading (or going to read). In my view the active verbal form is much more adapted for denoting the deictic centre than the passive form. It can be explained by the following factors: a) "iconicity" of the active construction in respect of the situation "I - Here - Now" and "non-iconicity" of the passive construction; b) the central and privileged position of denotation of a speaker (and a perceiver as well) in the active construction (Keenan 1976; Kozinskij 1983) and the peripheral and non-obligatory position of denotation of a speaker in passive construction; c) the personal coordination (personal agreement) between the denoted participant of a speech situation and the verb in the active construction and the absence of such a coordination in the passive construction (cf.: *Ja citajus', pis us' i t.d. '?! am being read, written, etc.'), d) the Russian active is not restricted lexically in comparison with passive (cf. a number of verbs which are activa tantum including all the intransitives and a part of the transitives). Hence both the imperfective aspect and active voice can be considered as the crucial grammatical conditions of denotation of the deictic centre. Note that the active

108 Langue and Parole in Synchrony and Diachrony and the passive seem not to constitute a privative or any other opposition in their semantics in respect of deixis. One can speak here only of a tendency in use: the active predominates in expressing the complex of the parameters "I - Here - Now", while the passive usually denotes events which do not coincide with this deictic complex (though the passive can denote each parameter separately).

2. Interrelation between deixis and aspectuality 2.1. Once again towards a semantics of verbs of the type pryti'to come1 The interaction between aspect (or, wider, aspectuality) and deixis is variable and manysided. Let us consider certain interrelations in the field of the Aktionsarten. The verbs of motion with the prefix PRI- '(to move) to1, cf. prijti, priexat', pribezat', priletetn\.o come, to arrive, to come running (up), to fly in' etc. are very interesting in this respect. Consider the following (rather unusual) example: (7) Odnazdy Gogol' pereodelsja Puskinym, prisjol k Puskinu i pozvonil. Puskin otkryl emu i kricit: "Smotri-ka, Anna Rodionovna, ja prisjol!" 'Once upon a time Gogol dressed up like Pushkin, came to Pushkin's place and rang the bell. Pushkin opened the door and cried out: "Look, Arina Rodionovna, I have come!'" (D.Xarms) One of the reasons which make this short story so funny is that the impersonated speaker and the perceiver do not coincide in the content of the last sentence. Example (7) can be explained by the fact that the wordya T does not denote the impersonated speaker (i.e. Pushkin), and, as a matter of fact, the real speaker coincides with the perceiver. The nature of non-coincidence is also clear in cases of a speaker's comments about the videos in which (s)he has filmed him(her)self (I call such cases "non-pathological alternating personality"), cf.: (8) Smotri, votja prisjol k Tane vgosti... 'Look, here I have come to Tanja's place for a party...' Here, the word "I" denotes an alter ego (the "second me") on the screen, whereas the real speaker and the perceiver also coincide. However, the situation when the speaker and the perceiver do not coincide is possible for each usage of the sentence Ja prisjol (literally: 'I have come' in sense of The person who has come is me1). This non-coincidence needs a discussion in cases of ordinary use, for example:

Aspect, Voice and Deixis in Russian 109 (9) Votjaiprisjol 'I have come /1 am here1 If the word "I" denotes the real speaker, the utterance Japrisjolcan be explained as Ja naxodilsja ne zdes', teper'ja naxozus'zdes"(SomQ time ago) I was not here, now I am here'. Hence, before the action prijti 'to come' ends, "I" as a point of departure of the personal deixis and "Here" as a point of departure of the spatial deixis do not coincide in space, and this fact seems to be paradoxical. This is because the parameter "Here" is usually interpreted as 'Where I am situated'. It is possible to say that the verb prijti 'to come' denotes a situation in which the place of observation is hardly connected with a point where the movement finishes. Ju.S.Maslov and Ju.D.Apresjan (Maslov 1984; Apresjan 1988) already noted that the final point of this movement is crucial, but my aim is to emphasise that its actuality has a deictic nature. The aspectual relevance of this fact is in the following. It is well-known that Russian imperfective verbs with the prefix PRI- cannot express the processual aspectual meaning, which is one of the main variations of imperfective semantics. In my opinion it is precisely because the point of departure of spatial deixis ("Here") is strictly attached to the end of such an action. But the interpretation of an action as a process requires its real or imaginary perception before it finishes (from other observation posts). In other words, a perceiver has to have a practical or theoretical opportunity to observe the gradual transition from earlier moments of the action to later ones, and this is why his/her "Here" has to be movable. However, the lexical semantics of such verbs with the prefix PRI- does not allow such observation. Thus, we can say that before the action prijti 'to come' ends, "I" as a point of departure of the personal deixis and "Here" as a point of departure of the spatial deixis do not coincide in space. In fact, it is necessary to note that they do coincide after the action finishes (cf. (9)) but not before. This explains why example (10) sounds rather strange in Russian: (\Q)Uanepr&jol 'I have not come', without special contextual efforts, cf.: (11) Oni men/a zdali, no ja ne pr&jol, potomu cto polucil ix zapisku sliskom pozdno They were waiting for me but I did not come, since I received their message too late' In the last example we can see a transitional case from the "primary" deixis to the "secondary" (narrative) one, which has to be a special topic. In the "secondary" (textual)

110 Langue and Parole in Synchrony and Diachrony deixis, "Here" is, certainly, not actual, but, in general, a significant place for narration (i.e. it is connected with the anaphoric and cataphoric expression of spatial localisation, and this is usually typical for secondary deixis). As to primary deixis, the example is doubtful and it seems to be an argument in favour of the unity of all the deictic parameters and the common point of departure for personal, temporal, and spatial axes. A similar feature has been noted for the English verb 'to come' (Fillmore, 1966; cf. Levinson, 1996), but it does not correspond to the Russian verb prijti absolutely accurately. The imperfective variation of the Russian analogue cannot express the processual meaning, whereas 'to come' seems to express it. Thus, example (12): (12) Look, he is coming '*Posmotri, on prixodit1, is acceptable and has processual meaning. In the Russian translation, we cannot take the verb prixodit'\o come1 for such a meaning and have to choose verbs with other prefixes, for instance POD-, cf: (13) Smotri, on podxodit k nam 'Look, he is coming to us (approaching us)' It makes it possible to conclude that the connection between the point of departure of the spatial deixis and the end of the action denoted by the English verb "to come" is not as strict as it is in the semantics of the Russian verb prixodit'.

2.2. Perception and semantics of ingressivity / completivity The notion of an observer allows us to perform a detailed lexico-grammatical analysis within the framework of different verbal groups. Thus, ingressive verbs can have two meanings, compare the verb zasumet'w\\h the strictly ingressive meaning ("to begin to rustle") or the meaning "to be rustling and to become available for perception by somebody". In Russian dictionaries these meanings are usually presented as different (cf. Slovar' russkogo jazyka, 1981-1984), e.g.: (14) Postepenno rassvelo. Zastrekotali kuzneciki, zapestreli cvety 'Gradually dawn has broken. The grasshoppers have begun to chirr, the flowers have started to be colourful', but:

Aspect, Voice and Deixis in Russian 111 (15) My vysliiz gluxoj cascinapoljanu. Zastrekotali kuzneciki, zapestreli cvety 'We have come out of the thicket into the glade. The grasshoppers have begun to chirr, the flowers have started to be colourful' In my view the main difference between the specified meanings of such verbs is determined by the fact that in the first utterance we find a fixed perceiver (a potential perceiver or the "omniscient" author's "I"), whereas, in the second utterance, the perceiver is moving. This difference can be seen among the verbs not only of the ingressive Aktionsart, but also, probably, of completive (also finitive) Aktionsart and the phase verbs nacinat' 'to begin' and koncat' 'to finish'. Depending on fixity / movability of the perceiver, the seme of ingressivity or completivity in the actional content of such predicates can be substituted by the seme "to become available for perceiving" (in the case of completivity "to cease to be available for perceiving"). Cf. the following example: (16) Oni exali bystro. Vperedina gorizonte nacalisijat'ogon 'ki They were driving quickly. Ahead of them on the horizon some lights have begun to shine' When we use the verbs nacat'sja, koncit'sja 'to begin (to be begun), to finish (to be finished)' we feel the presence of a certain observer. It was noted by Apresjan (1986) (this fact has served for Apresjan as an argument in favour of deicticness of the perfective aspect). In my view, using the imperfective variant of the verbs 'to begin, to finish' (nacinat'sja, koncat'sja) is always connected with perceptibility (and, consequently, with deicticness) as well. Let us consider example (17): (17) Dom byl odnoetaznyj, nedaleko ot ugla Tverskoj. Cuvstvovalas' blizost' Brestskoj zeleznoj dorogi. Rjadom nac mails'ejo vladenija, kazjonnye kvartiry sluzascix, parovoznye depo i sklady The house was one-story, it was located not far from the corner of Tverskaja Street. The nearness of Brest railway could be felt. Its properties began not far from here, there were state apartments of employees, railroad depots and storehouses' (B.Pasternak) If we change the imperfective form to the perfective form in this example, we will have the following sentence:

112 Langue and Parole in Synchrony and Diachrony (18) Dom byl odnoetaznyj, nedaleko of ugla Tverskoj. Pocuvstvovalas' blizost' Brestskoj zeleznoj dorogi. Rjadom nacalls' ejo vladenija, kazjonnye kvartiry sluzascix, parovoznye depo isklady The house was one-story, it was located near the corner of Tverskaja Street. The nearness of Brest railway was (literally: had been) felt. Its properties began (literally: had begun) not far from here, there were state apartments of employees, railroad depots and storehouses', and the only difference will be that in example (18) movability of the observer is presented, whereas in example (17) some additional context is needed to realise whether the observer is moving or not. However, the perceptibility itself is implied in both events. In general, it is possible to state that reflexive phase verbs of the type nacinat'sja, loncat'sja 'to begin (to be begun), to finish (to be finished)' can mark real initial or final phases of actions probably only in cases where the subject is expressed by a deverbal noun (or - more widely - a noun with the semantics of an action), cf.: Rassvet /poljot nacinalsja /koncalsja The dawn / the flight was beginning / finishing1 etc. On the other hand, if the subject is expressed by concrete nouns, either initial or final "subjective" phases of their perception are meant by the phase verbs. When we say: Gorod/'les nacinalsja /koncalsja The city / the wood began / finished' - we mean in fact that their observation began or finished. With the verb prodolzat'sja 'continue (to be continued)' the situation is more complicated, since examples of using it in the phase meaning occur rather rarely (this verb usually expresses the semantics 'to last', cf.: EtoprodoEalos'dva casa This lasted for two hours'). However, in the phase meaning it has a similar characteristic as well, for instance: (19) Onipovernulikreke. Vokrug prodoEalls' sady They have turned towards the river. The gardens around them continued ("were continued")' (i.e. the gardens were still being available for their direct perception) In some cases the imperfective verbs nacinat'sja 'to begin (to be begun)', koncat'sja 'to finish (to be finished)' are not connected with a single concrete perceiver. Usually this can be at the beginning of a text or an episode in which the author describes concrete or imaginary space, in which events will be developed. For instance:

Aspect, Voice and Deixis in Russian 113 (20) Dom, v kotorom zili Antipovy, naxodilsja v casti goroda, protivopoloznoj pristani. Dom bylposlednim na ulice. Za nim nacinalos'pole. Egoperesekala zeleznaja doroga. Bliz linii stojala storozka. Cerez rel'sy byl prolozen pereezd The house, where the Antipovs lived, was located in a part of the city opposite the pier. The house was the last in the street. A field began after it. The field was crossed by a railway. A gatehouse stood near the railway. Across the rails a crossing was built' (B.Pasternak) Here it is necessary to recall the notion of the "omniscient author" (Maslov, 1984). The "omniscient author", in my view, can be considered as a sort of "super-perceiver", who has originally seen a given space in his/her internal vision, and, thereafter, it "is to be", and any reader can see it, if (s)he would like to. In such cases using the verbs nacinat'sja 'to begin (to be begun)', koncat'sja 'to finish (to be finished)1 becomes possible because any potential perceiver can move in his/her imagination through a given space. In conclusion I would like to point out that the above discussed interrelations between perceptibility and the semantics of ingressivity / completivity can be interpreted more widely. As a matter of fact, here it is necessary to speak of the interconnection between perceptibility and the semantics of a phase character. This interconnection is, certainly, not casual. "Beginnings" and "ends" of actions fix borders between the action and its absence (or other action), and, naturally, present the most keenly perceived areas in the actional structure of the world. Interrelations of perceptibility with phases, in particular with initial and final borders of an action, are, in my opinion, a base on which regular correlations between the category of aspect and deixis are built.

2.3. The role of the perceiver in the expression of processual / relative meanings by verbs of motion ("displacement in space") The verbs of displacement in space are especially interesting for investigation from this point of view since several different variations of perceptibility (in particular, fixity / movability of the perceiver) can strongly influence their aspectual functions. These verbs are able not only to express a process developing towards its limit (presented by the corresponding perfective verb, in many cases), but, in a number of contexts, can express a relative meaning as well and denote a certain state, or, more exactly, a spatial position of a person / thing. For instance:

114 Langue and Parole in Synchrony and Diachrony (21) A Margarita veto vremja uze podnimalas'stremitel'no vverxpolestnice... 'and Margarita, meanwhile, was already going upstairs rapidly...' (M.Bulgakov) (22) Vdalisverkala 1'dami, podnimalas'iz-za rosci ostrympikomgornaja versina 'in the distance a lofty summit rose with its sharp peak behind a grove, and sparkled with ice1 (A.N.Tolstoy) In the Slovar' russkogo jazyka (1981-1984), processual and relative meanings of verbs of the type podnimat'sja 'to go up / to rise', oJcruzat"to surround', uxodjf'to go away / to leave' are connected, as a rule, with different lexical meanings of verbs. In my view, in the semantics of such verbs (as well of many other verbs of displacement in space), the lexical content is determined by their grammar (so-called aspectual particular meanings). Here we deal with the specific interaction of lexical and grammatical (as well as, probably, cognitive) regularities, which needs special analysis. Let us consider the realisation of the processual semantics of the given verbs in detail. For the analysis of processual situations presented by such verbs (as well as for the analysis of relative situations), we need to consider the following elements: 1) observer (perceiver) (P); 2) subject of displacement (S); 3) parameters specifying spatial deixis: a) place where the displacement occurs (Loc), b) direction in which the displacement is produced (Dir), c) place where the perceiver is located (PI). To consider the interrelation between processual and relative semantics of verbs of displacement in space in detail, it is possible to present the cases (21) and (22) as two extreme points, between which the whole gamut of transitional cases can be located: process

relative

Denoting a process in utterances with such verbs is usually connected with a really moving participant, for instance: (23) [Teterev.]...Ty idi, idi![Nil.]Nu, xoroso,ja uxozu. (Uxodit v seni, ne zametiv Tat'janu...) '[Teterev.] ...Go, go! [Nil.] O.K., O.K., I am going. (Moves into the hall, not noticing Tatyana...)' (M.Gor'kij). Note that it is not necessary for the moving participant to be animate, cf.:

Aspect, Voice and Deixis in Russian 115 (24) Solnce bystro uxodilo za gorizont, i takze bystro nadvigalis', tocno bessumno polzli, so vsex storon prezdevremennye sumerki The sun was quickly going behind the horizon, and the early twilight was in the same way quickly moving, as if it crept noiselessly from all sides' (N.G.Garin-Mixajlovskij) To realise processual semantics it is also essential that the observer either moves him/herself ('I was going along the road uphill'), or associates him(her)self with a moving subject (agent) (You / Sasha / A bus was going along the road uphill). One of the important stages between the processual and relative semantics in the chain of transitional cases is a type of verb where a special - "relative" - process is expressed. For these verbs it is distinctive that the observer is moving, whereas a thing which is interpreted in speech as a subject (agent) of displacement is not moving (though it seems paradoxical). For instance: (25) Navstrecuim vydvigaetsja, podnimajas'szemli,

tjomnyjxolm...

Towards them a dark hill is moving forth, rising from the ground...' (M.Gor'kij); (26) Samoljot zaxodil na posadku. S zemli podnimalis', sverkaja ognjami cut' pravee, neboskrjoby The plane was landing. From the ground a little to the right some skyscrapers rose up, sparkling with lights' (27) Ja netoroplivo posjol vles znakomqj dorogoj. Stanovilos'spokojnee, i berjozy vsjo tesnee okruzali men/a... 'I went leisurely into the wood by the familiar path. It became calmer, and some birches were surrounding me more and more closely...' (M.Prisvin) The specifics of such cases consists in the fact that the observer attributes a process of displacement to a perceived thing, whereas in reality (s)he him(her)self moves. The utterance does not denote the process which occurs in reality (moving of the perceiver), but expresses the other process, which is simulated (modelled) by a speaker (moving of the described thing/person). From the standpoints of physics we are dealing here with a simple and absolutely normal principle of relativity, not even in the style of Einstein, but of Newton. The moving observer has the right to consider (and to interpret) the world from his/her own point of departure, assuming that (s)he is at rest, and it is the things or persons which are moving relative to him/her. Such cases are usual in using the verbs uxodit'\o leave (to go away), upJyvat"sai\ away', ot"ezzat'\o depart (to drive off)', cf.:

116 Langue and Parole in Synchrony and Diachrony (28) Ja smotrju, kak uxodjat otnasiskryvajutsja vlesu ogon'kistancii 'I am looking at how some lights of the station are going away from us and disappear into the wood1 (I.Bunin) Probably one of the criteria for the ability of such verbs to express both the concrete-processual and a relative meaning (naturally in different conditions) is that their lexical semantics includes the possibility for the observer to attribute or to delegate his/her own moving to the perceived person/thing as a special "counteragent". However, I must say that this criterion is important, though incomplete. Thus, the verbs uplyvat' 'sail away', of'ezzat' 'to depart (to drive off)' correspond to this criterion, but they are not able to express statal-relative spatial semantics. It is also possible to select a type of case which demonstrates one more step towards the relative meaning of the verb podnimat'sja 'to go up / to rise1. Cf.: (29) Ja idu vsjo medlennee. Tropinkapostepennopodnimaetsja na serye skaly... 'I am going more and more slowly. The path is gradually going up / rising on the grey rocks...' The principle of relativity also works in such cases. In example (29) the observer goes up on the rocks, but he attributes a process of moving to the path, presenting it as a dynamic, unfolding in space and time. The point of departure is moving, but its dynamic character is determined by the observer or another moving person/thing, with which the perceiver associates him(her)self. Such "delegacy" of the standpoints to a stable spatial object (to a path, road, street, etc.), also becomes possible in cases where the observer does not really produce a process of movement, but this process can be produced by him/her potentially ("theoretically"), for instance: (30) Vdalidoroga podnimalas1na xolm 'Far off, the road was going uphill' In these cases, where there is only a possible, but not actually produced process, we are dealing already with strictly relative semantics, though it is connected with the processual meaning "genetically" and/or "associatively". More complicated semantic interaction may be illustrated by examples (31) and (32): (31) Vdali podnimaetsja basnja (gora i t.d.) 'Far off, a tower (a mountain, etc.) rises (goes up)1 (32) Smotrite: na gorodskoj okrainepodnimaetsja novoe zdanie 'Look, a new building is rising (going up) on the outskirts of the town'

Aspect, Voice and Deixis in Russian 117 In (31) and (32) there seems to be no trace of processual semantics, because both the perceiver and the object of perception do not move. If the verb of movement here expresses any moving, it can be only a glance (and/or the expansion of the field of vision, i.e. the integration of new fragments of the observed thing in that field). Only to this extent may we speak about the metaphorical processual meaning of the verb. The data discussed allow us to note the fact of interrelations between the processual and relative semantics of verbs of displacement in space. A relative semantics is based on a process which is connected either with a potential (theoretically possible) movement or with an act of observation. An analysis of the functions of verbs of displacement in space shows a close correlation between aspectual and deictic problematics. Moreover, it is crucial that the semantics of both processual and relative types concerns not only temporal, but also spatial deixis. The deictic approach to aspectual problems has much to offer. In an aspectual "calculus" of actions, conditions, relations, etc., the speaker has at his/her disposal a well organised system of points of departure and observation posts. Study of the system of such points of departure can lead perhaps to the revision of many questions concerning particular aspectual meanings of both the imperfective and perfective aspect.

REFERENCES Apresjan, Jurij D. (1986). Dejksis v leksike i grammatike i naivnaja model' mira. Semiotika iinformatika, vypusk 28. Apresjan, Jurij D. (1988). Glagoly momental'nogo dejstvija i performativy v russkom jazyke. In: Rusistika segodnja. Jazyk: sistema i funkcionirovanie. Moskva. Apresjan, Jurij D. (1995). Izbrannye trudy, t. II. Integral'noe opisanie jazyka isistemnaja leksikografija. Moskva. Bondarko, Alexander V. (1971). Vid i vremja russkogo glagola (Znachenie i upotreblenie). Moskva. Biihler, Karl. (1934). Sprachtheorie: Die Darstellungsfunktion derSprache. Jena. Clark, Herbert. (1973). Space, time, semantics and the child. In: Cognitive development and the acquisition of language (T.Moore, ed.), pp. 147-168. New York. Comrie, Bernard. (1985). Tense. Cambridge University Press, Cambridge. Fillmore, Charles G. (1966). Deictic categories in semantics of 'come'. Foundations of language, Vol. 2, 3, 219-227. Here and There: Crosslinguistic Studies on Deixis and Demonstration (1982). (J.Weissenborn and W.Klein, eds.). Amsterdam - Philadelphia. Jakobson, Roman O. (1971). Shifters, verbal categories, and the Russian verb. Selected Writings. II. Word and Language. Mouton, The Hague, 1971, pp. 130-147. Keenan, Edward. (1976). Towards a universal definition of "subject". In: Subject and topic (Ch.N.U, ed.). New York.

118 Langue and Parole in Synchrony and Diachrony Kozinskij, I.Sh. (1983). O kategorii podlezhashchee v russkom jazyke. Predvaritel'nye publikacii, vypusk 156. Leiss, E. (1992). Die Verbalkategorien des Deutschen. Berlin - New York. Levinson, S. (1996). Language and space. Annu. Rev. Anthropol., 25, 353-382. Maslov, Jurij S. (1984). Ocherki po aspektologii. Izdatel'stvo Leningradskogo universiteta, Leningrad. Paducheva, Elena V. (1996). Semanticheskie issledovanija: Semantika vremeni i vida v russkom jazyke. Semantika narrativa. Izdatel'stvo Shkola "Jazyki russkoj kul'tury", Moskva. Slovar1russkogo jazyka. V 4-x tomax. (1981-1984). Moskva.

10

THE INTEGRATION OF THE GERMAN MODALS INTO THE PARADIGM OF VERBAL MOODS Gabriele Diewald, Institutfur Germanistik, Universitat Erlangen, Germany

Abstract The German modals show two distinct uses: a typically narrow scope lexical reading, and a highly grammaticalized wide scope reading, in which the modals function as factuality markers. It is argued that the grammaticalized modals are deictic signs, and that this is the distinctive feature integrating them into the deictic category of verbal mood. The hypothesis is that the three inflectional moods (indicative, subjunctive I, subjunctive II) and the periphrastic modal constructions form a highly differentiated deictic grammatical category in present-day German. The distinctive values of this grammatical category are described with reference to Biihler's concept of deixis, which recognizes three different "modes of deixis".

INTRODUCTION The six German modals durfen 'to be allowed to', konnen 'can, to be able to', mogen 'to like, may', mussen 'must, to have to', sollen 'shall, to be to' and wollen 'to want1 show two distinct uses, which are illustrated by mussen in (1) and (2): (1) Aber jetzt muflt du natiirlich erst das Semester zu Ende bringen, ne? (Texte 63) 'but now, of course, you must finish the semester first, right?' (2) Dann mufi ihm langsam sein Kollege [...] unheimlich geworden sein. (Zeit 52) 'then his colleague [...] must gradually have given him the creeps' These two uses of the modals show different degrees of grammaticalization (Diewald, 1993; 1997; Heine, 1995; Valentin, 1984; cf. also Traugott, 1989 for the English modals). The use in (1), commonly referred to as "deontic" or "agent-oriented" use, shows the less grammaticalized stage. Here, the modals are lexical verbs with distinct semantic content. They have narrow scope, i.e. they make a predication on the subject of the sentence. Thus, mufit in (1) predicates the state of 'being obliged' of the subject. Sentence (2) illustrates the so-called "subjective epistemic" or "inferential" use, in which the modals are highly grammaticalized and approach the stage of auxiliaries. Mufi in (2) does not contribute to the prepositional content of the sentence, i.e., unlike mufit in (1) it does not express the obligation of the subject. Instead, it expresses an uncertain degree of the factuality of the whole proposition, i.e., it has wide scope. The use of the modals as factuality markers is the present endpoint of an ongoing grammaticalization process. The target category of this process, i.e., the grammatical category into which the grammaticalized modals are integrated, is verbal mood, which in present-day German consists of three inflectional members: the indicative, the subjunctive I, and the 119

120 Langue and Parole in Synchrony and Diachrony subjunctive II. This tripartite inflectional paradigm of mood is expanded by the additional values denoted by the periphrastic modal constructions. Verbal mood, like tense, is a deictic category expressing different values of speaker-based factuality judgements about the proposition (Jakobson, 1971/1957:135f; Rauh, 1984:52f; Donhauser, 1987:6). Therefore, it is argued here that the modals in their grammaticalized use, i.e. in their "subjective epistemic" reading, are deictic signs as well, and that it is this deictic function that - in various specifications - builds up the oppositional values in the category of mood. The claims put forward in this paper can be listed as follows: The modals in their grammaticalized use are deictic signs. Their deictic function integrates them into the grammatical paradigm of verbal mood. The three inflectional verbal moods (i.e. indicative, subjunctive I, and subjunctive II) and the periphrastic modal constructions (i.e. the grammaticalized uses of the six modals) form a highly differentiated deictic grammatical category in present-day German. The distinctive values of this grammatical category can be described with reference to Buhler's concept of deixis, which recognizes three different "modes of deixis": deixis proper, i.e. deixis in the actual deictic field of the utterance, "imagination-oriented deixis", and "the anaphoric use of deictic words" (1990/1934). The next section gives a brief sketch of the deictic process and the category of mood as a deictic category. After that the distinctive values of the six grammaticalized modals in this paradigm will be explored.

DEIXIS AND VERBAL MOOD Deictic signs link the utterance to the speech situation, i.e. to its nonlinguistic context. As the context of an utterance is constituted by and centered around its speaker, the concept of context can be narrowed down to the speaker, the "origo" as Buhler calls it. He goes on to define it as the source or zero point of the "coordinate system of 'subjective orientation', in which all partners in communication are and remain caught up" (1990/1934:118). From this definition it is evident that the deictic process can be described as a relational structure between a source, a path and a goal (Sennholz, 1985; Diewald, 1991). This basic relational structure is given in (3): (3)

Basic relational structure of the deictic process:

source origo

—> —»

path [deictic value]

-» —»

goal object of the deictic locating process

The distinctive feature of any deictic process is the fact that the position of the source is taken over by the origo. The remaining two positions, the path, i.e. the deictic value, and the goal, i.e. the deictic object, can be specified in various ways. This accounts for the different values concerning distance (near vs. remote) and dimension (e.g. temporal, personal, modal) which deictic signs can express. As Buhler points out, the relational structure of the deictic process can be applied to different deictic fields, i.e., the origo can be abstracted from the present speaker and transferred to some other point. These modifications create different "modes of deixis". The first mode is deixis proper, or "demonstratio ad oculos" as Buhler calls it (1990/1934:94). In this mode the deictic

German Modals and Verbal Mood 121 process takes place in the actual deictic field of the utterance. Deixis proper is the basic deictic mode, from which the other modes are derived. These derived modes are "imagination-oriented deixis" (which translates "Deixis am Phantasma") and "the anaphoric use of deictic words" (Buhler, 1990/1934:137ff.). All three modes of deixis are necessary to describe the differentiations of the system of verbal mood. As mentioned above, verbal mood is a deictic category. Therefore, the basic relational structure for verbal mood can be derived from the general deictic relation in (3) by the specifications shown in (4): (4) origo

Basic relational structure of verbal mood: —>

[deictic modal value]

—>

proposition

The speaker is the source of the locating process; the goal position is occupied by the whole proposition. The deictic value in the modal dimension is the degree of factuality of the proposition as it is judged by the speaker. It should be noted that the term "factuality" does not mean "objective truth", but refers to the observer-based (which without explicit marking of the contrary means speaker-based) judgement of the narrated event concerning its state of existence, reality, actuality (see e.g. Helbig and Buscha, 1986:506, who describe "Faktizitat" as concerning the "Tatsache-Sein der [...] Proposition"). The point of reference for the degree of factuality of the proposition is the degree of factuality that the speaker attributes to herself. As the speaker regards her existence, her state of being at the moment of utterance, as factual, this state represents the near value: a narrated event that is judged by the speaker to be on the same level of existence receives the value [factual], while any narrated event that is not judged that way receives a remote value (Rauh, 1984:52). The unmarked, deictically near value of factuality, i.e., the assessment of a proposition as factual, is represented by the formally unmarked mood, the indicative. All other members of the grammatical paradigm of mood apart from the indicative (the subjunctive I, the subjunctive II, and the deictic modals) express some marked value. The indicative signals that the speaker regards the proposition as factual, as "related to reality" (cf. "auf die Wirklichkeit bezogen", Heidolph et al, 1981:522). Thus, the meaning of the indicative can be paraphrased as: "A factuality value that from the position of the speaker is regarded by the speaker as 'near' and is attributed to the scene described in the proposition so that the scene is the goal of this deictic attribution of a factuality value." Its relational structure is shown in (5): (5) origo

Relational structure of the indicative: —>

[factual]

—»

proposition

Since the speaker is the source of the deictic locating process and the whole process goes on in the actual speech situation, the indicative represents the basic mode of deixis, i.e. deixis proper. Before it is possible to describe the specific values of the two subjunctives, we need to take a closer look at the modifications the deictic process can undergo. "Imagination-oriented deixis" is a form of deictic displacement ("Versetzungsdeixis", Sennholz, 1985:224). It differs from deixis proper because, rather than using the actual deictic field as context, it uses an imagined deictic field instead. The relational structure is displaced from

122 Langue and Parole in Synchrony and Diachrony the actual speaker to another imagined origo (Rauh, 1978:101). An example is (6): (6) We spent our Easter vacation in Italy. Here the sun was shining all day and we went for long walks across the hills. The origo of here obviously is not identical with the present position of the speaker; it is displaced to the location the speaker is talking about, i.e. Italy mentioned in the preceding sentence. The most important areas of deictic displacement are the different forms of reported speech, in which the category of mood and the modals are highly relevant. hi this domain, the most straightforward candidate for "imagination-oriented deixis" is the subjunctive I, as its primary function in present-day German is that of a quotative marker (Jager, 1971). The subjunctive I signals reported speech, as is illustrated by the subjunctive-I-form set in (7). The bracketed someone said in the translation is not present in the German sentence; it is implied by the use of the subjunctive I: (7) Nicht die Idee sei gescheitert, sondern ihre praktische Ausfuhrung. (Spiegel 172) '(someone said) it was not the idea that failed, but its realization in practice1 The subjunctive I signals that the factuality judgement does not come from the present speaker, but from a reported speaker, i.e., from a displaced origo. The present speaker is not committed to the factuality value of the proposition. Instead, she states that it is the reported speaker who judged the proposition as factual. This can be captured by the diagram in (8): (8)

Relational structure of quotative markers (e.g. subjunctive I):

r

deictic relation I

origo -> speaker

[factual]

—»

, embedded displaced relation n

(displaced origo —» [factual] —>• proposition) reported speaker

(8) shows that the subjunctive I as a quotative marker combines two relational structures: a deictic relation and a displaced relation, hi the deictic relation the present speaker states that there is a reported speaker who attributed a factuality value to the proposition. The second, embedded relation consists of the reported speaker judging the proposition as [factual]. For the subjunctive II, Biihler's third mode of deixis, the "anaphoric use of deictic words", is relevant, hi this third mode the deictic relation is displaced as a whole into the symbolic field, i.e., into the text itself. There is no anchoring in the surrounding nonlinguistic situation. I will refer to this as the phoric mode and distinguish between anaphoric and cataphoric according to the direction of pointing. The most obvious phoric elements are preforms like the personal pronoun it in (9), which anaphorically refers to the noun phrase a new bicycle that appears in the preceding text. (9) She bought herself a new bicycle. It was extra light and easy to ride. The phoric process is made use of in the subjunctive II as well, which, in the following, will be explained in some detail. The subjunctive II signals that the proposition is regarded as nonfactual by the speaker, hi a sentence like (10) Sie stiinde im Garten 'she would be standing in the garden' the subjunctive II signals first of all that the proposition:

German Modals and Verbal Mood 123 (1 Oa) Sie steht im Garten 'she is standing in the garden' is [nonfactual]. However, this is not its dominant function: unqualified nonfactuality is expressed by sentence negation, i.e. by (1 Ob) Sie steht nicht im Garten. 'she is not standing in the garden' The subjunctive II contains some further meaning, which makes up its distinctive function. As Kasper (1987:106) shows, this distinctive function of the subjunctive II is a reference to an unfulfilled condition, whereby the fact that the condition is unfulfilled causes the nonfactuality of the proposition expressed in the sentence. The unfulfilled condition is often represented in the surrounding text, and since the subjunctive II refers to that condition it qualifies as a phoric element, i.e., in addition to its deictic function it has the phoric function of connecting textual positions. In this respect the subjunctive II is comparable to the pluperfect, which beside its temporal deictic function (it always expresses past time in relation to the moment of speaking, i.e., in relation to the origo) also has an additional phoric function in that it evokes a secondary temporal reference point that is usually mentioned in the text. Similarly, in the modal domain, the subjunctive II also combines two relational structures: a deictic relation and a phoric relation. In the deictic relation the speaker points to the connection between the unfulfilled condition and the proposition. She states that it is true that there is a condition which influences the degree of factuality of the proposition. The embedded phoric relation connects the nonfactuality of the proposition to the unfulfilled condition. This is shown in (11): (11)

Relational structure of the subjunctive II:

I

deictic relation

, origo ->

[factual]

1

embedded phoric relation

1

-> (unfulfilled condition -> [nonfactual] -» proposition)

The diagrams (11) and (8) highlight the parallel structures of the two subjunctives. Both are combinations of two relational structures: the subjunctive I combines a deictic relation and a displaced relation; the subjunctive II combines a deictic relation and a phoric relation. It might seem strange that in this diagram the origo attributes the value [factual] to the structure in brackets. However, this attribution does not concern the proposition itself, but the relation between the unfulfilled condition and the proposition. Thus, the speaker states that it is true that there is a condition whose being unfulfilled causes the proposition to receive the value [nonfactual]. It is the existence of this relation that the speaker commits herself to by using the subjunctive II, and, since it expresses a definite factuality value, it can be questioned, denied, etc. just as indicative or negated indicative utterances can. Thus, if sentence (10) is denied by No, that is not true! the denial does not concern the proposition 'She is not standing in the garden' that is expressed by the subjunctive II via its second, embedded relational structure (cf. the feature [nonfactual] in the bracketed relation in 11). What is denied by No, that is not true! is the claim that there exists a factual relation between an unfulfilled condition and the proposition. This can

124 Lcmgue and Parole in Synchrony and Diachrony be paraphrased as "There is no unfulfilled condition that is relevant to the proposition 'She is not standing in the garden1." The basic functions of the verbal moods can be summed up as follows: the indicative expresses the deictic value of modal nearness, i.e. factuality, which is expressed by the feature [factual]. It contains one single relational structure and represents the deictic mode of deixis proper. The subjunctive I as a quotative marker indicates a dislocated origo and contains two relational structures: the first one points from the present origo to the fact that there exists a factuality judgement of the proposition by the reported origo; the second relational structure is a displaced relation which consists of the reported origo judging the factuality of the proposition as [factual]. The semantic content of the subjunctive II is a marked value that combines a deictic relation and a phoric relation, whereby the former states the factuality of the connection between the unfulfilled condition and the proposition, and the latter states that the unfulfilled condition is the source due to which the proposition receives the value [nonfactual]. The following section will show that the deictic use of the modals can be integrated into this verbal paradigm, and that the modals can be grouped under different modes of deixis.

THE INTEGRATION OF THE DEICTIC MODALS INTO THE PARADIGM OF MOOD Mussen and konnen The factuality value expressed by the deictic modals can be made explicit by paraphrases containing a sentence adverbial that conveys the meaning of the modal, or a superordinate clause. For the meaning otmufi in (2) this is illustrated by (2a) and (2b): (2) Dann mufi ihm langsam sein Kollege [...] unheimlich geworden sein. (Zeit 52) 'then his colleague [...] must gradually have given him the creeps' (2a) Ich halte es fur wahrscheinlich, daB ihm langsam sein Kollege [...] unheimlich geworden ist. 'I regard it as probable that his colleague [...] gradually gave him the creeps' (2b) Bestimmt ist ihm langsam sein Kollege [...] unheimlich geworden. 'certainly, his colleague [...] gradually gave him the creeps' In a similar way, the grammaticalized use of konnen in (12) can be paraphrased by (12a) and (12b): (12) Ich kann mich getauscht haben. (Radio) 'I may have been mistaken1 (12a) Ich halte es fur moglich, daB ich mich getauscht habe. 'I regard it as possible that I was mistaken' (12b) Vielleicht habe ich mich getauscht. 'perhaps I was mistaken' The paraphrases spell out the different degrees of doubt or certainty the speaker has concerning the factuality of the proposition. Stating that one regards something as probable, likely, possible and so on, amounts to stating that one - in fact - does not know, and therefore does not commit oneself to the truth of the proposition. The modals in this use signal that the speaker cannot attach a definite value of factuality to the proposition (Lyons 1977:800). From the speaker's perspective the proposition cannot be assessed as factual or nonfactual: it may be either.

German Modals and Verbal Mood

125

Thus, the meaning of the grammaticalized modals is that the speaker has tried to evaluate the factuality of the proposition but has not reached a definite conclusion: The speaker cannot (or does not want to) determine the "location" of the proposition in the deictic modal space. This indetermination of the speaker concerning the factuality of the proposition is the distinctive value of the modals. It is noted here as [factual/nonfactual]. The respective relational structure is given in (13): (13) origo

Relational structure of the deictic modals: —>

[factual/nonfactual]

—>

proposition

This deictic function, the indication that - seen from the speaker's perspective - the proposition has an indeterminate location in modal space, is the essence of this use of the modals. With the help of the relational diagrams, this deictic structure can be made explicit and comparable to deictic relations in other semantic domains. The semantic difference between konnen and mussen is a reflection of their non-deictic uses, where konnen expresses 'ability', 'possibility', etc., and mussen 'necessity' and 'obligation'. Accordingly, kann in (12) and mufi in (2), although sharing the feature [factual/nonfactual], express different affinities towards the values involved: deictic konnen signals a tendency or preference of the speaker towards judging the proposition as [factual], deictic mussen towards [nonfactual]. Mussen and konnen are the least complex of the grammaticalized modals, since they express deixis proper: they contain a single, purely deictic relation between the origo and the proposition. The remaining four modals contain additional information; they are composed of several relational structures. The following sections describe these additional distinctions as realizations of different modes of deixis.

Sollen and wollen The deictic use ofsollen and wollen in (14) and (15) is generally described as indicating reported utterances, which is spelled out in the respective paraphrases (14a) and (15a): (14) Der Arzt und die Kosmetikerin sollen 1993 noch zwei weitere Morde geplant haben. (FN 95) 'the doctor and the beautician are said to have planned two more murders in 1993' (14a) Jemand sagt: "Der Arzt und die Kosmetikerin haben 1993 noch zwei weitere Morde geplant". 'someone says: "The doctor and the beautician planned two more murders in 1993"' (15) Er will damals gehort haben, dafi Auslander bis zu 6000 Mark fur einen "Umverteilungsantrag" zahlen wiirden. (Zeit 74) 'he claims to have heard at that time that foreigners would pay up to 6000 DM for an application for "redistribution"1 (15a) Er sagt: "Ich habe damals gehort, dafi Auslander bis zu 6000 Mark fur einen Umverteilungsantrag zahlen wiirden". 'he says: "At that time, I heard that foreigners would pay up to 6000 DM for an application for redistribution'"

126 Langue and Parole in Synchrony and Diachrony The modals in this use are quotatives and, as has been shown for the subjunctive I, signal a displaced origo. Therefore, the basic structure of the quotative modals sollen and -wollen is the same as the one for the subjunctive I in (8). However, the quotative modals are not just substitutes for the subjunctive I, but differ from it in the explicitness of the location of the reported origo. As is well known, sollen specifies the modal source as external, which means the reported speaker can be anyone except the subject of the sentence. Wollen, on the other hand, specifies the modal source as internal, which means the reported speaker is identical with the subject of the sentence. This distribution, of course, mirrors the structure of the respective non-deictic uses of the two modals: sollen as a deontic modal implies an external source, wollen as a volitional modal implies an internal source. Different from quotative sollen and wollen, the subjunctive I does not specify the location of the reported speaker, it merely indicates the displacement of the reported origo. Thus, there is a subtle but clear-cut specialization of the three quotative markers subjunctive I, sollen and wollen.

Diirfte and mag The remaining modals dtirfen and mogen pose some problems for the exact description of their deictic use. They are often treated together with konnen and mtissen as merely expressing different degrees of likelihood, so that all four non-quotative modals are arranged on a scale from certainty to likelihood. However, the positions of dtirfen and mogen on such a scale are unclear and there are contradictory suggestions (see e.g. Buscha, 1984:215 vs. Ohlschlager, 1989:206). A four-step gradation of factuality values, therefore, is not very convincing, and I want to suggest a different solution that explains the meanings of these two modals as a combination of deixis proper with the phoric mode. Dtirfen and mogen show highly restricted formal paradigms. A deictic interpretation of dtirfen is possible only with its subjunctive II forms, while in the case of mogen it is all forms except the subjunctive II that express deictic meanings. This distribution hints at a functional specialization, which is made more likely by the fact that both modals in their deictic use occur only in restricted contexts. Diirfte is mainly used in written texts that report on opinions and attitudes of others. It often appears towards the end of a paragraph in sentences expressing a summary or prognosis. An example is (16), which is paraphrased in (16a) using the adverbial vermutlich to substitute deictic dtirfte: (16) In den Wintermonaten Januar/Februar allerdings diirfte die Kurve - wie immer in der kalten Jahreszeit - einmal noch weiter nach oben gehen. (FN 95) 'in the winter months January/February, though, the graph might - as always in the cold season - go up even higher once again1 (16a) hi den Wintermonaten Januar/Februar allerdings geht die Kurve vermutlich - wie immer in der kalten Jahreszeit - einmal noch weiter nach oben. 'in the winter months January/February, though, the graph presumably - as always in the cold season - will go up even higher once again' Typically, the factuality judgement expressed by dtirfte is not the independent judgement of the writer or speaker herself but a judgement that is derived from what has been said in the preceding text. Very often the present writer reports the judgement of someone else, who has been explicitly quoted before. In these cases dtirfte resembles the quotatives. However, there are two differences:

German Modals and Verbal Mood 127 first, while the quotatives signal that the reported speaker judged the proposition as [factual], durfte signals that the reported speaker judged the proposition as [factual/nonfactual], i.e. durfte reports the doubts or beliefs of someone else. Secondly, durfte in most cases signals that the present speaker not only reports the doubts of the reported speaker, but also shares them. Generally speaking, durfte represents the factuality judgement of the present speaker as being taken over from a factuality judgement mentioned or quoted in the preceding text. Durfte, thus, is a peculiar combination of the phoric and the deictic modes. At this point it becomes clear why it is only the subjunctive II of durfen that allows this use. In the preceding section the basic meaning of the subjunctive II was analyzed as containing a phoric relation, which shows that the factuality value of the proposition is connected to and influenced by some condition that is often mentioned in the preceding text. In the case of deictic durfte, this phoric pointing of the subjunctive II receives a specific interpretation: it points back to a given factuality judgement of someone else, which the present speaker takes up and adopts. Thus, the specific value of durfte is the combination of deixis proper and anaphoric backward pointing. An analogous interpretation can be given for deictic mag, the typical use of which is exemplified by sentence (17): (17) Die Leute mogen das so empfmden, richtig ist es dennoch nicht. (Spiegel 29) 'people may feel like that, it is not right, though1 The most obvious feature of deictic mag is its concessive meaning, which in (17a) (17a) Zugegeben, die Leute empfinden das vielleicht so, richtig ist es dennoch nicht. 'admittedly, people perhaps feel like that, it is not right, though' is paraphrased by a combination ofzugegeben and vielleicht. Although this concessive component has often been mentioned, it has usually been treated as some peripheral connotation of this modal. However, it is exactly this concessive component that accounts for the distinctive function of mag in the system of deictic modals, and it is this part of the meaning of mag that resists its being integrated into a linear scale of likelihood together with konnen and rnussen. Similar to the analysis of durfte, I suggest analyzing this concessive sense as the result of the combination of two relational structures: The first one is a deictic relation, by which the present speaker judges the proposition as [factual/nonfactual]. The second one is a phoric relation, which connects the proposition containing the modal to a contradicting proposition following it. Since the clause containing mag always precedes the clause containing the contradiction, this phoric relation can be specified as cataphoric, i.e. forward pointing. Thus, the problem of integrating the two deictic modals durfte and mag into the set of deictic modals is solved by assuming a combination of two different modes of deixis. Beside the purely deictic factuality judgement, both modals contain an additional phoric relation: durfte contains an anaphoric connection to a factuality judgement given in the preceding text, and mag contains a cataphoric relation to a contradicting proposition, following in the next sentence.

CONCLUSION It has been shown that the modals in their grammaticalized use, which can be illustrated by sentences such as Sie mufi zuhause gewesen sein 'She must have been at home', are deictic signs. They express a speaker-based factuality judgement and thus are integrated as periphrastic forms into the system of verbal moods. The distinctive values of the grammatical category of mood can

128 Langue and Parole in Synchrony and Diachrony be subclassified according to the modes of deixis suggested by Biihler. In the group of inflected moods deixis proper is expressed by the indicative, imagination-oriented deixis by the subjunctive I, and the phoric mode by the subjunctive II. The six modals in their deictic use lead to further differentiation. Konnen and mussen express the purely deictic mode. Sollen and wollen are quotatives, and thus represent the imagination-oriented mode of deixis. Durfte and mag, finally, make use of an additional phoric relation. Thus, the application of the concept of modes of deixis reveals the high degree of functional specialization in the category of mood, and makes it possible to describe the distinctive value of each modal verb as it has become integrated into this paradigm as a factuality marker. The treatment of the modals as deictic signs has consequences for the description of the system of the verbal categories of German as well as for the general treatment of modality. Concerning the first point, the interpretation of the grammaticalized modals as deictics highlights the close parallels between the categories mood and tense (both are deictic categories, both show inflectional as well as periphrastic realizations). This permits a more uniform and consistent description of the system of verbal categories. Secondly, in the general treatment of modality, the proposed interpretation of the modals as deictics shifts emphasis from the notion of epistemicity as concerning the speaker-independent expression of logical necessity and possibility to the basically "pragmatic" function of expressing subjective judgements by the speaker at the moment of utterance.

REFERENCES Text sources FN: Spiegel: Texte:

Zeit:

Further Nachrichten, 21/22 January 1995. Der Spiegel, Nr. 11,11 March 1996. Texte gesprochener deutscher Standardsprache. Erarbeitet im Institut fur deutsche Sprache Forschungsstelle Freiburg i. Br. Vol. 3: "Alltagsgesprdche" (H. P. Fuchs and G. Schank, eds.). Hueber, Miinchen 1975. Die Zeit, Nr. 5, 27 January 1995.

Literature Biihler, Karl (1990/1934). Theory of language. The representational function of language. Translated by D. F. Goodwin. Benjamins, Amsterdam/Philadelphia [first published 1934: Sprachtheorie. Die Darstellungsfuriktion der Sprache. Fischer, Jena]. Buscha, Joachim (1984). Zur Semantik der Modalverben. Deutsch als Fremdsprache, 21, 212217. Diewald, Gabriele (1991). Deixis und Textsorten im Deutschen. Niemeyer, Tubingen. Diewald, Gabriele (1993). Zur Grarnmatikalisierung der Modalverben im Deutschen. Zeitschrift fur Sprachwissenschaft, 12,218-234. Diewald, Gabriele (1997). Die Modalverben im Deutschen: Grammatikalisierung und Poly— funktionalitat. Habilitationsschrift, Universitat Erlangen [typescript]. Donhauser, Karin (1987). Mood and morphology: an alternative approach to the syntax and semantics of German moods. Lingua, 73, 53-77.

German Modals and Verbal Mood 129 Heidolph, Karl Erich, Walter Flamig and Wolfgang Motsch (1981). Grundziige einer deutschen Grammatik. Akademie, Berlin. Heine, Bernd (1995). Agent-oriented vs. epistemic modality - some observations on German modals. In: Modality in grammar and discourse (J. Bybee and S. Fleischman, eds.), pp. 17-53. Benjamins, Amsterdam/Philadelphia. Helbig, Gerhard and Joachim Buscha (1986). Deutsche Grammatik: Ein Handbuch fur den Auslanderunterricht. Enzyklopadie, Leipzig. Jager, Siegfried (1971). Der Konjunktiv in der deutschen Gegenwartssprache. Untersuchungen an ausgewdhlten Texten. Hueber/Schwann, Miinchen/Dusseldorf. Jakobson, Roman (1971/1957). Shifters, verbal categories, and the Russian verb. In: Selected writings, Vol. 2, pp. 130-147. Mouton, The Hague/Paris. Kasper, Walter (1987). Konjunktiv II und Sprechereinstellung. In: Satzmodus zwischen Grammatik und Pragmatik: Referate anlafilich der 8. Jahrestagung der Deutschen Gesellschaft fur Sprachwissenschaft, Heidelberg 1986 (J. Meibauer, ed.), pp. 98-113. Niemeyer, Tubingen. Lyons, John (1977). Semantics. 2 vol. Cambridge University Press, Cambridge. Ohlschlager, Gunther (1989). Zur Syntax und Semantik der Modalverben des Deutschen. Niemeyer, Tubingen. Rauh, Gisa (1978). Linguistische Beschreibung deiktischer Komplexitat in narrativen Texten. Narr, Tubingen. Rauh, Gisa (1984). Aspekte der Deixis. Sprachwissenschaft, 9,23-84. Sennholz, Klaus (1985). Grundziige der Deixis. Brockmeyer, Bochum. Traugott, Elizabeth Closs (1989). On the rise of epistemic meaning in English: An example of subjectification in semantic change. Language, 65, 31-55. Valentin, Paul (1984). Zur Geschichte der Modalisation im Deutschen. In: Deutsch-franzosische Germanistik: Melanges pour Emile Georges Zink (S. Hartmann and C. Lecouteux, eds.), pp. 185-195. Kummerle, Goppingen.


IV

MORPHOLOGY


11 NOUN CLASSIFIERS AND GENDER CLASSES ANOTHER LOOK K. Connors, Universite de Montreal, Canada

Abstract Given the contrast between noun-classifier and noun-class systems, this paper identifies the latter with gender classes. It is shown that nominalizing morphemes behave as classifiers, notably in assigning gender in gender-class languages; thus if their existence is universal, so is that of classifiers. Attention is drawn to the intersection of gender and declension classes. Though the combination of these two types of classification could generate as many sub-groups of nouns as in typical classifier systems, most of the possible distinctions in gender-and-declension languages are neutralized. Finally, the nonuniversality of genders and declensions is contrasted with the universality of classifiers, and a continuum of "classifier-language" features is postulated.

i. INTRODUCTION In a well known article, Dixon (1986) distinguishes "noun classifiers" from "noun classes", the latter including gender classes. In this paper, we will generalize the term "gender class" to include all noun classes of the type that Dixon clearly distinguished from classifiers. We will assume here that there is no difference in principle between such morphological classes and what has been known in many, especially Indo-European, languages as "gender". Greenberg (1978) rightly points out that a more etymologically oriented use of the term "gender", in the sense of genre 'kind' invites such a generalization, which at the same time distinguishes the two terms of central interest here more sharply. We do not solve here the question as to the exhaustive list of features that can determine a gender-class system.

Some "noun class"

systems seem to involve a greater number of features than those traditionally associated with "gender". Dixon distinguishes these two types of noun classification on a number of parameters. Classifiers are free morphemes and separate lexemes from the nouns which they classify, while noun (our "gender") classes are marked by bound morphemes (affixes), most essentially in concordial agreement with the nouns which they classify. Classifiers, on the other hand, can

133

134 Langue and Parole in Synchrony and Diachrony replace corresponding classified nouns (as exophors or in anaphoric agreement). They are often generic nouns, i.e. "superordinates" (Dixon, 1968 and 1986; Erbaugh, 1986) in relation to the classified noun. Dixon (1986:109) notes that in Yidiny, a classifier language, the classifier minya 'edible animal' occurs with, or as an anaphor for ganguul 'wallaby'. Dixon (1968:115) mentions that in Dyabugay: "a generic term and a specific term often occur together in an NP; thus mijia 'animal meat' and wawun 'turkey' occur together in a typical NP mijia wawun". Erbaugh (1986:402) observes the same phenomenon in Mandarin: nei

-san

-zhl

+mad

those three animal cat 'those three cats'. Noun (i.e., gender) classes are obligatory at the level oflangue in the sense that every noun in a gender language must be so classified, while in classifier languages, some nouns do not take classifiers. At the level of parole, the choice and use of classifiers admits of some interspeaker (and, as other sources will indicate, stylistic) variation, while such variation is decidedly marginal in the case of gender classes, for any given variety of a gender-class language. This variability in the use of classifiers is nevertheless of interest since, in some classifier languages at least (Allan, 1977; Demuth, 1988; Irvine, 1978) there is a default class into which, for example, loan-words go. The existence of default categories will also be pertinent to classifiers in non-classifier languages and to gender-class languages, in particular, below. Perhaps most striking is the difference in the numbers of categories: Dixon (1986:106) estimates that the number of noun classes in gender-class languages is between two and twenty, while dozens of classifiers have been described in typical classifier languages, by Dixon (1982) and other scholars (Derbyshire and Payne, 1990). This difference is not surprising, given the lexical nature of classifiers and the morphological nature of gender classes. In this connection, Dixon (1986:109) notes that classifiers are typical of isolating languages, and noun (gender) classes of agglutinative and inflectional languages. It would be a mistake, however, to forget that noun classifier systems, like gender systems, manifest some grammaticalization. Just as gender assignment (especially, but not exclusively, of inanimate nouns) is often semantically arbitrary and only partially predictable in typical gender-class languages, classifier assignment is in great part semantically arbitrary and unpredictable, even if it is possible to find post hoc justifications for classifier assignments (Lakoff, 1986). Like gender-class assignments (especially for inanimates), they have to be acquired or learned, and failure to do so perfectly is a sign of incomplete competence in the individual speaker, or of change in the language (Irvine, 1978). We will be interested here in this pervasive grammaticalization of all such systems.

Noun Classifiers and Gender Classes 135

ii. THE UNIVERSALITY OF CLASSIFIERS It is common linguistic knowledge that classifiers are universal in the sense that languages in general have measure words associated with numerals and mass nouns, as in expressions like 3 cups of sugar, or a pile of sand, and collectives associated with nouns designating entities that are (or come in) groups, like animal-classifier collectives. In this common linguistic knowledge, the pervasiveness and complexity of classifiers in non-classifier languages has been underestimated. Illustration of this requires a brief anecdote about individual native English speakers' knowledge in this area. In 1998, a Canadian Broadcasting Company radio programme aired a discussion, and a long series of enthusiastic listener comments, on names for collections of animals of the same species: a litter of kittens, a pack of wolves, a swarm of bees, a pride of lions, a gaggle of geese, a wedge or a herd of swans, and so on. The overall impression was of a part of the English lexicon whose use is variable because most speakers have incomplete knowledge of the ideal collective classifiers for animals with which they have little experience. This incomplete knowledge leads to the use of default classifiers, like English bunch. Classifiers have been studied more systematically, however, in languages where they are even more pervasive, i.e., numerous, obligatory in certain (such as numeral) contexts, and therefore extremely frequent. It is mainly in such languages that classifiers are generic nouns or clearly derived from them, as we have mentioned. Although the existence of such "classifier languages" cannot be denied, we will propose that (putatively universal) derivational and compositional nominalizing morphemes are classifiers. Like superordinate generic nouns, which are covert classifiers in "non-classifier" languages, they assign gender in gender-class languages. This is related to, but different from the claim in the literature that some classifiers are derivational morphemes (Derbyshire and Payne, 1990). Such proposals, it must be admitted, broaden the notion of "classifier" to include certain kinds of bound morphemes in inflectional and agglutinative languages. Yet it escapes the accusation of evacuating the concept of true "classifier language" if (at least some) classifiers in the latter type of language are characterized as free, lexical morphemes, associated necessarily with free-morpheme classified nouns, obligatory in certain contexts where they are not in other languages, and typically overt generic nouns. Thus, in languages such as English, French, and German, derivational and compositional nominalization provides examples of classifiers, and of the variability of their use and the inter-speaker differences in knowledge of the lexicon which it reflects. Pre-school English-speaking children thus generalize the productive -ness of goodness, kindness, etc. to the nominalization of adjectives of temperature, for example: *warmness, *hotness, etc. At another level of competence, Chomsky, in his early writings for a general linguistic audience, explicitly contrasted the use of the ideal classifier -ity in grammatically with his deliberate use of -ness in grammaticalness. This intra-speaker variability also parallels reports in the

136 Langue and Parole in Synchrony and Diachrony literature on some classifier languages indicating that speakers can deviate from the usual noun classifier for metaphoric rhetorical effect (Becker, 1986). The pejorative use, in particular, of classifiers in non-classifier languages is well known in, for example, the use of derivational suffixes whose basic meaning is said to be 'diminutive'. Thus in Canadian French, from job, jobinne classifies the job as not only small but, especially, of little economic worth. More striking, perhaps, is the deviation from the ideal nominalizing suffix, exemplified by gouvernage (szc--familiar style), in contrast to standard French gouvernement, to pejorative effect. This is related to the fact that -age is the nominalizer used to borrow and caique English -ing action nominalizations (Pupier et al., 1975:118), which are part of the non-standard lexicon: bddrage 'bothering' bargui(g)nage 'bargaining' botchage 'botching' to\vi(n)gnage 'towing' shippage 'shipping'. It also recalls the use of collective nouns designating human groups or activities which classify them pejoratively, if redundantly, as in a gang of thieves, and a pack of fools or lies. Of interest on the general topic of parallels and connections between classifiers and gender classes is the gender-assigning and gender-changing behaviour of derivational and compositional nominalizers in grammatical-gender-class languages, since, as mentioned above, the existence of compounding and derivation is thought to be universal (Greenberg, 1963). Thus, while the French suffixes -ment and -age, just mentioned, assign masculine gender in the nominalization of verbal forms such as gouvern(e), -ment changes the gender of ordonnance 'prescription, order' (f.) in forming ordonnancement 'ordering' (m.). As is also well known, German gender in complex nominalizations is determined by the rightmost nominalizer: Sprachwissenschaft 'linguistics' has the feminine gender of the derivational suffix -schaft, and not the neuter gender of the nominalized INF (in this case wisseri). Zubin and Kopcke (1986:176) show the prominence not only of such nominalizing compositional morphemes as -zeug and -tier, but also of the nominalizing derivational prefix ge-, in assigning neuter gender, for example. At the same time, the classifier function is more obvious in the fact that such compound or derived nouns thus become superordinate terms (as opposed to the "basic" and "subordinate" levels): das Flugzeug 'airplane' das Fahrzeug 'vehicle' das Spielzeug 'toy' das Kriechtier 'reptile' das Saugetier 'mammal' das Weichtier 'mollusk' das Gewdchs 'plant'

Noun Classifiers and Gender Classes 137 das Getrank 'beverage' das Gebaude 'building' (Zubin and Kopcke, 1986:147-173). More impressive in their coverage and more significant for the question of gender assignment are the covert classifiers studied by Zubin and Kopcke (1986). Principally, they contrast superordinates, which strongly tend to be neuter in German, with basic level masculines and feminines in the same lexico-semantic domains: thus das Vieh 'useful farm animals' and das Wild 'useful game animals', and many more specific forms. Further, in the context of hunting, there are neuter superordinates naming, for example, a huntable species, and masculine and feminine sub-terms for the male and female of the species, respectively (pp. 15 Iff): das Steinwild 'ibex' der Steinbock 'male ibex' die Steingeifi 'female ibex' das Gams-wild 'chamois' der Gamsbock 'male chamois' die Gamsgeifi 'female chamois' das Auerwild 'mountain cock' der Auerhahn 'male mountain cock' die Auerhenne 'female mountain cock'. Neuter gender also characterizes superordinate plant names, designating heterogeneous categories like das Kraut (p. 158), for plants that are neither bushes, trees, flowers, nor grass. For Zubin and Kopcke, "basic level" plant and animal terms, which are generally of masculine or feminine gender, must designate categories which are "imageable" and "internally homogeneous". They are the more specific animals or plants which ordinary farmers, gardeners, or cooks enter into contact with, in contrast to the neuter generics, which include often relatively recent technical terms (pp. 157-160): das Kraut 'plant' vs. der Klee 'clover', die Klette 'burdock' das Wurzkraut 'culinary herb' vs. der Dill 'dill', der Kerbel 'chervil' das Unkraut 'weed' vs. die Distel 'thistle', der Lowenzahn 'dandelion' das Rind 'head of cattle' vs. der Stier 'steer', der Bulle 'bull', die Kuh 'cow' das Pferd 'horse' vs. der Hengst 'stallion', die Stute 'mare' das Schwein 'pig' vs. der Eber 'boar', die Sau 'sow' das Schaf sheep' vs. der Bock 'ram', die Zibbe 'ewe'. Although it seems clear that noun classifiers are a lexico-semantic phenomenon, they become partially grammaticalized, as we have mentioned. On the one hand, the semiobligatory connection of nouns in general with a much smaller number of classificatory lexemes means that the latter may no longer have the (presumed original) literal, specific meaning (Derbyshire and Payne, 1990; Dixon, 1982; Lakoff, 1986; Seiler, 1983). This is of course the "desemanticization" aspect of grammaticalization (Croft, 1990): G. Lakoff (1986:14-15), cites and summarizes Dixon (1982), who, however, does not seem to have placed Dyirbal among classifier languages, because of the small number of classes:

138 Langue and Parole in Synchrony and Diachrony (I) bayi: used with nouns designating men, various animals, some forces of nature, and some weapons; (II) balan: used with nouns designating women, various animals, some trees, some forces of nature, and some weapons; (III) balam: some plants, some sweets, wine, and cigarettes; (IV) bala: the "elsewhere" class in Dixon's (1982) analysis. In addition, however, although there is divergence in the literature as to whether classifiers typically are or are not "full lexemes" (Allan, 1977; Dixon, 1982 and 1986; Seiler, 1983), they are sometimes incorporated (Mithun, 1986) or reduced forms in relation to the corresponding non-classifier lexemes (Derbyshire and Payne, 1990; Dixon, 1968; Payne, 1986 and 1990; Seiler, 1983): Payne (1986:120-121) discusses classifiers, derived historically from nouns, in Yagua: "classificatory formatives correspond to the last one or two syllables of a nominal root [if not] to an entire nominal root": dasiy 'palm trunk' > dasiy 'thin pole'; jay 'skin' > jay 'pelts, cloth'; but dooriy 'house' > do (house classifier in dialect of San Jose de Loretoyacu). Mithun (1986:381) notes that "in the Munduruku sentence below, the water is first introduced by a separate noun, then, in the next clause, it is incorporated": Ti dojot puye, o'timog ip baseya'a be. water bring when they-water-place they basin in 'When they brought water, they placed it in a basin.' The incorporated classifier is sometimes a shorter form, and not always obviously related to the noun of the same meaning (Ibid.:388): "In Gunwinggu...the combining form of the term for 'water' (and 'liquid') is -bo:-, while the free noun is guggu: Gugu gabo:mangan. water it-liquid-fall 'Water is falling.'" It is surely this partial grammaticalization of classifiers which has prevented some scholars (Dixon, 1968; Irvine, 1978; Seiler, 1983) from distinguishing qualitatively, as Dixon (1982 and 1986) so explicitly does, between noun classifiers and noun or "gender" classes. There are also scholars such as Payne (1987) and Derbyshire (with Payne, 1990), who are fully aware of Dixon's distinction but call it into question.

Noun Classifiers and Gender Classes 139 III. "NOUN" OR "GENDER" CLASSES IN CONTRAST TO CLASSIFIERS A more positive and superficial reason for the confusion in part of the literature between noun classifiers and noun classes it that while the numbers of the former have been estimated at least in the dozens in studies devoted to specific languages (Derbyshire and Payne, 1990; Dixon, 1982), the label "gender" has sometimes been reserved for systems exhaustively describable in terms of approximately two binary features, generally [±feminine] and [±animate] or [±human], though in grammatical-gender systems, these features, too, are partially grammaticalized, as is well known. Even "animacy", in fact, has often been implicitly treated as a feature separate from gender, with such morphosyntactic reflexes as the form of interrogative and relative pronouns in many languages, such as English, for example. But students of noun classification in non-Indo-European languages have noticed that noun-class systems with a greater number of categories can have the same morphosyntactic properties as the gender classes known from the Indo-European family: A. Totally obligatory status, virtually without inter-speaker or stylistic variation in the choice of class markers; B. Expression by (bound) inflectional morphology; C. Obligatory marking in concordial (and not just anaphoric) agreement—a property disqualifying Modern English as a gender-class language; D. Some degree of grammaticalization, specifically involving desemanticization: "Gender" assignment is never predictable in all noun classes on semantic grounds, such that it is analyzed, by different specialists of different languages as, in part, morphologically based, an arbitrary lexical feature, or, occasionally, phonologically or phonetically motivated. We will return to this question. E. Finally, in declension-class languages, such as Indo-European and most of its daughter languages, classification by gender necessarily intersects with classification by declension. There seems always to be a relation of partial redundancy between these two noun classificatory systems, such that gender is partially predicted on the basis of declension, or vice versa, according to the analysis in question (Brosnan, 1979 and 1987; Corbett, 1982 and 1989; Cummins, 1991; Durrell, 1990; Janda, 1996; Kastovsky, 1995; MacAulay, 1992; Neidle, 1988; O'Dochartaigh, 1992; Plank, 1979). There is a striking corollary to this: Dixon (1986) and others have contrasted the great numbers of classifiers in typical classifier languages with the much more modest numbers of "noun" (our "gender") classes in the languages typifying this second phenomenon. In the gender-and-declension languages, the hundreds of theoretically possible inflected forms of individual nouns and their agreement targets are never realized. One could cite the well known example of Classical Latin, where there was already syncretism in the declensional paradigms based on gender, number, and case. Nyman (1987) summarizes

140 Langue and Parole in Synchrony and Diachrony the singular and plural declensional suffixes, mainly distinguishing cases in the five declensions, of which II, HI, and IV contained nouns of each of the three genders, but I and V had no gender distinction: Classical Latin: Case/number suffixes in declensions containing nouns of 3 genders (from Nyman 1987, Table 1, p. 252)

though the situation was not typical of the massive syncretism by which, in many languages, gender and case distinctions are neutralized in the plural. The partial redundancy of gender in relation to declension class was effected most importantly in Classical Latin by the abovementioned fact that while the second, third, and fourth declensions distinguished genders, the first and fifth did not. In the case of Modern German, recent analysis has tended to minimize the role of declension in the narrower sense of morphological case marking (Durrell, 1990; Leibiger, 1991). The neutralization of case as well as of gender marking in the plural is such that Leibiger (ibid.: 122), for example, sides with the scholars who analyze the relatively numerous pluralization patterns (seven in Durrell, 1990) as part of the derivational morphology, and, citing Antonsen (1973), postulates (p. 119) only three noun "inflectional paradigms", one for most masculines and all neuters in the singular, one for the plural, and one for uninflected nouns:

Noun Classifiers and Gender Classes 141 Modern German Noun Inflectional Paradigms (after Leibiger 1991:119) most masc.+all neuter sg. -0

pi. -0

uninflected -0

Norn.

-(a)s

-0

-0

Gen.

-(a) -0

-n -0

-0 -0

Dat. Ace.

Thus it is not that gender-class systems could not, in principle, generate huge numbers of form differences on the basis of noun classification, if combined with declension, especially when the morphological reflexes of agreement are counted. The great difference in the number of forms expressing classifier systems, on the one hand, and gender and declension-class systems, on the other, seems to reflect the essential fact that classifier systems are lexical, while genderand-declension-class systems are morphological. The latter tend toward reduction of the number of form distinctions, in the face of countervailing sources of redundancy in grammatical information. Characterizing gender-class systems as morphological, however, does not solve the problem of the basis of gender assignment. It is widely recognized that gender assignment is fundamentally semantic (Chini, 1993; Corbett, 1991; Denny and Creider, 1986), but seems everywhere to lose its semantic motivation in part, especially in the designation of (semantically) inanimate or non-human nouns (Corbett, 1989; Demuth, 1988; Herbert, 1985; Marchese, 1988; Osam, 1993; van Marie, 1995). This is the partial desemanticization that gender systems have in common with classifier systems (Allan, 1977; Irvine, 1978; Payne, 1990; Seiler, 1983). We have seen how complex nouns may receive gender assignment from declension-class assignment (according to one type of analysis), or from derivational or compositional morphemes acting as classifiers, and thus gender assigners. But what of morphologically simpler or more opaque words to which the noun-class system does not assign a semantically motivated gender, such as inanimates in binary gender systems, those of the modern Romance languages, for example? If gender assignment is not done semantically or morphologically, how is it done? Morphologically unassimilated borrowed nouns may, as in classifier languages (Allan, 1977; Dixon, 1982; Irvine, 1978; Seiler, 1983), go into a default class. This seems in general to be the solution adopted for English nouns entering Portuguese, where they are almost all assigned masculine (Teresa Pinto, personal communication). Thus gender assignment can, marginally at least, be purely lexical. Where the literature does not seem decisive is on whether gender and classifier (Allan, 1977; Irvine, 1978) assignment can normally be phonological (Karmiloff-Smith, 1979) or in the more usual formulation, phonetic (Tucker et al., 1977), thus showing a tendency associated historically with declension assignment (Brosnan, 1987; Durrell, 1990; Janda, 1996; MacAulay, 1992).

142 Langue and Parole in Synchrony and Diachrony The question has come up repeatedly in the literature for French (Desrochers, 1986; Seguin, 1969; Tucker et al., 1977). It is sometimes said vaguely that noun "endings", despite historical losses, assign gender or, from a psycholinguistic viewpoint (Desrochers, 1986; Karmiloff-Smith, 1979; Tucker et al., 1977), provide the necessary information for child acquirers to learn the gender assignment of most nouns designating non-humans (which create most of the problem). This gender assignment on the basis of "endings" has been labelled "phonetic" by Tucker et al. (1977) and Poplack et al. (1982) and phonological by KarmiloffSmith (1979). In fact, as Desrochers (1986, citing Seguin, 1969) has also noticed, the gender of nouns in French becomes more predictable as more and more sounds (or letters, in the written language) are considered from right to left, in such cases as -tion>-ion>on, for example. The probability of correct assignment is based on the relative likelihood that one has, in fact, isolated the feminine-gender-assigning derivational nominalizing suffix -tion. In the case of some other so-called "endings", it is much less clear that they are suffixes and thus gender assigners, as Desrochers, citing Tucker et a/.'s and Seguin's data, shows for the written language: Endings on nouns: percentage masculine (from Tab. I, Desrochers (1986:230))

Ending -er -oir -um -us • -al -en -ais • -ail • -et • -pe -re -que • -gue . -te

!Seguin (1969) corpus 100 100 100 100 100 100 100 100 99 52 47 45 44 44

Tucker et al. (1977) corpus 100 100 99 99 100 100 100 100 100 42 53 56 58 48

If the "ending" is not associated with categorical gender predictability, it is not the "ending" which has been isolated that is assigning gender. In such morphologically opaque words designating inanimates, assignment must, in synchrony, be called lexical. The results of psycholinguistic experiments in which child acquirers or older speakers must assign gender to nonsense or otherwise unknown nouns represent their relative success in recognizing or guessing at a possible form of a gender-assigning affix. In evaluating such data, it must also be

Noun Classifiers and Gender Classes 143 remembered that in French, as in so many other languages, most nouns are masculine (Poplack et al, 1982). Desrochers (1986) also notes what seems to be a more serious challenge from the phonological-gender-assignment school of thought. In non-standard Canadian French, certain inanimates which are masculine in the standard have been assigned invariable feminine gender (Vincent, 1991:35-40): accident, age, air, autobus, avion 'airplane', erable 'maple (tree)', escalier 'staircase', habit 'suit', hopital, hiver 'winter', etc 'summer', automne. In addition, there are standard masculines with variable gender in this variety (Ibid.:29-34): amour 'love', argent 'silver, money' avant-midi 'morning' ouvrage 'work'. What these non-standard feminines have in common is that they begin with a vowel, even when the graphic word starts with the letter h, which in the relevant cases is non aspire, i.e., the initial vowel is obligatorily permeable to liaison and elision. A reasonable speculation on the basis of the literature may be that purely lexical gender or classifier assignment (e.g., placement of all loan-words into a default class), and truly phonological gender assignment are characteristic of class-assigning systems whose semantic basis has opacified, and of nouns which are formally opaque (i.e., foreign or otherwise bereft of gender-assigning morphemes such as derivational affixes). With only two genders, roughly representing the sexes of animates, French, like many languages with a greater number of genders, has no single solution for nouns designating non-humans.

iv. THE CONTINUUM OF "CLASSIFIER LANGUAGE" FEATURES In contrast to the non-universal morphological classification of nouns, languages in general have a continuum of lexical classifier features. The "par excellence classifier language" end of the continuum could be characterized by A. The great number, relative frequency, and relative obligatoriness of classifiers (e.g., in the context of numerals or quantifiers); B. The systematicity of the use of generic classifiers as anaphors and exophors, with the concomitant low frequency of occurrence of non-classifier nouns; C. The use of superordinate, generic category words as classifiers, rather than limiting classifiers to measure words and collective nouns; D. The possible distinctness of the classifier from its historical source noun or from the semantically corresponding non-classifier noun; E. The relative predictability of classifier : noun matching on semantic grounds;

144 Langue and Parole in Synchrony and Diachrony F. The predictability of the contexts of occurrence and of variability of classifiers in langue and in discours.

v. CONCLUSION The more general problems underlying the distinction between classifier systems, on the one hand, and gender-class systems, on the other, could be formulated as questions: A. Are there qualitative differences in the grammaticalization of the two types of systems? Both show deviation from semantic motivation in the direction of arbitrary lexical classification. Do the languages near the "par excellence classifier language" end of the continuum show a different kind of grammaticalization of classifier nouns than do languages at the low end of the continuum? Is this classifier grammaticalization the first step in the evolution of classifier systems towards "noun-class" (which we equate with "gender") systems? What are the grammatical specificities of this kind of historic change? What kinds of classifier systems become gender systems, and what kinds of gender systems come from classifier systems? In other terms, what are the diachronic relations between these two types of noun classification? B. If the widely accepted postulates of the unidirectionality of diachronic change, and of the direction of change being from lexical to grammatical morphemes and meanings are right (Croft, 1990), how can changes from more grammaticalized to more semantically motivated noun-class systems be accounted for, such as the case of the gender system in English, and as has been postulated in analyses of other languages (Dixon, 1986; Herbert, 1985; Ibrahim, 1973)? Must such change be cataclysmic rather than gradual, involving the loss of all bound-morphological marks of the system? Is the new semantic motivation necessarily based on new principles (cf. Kuryiowicz, 1964; Marchese, 1988; Osam, 1993; Seiler, 1983; van Marie, 1995)? Such are the perplexing questions raised by the study of noun classification. If we are right in postulating that noun-classifiers are universal, then all languages are burdened with this puzzling type of partial redundancy: Such systems are always semantically motivated to some extent, but never entirely, in a given synchronic state. Semantic motivation seems to default mainly to morphological, or to arbitrary lexical classification, or more marginally, to classification based on phonological characteristics of the classified noun.

Noun Classifiers and Gender Classes 145

REFERENCES Allan, Keith. (1977). Classifiers. Language, 53, 285-311. Antonsen, Elmer H. (1973). Inflection and derivation in German. In: Issues in Linguistics: Papers in Honor of Henry and Renee Kahane (Braj B. Kachru, Robert B. Lees, Yakov Malkiel, Angelina Pietrangeli, and Sol Saporta, eds.), pp. 32-40. University of Illinois Press, Urbana. Becker, A.L. (1986). The figure a classifier makes: describing a particular Burmese classifier. In: Noun Classes and Categorization (Colette Craig, ed.), pp. 327-343. Benjamins, Amsterdam and Philadelphia. Brosnan, Paul W. Jr. (1979). The semantics of the Hittite gender system. Journal of IndoEuropean Studies, 7, 227-236. Brosnan, Paul W. Jr. (1987). The Latin fifth-declension nouns with nominative singular ies. Journal of Indo-European Studies, 15, 327-340. Chini, Marina (1993). Aspetti teorico-descrittivi e tipologici della categoria del genere grammaticale. Lingua e Stile, 28, 455-486. Corbett, Greville G. (1982). Gender in Russian: an account of gender specification and its relationship to declension. Russian Linguistics, 6, 197-232. Corbett, Greville G. (1989). An approach to the description of gender systems. In: Essays on Grammatical Theory and Universal Grammar (Douglas Arnold, Martin Atkinson, Jacques Durand, Claire Grover, and Louisa Sadler, eds.), pp. 53-89. Clarendon, Oxford. Corbett, Greville G. (1991). Gender. Cambridge University Press, Cambridge. Croft, William (1990). Typology and Universals. Cambridge University Press, Cambridge. Cummins, George (1991). Plural, gender, and pluralia in Czech. Slavic and East European Journal, 35, 254-27'1. Demuth, Katherine A. (1988). Noun classes and agreement in Sesotho acquisition. In: Agreement in Natural Languages: Approaches, Theories, Descriptions (Michael Barlow and Charles A. Ferguson, eds.), pp. 305-321. Center for the Study of Language and Information, Stanford, CA. Denny, J. Peter and Chet A. Creider (1986). The semantics of noun classes in Proto-Bantu. In: Noun classes and Categorization (Colette Craig, ed.), pp. 217-239. Benjamins, Amsterdam and Philadelphia. Derbyshire, Desmond C. and Doris L. Payne (1990). Noun classification systems of Amazonian languages. In: Amazonian Linguistics: Studies in Lowland South American Languages (Doris L. Payne, ed.), pp. 243-271. University of Texas Press, Austin. Desrochers, Alain (1986). Genre grammatical et classification nominale. Canadian Journal of Psychology, 40, 224-250.

146 Langue and Parole in Synchrony and Diachrony Dixon, Robert M.W. (1968). Noun classes. Lingua, 21, 104-125. Dixon, Robert M.W. (1982). Where Have all the Adjectives Gone? De Gruyter, Berlin. Dixon, Robert M.W. (1986). Noun classes and noun classification in typological perspective. In: Noun Classes and Categorization (Colette Craig, ed.), pp. 105-112. Benjamins: Amsterdam and Philadelphia. Durrell, Martin (1990). German noun inflexions: synchrony and diachrony. German Life and Letters, 43, 113-124. Erbaugh, Mary S. (1986). Taking stock: the development of Chinese noun classifiers historically and in young children. In: Noun Classes and Categorization (Colette Craig, ed.), pp. 399-436. Benjamins, Amsterdam and Philadelphia. Greenberg, Joseph H. (1963). Some universals of grammar with particular reference to the order of meaningful elements. In: Universals of Language (Joseph H. Greenberg, ed.), pp. 73-113. MIT Press, Cambridge, MA. Greenberg, Joseph H. (1978). How does a language acquire gender markers? In: Universals of Human Language III: Word Structure (Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravcsik, eds.), pp. 47-82. Stanford University Press, Stanford, CA. Herbert, Robert K. (1985). Gender systems and semanticity: two case studies from Bantu. In: Historical Semantics: Historical Word Formation (Jacek Fisiak, ed.), pp. 171-197. Mouton, Berlin. Ibrahim, Muhammad H. (1973). Grammatical Gender: Its Origin and Development. Mouton, The Hague. Irvine, Judith T. (1978). Wolof noun classification: the social setting of divergent change. Language in Society, 1, 37-64. Janda, Laura A. (1996). Figure, ground, and animacy in Slavic declension. Slavic and East European Journal, 40, 325-355. Karmiloff-Smith, Annette (1979). A Functional Approach to Child Language: a Study of Determiners and Reference. Cambridge University Press, Cambridge. Kastovsky, Dieter (1995). Morphological reanalysis and typology: the case of the German rplural and why English did not develop it. In: Historical Linguistics 1993: Selected Papers from the llth International Conference on Historical Linguistics. (Henning Andersen, ed.), pp. 227-237. Benjamins, Amsterdam and Philadelphia. Kuryiowicz,

Jerzy

(1964).

The Inflectional

Categories of Indo-European. Winter,

Heidelberg. Lakoff, George (1986). Classifiers as a reflection of mind. In: Noun Classes and Categorization (Colette Craig, ed.), pp. 13-51. Benjamins, Amsterdam and Philadelphia. Leibiger, Carol A. (1991). Inflections and paradigms in German nominal declension. In: Stcefcrceft: Studies in Germanic Linguistics (Elmer H. Antonsen and Hans Henrich Hock, eds.), pp. 115-123. Benjamins, Amsterdam and Philadelphia.

Noun Classifiers and Gender Classes 147 MacAulay, Donald (1992). The Scottish Gaelic language. In: The Celtic Languages (Donald MacAulay, ed.), pp. 137-248. Cambridge University Press, Cambridge. Marchese, Lynell (1988). Noun classes and agreement systems in Kru: a historical approach. In: Agreement in Natural Languages: Approaches, Theories, Descriptions (Michael Barlow and Charles A. Ferguson, eds.), pp. 323-340. Center for the Study of Language and Information, Stanford, CA. Mithun, Marianne (1986). The convergence of noun classifier systems. In: Noun Classes and Categorization (Colette Craig, ed.), pp. 379-397. Benjamins, Amsterdam and Philadelphia: Neidle, Carol J. (1988). The Role of Case in Russian Syntax. Kluwer, Dordrecht, Boston, and London. Nyman, Martti (1987). Is the paradigm economy principle relevant? Journal of Linguistics, 23,251-267. O' Dochartaigh, Cathair (1992). The Irish language. In: The Celtic languages (Donald MacAulay, ed.), pp. 11-99. Cambridge University Press, Cambridge. Osam, Emmanuel K. (1993). The loss of the noun class system in Akan. Acta Linguistica Hafniensia,26,Sl-\Q6. Payne, Doris L. (1986). Noun classification in Yagua. In: Noun Classes and Categorization (Colette Craig, ed.), pp. 113-131. Benjamins, Amsterdam and Philadelphia. Payne, Doris L. (1987). Noun classification in the Western Amazon. Language Sciences, 9, 21-44. Payne, Doris L. (1990). Morphological characteristics of Lowland South American languages. In: Amazonian Linguistics: Studies in Lowland South American Languages (Doris L. Payne, ed.), pp. 213-241. University of Texas Press, Austin. Plank, Frans (1979). The functional basis of case systems and declension classes: from Latin to Old French. Linguistics, 17, 611-640. Poplack, Shana, Alicia Pousada, and David Sankoff (1982). Competing influences on gender assignment: variable process, stable outcome. Lingua, 57, 1 -28. Pupier, Paul, Suzanne Lepage, and Bernard Courte (1975). -Age dans le systeme suffixal du franfais quebecois. Montreal Working Papers in Linguistics, 4, 117-157. Seguin, Hubert (1969). Les marques du genre dans le lexique du francais ecrit contemporain: compilation des cas et essai de classement. Memoire de D.E.S. (linguistique). Universite de Montreal. Seiler, Walter J. (1983). The acquisition of a noun classification system by a language. Language and Linguistics in Melanesia, 14, 76-97. Tucker, G. Richard, Wallace E. Lambert, and Andre A. Rigault (1977). The French Speaker's Skill with Grammatical Gender: an Example of Rule-Governed Behavior. Mouton, The Hague.

148 Langue and Parole in Synchrony and Diachrony van Marie, Jaap (1995). On the fate of adjectival declension in Overseas Dutch (with some notes on the history of Dutch). In: Historical Linguistics 1993: Selected Papers from the llth International Conference on Historical Linguistics (Henning Andersen, ed.), pp. 283-294. Benjamins, Amsterdam and Philadelphia. Vincent, Nadine (1991). Le genre grammatical au Quebec: deviation ou persistance ? M.A. thesis (linguistics). Universite de Montreal. Zubin, David and Klaus Michael Kopcke (1986). Gender and folk taxonomy: the indexical relation between grammatical and lexical categorization. In: Noun Classes and Categorization (Colette Craig, ed.), pp. 139-180. Benjamins, Amsterdam and Philadelphia.

12 WASISTiSTMA/M?

JozefDarski, Adam Mickiewicz University, Poznan, Poland

Abstract The author discusses briefly the history of the concept of the stem/root (German: Stamm/ Wurzel). From the beginning the concept has been understood in purely intuitive terms. Common to all of its uses is the assumption that analysable words consist of a basic part and an adjoining part. However, a definite solution as to what constitutes the basic part and the adjoining part is lacking. Subsequently, different approaches to the notions stem/root and ending (German: Endung/Sufflx) have been proposed not only by different authors but also by the same author. In order to save the notion of the ending, a number of roots for the same lexeme have been postulated. First, the author defines the notion of the lexeme in singular and in plural. Next, it is clearly stated what constitutes the fixed element of a given lexeme in singular and in plural. The author distinguishes several kinds of stems ((neutraler) referentieller Singularund Pluralstamm, (neutraler) Flexionsstamm). These notions are defined more precisely in context with a relevant part of speech (e.g. substantivischer (neutraler) Flexionsstamm) or ((neutraler) Flexionsstamm des Substantivs). In the analysis of verbs, tense and mood are also considered, e.g. (neutraler) referentieller Singularstamm im Prasens Indikativ. On the basis of the selected number of examples, the author demonstrates how a more prototypical understanding of a particular part of speech may be arrived at, e.g. (neutraler) Verbstamm. The next step involves the establishment and definition of the most general concept, which is not prejudiced by any part of speech, namely the stem/root.

Der Begriff Stamm bzw. Stammwort wurde von Schottel (1663) in Anlehnung an das Hebraische eingefuhrt: Die Ausdriicke Stammwort und Stamm (in sprachwissenschaftlicher Bedeutung) sind bei Schottel zuerst belegt (Grimm/Grimm 1984, 17:643 und 673). Dies ist ein Indiz dafttr, dafi die Idee, Stammworter zu isolieren und zwischen Stamm und Endung zu unterscheiden, von Schottel stammt (vgl. auch HaaBengier 1993). (Weber 1995:377) Diese Feststellung findet man auch bestatigt bei Jellinek (1914:141). Schottels Neuerungen werden von Weber folgendermafien beschrieben:

149

150 Langue and Parole in Synchrony and Diachrony Schottel trennt also die Endungen von den Stammwortern und listet sie getrennt auf. Zugleich identifiziert er die Endungen mit den Akzidentien, wie sich an den Formulierungen ,,zufallige Endungen" und ,,literas accidentales" zeigt. AuBerdem unterscheidet er zwischen den Bedeutungsanderungen, die von den Endungen verursacht werden, und der Konstanz der ,,Stammworter" oder der ,,litere radicales". Er nimmt so eine Segmentierung vor, die vorher nicht gemacht wurde, und veranschaulicht diese Neuerung am Beispiel des Stammwortes Reich: zum exempel will ich Reich (welches dreierley bedeutet/Z)/ves, regnum, porrige) anhero setzen/die Mannigfaltigkeit der Enderungen in significationibus ob adpositas literulas illas accidentales in etwas anzudeuten [die lat. Bedeutungsangaben sind weggelassen; start Zeilenwechsel steht Komma]: Reich, ich Reich e, im Reich e, die Reich e, Reich en, Reich em, Reich es, Reich er, Reich est, Reich et, Reich erster, Reich erste, Reich erstes, Reich erstem, Reich ersten, Reich etc, Reich eten, Reich etet, Reich ere, Reich erer, Reich eren, Reich erem, Reich ernd, Reich ender, Reich ende, Reich endes, Reich endem, Reich enden (Schottel 1663:68-69) Das Wesen-und-Akzidenz-Modell bleibt bei Schottel erhalten; das Wort-und-ParadigmaModell wird dagegen erganzt durch das Stammwort-und-Endungs-Modell. (Weber 1995:377-378) Als Stamm gait bei Schottel ein semantisch welter nicht analysierbares Simplex und als Endung all das, was auf den Stamm folgte. Erst seit Ende des 18. Jahrhunderts wurde nur die Schottelsche Endung von Meiner, Adelung und ihren Nachfolgern weiter differenziert. Schwierigkeiten hat es gegeben bei der Bestimmung der Verbstamme. Adelung (1801:258) und anfangs auch Bauer (1830:111) gehen bei der Bildung des Prasens und Prateritums vom Imperativ aus als dem einfachen Simplex. Sie betrachten diese einfache Verbform als die Wurzel. Spater gehen Bauer (1830:273) und seine Nachfolger vom hifinitiv I Aktiv aus, von dem durch Wegstreichen des n oder en der Stamm gewonnen wird. Vergleicht man sowohl in alteren als auch in neueren Grammatiken, was als Stamm/ Wurzel zu gelten habe, so kommt man zu dem Schlufl, daB dieser Begriff nicht einheitlich verwendet wird. Man hat vom synchronischen Standpunkt aus nicht reflektiert, was unter diesem Begriff zu verstehen ist, weil man ihn meistens als nicht erklarungsbedurftig betrachtet. Wird dieser Begriff ausnahmsweise erklart, so wird entweder an die Intuition des Lesers appelliert: Let us call the underlying form, which is used in the infinitive, the base root and the variations in preterite and participle, respectively, pret and part roots. (Wunderlich 1992:5) oder auf die diachronische Grammatik Bezug genommen. So heifit es z. B.: Morpheme sind also: - das Stammorphem (oder auch: Wortwurzel, Basismorphem), das direkt erkennbar ist oder aus der Sprachgeschichte erschlossen werden kann und den Ausgangspunkt aller Wortbildungen darstellt: geh-: gehen, Geher, Gang, Gehweg, Vorganger usw. (Gotze/Hess-Luttich 1989:89)

WasistStamm?

151

Der Begriff Stamm bzw. der diachronisch fundierte Begriff StammorphemlStamm-Morphem oder eine ihrer synonymen Bezeichnungen (Hauptsilbe, Stammsilbe, Wurzelsilbe, Wurzel, Wortwurzel, Stammwort, Wortstamm, Basismorphem, Grundmorphem) wird seit der 2. Halfte des 19. Jahrhunderts auch in alien anderen Arbeiten verwendet. Auf den terminologischen MiBstand machte schon 1927 Saran aufinerksam, aber sein Aufsatz wurde kaum beachtet, und so ist alles beim Alten geblieben. Die Verwirrung bei Verwendung des Begriffes Stamm wird von Gunther in Gliick 1993 treffend charakterisiert: Der Ausdruck S. ist unterschiedl. (oder uberhaupt nicht) terminologisiert und wird, oft sogar im gleichen Text, unterschiedl. verwendet. Allen Verwendungen gemeinsam ist die Vorstellung, daB ein morphol. analysierbares Wort aus einem Grundbestandteil und dem Hinzugefugten besteht; uneinheitl. ist die Auffassung dariiber, was als Grundbestandteil mit S. bezeichnet werden soil. Verunklart wird die Situation weiter durch Uberlappungen mit dem ebenfalls unterschiedlich gebrauchten Ausdruck /- Wurzel. (Gliick 1993:600) Der grammatischen Tradition folgen dennoch am Ende des 20. Jahrhunderts fast alle Grammatiker. Bei Drosdowski wird der Stamm, sieht man von der Terminologie ab, genauso wie bei Bauer 1830 bestimmt: Den Wortstamm, an dem sich die Formenbildung des Verbs vollzieht, erhalt man durch die Kiirzung des Infinitivs um seine Endung -[e]n: fahren: fahr-, wohnen: wohn-, hakeln: hakel-, klappern: klapper-. (Drosdowski 1995:116, Ziff. 211) [...] als Wurzelform der Verben muB man eigentlich nicht nach § 535 den Imperativ, sondern den Infmitiv, nach Wegwerfung der allgemeinen Verbalendung en oder n, annehmen, und der Imperativ kann nur die Wurzelform heifien, insofern derselbe mit dieser Abkurzung des Infinitivs iibereinstimmt, was freilich fast immer der Fall ist: laufen, lauf, stehen, steh, reiten, reit, fahren, fahr, trotzen, trotz. Demnach ist von geben, stehlen, flieBen, schielien usw. die Wurzelform nicht der Imperativ gieb, stiehl, fleulJ, scheuB, sondern die Abkurzung des Infmitivs geb, stehl, flielJ, schiefi. (Bauer 1830:273) Es ist manchmal auf Grund der Paradigmen schwer zu sagen, was als Stamm anzusehen ist. Man sehe sich das Konjunktivparadigma der Verben auf ein im Prasens bei Engel an: (ich) (du) (sie) (wir) (ihr) (sie)

lachl-e lachle-st lachl-e lachel-n lachl-e-t lachel-n (Engel: 1996:418)

Nimmt man an, daB die sprachlichen Mittel, die vor dem Bindestrich stehen, den Stamm bilden, so hatte man es hier, orthographisch betrachtet, mit drei Stammen zu tun: lachl-, Idchleund lachel-. Drei Stamme bei der Deklination der polnischen neutralen Substantive des Typs ciel% 'Kalb' und rami% ,Arm' (Klasse 27) nimmt auch Laskowski (Grzegorczykowa et al.

152 Langue and Parole in Synchrony and Diachrony 1998:315f.) explizite an, um fur diese Deklinationsklasse dieselben Endungen zu erhalten. (Die drei Stamme sind im Paradigma durch Unterstreichen gekennzeichnet.)

N. G. D. A. I. L. V.

Singular rami-g /ramie/ ramieni-a /raWejia/ ramieni-u /raWepu/ rami-? /ramie/ ramieni-em /ra'm>enem/ ramieni-u /ra'mJejiu/ rami-? /ramie/

Plural ramion-a ramion-0 ramion-om ramion-a ramion-ami ramion-ach ramion-a

/ra'mbna/ /'rambn/ /ra'mbnom/ /ra'mbna/ /ramb'nami/ /ra'mbnax/ /ra'mbna/

Mehrere Stamme ein und desselben Verbs werden auch von Aronoff (1993:41, 123ff.) angenommen. Die Stamme in all diesen Arbeiten konnen aber nicht als Suppletivstamme bezeichnet werden. Am Ende des 20. Jahrhunderts halt man sich also weiterhin an die traditionelle Bestimmung des Stammes: I will adopt the traditional definition of a stem as the part of a complete word form that remains when an affix is removed. (Aronoff 1993:31)

Der Begriff Affix wird in der Arbeit von Aronoff nicht bestimmt. In der grammatischen Tradition werden Affixe als Morpheme betrachtet, die an den Stamm gefiigt werden und ihn in seiner Bedeutung modifizieren. Hier mochte ich meine Konzeption der Wortanalyse, besonders der Stammauffassung, fur sich selber sprechen lassen. Eine rein synchronische Analyse der Wortformen habe ich 1986 und noch genauer 1987:55-71 dargestellt. Im folgenden werden die wichtigsten Defimtionen angefuhrt, auf denen meine Begriffsbestimmungen fuBen. Wortformen, die in Texten im Singular beziehungsweise im Plural immer dieselbe Bezeichnung aufweisen, die aber in ihrer Ausdrucksform im verschiedenen Grade voneinander abweichen, oder sich sogar total unterscheiden, werden als eine Klasse von Wortformen betrachtet, die je nach der durch sie in Texten reprasentierten Zahl als Singular- beziehungsweise Plurallexem bezeichnet wird. (Darski 1987:56)

Diese und auch die folgenden Defmitionen sind zweifelsohne nicht sprachgebunden, worauf der Begriff Bezeichnung deutlich hinweist. Der Begriff Bedeutung wird anstelle von Bezeichnung bei einzelsprachlichen Analysen verwendet. Die Begriffe Bezeichnung und Bedeutung sind folgendermaBen festgelegt worden: Die Widerspiegelung eines Ausschnittes der auliersprachlichen Realitat durch den in irgendeiner Sprache realisierten Text nennen wir seine Bezeichnung68. Die Widerspiegelung eines Ausschnittes der auBersprachlichen Realitat durch den in nur einer Sprache realisierten Text nennen wir seine Bedeutung68. (Darski 1987:37)

WasistStamm?

153

Mit der Unterscheidung zwischen Bezeichnung und Bedeutung lehne ich mich an Coseriu (1981: 47ff.) an, worauf die Ziffer 68 im Zitat hinweist. Im Deutschen weisen im Singular z. B. folgende Wortformen dieselbe Bedeutung auf: 1. 2. 3. 4. 5.

Wind, Windes, Winde du, deiner, dir, dich jener, jenes, jenem, jenen (ich) male, (du) malst, (er/sie /es) malt (ich) rate, (du) ratst, (er/sie /es) rat.

Die jeweils unter (1) bis (5) genannten Wortformen gehoren also zu ein und derselben Klasse der Singularlexeme. Analog werden zu derselben Klasse der Plurallexeme z. B. folgende Wortformen gezahlt: 1. 2. 3. 4. 5.

Winde, Winden ihr, euer, euch jene, jener, jenen (wir/sie) malen, (ihr) malt (wir/sie) raten, (ihr) ratet.

In den meisten

Wortformen eines

Singular- bzw.

Plurallexems

lafit

sich ein

Grundbestandteil und etwas Hinzugefugtes feststellen. Der Grundbestandteil wird im allgemeinem in der grammatischen Tradition als Stamm und das Hinzugefugte als Endung bzw. Suffix bezeichnet. Uneinheitlich ist die Auffassung, was als Grundbestandteil anzusehen sei. Der Grundbestandteil wird hier folgendermafien bestimmt: Die sprachlichen Mittel, die sich in alien Wortformen eines Singularlexems wiederholen, die im Falle eines regelmaBigen Wechsels durch eine angenommene {Convention (zum Beispiel durch (einen) groBe(n) Buchstaben) neutral dargestellt werden, die also dieselbe Bezeichnung aller Wortformen eines Singularlexems sichern, bilden den (neutralen) referentiellen Singularstamm. (Darski 1987:60) In dieser und den folgenden Defmitionen wird der Begriff sprachliche Mittel verwendet. Es gibt heutzutage zwei Mb'glichkeiten der schriftlichen Fixierung der sprachlichen Mittel: die Orthographic mit alien ihren Unzulanglichkeiten und die phonetische bzw. die phonologische Umschrift Zu den orthographischen sprachlichen Mitteln gehoren die Buchstaben und ihre Anordnung. Der orthographische Wechsel zwischen einem einfachen Konsonanten und einer gleichnamigen Geminate wird nicht als Erweiterung des neutralen

referentiellen

Singularstammes betrachtet, weil dieser Wechsel nur orthographisch die Kiirze des vorangehenden Vokals markiert und als ein Laut gesprochen wird, z. B.: ich nehme; du nimmst; er, sie, es nimmt; ich trete; du trittst; er, sie, es tritt.

154 Langue and Parole in Synchrony and Diachrony Der Begriff

(neutraler)

referentieller

Singularstamm mufi abhangig von der zu

analysierenden Wortart noch genauer bestimmt warden. Analysiert man z. B. Substantive, so muB noch vor dem Begriff das Adjektiv substantivisch oder nach dem Begriff des Substantivs gesetzt werden. Bei der Analyse der einfachen Verbformen miissen die grammatische Zeitform und der Modus angegeben werden, z. B. prdsentischer, indikativischer, (neutraler) referentieller Singularstamm oder (neutraler) referentieller Singularstamm im Prasens Indikativ. Analysiert man die Singularparadigmen der Verben machen, gehen, lernen und arbeiten im Prasens Indikativ: Singular 1. 2. 3.

mache machst macht

gehe gehst geht

lerne lernst lernt

arbeite arbeitest arbeitet

so sind den obigen Festlegungen nach: mach, geh, lern und arbeite ihre orthographischen referentiellen Singularstamme. Analysiert man die Singularparadigmen derselben Verben im Prasens Konjunktiv, so kommt man zu teilweise anderen Ergebnissen: Singular 1. 2. 3.

mache machest mache

gehe gehest gehe

lerne lernest lerne

arbeite arbeitest arbeite

Orthographische referentielle Singularstamme sind im Prasens Konjunktiv meiner Definition nach: mache, gehe, lerne und arbeite. Nur beim Verb arbeiten ist im Prasens Indikativ und Konjunktiv der Orthographische referentielle Singularstamm formengleich. Die Verben raten, stofien, geben und treten weisen im Prasens Indikativ Vokalwechsel auf: Singular 1. rote 2. rotst 3. rot

stofie stofit stoBt

gebe g/bst g/bt

trete trfttst trftt

Orthographische neutrale referentielle Singularstamme im Prasens Indikativ werden fur die Verben raten, stofien, geben und treten folgendermaBen dargestellt: rAt, stOfit, gib, trIT.

Was ist Stamm? 155 Die grofien Buchstaben markieren den orthographischen regelmaBigen Wechsel: A von a zu a; O von o zu o; I von e zu / und IT von et zu itt. Der Wechsel wird als regelmaBig bezeichnet, wenn er wenigstens in zwei Singular- bzw. Plurallexemen feststellbar ist. Der Wechsel von a zu a wiederholt sich auch beim Verb tragen; der Wechsel von o zu o bei verstofien; der Wechsel e zu / bei nehmen; der Wechsel von et zu itt bei betreten. Der Einfachheit halber wird der regelmaBige Wechsel immer durch den groBen Buchstaben bzw. die groBen Buchstaben gekennzeichnet, die im Singular- bzw. Pluralparadigma der Verben in den Wortformen der entsprechenden Singular- bzw. Plurallexeme meistens zweimal vorkommen. Dadurch konnen die neutral dargestellten referentiellen Singular- und Pluralstamme bzw. die neutralen Flexionsstamme mit der orthographischen Wiedergabe ihrer Wortformen ohne groBere Miihe assoziiert werden. Orthographische referentielle Singularstamme im Prasens Konjunktiv sind fur die Verben raten, stoflen, und treten: Singular 1. rate 2. ratest 3. rate

stofie stoBest stoBe

trete tretet trete

die sich in alien Wortformen wiederholenden sprachlichen Mittel: rate, stofie, trete. Zu den phonetischen bzw. phonologischen sprachlichen Mitteln gehoren die Phone (Laute als Klassen) bzw. Phoneme, ihre Anordnung und die prosodischen Merkmale. Die phonetischen bzw. phonologischen sprachlichen Mittel werden mit Hilfe der Zeichen der IPA wiedergegeben. Es ist am einfachsten, von der ,,phonologischen Aussprache" (Basb0 11/Wagner 1985:8f.) bzw. ,,Explizitlautung" (Drosdowski 1995:48f., Ziff. 44) auszugehen. In meinen Analysen wird von der phonologischen Transkription auf der phonostilistischen Ebene I (man vergleiche dazu Meinhold/Stock 1980:95ff. und 140ff. sowie ,,Gro6es Worterbuch der deutschen Aussprache" (GWdDA) (1982:73f.) ausgegangen, weil auf diese Ebene immer beim MiBverstandnis der ubermittelten Information, also in zusammenhangender Rede, zuruckgegriffen wird. Bei den Transkriptionen richte ich mich vor allem nach dem GWdDA. Zu Rate gezogen wurde aber auch DUDEN. Das Ausspracheworterbuch. In meinen Transkriptionen bleibt beim Weglassen des /9/ lediglich die silbenbildende Funktion

der

Nasale

und

des

Laterallautes

unbezeichnet,

weil

dadurch

ihre

Phonemzugehorigkeit nicht gedndert wird Es gibt viele Phonemdefinitionen, die hier nicht diskutiert werden konnen. Ich stutze mich auf meine Phonemdefinition:

156 Langue and Parole in Synchrony and Diachrony Bin Phon/Mehrere Phone, das/die an derselben Stelle/denselben Stellen des ,,mit hoherem Spannungsgrad vor grofierer Horerzahl" normgerecht realisierten Textes immer dieselbe Bedeutung des Textes gewahrleistet/gewahrleisten, bildet/bilden eine Klasse, die Phonem genanntwird. (Darski 1987:30) Die sich in den phonologischen Paradigmen der Verben: machen, atmen, gehen und bauen: Singular 1. / 'rnaxa / 2. /'maxst/ 3. /'maxt/

/ 'atma / /'atmast/ /'atmat/

/ 'ge:a / /'ge:st/ /'ge:t/

/ 'bapa/ /'ba_pst/ /'bapt/

wiederholenden sprachlichen Mittel bilden die phonologischen referentiellen Singularstamme: /'max/, /'atms/, /'ge:/ und /'bao/. Die Konjugationsparadigmen der Verben: spuren, kosen, rudern und geben weisen Phonemwechsel (Vokal- und/oder Konsonantenwechsel) auf: Singular 1. /' Jpu:ra / 2. /'JpuiBSt/ 3. /'Jpu:et/

/ 'ko:za / /'ko:st/ /'ko:st/

/ 'ru:dar3 / /'ru:dest/ /'nrdut/

/ 'ge:ba / /'gi:pst/ /'gi:pt/

Die phonologischen neutralen referentiellen Singularstamme der Verben: spuren, kosen, rudern und geben werden so dargestellt: /'Jpu:R/, /'ko:S/, /'ru:dER/ und /'gIB/. Bei der Analyse der orthographischen und phonologischen Konjugationsparadigmen wurden auch sprachliche Mittel festgestellt, die nicht zum referentiellen Singularstamm gehoren. Diese sprachlichen Mittel erfullen syntaktische Funktionen: Die sprachlichen Mittel, die dieselbe Bezeichnung eines Singularlexems nicht sichern, die aber in den Realisierungen der Wortformen desselben Singularlexems vorhanden sind, werden als syntaktische Singularexponenten bezeichnet. (Darski 1987:61) Diese Definition soil vor allem folgendes klaren: Es gibt bei der Bestimmung der verschiedenen Stammarten keine Zirkeldefmition. AuBerdem ist es moglich, die syntaktischen Exponenten von anderen Wortbildungsmitteln sauber zu unterscheiden. Auf dieselbe Art und Weise werden auch der (neutrale) referentielle Pluralstamm und die syntaktischen Pluralexponenten defmiert:

Was ist Stamm? 157 Die sprachlichen Mittel, die sich in alien Wortformen eines Plurallexems wiederholen, die im Falle eines regelmaBigen Wechsels durch eine angenommene Konvention (zum Beispiel (einen) groBe(n) Buchstaben) neutral dargestellt werden, die also dieselbe Bezeichnung aller Wortformen eines Plurallexems sichern, bilden den (neutralen) referentiellen Pluralstamm.

Der Begriff (neutraler) referentieller Pluralstamm mufi, wie beim Begriff (neutraler) referentieller Singularstamm noch genauer durch die Angabe der Wortart, bei Verben auch noch durch die Angabe der einfachen Zeitform und des Modus spezifiziert werden. Orthographische referentielle Pluralstamme der Verben: machen, gehen, lernen, raten, stofien, treten sind im Prasens Indikativ: mach, geh, lern, rate, stofi, trete: Plural 1. 2. 3.

machen macht machen

gehen geht gehen

lernen lernt lernen

raten ratet raten

stoBen stoBt stoBen

treten tretet treten

und im Prasens Konjunktiv: mache, gehe, lerne, rate, stofie, trete: Plural 1. 2. 3.

machen machet machen

gehen gehet gehen

lernen lernet lernen

raten ratet raten

stoBen stoBet stoBen

treten tretet treten

Orthographische neutrale referentielle Pluralstamme sind im Deutschen nicht vorhanden. Sie kommen aber in ihren phonologischen Reprasentationen vor. Phonologische referentielle Pluralstamme der Verben: machen, atmen, gehen und bauen lauten im Prasens Indikativ: /'max/, /'alms/, /'ge:/ und /'bao/. Plural 1. / 'maxn / 2. /'maxt/ 3. /'maxn/

/ 'atman /, /'atmat/ /'atman/

/ 'ge: an / /'ge:t/ /'ge:an/

/'bapan/ /'bapt/ /'bapan/

Sie sind mit den phonologischen referentiellen Singularstammen formengleich. Phonologische neutrale referentielle Pluralstamme der Verben: spuren, kosen und geben werden so dargestellt: /'Jpu:R/, /ko:S/, und /ge:B/.

158 Langue and Parole in Synchrony and Diachrony Plural 1. /'Jpu:ran/ 2. /'Jpu:et/ 3. /'Jpu:ran/

/'ko:zan/ /'ko:st/ /'ko:zan/

/'gerban/ /'ge:pt/ /'ge:bsn/

Ahnlich wie die syntaktischen Singularexponenten werden auch die syntaktischen Pluralexponenten bestimmt. Die sprachlichen Mittel, die dieselbe Bezeichnung eines Plurallexems nicht sichern, die aber in den Realisierungen der Wortformen desselben Plurallexems vorhanden sind, werden als syntaktische Pluralexponenten bezeichnet. (Darski 1987:62)

Da die referentiellen Singular- und Pluralstamme nicht immer formengleich sind, mussen noch der (neutrale) Flexionsstamm und die Singular- und Pluralmarker folgendermafien defmiert werden: Die sprachlichen Mittel, die sich sowohl in dem (neutralen) referentiellen Singularstamm als auch in dem ihm in der Bezeichnung entsprechenden (neutralen) referentiellen Pluralstamm wiederholen beziehungsweise einen regelmafiigen Wechsel aufweisen, der durch (einen) (beim Zusammenfall mit dem regelmaJJigen syntaktischen Wechsel indizierte(n) grofle(n) Buchstaben neutral dargestellt wird, bilden den in der Zahl nicht markierten referentiellen Stamm, der der Einfachheit halber kurz (neutraler) Flexionsstamm genannt wird. (Darski 1987:64).

Es wird wieder, abhangig von der Darstellungsweise der sprachlichen Mittel, zwischen orthographischen und phonologischen (neutralen) Flexionsstammen der zu analysierenden Wortarten unterschieden. Sind die referentiellen Singular- und Pluralstamme formengleich, so werden sie auch als Flexionsstamme gewertet. Flexionsstamme sind im Prasens Indikativ je nach der Fixierung der sprachlichen Mittel z. B.: geh /'ge:/, lern /'lern/, schneide /'Jnaeda/, mal /'ma:I/, tu /'tu:/.

Neutrale Flexionsstamme werden im Prasens Indikativ z. B. bei den Verben: geben, lesen und laden folgendermaflen prasentiert: gEB /'gEB/, IBS /'1ES/, IAD /'IAD/.

Der Vergleich des (neutralen) Flexionsstammes mit dem (neutralen) referentiellen Singularund Pluralstamm erlaubt es, die Zahlmarker festzustellen. Die Zahlmarker wurden folgendermaflen defmiert:

WasistStamm?

159

Das sprachliche/Die sprachlichen Mittel, durch das/die sich der (neutrale) referentielle Singular- und Pluralstamm vom (neutralen) Flexionsstamm unterscheidet/unterscheiden, wird/werden als Zahlmarker bezeichnet. Der Zahlmarker kann in zwei Subklassen gespalten werden: Das sprachliche/Die sprachlichen Mittel, durch das/die sich der (neutrale) referentielle Singularstamm vom (neutralen) Flexionsstamm unterscheidet/unterscheiden, wird/werden als Singularmarker bezeichnet. Das sprachliche/Die sprachlichen Mittel durch das/die sich der (neutrale) referentielle Pluralstamm vom (neutralen) Flexionsstamm unterscheidet/unterscheiden, wird/werden als Pluralmarker bezeichnet. Sind der Flexionsstamm, der referentielle Singular- und Pluralstamm formengleich, dann kennzeichnen die Zahl die syntaktischen Singular- beziehungsweise Pluralexponenten. Die syntaktischen Singular- und Pluralexponenten sind also in solchen Fallen bifunktional. (Darski 1987:64f.) Als Beispiel fur Zahlmarker sei zuerst das Paradigma von Bein angefuhrt: Singular M. D. C. B.

Bein Beines Bein Bein

/ 'baen / /'baenas/ /'baen/ /'baen/

Plural Beine Beine Beinen Beine

/ 'baena / /'baena/ /'baenan/ /'baena/

Referentieller Singularstamm 1st Bein /"baen/, referentieller Pluralstamm Beine /baens/. In beiden Stammen wiederholen sich die sprachlichen Mittel Bein /baen/. Sie bilden den Flexionsstamm. Singularmarker ist 0, Pluralmarker e fel. Die Zahlmarker lassen sich im Deutschen nur im Prasens Indikativ bei einer kleinen Gruppe der unregelmafiigen Verben feststellen, z. B. bei raten. Singular 1. rate 2. ratst 3. rat

/ 'ra:te / /'re:tst/ /'re:t/

Plural raten ratet raten

/ 'ra:ten / /'ra:tet/ /'ra:tan/

Neutraler referentieller Singularstamm ist rAt /"rAt/, referentieller Pluralstamm rate /raits/. Neutraler Flexionsstamm ist rAt /'rAt/. Singularmarker ist die Realisierung des A als a /a:/bzw. a /e:/, Pluralmarker ist die Realisierung des A nur als a Id und e bl, das nach t l\l realisiert wird. Die Konjugationsparadigmen im Prateritum Indikativ und Konjunktiv miissen genauso analysiert werden wie im Prasens. Die in meinen Definitionen neu eingefuhrten Begriffe fur die jeweiligen theoretischen Festlegungen sollten als Merkhilfen betrachtet werden. Wichtig sind also nicht die Merkhilfen, sondern die theoretischen Festlegungen, die auch anders, vielleicht viel treffender, werden konnten.

benannt

160 Langue and Parole in Synchrony and Diachrony Meine explizit dargestellte Theorie kann in alien Sprachen auf die synchronische Analyse der Wortformen eines Singular- bzw. Plurallexems angewandt werden, denn ich mache keinen Unterschied zwischen flektierenden und nicht flektierenden Sprachen. Ich habe in meiner Theorie auch die Tatsache beriicksichtigt, dafi die syntaktischen Exponenten nicht unbedingt am Ausgang des Flexionsstammes stehen miissen, worauf schon Meiner (Jellinek 1914:184) und spater auch Schlegel (1808:47f.), wohl in Anlehnung an Meiner, hingewiesen haben. Als Beispiel eines nicht am Wortende stehenden syntaktischen Exponenten kann die polnische Wortform marszatek 'MarschalP angefuhrt werden, in der das e als syntaktischer Exponent des Nominativs Singular zu werten ist. Alle dieselbe Bedeutung aufweisenden Stammarten einer Wortart mussen miteinander verglichen werden. Beim Verb werden zuerst die prasentischen und prateritalen neutralen Flexionsstamme im Indikativ und Konjunktiv miteinander verglichen. Sie konnen formengleich, z. B. arbeite /'arbaete/ (Prasens Indikativ und Konjunktiv) und arbeitete /'arbaeto te/ (Prateritum hidikativ und Konjunktiv) oder verschieden sein. Sind sie verschieden, so mussen die (neutralen) prasentischen Flexionsstamme aufgrund der sich wiederholenden sprachlichen Mittel festgestellt werden. Die konjunktivischen prasentischen und prateritalen Flexionsstamme sind in ihrer Zeitform immer formengleich und gehen auf e fa/ aus. Die konjunktivischen prasentischen Flexionsstamme unterscheiden sich von den ihnen in ihrer Bedeutung entsprechenden indikativischen prasentischen Flexionsstammen bei alien regelmaBigen und unregelmaBigen Verben ohne Vokalwechsel im Singular durch das zusatzliche e fa/, das als Konjunktivmarker gewertet werden muB. Zum Beispiel sind mal /'ma :!/ und geh /'ge:/ indikativische prasentische, male /'mails/ gehe /geis/ konjunktivische prasentische Flexionsstamme. In den beiden Stammen wiederholen sich die sprachlichen Mittel mal /'ma:l/ und geh /'ge:/. Die sich wiederholenden sprachlichen Mittel werden als prasentischer Flexionsstamm bezeichnet. Es wird also von der Opposition Indikativ : Konjunktiv abstrahiert. Die unregelmaliigen Verben mit Vokal- und/oder Konsonantenwechsel im Prasens Indikativ weisen im Prasens Konjunktiv denselben Wechsel wie im Indikativ nicht auf und gehen im Konjunktiv auch auf e fa/ aus. Deshalb wird der prasentische neutrale Flexionsstamm so wie der neutrale Flexionsstamm im Prasens Indikativ dargestellt. Ahnlich ist es auch im Prateritum Indikativ und Konjunktiv. Die prateritalen indikativischen und konjunktivischen Flexionsstamme sind bei den meisten regelmafiigen Verben formengleich. Da das Prateritum Konjunktiv der unregelmafligen Verben im Vergleich mit dem Prateritum Indikativ immer urn e fa/ erweitert ist und auch Vokalwechsel aufweist, wird der Einfachheit halber der indikativische neutrale Flexionsstamm als prdteritaler neutraler Flexionsstamm gewertet. Es wird also genauso wie im Prasens von der Opposition Indikativ : Konjunktiv abstrahiert. Jeder indikativische prasentische bzw. prateritale neutrale referentielle Singular- und/oder Pluralstamm hat seine konkreten Realisierungsvarianten, die am genauesten in der

Was ist Stamm? 161 phonologischen Transkription zum Vorschein kommen. Zum Beispiel wurde fur das Verb geben der neutrale indikativische prasentische referentielle Singularstamm /gIB/ und der neutrale indikativische prasentische referentielle Pluralstamm /'ge:B/ postuliert. /'gIB/ hat zwei Realisierungsvarianten: /'gi:p/ und /geib/, die die Wahl der syntaktischen Singularexponenten festlegen: Nach /'ge:b/ kann nur /a/ folgen, nach /'gi:p/ nur /st/ bzw. lit.

Ahnlich ist es auch im Plural, wo /'ge:B/ als /ge:b/ oder /"ge:p/ realisiert wird:

Nach dem pluralischen /'ge:b/ folgt /an/, nach /'ge:p/ nur /t/. Die neutralen indikativischen prasentischen referentiellen Singular- und Pluralstamme wurden miteinander verglichen, um festzustellen, welche sprachlichen Mittel sich wiederholen. Da sowohl die indikativischen prasentischen neutralen referentiellen Singular- als auch die ihnen in ihrer Bedeutung entsprechenden indikativischen prasentischen neutralen referentiellen Pluralstamme verschiedene Formen aufweisen, muBte noch eine zweite Abstraktionsstufe postuliert werden. Der neutrale indikativische prasentische Flexionsstamrn von z. B. geben wurde so dargestellt: /'gEB/. Dieser neutrale Flexionsstamrn hat seinerseits zwei Realisierungsvarianten: /'gIB/ und /'ge:B/. Schematisch konnte dies folgendermafien prasentiert werden:

Der neutrale indikativische prateritale Flexionsstamrn /'ga:B/ hat im Singular nur die Realisierungsvariante /'ga:p/, aber im Plural zwei Realisierungsvarianten /gcrp/ und /'ga:b/. Schematisch kann es so dargestellt werden:

162 Langue and Parole in Synchrony and Diachrony

Urn den neutralen Flexionsstamm dieses Verbs zu erhalten, mufl man noch eine Abstraktionsstufe postulieren. Der neutrale Flexionsstamm wird so dargestellt: /'gAB/. Er hat zwei Realisierungsvarianten: die prasentische /'ge:B/ und die prateritale /'ga:B/. Schematisch wird dieser Sachverhalt so dargestellt:

Der Vergleich des prasentischen neutralen Flexionsstammes mit dem prateritalen neutralen Flexionsstamm zeigt den Unterschied zwischen Prasens und Prateritum an. Die beiden grammatischen Tempora unterscheiden sich beim Verb geben durch den Vokalwechsel, bei arbeiten durch das Prateritalsuffix te. Auf dieselbe Art mussen noch alle Prafigierungen des Verbs geben dargestellt werden. Zum Beispiel werden die neutralen Flexionsstamme der Verben: begeben, vergeben, umgeben so dargestellt: /bs'gAB/, /fte'gAB/, /um'gAB/. Die sich bei alien diesen Verben wiederholenden sprachlichen Mittel /'gAB/ werden kurz als neutraler Verbstamm bezeichnet. Dies ist die hochste Abstraktionsstufe bei der Analyse der verbalen Stamme. Der Vergleich eines neutralen Flexionsstammes mit dem ihm in seiner Bedeutung entsprechenden neutralen referentiellen Singular- und Pluralstamm erlaubte festzustellen, ob die syntaktischen Exponenten mono- oder bifunktional sind. Es envies sich, dafi man auch bei einer kleinen Anzahl von Verben, ahnlich wie beim Substantiv, sprachliche Mittel feststellen kann, die den Unterschied zwischen Singular und Plural allein ausdriicken. Daraus folgt, daB die Festlegung der neutralen referentiellen Singular- bzw. Pluralstamme sowie der neutralen Flexionsstamme nicht reine Spielerei, sondern von grofiem theoretischen Wert ist. Es ist auch moglich, die (neutralen) Stamme einer Wortart mit den ihnen in ihrer Bedeutung entsprechenden (neutralen) Stammen anderer Wortarten zu vergleichen. Zum Beispiel kann der Verbstamm arbeite /'arbaeta/ mit den Substantivstammen Arbeit /"arbaet/, ^ ^ Arbeiter /'arbaete/, -> Arbeiterin /'arbaetsnn/ u.a. sowie dem Adjektivstamm arbeitsam /'arbaetzam/ u.a. verglichen werden. Die sich in all diesen Stammen wiederholenden sprachlichen Mittel bilden den Stamm

Was ist Stamm? 163 /'arbaet/. Dieser Stamm stellt die hochste Abstraktionsstufe der in verschiedenen Wortarten *-/ vorkornmenden Stamme dar. So einfach wie beim Stamm /'arbaet/ ist der Sachverhalt bei anderen Stammen nicht. Es sei hier nur an den neutralen Verbstamm /'gAB/ erinnert. Zuerst muBten noch die neutralen Stamme anderer Wortarten (z. B. die neutralen Substantiv-, Adjektivstamme) festgestellt werden, dann miteinander verglichen werden, um den neutralen Stamm zu erhalten. Aus meinen bisherigen Ausfuhrungen diirfte hervorgehen, daB ,,das Stammorphem (oder auch: Wortwurzel, Basismorphem)" nicht immer ,,direkt erkennbar ist oder aus der Sprachgeschichte erschlossen werden kann" (Gotze/Hess-Luttich 1989:89), sondern daB es durch synchronische Analysen und stufenweise Abstraktionsschritte gewonnen werden muB, wenn man den Sachverhalt korrekt prasentieren will. Der Begriff Stamm kann jetzt folgendermaBen defmiert werden: Die sprachlichen Mittel bzw. deren neutrale Reprasentationen, die sich in alien Wortformen verschiedener Wortarten wiederholen und immer dieselbe Bezeichnung aufweisen, werden als Stamm bezeichnet.

LlTERATUR Adelung, Johann C. (1801). Deutsche Sprachlehre fur Schulen, vierte verbesserte und mit einer kurzen Geschichte der Deutschen Sprache vermehrte Auflage. Vossische Buchhandlung, Berlin. Aronoff, Mark (1996): Morphology by Itself. Stems and Inflectional Classes. MIT, Cambridge/Massachusetts/London: (= Linguistic Inquiry Monograph Twenty-Two) Basb011, Hans and Johannes Wagner (1985). Kontrastive Phonologic des Deutschen und Danischen, segmentale Wortphonologie und -phonetik. Niemeyer, Tubingen. (= Linguistische Arbeiten 160). Bauer, Heinrich (1830). Vollstdndige Grammatik der neuhochdeutschen Sprache. Dritter Band. Reimer, Berlin. Coseriu, Eugenio (1981). Textlinguistik. Eine Einfuhrung, 2., durchgesehene Auflage. Herausgegeben und bearbeitet von Jorn Albrecht. Narr, Tubingen. (= Tiibinger Beitrage zur Linguistik 109). Darski, Jozef (1986). Syntaktische Exponenten der Lexeme. Studia Germanica Posnaniensia XV, 17-25. Darski, Jozef (1987). Linguistisches Analysemodell. Deflnitionen matischer Begriffe. Wydawnictwo Naukowe UAM, Poznan.

grundlegender gram-

164 Langue and Parole in Synchrony and Diachrony Drosdowski, Giinther (ed.) (1995). DUDEN. Grammatik der deutschen Gegenwartssprache, 5. vollig neu bearbeitete und erweiterte Auflage. Bibliographisch.es Institut & Brockhaus, Mannheim. Eisenberg, Peter (1994). Grundrifl

der deutschen Grammatik. 3., uberarbeitete Auflage.

Metzler, Stuttgart/Weimar. Engel, Ulrich (1996). Deutsche Grammatik, 3., korrigierte Auflage. Groos, Heidelberg. Gliick, Helmut (ed.) (1993). METZLER. Lexikon Sprache. Metzler, Stuttgart/Weimar. Gotze, Lutz and Ernest W. B. Hess-Liittich (1989). Knaurs Grammatik der deutschen Sprache. Sprachsystem und Sprachgebrauch. Droemer/Knaur, Munchen. Grimm, Jakob and Wilhelm Grimm u.a. (1984). Deutsches Worterbuch. 33 Bande. Neudruck Munchen [1. A. Leipzig 1854-1960]. Grofles Worterbuch der deutschen Aussprache (zitiert als GWdDA) (1982). Bibliographisches Institut, Leipzig. Grzegorczykowa, Renata, Roman Laskowski and Henryk Wrobel (eds.) (1998). Gramatyka wspolczesnego j?zyka polskiego. Morfologia. Wydanie drugie, zmienione. Wydawnictwo Naukowe PWN, Warszawa. HaaBengier, Ralf M. (1993). Vom Wort zum Morphem. Segmentierung und Klassifizierung in einigen dlteren und neueren Grammatiken. Magisterarbeit, Tubingen (Typoskript). Jellinek, Max H. (1913). Geschichte der neuhochdeutschen Grammatik von den Anfangen bis auf Adelung. Bd. 1. Winter, Heidelberg. Jellinek, Max H. (1914). Geschichte der neuhochdeutschen Grammatik von den Anfangen bis auf Adelung. Bd. 2. Winter, Heidelberg. Kloeke, Wus van Lessen (1982). Deutsche Phonologic und Morphologie. Merkmale und Markiertheit. Niemeyer, Tubingen. (= Linguistische Arbeiten 117). Kloeke, Wus van Lessen (1994). Die finiten Formen des deutschen Verbs. Eine morphologische Analyse. In: Laisina, Inge, Wus van Lessen Kloeke and Hans Elster: Zdsur. Zum Abschied von Gregor Pompen am 1. September 1993, 81-93. Vakgroep Duits, Nijmegen. Lieb, Hans-Heinrich (1980). Words as Syntactic Paradigms. In: Brettschneider, Gunter and Christian Lehmann (eds.). Wege zur Universalienforschung, 115-123. Narr, Tubingen Mangold, Max (1990). DUDEN. Ausspracheworterbuch. Worterbuch der deutschen Standardaussprache,

3.,

vollig

neu

bearbeitete

und

erweiterte

Auflage. Duden,

Mannheim/Wien/Zurich. Meinhold, Gottfried and Eberhard Stock (1980). Phonologic der deutschen Gegenwartssprache. Bibliographisches Institut, Leipzig. Richter, Helmut (1982). Zur Systematik der Personalendungen des deutschen Verbs. In: Detering, Klaus, Jiirgen Schmidt-Radefeldt and Wolfgang Sucharowski (eds.). Sprache beschreiben und erkldren. Akten des 16. Linguistischen Kolloquiums, Kiel 1981. Band 1. Niemeyer, Tubingen.

WasistStamm? 165 Saran, Franz (1927). Stamm, Wurzel, Hauptsilbe. In: Schadel, B. and W. Mullert (eds.)Philologische Studien aus dem romanisch-germanischen Kulturkreise. Karl Voretzsch zum 60. Geburtstage und zum Gedenken an seine erste akademische Berufung vor 35 Jahren. Niemeyer, Halle/Saale. Schlegel, Friedrich (1808). Ueber die Sprache und Weisheit der Inder. Ein Beitrag fuer Begruendung der Althertumskunde. Nebst metrischen Uebersetzungen indischer Gedichte. Mohr und Zimmer, Heidelberg. Schottelius, Justus G. (1663). Ausfuhrliche Arbeit von der Teutschen HauptSprache. 2. Bde. Braunschweig. [Neudruck Tubingen 1967]. Thieroff, Rolf (1992). Das finite Verb im Deutschen. Tempus - Modus - Distanz. Narr, Tubingen. (= Studien zur deutschen Grammatik 40). Weber, Heinrich (1995). Worter fur den Satz. Alte und neue Formen der Wortgrammatik. In: Hindelang, Gotz Eckard Rolf and Werner Zillig. Der Gebrauch der Sprache. Festschrift fur Franz Hundsnurscher zum 60. Geburtstag. LIT, Munster. Wiese, Bernd (1994). Die Personal- und Numerusendungen der deutschen Verbformen. In: Kopcke, Klaus-Michael (ed.). Funktionale Untersuchungen zur deutschen Nominal- und Verbalmorphologie, 161-191. Niemeyer. Tubingen. (= Linguistische Arbeiten 319). Wunderlich, Dieter (1992). A Minimalyst Analysis of German Verb Morphology. In: Theorie des Lexikons. Arbeiten des Sonderforschungsbereichs 282. Nr. 21. Heinrich-Heine-Universitat, Diisseldorf. Zifonun, Gisela, Ludger Hoffmann, Bruno Strecker, Joachim Ballweg, Ursula Braufie, Eva Breindl, Ulrich Engel, Hehnut Frosch, Ursula Hoberg, Klaus Vorderviilbecke (1997). Grammatik der deutschen Sprache, Bd. 1- 3. de Gruyter, Berlin/New York.


V

SEMANTICS


13

ONE SIZE DOES NOT FIT ALL: A SEMANTIC ANALYSIS OF 'small/large' VS. 'little/big Yishai Tobin, Ben-Gurion University of the Negev, Be 'er Sheva, Israel

Abstract This paper presents a semantic analysis of the so-called synonymous pairs: 'small/large' vs. 'little/big' which will explain their subtle semantic distinction as found in the following examples: (1) (2) (3) (4)

I had a little pizza/brandy. I had a small pizza/brandy. He has a big mouth. He has a large mouth.

This semantic analysis is based on the sign-oriented, functional hypotheses that these forms differ in their invariant meaning or signifie. This difference is expressed by a marked distinctive semantic feature (Semantic Integrality) which is related to the perception of entities in continuous versus discontinuous space, time, or existence. According to this analysis the forms small/large are marked for Semantic Integrality and therefore describe entities collectively: i.e., as part of a set viewed in continuous space, time or existence. The data will be analysed on the microlevel: i.e., sentences in context as well as on the macrolevel: i.e., in the form of semiotic-oriented discourse analyses in the 'from sign to text' and 'from text to sign' approaches. True, these are subtle differences, but who says semantic distinctions have to be gross? (Bolinger 1977:17)

INTRODUCTION In this paper I will present a semantic analysis of the so-called synonymous pairs: 'small/large' vs. 'little/big' which will explain their subtle semantic distinction as found in the following examples (1-4): (1) (2) (3) (4)


This semantic analysis will be based on the fundamental sign-oriented, functional hypotheses that: (a) these forms differ in their 'core', or invariant meaning, or signifie (e.g., Saussure 1913/1959, Waugh and Rudy 1991); (b) this difference in meaning is expressed by a marked distinctive 169

170 Langue and Parole in Synchrony and Diachrony semantic feature in the sense originally postulated by the Prague School (e.g., Andrews and Tobin 1996, Tobin 1988); (c) this marked distinctive feature (Semantic Integrality) (e.g., Tobin 1990, 1994) is based on the perception of entities in continuous versus discontinuous space, time, or existence associated with the Psychomechanics School of Language (e.g., Hirtle 1982); (d) according to this analysis the forms small/large are marked for Semantic Integrality and therefore describe entities collectively: i.e., as part of a set viewed in continuous space, time or existence; (e) these forms are part of a semantic system in the sense of the Columbia School (e.g., ContiniMorava 1989, Garcia 1975, Huffman 1996, Kirsner 1979, Reid 1991). The data will be analysed on the microlevel: i.e., sentences in context as well as on the macrolevel: i.e., in the form of semiotic-oriented discourse analyses originally inspired by Diver (1969) and further refined in the concepts of word systems and communicative strategies (Aphek and Tobin 1988, 1989) and in Tobin (1989,1993) in the 'from sign to text1 and 'from text to sign' approaches.

SEMANTIC INTEGRALITY The semantic feature relevant to the present analysis, Semantic Integrality, has been previously postulated as a semiotic feature of human cognition and perception. The marked feature Semantic Integrality is based on the assumption that there are two alternative ways of perceiving entities in space, time, or existence either as discrete entities (a + b = a + b) or as potentially discrete entities perceived as part of a continuous set (a + b = [ab]). Semantic Integrality is illustrated in Figure 1: Alternative Ways of Perceiving Entities in a Spatio-Temporal-Existential Cline a + b = a + b (discontinuous space, time, or existence] or a + b = [ab] (continuous space, time, or existence] SEMANTIC INTEGRALITY: THE PERCEPTION OF POTENTIALLY DISCONTINUOUS ENTITIES VIEWED IN A CONTINUOUS SET Figure 1 - Semantic Integrality Semantic Integrality may be viewed as a marked distinctive feature that can be used to explain linguistic concepts both in the grammar and the lexicon in a unified and systematic way that reflects the human perception of entities in space, time, and existence. Indeed, there may very well be a universal semantic development of linguistic forms on a spatio-temporal-existential cline going from the most concrete spatial messages to the more abstract temporal, to the most abstract existential kinds of messages: (a) existence in space or in a place (e.g., in the room), (b) existence in time (e.g., in the morning), (c) abstract existence (e.g., to be in trouble, to be in the in-group) (Tobin 1990:ch. 3). The distinctive feature of Semantic Integrality has served as the basis for analyzing the following linguistic phenomena in English and Hebrew: systems of grammatical number in general and so-called "irregular" plurals, mass/count nouns, agreement in English and the dual number in Hebrew; the quantifiers many vs. much; each vs. every; and some vs. any; the adverbs of comparison like vs. as; the connectives also vs. too; the restrictives only

A Semantic Analysis: small/large vs. little/big 171 vs. just; the deletives without vs. -less; the synthetic vs. analytic systems of comparatives and superlatives (the) more/most XJ -er/-est X; the diminishers few vs. less; the conditionals if vs. whether; possessive constructions; and contracted vs. non-contracted forms in English and Hebrew in Tobin (1990,1994). The present analysis is based on the Jakobsonian notion of markedness as it is used to distinguish between opposed members of a lexical or grammatical system (Andrews 1990, Battistella 1990, 1996). The asymmetric value relationship of the concept of binary markedness may be summarized as follows for the distinctive feature of Semantic Integrality: (a) The unmarked (U) member of the opposition is neutral. It makes no claim with regard to Semantic Integrality and is less complex cognitively. It means: Perceive Entities in Continuous or Discontinuous Space, Time, or Existence, (b) The marked (M) member of the opposition makes a specific claim regarding Semantic Integrality. It is more complex cognitively. It means: Perceive Potentially Discontinuous Entities in Continuous Space, Time, or Existence. The unmarked member of an opposition which makes no specific claim concerning the absence or the presence of a distinctive semantic feature is usually more flexible and open-ended than the marked member of an opposition; allows for any and all possibilities; is very often the first to be acquired and the one more frequently used although this does not have to be an absolute rule. The marked member of an opposition which does make a specific claim for the presence of a distinctive semantic feature is less flexible and usually more limited in its distribution because of its greater precision in meaning; is often acquired later and generally may be used less frequently than the unmarked member of an opposition although this is not necessarily an absolute rule. Therefore it should always be remembered that neither order of acquisition nor the rate of frequency is the sole criterion for choosing the markedness value. However, the asymmetric markedness relationship implies that the marked members usually may be replaced by the unmarked members, while the opposite is not necessarily true. The feature of Semantic Integrality which makes a claim for the perception of potentially discrete entities as a single continuous set has been chosen as the marked feature for the following reasons: (i) The process of first perceiving concrete or abstract entities as they appear discretely in discontinuous space, time, or existence is the fundamental perception of identification. The later perception of these potentially discrete entities as forming part of an integral set occupying continuous space is a more complex cognitive task. This greater cognitive complexity underlies and justifies the choice of Semantic Integrality as being the marked feature synergetically. (ii) If the notion of isomorphism is taken seriously, it becomes evident that the notion of integrality or fusion functions on all the levels of language as well. In phonology there is an opposition between vowels and diphthongs, non-geminated and geminated consonant phonemes, and stops/fricatives and affricates forming an obstruent consonant class, hi all of these cases the second or marked member of the opposition represents a more complex unit or sound nucleus composed of the merger, fusion, or integration of two vowels or consonants and may be marked for Integrality (Tobin 1997). (iv) hi morphology or syntax there are also simple, analytic, independent units versus more synthetic, agglutinative, or compound units composed of the fusion, merger, or integration of different bound or unbound morphemes or clause types, etc., both in word formation and in clause and sentence formation, hi all of these cases, the fused, merged, or integrated units are perceived as being more complex or marked as opposed to the simple or more neutral counterparts, (iii) There is usually some sort of iconic connection between the signals themselves and their markedness values. The signs marked for the more complex or specific

172 Langue and Parole in Synchrony and Diachrony meanings are usually larger or more complex in their signal or form, while the signs which are unmarked and are simpler and less complex in their meanings are usually simpler in their signals and forms as well.

THE ANALYSIS The marked feature of Semantic Integrality may explain the semantic distinction between 'small/large' vs. 'little/big'. Specifically, the analysis is based on the following five principles: (i) The so-called synonymous 'small/large' vs. 'little/big' share a semantic domain dealing with the relative size or the dimensions of entities, (ii) These pairs are not synonyms, however, but each member of the pair possesses a single invariant meaning which distinguishes it from the other, and will motivate its distribution in the language, (iii) These invariant meanings are in an asymmetric marked/unmarked relationship revolving around the feature of Semantic Integrality, i.e., the perception of the description of the relative size of entities that are perceived within continuous or discontinuous space, time, or existence, (iv) The marked forms small/large describe the relative size of entities perceived as sets in continuous space, time, or existence only: e.g., clothing sizes (all of the individual pieces of clothing marked small/large belong to the same set based on their dimensions), (v) The unmarked forms little/big can describe entities in sets or focus on the dimensions of individuated items perceived discretely in discontinuous space, time, or existence or viewed as being unique: e.g., to describe the dimensions of an individual's anatomy: What big feet, hands, genitals, etc., s/he has!

Microlevel Analysis Minimal Pairs I shall first return to the minimal pairs presented in the introduction: (1) (2) (3) (4)


Example (1) with the unmarked form little is used to describe the meagre amount of pizza or brandy consumed by an individual in a single act of eating or drinking viewed discretely in discontinuous space, time, or existence. Example (2) with the marked form small designates a class of pizza pies or a glass of brandy based on size (small/medium/large) to which different pizzas or glasses of brandy are grouped and classified, similar to clothing sizes: i.e., viewed as a set occupying continuous space, time, or existence. Example (3) with the unmarked form big describes the mouth of an individual not only in its dimensions but in its output: i.e., identifying it as being unique and focusing on it as opposed to other mouths. Example (4) with the marked form large describes a class of mouths based on physical dimensions (small/medium/large) (in a way similar to clothing sizes and kinds of pizza pies or glasses of brandy above).

A Semantic Analysis: small/large vs. little/big 173 Non-Free Variation In the following examples (5-9) both the marked and unmarked forms are used in the same passage or sentence to describe the same object. In addition to these examples illustrating the same communication strategies shown above, they also illustrate the concept of non-free variation in the same utterance. Example (5) shows two alternative ways of viewing a car. It is a small red car when it was rented (rental cars are classified according to their size, model and dimensions) using the marked form; but it becomes a little red car when it is involved in a near traffic accident (when the specific car and what happens to it is being described in greater detail and focused upon): (5)

The next day Sholto picked up a small red car from somewhere and drove very fast from Rome. He called out things of interest that they had passed on the way, but wouldn't stop or even slow down enough for her to see. When he suddenly spun the wheel and sent the little car hurtling up a track, she squealed, believing they had crashed ... the little red car squealed, lurched, and was saved from disaster only by the avoiding action of the donkey. (Walker 1992:371)

Example (6) describes a specific sugar caster, first individualized and focused upon by using the unmarked form and then described in more general terms as a prototype by using the marked form (in the reverse order of example (5) above): (6)

She gave a little Scottish sugar caster unusual prominence because it was small and simple. (Walker 1992: 65)

Example (7) from The Kon-Tiki Expedition (Heyerdahl 1965) juxtaposes big (specific) logs for a raft which could not be constructed because they could not find large (a class of) balsatrees. (7)

A few days later we tried to buy some balsa-wood. We wanted big logs of the wood, but there were no large balsa-trees on the coast. (39)

Example (8) contrasts small talk (a kind of conversation type) with large talk, a new term created by the author to describe another kind of conversation type: (8)

Terence was everyone's least favorite dinner guest. He had no small talk and his large talk took the form of argument. (Alther 1990:262)

Example (9) represents another example of non-free variation in the same utterance from an advertisement taken from the Sesame Street Parents Guide for a new children's toothbrush. The marked form is used for the background information on toothbrushes in general and other children's toothbrushes as a class. The forms which separate this new kids' toothbrush from all the others and accentuate its unique qualities for the specific population it is designed for are the unmarked forms: bigger and little.

174 Langue and Parole in Synchrony and Diachrony (9)

You see, unlike most kids (tooth) brushes that are merely smaller versions of adult brushes. We made our handles bigger so that it would be just right for fitting into little hands.

Idioms and Familiar Expressions It is often assumed in the linguistic literature that idioms, familiar sayings or proverbs, and other examples of frozen or fixed language, are out of bounds for linguistic analysis in general and semantic analyses in particular. The following examples present such data with the original expression accompanied or followed by a minimal pair. I maintain that the choice of the marked or unmarked form in the original expression is not arbitrary but motivated by the invariant meaning of each form; and the acceptability, or lack thereof, of the alternative expression may be explained by the invariant meanings postulated for these signs. In those cases where the signs are interchangeable, different messages are obtained according to these same invariant meanings and the communication strategies that have developed around them. (10) a. Who's afraid of the big bad wolf? (a specific canine) b. ? Who's afraid of the large bad wolf? (a genus, species) (11) a. It's a small (l*little) world... (a kind/genus) b. He lives in his own little (7smalF) world, (a specific world) (12) Hush, little (7small) baby, don't you cry... (specific child versus class of child) (13) Of All God's miracles, large and small (7big and little) (kinds vs. specific instances) (14) a. The big day arrived, (specific, great day) b. The large day arrived, (a class of day) (15) a. Thank God for small favors, (class) b. ?Thank God for little favors, (specific instance) (16) a. a small kindness (a type of kindness) b. a little kindness (a meagre amount of kindness) (17) a. Don't make a big deal over it! (a specific deal) b. ?Dorit make a large deal over it! (a complicated class) (18) a. little by little (a specific process) b. * small by small (counting entities in a class) (19) a. A little goes a long way. (a meagre amount) b. ? A small goes a long way. (a kind/class of thing) (20) a. a little love goes a long way (a meagre amount) b. ?a small love goes a long way (a kind/class) (21) a. Little Bo Peep, Big Jim (specific people) b. 1'Small Bo Peep, Large Jim (classifications of people) (22) a. a big shot (a specific person) b. a large shot (of whiskey) (an established amount) (23) a. big brother (an older brother) b. large brother (any brother of any age meeting of certain physical dimensions) hi all of the above examples the choice of one form over the other is motivated by the semantic distinction postulated for the marked/unmarked pair and conforms to the communication strategies that have been uncovered for their use in the other examples found in this paper.

A Semantic Analysis: small/large vs. little/big 175 Macrolevel Analysis From Sign to Text As was seen in examples (l)-(23) the choice of a marked versus an unmarked form is not arbitrary but motivated by a subtle semantic distinction which can be traced to the invariant meanings of the forms in question, hi other words, one opposed meaning may be seen to be more suitable to express a specific kind of message rather than the other. The repeated exploitation of this meaning to communicate that specific kind of message can be related to or associated with a particular leitmotif in both spoken and written discourse. If a specific sign or a set of signs containing the same or a similar semantic feature consistently occurs in contexts where a particular theme or leitmotif is being discussed, this will give the discourse or text a sense of coherence. After all, a text, like language itself, may be defined as a system of systems used by human beings to communicate. The ability to trace the consistent exploitations of particular signs in specific contexts within a text is called the "from sign to text" approach. It allows one to view the text in the hierarchical ascending order of sign and system to context and text. In the present paper, the "from sign to text" approach implies: (i) The distribution of the marked versus unmarked forms is not random, but is skewed along thematic lines within a specific text, (ii) The choice and preference of the marked or unmarked forms can be related directly to particular themes, specific characters, and recurring or related events in the plot or subplots within the text, (iii) Therefore the choice and preference of the marked or unmarked forms can serve as part of a larger system of textual coherence and cohesion. hi the text, Alice's Adventures in Wonderland (Carroll 1865/1970), size and changes in size play a major role in the text and may be viewed as a recurrent theme or leitmotif. Alice, herself, grows smaller and larger several times throughout the text, and each time a change in her dimensions is presented, the marked forms are used, indicating her body as a whole, integral part of herself: i.e., her individual identity. Alice, however, is not the only entity whose dimensions form a recurrent theme throughout the text. Alice initially enters the strange world of Wonderland via a passage, first introduced using the marked form small, a general description classifying the passage (small vs. large), but later on in the text, when she returns to the same passage, it is now individualized and focused upon, based upon her previous experience, making it a unique and one-of-a-kind passage, and thus it now appears with the unmarked form little: (24) (25)

Alice opened the door and found that it led to a small passage ... she knelt down and looked along the passage into the loveliest garden you ever saw. (30) Once more she found herself in the long hall... then she walked down the little passage and then — she found herself at last in the beautiful garden. (104)

Another recurrent theme in the book involves keys, locks, and doors leading to different places in Wonderland. Once again, when Alice first encounters keys, locks, and doors, she views them as generic items, and the marked form small is used. On her second encounter, with the same key, lock, and door, they become individualized and unique and the unmarked form little is used.


There was nothing on it but a tiny golden key, and Alice's first idea was that this might belong to one of the doors of the hall; but alas! either the locks were too large, or the key was too small, but at any rate it would not open any of them. However, on the second time round, she came upon a low curtain she had not noticed before, and behind it was a little door about fifteen inches high: she tried the little golden key in the lock, and to her great delight it fitted! (19-20)

The same key, originally introduced generically and then discretely in this passage, is consistently referred to by the unmarked form throughout the rest of the text (pp. 32, 36, 39, 104), thus strengthening this communicative strategy motivated by the invariant meanings of the marked and unmarked forms discussed above. Birds are another recurrent leitmotif within the text. Once again, when they first appear in the text they are perceived en masse, as a flock or group, unindividuated by Alice, the narrator, or the reader. Not surprisingly, this initial encounter describes their dimensions by exploiting the marked forms small/large (50); but later on in this passage, Alice describes how her cat Dinah catches and eats birds, causing consternation among the flock. The specific bird that Dinah catches and eats in the real world (rather than the mass of birds in Wonderland) appears with the unmarked form little as may be expected by the meanings of the forms (53). Eyes are another recurrent leitmotif in this text. Once again, when she encounters the eyes of an individual (in this case, a mouse) the unmarked form little is used (41); but when she discusses the size of eyes in a generic sense (the eyes of babies) she uses the marked form small (86). Throughout the text, size and dimensions serve as a major and recurrent theme. Wonderland is a unique world in miniature accidently discovered by Alice. The most frequently used member of the system is little (93X) which is consistently used to point out and focus on the diminutive dimensions of specific entities while the marked form small appears only 13 times, all in a generic sense. Moreover, most of the entities Alice encounters in Wonderland are different from their parallel counterparts in the real world to which Alice constantly tries to compare them, especially those entities that are not perceived in miniature. The marked form large is consistently used in these generic-like comparisons (rather than the individualizing unmarked form big which does not appear in the text) for the following items: a rabbit hole (26), It's like other rabbit holes of that size (29); locks (29); a rat-hole (30); letters (31); a telescope (35); a pool (36); a fan (39); a ring (46); birds (50); a mushroom (66); a caterpillar (66); eyes (79); a letter (79); a plate (81), a kitchen (82); a cauldron (82), a cat (83); a saucepan (84); a house (91); a table (93); an arm chair (93); a rose tree (105); a flowerpot (110); a crowd (115); eyes (126); a piece (147); and a canvas bag (149). These recurrent communicative strategies consistently exploiting the most suitable invariant meanings of the marked and unmarked forms in different contexts to communicate similar messages throughout the text give a sense of coherence and cohesion to the text: i.e., "from sign to text". From Text to Sign The "from sign to text" approach is only one way to analyze a text semiotically. It is also possible to view an entire discourse or a text itself as a unified whole composed of a signal and a meaning: i.e., a sign in its own right where the message of the text represents its meaning or signifie, and the nonrandom distribution of the language in the text represents its form, signal, or signifiant.

A Semantic Analysis: small/large vs. little/big 111 When it is possible to view a text as a sign in its own right, it then becomes possible to analyze that text and the nonrandom distribution of the signs within that text in a method called the "from text to sign" approach. This approach attempts to show that the consistent choice of a marked or an unmarked form may be directly determined and motivated by the message of the text itself as it functions as the meaning half of the text as a sign: i.e., the larger textual message will motivate the choice of the marked or the unmarked form throughout the text. The message of the text, functioning as the meaning of the "text as sign", involves a conflation of elements including the characters, the plot, the themes and leitmotifs that all function together within the larger system of the text. This approach involves the hierarchical descending order from text and context to system and sign. The "from text to sign" approach has explained the use of unusual marked forms in classic and contemporary texts (e.g., curiouser and curiomer in Alice's Adventures in Wonderland, darlingest in Social Disease (Rudnick 1986) and worse/worser/worst in the Jerusalem Post) and in other texts of literary and journalistic prose as well as poetry in English and Hebrew (Tobin 1990, 1994). In The House at Pooh Corner (Milne 1981), size is a central concept. The plot revolves around a boy creating his own miniature imaginary world as a means to come to terms with the real world. This text, a world in itself, presents the real world and The Hundred Acre Wood as alternative worlds. The words small/large and little/big appear 140 times in this 179 page text (almost once per page: a word per 1.27 pages). There are 46 instances of the use of the marked forms. These four words are used systematically throughout the text and define and describe most of the major characters and scenes within the text. Furthermore, these words are often capitalized for emphasis and are even used as parts of proper names, or even as the name of a minor character: (27)(a) Pooh = 'Bear of Very Little Brain' (b) Piglet = 'Very Small Animal' (c) Rabbit = 'Small and Sorry Rabbit' (28)

"What I came to say was: Have you seen Small anywhere about?" "I don't think so," said Pooh. And then after thinking a little more, he said: "Who's SmallT "One of my friends and relations," said Rabbit carelessly. (39)

The setting of the book, The Hundred Acre Wood, is described as a single integral unit belonging to a generic class with the marked form large, which is then divided into separate corners, individualized by the use of the unmarked form little (104). The animals of the Hundred Acre Wood are introduced to the character Tigger earlier in the text in chapter 2, where the marked and unmarked forms are chosen to describe him in the following way. First, tigers in general are classified by Piglet genetically as being small (26) and then Tigger asserts his individuality by referring to himself with the unmarked form big (32), which becomes the common way to refer to him except when classifying him genetically as a large (127) tiger using the marked form: One of the episodes in the book reflecting how the animals' world may be viewed as a microcosm of the real world is found in a game inadvertently invented by Pooh. Pooh, standing on a bridge over a river, drops two objects off the bridge to see which one will reach the other

178 Langue and Parole in Synchrony and Diachrony side of the river first. He first throws two fir-cones of equal size across the river, but he couldn't tell them apart. He then chooses fir-cones of different dimensions that are described by the unmarked forms little/big (94) which focus on the unique and individual perception of each as a competitor in his Newtonian-like experiment. When other animals join the game, Pooh substitutes the fir-cones with sticks, which are not only differentiated by size, but also by color, again using the unmarked forms. One of these individuated sticks appears to get stuck under the bridge, but it turns out not to be the 6/g-grey stick, but rather Eeyore, the (big-grey) donkey (97). hi order to free Eeyore from being stuck under the bridge, Pooh gets the idea of throwing a stone in the water to create waves to wash Eeyore to the shore. The specific stone he chooses is described by the unmarked form biggest (100) in the superlative. Eeyore disappears underwater, and Pooh is worried that he may have hurt his friend by choosing 'the wrong stone'. Eeyore, on the other hand, reappears unscathed and unconcerned, and then refers to the very same stone as part of a general class, a generic stone, using the marked form large (102), to allay Pooh's concern. Throughout this text, as in all the above examples, the characters seem to manage very well in their microcosmic world, thus reassuring the little boy, and the reader, that they, too, will be able to function successfully in the real world. The systematic interplay of the multidimensioned characters and entities in these and other episodes conveys this message via the nonrandom distribution of these marked and unmarked forms in a "from text to sign" approach. From Text to Sign Revisited The third text to be analyzed, The Little Prince (Saint Exupery 1943), is another classic of children's literature. In this text as well, size is a major thematic concept: the message of the text may be viewed as children see the world in a more realistic and less hypocritical way than adults. Certain communication strategies centering around the opposition of small/large versus little/big previously discussed can be traced in this text, hi this text when size is viewed generically (i.e. in continuous space, time, and existence) the forms marked for Semantic Integrality (small/large) are consistently used, and when specific entities are being focused upon the forms unmarked for Semantic Integrality (little/big) are consistently used. The little prince is first introduced in the text as a small person (generic) (10), then as a more specific little man/chap/fello\v (10. 83, 84) and as the little prince and the dear little prince throughout the text. Other major themes appearing in the text such as the sheep and the little prince's planet follow the same strategy: sheep are small (generically) and everything on his planet was small (generically), nothing was larger than a house there (12, 13, 16), but when a single sheep is focused on later hi the text we get one little sheep capable of eating a unique flower that only grows on the little prince's planet (28). The more specific information we get on the little prince the more the phrase is repeated. When the little prince's planet is specifically compared with him in size, the unmarked form big is used (18). The next major leitmotif is the catastrophe of the baobabs, which are big trees (specific) rather than little bushes or charming little sprigs (specific) or small sprouts (generic), and the problem of good versus bad seeds (specific) which make the baobabs too big for the small planet and the role of plants in general (19-22). The same strategy holds for he little prince's sad little life on his tiny planet (24-25). The little prince visits a series of planets whose sizes are viewed generically as small/large with the forms marked for Semantic Integrality when they are compared together (32, 38, 47, 50-54). The inhabitants or entities occupying these planets follow

A Semantic Analysis: small/large vs. little/big 179 the same discursive strategy: large animals (34); large mountain, large stones (53); very small place (57); (T)he earth is large (57); a little boy/other little boys (66); little bells are turned to tears (91). Emotions and states characterizing specific individuals and entities also appear with the unmarked forms: so little understood (22); a little ashamed (27), a little bored/A little sunshine (75); a little frightened (80); look a little as if I were dying (86); a little discouraged (87). In this text as well, all the major themes and the major character are described through these oppositions and communication strategies contributing to the message of the text.

SUMMARY AND CONCLUSIONS This paper presents a sign-oriented, semantic analysis of the forms small/large vs. little/big based on the theoretical tenets of invariance, markedness and distinctive feature theory. The methodology employed includes both micro-sentence and macro-discourse level analyses following semiotic approaches to text analysis, hi every case it was shown that the choice of one linguistic sign over another was not arbitrary but motivated by a subtle semantic distinction in the signifie of that sign, based on how the entities whose size and dimensions were being described were perceived either in continuous or discontinuous or discrete space, time, or existence. Despite the fact that this semantic distinction is, indeed, a subtle one, in the words of Dwight Bolinger (1977:17), found in the epigraph of this paper: True, these are subtle differences, but who says semantic distinctions have to be gross?

REFERENCES Andrews, Edna (1990). Markedness theory: The union of asymmetry andsemiosis in language. Duke University Press, Durham, NC. Andrews, Edna and Yishai Tobin (1996). Towards a calculus of meaning: Studies in markedness, distinctive features and deixis. John Benjamins, Amsterdam. Aphek, Edna and Yishai Tobin (1988). Word systems in modern Hebrew: Implications and applications. E.J. Brill, Leiden. Aphek, Edna and Yishai Tobin (1989). The semiotics of fortune-telling. John Benjamins, Amsterdam. Battistella, Edwin (1990). Markedness: The evaluative superstructure of language. SUNY Press, Albany. Battistella, Edwin (1996). The logic of markedness. Oxford University Press, Oxford. Bolinger, Dwight (1977). Meaning and form. Longman, London. Contini-Morava, Ellen (1989). Discourse pragmatics and semantic organization: The case of negation and tense-aspect with special reference to Swahili. Mouton de Gruyter, Berlin. Diver, William (1969). The system of "Relevance" in the Homeric verb. Acta Linguistica Hqfhiensia 12,45-68. Garcia, Erica (1975) The role of theory in linguistic analysis: The Spanish pronoun system. North Holland, Amsterdam.

180 Langue and Parole in Synchrony and Diachrony Hirtle, Walter (1982). Number and inner space: A study of grammatical number in English. Presses de 1'Universite Laval, Quebec. Huffman, Alan (1996). The categories of grammar: French lui andle. John Benjamins, Amsterdam. Kirsner, Robert (1979). The problem ofpresentative sentences in modern Dutch. North Holland, Amsterdam. Reid, Wallis (1991). Verb number in English: A functional explanation. Longman, London. Saussure, Ferdinand de (1913/1959). A course in general linguistics. Philosophical Library, New York. Tobin, Yishai (1988). The Prague school and its legacy. John Benjamins, Amsterdam. Tobin, Yishai (1989). From sign to text: A semiotic view of communication. John Benjamins, Amsterdam. Tobin, Yishai (1990). Semiotics and linguistics. Longman, London. Tobin, Yishai (1993). Aspect in the English verb: Process and result in language. Longman, London. Tobin, Yishai (1994). Invariance, markedness and distinctive feature analysis: A contrastive study of sign systems in English and Hebrew. John Benjamins, Amsterdam. Tobin, Yishai (1997). Phonology as human behavior: Theoretical implications and clinical applications. Duke University Press, Durham, NC. Waugh, Linda and Stephen Rudy (1991). New vistas in grammar: Invariance vs. variation. John Benjamins, Amsterdam.

CORPUS Alther, Lisa (1990). Bedrock. Penguin, London. Carroll, Lewis (1865/1970). Alice's adventures in Wonderland. Penguin, London. Heyerdahl, Thor (1965). The Kon-Tiki expedition. Longman, London. Milne, A. A. (1981). The House at Pooh Corner. Dell, New York. Rudnick, Paul (1986). Social disease. Ballantine, New York. Saint-Exupery, Antoine de. 1943. The Little Prince (translated from the French by Katherine Woods). Reynal & Hitchcock, New York. Sesame Street parents. April. 1994. Walker, Elizabeth. 1992. Hallmark. Headline Book Publishers, London.

14

COLOUR CATEGORIZATION AND NAMING IN FRENCH AND HUNGARIAN Isabel Forbes, SAILLS (St Andrews Institute of Language and Linguistic Studies), University of St Andrews, Scotland/UK, and Gabor Kiss, Institute for Linguistics of the Hungarian Academy of Sciences, Hungary

Abstract Ideally we should find isomorphy between colour category and colour name, at least as far as basic colours are concerned. In most languages this is the case, but in some languages we find one basic term to denote two categories or one category denoted by two terms. The latter is the case with French and Hungarian, and the problem is to identify the relationship between the two terms. Staying within the language system, we have to consider whether the relationship is one of hyponymy or synonymy. We then have to consider whether the two terms are contextual variants. The status of the two variables is not exactly the same in French as it is in Hungarian, but nevertheless there are underlying features which suggest that the colour vocabularies of both languages may be best described within the framework of a unified theory.

The basic colour vocabularies of French and Hungarian are similar in one important respect and the parallelism illustrates an evolution in the theory of colour semantics. Research on colour vocabularies can be roughly divided into two periods, before and after 1969. Structural linguists from Saussure on held the view that the signs of language are arbitrary and this view was applied to colour vocabularies as well as to other domains. Early linguistic textbooks of the forties, fifties and sixties very often contained Hjelmslev's depiction of the colour spectrum as being divided in an arbitrary way, each language making its own divisions. Along with this view went the notion that it is language that determines the way man looks at the world. Most people would now agree that this extreme form of the Sapir Whorf hypothesis is no longer tenable. In 1969 there was a Kuhnian revolution that overturned the reigning paradigm. Berlin and Kay published Basic Color Terms: their Universality and Evolution. (Berlin and Kay, 1969). The title tells it all. The suggestion is now that basic colour terms at least are not arbitrary and language specific but are universal. That theory has of course been challenged, but it is true to say that since 1969 most work on colour vocabularies has taken the theory as a starting point. The two hypotheses on which the theory rests and which are there to be tested are:

181

182 Langue and Parole in Synchrony and Diachrony (a) There is a set of eleven basic colour categories which are universal and from which each language draws and chooses to name using a basic colour term. (b) Not all languages have all eleven basic terms and as they evolve they do so in a certain fixed sequence. The original evolutionary sequence as it appeared in 1969 is shown in Figure I below: Figure I. The evolutionary sequence of basic colour terms. WHITE

GREEN RED

BLACK

BLUE YELLOW

BROWN

PURPLE PINK ORANGE GREY

The sequence is implicational: that is, if a language has only two basic colour terms they always denote the categories BLACK and WHITE; if a third term is added it is always the term for RED and so on. The original sequence has been revised, but the revisions are not important for our purpose here, since both French and Hungarian have fully developed colour vocabularies. Indeed, as we shall see, they are perhaps overdeveloped. One thing that should be pointed out about the sequence at this stage is that there is now considered to be a break between the first six terms and the rest; there are primary basic terms and secondary basic terms. Berlin and Kay do not give a unique definition of basic colour terms, but they suggest the following criteria: 1. 2. 3. 4.

A basic colour term is monolexemic. The meaning of a basic colour term is not included in the meaning of another term. A basic colour term is psychologically salient. A basic colour term is not restricted to a narrow class of objects.

Hardin and Maffi (1997) point out that these criteria have been questioned. Some researchers have fewer, retaining, for example, inclusiveness, generality of reference and salience (Crawford, 1982); inclusiveness and generality (MacLaury, 1982); salience (Bolton, 1978; Corbett and Davies, 1997). Although we have taken all criteria into account, in this paper particular attention will be paid to salience as measured by frequency of occurrence . French and Hungarian colour vocabularies are of particular interest because both languages appear to have eleven basic categories encoded by twelve basic terms. In 1969 Berlin and Kay recognised that some languages may have more than eleven basic terms and they cited Russian, which appears to have two basic terms for the category BLUE, sinij and goluboy (dark blue and light blue), and Hungarian, which has two terms for RED, piros and voros. They did not mention French. Forbes (1976) noted that French appeared to have two basic terms for BROWN , brun and marron. In subsequent works this was tested and confirmed (Forbes 1979, 1986, 1997a,b). The data for French discussed in this article were collected by Forbes, and the Hungarian data were collected jointly by Forbes and Kiss in Budapest in 1997. The main theoretical point to be tested both in French and Hungarian is whether there are indeed twelve basic colour terms in French and Hungarian and, if so, whether there are twelve categories or only eleven categories, with some people naming the same category using two different names. This is an important point for theories of cognitive linguistics in general and

Colour Categorization and Naming 183 theories of categorization in particular, and it has been taken up by other researchers. It has been applied to theories of colour semantics. MacLaury's vantage theory is of special interest (MacLaury, 1997) and he has published, along with two Hungarian colleagues, an article on Hungarian piros and voros which we had not read at the time of our research ( MacLaury et al, 1997). For both French and Hungarian the methodology was the same as that used by Berlin and Kay (1969) with, in addition, one other test. The first test was an elicitation test where informants were asked to write down the names of colours. The second test was a mapping test. Informants were asked to mark on an array of 329 colour chips the areas covered by the ten basic terms plus brun and marron in the case of French and piros and voros in the case of Hungarian. The third test was one in which informants were asked to write in two columns lists of things that could be called brun and marron and things that could be called piros and voros. This test enables us to test one of the criteria of basicness, namely use over a wide range of objects. It also helps to establish the kind of relationship that exists between the words in the pair. Are they synonyms, interchangeable in all contexts? Are they contextual variants, mutually exclusive? This attention to the use to which a term is put is not stressed in all research on colour vocabularies, at least not explicitly, but its importance has been mentioned by Forbes (1976) and more recently by Lucy (1997). In the first test, the elicitation test, we got hundreds of terms in spite of asking for 'principal colour terms'. Some individual informants gave longer lists than others. Forbes (1980) found that, for French, women usually gave longer lists than men and gave more non-basic and peripheral terms. The complete lists of elicited terms would be too long to include here and are not necessary for our immediate purpose. We were interested mainly in the most frequently mentioned terms because it has been observed by researchers that basic terms are used more frequently than non-basic terms and that furthermore there is a correlation between frequency and evolutionary order. The order of frequency in Hungarian corresponds largely to that found by Forbes (1976,1997b) for French and other researchers for other languages. There is a distinct drop in frequency between the eleven (or twelve) terms which name the eleven basic categories and the rest. The next three or four most frequent terms have been included for comparison. We can call them contenders for basicness. There is also a drop in frequency between the six primary basic terms and the next five or six secondary basic terms. These observations apply more clearly to the Hungarian data than to the French data, but it is not obvious why this should be so. Berlin and Kay (1969) refer to 'wildcards', that is terms that seem to turn up in the 'wrong' place in the evolutionary sequence. What is clear both from the French data and the Hungarian data is that some terms are mentioned more readily than others and those terms appear early in elicited lists. This is a universal finding and, as far as the first six categories are concerned at least, this may be determined by neurophysical mechanisms of colour perception and categorization. Figure II below shows the salience of colour terms in French and Hungarian as measured by frequency.

184 Langue and Parole in Synchrony and Diachrony Figure II.

Category

Salience of colour terms as measured by frequency.

French

No. of mentions

noir blanc RED rouge vert GREEN YELLOW jaune BLUE bleu

86 85 99 89 89 97

brun marron PURPLE violet PINK rose ORANGE orange GREY gris

8 41

BLACK WHITE

BROWN

beige mauve bordeaux

N=98

34 52 35 8 8 5

rank Hungarian

No. of mentions

rank

kek

87 83 82 81 71 75

1 2 3eq. 5 9 7

12eq.

barna

82

3eq.

9 8 11 7 10

lila rosaszin narancsarga szurke

78 67 64 77

8 10 11 6

bordo voro's citrom(sarga) turkis

46 41 28 14

12 13eq. 13eq. 15

5 6 1 3eq. 3eq. 2

12eq. 12eq.

15

fekete feher piros zold sarga

N=99

The French results are neater in a way, but we cannot always be neat in language. In the French data the six primary basic terms are much more frequent than the next six and both brun and marron are among the first twelve. Then there is a drop in frequency and so we could go on. Brun is down at number twelve. Forbes (1976, 1979, 1986, 1997b) has suggested that marron is taking the place of brun as a basic term to denote the category BROWN. Is the same thing happening with piros and voros? There are certain differences in the Hungarian data. There is a distinct drop in frequency between the first eleven terms and the rest, but voro's is among the rest. It is not even within the first twelve terms; it has been overtaken by bordo (dark red). However it is still a salient and important term to name the category RED, just as brun is still an important term to name the category BROWN. In the second test informants were shown a colour array containing 139 chips graded and numbered according to the Munsell system of colour notation. The advantage of using this type of colour stimulus is that it allows researchers to compare results across languages. Tests done by Berlin and Kay (1969) on 20 different languages from different language families show that informants readily mark the extent of the eleven basic categories and the best examples of these categories. There is more agreement about the best examples, the foci, than about the extent. We found this to be the case also for French and Hungarian. By and large the mappings are consistent, French with Hungarian and both with the languages tested by Berlin and Kay.

Colour Categorization and Naming 185 Informants know which chips are blue, which are green and so on. They also know which chips represent the best blue, the best green and so on. What we want to know is what French informants do when asked to map brun and marron, and what Hungarian informants do when asked to map piros and voros. There are two words certainly but do they denote two different colours? Forbes (1979) has shown that in French the two words cover the same area of the colour space. It is not the case that one word denotes light brown and the other dark brown or that one is a reddish brown, the other a yellow brown for example. Neither is it the case that one term includes the other. Both terms are used throughout the brown area. MacLaury (1997) calls this coextension. In Hungarian we also seem to have a case of coextension, although the picture is not quite the same in both languages. Piros is much more extensive than voros and its foci spread beyond the composite foci for RED in the 20 languages mapped in Berlin and Kay (1969). Voros is more restricted and we find its foci towards the dark area (the foot) of the chart and also at the PURPLE-RED end of the array. No examples of piros were found at the PURPLE-RED end. However, voros does denote RED, and indeed its foci fall more within the Berlin and Kay composite foci for RED than do the foci of piros. The third test was designed to establish if there are collocational differences between the terms in each pair. Forbes (1976, 1979) found that both brun and marron are used for a wide range of things. More recent research by Forbes, however, suggests that brun is becoming more and more restricted to hair and physical type. It is still, however, retained in certain fixed expressions and one would expect this to continue. In the Hungarian test, where informants were asked to list all the things that could be called piros and all the things that could be called voros, a wide range of objects appeared under each heading, 187 under piros and 183 under voros. There was considerable overlap, 87 things appearing in both lists. Several informants gave identical lists for both terms, often simply writing "the same". Some things were intrinsically red, for example ceresznye (cherry) and ver (blood); others could be any colour, for example kabat (coat) and toll (pen). As was found to be the case with French brun, there are certain things that can only be voros, for example things with socialist connotations such as oktober, komunismus, and zaslo when it is the Soviet flag. The Hungarian flag, on the other hand, is piros. Only voros is used to describe the colour of the fox (rokd). It may be that this older word is the word more commonly used for animal coats as is the older French term brun. In French ours brun (brown bear) is a fixed expression.

186 Langue and Parole in Synchrony and Diachrony Figure III The distribution of brun and matron in informants' lists. brun marron Category of object No. of mentions No. of mentions Hair 60 7 Eyes 20 45 Skin and other physical attributes 20 1 Persons (physical type) 12 0 Animals (including horses) 22 4 Trees and parts of trees 23 38 Clothes and material 23 41 Leather 14 22 Paint colour 11 18 Earth and Terra Cotta 9 4 (Wooden) furniture 8 15 Food 3 10 Tobacco 3 0 Miscellaneous 4 1 Fixed expressions with brun include: Tabac brun, biere brune, sauce brune, ours brun (Table reproduced from Forbes, 1979)

Figure IV The distribution ofpiros and voros in informants' lists. piros voros No.of mentions Category lexeme No. of mentions Hair Animals Blood Fruit

Miscellaneous

haj (hair) roka (fox) ver (blood) alma (apple) eper (strawberry) czeresznye (cherry) meggy (cherry) Mikulas (Santa Claus) csillag (star) zaszlo (flag) lampa (light)

40 18 39 1 0 3 8 0 16 13 0

4 0 21 49 14 19 3 13 1 5 14

Fixed expressions with voros include: Voros csillag (the red star); voros poszto (red rag (to a bull)); voros zaszlo (the red flag)

As can be seen from Figures III and IV above, some things are more frequently mentioned under one term than under the other. In French brun is the favoured term for hair and the seven mentions of marron to denote hair are highly unusual. It may be, and I am indebted to one of the conference participants for this observation, that marron is the term used to denote a colour

Colour Categorization and Naming 187 of hair dye or dyed hair. In French brun is also the favoured term, indeed the only term, used to denote physical type. It is also the favoured term used to describe skin colour. Fixed expressions with brun include 'tabac brun' (dark tobacco), 'biere brune' (brown beer), 'sauce brune' (brown sauce) and 'ours brun' (brown bear). In Hungarian, hair, as was the case in French, is much more commonly described using the older term. Fruit is more likely to be described using piros, with the exception of meggy, which is a dark red cherry. This latter collocation bears out an observation made earlier, that for many people voros is the term used to denote dark red. The cherry which is a lighter red, on the other hand, is more likely to be described as being piros. Fixed expressions with voros, apart from the expressions referring to socialism mentioned above, include 'voros poszto' (red rag (to a bull)). There are certain fixed expressions in Hungarian with piros and these are things of recent origin such as piros lap (the red card in football). The word 'MacDonalds' appeared only under piros and never under voros. Rather than posit two separate categories for French brown and Hungarian red it seems plausible to suggest that there is only one category labelled by two highly salient terms. Present theories of colour semantics do not rule this out (cf. MacLaury, 1997). The two pairs in French and Hungarian appear to work in the same way. The two terms are not exact synonyms. They are not in complementary distribution. They are not in a relation of inclusion. However they are not equal. The way MacLaury puts it is that one member of the pair is dominant and the other is recessive, depending on the viewpoint of the speaker. The choice of term is determined not only by perceptual factors but by social and cultural factors, and the meaning of the four colour terms under discussion here, and indeed all colour terms, includes their reference, their meaning in use, as well as their denotation.

References Berlin, B. and P. Kay (1969). Basic Color Terms: their Universality and Evolution. University of California Press. Reprinted 1991. Bolton, R.(1978). Black,white and red all over: the riddle of color term salience. Ethnology, 17(3), 287-381. Buky, Bela and Gabor Kiss (1997). The main statements of theories as determined by a semiotic method. Semiotics around the World: Synthesis in Diversity. Mouton de Gruyter, Berlin, New York. Buky, Bela and Gabor Kiss (1997). The use of the verbal prefixes in late old Hungarian and in the 20th century Hungarian. Paper given at the XVIth International Congress of Linguists, Paris. Forthcoming. Corbett, Greville G. and Ian R.L. Davies (1997). Establishing basic colour terms. Color Categories in Thought and Language, edited by C.L. Hardin and Luisa Maffi. Corbett Greville G. and G. Morgan (1988). Colour terms in Russian. Journal of Linguistics, 24, 31-64. Crawford T.D. (1982). Defining 'basic color term'. Anthropological Linguistics, 24(3), 338-343. Forbes, Isabel. (1976), Structural Semantics with Particular Reference to the Colour Vocabulary of Modern Standard French. PhD thesis, University of Edinburgh.

188 Langue and Parole in Synchrony and Diachrony Forbes, Isabel (1979). The terms 'brun' and marron' in modern standard French. Journal of Linguistics, 15, 295-305. Forbes, Isabel (1980). "Coq-de-roche" and "bleu layette": sex differentiation in the use of colour terms in French. Paper given at the Sociolinguistics Symposium, Walsall. Forbes, Isabel (1986). Variation and change in the basic colour vocabulary of French. Sigma, 10, 81-103. Publications de la C.E.L.A.M., Universite de Provence, Aix-enProvence and Montpellier. Forbes, Isabel (1997a). The notional domain of colour. In: La Notion (C.Riviere and M-L Groussier eds). Paris: Ophrys. Forbes, Isabel (1997b). The evolution of the basic colour vocabulary of French. Paper given at the XVIth Congress of Linguists, Paris. Hardin,C.L.and Luisa Maffi (eds) (1997). Color Categories in Thought and Language. Cambridge University Press, Cambridge. Hurvich, L.M. and D. Davidson (1957). An opponent process of colour vision. Psychological Review, 64, 384-404. Kiss, Gabor (1997). Kiserlet a Foldrajzi nevek etimologiai szotaranak szamitogepes feldolgozasara. Szavak - Nevek- Szotarak: Irasok Kiss Lajos 75. Szuletesnapjara. A Magyar Tudomanyos Akademia Nyelvtudomanyi Intezete, Budapest. Lucy, John .A. (1997). The linguistics of "color". Color Categories in Thought and Language, edited by C.L. Hardin and Luisa Maffi. MacLaury, Robert E. (1982). Prehistoric Mayan color categories. MS. MacLaury, Robert E. (1997). Color and Cognition in Mesoamerica. University of Texas Press, Austin. MacLaury, Robert E., Judit Almasi and Zoltan Kovecses (1997). Hungarianpiros and voros: Color from points of view. Semiotica, 114-1/2, 67-81. Morgan, G. (1993). Basic colour terms: comparative results for French and Russian. French language Studies, 3, 1-17. Riviere, Claude and Marie-Line Groussier, eds, (1997). La Notion. Ophrys, Paris.

15

ASSOCIATIVE AND SEMANTIC WORD FIELDS IN BlLINGUALS: THE CASE OF RUSSIAN-HEBREW BlLINGUALISM1

Larissa Naiditch, The Hebrew University of Jerusalem, Israel

Abstract The paper deals with a word association test on Russian-Hebrew bilinguals and is based on an experiment. The respondents (repatriates from Russia living in Israel, aged 12-22) were given stimulus words in Russian and requested to write down in the space of one minute all words which came to mind. The results were compared with those from two associative dictionaries of Russian. Shifts in associative and correspondingly in semantic fields of lexemes are explained by linguistic and extralinguistic factors concerning modern Russian in the metropolis (i.e. in Russia) and abroad: changes in Russian itself, changes in life conditions of the immigrants, and the influence of Hebrew. Interference on the semantic level was also considered.

BACKGROUND The word association test (WAT) is an important means for semantic, psycholinguistic and sociolinguistic studies. The investigation of associations has a long tradition (Cramer, 1968, 3). Early experimental research of associations was conducted by F.Galton in 1879-1880 and by W.Wundt in 1883; the associative experiments of the latter influenced many famous psychologists and psychiatrists (Szalay and Deese, 1978, 10). As early as 1910 the first dictionary of associative norms of English was published. Nowadays there are various dictionaries of this type. The associative norms of English, French, German, Polish have been well described. As for East Slavonic languages relevant to this research, there are two chief dictionaries of Russian: Leont'ev, 1977 (hereafter termed L), and Karaulov et al., 1994, 1996 (hereafter termed K.). While the material for the former was collected during the period 1969-1972, the data for the latter were gathered in 1986-1991. There is also an Ukrainian associative dictionary (Butenko, 1979) and a Byelorussian one (Citova, 1981). For the linguist, the WAT primarily enables the drawing of conclusions regarding the semantic 1 The research was supported by the Ministry of Science and Arts in Israel and by the Franz Rosenzweig Research Center for German-Jewish Literature and Cultural History (the Hebrew University of Jerusalem).

189

190 Langue and Parole in Synchrony and Diachrony structure of lexemes, the organization of verbal memory, and the internal structuring of the lexicon. Thus, the WAT can provide information on word-webs present in language and mind not always, however, in a direct and straightforward manner, since a mixture of heterogeneous data is obtained, and responses cannot be considered as a direct reflection of the mental representation pattern of a word. Let us take, e.g., the Russian word voda 'water' as in L. The most frequent reaction is xolodnaja 'cold', which may be defined as a syntagmatic or collocative reaction (because of cooccurence of certain words, a word comes to mind that is often used in the same syntagma). The reaction zemlja 'earth' is a coordinative response. The response vescestvo 'matter, substance' may be seen as superordination. Other kinds of reactions are possible, e.g. synonymy (the response FLO can, from a certain point of view, be regarded as a synonymous reaction or as an explanation, definition, etc.). J.Aitchison (1987, 74-75) explains the same notions using the example of the English word "butterfly". All these responses, while differing from the point of view of logic and based on different logical principles, contribute to our knowledge of the semantic structure of lexemes and of the organization of the lexicon. However, the set of responses also encompasses another kind of information, not corresponding to logical principles at all. It may include words indirectly connected with the stimulus word, sometimes an understood linking element, e.g., voda tverdost' ('water' - 'hardness'), probably connected through the word 'earth', or voda - vesc 'water' - 'thing'. Sometimes phonetic and morphological similarities also play a certain role, e.g., voda - vodka. A significant number of responses consist of parts of phrasemes and idioms (or the whole idiom), or, for instance, of hints on titles of books, films, etc. (Dobrovol'skij and Karaulov, 1992). These cliches also contribute to our knowledge of associative domains and of the cultural background of the informants. However, they also provide information of other kinds. Thus, the WAT gives us different kinds of data, and the task of the researcher is to find suitable means for their processing. As for semantic research, the WAT can indirectly demonstrate the proximity of lexemes to each other and contribute to componential analysis - semantic paradigmatics. As mentioned above, the WAT also gives collocations, thus contributing to the investigation of linguistic syntagmatics: e.g., 'red' - 'flag', 'white' - 'snow'. Besides this, another type of data provided by the WAT can also be of interest to linguists - the information reflecting other levels of the informant's knowledge, connected with cognitive and pragmatic data. The cognitive aspects of associations reflect the informant's naive world picture as a whole, as well as various extralinguistic data. The latter are often very individual and reflect individual experience, interests, knowledge, etc. These individual features also have to be taken into account by linguists; they show the flexibility of what we should include in the notion of meaning. Besides individual experience, these aspects of the WAT also demonstrate grouping within a given language community (according to age, sex, education, place of residence, interests, etc.). The concept of personality, important for the WAT, is linked with that of group mentality, as well as national and cultural peculiarities. In this connection the works of several anthropologists can be mentioned (see: Szalay and Deese, 1978, 18). As for ethnolinguists, they were to a large extent influenced by the Sapir-Whorf hypothesis, based on the idea of dependence of thought on language, as in the works by Whorf. Cultural aspects of associations are investigated in several works. An example given by Szalay and Deese consists of a

Associative and Semantic Word Fields in Bilinguals 191 comparison of the associations of the word "freedom" for Afro-Americans and white persons; Afro-Americans giving the words: "free, slave, blacks, love, peace, good, jail, whites, speech", and white Americans giving - "free, speech, life, US, love, liberty, religion, flag" (Szalay and Deese, 1978, 20). This example is based on experimental data and "serves to illustrate that free associations inform us about subjective culture" (ibid.); we would add "and group culture". Cultural aspects of the WAT become clear, when at least two national and cultural stereotypes are compared. Thus, A.A.Zalevskaja pointed out that the most frequent reactions to the stimulus "bread" are "butter, food, eat" in American groups of informants, "eat, butter" etc. in German groups, "wine, white" in French, "white, salt, butter" in Russian, "butter, water, flour" in Kazakh, "white, wheat, butter" in Kirgiz, and "hungry, tea, water" in Uzbek (Zalevskaja, 1977; see also: Ruke-Dravina, 1971; Dmitrjuk, 1985). What do these differences demonstrate? Firstly, differences in every-day life and images, customs and technology, all reflected in associations and probably contributing to the semantic structure of the corresponding words. Another example given in the same article by Zalevskaja shows that national and cultural aspects are connected with more profound principles of associativity. The stimulus word 'eagle' gave 55% superordinative answers (such as "bird") in American groups, 20,5% in German groups, 16% in Polish groups, 14% in Kazakh, 11% in Kirgiz, 10% in Russian, 7% in Uzbek. These data reflect, of course, as in the former case, nature, traditions and customs of the corresponding countries and peoples. On the other hand, they show even more clearly than in the former case the dependence of general directions of associations on ethnicity and, we could say, on language. In the recent article by N.V.Ufimceva (1995) the word associations in Russian, Lithuanian and Georgian groups of informants were used to reveal the ethnocultural stereotypes characterizing the human body. The results of such kind of study provide important data for semantic research as well. The considerations mentioned above primarily concern monolingual respondents. In the bi- and multilingual WAT additional problems arise, connected with cognitive and cultural aspects of the associations. The chief goal of many investigations is to understand how words are stored in the memory of bilinguals: do they lie in a single, shared conceptual store (all information stored centrally), or does each language have a separate conceptual representation (the common code or the separate code hypotheses) (Paradis, 1986, 1987; Macnamara, 1986)? One of the first experimental studies devoted to this question is the famous article by Kolers (1968). Since then a vast literature has accumulated on this issue, yielding rather contradictory data, hi a recent study, A.de Groot (1993) arrived at the conclusion that the memory of a bilingual may contain a mixture of different representational forms and that the storage of the data may depend on the word type. Whereas concrete words are quite often stored in a compound fashion, abstract words are more likely to be stored in coordinate (separate) form. It is supposed that there are two levels of representation and two corresponding levels of memory: lexical and conceptual. J. Kroll (1993: 74) demonstrates "that fluent bilinguals can sometimes use conceptual representations that are shared across their two languages". However, "shared conceptual-level memory representations are not the only form of interlanguage connection that the bilingual has available. Whether cross-language interactions are conceptually or lexically mediated appears to be determined by the level of proficiency in second language, by the conceptual demands of situation [...], and by the type of material" (ibid.). To assess the degree of overlap between the bilingual speaker's two lexical systems, different methods are used, the WAT being one of them.

192 Langue and Parole in Synchrony and Diachrony The other possible goal of the WAT in bilinguals consists of comparisons of its results with the associative norms of monolinguals. Our investigation deals mainly with this aspect of the problem. We tried to find changes in the semantic structures of lexemes caused by extralinguistic factors (repatriation to Israel) and by linguistic factors (influence of the second language - a kind of interference). The former could lead to a well known type of semantic change, whereby an object changes, while its nomination remains the same. The shifts in word associations in bilinguals, in comparison with those in monolinguals, can be revealed by comparing the responses in our tests with those in dictionaries and by comparing the responses of different groups of informants. Thus the object of our study is primarily the Russian language in Israel. In speaking about Russian-Hebrew bilingualism in Israel, the role of each of these languages should be understood. Special attention should be paid to the cultural connotations of each language spoken by the bilingual individual. Questions regarding specific features of Russian abroad and Russian as a mother tongue of a minority arise in this connection. This problem has recently drawn the attention of several scholars (Moskovich, 1978, 1990; Orel, 1994). The situation can be compared to the use of regional variants of standard languages, also called pluricentric languages (e.g. German in Germany, Switzerland and Austria; English in England, the USA, Canada, Australia, New Zealand and South Africa, etc.) (Clyne, 1992; Wiesinger, 1997). Usually these terms are applied to cases of state languages, i.e. of several varieties of the language. From this point of view Russian is a monocentric language. However, Russian in Israel, which to a large extent has become a language of culture and everyday communication for a minority community, has developed several peculiarities making it close to a variant of a language. In investigating languages in contact, linguists pay attention to different kinds of interference in phonology, grammar and lexicon, including such phenomena as code switching and mixing. In our case single lexemes usually fully coincided with the corresponding words in the standard language of the metropolis (i.e. in Russian in Russia) from the point of view of their phonological and grammatical structure; however, deviations from usual associations could be interpreted as shifts in the semantic structure of lexemes. Thus we have here an additional kind of interference, manifesting itself in the semantic structures of lexemes.

DESCRIPTION OF THE EXPERIMENT The subjects of my experiment were predominantly schoolchildren (plus a small group of students) living in Israel, all born in Russia and having immigrated to Israel. Their ages ranged from 12 to 22, but most were 13-17 years old. There were altogether 200 subjects. All use Hebrew at school (or in the university); most speak Russian at home with their parents and grandparents. Their proficiency in Hebrew varied. The subjects were given several general questions: they had to write down their age, date of immigration, language preferences in speaking with friends, in reading books and watching TV. Later the respondents were given the stimulus words (in oral form, word by word) in Russian and requested to write down in the space of one minute all words which came to mind (free, continued association test). In principle the subjects had to respond in Russian, and in most cases they did so. But several persons refused to answer in Russian, in which case they were permitted to write down their answers in Hebrew.

Associative and Semantic Word Fields in Bilinguals 193 Following the processing of the questionnaires, four groups of respondents were singled out according to their integration into Israel society and their knowledge of Hebrew, the last of these groups being characterized by a high level of integration and by a preference for Hebrew in most domains (Russian was preserved as a home language, predominantly in conversation with parents and grandparents). The association tests show substantial differences between the informants of groups 1 - 3 and those of the last (most integrated) group, in which the informants often could not answer in Hebrew to the stimulus word in Russian, their associations were scarce, and they also had difficulties in writing Russian words. This, of course, reflects a trend towards the deletion of the mother tongue and the shift from Russian to Hebrew as the first language. Differences in associations between these groups were also assessed, but results are not presented here, because the material is statistically irrelevant. The list of stimuli included 11 words, comprising 8 nouns, 2 adjectives ('red' and 'white') and one verb form. The choice of this material was based on our hypotheses on linguistic and ethnocultural differences of associations for these words and on their representation in dictionaries. The results of the test were computer-processed, distinguishing between two kinds of data: the first reaction (hereafter T 7), and the sum reaction (hereafter T 2). The former results were compared with the data in dictionaries L. and K. , since both dictionaries and this test were based on the first reaction response, whereas T 2 provided important additional information.

RESULTS Let us consider several results of our WAT. The first word we will discuss is the stimulus xleb 'bread'. According to the data of both associative dictionaries of Russian (L. and K.) the foremost reaction to this word is the word nasuscnyj meaning 'urgent' or 'vital'. The association xleb - nasuscnyj is based on a phraseological unit going back to the Paternoster and meaning 'daily bread'. In our experiment this answer was not present, which can be explained not by a poor knowledge of Russian, but rather by the fact that the expression is gradually falling out of usage in the colloquial language of the young generation. Instead of this, the most frequent reaction to xleb 'bread' in our WAT, both in T 1 and T 2, was eda 'food'. In both dictionaries this reaction has a low frequency. Regarding traces of phraseology in our answers, we should point out the high rank of the expression vsemu golova 'the head of all', the answer golova was also sometimes encountered. The reaction is based on the proverb Xleb - vsemu golova, literally 'Bread is the head of all', stating the importance of bread as food and as the result of human work. As this reaction is not present in both dictionaries, we could suppose that the proverb has recently become more common. In reality, however, this reaction reflects the traces of Soviet propaganda rather than trends in the usage of phraseology. For several years proverbs about the importance of bread were used in propaganda, especially in school. The association xleb - sol' 'bread' - 'salt' has a very high rank in L, is in the tenth place in the K. and in the seventh and the twelfth place in our T1 and T2 respectively. Thus, once more our experiments reflect trends in the usage of phraseology in Modern Russian. The considered association is based on idioms, e.g., vstrecat' xlebom s sol'ju 'to meet with bread and salt, to demonstrate hospitality', which in turn corresponds to the popular tradition of giving bread and

194 Langue and Parole in Synchrony and Diachrony salt to guests as a gesture of hospitality, now rarely used (preserved only in official practice imitating traditional folk ceremony). It is worth mentioning that in Hebrew traditions the symbolism of bread and salt also exists, but has a different function connected with the ritual of the Sabbath meal. As for syntagmatic reactions, the most frequent were belyj 'white', cernyj 'black', vkusnyj 'tasty'. Interestingly, the two former have the same frequency in our experiments, whereas in L. and in K. the reaction 'black' is more frequent than 'white', a phenomenon which can be explained by specifics in food traditions - in Russia black bread is widely used, whereas in Israel white bread is more common. The association cerstvyj 'stale' occupying the second place in K. (not by chance since the tests for this dictionary were conducted during the food deficit) is absent in our test - once more reflecting the differences in lifestyle and traditions. A more detailed analysis shows two different directions of thought reflected in the associations, connected either with Russian or with Israeli life. The first includes traces of Soviet life, Soviet cliches, as well as knowledge of the recent history of Russia known to the informants from school, from books and from the recollections of elder members of the family, for example, the Second World War, and especially the famine in Leningrad during its blockade by German troops (e.g. xleb - vojna, xleb - Leningrad 'bread war', 'bread - Leningrad'). The Hebrew (or Israeli) connotations can be observed in such associative reactions to the word xleb 'bread' as xleb s xumusom, pita, maca, Tora, pasxa, subbota 'bread with humus, pitta, unleavened bread, Torah, Passover, Saturday'. Another interesting example showing shifts in association norms is the word ljubov' 'love'. In L. the foremost responses to this word are nenavist', druzba 'hatred, friendship' (paradigmatic reactions) and bol 'Sdjd 'big' (a syntagmatic reaction); in K. the first associations are parts of phrasemes - do groba, s pervogo vzgljada 'till death, at the first sight', followed by the noun cuvstvo 'feeling'. In both our tests the first place is occupied by the word seks 'sex', sometimes written in English, the second one by the word scast'e 'happiness'. Other common reactions were: cuvstvo, serdce, bolezn', romantika 'feeling, heart, disease, romance' (see Appendix). These striking differences between our results and the data in the dictionaries are of course to be explained by shifts in mentality, probably observed in Russian society as well and accelerated under the conditions of another country. The results indirectly confirm the above mentioned hypothesis of A. de Groot regarding the separate storage of abstract notions demonstrating significant differences between the languages of a bilingual person. In this case the comparison of our material with the association norms in other languages is significant. The most frequent associative responses to the word "love" in American groups of informants were "sex, peace, marriage, happiness, life, family, girl, mutual" (Szalay and Deese, 1978: 124). These results differed from the associations of Korean informants: 'mother and offspring, offspring, marriage, happiness, woman, friend, brother, parent and offspring'. "The greater Korean emphasis upon the family is obvious" (ibid.). Several of our results coincide with the American norms: 'sex, happiness, girl, family', whereas the data of the dictionaries L. and K. were quite different, since the word 'love' was understood in its most abstract sense - in its paradigmatic relations with the words 'hatred' and 'friendship'. Finally, krasnyj 'red' is an example of an adjective as a stimulus word. Adjectives often evoke syntagmatic reaction, the paradigmatic ones not being excluded. The associates to this word are usually connected with the following domains of meaning: Soviet, nature, parts of the body,

Associative and Semantic Word Fields in Bilinguals 195 and various objects' (e.g. "car", "pencil", etc.) (Alimpieva, 1974). As for the former, in K. the reaction flag 'flag' occupies the first place; in our tests the same word is not in the first, but still in the foremost places (see Appendix). A difference between the data in K. and our results that is worth noting is the presence and high frequency of the word krov' 'blood' in our responses. The influence of Hebrew is evident: the Hebrew word adorn 'red' evokes the association dam 'blood' not only because of semantic, but also because of phonetic closeness. Several WAT's in Hebrew, conducted partially with the same informants, partially with Hebrew speaking monolinguals, showed the high frequency of the association adorn - dam. The association krasnyj - zeml'a 'red' - 'earth' encountered in one of the questionnaires (in the most integrated group of the informants) also reflects the influence of Hebrew where these words are phonetically close: adorn - adama. The emotive or qualitative connotations of the adjective krasnyj 'red' also must be taken into account. It is known that adjectives tend to develop additional qualitative semantics. This aspect of meaning is reflected in our test by several (sometimes single) reactions, most of them adjectives and nouns, e.g., krasivyj, krasavica, sil 'nyj, radostnyj, jarkij, aktivnyj, vyzyvajuscij, intensivnyj, zlost', I 'ubov', besenyj, krik, sila, strast', nasilie' 'beautiful, beauty (fern.), strong, joyful, bright, active, provocative, intensive, angry, love, mad, shout, strength, passion, violence'. As seen from these examples, the adjective evokes two kinds of emotive associations, positive connotations being connected with beauty and negative ones connected with violence. One of the informants wrote instead of a word a more detailed answer: cvet, oboznacajuscij hah krasotu (roza), tak i uzas 'the color designating both beauty (rose) and horror'. The common sememe of both these aspects is probably 'high intensity'. The symbolics of the color red should be remembered in this connection. In Russian krasnyj is etymologically connected with krasivyj 'beautiful' and has the meaning of positive evaluation in several expressions: compare the epitethon ornans in krasna devica 'beautiful maid' in folklore, krasnyj ugol literally 'red corner' (place in the room where the icons were hung), also Krasnaja ploscad' Red Square (the central square in Moscow). The symbolism goes back to Byzantine traditions and is reflected in Russian painting, from icons to the avant-garde (e.g,.the famous "Krasnyj kvadrat" by Kazimir Malevich). It was adopted and reinterpreted in Soviet symbolism2. The "negative" sides of the associations are almost absent in K.; thus it can be supposed that in our T 2 they were provoked by the associate krov' 'blood' and caused by associative chains (e.g., 'red - blood war - horror'). The extralinguistic cause of these "negative associations" is possibly the danger of everyday life in Israel; thus such notions as 'violence, wound, war, murder' are very near the surface of public consciousness. The shifts in the associative field considered above are caused by both linguistic and extralinguistic factors, sometimes difficult to differentiate from each other. Several cases of linguistically caused peculiarities of the WAT in bilingual informants in comparison with monolinguals seem to confirm the common representation hypothesis (compare several reactions to the word 'red' showing the influence of the corresponding Hebrew word). We tried to find differences caused by the peculiarities of Russian in Israel, but not all of these shifts can be attributed to this cause. Several of them are due to the changes in Russian itself; they were probably only accelerated or enhanced in Israel. Because the associative field reflects in some 2

The importance of the symbolism of the colour red in Russian tradition can be demonstrated by the fact that it became the subject of the interesting and stimulating exhibition "Rot in der russischen Kunst" in Vienna (September - November 1998).

196 Langue and Parole in Synchrony and Diachrony direct or indirect way the semantic field of a word, the shift in the semantic structure of the corresponding lexemes can be assumed. Thus, what we have observed is either a shift or interference on the level of semantic structure. In cases where the changes or differences of this structure are striking (cf. the associative responses to the word 'love' considered above), it could even be asked whether we are on the way to a kind of homonym. In trying to solve this problem, not only semantic but also sociolinguistic factors should be taken into account. The associative and semantic structure of lexemes reflects the experience of a given group of speakers using a language. Thus, the language of a new generation of speakers or a community living in another country shares profound peculiarities that can be demonstrated by help of linguistic experiments. Because many languages today are pluricentric, the groups of speakers living outside the metropolis and having a slightly different experience (or different consciousness, mentality, or convention on which the semantics of corresponding words is based) also deserve the attention of linguists.

REFERENCES Aitchison, J. (1987). Words in the Mind. An Introduction to the Mental Lexicon. Basil Blackwell. Oxford. Alimpieva R.V. (1974). Realizacija komponentov semanticeskoj struktury slova "krasnyj" v sisteme obrazno-poeticeskoj reci. In: Voprosy semantiki. Leningrad, 107-128. [Realization of the components of the semantic structure of the word "krasnyj" 'red' in the system of the figurative poetic language]. Butenko N.P. (1979). Slovnik asociativnix norm ukrains'koi movi. Visca skola, L'vov. [Dictionary of Associative Norms of the Ukrainian Language]. Citova A.I. (1981). Asacyjatyuny slounik belaruskaj movy. Vydavectva BDU imja U.I.Lenina, Minsk. [Associative Dictionary of the Byelorussian Language]. Clyne M. (1992). Pluricentric Languages - Introduction. In: Pluricentric Languages. M.Clyne ed. Mouton de Gruyter, Berlin, New York. Contributions to the Sociology of Language. 62,1-9. Cramer, Ph. (1968). Word Association. Academic Press, New York and London, de Groot, A.M.B. (1993).Word-Type Effects in Bilingual Processing Tasks. In: The Bilingual Lexicon. R. Schreuder and B.Weltens eds., Benjamins, Amsterdam/Philadelphia, 53-81. Dmitrjuk N.V. (1985). Nacional'no-kul'turnaja specifika verbal 'nyx associacij. Avtoreferat kandidatskoj dissertacii. Moskva. [National-cultural specifics of word associations]. Dobrovol'skij D.O. and Karaulov Ju.N. (1992). Frazeologija v associativnom slovare. Izvestija RAN. Serija literatury ijazyka, 51, 6,4-13. [Phraseology in an Associative Dictionary]. Karaulov Ju.N. et al. (1994). Russkij associativnyj slovar'. Kniga 1. Prjamoj slovar': ot stimula kreakcii. Moskva. [Russian Associative Dictionary]. Karaulov Ju.N. et al. (1996). Russkij associativnyj slovar'. Kniga 3. Prjamoj slovar': ot stimula kreakcii. Moskva. [Russian Associative Dictionary]. Kolers P. A. (1968). Bilingualism and information processing. Scientific American, March, 7885. Kroll, J. F. (1993). Accessing Conceptual Representations for Words in a Second Language. In: The Bilingual Lexicon. R. Schreuder and B.Weltens eds. Amsterdam/Philadelphia, 5381.

Associative and Semantic Word Fields in Bilinguals 197 Leont'ev, A. A. ed. (1977). Slovar' associativnyx norm russkago jazyka. Moskva. Izdatel'stvo Moskovskogo Universiteta. [Dictionary of Associative Norms of Russian]. Macnamara, J. (1986). Bilingualismus und Denken. In: Grundlagen der Mehrsprachigkeitsforschung. Zeitschrift fur Dialektologie und Linguistik, Beihefte, 52, 26-39. Moskovich, W (1978). Interference of Hebrew and Russian in Israel. In: Slavica Hierosolimitana, 2, 215-234. Moskovich, W (1990). Russko-ivritskie jazykovye kontakty. In: Ivrit - jazyk vozrozdennyj. Biblioteka-Alija. [Jerusalem] 157-175. [Russian-Hebrew Linguistic Contacts]. Orel, W (1994). Russkij jazyk v Izraile. In: Slavjanovedenie, 4, 35-43. [The Russian Language in Israel]. Paradis M. (1986). Sprache und Denken bei Zweisprachigen. Grundlagen der Mehrsprachigkeitsforschung. Zeitschrift fur Dialektologie und Linguistik, Beihefte, 52, 40-50. Paradis M. (1987). The Assessment of Bilingual Aphasia. Chap.l. Neurolinguistic Perspectives on Bilingualism. London, 1-17. Ruke-Dravina V. (1971). Word associations in monolingual and multilingual individuals. Linguistics, 74, 66-84. Szalay L. B. and Deese J. (1978). Subjective Meaning and Culture: An Assessment Through Word Associations. Lawrence Erlbaum Associates, Hillsdale, New Jersey. Ufimceva N.V. (1995). Etniceskie i kul'turnye stereotipy: kross-kul'turnoe issledovanie. In: Izvestija RAN, Serija literatury i jazyka, 54, 3, 55-62. [Ethnic and cultural stereotypes: a cross-cultural study]. Wiesinger P. (1997). Sprachliche Varietaten - Gestern und Heute. In: Varietdten des Deutschen. Stickel G. Hrsg., de Gruyter, Berlin. New York, 9-45. Zalevskaja A.A. (1977). Mezkul'turnyj aspekt problemy associacij. In.: Slovar' associativnyx norm russkogo jazyka. A.A.Leont'ev ed. Izdatel'stvo Moskovskogo Universiteta, Moskva. [Intercultural Aspect of the Problem of Associations].

APPENDIX i: MOST FREQUENT ASSOCIATIVE RESPONSES TO THE \VORDXLEB 'BREAD' From Leont'ev ed. 1977 (L.) HacymHbiH 123, cojib 119, qepHbiii 49, 6ejibiH 38, CBCHCHH 19, HMJI cymecTBHTejibHoe 17, BKycHO 16, ctecTb, nepcTBbiH 15, 6ynKa 14, nmeHHija, p^canoH, po>Kb 12, MflrKHH, none 8, e#a 7, nniqa, poabi 6, >KH3Hb, yicpaHHCKHH 5, ropflHHH, KyiiiaTb, Macjio, ne^b, CTOJI 4, 6yjiOMHaa, ropbKHH, KynHTb, Mara3HH, MOJIOKO, HO>K, xanBa 3. Literal Translation: vital 123, salt 119, black 49, white 38, fresh 19, noun 17, tasty 16, eat up, stale 15, bun (white bread) 14, wheat, rye [noun], rye [adj.] 12, soft, field 8, meal (food) 7, food, rose 6, life, Ukrainian 5, hot, eat, butter, bake, table 4, bakery (shop), bitter, buy, shop, milk, knife, khalva 3.

198 Langue and Parole in Synchrony and Diachrony From Karaulov et al., 1994 (K.). HacymHbiH 58, nepcxBbiH 37, nepHbiH 30, CBOKHH 28, GejibiH 23, MflncnH, p^canoH 22, BKycHbiH 21, 6yjiKa, cojib 14, sceMy rojiOBa 12, xenjibifi 10, ecTb 9, ropaHHH, Macjio 7, GorarcTBO, BKycno, ronoBa, flymHCTbiH, >KH3Hb, nmija, nojie 6, KapasaH, MOJIOKO 5, 6aTOH, Gyxanna, eâ, MarasHH, nmeHHMHbiH 4, sanax, H cojib, HMH cymecrBHTejibHoe, KpyrjibiK, KycoK 3. Literal Translation: vital 58, stale 37, black 30, fresh 28, white 23, soft, rye [adj.] 22, tasty 21, bun (white bread), salt 14, head of all 12, warm 10, eat 9, hot, butter 7, wealth, tasty [adv.], head, delicious (smell), life, meal, field 6, round loaf, milk 5, long loaf, loaf, food, shop, wheat (adj.) 4, smell, and salt, noun, round, slice 3.

Tl eâ 40, BceMy ronoBa 10, qepHbifi, 6e;ibiH 9, BKycHbiH 8, nmeHHija, cojib 6, 6yjiKa, MyKa, Kymaxb 5, 6oraTCTBO 4, rojiosa, KpouiKH 3. Literal Translation: food 40, head of all 10, black, white 9, tasty 8, wheat, salt 6, bun (white bread), flour, eat 5, wealth 4, head, crumbs 3. T2 eâ 50, 6ejibiH, nepHbiH 44, BKycHbiH 29, nuieHHua, MarKHH 20, CBOKHH 17, BKycno, MyKa 14, p>KaHOH 11, coiib, BceMy rojiosa 10, pom*, MarasHH, KymaTb, xenjibiH, nojie, »(H3Hb, nHina, TpyA 9, nenb, MepcxBbiH 8. Literal Translation: food 50, white, black 44, tasty 29, wheat, soft 20, fresh 17, tasty [adv.], flour 14, rye [adj.] 11, salt, head of all 10, rye [n.], shop, eat, warm, field, life, meal (food), job 9, bake, stale 8.

APPENDIX ii: MOST FREQUENT ASSOCIATIVE RESPONSES TO THE WORD LJUBOV 'LOVE' From Leont'ev ed. 1977 (L.) HCHaBHCTb 15, flpy»6a 13, Gojibiuaa, Bepnafl 12, ,0,0 rpo6a, >KH3Hb 6, MO«, CMepxb, HHCTaa, HyBCTBO 4, BsaHMHaa, ô KOHqa, 3Jiaa, KOJibuo, cepAue, cnacxbe 3. Literal Translation: hatred 15, friendship 13, great, true 12, till death, life 6, my, death, pure, feeling 4, reciprocal, until the end, evil, ring, heart, happiness 3. From Karaulov et al., 1994 (#.)• c nepBoro Bsrjiflfla 9, #o rpoGa 8, nyBCTBO 5, sna, cnacxbe 3. Literal Translation: at the first sight 9, till death 8, feeling 5, evil, happiness 3.

Tl CCKC 15+4 [nocjieflHee HanHcano no-anrjiMHCKH], cnacTbe 12, HVBCTBO, cep/me 11, 6one3Hb 5, poMaHTHKa, posa, MaMa, ôêJIyH, aexH 4, (fwnbM, Apy^Sa, êByiiiKa, ceMba, paÔCTb, MaJIbMHK 3.

Associative and Semantic Word Fields in Bilinguals 199 Literal Translation: sex 15 + 4 [the latter written in English], happiness 12, feeling, heart 11, disease 5, romance, rose, mum, kiss, children 4, movie, friendship, girl, family, joy, young man 3.

T2 CCKC 22 + 5 [nocjie^Hee HarmcaHO no-aHrnnHCKH], cnacxbe 22, nyBCXBO 20, flexH 19, CCMBH 18, paflocxb 16, noqejivH 14, cepûe 12, Ljaexbi 11, >KH3Hb, cjiesbi 10, MaMa, 6ojie3Hb 9, pCBHOCTb 8.

Literal Translation: sex 22 + 5 [the latter written in English], happiness 22, feeling 20, children 19, family 18, joy 16, kiss 14, heart 12, flowers 11, life, tears 10, mum, disease 9, jealousy 8.

APPENDIX III: MOST FREQUENT ASSOCIATIVE RESPONSES TO THE WORD KRASNYJ 'RED1 From Karaulov et aL, 1996 (K.\ 4)jiar 15, CBCT 11, UBCX 8, rancxyK 7, noMH/top 4, KBa^pax, HOC, nepeu,, map 3. Literal Translation: flag 15, light 11, color 8, tie 7, tomato 4, square [geometry], nose, pepper, balloon 3.

Tl UBCX 43, (Jwiar 23, KpoBb 18 + 4 [nocjie^Hee HanncaHO Ha HBpnxe], sejienbiH 8, noMH,o,op 6, CB6TO(|)Op 5, 3HaMH, CHHHH, CBCT 4, HOC, flpKHH, 6ejIbIH 3.

Literal Translation: color 43, flag 23, blood 18+4 [the latter written in Hebrew], green 8, tomato 6, traffic lights 5, banner, blue, light 4, nose, bright, white 3.

T2 KpOBb 76 + 4 [nocjieAHee HanHcano Ha HBPHTC], UBCT 75, (Jmar 51, noMHôp 21, CBCTO(J)op 18, CBCT 16, KpacHBbiH 14, sejieHbiH, apKHH 12, SHavifl, cojiHu,e, HOC 11, oroHb, saKax, UBCXOK 10, CHHHH 9, rajicxyK, fl6jioKO, KapaHAam, posa, ^cejixbiH, 3Jiocxb 8. Literal Translation: blood 76 + 4 [the latter written in Hebrew], color 75, flag 51, tomato 21, traffic lights 18, light 16, beautiful 14, green, bright 12, banner, sun, nose 11, fire, sunset, flower 10, blue 9, tie, apple, pencil, rose, yellow, anger 8.


VI

PHRASEOLOGY


16

ON THE CROSS-LINGUISTIC EQUIVALENCE OF IDIOMS Dmitrij Dobrovol'skij, Institute of Russian Language, Russian Academy of Sciences, Moscow

Abstract In this paper, some general issues of contrastive idiom analysis will be discussed. The main problem in this field stems from the fact that most idioms which have traditionally been regarded as absolute parallels, in reality, are far from being equivalent. To find out all types of differences between idioms which are intuitively perceived as being semantically similar is the most important aim of contrastive idiom analysis. Its achievement would help to distinguish real cross-linguistic equivalents from pseudoequivalents and to explain the nature of idiom syntactics, semantics and pragmatics. A typology of relevant cross-linguistic differences is put forward here.

1 Preliminary Remarks Methods and tools of cross-linguistic idiom analysis vary widely according to their aims. The most general and fundamental line of research in this field deals with the relations between idiosyncratic phenomena and universal features of idioms (for details cf. Dobrovol'skij, 1988; 1992; Abraham, 1989). In recent years cognitive and cultural aspects of these relationships have become especially crucial (Dobrovol'skij and Piirainen, 1997). This area of research includes many concrete questions. If, for example, two languages show major similarities in the field of idioms, it is important to know the reasons for these similarities. In principle they could be traced back to accidental parallels, to borrowing processes, to genetic factors, or even to the universal character of conceptualization of the given entities. Also, the cross-linguistic contrasts have to be questioned. They can go back to coincidences of metaphorization, to consistent preferences of certain conceptual metaphors (in the sense of Lakoff, 1987) by the linguistic communities in question, i.e. to cognitive factors, or to relevant differences in the given cultures. It is evident that answering these questions requires interdisciplinary approaches. Investigations of this kind would contribute not only to the theory of phraseology, but also to cognitive linguistics, language universals research and the semiotics of culture. However, there are also more practically oriented tasks in the field of cross-linguistic idiom analysis, namely to develop appropriate descriptive tools for evaluating semantically similar idioms of both source and target language (LI and L2) in terms of their functional

203

204 Langue and Parole in Synchrony and Diachrony equivalence. This more practically oriented cross-linguistic approach to idioms is often called contrastive idiom analysis. The main problem in this field stems from the fact that most idioms which have traditionally been regarded as absolute parallels, in reality, are far from being equivalent. For this phenomenon some explanations can be put forward. First of all, the more progress in semantic analysis that is achieved, the more items of the lexicon turn out to be languagespecific, because ,,every language draws semantic distinctions which other languages do not" (Wierzbicka, 1996: 15). Cf. also Givon (1995: 18): the lexicon represents the bulk of culture-specific cross-linguistic diversity. The diversity represented in the lexicon is that of world-view, i.e. a group's perspective on its relativelystable conceptual universe. Parallels in the ,,core meaning" do not necessarily mean perfect equivalence in language use. This accounts especially for lexical items denoting non-physical entities (e.g., emotion concepts or mental predicates), which tend to depend much more on linguistic factors than words denoting physical entities (e.g., artefacts or natural kinds)1, because the latter have a different ontological status from the former. Since most idioms are predicates with regard to their semantics (i.e. lexical items denoting properties, states, actions, events and the like), they display a high degree of idiosyncrasy. Another factor of the tendency towards idiosyncratic configurations in the semantic structure of idioms has to be mentioned: the image component of idioms' plane of content. As an essential part of the conceptual structure (cf. Lakoff, 1987: 450 ff; Gibbs, 1993), mental images evoked by idioms influence their actual meanings. Furthermore, the image component is often responsible for relevant restrictions in the usage of idioms (for details cf. Dobrovol'skij, 1996; 1997a: 171-203; Baranov and Dobrovol'skij, 1998). On the other hand, it should be emphasized that it is not possible to predict the actual meaning of an idiom on the basis of its image, i.e. its literal meaning. The same image can often be traced back to different conceptual metaphors, that is to say, the lexical structure of an idiom does not tell us which conceptual metaphor has to be taken as the framework within which this idiom has to be interpreted. The result is that - both within the same language and cross-linguistically - semantic differences between idioms cannot be predicted. To reveal those differences is a purely empirical endeavour, i.e., they can be discovered only by an investigation of their range of use. Hence, given an LI-idiom and some L2-idioms which are considered near-equivalents to the LI-idiom, the only way to find out possible differences between them is to inquire into their functioning in authentic texts. This accounts not only for cross-linguistic non-parallelism in the meaning of idioms in question (idiom semantics), but also for relevant differences in their combinatorial and transformational properties (idiom syntactics) as well as for specifics of the situation in which they can be used (idiom pragmatics). For both theoretical and practical issues, it is much more important to find out nontrivial differences between lexical units of LI and L2 than to point to their similarities (cf. in this regard Tobin, 1989: 31). A systematic 1 The claim that cases of cross-linguistic full-equivalence are rare holds also for words denoting physical entities, cf. Jakobson's (1966) observation that even such a simple word as cheese has no full equivalent in Russian, because cottage cheese is regarded (and named) in Russian not as a kind of cheese, but as a dairy product of its own.

On the Cross-Linguistic Equivalence of Idioms 205 empirical search for relevant cross-linguistic differences in idiom semantics, syntactics and pragmatics needs certain theoretical guidelines, which would ensure its systematic character. To provide these guidelines is a task of the theory of contrastive idiom analysis, which has to elaborate parameters and procedures of comparison. Before going into detail about relevant parameters of cross-linguistic idiom comparison, let us consider the role of idiom equivalents in the translation of texts.

2 Translation and Idioms If we consider the tasks of contrastive idiom analysis from a purely practical perspective we have, first, to look for equivalents which can be used in the translation of LI-texts containing idioms into L2. If an LI-text containing an idiom is to be translated into L2, as a rule, some ad hoc solutions must be found, because equivalents provided by dictionaries (i.e. L2-idioms which semantically correspond to a given LI-idiom to a certain extent), in most cases, do not meet relevant contextual conditions. Whereas bilingual dictionaries of idioms mostly try to find an L2-idiom (and not a single word or a free word combination) to explain the meaning of the corresponding LI-idiom, for translation adequacy it is totally irrelevant whether a given Llidiom is translated into L2 by an idiom, a word or a free word combination. Moreover, even the imagery basis of corresponding lexical units (idioms or one-word metaphors) does not matter in most cases, except for puns and wordplays (see also section 3.2.2). Consequently, the German idiom (1), in most cases, can be translated into Russian by collocation (2), however not in contexts like (3). (1)

(2)

(3)

German jmdm. den schwarzen Peter zuschieben "to give someone the black Peter" 'to give someone the blame for a negative situation' Russian svalit Vperelozit' vinu za cto-libo na kogo-libo "to put the blame for something on someone" 'to give someone the blame for something' Und als Frage verpackt gibt er den schwarzen Peter an die Tarifpartner wetter. Ob das mit der Arbeitszeitverkiirzung 'ne gltickliche Sache gewesen sei? (Mannheimer Morgen, 28.03.1989) [And he passes the black Peter, packed as a question, on to the employers and unions, whether it was a good thing to reduce working hours?]

The NP-constituent of the Russian collocation (2) vina "blame" denotes a mental entity and does not combine with verbs like upakovat' "to pack" because the blame (unlike the black Peter, which is the name of a playing card) cannot be packed. In cases like this the image component of idiom semantics plays a crucial role in finding a functionally acceptable translation equivalent. The view that for an adequate translation it is important to have the same amount of idioms in the target text as in the source text (which can often be found in the literature on phraseology) is absurd. Compared to other lexical units, idioms have no special value, they can be translated by all possible means. A word of L2 having the same functional properties as the LI-idiom in question is a much better translation equivalent than an L2-idiom which, for one

206 Langue and Parole in Synchrony and Diachrony reason or another, sounds strange in the given context. Such factors as frequency are enormously important in this regard (cf. section 3.2.3). Moreover, not only is the translation of idioms by idioms a phantom ideal, but also the equivalence of isolated lexical items (no matter whether they are idioms or words) is not of any relevance for the quality of translation. The only point that matters is the translation adequacy of the whole text. As Jakobson (1966) pointed out, most lexical units are untranslatable, but every utterance can be translated. Since all possible contextual conditions, including implicatures of discourse, cannot be predicted, the bilingual dictionaries of idioms are basically incapable of registering all L2-equivalents of a given LI-idiom which might be preferable in all potential translations from LI into L2. Thus, the best instruments to serve the needs of a translator are parallel textual corpora and other text-oriented tools. The comparison of original texts and their translations into other languages reveals a broad range of potential translations with regard to idioms: from word-by-word translations to the use of other types of lexical units. From the point of view of translation it seems to be more important to know all syntactic, semantic and pragmatic properties of LI-idioms in question than to have a list of possible L2-equivalents, because the choice of equivalents each time depends on contextual conditions and cannot be predicted. Consequently, questions of the cross-linguistic equivalence of idioms are of no interest for the practice of translation. From this point of view, the scientific legitimacy of investigations into this area is dubious in itself. However, there are other fields of linguistic activities for which the aim of comparing all similar idioms (i.e. idioms that are intuitively perceived as equivalents) and describing all relevant differences between them seems reasonable. Compare, first of all, bilingual lexicography and second language acquisition. Also from a purely theoretical point of view, it is crucial to elaborate on the question of which idioms are at the speakers' disposal in LI and L2 for denoting certain concepts. This requires the description of LI- and L2-idioms according to their semantic fields in the first place (cf. Dobrovol'skij, 1995), i.e. it must be found out which LI- and L2-idioms are closest in meaning and function. As the second step, an apparatus for indicating parameters of nonequivalence has to be developed, in order to distinguish between semantically similar idioms. Thus, a typology of relevant differences between heuristically similar idioms of LI and L2 is needed for both practically oriented tasks (especially in the field of bilingual lexicography and second language acquisition) and for the theoretical aims of cross-linguistic idiom analysis.

3 Parameters of Idiom Comparison The following will concentrate on some parameters of cross-linguistic comparison, trying to offer a typology of non-parallel features between LI- and L2-idioms. On the one hand, a typology of this kind can help to distinguish real cross-linguistic equivalents from pseudoequivalents and, in this way, improve bilingual dictionaries. On the other hand, it could help in explaining the nature of idiom syntactics, semantics and pragmatics because, being concerned with relevant differences in idiom functioning, comparison is an instrument of investigation not only into cross-linguistic issues, but also into intralinguistic features and properties of the items in question. Relevant parameters can be roughly divided into three groups according to three wellknown semiotic dimensions: syntactics, semantics and pragmatics. Although it is hardly

On the Cross-Linguistic Equivalence of Idioms 207 possible to draw a clear borderline in every concrete case (because, for example, nearly all syntactic transformations have certain semantic and/or pragmatic entailments), it seems to be reasonable to group relevant parameters of comparison according to their fundamental semiotic and linguistic features. The primary aim is not to classify concrete cases of cross-linguistic near-equivalence, but to offer a typology of parameters which would allow us to describe all idioms under consideration more consistently.

3.1 Syntactics In order to describe syntactic properties of an idiom, we basically have to answer the following three questions: (a) With which words does it combine? (b) Into which syntactic models can it be embedded? (c) What kinds of transformations does it undergo? For space reasons it is not possible to analyze each of these questions in detail here. Let us briefly mention the corresponding types of non-equivalence and give a few examples. (a) Sometimes idioms which are very close semantically can be found in LI and L2. Compare, for instance, the German idiom (4) and the semantically equivalent Russian idiom (5). (4) (5)

German in den Tag hineinleben "to live into the day" Russian zit' odnim dnem "to live by one day" both idioms meaning 'to live for the moment'

Still these idioms are not fully equivalent from a functional perspective, because the German idiom combines with adverbial adjuncts, such as sorglos 'happy-go-lucky', flott 'jolly', gedankenlos 'thoughtlessly' (cf. (6)), whereas the Russian idiom does not. (6)

Sparen wird mit Worten gelobt, heimlich als "Burgerlichkeit" mifiachtet, als Dummheit verlacht; - man lebt gedankenlos in den Tag hinein . (Jaspers. Die Atombombe und die Zukunft des Menschen) [ economizing is verbally praised, but in secret disregarded as being "bourgeois", it is mocked as stupidity; - one has to thoughtlessly live for the moment .]

These constraints on the use of the Russian idiom (5) can be explained on the basis of its formal structure. Formally, this Russian idiom contains a /zow-adjunct (odnim dnem "by one day"). This structural property makes it impossible to attach another /zow-adjunct within the same sentence. The German idiom (4) formally contains not a /zow-adjunct, but a \vhereadjunct (in den Tag hinein "into the day"). This allows the German idiom to easily attach a /zow-adjunct. Observations of this kind can be used as additional proof of the psychological reality of the mental image underlying the figurative (actual) meaning of idioms and its crucial

208 Langue and Parole in Synchrony and Diachrony role for real usage. Although, from the point of view of the actual meaning, the constituent group in den Tag hinein of the German idiom (4) is a /zow-adjunct (just the same as in the lexical-semantic structure of its Russian near-equivalent (5)), the literal interpretation of the PP in den Tag hinein leads to a different combinatorial situation, hi terms of conceptual metaphors, we are dealing with an "orientational metaphor" in the case of (4), and with an "instrumental metaphor" in the case of (5). From the perspective of the cognitive theory of metaphor, examples like this provide striking evidence for the role of linguistic properties of the single constituent parts taken literally, as well as the role of the underlying conceptual metaphor for the interpretation of the whole idiom taken in its actual meaning. Often a given LI-idiom and its L2-near-equivalent differ with regard to filling in their valencies. Cf. the discussion of examples (7-12) under (b). Since, in most cases, the translation equivalence can be provided by the diathesis-modification, cross-linguistic differences of this kind are regarded here as contrasts in syntactic embedding. (b) To illustrate the importance of differences in syntactic embedding of the idiom in question, let us consider the German idiom (7) and the semantically equivalent Russian idiom (8).2 (7) (8)

German zwei Fliegen mit einer Klappe schlagen ,,to hit two flies with one flap" Russian ubit' dvux zajcev (odnim udarom/vystrelom) ,,to kill two hares (with one blow/shot)" both idioms meaning 'to accomplish two important things at once, realize two goals through a single action' (cf. Lubensky, 1997: 244)

In standard cases, both idioms (7) and (8) require an NP denoting a human being in the subject position, compare (9) and (10), where it is possible to translate the idioms in question by each other, as well as by the English idiom to kill two birds -with one stone. (9)

2

Oberburgermeister Hans Martin schldgt zwei Fliegen mit einer Klappe: Mit dem BHKW Hanau wird aktiv Umwelt geschutzt und gleichzeitig Energie preiswert erzeugt: 88 bis 90% der Zufuhrenergie wird in Nutzenergie umgesetzt. (Mannheimer Morgen, 05.07.1989) [Lord Mayor Hans Martin kills two birds with one stone: with the BHKW Hanau, the environment is actively protected and at the same time energy is produced at low cost: 88% to 90% of supply energy is transposed into utilizable energy.]

Differences in the image component of both idioms do not seem to be of any importance for their semantic equivalence, except for wordplays (cf. for details section 3.2.2).

On the Cross-Linguistic Equivalence of Idioms 209 (10)

Ja pomnju rasterjannyj vid M[andelstama], kogda my vernulis' domoj, pogljadev, kak proisxodit iz"jatie [cerkovnogo imuscestva]... Predlozenie [patriarxa] Tixona [organizevat' pomos' golodajuscim] otklonili, a teper' vopjat, cto cerkovniki ne zalejut golodajuscix i prjacut svoi sokrovisca. Odnim udarom ubivali dvux zajcev. zagrebali zoloto i porocili cerkovnikov. (Nadezda Mandelstam. Vtoraja kniga) [I remember the appalled look on M[andelshtam]'s face as we returned home after watching the Confiscation" [of church property] for a while... [Patriarch]... Tikhon's proposal [to organize aid for the starving] had been turned down, and now they were shrieking that the church people have no pity for the starving and were hoarding their treasures. In this way they were killing two birds with one stone: while vilifying the church, they were also grabbing its gold. (Nadezhda Mandelshtam. Hope Abandoned)] (cited after Lubensky, 1997: 224)

In contexts like (11), the German idiom (7) has a different argument structure: the subject position is occupied by an NP with an abstract meaning. (11)

Gase aus Miilldeponien wirken sich auch auf das Weltklima aus. Eine Nutzung dieser Energie schldgt zwei Fliegen mit einer Klappe. (Mannheimer Morgen, 23.05.1989) [ Gases from waste areas influence the world climate. Utilization of this energy kills two birds with one stone.}

To be able to use idiom (8) in the translation of context (11) into Russian,3 here we have to alter the syntactic construction and to transform the diathesis of this VP-idiom into something like (12). (12)

ispol'zuja etu energyu, my ubivaem dvux zajcev [ using this kind of energy, we are killing two birds with one stone]

(c) We deal with transformational factors of non-equivalence in cases in which a given Llidiom and the corresponding L2-idiom are equivalent along all parameters but one, namely the LI-idiom does not undergo the same transformations as the L2-idiom: e.g., Ll-idiom can be nominalized, and L2-idiom cannot. In other words, the ability to modify the idiom structure according to relevant contextual conditions can differ, even if the idioms in question are "fully equivalent" in the traditional sense. This can be illustrated by the German idiom (13) and its Russian counterpart (14). Whereas (13) can easily be passivized, (14), which is semantically equivalent to (13), cannot (cf. (15) and (16)). The only way to translate sentence (15) into Russian preserving the same topic-comment-structure is to use the impersonal construction (18), which, strictly speaking, corresponds not to (15), but to (17). (13)

3

German jmdm. das Fell tiber die Ohren ziehen "to pull someone the skin over the ears"

Although many native speakers of German judged this context as a stylistic failure, its word-by-word translation into Russian seems to be still much less acceptable.


(15) (16) (17) (18)

Russian drat '/sodrat' tri skury s kogo-libo ,,to take off three skins from someone" both idioms meaning 'to exploit someone pitilessly by making him pay large taxes, very high interest, exorbitant prices etc.' German Ihm wurde das Fell iiber die Ohren gezogen. ,,The skin was pulled him over the ears." Russian ? S nego byli sodrany tri skury. ,,Three skins were taken off from him." German Man hat ihm das Fell uber die Ohren gezogen. ,,One pulled him the skin over the ears." Russian S nego sodralli tri skury. ,,One took off three skins from him."

3.2 Semantics Basically the content plane of idioms consists of two macrocomponents: (i) actual meaning, and (ii) mental image, being both starting-point and the conceptual foundation of semantic reinterpretation. These two macrocomponents are independent of each other to a certain extent. One consequence of this fact is the existence of idioms which have (nearly) the same image, but differ with regard to their actual meanings, as well as the existence of idioms which have (nearly) the same actual meaning, but differ with regard to their images. Hence, these two major types of non-equivalence and their different combinations can be distinguished. It goes without saying that a certain degree of similarity between the idioms under consideration is a crucial precondition of their analysis in terms of cross-linguistic equivalence vs. nonequivalence. If we take two idioms which significantly differ in both actual meaning and image component, there is no reason to compare them at all, not to mention regarding them as possible candidates for cross-linguistic equivalence. Let us first consider the types of non-equivalence which are characterized by a similarity of images and non-parallel features of actual meaning. 3.2.1 Contrasts in actual meaning. Here at least the following types can be distinguished: (a) idiomatic ,,false friends"; (b) cross-linguistic near-synonyms; (c)"asymmetrical polysemy". (a) The most striking cases of semantic non-equivalence are the so-called "false friends" (cf, e.g., Piirainen 1997; Burger, 1998: 200). Compare the English idiom (19) and its Russian pseudo-equivalent (20). (19)

English to throw dust in/into someone's eyes 'to confuse (someone) or take his attention away from something that one does not wish him to see or know about'(Longman DEI, 1979: 90)

On the Cross-Linguistic Equivalence of Idioms 211 (20)

Russian puskat' pyl' v glaza komu-libo ,,to throw dust in/into someone's eyes" 'trying to impress other people, show oneself or his/her position in a much better light than it really is' (cf. Gurevic and Dozorec, 1995: 387).

Both idioms consist of analogous constituents and have an identical image basis. Nevertheless, they reveal significant differences in actual meaning. From a linguistic viewpoint phraseological "false friends" turn out to be a more sophisticated problem than "one-word false friends". The latter are words of LI and L2 which sound similar and often go back to the same etymological roots, such as German vital, eventuell, Fabrik and English vital, eventual, fabric. Such cases are fairly well described and can be found in every good bilingual dictionary. On the other hand, idioms which are classified here as "false friends" resemble each other not on the phonetic and/or graphic level, but on the level of lexical constituent parts and mental images, i.e. on the level of "inner form". Hence investigations into this problem require a semantic and conceptual analysis of given idioms, for example in terms of the cognitive theory of metaphor. (b) Another type of semantically based non-equivalence can be exemplified by the German idiom (21) and its Russian counterpart (22). (21)

(22)

German den Bock zum Gartner machen "to make the ram/he-goat into the gardener" 'to allow a person, who seems to be able to do much harm in a given field of activities, to do just these things' Russian puskat' kozla v ogorod "to let the he-goat into the kitchen-garden" 'to allow a person, who seems to be able to do much harm in a given field of activities and to derive benefit for himself/herself, to do just these things'

The subtle semantic differences which become evident as a result of contrasting idioms, such as (21) and (22), have to be analyzed in more detail. The semantic explications of (21) and (22) differ with regard to the semantic element 'to derive benefit for himself/herself, which is a part of the meaning of the Russian idiom (22); compare also the English idioms to put the cow to mind the corn, to put the wolf in charge of the sheep, to put the fox to guard the henhouse/the chickens, which are equivalent to the Russian idiom (22) rather than to the German idiom (21). The meaning of the corresponding German idiom (21) does not include this semantic element. Therefore idiom (21) can be used in a broader range of contexts than (22). Cf. example (23), which cannot be translated into Russian with the help of idiom (22). (23)

einige Zeit spater stieB er darauf, dafi er sozusagen den Bock zum Gartner gemacht hatte, derm es stellte sich heraus, dafl bei der Planung der Flachdacher der Architekt fehlerhaft gearbeitet hatte. (Mannheimer Morgen, 26.04.1986) [ some time later he found out that he, so to speak, had made the ram/he-goat into the gardener, since it turned out that while planning the flat roofs the architect had made a mistake]

212 Langue and Parole in Synchrony and Diachrony This type is rather similar to (a) in the sense that being based on very similar mental images, which go back to the same conceptual structure, the actual meanings of these two idioms are not completely identical. However, there is an important difference between type (a) and type (b). Whereas the so-called "false friends" display significant semantic differences, often developing meanings which have nothing in common (cf. (19) and (20)), the idioms which are subsumed under (b) have the same core meaning. The meanings of such idioms differ only slightly, and these differences, as a rule, concern peripheral elements of their semantics. Since near-equivalents of this kind traditionally were treated as "full equivalents" it seems especially important to draw attention to them from both a theoretical and a practical point of view. For a semantic theory it is a reasonable task to single out all ostensible crosslinguistic equivalents which, from a functional perspective, are not fully equivalent, and to describe all relevant semantic differences. It would help to find out the function of nontrivial semantic components, such as presuppositions, semantic consequences, the speaker's attitude, and the like. From the point of view of bilingual lexicography and second language acquisition, there is a need for developing metalinguistic tools which would enable the user to get access to functional properties of the idioms in question. (c) The next type of non-equivalence I would like to mention here is based on a semantic property which I would call "asymmetrical polysemy". We deal with this semantic property in all cases in which a given LI-idiom has more than one meaning, whereas the corresponding L2-idiom has only one meaning. This monosemic L2-idiom may fully correlate with one of the meanings of the polysemic LI-idiom, but even in this case both idioms, taken as a whole, are not fully equivalent. Their contrastive description and the relevant dictionary entries have to contain information about such cross-linguistic semantic asymmetry. Examples of this kind can be taken from the semantic field 'death/murder'. Many German idioms of this field have besides the meaning of 'death/murder' a second meaning, something like 'dysfunction'. For example, the German idiom (24), according to Duden 11 (1992: 245), has two meanings: (compare the English idiom to give up the ghost). The Russian idiom (25), which fully corresponds to this German idiom in terms of lexical constituent parts and the image basis, has only one meaning (FSRJ, 1986: 187). (24)

(25)

German den Geist aufgeben "to give up the mind/spirit" 1. (obsolete) 'to die', and 2. (coll.) 'to stop working' Russian ispustit' dux "to give up the mind/spirit" 'to die'

Thus, all bilingual dictionaries which consider these two idioms absolute equivalents (cf, for example, Binovic and Grisin, 1975: 207) give the user misleading information. As a matter of fact, these idioms are far from being equivalent. Even taken in the corresponding meaning 'to die', they turn out to be not fully equivalent from the functional point of view, because the German idiom (24) is practically never used in this meaning today, whereas for the Russian idiom (25) this meaning is the only possible one. All contexts containing the idiom (24) which could be found in the textual corpora of the Institute of German Language in Mannheim

On the Cross-Linguistic Equivalence of Idioms 213 (Mannheimer Korpora geschriebener Sprache) illustrate only the second meaning of this idiom, cf., for example, (26). (26)

Sollte der Scheibenwischer einmal seinen Geist aufgeben, hilft der Apfel-Trick: Man schneidet einen Apfel in der Mitte durch und reibt mit den Halften uber die Scheibe. (Mannheimer Morgen, 31.05.1995) [Should the windscreen wiper give up its ghost someday, the "apple-trick" helps: cut through an apple in the middle and rub with the halves over the windscreen.]

3.2.2 Contrasts in images. Now let us consider the idioms of LI and L2 which are similar with regard to their actual meanings, but differ with regard to their images. In such cases we are dealing with semantic similarity of a high degree, since idioms having identical actual meanings and similar syntactic and pragmatic features can basically be treated as real equivalents. But in certain contexts the image component of the plane of content of a given idiom plays a crucial role in its functioning. In such cases the idiom cannot be translated into L2 by the corresponding idiom which differs with regard to its imagery (cf. similar ideas in Eismann, 1995: 100-101). For example, in standard contexts the German idiom (27) can be translated into Russian by the idiom (28). (27) (28)

German nicht alle Tassen im Schrank haben "to have not all cups in the cupboard" Russian ne vse doma u kogo-libo literally "not all people at home with someone" evoking a mental image of a situation of "not all people being at home at someone's place" both idioms meaning 'to be mad'

Although both images can be traced back to the same basic conceptual structure, something like "absence of completeness or absence of entities which have to be there in a standard case" being the source concept for the target idea of 'madness', they reveal significant differences on the "rich" conceptual level (the image of cups being in the cupboard vs. people at home). In contexts containing a wordplay, such as (29), these differences turn out to be crucial. (29)

DaB er noch alle Tassen im Schrank hat, stellte der Mannheimer Fotograf Hartmut Suckow iiber Jahr und Tag unschwer fest: Da sein Atelier genau gegeniiber dem neuen Theaterhaus liegt, machte sein Geschirr alle baulichen Kraftakte bei der Sanierung scheppernd mit. (Mannheimer Morgen, 22.05.1989) [For the photographer from Mannheim, Hartmut Suckow, it has for years been not difficult to state that he still has all cups in the cupboard: Since his studio is situated exactly opposite the new theatre, his crockery continually clinked in unison with all the powered acts of building work going on during the renovations.]

In cases like this the German idiom in question cannot be translated into Russian by the idiom (28).

214 Langue and Parole in Synchrony and Diachrony 3.2.3 Combined semantic contrasts. Concluding the description of semantically based types of cross-linguistic non-equivalence, I would like to mention that many different combinations of the types discussed in this section are possible. For space reasons let us consider only one example. Although the German idiom (30), the English idiom (31) and the Russian idiom (32) are traditionally regarded as full equivalents (cf. Binovic and Grisin, 1975: 377), they differ with regard to both actual meaning and image component. (30) (31) (32)

German mit dem silbernen Loffel im Mundgeboren sein ,,to be born with the silver spoon in the mouth" English to be born with a silver spoon in one's mouth Russian rodit 'sja v sorocke "to be born in (a) shirt"

Whereas the German idiom (like the English idiom which has the same lexical structure and is based on the same mental image) means 'to have wealthy parents, to be born into a rich family', the Russian idiom means something like 'to be lucky, esp. to escape a dangerous situation which seemed to be inevitable, or to achieve something desired that seemed to be unattainable' (cf. Gurevic and Dozorec, 1995: 403). The images standing behind these idioms are also not identical, though, from a cognitive point of view, there is a certain similarity between them. In terms of the cognitive theory of metaphor, both mental images can be traced back to an identical conceptual structure, namely the idea of being born with a special facility (and, therefore, to be something special, not like other people). This conceptual structure serves as the source domain for the metaphoric inference in both cases. This seems to give rise to treating these idioms as equivalents in bilingual dictionaries, and proves at the same time the psychological reality of conceptual metaphors. From the perspective of linguistic theory cases like this are of prime interest, because they demonstrate the need of an additional tool for analyzing the image component of the content plane of idioms. In Dobrovol'skij and Piirainen (1997) we suggested the inclusion of a so-called "symbolic component" in the description of figurative language, which helps to explain phenomena that could not be explained on the basis of conceptual metaphors. Compare, in the case above, silver as a symbol of wealth. The corresponding constituents in the lexical structure of the German idiom (30) and the English idiom (31) activate this "symbolic" knowledge, which in its turn correlates with the actual meaning 'to have wealthy parents, to be born into a rich family'. The absence of such a constituent in the lexical structure of the Russian idiom (32) explains the absence of the sense 'wealth' in its semantic structure.

3.3 Pragmatics Since among the factors of non-equivalence the pragmatic ones have been investigated most intensively, and described in a most explicit and accurate way (cf, for example, Burger, 1991; Dobrovol'skij, 1997a: 104-106; 1997b), let us simply enumerate them here and give a few examples. The pragmatically based differences between semantically and syntactically similar idioms of LI and L2 can go back to:

On the Cross-Linguistic Equivalence of Idioms 215 (a) differences in their stylistic properties; (b) differences with regard to the degree of their familiarity and/or textual frequency; (c) differences in the cultural component of their plane of content; (d) differences in their illocutionary function. (a & b) Differences in stylistic properties and in the degree of familiarity can be exemplified by idioms (33) and (34). (33) (34)

German einen Bock schiefien "to shoot a ram/he-goat" Russian ubit' bobra "to kill a beaver" both idioms meaning 'to make a bad mistake'

Idioms (33) and (34) are not equivalent, as erroneously assumed, for example, in Gladrow (1993: 114), because (33) is a familiar, widely known idiom of current German, whereas (34) is totally obsolete (and not colloquial, as stated in Lubensky, 1997: 17). Most of today's native speakers of Russian do not even know that this idiom exists. Using this idiom, the speaker would achieve a special stylistic effect which would be totally different from the stylistic properties of the corresponding German idiom (33). (c) Differences in the cultural component of the content plane of near-equivalent idioms can be exemplified by (35) and (36). (35) (36)

Russian cernyj voron ,,a black raven" German die grune Minna ,,the green Minna" both idioms denote a police vehicle for the transportation of prisoners and arrested people

Being an idiom with identifying semantics, i.e. an idiom which allows referential use (cf. Dobrovol'skij, 1992), (35) denotes a concept which evokes a lot of cultural and historical associations. In this sense, the referent itself is unique and culturally marked. On the other hand, the image component of the idiom's meaning corresponds with a concept that plays an important role in Russian folklore, i.e. with the concept RAVEN. This allows us to consider (35) a culture-specific idiom of Russian with regard to other languages (for details cf. Dobrovol'skij, 1997b). A similar case is discussed in Segura Garcia (1997: 224). Idiom (37), which is common in Peru, is not fully equivalent to (38); though both idioms have the same denotational meaning. The reason is that (37) is culture-specific in comparison to (38), and vice versa.

216 Langue and Parole in Synchrony and Diachrony (37) (38)

Spanish el chino de la esquina ,,the Chinese at the corner" German Tante-Emma-Laden ,,aunt-Emma-shop" both idioms meaning 'a small shop of traditional type'

(d) In every known language there are idioms whose illocutionary potential is more or less fixed and bound to certain speech act types. For example, Burger (1991: 21) points out that idiom (39) is normally used in utterances with the illocutionary function of warning or threat, such as (40). (39)

(40)

German blaues Wunder ,,blue miracle" 'a big unpleasant surprise' Da kannst/wirst du dein blaues Wunder erleben. ,,Then you can/will see your blue miracle." « 'you will get what's coming to you'

If a given LI-idiom tends to be used in certain speech acts, and its near-equivalent L2correspondent does not, or is bound to speech acts of different types, the idioms in question cannot be considered absolute equivalents. Compare idioms (41) and (42), which have the same semantics (with regard to both actual meaning and image component) and a very similar lexical structure. (41) (42)

German im Klartext ,,in the clear text" Russian othytym textom ,,by (the) open text" both idioms meaning 'in a very clear and explicit way'

The German idiom (41) is mostly used to introduce an utterance which reformulates the previous utterance in a more explicit and exact way, i.e. in the function of ,,self-correction", something like (43), whereas the Russian idiom (42) is normally used in statements like (44). (43)

(44)

Im Klartext heifit es... ,,In the clear text, it means..." Um das noch einmal im Klartext zu sagen... ,,To say it once more in the clear text..." Das bedeutet also im Klartext... ,,Hence, this means in the clear text..." 'to put it more clearly...' Ona skazala emu othytym tekstom, cto... ,,She told him by (the) open text that..." 'she let him explicitly and clearly understand that...'

On the Cross-Linguistic Equivalence of Idioms 217 Some nontrivial pragmatic properties of idioms become evident only when analyzing big textual corpora. Taken in itself, the German idiom (45) seems to be easy to translate in every language by means of corresponding idioms, because in all languages investigated from the point of view of figurative lexical units, there are many idioms having the core meaning 'to be mad'. (45)

German von alien guten Geistern verlassen sein ,,to be left by all good spirits/genii" 'to be mad';«to have taken leave of one's senses

However, an investigation of the textual corpora of the Institute of German Language in Mannheim has shown that (45) is predominantly used in hypothetical utterances, such as (46). (46)

Zugleich sprach sich der Finanzminister deutlich gegen Forderungen nach einem Sonderopfer der Bundesbiirger fur die DDR in Form von Abgabenerhohungen oder als Verzicht auf die Steuerreform aus. "Wir waren von alien guten Geistern verlassen, wenn iiber den Umweg der Deutschlandpolitik sozialistische Vorstellungen eingebracht wtirden", kommentierte Waigel dazu. (Mannheimer Morgen, 15.11.1989) [At the same time, the Minister of Finance clearly expressed himself against claims for a special sacrifice by the citizens of the Federal Republic for the GDR in the form of increased payments or abandoning the tax reform. ,,We would be left by all good spirits/genii if we indirectly brought in socialist ideas by way of the Germany policy", commented Waigel.]

In Russian no single idiom with the same illocutionary potential could be found among idioms denoting abnormal mental states. Consequently, no Russian idiom from the semantic field 'abnormal mental states' can to be regarded as fully equivalent to (45).

4 Conclusion A systematic description of functionally relevant factors of the use of idioms allows us to develop a typology of parameters for their cross-linguistic comparison. First of all, such a typology aims at improving the theoretical background of contrastive idiom analysis. In practice, such a typology helps to find out all subtle differences in the semantic, syntactic and pragmatic properties of LI- and L2-idioms. This is especially important in those cases where traditional analysis and bilingual dictionaries postulate a relation of "full equivalence" between a given pair of idioms, disregarding their functional non-parallelism.

Acknowledgment Preparation of this paper was supported in part by RFFI-grant 98-06-80081.

218 Langue and Parole in Synchrony and Diachrony References Abraham, Werner (1989). Idioms in contrastive and in universally based typological research: toward distinctions of relevance. In: Proceedings of the first Tilburg workshop on idioms (Martin Everaert and Erik-Jan van der Linden, eds.), 1-22. ITK, Tilburg. Baranov, Anatolij N. and Dmitrij O. Dobrovol'skij (1998). Vnutrennjaja forma idiomy i problema tolkovanija. Izvestija RAN, serija literatury ijazyka. 1, 36-44. Binovic, Leonid E. and Nikolaj N. Grisin (1975). Nemecko-russkij frazeologiceskij slovar' = Deutsch-russischesphraseologisches Worterbuch. Russkij Jazyk, Moskva. Burger, Harald (1991). Phraseologie und Intertextualitat. In:. EUROPHRAS 90: Akten der internationalen Tagung der germanistischen Phraseologieforschung (Christine Palm, ed.), 13-27. Uppsala Univ., Uppsala. Burger, Harald (1998). Phraseologie: eine Einfuhrung am Beispiel des Deutschen. Erich Schmidt, Berlin. Dobrovol'skij, Dmitrij (1988). Phraseologie als Objekt der Universalienlinguistik. Enzyklopadie, Leipzig. Dobrovol'skij, Dmitrij (1992). Phraseological universals: theoretical and applied aspects. In: Meaning and grammar. Cross-linguistic perspectives (Michel Kefer and Johan van der Auwera, eds.), 279-301. Mouton de Gruyter, Berlin & New York. Dobrovol'skij, Dmitrij (1995). Kognitive Aspekte der Idiom-Semantik. Studien zum Thesaurus deutscher Idiome. Narr, Tubingen. Dobrovol'skij, Dmitrij (1996). Obraznaja sostavljajuscaja v semantike idiom". Voprosy Jazykoznanija, 1, 71-93. Dobrovol'skij, Dmitrij (1997a). Idiome im mentalen Lexikon: Ziele und Methoden der kognitivbasierten Phraseologieforschung. WVT Wissenschaftlicher Verlag Trier, Trier. Dobrovol'skij, Dmitrij (1997b). Nacional'no-kurtiirnaja specifika vo frazeologii (I). Voprosy Jazykoznanija, 6, 37-48. Dobrovol'skij, Dmitrij and Elisabeth Piirainen (1997). Symbole in Sprache und Kultur: Studien zur Phraseologie aus kultursemiotischer Perspektive. Brockmeyer, Bochum. Duden 11 (1992). Duden. Redewendungen und sprichwortliche Redensarten. Worterbuch der deutschen Idiomatik. Dudenverlag, Mannheim etc. Eismann, Wolfgang (1995). Pragmatik und kulturelle Spezifik als Problem der Aquivalenz von Phraseologismen. In: Von der Einwortmetapher zur Satzmetapher (Rupprecht S. Baur and Christoph Chlosta, eds.), 95-119. Brockmeyer, Bochum. FSRJ (1986). Frazeologiceskij slovar' russkogo jazyka (Aleksandr I. Molotkov, ed.). 4th ed. Russkij Jazyk, Moskva. Gibbs, Raymond W. (1993). Why idioms are not dead metaphors. In: Idioms: processing, structure, and interpretation (Cristina Cacciari and Patrizia Tabossi, eds.), 57-77. Lawrence Erlbaum, Hillsdale (NJ). Givon, Talmy (1995). Functionalism and grammar. Benjamins, Amsterdam & Philadelphia. Gladrow, Wolfgang (1993). Faktoren der zwischensprachlichen Aquivalenz in der Phraseologie. In: Phraseology in education, science and culture (E. Kroslakova and P. Durco, eds.), 110-121. VSPN, Nitra. Gurevic, Valerij V. and Zanna A. Dozorec (1995). Kratkij russko-anglijskij frazeologiceskij slovar'. Vlados, Moskva.

On the Cross-Linguistic Equivalence of Idioms 219 Jakobson, Roman (1966). On linguistic aspects of translation. In: On translation (R.A. Brower, ed.), 232-239. Oxford University Press, Oxford & New York. Lakoff, George (1987). Women, fire, and dangerous things. The University of Chicago Press, Chicago & London. Longman DEI (1979). Longman dictionary of English idioms. Longman Group Limited, Harlow and London. Lubensky, Sophia (1997). Random House Russian-English dictionary of idioms. Random House & Jazyki russkoj kul'tury, New York & Moskva. Piirainen, Elisabeth (1997). ,,Da kann man nur die Hande in den Schofi legen". Zur Problematik der falschen Freunde in niederlandischen und deutschen Phraseologismen. In: Nominationsforschung im Deutschen (Irmhild Barz and Marianne Schroder, eds.), 201-211. Lang, Frankfurt etc. Segura Garcia, Blanca (1997). Kulturspezifische Phraseologismen in literarischen Texten und ihre Interferenzen beim Ubersetzen vom Spanischen ins Deutsche. In: Phraseme im Text: Beitrage aus romanistischer Sicht (Annette Sabban, ed.), 221-236. Brockmeyer, Bochum. Tobin, Yishai (1989). Invariant meaning and non-equivalence in language and translation: the two ,,hows" in Modern Hebrew - a case in point. In: Translation and lexicography (Mary Snell-Hornby, E. Pohl and B. Bennani, eds.), 31-43. Benjamins & Paintbrush & Euralex, Missouri. Wierzbicka, Anna (1996). Semantics: primes and universals. Oxford University Press, Oxford & New York.


17

TOWARDS A DESCRIPTION OF ENGLISH AND FRENCH PHRASEOLOGY Christopher Gledhill, Department of French, University of St. Andrews, Scotland/UK

Abstract This paper sets out to establish the relationship between idioms and collocations and to explain how these terms relate to a general model of phraseology. We see phraseology as the relationship between a specialist language and the general or core language. We further argue that it is essential to ground notions such as phraseology in terms of discourse, an approach which emphasises the pragmatic and rhetorical functions of fixed expressions within the phraseological system, rather than simply their syntactic or semantic features. We argue that phraseology depends on the interplay between pragmatically marked expressions on the one hand (idioms), and their unmarked core equivalents on the other (collocations).

INTRODUCTION Linguists and non-linguists alike use a wide number of terms to express what are commonly thought of as chunks or strings in language. The following sample of terms express different aspects of this basic idea: ...cliche, collocation, compound word, dictum, fixed expression, formula, formulaic expression, idiom, lexical phrase, lexical unit, locution, phrase, phraseme, polyword, prefabricated expression, proverb, turn of phrase, word complex... (compare with the French variants: phrase toute faite, parlure (in Canadian French), tour de phrase, tournure, etc.)

These terms essentially capture the intuitive idea that speakers select sequences of words as a whole. While the internal structure of expressions may obey the usual principles of grammar, they are also recognised as cultural artefacts rather than simply sequences or syntagms. Fillmore, Kay and O'Connor encapsulate the lexical nature of fixed expressions when they describe them as: "...phenomena larger than words, which are like words in that they have to be learned separately as individual facts about pieces of the language, but which also have grammatical structure [and] interact in important ways with the rest of the language." (Fillmore et al. 1988:501) 221

222 Langue and Parole in Synchrony and Diachrony The idea that lexical sequences behave like words has been widely propagated by linguists such as Firth (1957) and Makkai (1992) and can be seen to have spread beyond the confines of lexicology and lexicography. For example, 'prefabs' are referred to in language acquisition theory (Granger 1994) and represent the type of expression that language learners can expect to use safely with little mastery of a language (such as: How about a ?, Pourquoi pas?). Psychologists in turn refer to 'formulae1 to describe the extent to which speakers access and predict sequences of words (Clark 1985). And the 'lexical phrase', an expression with a specific rhetorical function, is now seen as an important unit in text and discourse analysis (McCarthy and Carter 1994). In addition, there is much evidence in mainstream linguistics to suggest that multiword items behave as single words. For example, Firth (1957) proposed that grammatical features and categories form predictable sequences (colligations) in much the same way that single words form collocations. The idea has been pursued recently by van der Wouten (1997) in his discussion of long-range collocations and colligations extending beyond the boundary of the phrase (such as the negative associated with certain moods and verb forms). The theory of grammaticalization in Creole studies similarly emphasises the evolutionary conversion of lexical items into fully grammatical forms (Schwegler 1990). Many of the studies cited above imply that the fixedness of certain expressions eventually leads to word formation, as can be seen in because, parce que, of course, d'accord, maybe, peut-etre, today, aujourd'hui etc. and in the existence of well-known historical fusions (lord derived from loaf+ ward, vinaigre from vin + aigre) (Gross 1996). Idioms, collocations and other expressions therefore exist on a different linguistic level than the simple word, although in time they are used and become recognised as though they were simple lexical items, a process known as lexicalisation (Picoche 1992). Idiomatic expressions thus embody the Saussurian principle of arbitrariness, whereby each expression is a sign composed of more than one word form with either a conventional meaning (its semantic component) or conventional formulation (a syntactic component). These conventional expressions can be said to belong to a general system of phraseology, which we define here as 'the preferred way of saying things in a particular discourse'. Within this broad definition, the concept of phraseology extends from the basic terminology belonging to a specific field (such as the fixed compounds and jargon of genetics: gene expression, Vexpression du gene) to longer stretches of language which are typical of expository prose, or in our specific area, that of research articles (Results have shown that + X, ces observations ont demontre que + Y). Whereas terms such as idiom, cliche, lexical phrase and collocation often refer to discrete entities, phraseology denotes a broad system of expression. We argue here that phraseology is a continuum along which various types of expression are situated. For the purposes of this paper, we identify collocations and idioms as the opposite ends of this continuum. We attempt in the first instance to establish the continuum idiom - collocation according to pragmatic and rhetorical criteria. We then describe the role of collocations in specialist language, and attempt to demonstrate that 'collocational shift1 is key to our understanding of core and periphery in language.

English and French Phraseology 223 IDIOMS AND IDIOMATICITY It is often noted that idiomatic and multi-word expressions are lexically special, or have a special grammatical status. Idioms traditionally involve at least one central lexical item (usually a 'dead' or fixed metaphor: it's raining cats and dogs, il pleut des cordes), or an unusual formulation, for example to dress up to the nines ('to dress in one's best clothes') or its French equivalent etre sur son trente et tin (literally 'to be on one's thirty one'). At times the idiom is somewhat more motivated (i.e. transparent or easily interpreted) than its foreign counterpart which appears relatively opaque and arbitrary. For example, (1) a fat lot of good that'll do me (meaning ironically and informally 'that is completely useless') is easier to explain than the French equivalent (2) cela me fait une belle jambe, literally 'that gives me a nice leg'. According to traditional accounts, idioms resist changes in word choice or differ in the extent to which the expression can be transformed. Thus because one cannot say: II am done a fat load of good by that, or la great load of good that might do me, example (1) can be said to be idiomatic on lexical and syntactic grounds. Cruse (1982) summarizes the distinction between idioms and collocations very simply: collocations are grammatically simple but semantically complex (i.e. syntagmatic units, such as 'to take a break, faire une pause'), while idioms are semantically simple but grammatically complex (i.e. semantic units 'to kick the bucket, casser sa pipe'). The statistical analysis of collocations by Smadja (1993), or the criteria for idiomaticity set down by Fernando (1996), for example, are also based on purely semantic and syntactic grounds. Idioms, collocations and the other terms we mentioned in the Introduction often appear to be ad hoc lexical items with little relation to the rest of the language system (or to each other), and have often been seen as an interesting but marginal topic in grammatical theory. Fernando (1996) makes the point that while idioms have been widely analyzed in terms of their syntactic transformability, they are seen by definition as marginal to the general principles of syntax, whereas collocations are seen by grammarians as merely syntactic restraints (most usually on verbs, as in the principle of lexical projection). Furthermore, the relation between different types of expression has been obscured, and collocations are at times presented as subcategories of idioms (as in Fernando 1996) or at other times the other way round (as in Gross 1996). Idioms and collocations are seldom considered in terms of the 'norm' or varieties of language. For example, collocations such as to acknowledge receipt/accuser reception are presented as equivalent and 'bound collocations' involving restricted lexical items. However, while the English form is generally recognised, not all French speakers recognise accuser reception, and it appears that du beurre rand is rather technical and belongs to the category of LSP collocations ('specialised' collocations, a concept in the field of terminology). We return to this fundamental mismatch and notion of 'typicality' below. Lexical and semantic properties or a sense of 'uniqueness' are not the only defining features of these expressions. Expressions, as implied by Fillmore et al. have a life of their own in the language, and can survive even when truncated or reformulated. While the two grammatically similar expressions (3) 'I have had it' and (4) 'I have done it1 have the same grammatical structure, only one of these is recognised as an expression as such: (3) 'I have had it' can be taken to mean 'I have had enough', and this intention can not be translated literally into another language with the same effect, so not for example ?Je I'ai eu, but £a me suffit or J'en ai marre

224 Langue and Parole in Synchrony and Diachrony or J'en ai par dessus la tete. Example (3) specifically demonstrates that even formulations composed of grammatical items can be idiomatic as Mochet (1997) shows for collocations involving the French word qa (e.g. qa va, qay est, on nefait que qa, il n'y a que qa afaire...). But example (3) is also part of a longer expression 'I have had it up to here1 (again, a long sequence of closed class items), although native speakers do not have to access the whole expression to realise its rhetorical potential. Clearly in order to have this effect 'I have had it' can not be changed radically in terms of its word order or vocabulary (although the longer version can exist in various truncated parts: 'I have had it', 'I've had it', and (accompanied by an appropriate gesture) 'Up to here'). But this is not the most salient feature of the idiomatic expression. The main difference between (3) and (4) is that utterance (3) has a conventional rhetorical meaning. The move from word sequence (as a sentence) to rhetorical unit (as an utterance) is a central tenet of speech act theory, and certain linguists have claimed that this property, sometimes termed idiomaticity, is more central to a concept of native-like language use than the principles of grammatical competence proposed in mainstream linguistics (proponents include Yorio 1980, Pawley and Syder 1985, Sinclair 1991, Makkai 1992). Sinclair's idiom principle, for example, posits that language is in constant flux (synchronically and diachronically), in a cycle between the compositional 'open choice' of single words, to the automatic and 'idiomatic' use of whole expressions. While traditional accounts of idiom concentrate on semantic transparency or lexico-syntactic variation, others have explored the role of idioms in discourse. Makkai (1972), for example, emphasized the distinction between lexemes (compounds with predictable semantics, such as fly °fftne handle) and sememes (expressions with some rhetorical force, such as not a mouse stirred). The correct interpretation of either (3) or (4) above, equally depends on the extent to which they obey the general Gricean principles of conversation. As Moon (1992) points out, when utterances such as (3) or (4) appear to contravene the principles of relevance, the reader or interlocutor is justified in searching for an alternative interpretation. This shift in emphasis has the advantage of making the concept of idiom less categorical. It means that sentence (3) is 'typically' interpreted as 'I have had enough' unless the literal sense 'I have had it' may make sense in context (for example as a response to (3a) Have you had your measles injection?). Utterance (4) 'I have done it' is typically interpreted as literal, if no relevant interpretation is forthcoming; and indeed it is difficult to invent a context in which (2) may be interpreted as a rhetorical utterance with some indirect meaning, and which is not the response to the question (2a) 'Have you done it?'. The idea of 'authenticity1 and 'naturalness', as with 'typicality' is a principle enshrined by empirical linguists such as Sinclair, and we return to them below. According to Moon (1992), idioms play a vital role in encoding modality not only as potential speech acts, but as alternative and marked formulations in a system of choices of expression. For example, 'to walk slowly' can be encoded subjectively as 'to walk at a snail's pace', where the use of the idiom can be interpreted as an additional, subjective evaluation of the proposition. For Moon, the paradigmatic choice of expression by an idiom as opposed to a more literal expression always implies some rhetorical force, and this explains the large number of idioms used as euphemisms or intensifying expressions (one thinks here of idioms for taboo subjects such as death to shuffle off the mortal coil / manger son bulletin de naissance, emotional states to live it up /peter lefeu, relative success to bark up the wrong tree /faire fausse route and conversational gambits, do you come here often? / vous habitez chez

English and French Phraseology 225 vos parents?). Moon (1992) and Fernando (1996) further classify idioms according to Halliday's three 'metafunctions' (expressions which convey (a) ideational or conceptual information such as down in the dumps, broyer du noir, (b) interpersonal or dialogic information as in the ironic expression cause toujours, tu m'interesses ('tell me about it') and (c) textual or relational information at the end of the day, en fin de compte). From this perspective, dictums, cliches and 'turns of phrase' can be seen to be archetypal idioms. Dictums and proverbs differ from other expressions in that while they share the same complex semantics of idioms, they are often seen as purely rhetorical devices where their function is to exhort or provide a metacomment (more haste less speed meaning roughly 'take your time' or il faut semer le ban grain meaning roughly 'spread the good word'). Cliches in turn have negative pejorative connotations attached to their context of utterance, and are often avoided or reformulated sometimes for humourous effect (e.g. Worse things happen at sea, Jamais deux sans trois). Similarly, proverbs have a marked rhetorical function of'advice'. Although both Moon and Fernando point out the rhetorical role of idioms, they nevertheless stick to the traditional criteria for inclusion into the category: syntactic and semantic uniqueness. We would argue that it is equally valid to see rhetorical function and pragmatic force as determining factors in the classification of idioms, although this point can only be clarified in the light of our discussion of the related concept of collocation. For the moment, it is sufficient to point out that traditional accounts using syntactic and semantic criteria (and even more radical accounts, such as Makkai's) fail to include as 'idiomatic' such expressions as 3) 'I have had it' and those suggested by Mochet such as gay est.

COLLOCATIONS AND COLLOCABILITY Collocations (such as ask a question, poser une question, high winds, vents forts, on foot, a pied) are similar to idioms in that they involve relatively fixed sequences of words, but differ in that they are not recognised culturally or stylistically as expressions in themselves. We stated above that some linguists prefer to distinguish collocations and idioms on syntactic and semantic grounds. According to Cruse (1982), Benson et al. (1986:252) and others, collocations are syntactic units which can be broken down intuitively into smaller recognisable and independent semantic units (ask + a question, ask + the price). Other linguists refuse to distinguish between idioms and collocations, on the grounds that they often see one form as a subordinate category of another (e.g. Moon 1992, Fernando 1996, Gross 1996). Van der Wouden (1997), for example, states: "...I will use the term collocation as the most general term to refer to all types of fixed combinations of lexical items; in this view, idioms are a special subclass of collocations, to wit, those collocations with a non-compositional, or opaque semantics." (van der Wouden 1997: 9) Van der Wouden does point out that this entails problems. He cites the example of commonly considered collocations such as a murder of crows which happen to be opaque (i.e. interpretable as 'the slaughter of crows' and thus idiomatic). Similarly, the formulation ask for money is considered to be a collocation, although it is not completely compositional. The expression can not be broken down further than: ask for + money, so that 'for' appears to be stuck, morpheme-like, to the verb.

226 Langue and Parole in Synchrony and Diachrony By seeing idioms as essentially 'marked' expressions and collocations as 'unmarked' or normal means of expressing a concept, we are trying to make a distinction that is not categorical or binary and which lends itself to the notion of a continuum. Very common collocations such as Wierzbicka's examples of prepositional phrases (in April, on Thursday, at ten o'clock}, are clearly unique formulations in that the prepositions are obligatory for each formulation, but are also unmarked, standard ways of expressing those concepts. Bound collocations are a little more unusual (blond hair, nez aquilin) and yet these represent the preferred way of saying things in general discourse. To take Moon's (1992) examples, out of the blue, to call the shots, foot the bill: all of these are of course semantically opaque, but they are also marked forms of more prosaic formulations, namely: 'unexpected', 'to take command', 'to pay the bill'. These phrases are idioms, because they bring some rhetorical force to the basic expression (usually by the use of explicit metaphors: the first two expressions increase the intensity of the expression, while to foot the bill also implies a reluctance to pay). At times it may also be the case that there is a cluster of related core statements with no central phrase (such as finally, in summary, at last, enfin, en somme, pour terminer) which coexist with more idiomatic expressions ('at the end of the day, when all is said and done, all's well that ends well, en fin de compte, tout comptefait, tout est bien quifinit bien'). It is also perhaps worth noting that collocations appear to be neutral in terms of style, whereas idioms can be seen as at times inappropriate in terms of formality (e.g. randy devil, chaud lapin). We have also discussed in the previous section the possibility of analysing as idioms for the same reasons more 'transparent' expressions such as I've had it up to here - J'en ai eu assez -1 have had enough, Here we go again - Nous y voila = "We must start from the beginning", It's fine by me, £a me va tres bien = I agree. It should be noted that we intend these judgements to be relative: the 'unmarked' forms at times coincide with even less marked forms, and will certainly change in status from one discourse or register to another. In the following discussion, we attempt to establish the extent to which the concept of collocation can be applied to various features of language, and assume that all of the examples of collocation cited are unmarked in the general language, although we argue later that the norm will change according to context (thus scientific formulations will appear marked in general contexts, and informal forms will appear marked in formal registers). In addition, while some linguists see collocation as a rather restricted category (perhaps extending only to lexical compounds and formulations such as addled brains, le cerveau fele), we argue below that collocation extends to grammatical items as well as grammatical categories. The 'strong hypothesis' of collocation, especially stated by Sinclair (1991), is that every syntagm is a collocation, including even those formulations which display 'weak' collocational restrictions (such as he forgot his keys, il a oublie ses cles) and can be seen to enter into a default schema or colligation (S V O). In fact, Sinclair predicts that these so-called open expressions are more likely to occur in running text than canonical or stereotypical idioms (such as it's raining cats and dogs). According to this view, collocations are not seen as 'items' or units in the traditional sense, but underlying patterns of lexical attraction, a concept evoked by Mel'chuk's (1984) 'lexical functions'. Firth (1957) promoted the concept of collocation in order to relate the combinatory features of words to the rest of the language system. This idea was pursued by his students who were later to develop the systemic model of language (Halliday 1985 and Sinclair 1991). As Moon

English and French Phraseology 227 (1992), Fernando (1996) and Gross (1996) point out, collocation is simply a restriction of expression. For example, 'to bake a cake' (an unmarked expression) and 'to curry favour1 (a marked expression, 'to seek to gain favour') are verbal phrases with varying degrees of lexical restriction ('bake' is relatively free as a verb, 'curry' as a verb is extremely limited). Collocations are important to the contrastive analysis of languages, since they reveal fundamental mismatches between lexical systems, largely in relation to these differing ranges of lexical restriction. The English and French expressions to hammer a nail, enfoncer un clou are 'free1 in that both the verbs and their complements can be used with other words. However, to hammer and enfoncer are also used with different sets of collocates in the rest of the language system and the French verb, for example, differs markedly from other English equivalents: enfoncer la porte (smash a door down), enfoncer un bouchon (to push a cork in tight), enfoncer un poignard (to plunge a dagger in). Similarly the expression to jog one's memory is relatively restricted in English as there are few other possible complements of the verb jog. The French equivalent rafraichir la memoire is not as restricted (rafraichir simply signifying 'to refresh'). Jog one's memory and curry favour are therefore traditionally known as bound collocations (and are thus more idiom-like), whereas rafraichir la memoire and bake a cake are said to be 'free' collocations. While purely lexical collocations such as verb + noun, adjective + noun etc. have long been recognised, especially in the fields of terminology and lexicography, it has only recently been possible to gauge the extent to which collocation has a role to play in the grammatical system with the advent of corpus linguistics. The computational analysis of text corpora has enabled linguists to search very large text archives systematically, and according to Stubbs (1996) the computer has afforded linguistics the same degree of data processing potential as the telescope did for astronomy. There are two assumed advantages of computer-based corpus analysis: (i) as with the astronomer, the linguist can test theoretical 'armchair' hypotheses by examining authentic data, and (ii) the size of the data base can provide insights into language that had not been previously envisaged. In particular, corpus analysis has shifted the emphasis in lexicological studies away from the analysis of idioms and the transformational or semantic properties of fixed expressions, towards the analysis of collocations and the distribution patterns of stereotypical phrases. For example, dictionaries now rely on corpus evidence, not only for the existence of words and phrases but for their use and distribution patterns (Mel'chuk 1984, Sinclair 1991, Correard and Grundy 1994). In addition, corpus evidence on the distribution of idioms suggests that idioms are much less widespread and more variable than previously thought. Moon (1987) has found that of 2265 idioms (including a mixed category of metaphors) identified in the 323 million word Bank of English (held at Birmingham University, England), 47% occur less than once per 4 million words. Of all the idioms examined, only 135 occur more than twice per million words (among these items Moon includes out of the blue, call the shots, foot the bill). This can be compared with the collocation of course, which occurs over 240 times per million words. Moon concludes that 'pure' idioms are somewhat marginal in nature, but are likely to be reformulated for stylistic effect (thus to be a penny short of a sixpence 'to be mentally deficient1 is virtually always reformulated, e.g. to be a trunk short of a tree). The original idiom becomes obscured, and all that is left is a framework: to be an Xshort ofaY.

228 Langue and Parole in Synchrony and Diachrony On the basis of such corpus evidence, Sinclair and his co-workers (Moon 1987, Renouf and Sinclair 1991, Francis 1993, Fernando 1996, Hunston and Francis 1998 inter alia) have demonstrated that collocations are more systematically organised in language than had previously been thought and have argued that collocation is more central to the 'idiom principle' than traditional idioms themselves. Instead of seeing collocations such as rancid butter, du beurre ranee as items, the collocation has been extended to a more abstract notion of the lexico-syntactic sequence. For example, Sinclair (1991) found systematic patterns of cooccurrence for prepositional verbs such as to set in. To set in typically has negative NP subjects: [bad weather, disease, depression, gangrene] + sets in. In French, we find a similar negative semantic set emerges for essuyer + [une defaite, un affront, une crise, des pertes] ('to undergo, to soak up' + 'defeat, an affront, crisis, losses' etc.). The negative semantic context of both of these terms is known as 'semantic prosody'. A semantic prosody is the net effect of an accumulation of collocations, and the lexico-grammar of the expression is inseparable from the semantics of its constituents. Any new constituents are interpreted in the light of the established collocational pattern (thus The Tory Party had set in has the intended implication that Tory Party is a negative item). New elements in the prosody can be seen to integrate but also to impregnate the pattern: thus collocations provide a framework for gradual language change. The idea that idiomatic expressions carry with them a skeletal grammatical structure together with the notion that collocation extends beyond individual collocates to a more abstract semantic prosody leads us to the concept of the 'collocational framework' (Renouf and Sinclair 1991): an interrupted sequence of grammatical items (such as anXof, the X-er the Y-er) where the gap is filled by a restricted set of possible lexical words. In their 'pattern grammar' Hunston and Francis (1998) have similarly posited that most grammatical items become fixed to specific schemata and that these serve as redundant frames for the intervening lexical items, while the intervening lexical items often belong to restricted semantic sets. Thus the frame NP is X-ed as being only allows verbs with similar semantics such as considered, viewed, seen, while the collocational framework the X that NP has/have to V only permits sentiments as the main metalanguage noun (X): the wish, the desire, the need (Winter 1996) and is usually followed by verbs such as to succeed, to enjoy, to be loved (etc.). Hunston and Francis point out that in the potential patterns they examine in the corpus, many are 'fulfilled' by a small number of probable formulations. Thus N in N is usually increase in N, N into N is usually insight into N and so on. If the principle of collocational frameworks is applied to longer stretches than phrases and sentences, it should be possible to arrive at some meaningful reading of a text with no lexical items available (as in a cloze-test), or to interpret foreign text, or texts where the lexical items are obscured by nonsense words (the 'Jabberwocky' phenomenon, described by Hoey 1991). In the following extracts, for example, it should be difficult to guess many of the missing lexical items, but it is more likely that the reader will be able to assign a genre or texttype label to the extract (each lexical item has been replaced by one X, the morphology has been left more or less intact): (5) X the Xs with a X X X. X the X with X and X to X, and X each X in it. X the X, X with a X X or X and X over each X. X the Xs with Xs, X on Xly and X in X X until Xly Xed.

English and French Phraseology 229 (6) X. An X of X X which has been Xen off or Xed from the X of an X or from a X X, and is X in the X. When X enters the X, the X is Xed up by the X, and a X of the X is easily Xen off and Xs away. A X X is X in X; a X from an X X is X in X, often very X, and is X of the X. (7) Most X and X Xed Xs X X Xs and X Xs have Xed that X might be Xed by the X of Xs such as Xs that X the X of these X Xs. The X Xs Xed were X and X, but X has Xed their X in X. A X X of X X Xs has been Xed, among which X is Xly X since it is Xly Xed by X Xs in X. The native or fluent speaker of English should be able to identify that (5) is some form of instructional text, (6) is a definitional text and (7) a technical introduction [The original texts are presented in the Appendix]. Other genres, such as narrative and persuasion are equally identifiable, although it is usually impossible to guess the degree of lexical specificity of the texts. Once these extracts have been checked against the full originals in the Appendix, it becomes clear that certain clues (such as lexical repetition of items) are also missing, and play an important role in the formulation of these texts (Hoey's original point). What is interesting from the collocational point of view, however is that the grammatical items provide a series of coherent links, almost establishing a rhythm within each extract, and they allow the reader to predict certain key phrases. For example, in (5), X the Xs with a XXX, the first X must be grammatically an imperative verb, i.e. we must obtain the pattern grammar: V the N with a (Adj. or N) N. In (6) similarly, we interpret the final two missing items as prepositional verbs which must have to do with breaking and splitting off: and a X of the X is easily Xen off and Xs away. The rest of the text can then be reconstructed on this basis, and of course it can be seen that given one or two lexical cues, it would be possible to build a coherent text as one expression leads to another. The final text extract (7) is highly technical: so much so that having access to the lexical items themselves would hardly make much difference to the uninitiated. It is the case, however that non-specialists can read very specialist scientific prose (as in (7)) as though the lexical items were obscured in this way, and we are still able to build coherent interpretations on the basis of recognisable collocational frameworks. Most of the typical grammatical features of scientific text can be seen in this short extract: defining relative clauses, hedging (use of 'might'), complex nominals, passives, and so these together with a sense of some collocational frameworks lead the reader to impose a coherent reading on the text. To summarise, although collocations and corpus evidence have mainly been exploited in lexicography, Sinclair and his colleagues have put forward a grammatical theory of collocation which attempts to reassign the relationship between lexis and grammar in the language. Francis conceives of collocational frameworks as integral to utterances: "As communicators we do not proceed by selecting syntactic structures and independently choosing lexis to slot into them. Instead, we have concepts to convey and communicative choices to make which require central lexical items, and these choices find themselves syntactic structures in which they can be said comfortably and grammatically." (Francis 1993:142). Furthermore, corpus evidence has been able to challenge the general observation that grammatical items do not have general collocational properties. Even Halliday and Hasan (ironically, having proposed a theory of lexico-grammar) at one time claimed that a

230 Langue and Parole in Synchrony and Diachrony fundamental property of grammatical items was that they have few collocational restrictions and have little to contribute to the cohesion of text (1976:290). Yet it is possible to demonstrate that even grammatical classes such as the preposition have highly idiosyncratic collocational properties, especially for very high frequency items such as of, which has often been seen as atypical. Again, Francis advances this hypothesis: "If we take any one of a huge range of the more frequent words in English, and examine its citations en masse, it will emerge that it, too, has a unique grammatical profile, which certainly cannot be encapsulated by calling the word in question an adjective or a noun or a preposition." (Francis 1993: 147). Halliday points out that no one feature can characterise a register, and that a register is simply a set of statistically probable features: "... In fact lexis and grammar are not different phenomena; they are the same phenomenon looked at from different ends. There is no reason therefore to reject the concept of the overall probability of terms in grammatical systems, on the grounds of register variation. On the contrary; it is the probabilistic model of lexicogrammar that enables us to explain register variation. Register variation can be defined as the skewing of (some of) these overall probabilities, in the environment of some specific configuration of field, tenor and mode. It is variation in the tendency to select certain meanings than others, realising variation in the situation type." (Halliday 1991 :57) Having raised some of the implications of a collocational approach to language, we can now turn to some of our own corpus data, and while we can not hope to establish general phraseological differences in French and English, then at least we may show ways in which a phraseological account could be used to describe differences in the general language and specific genres.

PHRASEOLOGY AND GENRE In the cloze-texts above, we hypothesised that a text-type or genre can be identified solely by recognising key collocational frameworks. I have previously described the collocational behaviour of grammatical items in science writing, with the suggestion that grammatical items have radically different behaviour in different registers, even in different sub-genres (Gledhill 1995). The data discussed in Gledhill 1995 show that grammatical items are not equally distributed across the language, and that the collocational patterns of grammatical items correspond systematically to register or text type, as Biber has argued from the perspective of more general grammatical categories (1986). Since the emphasis has until very recently been on the phraseology of the general language, very little work has been done on the comparison of collocational patterns between more specialised language varieties, especially those in different languages, and so we set out here some sample analyses to demonstrate the principles of our methodology. As mentioned above, for enfoncer and to hammer, the differences between cognate terms in different languages can be particularly unpredictable. We set out to compare similar patterns from our English language corpus of scientific texts (the Pharmaceutical Sciences Corpus, PSC

English and French Phraseology 231 500 000 words) and a recently designed French corpus (Corpus de 1'Institut Pasteur, CIP 250 000 words). One reason for analysing scientific prose lies in the fact that science writing is highly devoid of idioms of the traditional kind, but is rich in metaphor and collocational restrictions. Science writing also has the advantage of belonging to a community of writers who are highly aware of the power and pitfalls of terminology and written expression (Gledhill 1997). From both corpora, it was possible to identify phraseologies on the basis of simple computer-generated concordances (we underline the lexical collocations and artificially limit the number of concordances to five examples per word). To distinguish between demontrer I demonstrate, for example: CIP) <demontrer> nous avons pu demontrer 1'existence d'autoanticorps ces resultats demontrent P existence de competition cellulaire sa decouverte a permis de demontrer 1'existence d'une nouvelle famille de genes nos experiences ont demontre que plusieurs mutations de cx32... entrainent une perte totale de fonction les resultats de cette etude ont permis de demontrer les proprietes hypolipidemiaires des huiles PSC) <demonstrate> the present study failed to demonstrate a sustained cell proliferation we could in no case demonstrate expression of the papillomavirus the high optical absorption spectra demonstrated that HUM does not directly decay the fact that we cannot demonstrate this change might be due to insufficient sensitivity of our method we have been unable to demonstrate methylene chloride adduction to heptocyte DNA Both verbs split very distinctly in terms of their lexico-grammar and semantic prosodies. In the English corpus, as can be seen, 'demonstrate' is almost uniquely used in negative contexts, usually where the researchers failed to demonstrate some spread of data. On the other hand, the French use of 'demontrer1 shows a pattern with a strong lexical collocation: demontrer + evidence. Since we are dealing with a highly specialized form of writing (the research article in the biomedical sciences), negative data and failure are not perceived as necessarily bad. The expression Tailed to demonstrate' is thus very frequent in this type of English discourse, but may not carry into other fields (such as linguistics). We suggested above that collocations and idioms exist in relation to or in competition with clusters of related expressions. In the case of 'demontrer', we can examine a number of related verbs in the French CIP corpus all relating to the biomedical preoccupation with empirical demonstrations of evidence (preciser 'to point out', etudier 'to study1, montrer 'to show', indiquer 'to indicate'). The collocational patterns for these words are also divergent: CIP <preciser> Afin de preciser le role de phenomenes d'amplification L'objectif des travaux menes ...est de preciser le potentiel offert par un arcomycete Plusieurs travaux ont permis de preciser le mode d'activation de ces proteines Nous precisons actuellement leur role dans la pathogenic analyses d'ARN ribosomique 16s ont ete realisees permertant de preciser les relations phylogeniques

232 Langue and Parole in Synchrony and Diachrony CIP <etudier> Un premier groupe etudie les bacteries fixatrices nous avons egalement etudie la regulation des genes nous avons etudie la reponse des lyphocytes .. .nous fournit des marqueurs interessants pour etudier la morphogenese le diabete insulo-dependant est etudie a travers la souche NOD CIP <montrer> nos resultats montrent I'importance de la structure .. .de la regulation 1'analyse genetique montre que sapl est essentiel a la vie de la cellule ces resultats... montrent de plus qu'il devrait etre possible de vacciner centre le cholera cette observation montre 1'importance de pax-6 dans la formation des yeux des recherches en region endemique montrent un polymorphisme important dans les .. .parasites CIP 1'ensemble des donnees cliniques indique qu'il s'agit d'une anomalie 1'ensemble de ces donnees indique done qu'il existe une regulation. 1'ensemble des donnees dont nous disposons indique que 1'anticorps selectionne un confomere les donnees epidemiologiques indiquent que ce type de cancer est tres frequent la comparaison de ces donnees genetiques ... a indique qu'un gene unique devrait etre en cause These words appear to have found their own small but significant collocational niche in French science writing, each with different degrees of collocational fixedness and grammatical role. 'Preciser' has no English equivalent expression in the PSC corpus, but is used in French to state the aims of the research institute (usually in terms of its main collocation, its role). 'Etudier' is used with technical biochemical entities, and with less abstract, research-oriented words than 'preciser', while 'montrer' collocates lexically with 'important' but emphasises specifically the importance of a new model. Conversely, 'indiquer' is almost exclusively introduced by 'the data set' (1'ensemble des donnees) and is followed by a projecting V-complement clause. It is interesting to return to the English technical corpus to examine the equivalent expressions, to see whether they occupy the same phraseological space. Perhaps predictably, some share the same niche as the French expressions while others do not. Of the two that do, 'show' appears to have the same role as 'montrer' in terms of reporting results, but does not have the same collocates, and the same is also true of'indicate': PSC <show> (c.f. 'montrer') The studies reproducing elevated TNFa induction showed no correlation HPC was shown to be topically active It was shown to inhibit 12-0-tetrachloryl... compounds PHEPC does not show any get-to-liquid planar transition about 0 degrees C. These results show a dramatically reduced resistance to N,N-dimethylated antracyklines PSC (c.f. 'indiquer') This result may indicate that AJ-1 is a very distant exon Combined with present data, this would indicate that about 50% of the compound is present these findings indicate that it is extremely difficult to immobilize named human cells these results indicate that distinct metastasis is significantly associated Data from other investigators .. .may also indicate the occurrence of some microciculatory events

English and French Phraseology 233 We claim that the similarities between these expressions should be seen as significant evidence of a consistent discourse of science. Both specialist corpora involve texts by multiple authors, and texts on a wide number of issues (within the specialism of cancer research or biomedical sciences). Thus such similar phraseologies for what are known as 'semi technical' lexical items indicate an established way of writing which appears, somehow, to have been propagated within the discourse community. However, it is necessary to distinguish between phraseological systems which appear to become established in a genre or a small specialism, and the possible regularities of the general language. To what extent are these systematic patterns (within French and within English, not necessarily between the two) different to those of the general language? This question is unfortunately difficult to answer, not least because there is currently no French equivalent of the accessible Bank of English (ex-Cobuild corpus). We have built however a control corpus for the purposes of comparison from Le Monde (one million words). The following concordances, this time focusing on 'suggerer' (suggest), show that there are significant differences in French: CLM) <suggerer> j ournalistique: tes ". L'Academie de medecine suggere en troisieme lieu une revision des resp timisation. Le simple bon sens suggere d'agir le plus possible durant la perio La pratique mediatique actuelle suggere une autre reflexion sur elle-meme, ne s 6 ues d'un " Munich social". II suggere, desormais, 1'organisation d'un referendum 22 t qu'aujourd'hui. Les mesures suggerees ici pourraient permettre de redistribuer... CIP) <suggerer> scientifique: S/BvgA. Une observation recente suggere qu'il existe une autre voie de regulation Montevideo, Uruguay). Le modele suggere que la specificite anti-Tn est associee Ces resultats suggerent que 1'expression de CD26 joue un role important 75 1'infection. Cette observation suggere que le systeme immunitaire joue un role 96 gillus fumigatus. Ces resultats suggerent que plusieurs facteurs sont necessaires... Our choice of examples is of course selective (we have taken the most frequent patterns to show the typical collocates). However, journalistic 'suggerer' clearly requires nominal complements (semantically: political acts and decisions), whereas scientific 'suggerer1 overwhelmingly (there are virtually no exceptions in its 25 occurrences) requires active clause complements, placing the onus on the act of suggestion (suggestion is an important part of academic hedging or modality). Similar patterns can be seen for other cognate pairs (including nouns), lexical phrases (such as au cours de) and collocational frameworks (Gledhill 1997). From this research, the collocational patterns of grammatical items and lexical items appear to become more stable and fixed as we observe more specialized genres (leading to the notion of sublanguage) which suggests that the relationship between the periphery and the core of language can be reviewed: the core, the langue may be seen as determined by the idiosyncrasies of the periphery, parole. It would however be a mistake to conclude that the general effect of collocational studies emphasizes the repetitive, stereotypical nature of language and also the extreme conventionality and conformism of specific genres. The pressures to conform in speech and writing styles are

234 Langue and Parole in Synchrony and Diachrony well known in discourse communities, as Swales (1990) points out. But the collocational patterns we have been exploring should be seen as a backdrop on which novel writing and innovation are able to develop: clearly, by our own definition, everything that is not phraseological is not the 'preferred way of saying things'.

CONCLUSION While lexicologists use the term 'phraseology' to refer to lexical co-occurrences (Thoiron, Hartley), we refer to the phraseology of a word or single expression as its rhetorical effect or pragmatic application of use. In contrast, the term lexico-grammar, a central term in Hallidayan grammar (Halliday 1985) indicates the strict interrelationship between lexical form and syntactic formulation. This in turn allows us to distinguish the phraseology and lexico-grammar of an expression from its semantic prosody, its typical semantic context. Our motivation for revising the distinction between idiom and collocation lies in the recent development of corpus linguistics. By attempting to fix the applications of the terms 'phraseology, lexicogrammar and semantic prosody' in relation to each other, we envisage a model of language in which phraseology embodies a continuum of expressions from pragmatically marked forms (idioms) to pragmatically unmarked or normal expressions (collocations). Unlike mainstream models of idiomatic expressions therefore, we use discourse criteria to determine the idiomatic status of an expression. This model presupposes that there are two forms of expression: a norm and a variant. This view of idioms and collocations has the advantage of being able to review the status of an expression (unmarked / marked) according to its context. The underlying assumptions are that a norm can be established and that the speaker has available to him or her a variety of expressions, of which many can be identified as pragmatically marked. While we have had the space to enumerate only a small number of authentic corpus examples of collocation, we hope to have shown that collocational norms (and therefore phraseological systems) are dependent on the context of situation in which they are produced. Any concept of core language (a concept of langue that assumes that peripheral forms are marked or 'special' as in the term Languages for Special Purposes) appears to be incompatible with the fact that in any particular discourse, new norms are forged and become effectively the new core for that particular register or genre. We have seen that this new core can be particularly fixed and systematic, with unexpected symmetries or 'collocational shifts', and this suggests that the underlying collocational patterns of a particular discourse are more prevalent and sensitive than they appear on the surface. The fact that certain discourses may be devoid of traditional idioms (as in science writing or Moon's Bank of English) also suggests that rhetorical expressions are not absent, but that the new discourse has developed its own rhetorical devices.

English and French Phraseology 23 5

REFERENCES Benson. M., Benson., E. and Ilson R. (1986) The Lexicographic Description of English. John Benjamins, London. Biber D. (1986) Variation across Speech and Writing. Cambridge University Press, Cambridge. Clark E. (1985) The acquisition of Romance, with special reference to French.' In D. I. Slobin (ed.) The Crosslinguistic study of language acquisition. Vol.1 The data. Hillsdale, NJ: Erlbaum. Correard M. H. and Grundy V. (eds.) (1994) Oxford Hachette French Dictionary. Oxford University Press. Cruse D. A.(1986) Lexical Semantics Cambridge University Press, Cambridge. Fernando C. (1996) Idioms and Idiomaticity. Oxford University Press, Oxford. Fillmore C. J., Kay P. and O'Connor M.C. (1988) 'Regularity and idiomaticity in grammatical constructions.1 in Language Vol. 64 :501-538 Firth J.R. (1957) Papers in Linguistics 1934-195I.Oxford University Press, Oxford. Francis G. (1993) 'A Corpus-driven approach to grammar'. In Baker M., Francis G. and Tognini-Bonelli E. (eds.) (1993) Text and Technology. :137-156. John Benjamins, Amsterdam. Gledhill C. (1995) "Collocation and genre analysis. The discourse function of collocation in cancer research abstracts and articles." In Zeitschrift fur Anglistik und Amerikanistik. Vol. 1/1995:1-26. Gledhill C. (1997) "Les collocations et la construction du savoir." In Anglais de Specialite. GERAS: Presses de 1'Universite Victor-Segalen, Bordeaux No. 15-18:85-104. Granger S. (1993) The International Corpus of Learner English.' in J. Aarts, P. de Haan and N. Oostdijk (eds.) English Language Corpora: Design, Analysis and Exploitation. :56-69. Rodopi, Amsterdam. Granger S. (1996a) 'Romance words in English: From history to pedagogy.' in J. Svartvik (ed.) Words. Alanqviar and Wikall: Stockholm. Granger S. (1996b) 'Prefabricated patterns of advanced EFL writing: collocations and lexical phrases.' in Cowie A. (ed.) Phraseology. Oxford University Press, Oxford. Gross G. (1996) Les Expressions figees enfranqais. Ophrys, Paris. Halliday M.A.K. (1985) Introduction to Functional Grammar London: Edward Arnold Halliday M.A.K. (1991) "Towards probabilistic interpretations." in E. Ventola (ed.) 1991 Functional and Systemic Linguistics: Approaches and Uses :39-61. Mouton de Gruyter, Den Haag. Halliday M. A. K. (1993) 'Quantitative studies and probabilities in grammar' in M. Hoey (ed.) 1993 Data, Description, Discourse: Papers on the English Language in Honour of John McH Sinclair. HarperCollins, London. Halliday M.A.K. and Hasan R. (1976) Cohesion in English Longman, London. Hoey M. (1991) Patterns of Lexis in Text Oxford: Oxford University Press Hunston S. and Francis G. (1998) 'Verbs observed: A Corpus-driven pedagogic grammar.1 in Applied Linguistics Vol. 19/1:45-72. Jacobi D. (1994) 'Lexique et reformulation intradiscursive dans les documents de vulgarisation scientifique.' In Candel D. (ed.) (1994) Frangais scientifique et technique et dictionnaires de langue. Didier Erudition, Paris.

236 Langue and Parole in Synchrony and Diachrony Makkai A. (1972) Idiom Structure and Idiomaticity. Mouton, Paris. Makkai A. (1992) 'Idiomaticity as the essence of language'. In Actes du XVeme congres international des linguistes. :361-365. C. I. L, Presses universitaires de 1'Universite laval, Quebec. McCarthy M. and Carter R. (1994) Language as Discourse. Longman, London. Mel'chuk I. (1984) Dictionnaire explicatif et combinatoire du francais contemporain. Recherches lexico-semantiques. Vol. 1. Presses de 1'Universite de Montreal, Montreal. Mochet M-A. (1997) "Expressions et groupements discursifs de 1'oral: questions d'inventaire et propositions didactiques.' Unpublished paper presented at Colloque Triangle. ENS Fontenay / St. Cloud, 7-8th March 1997. Moon R. (1987) The analysis of meaning'. In Sinclair J. McH. (ed.) (1987) Looking up: an Account of the Collins COBUILD Project. :86-103. Collins ELT, London. Moon R. (1992) There is reason in the roasting of eggs. A comparison of fixed expressions in native speaker dictionaries.' in Euralex '92 Proceedings Oxford University Press :493502 Nattinger J., and De Carrico R. (1992) Lexical Phrases and Language Teaching. Oxford University Press, Oxford. Pavel S. (1993) 'Neology and phraseology as terminology-in-the-making.' In H. B. Sonneveld and K. L. Loening (eds.) Terminology: Applications in Interdisciplinary Communication. :21-34. John Benjamins, Amsterdam. Pawley A. and Syder F. H. (1985) Two puzzles for linguistic theory: nativelike selection and nativelike fluency'. In J. C. Richards and R. W. Schmidt (eds.) (1989) Language and Communication. : 191-226. Longman, London. Picoche J. (1992) Precis de lexicologie francaise. L'etude et I'enseignement du vocabulaire. Nathan, Paris. Renouf A. and Sinclair J. McH. (1991) 'Collocational frameworks in English.' in K. Aijmer and B. Altenberg (eds.) (1991) English Corpus Linguistics : 128-144. Longman, London. Rey A. (1977) Le Lexique. Images et Modeles. Armand Colin, Paris. Schwegler A (1990) Analyticity and Syntheticity: A Diachronic Perspective with Special Reference to Romance Languages. Empirical Approaches to Language Typology, 6. Mouton de Gruyter, Berlin/New York. Sinclair J. McH. (1991) Corpus, Concordance, Collocation. Oxford University Press, Oxford. Sinclair J. McH., Fox G. and Hoey M. (eds.) (1993) Techniques of Description. Routledge, London. Smadja F. (1993) 'Retrieving collocations from text: Xtract.' in Computational Linguistics Vol. 19/1 :143-177. Stubbs M. (1996) Text and Corpus Analysis. Routledge, London. Svartvik J. (ed.) (1992) Directions in Corpus Linguistics. Proceedings of the Nobel Symposium 82, 4-8 August 1991, Stockholm. Swales J. (1990) Genre analysis. English in Academic and Research Settings. Cambridge University Press, Cambridge. Titone D. A. and Connie C. M. (1994) The comprehension of idiomatic expressions: Effects of predictability and literality.1 Journal of Experimental Psychology: Learning, Memory and Cognition, No. 20 :1126-1138 van der Wouden T. (1997) Negative Contexts. Collocation, Polarity and Multiple Negation. Routledge: London.

English and French Phraseology 237 Yorio C. (1980) 'Conventionalized language forms and the development of communicative competence.' TESOL Quarterly. Vol. 14/4:433-422. Wierzbicka A. (1993) 'Why do we say in April, on Thursday, at ten o'clock? In search of an explanation.' In Studies of Language 1993 Vol. 17/2:437-454. Winter E. (1996) 'Metalanguage nouns of clause relations', unpublished paper presented at Corpus Research: Sharing Interpretations 20th Sept. 1996, University of Birmingham.

APPENDIX Cloze Texts and Collocational Frameworks The reader is invited to attempt to find the lexical items (each replaced by one X) and attempt to assign a genre or text-type label to the following extracts. The full texts follo\v these amended extracts (reference numbers 5-7 refer to the examples discussed in the paper): (5) X the Xs with a X X X. X the X with X and X to X, and X each X in it. X the X, X with a X X or X and X over each X. X the Xs with Xs, X on Xly and X in X X until Xly Xed. (6) X. An X of X X which has been Xen off or Xed from the X of an X or from a X X, and is X in the X. When X enters the X, the X is Xed up by the X, and a X of the X is easily Xen off and Xs away. A X X is X in X; a X from an X X is X in X, often very X, and is X of the X. (7) Most X and X Xed Xs X X Xs and X Xs have Xed that X might be Xed by the X of Xs such as Xs that X the X of these X Xs. The X Xs Xed were X and X, but X has Xed their X in X. A X X of X X Xs have been Xed, among which X is Xly X since it is Xly Xed by X Xs in X. (5) Wipe the fillets with a clean dry cloth. Season the flour with salt and pepper to taste, and dip each fillet in it. Beat the egg, mix with a little milk or water and brush over each fillet. Coat the fillets with breadcrumbs, press on firmly and fry in hot fat until nicely browned. (Mrs Beeton's Cookery Book) (6) Iceberg. A mass of land ice which has been broken off or carved from the end of a glacier or from an ice barrier, and is afloat in the sea. When a glacier enters the sea, the ice is buoyed up by the water, and a portion of the glacier is easily broken off and floats away. A glacier berg is irregular in shape; a berg from an ice barrier is rectangular in shape, often very large, and is characteristic of the Antarctic. (W. G. Moore's Dictionary of Geography) (7) Most rodent and human xenografted tumours contain hypoxic cells and clinical studies have suggested that radiotherapy might be improved by the use of agents such as nitroimidazoles that increase the radiosensitivity of these hypoxic cells. The first agents evaluated were metronidazole and misonidazole, but neurotoxicity has limited their use in radiotherapy. A second generation of hypoxic cell sensitisers has been developed, among which pimonidazole (PIMO) is particularly interesting since it is preferentially accumulated by tumour cells in vitro. (Cancer research article introduction from the Pharmaceutical Sciences Corpus: Gledhill (1995)).


VII

DISCOURSE ANALYSIS


18

REGISTER-MOTIVATED VARIATION or TENSE AND MOOD IN GERMAN FINAL CLAUSES INTRODUCED BY DAMIT Sven-Gunnar Andersson, University ofGoteborg, Sweden

Abstract The variation pattern of tense and mood in German final clauses introduced by damit in past contexts is discussed on the basis of data from written texts and oral elicitation tests. The former by and large fit the traditional description, i.e. with the past indicative most strongly represented. The latter deviate to a great extent, since they contain both a high percentage of the present indicative and of the present and the past subjunctive with the past indicative almost lacking. This is interpreted as a recategorization in spoken language of the final clause with damit as a case of indirect speech.

INTRODUCTION According to standard grammars and handbooks on German usage there is a certain variability of tense and mood in damit-clauses in past context, as illustrated by (1) Er schickte den Brief schon anderthalb Wochen friiher ab, damit er wirklich rechtzeitig ankam/ankomme/ankame. The variation is, if at all, described as diachronically motivated, the past indicative (ankarri) being the younger and most common variant, the present or past subjunctive (ankomme, ankdme) the older and less common variants. Or, the choice is said to be made according to the general rules for the different forms, i.e. the indicative as the unmarked case, the present subjunctive for indirect speech and in the adhortative function, and the past subjunctive as a substitute for the present subjunctive in indirect speech or to signal counterfactual meaning. Neither of the descriptions gives an exhaustive account of the distributional pattern. This seems to involve choices of perspectives differing in written and spoken varieties of German. In this article the results of two investigations are presented and discussed together with some hypotheses about factors influencing the use of forms. The empirical material of the first investigation is taken from written texts in the computerized COSMAS corpus in Mannheim. The second investigation is based on elicitation tests with native speakers carried out in three different cities from North to South in the German language area (Berlin, Bamberg, Salzburg).

241

242 Langue and Parole in Synchrony and Diachrony In the elicitation material, the most striking difference from the written texts is the use of the present indicative in damit-clauses in past contexts, which corresponds to (la) Er schickte den Brief schon anderthalb Wochen friiher ab, damit er wirklich rechtzeitig ankommt. It should be noted that in (la) as well as in (1) the whole passage belongs to the past, so that the present tense does not refer to a point in time simultaneous with or later than the time of utterance for (la). Throughout the elicitation material, this use of the present indicative is by far the most common choice (cf. Table 2, below). There is, however, a clear sociolinguistic variation pattern within the elicitation material. In the written texts some few instances (7 out of 105) were found, too. It is an interesting fact that the use of the present indicative in past contexts is nowhere reflected in the literature on German damit-clauses. The results of the investigations give rise to some hypotheses about the direction of change in the use of tense and mood in damit-clauses in past contexts and, more generally, about the succession of stages in the long-term process by which the subjunctive is replaced by the indicative in German.

PREVIOUS RESEARCH ON TENSE AND MOOD IN GERMAN FINAL CLAUSES The only comprehensive study to this day was carried out by Flamig (1964). The main objective of his investigation was to work out the differences between two synchronic samples of written German from the periods 1760-1820 and 1900-1960, respectively. The sample from the latter period contains 920 instances of damit-clauses. Unfortunately, much important information about this material is lacking. There is no list of sources and no information on classification according to type of text. Flamig distinguishes between the following types: literary prose, scientific prose, popular prose, and newspapers. Nothing is, however, said about the relative proportion of these types of texts in the sample, neither is there an account of the number or percentage of the 920 dawzY-clauses for each type of text. It is nowhere stated how modally ambiguous verb-forms were treated - some examples give the impression that they were interpreted and included in the statistics. Except for literary prose there is no distinction made according to the tense-form or the time reference of the superordinate clause. For the indicative in the damzY-clause no information about tense is given. Consequently, many things are missing which are a necessary prerequisite for a comparison with the data presented below. The main objective of Flamig's study partly explains why this is so. His aim was to establish the relative distribution of the indicative, the present subjunctive and the past subjunctive in the two samples with special emphasis on the different types of texts, hi a way he did so, but for the reasons just mentioned his data cannot be controlled or used for a different patterning. The fact that the tense forms of the indicative are placed in only one indicative class is particularly negative for a comparison with the results presented in this article. However, comparisons are made wherever possible.

Tense and Mood in German Final Clauses 243 /X4M/r-CLAUSES IN WRITTEN TEXTS FROM THE COSMAS CORPUS For the investigation of tense and mood form in damit-clauses in past contexts, fictional and non-fictional texts from the period 1950-1990 were used. The finite form had to be unambiguous as regards tense and mood. The restriction to past contexts and unambiguous forms substantially narrowed the number of examples included, yielding a total of 105. The distribution of finite forms in the damit-clause of these 105 examples is shown in Table 1:1.

Table 1. Written Texts. Tense and mood in final clauses in past contexts 1:1 Sample from the COSMAS-corpus (n = 105) Fictional prose

Pres. Ind. 4 (= 6,3%)

Non-Fictional prose 3 (= 8%) 7 (=6,5%)

1:2

Past Ind. 42 (=65,6%)

Pres. Subj. 14 (=21,8%)

Past Subj. 4 (= 6,3%)

Total = 64 (=100%)

27 (= 66%) 69 (=66%)

7 (= 17%) 21 (=20%)

4 (= 9%) 8 (= 7,5%)

= 41 (= 100%) = 105 (= 100%)

Percentages for literary texts (COSMAS and Flamig) Indicative

Pres. Subj.

Past Subj.

Total

COSMAS-corpus (n = 64)

71,9%

21,8%

6,3%

=100%

Flamig 1964 (n = ?)

48%

36%

16%

=100%

For the category fictional prose it is possible to compare the percentages with Flamig's, although the number of instances in the latter case cannot be established. This comparison is made in Table 1:2. The correspondence is fairly good, the differences probably being due to the difference in time covered by the two samples (1950-1990 for the COSMAS corpus, 19001960 for Flamig's material). The correspondence with what is said in handbooks and grammars (cf. Introduction) is also good. As can be seen in table 1:1, however, there are seven instances of the present indicative in a damit-clause in a past context. These instances will be accounted for in detail in the next section, after the discussion on the use of the present indicative in the elicited material. Before turning to the analysis of examples from the COSMAS corpus, Flamig's discussion of the functional distinctions between the indicative and the two subjunctives in damit-clauses will be summarised. Examples from the COSMAS material will then be analysed with a view to seeing how they fit in with Flamig's description. Flamig starts by noting that a choice between the three modal categories is possible practically only in literary prose. In popular prose the indicative makes up 100% of the singular (the plural is not accounted for). For scientific prose the subjunctive is described as very rare. With the indicative, so Flamig says, no special modal meaning is conveyed in addition to that of the conjunction damit. This means that the content of the damit-clause is to be viewed as the intended consequence of the state of affairs expressed in the superordinate clause. The present subjunctive is used to mark the damit-clause as indirect speech if the whole passage is in indirect speech. Otherwise it has an adhortative function: In addition to the meaning 'intended consequence' of damit, the speaker views the content of the da/mY-clause as a demand (his own or someone else's) to be fulfilled by the clause subject. The latter statement should, as I see it,

244 Langue and Parole in Synchrony and Diachrony be slightly modified, cf. below. The past subjunctive could be a substitute for the present subjunctive in cases where the latter would coincide with the indicative. Otherwise it would convey a counterfactual meaning, especially when the whole passage is in the past subjunctive. In literary prose the writer would have a choice according to what nuance he wants to bring out. Different writers would decide differently, thus leaving room for idiolectally motivated differences of frequency. This description of Flamig's is now to be compared with some examples from the COSMAS material. Since things stand out more clearly in marked than in unmarked cases, examples with the subjunctive will be considered first. (2) Er war bei bester Laune, denn sein Beauftragter, den er nach Kirchbach geschickt hatte, damit er dort das Terrain erkunde, war mit allerlei interessanten Nachrichten heimgekommen. (Die Magd von Zellerhof) This is an example from trivial literature. The context gives full evidence that the intention expressed in the damit-clause was fulfilled before the time expressed by the first main clause (Er war bester Laune). The positive result of the mission is stated as the reason for the protagonist's high spirits. The mission and its result are both told in retrospect, as marked by the pluperfect. The use of the present subjunctive in the damit-clause is therefore to be taken as an indication that the narrator here chooses to focus the intention as a plan still to be carried out. He steps down from his position of omniscience and takes as his point of empathy the position of the protagonist when sending his assistant away. Also for (3) and (4) the same kind of interpretation seems appropriate: focusing the intention as a plan still to be carried out. (3) SchlieMch ziindete er verzweifelt ein Haus an und alarmierte sofort die Feuerwehr, damit niemand verletzt werde. (Der Spiegel) (4) Auch gab man ihm nach drei Wochen ein Bett, zwei weiBe Hemden und Tiicher, damit er im Unrat nicht verderbe. (Biographic ttber Schiller) Flamig describes the adhortative function of the present subjunctive in damit-clauses as a demand to be fulfilled by the damit-clause subject. This would certainly do for example (2), but it obviously only fits in with (3) and (4) on the condition that "fulfil a demand" is understood in a very abstract sense and not as a normal obligation. And a damit-clause without a subject would in no way meet the subject condition, as for instance (2a), which is perfectly possible: (2a)

Er hatte ihn nach Kirchbach geschickt, damit am Ort erkundet werde.

Instead I would propose to leave out of the definition by whom the demand is to be fulfilled: 'a demand to be fulfilled' or 'a plan still to be carried out' seems to go with the present subjunctive in all of the above-mentioned examples, i.e. (l)-(4). For the past subjunctive there is only one instance in the COSMAS material where it does not serve as a substitute for the present subjunctive, viz. (5) ,,Komm", sagt sie, ,,mach es dir bequem."[—]„ Komm", sagt sie, ,,setz dich." [—] ,,Walter", sagt sie, ,,warum setzt du dich nicht?" Mein Trotz, zu stehen -. [—]; sie raumte Biicher aus den Sesseln, damit ich mich setzen konnte. ,,Walter", fragt sie, ,,hast du Hunger?" (Max Frisch, HomoFaber)

Tense and Mood in German Final Clauses 245 Here Flamig's description seems to obtain. There is full justification from the context for the claim that the past subjunctive konnte signalises counterfactual meaning: the person stubbornly refuses to sit down. When focus is not on the intention as a demand to be fulfilled or as a plan still to be carried out, the content of the damit-clause is rather more in the background. Its function then seems merely to be to give information about the intended consequence of what is conveyed in the superordinate clause, just as Flamig claims for the use of the indicative. If the intended consequence quite obviously comes about, the indicative would be an appropriate choice, as in (6), taken from a newspaper text: (6) Hochst erschrocken bastelte der Mann aus Tubingen, damit er noch vor Einbruch der Dunkelheit sein Fahrzeug einigermafien beieinander hatte. Als der Gast aus Siidwestdeutschland endlich seine Fahrt fortsetzte, beunruhigte ihn ein merkwiirdiges Klappera im Kofferraum. (Die Welt) Here the content of the damit-clause functions as a link in a chain of situations following each other. The situation described in the damit-clause comes about with absolute certainty, because its realisation is necessarily the starting-point for what is told in the following sentence. It would seem natural in cases like (6) to choose the 'chain of events '-alternative, thus using the indicative, and not dwell on the first situation, which would focus the intention as something hopefully to be fulfilled and call for the present subjunctive in the damit-clause. In (2), however, where there is a similar chain of events in the retrospective, the present subjunctive occurs. Thus there is a principal freedom of choice of the point of empathy. In the following example the intended consequence could certainly be said to be counterfactual. Still, the indicative, and not the past subjunctive is used: (7) Nein, ich hatte da keinen Ehrgeiz. Oskar wollte nicht heilig gesprochen werden. Ein kleines privates Wunderchen wollte er, damit er horen und sehen konnte, damit ein fur allemal feststand, ob Oskar dafiir oder dagegen trommeln sollte, damit laut wurde, wer von den beiden Blauaugigen Eineiigen sich in Zukunft Jesus nennen durfte. (Giinter GraB, Die Blechtrommel) Being the unmarked mood, the indicative can principally occur also in contexts fitting the marked mood. In (7) the context conveys the counterfactual meaning without any extra marking on the verb put on top of it. Summing up the discussion of example (2)-(7), the characterisation given by Flamig seems by and large to be applicable to the COSMAS material.

DAMIT-CLAUSES IN THE ELICITATION TEST The elicitation test was carried out in 1995 by three 3rd-semester students of German from Goteborg studying in Berlin, Bamberg, and Salzburg. As North to South is by far the most important geographical parameter for German, the distribution of testing points seems adequate. In each city 13 sentences (listed at the end of this article) were presented, one at a time in random order, to 20 informants (male and female, with and without secondary education). These orally supplied a finite form of the verb given in the infinitive, and the answer was written down by the test leader. This procedure yielded a total of 752 processable answers with a non-ambiguous finite verb form.

246 Langue and Parole in Synchrony and Diachrony The test sentences were chosen according to certain hypotheses about tense and mood in final clauses in past contexts and according to general hypotheses about factors influencing the use of the indicative and the subjunctive, respectively. The following was considered: explicit fulfilment/non-fulfilment of the intention, informal vs. formal register, direct narration vs. indirect speech, first vs. third person singular. The first element in each pair would call for a higher frequency of the indicative, the second one would promote the use of the subjunctive. Since the test sentences and the embedding situations were by and large regarded as close to what Flamig called popular prose, where in his material only the indicative was attested, the outcome of the elicitation test was expected to be a very high percentage of the past indicative with the subjunctive represented, perhaps, in those test sentences expressly made up to elicit it (cf. above). The result, however, was a quite different pattern of distribution than the expected one: 1. The indicative and the subjunctive were used in almost equal proportions, 57% vs. 43%. 2. Of the 434 instances of the indicative, no less than 334, i.e. 77% of the indicatives and 44% of the total of processable verb forms, were in the present tense. The informants were evidently extremely reluctant to use the past indicative, thereby strongly deviating from the usage in written texts. Use of the present indicative in damit-clauses in past contexts, when the situation referred to does not obtain at the moment of utterance, is nowhere mentioned in grammatical literature. Flamig does not mention it and has no examples. It is to be recalled that in the COSMAS material only seven instances were found out of a total of 105. Of these, only three are absolutely clear cases of a past situation (cf. below). 3. The hypotheses about factors influencing the choice of mood were not supported by the elicitation test, the pattern of distribution being roughly the same for all test sentences regardless of content. 4. The distributional pattern that emerged from the elicitation material is quite obviously sociolinguistic in character. Details are given in table 2. The pattern is interpreted in the next section.

Tense and Mood in German Final Clauses 247 Table 2. Elicitation Test. Tense and mood in final clauses in past contexts. 13 sentences presented to informants in Berlin, Bamberg and Salzburg. (Abbreviations: m = male; f = female; +/- = with/without a secondary education; n = number of informants) Verb forms Informants m + (n= 18)

Pres. Ind

Past Ind

79

PresSubj.

PastSubj.

Total

(konnte(n) 18

Past Ind other than konnte(n) 7

87

36

= 227

34,8%

8%

3%

38,4%

15,8%

58

17

9

32

46

= 161

5,5%

20%

26,5%

= 100%

21

47

39

= 227

9,2%

20,7%

17,6%

= 100%

4

15

16

= 137

2,9%

11%

11,6%

= 100%

41

181

137

= 752

19%

= 100%

54,2%

11% f+(n=13)

38,8%

9,2% 14,7%

f-(n=18)

106

46,6%

13

5,9%

46,5%

15,1% m- (n=ll)

91

66,4%

11

8,1%

38,3%

22,6%

11%

(n = 60)

334

59 100

44% Informants

Present Indicative

100%

13% Past Indicative

318 24% 43% Subjunctive

Total

INTERPRETATION OF THE DISTRIBUTION OF FORMS IN THE ELICITED MATERIAL The relevant comparison is for reasons given below the one between the figures for the present indicative and those for the added results of the two subjunctives. It gives rise to the familiar picture of the sociolinguistic variation of tense and mood for past situations in indirect speech in informal spoken German of today, also well attested in modern literary texts reflecting informal speech. The use of the present indicative in indirect speech in past contexts means that the indicative tense of direct speech is taken over into indirect speech the same way as in Slavic

248 Langue and Parole in Synchrony and Diachrony languages, where it is, however, the norm. In the case of the damit-clause this would mean that the same tense and mood is used as the bearer of the intention would have used in the original situation, had he uttered his intention in speech. The interpretation as indirect speech is not only motivated by the distributional pattern for the present indicative versus the subjunctives but also by the quite plausible interpretation of the final clause as a case of indirect speech, taken in the wide sense in which this term is normally used. An intention is quite obviously to compare with the thought of a person and therefore the rendering of it can be formulated as indirect speech. This interpretation is, as I see it, also justified by the internal distribution and the high frequency of the subjunctives. The reverse sociolinguistic ordering of the subjunctive in relation to the present indicative is typical of indirect speech as is the higher proportion of the past subjunctive in lower sociolinguistic strata. For indirect speech it is hardly possible to uphold a functional difference between the two subjunctives, the present subjunctive being considered more formal and typical of written language, whereas the past subjunctive is more colloquial. The indirect speech hypothesis would explain the extremely low figures for the past indicative which deviate strongly from those of the written material (cf. table 1). The past indicative in the elicited material is, however, probably not to be interpreted as indirect speech but as the application of the same principle as in the written material. In indirect speech in past contexts the past indicative would correspond to the use of the past in English and Dutch as well as in Scandinavian and Romance languages. Such a use is rarely attested for Standard German of the northernmost area, where it is probably ultimately due to influence from Low German, which adheres to the same principle as Dutch and Scandinavian. In the elicited material, however, instances of the past indicative are registered not only from Berlin but also from Bamberg and Salzburg, which makes the indirect speech hypothesis implausible for this form. The low figures for the past indicative can be taken as an indication of how small the degree of overlapping is with the usage in written texts. It might be argued, theoretically, that the subjunctives of the elicited material are chosen not as exponents of indirect speech but as exponents of intentionality as in written texts, and that the high percentages compared to those for the subjunctives in the COSMAS material are due either to the difference in size between the two types of material (the COSMAS being rather small) or to an artificial rise in frequency, caused by the test situation. It is difficult to estimate the validity of such objections. However, it seems somewhat implausible that a marked case of written language (viz. the subjunctive as an exponent of intentionality) should be overused in the elicitation test when at the same time the colloquial present indicative is attested to such a high degree in all sociolinguistic informant strata of it. There is also no clear pattern for the choice of the subjunctive in the elicited material: neither in answers given by certain informants, nor in certain test sentences. The general picture fits the interpretation that for the informants in the elicitation situation the damit-clause was in the overwhelming majority of cases given the value of indirect speech with the present indicative or the two subjunctives as exponents of it.

OCCURRENCES OF THE PRESENT INDICATIVE FOR PAST SITUATIONS IN THE COSMAS MATERIAL There are in all seven instances in which the present indicative is used. In three of them the intention is quite clearly still valid at the time of text production and, consequently, does not refer to a past situation. One is ambiguous as to time reference:

Tense and Mood in German Final Clauses 249 (8) Und wenn sie ihm beim Robben zuriefen, daB er die Hacken einziehen sollte, fragte er ziemlich bescheiden: warum? Er wuBte aber warum. Damit der Feind ihm nicht den Knochel zerschieBt. Dann traten sie ihm mit den schweren Schuhen die Knochel flach, und er sagte: ach so. (Uwe Johnson, Das dritte Buch iiber Achim) Reference time of the damit-c\ause in (8) could be the actual situation described. But the proposition also expresses knowledge of general validity and (8) is therefore no clear case. The remaining three instances, all of them from the novel Homo Faber by Max Frisch (published 1957), are no doubt perfect parallels to those of the elicitation material, cf. (9) - (11): (9) Als ich in die Kabine zuruckkehrte, rasiert, so daB ich mich freier fuhlte, sicherer - ich vertrage es nicht, unrasiert zu sein - hatte er sich gestattet, meine Akten vom Boden aufzuheben, damit niemand drauf tritt, und uberreichte sie mir, seinerseits die Hoflichkeit in Person. (10) Ich brauchte nur daran zu denken - und es tippte plotzlich wie von selbst, im Gegenteil, ich muBte auf die Uhr sehen, damit mein Brief noch fertig wird, bis der Helikopter startet. Sein Motor lief bereits. (1 l)Es amusierte sie, aber das anderte nichts daran, daB Sabeth eigentlich verschwinden muBte. Ich saB, ich hatte ihren Unterarm gefaBt, damit sie nicht davonlauft. ,,Bitte", sagte sie, "bitte". The content of Homo Faber is in a very subjective way narrated by its main protagonist Walter Faber. Everything is told from his perspective. This has correlates at the level of expression which shows many parallels to the style of informal oral narrative. One example is the way tense and mood are treated.

SOME GENERAL REMARKS ON THE REPLACEMENT OF THE SUBJUNCTIVE BY THE INDICATIVE IN STANDARD GERMAN The use of the subjunctive in final clauses in German was originally a case of the use of the subjunctive in subordinate clauses whose content is in some way non-factual (imagined, hypothetical, matter of opinion or dispute, hoped for a.s.o.) Quite clearly, intention is part of non-factuality. The replacement of the subjunctive by the indicative in subordinate clauses of this kind has gone very far in present contexts, where in most cases the subjunctive can no longer be used. In damit-clauses in present contexts (not treated in this article) the present subjunctive is very rare today. In Flamig's literary texts from 1900-1960 it takes up 20% of an unknown number of damit-clauses in present contexts. In past contexts there seems to be a dichotomy between written texts and informal spoken language as shown above. With the former, the replacement process has not gone equally far as in present contexts. Replacement here means a shift from the subjunctive to the past indicative, that is to the unmarked indicative tense for past situations. This is a parallel to the shift to the unmarked indicative tense in present tense contexts. In informal spoken language, damit-clauses in past contexts seem to have been recategorized as a case of indirect speech with the tense and mood forms chosen accordingly. The dichotomy for past contexts fits in very well with a well-known difference of attitude between spoken and written language. In spontaneous speech there is as a rule an attitude of

250 Langue and Parole in Synchrony and Diachrony engagement, proximity and empathy towards the things referred to, whereas writing rather goes with an attitude of distance and analysis. In English linguistic literature this dichotomy is known as the involvementdetachment opposition (cf. e.g. Chafe 1982, Chafe & Danielewicz 1987, Biber 1988). It seems to obtain for the interpretation of damit-clauses as indirect speech (being involvement in somebody's thoughts) versus just stating the intended consequence of what is expressed in the governing clause. It also obtains for the difference in da/mY-clauses between the present indicative in spoken language and the past indicative in written texts. There are to my mind good reasons for viewing the difference between the present and the past tense, especially in languages without a subjunctive, but also in German, as fundamentally based on the dichotomy proximity/distance, where the time reference interpretation is the default case and not the fundamental category. "Proximity" is a near synonym of "involvement", whereas "distance" is a variety of "detachment", which brings the two pairs of concepts very close to each other. The present indicative having replaced the subjunctive in final clauses in present contexts and the dichotomy shown for past contexts are in my view another two indications of the importance of the proximity/distance opposition in the tense and mood system.

APPENDIX List of test sentences for the elicitation test (umkippen)

1.

Der Chemielehrer sagte dem Schiiler, daB er aufpassen musse, damit sein Glaskolben nicht .

(umkippen)

2.

Der Chemielehrer mahnte den Schiiler im strengen Ton, ja nur aufzupassen, damit sein Glaskolben nicht .

(konnen)

3.

Sie hatte den Braten ordentlich weggestellt, damit ihn der Hund nicht anruhren .

(fallen)

4.

Sie hatte das Kalbskotlett ordentlich weggestellt, damit es nicht dem Hund zum Opfer .

(behalten)

5.

Ich brachte meinen Wellensittich zur Grofimutter, damit sie ihn fur einige Zeit .

(behalten)

6.

Fritz brachte seinen Wellensittich zur Grofimutter, damit sie ihn fur einige Zeit .

(einschlafen)

7.

Das Kind konnte nicht einschlafen. Die Mutter las dem Kind ein Marchen vor, damit es , und es ist dann auch eingeschlafen.

(schreiben)

8.

Er schenkte ihr Briefpapier, damit sie ihm otter tat sie auch.

, und das

(schreiben)

9.

Er schenkte ihr Briefpapier, damit sie ihm otter

.

Tense and Mood in German Final Clauses 251 (schreiben)

10.

Er schenkte ihr Briefpapier, damit sie ihm 6'fter aber nichts geholfen.

,dashat

(einschlafen)

11.

Das Kind konnte nicht einschlafen. Die Mutter las dem Kind ein Marchen vor, damit es , aber das hat nichts geholfen.

(zugehen)

12.

Freilich gab es bestimmte vornehme Frauen, unter ihnen die Ra'tin Seideleben, die es gerade fur ihre Pflicht erachteten, bei diesen Gelegenheiten anwesend zu sein, damit es recht .

(zugehen)

13.

Na, klar hat's auch Frauen gegeben, die meinten, dabei sein zu mussen, damit es recht .

REFERENCES Biber, D. (1988) Variation Across Speech and Writing. Cambridge University Press, Cambridge. Chafe, W. L. (1982) Integration and involvement in speaking, writing and oral literature. In: Spoken and Written Language: Exploring Orality and Literacy (D. Tannen, ed.), 35-53. Ablex, Norwood. Chafe, W. L. and J. Danielewicz (1987) Properties of spoken and written language. In: Comprehending Oral and Written Language (R. Horowitz and S.J. Samuels, eds.), 83-113. Academic Press, San Diego. Flamig, W. (1964) Untersuchungen zum Finalsatz im Deutschen.(Synchronie und Diachronie). Sitzungsberichte der dt. Akademie der Wissenschaften zu Berlin. Klasse fur Sprachen, Literatur und Kunst. Jahrgang 1964, Nr. 5. Akademie-Verlag, Berlin.


19

MODALS, COMPARATIVES, AND NEGATION Igor Boguslavsky, Institute for Information Transmission Problems, Russian Academy of Sciences, Moscow, Russia

Abstract It is shown that a certain class of sentences with the modality of possibility can have two types of interpretation - a boundary and an interval one. Under the first interpretation, an extreme pole on the scale of possibilities (either maximum or minimum) is referred to. Under the second, a possibility is viewed irrespective of how it relates to other possibilities. The boundary interpretation is a result of an interaction of lexical semantics, communicative structure of the sentence and conversational maxims. This distinction is used to describe a well-known ambiguity of sentences with the comparative.

i. INTRODUCTION It was noticed long ago that comparative constructions with the English adverb less and its Dutch counterpart minder exhibit an interesting ambiguity. As far as I know, it was first noted by P. Seuren (1979) and later described by H. Rullmann (1994, 1995). So a sentence like (1) The helicopter was flying less high than a plane can fly. may be interpreted either as (2) 'The helicopter was flying at an altitude below the maximal altitude at which a plane can fly' or as (3) 'The helicopter was flying at an altitude below the minimal altitude at which a plane can fly' Alternative ways of conveying these two interpretations are as follows: (4) A plane can fly higher than the helicopter was flying. (5) A plane cannot fly as low as the helicopter was flying. The first reading can be referred to as the maximum interpretation and the second one as the minimum interpretation. Concrete discourse contexts favour either one or the other: (6) Because the helicopter was flying less high than a plane can fly, the jet fighter could easily attack it from above. (7) The jet fighter was trying to chase the helicopter, but because the helicopter was flying less high than a plane can fly, the jet fighter crashed into a building. 1

This study was supported by RBRF grants N 98-07-90072 and 99-06-80277. 253

254 Langue and Parole in Synchrony and Diachrony Another example (from Dutch) that brings out the same ambiguity is (8) (Seuren 1979): (8) Jan reed minder snel dan hij mocht. 'Jan drove less fast than he could' (8a) 'Jan could drive faster than he did' (8b) 'Jan could not drive as slowly as he did' Only under the second interpretation was Jan breaking the regulations. The same two types of interpretation can be found in Russian: (9) Ivan exal medlennee <s men 'shej skorostju>, chem bylo mozhno. 'Ivan drove slower than he could' (9a) 'Ivan could drive faster than he did' (9b) 'Ivan drove too slowly, he could not drive as slow as he did' Let us split the meaning of a comparative construction into three parts to explicitly convey the meaning of comparison proper and the entities which are compared. The sentence (1) can then be represented as (10): (10) a. 'the helicopter was flying at altitude A'; b. 'a plane can fly at altitude B'; c. 'A is less than B'. Hotze Rullmann (1994, 1995) attributed the ambiguity exhibited in (1) exclusively to the properties of the comparative. He claims that it is the comparative meaning that brings out the maximum and the minimum readings. In other words, he locates the ambiguity in (lOc). In this he follows an earlier proposal by P. Seuren (1979), although the theories of the comparative adopted by these authors are different. I have several objections to these descriptions. Suffice it to mention here only one. Both authors note that one of the interpretations, namely the minimal one, is only possible with 'less' but not with 'more'. This fact is crucial for their descriptions: they are designed in such a way as to predict this difference. However, one can find the minimal meaning even in the case of 'more', at least in Russian: cf. (1 la) and (1 Ib) that are clearly opposed. (1 la) Samolet letel vyshe, chem bylo mozhno. 'the plane was flying higher than was allowed' (lib) Samolet letel vyshe, chem mogby. 'the plane was flying higher than it could' (11 a) is naturally interpreted with the maximum reading: the plane exceeded the maximum permitted height. Sentence (lib) obviously exhibits the minimum reading. It is intended, for example, for a situation where it was desirable for the plane to fly as low as possible. The minimal altitude at which the plane could fly was H, but the pilot did not fly as low as he could, he was flying at a higher altitude. Therefore, an account that is based on the impossibility of the minimum interpretation for 'more' cannot be accepted as a general solution. In what follows a different description will be proposed. First, I will argue that the main source of the ambiguity in question is not located in the comparative but in the modal phrase, i.e. in (lOb). Second, I will show that the distinction should be drawn not between the

Modals, Comparatives, and Negation 255 maximum and minimum interpretations but between two other types of meaning: a BOUNDARY reading (which can be either maximal or minimal) and an INTERVAL reading (which has no bearing on any extreme). After that, I will discuss these data in the context of negation. Our discussion in this paper will be based on Russian data. Sometimes Russian sentences have direct English equivalents. Sometimes they do not. In the latter case English translations will serve as glosses not purporting to be idiomatic English.

2. THE MODAL PHRASE Let us begin with phrases containing the modality of possibility (deontic or epistemic) in isolation, i.e. beyond the context of comparison. The sentence The plane can arrive on schedule describes one possible state of affairs: the plane can arrive on schedule but it can also arrive earlier or later. The range of possible arrival times is not restricted, either from above or from below. In the following sentences this is not the case. (12) Etot kran mozhet podnjat' desjat' tonn. 'this crane can lift ten tons' (13) Etot samolet mozhet letet' na vysote desjat' kilometrov. 'this plane can fly at an altitude often kilometres' These sentences can be interpreted in two essentially different ways depending on the intonation and the communicative function.

2.1. The boundary interpretation If pronounced with neutral intonation, (12) - (13) are normally meant to answer questions of the type What is the lifting capacity of the crane? What altitude can the plane reach? The answers that (12) - (13) suggest evidently denote the MAXIMUM value on the scale of possibilities. The crane cannot lift more than ten tons and the plane cannot fly at an altitude that exceeds ten kilometres. On the other hand, it is not always the case that the value referred to really EXCEEDS all other possible values. Sometimes what is meant is not the maximum but the minimum value. In most situations only one of these is relevant, but one can easily find cases where the range of possible values is restricted on both sides. For example, a plane is characterized not only by its ceiling, i.e. the maximum altitude, but also by the minimum altitude at which it can fly. Anybody who knows at what height planes fly will interpret the sentence (14) Etot samolet mozhet letet' na vysote sto metrov 'this plane can fly at an altitude of one hundred metres'

256 Langue and Parole in Synchrony and Diachrony as referring to the minimal altitude: the plane cannot fly lower, though of course it can fly higher. Only a background knowledge of the range of possible heights at which a plane can fly gives us an opportunity to decide whether we are dealing with the maximum or minimum reading. The difference between them is not marked by linguistic means and the choice is made on purely extralinguistic grounds. Therefore, from the linguistic point of view, these readings should be merged together. There is only one meaning associated with the idea of a limit, an extreme point on a scale, and whether it is the upper limit or the lower limit is a matter of background knowledge and the extralinguistic world. In this respect the sentences in question are not linguistically ambiguous but pragmatically vague. This generalised maximum/minimum or BOUNDARY reading is linguistically opposed to another interpretation that appears in a different communicative situation and is expressed by a different intonation.

2.2. The interval interpretation If the modal verb in (12) - (13) is marked with a phrasal stress (cf. (15) - (16)), the communicative function is different: such sentences are normally used to answer the question whether or not the crane can lift as much as ten tons (or whether or not the plane can fly as high as at ten kilometres). (15) Etot kran -Imozhetpodnjat' desjat' tonn. 'this crane can lift ten tons and maybe even more' (16) Etotsamolet J-mozhet letet' na vysote desjat' kilometrov. 'this plane can fly at an altitude often kilometres and maybe even higher' This question is answered in the affirmative. The lifting capacity of the crane does ensure lifting ten tons. Maybe it can lift even more - the question of the maximum is not at issue. What is relevant is whether or not ten tons belong to the set, or interval, of possibilities. Therefore we will call this interpretation INTERVAL, as opposed to the BOUNDARY interpretation discussed above. To summarise, sentences with the possibility modality of moch' or mozhno type, a parametric verb and a quantitative noun phrase can have two types of interpretation. Under the first interpretation the quantitative noun phrase denotes an extreme pole on the scale of possibilities (either maximum or minimum). Under the second it is only considered as one of the possibilities, without any indication as to how it relates to other possibilities. The interpretations differ with respect to the communicative function of the sentence. The boundary interpretation is intended when the quantitative noun phrase is in the focus (= rheme) and the sentence is intended to answer the question "how much P?" The second - interval - reading

Modals, Comparatives, and Negation 257 implies a focus on the modal verb and the sentence rather answers the question "whether or not P?"

3. SOURCE OF THE DIFFERENCE When we encounter a multiple interpretation of a word or a phrase, the first question to be answered is what the source of the ambiguity is. To account for the difference between the interval and the boundary interpretations, one could certainly postulate two lexical meanings of the modal verb, but this decision is obviously counterintuitive. In our opinion, the lexical meaning of the verb is the same - that of mere possibility. The boundary reading comes as a result of certain communicative mechanisms that operate upon this meaning. The most important component of the explanation is the Gricean maxim of quantity. As is well known, it obliges the speaker to give as much information as he/she possibly can in order to meet the informational requirements of the addressee. A minor refinement of the maxim will be appropriate: it only applies to the focus of the sentence that actually carries the most informative load. When we utter sentence (12) with neutral intonation, literally we only claim that the ten-tons weight is possible among others, and do not claim anything else. But the communicative function of the sentence is to characterise the lifting capacity of the crane. If we chose to do so by indicating a possibility, then, according to the maxim of quantity, we would have to mention the most informative possibility. In this case it would be the maximum weight. This - maximum - interpretation is supported by the fact that all artefacts are produced to fulfil a certain function and are mostly characterized by some parameters that describe their ability to fulfil this function (speed, power, range, altitude, capacity, tonnage, etc.). Seen as characteristics of things, the values of these parameters are important as representing the EXTREME ability, and not just one of the possibilities. To better grasp the difference between a linguistic meaning and an interpretation provided by a pragmatic maxim, let us compare sentences (17) - (18): (17) The lifting capacity of the crane is ten tons. (18) [=(12)] The crane can lift ten tons. If we utter these sentences to refer to the crane capable of lifting as much as fifteen tons, sentence (17) will be simply false, while in the case of (18) it is a communicative failure due to violation of the quantity maxim. In a different pragmatic situation the most informative answer may be represented by the minimum rather than by the maximum (cf. (14)). It follows from the preceding discussion that the boundary meaning results from the interaction of three different sources: the lexical meaning of the modal verb ("one of the

258 Langue and Parole in Synchrony and Diachrony possibilities"); the communicative function of the sentence ("How much P?"); and the pragmatic maxim of quantity ("Give the most informative answer").

4. COMPARATIVES Let us come back to sentence (1). Our hypothesis is that its ambiguity results precisely from the two types of the modal clause interpretation we discussed above. Reading (3) is a clear instance of the boundary interpretation: the plane cannot fly as low as the helicopter was flying. The helicopter was flying below the minimal height at which the plane can fly safely. In reading (2) we are dealing with the interval interpretation: the height at which the helicopter was flying belongs to the range of possible heights of the plane. It should be emphasised that no reference is made to the maximum altitude of the plane, as was assumed at the beginning of the paper. The speaker says that some altitude exceeding the altitude of the helicopter is accessible for the plane. He does not refer to the upper limit of the plane flight. He does not even presuppose that such a limit exists. He only says that it is possible for a plane to fly at a higher altitude than that of the helicopter. It is easy to distinguish between two readings of the modal verb when it is in the main clause. The communicative structure of such a clause is clear-cut and the semantic difference in question is marked by the intonation. It becomes more difficult when the clause is in the subordinate position, as is the case in (1). Here the communicative and intonational distinctions are neutralised and hence the semantic difference is not marked.

5. MODALITY AND NEGATION It is interesting to discuss the boundary / interval meaning opposition in a broader context not related to modal words. Sentence (19) under a boundary interpretation (19) [=(12)] Etot kran mozhet podnjat' desjat' tonn. 'this crane can lift ten tons' ascribes a certain value (ten tons) to a certain parameter (lifting capacity of the crane). Let us compare this expression with two other, more prototypical ways of ascribing values to parameters. First of all, the parameter can be expressed by a noun, as in (20) - (21). Second, it can take a verbal form, as in (22). (20) Ves gruza raven desjati tonnam. 'the weight of the cargo is ten tons'

Modals, Comparatives, and Negation 259 (21) Gruzopodjemnost' krana ravna desjati tonnam. 'the lifting capacity of the crane is ten tons' (22) Gruz vesit desjat' tonn. 'the cargo weighs 10 tons' We will compare these sentences with respect to their behaviour under negation. If we want to negate a whole sentence, most often we will have to add the negative particle ne 'not' to the main predicate of the sentence. This is just what we see in (23) - (24). (23) Ves gruza ne raven desjati tonnam. 'the weight of the cargo is not ten tons' (24) Gruzopodjemnost' krana ne ravna desjati tonnam. 'the lifting capacity of the crane is not ten tons' As for sentence (22), the situation is more complicated. If we place the negative particle in front of the verb group, as we have just done for (20) - (21), the resulting sentence (25) will not have the meaning expected from the negative correlate of (22). Rather, (25) means 'the cargo weighs less than ten tons' and not 'the weight of the cargo is not ten tons'. (25) Gruz ne vesit desjati tonn. 'the cargo does not weigh 10 tons' On these grounds it is sometimes claimed that sentences of the (22) type have an idiomatic negation: 'less' instead of 'either more or less'. We argue that this is wrong. The picture is much more interesting. If we expect (25) to express the idea 'the weight of the cargo is not ten tons', i.e. 'either more or less than ten tons', it means that we consider it as the negative correlate of (22), which is just wrong. The general rule of negating sentences in Russian claims that the negative particle ne 'not' should be attached to the rhematic part of the sentence, be it a verb group or not. Very often the rheme is expressed by the main verb group, as is the case in (20) - (21). But it is not always so and, in particular, it is wrong for (22). The verb vesit 'weighs' that takes on the negation in (25) does not make part of the rheme in (22). Sentence (22) is normally intended to answer the question "How much does the cargo weigh?" Hence, it is the quantitative noun phrase desjat' tonn 'ten tons' that is the rheme in (22). If we follow the general rule and attach the negation to this group, we obtain a quite correct sentence (26) that is a regular negative correlate of (22). (26) Gruz vesit ne desjat' tonn. lit. 'the cargo weighs not ten tons' 'it is not ten tons that the cargo weighs' If (25) cannot be considered the negation of (22), then what is it? Is it simply a sentence containing the negative particle without an affirmative correlate, or is it still a regular negation of some other sentence? It is the latter possibility that is true.

260 Langue and Parole in Synchrony and Diachrony Actually, the negated verb vesit' 'to weigh' in (25) does not belong to the thematic part of the sentence, as it does in (22) and (26). Sentence (25) answers the question whether or not the weight of the cargo attains ten tons. In this case the negated verb group belongs to the rheme and the sentence turns out to be built in conformity with the general rule mentioned above. If sentence (25) has a corresponding affirmative counterpart, the latter must be obtainable by deleting the negative particle WHILE RETAINING THE RHEMATICITY OF THE VERB. As we have seen, (22) cannot be considered to be such an affirmative counterpart. In the communicative structure of sentences with parametric verbs (such as weigh (ten tons), last (two hours), equal (twenty), contain (three litres), etc.) these verbs are typically relegated to the thematic part. If one wants to shift the verb to the rheme, one has to mark it with a specific intonation pattern. This intonational modification of (22) results in (27) tDesjat' tonn gruz 4-vesit. 'the cargo does weigh at least ten tons', i.e. 'the cargo weighs ten tons and maybe even more' which turns out to be the true affirmative correlate of (25), both formally and semantically. Formally, (25) differs from (27) in that it contains the negative particle attached to the verb which constitutes the rheme of (27). Semantically, (25) is a standard negation of the meaning expressed in (27): (27) = 'the cargo weighs ten tons or more'; (25) = 'the cargo weighs less than ten tons'. Sentences (22) and (27) have different intonation patterns (neutral vs. marked), different communicative properties (vesit' belongs to the theme vs. to the rheme), different meaning ('ten tons' vs. 'ten tons or more') and different negative correlates ((26) vs. (25)). Furthermore, they have different interrogatives. Cf. dialogues (28) and (29): the questions are interrogative forms of (22) and (27), respectively. Both questions refer to the same cargo and still they are given opposite (negative and positive) answers. (28) - Etot gruz vesit desjat' ftonn? 'Is it the case that this cargo weighs ten tons?' - Net, bol 'she. 'No, more than that' (29) - Etot gruz desjat' tonn Tvesit? 'Does this cargo weigh as much as ten tons?' - Da, dazhe bol'she. 'Yes, even more' Now we are ready to go back to the modal sentence (19) and subject it to the negation test. Strange as it may seem, it cannot be negated at all! It should be remembered that (19) is considered under the boundary interpretation and hence is synonymous with (21). The latter is perfectly negated - cf. (24). But if we try to add the negation to (19), we obtain a sentence that does not carry the meaning expected. (30a) Etot kran ne mozhetpodnjat' desjat' tonn. 'this crane cannot lift ten tons' (3 Ob) Never no, chto etot kran mozhet podnj at' desjat' tonn. 'it is not the case that this crane can lift ten tons'

Modals, Comparatives, and Negation 261 On no account do (30a,b) mean that the lifting capacity of the crane is different from ten tons, i.e. is either more or less than that. These sentences only mean that ten tons do not belong to the range of weights that can be lifted by the crane. It follows that the crane can most likely lift less than ten tons but in no way more. This brings us to the conclusion that (30a,b) is the negative form of the INTERVAL reading (15) but not of the boundary reading (19). Noteworthy is a striking parallelism between (19) and (22). First, as mentioned above, along with sentences (20) - (21), they describe similar situations: they ascribe a certain value to a parameter. Second, with the phrasal stress on the main verb (vesit in (22), mozhet in (19)), the same semantic effect occurs: (31) [=(27)] Gruz 4-vesit dejsat' tonn. 'the cargo weighs ten tons and maybe even more' (32) [-(15)] Etot kran -tmozhet podnjat' desjat' tonn. 'the crane can lift ten tons and maybe even more' In both (31) and (32) the interpretation of the noun phrase shifts from an exact value of the parameter to an open interval value ('ten tons or more'). Third, in both (19) and (22), attaching the negative particle to the main verb does not result in the negation of the initial sentence. In both cases, the resulting sentence is the negation of the stressed variant (31) or (32), respectively. Still, in spite of these similarities, pairs (20)/(21), (19)7(22) and (31)7(32) differ in an interesting way. The first difference concerns the correlation of the semantic interpretation of the sentence with the dictionary semantic definitions of words. The case of (20) - (21) is the simplest one: the meaning of these sentences is presented in the most straightforward way. There is the name of a parameter (weight, lifting capacity) and this parameter is directly ascribed a value by means of the copula or a copula-like verb. Sentence (22) has a similar semantic structure. As noted above, the verb vesit' 'to weigh' denotes nothing else than the same parameter of weight but presents it in a verbal form: X vesit P 'X weighs P' means exactly the same as Ves X-a raven P 'the weight of X is P'. The interpretation of sentence (19) is not obtained so easily. As we saw above, the boundary reading of this sentence requires interaction of the dictionary definition, communicative function of the sentence and a conversational maxim. For the stressed variants (31) and (32), it is the other way round. (32) is a direct (emphatic) realisation of the dictionary definition of moch' 'can': ten tons is one of the possibilities. Since the ability of the crane to cope with lighter weights trivially follows from this fact, other possibilities worth having in mind necessarily exceed ten tons. As for (31), its interpretation cannot be obtained directly from the definition of vesit' 'to weigh'. To interpret the phrase vesit' desjat' tonn 'to weigh ten tons' as 'to weigh ten tons or more', it is necessary to postulate a special modification of the dictionary definition which operates under the given intonation conditions (see Boguslavsky 1985: 28 - 29).

262 Langue and Parole in Synchrony and Diachrony The second difference between (19) and other sentences is really surprising and requires an explanation: (19) has no negation, in spite of the fact that all other sentences, including its synonym (21) Gruzopodjemnost' krana ravna desjati tonnam 'the lifting capacity of the crane is ten tons', are easily negated The solution to this seeming inconsistency can be found in the way the meaning is encoded in (19). Let us recall how the interpretation of this sentence is obtained. It can be represented as consisting of three steps: 1. Literal meaning provided by the dictionary definition of the modal verb: 'it is possible for the crane to lift ten tons'. 2. Interpretation obtained by the maxim of quantity: 'if the crane were able to lift more than ten tons , the speaker would have mentioned this possibility'. 3. Hence: 'the maximum weight the crane can lift is ten tons'. The resulting meaning obtained at the third step has no contraindications for negation. What is essential in this interpretation is the fact that the maxim of quantity operates on a definite value that is already presented as true. This mechanism simply cannot apply to a negative sentence. If no definite value is proposed as a possibility, then THE QUANTITY MAXIM HAS NOTHING TO OPERATE UPON and the maximum meaning cannot appear.

REFERENCES Boguslavsky I. M. (1985). Issledovanija po sintaksicheskoj semantike: sfery dejstvija logicheskix slov. Nauka Publishers, Moscow. (Studies in Syntactic Semantics: Scope of Logical Words) Rullmann, H. (1994). De ambiguiteit van comparativen met MINDER. TABU: Bulletin voor taalwetenschap, 24, 79-101. Rullmann, H. (1995). The Ambiguity of Comparatives with Less. In: ESCOL'94: Proceedings of the Eleventh Eastern States Conference on Linguistics. (Janet M. Fuller, Ho Han, and David Parkinson, eds.), pp. 258 - 269, DMLL Publications, Cornell. Seuren, P.A.M. (1979). Meer over minder dan hoeft. De nieuwe taalgids 72, 236 - 239.

20

LANGUE AND PAROLE IN SPEECH ACT THEORIES: SOME CONSIDERATIONS AND A PROPOSAL Elisabetta Fava, Dipartimento di Linguistica, Universita degli Studi di Padova, Italy

Abstract Aim of this paper is to enlighten the complexity of the langue phenomena, on one hand, and the parole phenomena, on the other hand, that a speech act theory has to explain. This complexity may have brought about confusion, leading to rather approximate linguistic descriptions. By defending a proposal of two levels of representation of illocutionary force devices, an explanation of the character and of the organisation of a series of phenomena of variation in some Indo-European languages is offered, allowing for the descriptive adequacy of some grammatical descriptions, linking form and function, langue and parole.

i. INTRODUCTION Although relevant insights into the action-character of linguistic phenomena and the idea of speaking as a form of action dates back to Greek grammatical speculations, it was only in the second half of this century that this approach became standard and influential. There has been a wide application of speech act theory to many different fields, psychology, sociology, cognitive science. This research, involving many different languages and mostly aimed at parole theories, has stressed the importance of the role of communication in language, often distinguishing a communicative, or, in a rather broad sense, linguistic capacity from the capacity to have linguistic representations in a narrow sense. However, in one of the first systematic proposals it was argued that although it might seem that the speech acts approach is simply, in Saussurian terms, a study of parole, rather than langue, "an adequate study of speech acts is a study of langue" (Searle, 1969, p. 17). Nevertheless, contrary to possible expectations, the state of the art suggests that this interface has been only partially explored. On the one hand, the literature on speech acts has revealed only desultory interest in linguistic theories. On the other hand, only a few of the linguistic phenomena dealt with in speech act theory have been the object of intensive research in linguistic theory. The contribution of grammatical theory in proposing and developing hypotheses about the recognition and the interpretation of speech acts, such as the analysis of the linguistic features relevant to their formulation and to their interpretation, is somewhat

263

264 Langue and Parole in Synchrony and Diachrony controversial and less developed than expected. This might seem curious, considering that these studies, which deal with problems concerning the nature of the rules and of the conventions which make our conversations effective and appropriate, almost always presuppose grammatical competence. In many cases, the specification of the type of act performed in saying something, whether it is a question, a request for action, or a wish, has a grammatical codification: for example, a series of grammatical considerations also play a part in interpreting utterances as acts of a certain type whether in English, Russian, or in one Northern Eastern Italian dialect (Central Veneto variety): (1) a You open it (I believe). (2) a Ty otkryvaes 'ego (ya dumayu). (3) a Lo verzi (penso).

b Do you open it? c Open it! English b Ty otkryvaes 'ego? c Otkryvaj ego! Russian b Lo verzio? c Verzilo! Central Veneto

However, the linguistic description of the cluster of morphosyntactic features which, without reliance on any extralinguistic context, permit the interpretation of an utterance as an assertion in a, or as a question in b, or as an order in c^ is instead very often unsatisfactory. Not only has the range of variations of different grammatical features involved not been systematically discussed in a cross language perspective, but there is also a great deal of variation in form and function within a given language, whose role has to be reconsidered. An interrogative structure presenting the clitic particle li distinguishes a special class of questions (doubt/uncertainty questions) from canonical ones in Russian, while the main clause infinitive in Central Veneto characterises orders of a very general nature: (2) b' Tyli otkryvaes 'ego? Should you open it?

(3) c' Verzerlo (sempre, el vin)! (It is better to) Open[+infinitive] it (always, the wine)!

To complicate matters, context variation also has to be considered. A sentence with grammatical features of the interrogative type such as (4) a When do you think the window could be opened ? b Quando ze che te pensarissi che se podaria verzer sta finestra? c Kogda, dumaes', mozno otkryt1 okno? may be imagined in at least three different contexts: a) addressed with kindness, to a convalescent interlocutor, recovering from a serious illness or b) said with sarcasm, by someone suffering from the heat in a stuffy room, c) or, with a pleading tone, by a smoker. Hence, the same sentence can be interpreted not only as a question, but as a request for action, a sarcastic protest, or as an excuse. The fact that the same utterance can be interpreted in different ways according to the context in which it is said, justifies, if not explains, the widespread lack of methodologically clear-cut distinctions between the grammatical and pragmatic properties of what is said. The reasons why speech acts are interesting as a link

Langue and Parole in Speech Act Theories 265 between language and communication, language and action, are also a cause of theoretical and methodological difficulties in attaining descriptive adequacy, when dealing at the same time with both sentences and utterances, and when applying controversial distinctions at the boundaries between the grammatical and pragmatic properties of what is said. I will argue that the complexity of the langue phenomena, on one hand, and the parole phenomena, on the other hand, that a speech act theory has to explain, or at least accomodate may have brought about confusion and imprecision leading to rather approximate linguistic descriptions of the illocutionary force devices, or of the candidates for such a role, that is the role of giving an utterance its literal force. It has encouraged, at the same time, the prevalence in the recent literature of positions which neglect or reject the literal aspect of speech acts and give preferential analysis to the inferential and contextual aspects. The aim of this paper is to defend the importance of an adequate descriptive analysis of the set of morphosyntactic criteria which have developed around the notion of linguistic illocutionary force devices. They are intended as the linguistic elements that permit the interpretation of an utterance as an act of a particular kind, i.e. as a request for action, a question, an assertion, urging, and so on. Following a tradition of research which clearly associates grammatical properties with types of speech acts, the descriptive adequacy of some of the linguistic categories most commonly adopted in the representation of illocutionary force devices is discussed with regard to some Indo-European languages. The paper is organised in the following way. While Austin appears to be fully aware of the complexity of an appropriate identification of the categories and strategies which may be used to indicate particular types of acts, his few and cautious observations have been taken up in the subsequent literature in a rather oversimplified way. Despite the diversity of solutions proposed in the literature, from that of Benveniste to the most recent, many current formulations of illocutionary force devices reveal a conception that systematically overlooks the possible scope of variation of the different devices. There has been a tendency to limit the correlation of grammatical structure and illocutionary force either to just one single feature, mood, or group them into "types and main moods", leaving them undefined or organizing them into broad categories, excluding in both cases the phenomena of variation (§ 2, 3). A reconsideration of the modal distinctions inherited by Western grammatical tradition from Apollonius Dyscolus1 theory of moods will allow for the comparison of different IndoEuropean languages as well as the analysis of the structure of a given language. It will be highlighted that the phenomena of variation in grammatical forms across and among these different languages reflect a non-isomorphic correlation of one feature-one force. As the wealth and subtlety of linguistic description have increased with respect to Romance, Slavic or Ancient Greek, some aspects of the relations between langue and parole have to be reconsidered (§ 4, 5, 6).

2. AUSTIN ON MOODS AND OTHER ILLOCUTIONARY FORCE DEVICES Research into the criteria, or rather the set of criteria, simple or complex, which involve both grammar and vocabulary, and which permit the interpretation of an utterance as an act of a certain type, such as an order, a question, or an assertion, has its starting point in Austin. Although his analyses are developed around the explicit performatives, a hybrid of grammatical

266 Langue and Parole in Synchrony and Diachrony and lexical organisation, rather than around the primary performatives (Austin, 1975, p. 69), there are several passages of his work, in my view inadequately discussed in the literature, where he appears to be aware of the complexity of the problems raised by the identification of the grammatical categories and strategies which languages make use of to indicate illocutionary force. For instance, Austin (1975, pp. 1-2) notes that "... both grammarians and philosophers have been aware that it is by no means easy to distinguish even questions, commands and so on from statements by means of the few and jejune grammatical marks available, such as word order, mood and the like: though perhaps it has not been usual to dwell on the difficulties which this fact obviously raises. For how do we decide which is which? What are the limits and definitions of each?" Although it is well known that word order variation, special particles and conjunctions, intonational patterns and other elements may be involved, mood is the grammatical category which seems to have received the most attention in speech acts theory, even though it is far from being a well established linguistic category. It is adopted in rather different ways, depending on the language described and the theoretical framework adopted. There is a rather interesting passage in Austin, where he seems to leave wide open the problem of how to deal with this grammatical category: this leads, however, to many troubles over, for example, when a verb is in the imperative mood and when it is not...(Austin, 1975, p. 59-60) A rather interesting and revealing case of the difficulties in deciding, to rephrase Austin, "when a mood is interrogative and when it is not" comes from the controversial attribution of an interrogative mood to Question marker characterising some Northern Italian dialects. In the Central Veneto variety, for instance, the question markers involve distinctions of gender, number and person (first, second, third sing fern., third sing, pi., etc.). Their distribution is rather complex in different verbal tenses (present, imperfect, etc.) and in different syntactic contexts. Already attested in the sixteenth century, they are still a very productive and widespread syntactic strategy, characterising both direct yes/no questions and WH- questions. While there is substantial and uncontroversial evidence that these Question markers derived diachronically via the enclisis of subject pronouns, there is considerable fluctuation in their synchronic descriptions, not yet sufficiently standardized to avoid contradictions, not only in their formal representation, but even in their terminology (Fava, 1997). The different proposals oscillate between a description of these question markers as involving either a strategy which makes use of variations in the morphological structure of the verb (interrogative mood) or a strategy of pronominal inversion. Moreover, when considered pronouns, they are treated either as Clitics with at least two radically different interpretations, phonologically Clitic Pronouns, but syntactically in Inflection, or syntactically Full Pronouns in NP position (Fava, 1993, 1999). These oscillations and inconsistencies in the categorization of this single illocutionary force device are not only problematic on a pre-theoretical level, but they are also relevant to a theoretical approach. They not only reflect a terminological problem, but lie in the theory of grammar, by reflecting the different weight given to different orders of empirical generalizations.

Langue and Parole in Speech Act Theories 267 The grammatical research into the illocutionary force devices, inasmuch as they involve a comparison of several levels of grammatical description - morphological, syntactic and lexical having a unifying explanation in function, calls for an assessment of their organization and interaction within a given theory. In verifying the appropriateness of the attribution of an interrogative mood in Northern Italian dialects, the heterogeneity of the grammatical criteria, explicit or implicit, which are used to deal with the relationship between mood and illocutionary force devices, renders unsatisfactory any pre-theoretical and intuitive utilization of these notions. Moreover, it is far from clear how mood interacts with other illocutionary force devices. For instance, although grammatical considerations of mood and word order variations also play a part in interpreting the utterances in (l)-(3) a, b and c as acts of a certain type, the way these grammatical devices are involved is different in English, Russian, or Central Veneto. By simply considering the minimal pairs proposed in (l)-(3), Russian does not make use of word order variation, but both English and Central Veneto do. However, the English post-verbal position of the Subject pronoun you pertains to the interrogative structure, while pronoun post-verbal position characterises the iussive sentence in Russian and Central Veneto. Affix variation characterises both Russian and Central Veneto, even though it concerns the declarative/interrogative pair in Russian, but the declarative/interrogative minimal pair in Central Veneto (dizi -dizio). The clitic particle // in (2) b1 distinguishes a special class of questions (doubt/uncertainty questions) from canonical ones in Russian, while the presence of the infinitive in Central Veneto main clause in (3) c' may characterise orders of a very general nature. The ways the categories and strategies involved in speech act distinctions differ in different languages and within a given language are not systematically described. Moreover, variation phenomena characterising speech acts should be reconsidered not only from a synchronic perspective: there is also a great deal of variation in form and function from a diachronic perspective to be dealt with. Despite Austin's few but careful observations about the complex problem of the relationship between grammatical categories and speech acts, in the subsequent literature little attention has been devoted to the problems raised by a grammatical analysis of illocutionary force devices. Different theoretical frameworks have not developed enough for a comparative study across languages. In any case, in order to reframe the role of speech acts theory with respect to grammatical theory some langue and parole distinctions have to reconsidered.

3. THE IMPERATIVE AS NAKED SEMANTEME IN BENVENISTE The question of which boundaries of langue phenomena in speech act theory should be considered relevant both to philosophers and linguists is dealt with in Benveniste, who reconsiders the appropriateness of the distinction between performative and constative utterances, as well as the relationship between moods and acts, outlined in a first paper by Austin (1962). While Austin came to consider the performative and constative antithesis inadequate as its boundaries seemed to him to be too unclear, Benveniste seems to regard this distinction as the only one relevant to grammatical theory. On the other hand, he rejects Austin's proposal of morphosyntactic illocutionary force devices, on the basis of a series of argumentations centred on mood, and in particular on the imperative, as "naked semanteme" (semanteme nti). In this framework, the complex relationship between performative verbs and

268 Langue and Parole in Synchrony and Diachrony grammatical devices, or rather, between explicit, as in (5)a and primary performatives, as in (5)b (Austin, 1962) is reconsidered by Benveniste (1963, p. 274): (5) a Je vous ordonne de fermer la porte. b Fermez la porte. While conventional linguistic considerations about this relationship, together with an evaluation of the "infelicities of an utterance", related to the consequences and to the effects of a speech act, persuaded Austin to abandon this constative-performative antithesis, in favour of a different distinction, according to Benveniste this contrast between constative and performative need not be abandoned as long as precise criteria of a formal, linguistic nature are respected. Moreover, he maintains that it would be a mistake to unify performative verbs into a more general category, treating the performative utterance (5) a and the non-performative (5) b in the same way, as only the first is of strictly linguistic pertinence: only the performative utterance has the self-referential character that distinguishes it. The imperative, for Benveniste, is not a perfomative utterance, because it is neither an utterance, in that it does not serve to construct a clause with a personal verb, nor is it performative, in that it does not name the act to be carried out. The imperative is not denotative and does not aim to have an effect on the addressee, that is to order him to carry out a task. For Benveniste, it is only on the basis of grammatical considerations of imperative forms that the correlation between acts and grammar has to be limited to explicit performatives. Observations about the restrictions of imperative forms, which hardly ever carry indications of time or person, occur fairly frequently in the literature: commands and requests are restricted as regards the semantic distinctions that are grammaticalized for time and person, since one cannot rationally order someone to do something in the past, but only in the immediate or distant future, and one does not command oneself, or in the third person - as non-person (Benveniste, 1956) - unless incorrectly. These linguistic considerations, mainly based on a few characteristics of the imperative (which lacks temporal and personal distinctions) and reflecting a rather common conception of the imperative as a pure, absolute form of the verb, almost a non-syntactic form, led Benveniste to reject the notion of illocutionary act even on a level of macroanalysis, thus withdrawing from a conception which has proved to be as fruitful as that of performative verbs, which, instead, he had accepted. Although Benveniste's approach has been left on the sidelines in the current literature, since the notion of illocutionary force devices has had general consensus, many of the subsequent formulations share with him the same tendency, that is of limiting the correlation between grammatical structure and illocutionally force to a single feature, concentrating on mood, and neglecting the possible co-existence of more than one feature, and the scope of variation that this implies. 3.1. Form and functions of the imperative in Slavic languages. A first argument for the descriptive inadequacy of Benveniste's proposal of the imperative as naked semanteme and of his refusal, tout court, of moods as illocutionary force devices comes from the the morphological paradigms of the imperative in Slavic languages. They do not represent the evolution of the Indo-European imperative, which was lost in these languages as it was in the Baltic languages, but rather they represent the ancient optative, which these

Langue and Parole in Speech Act Theories 269 languages adopted with a iussive function (Vaillant, 1966, p. 28-44; Szemerenyi, 1985, p. 300, Fava and Benacchio, 1991). Although the imperative of modern Slavic languages, compared to Old Church Slavic, appears to have progressively undergone an ever increasing simplification of morphological features (Jakobson, 1965 and Isacenko, 1960, p. 478), a series of distinctions of forms and functions should not be overlooked. The imperative today has a paradigm, which, however reduced, still possesses its own morphemes and represents the evolution of a mood with regular personal endings. It constitutes a class of well-defined morphological forms, not reducible to the pure stem, extended at most, with a morphological device of the plural. Many Slavic languages still have, either alone (in the singular), or before the plural inflectional ending, a vowel which is characteristic of the imperative, and which may happen to be the same as that characterising the stem of the present from which the imperative is formed. Furthermore, in some languages and for some classes of verbs, this vowel is different in the singular and the plural. In Bulgarian and Macedonian, for example, some classes of verbs are characterized by a vowel of the singular (/) still different from that of the plural (e). These languages, compared to Old Church Slavic, present an extension of vocalic alternation even in a class of verbs that originally did not have it, namely the verbs whose stem ends in /'. Only if the stem of the present ends in a preiotized vowel, do the imperative forms present a zero morpheme in the singular with the addition of te in the plural. A second argument for the descriptive inadequacy of Benveniste's proposal conies from the relationship between illocutionary force devices and aspectual distinctions also characterising, among Indo-European languages, many Slavic languages. Russian, for instance, has an aspectual opposition which has been considered relevant, at least for events verbs, both at the locutionary and on an illocutionary level (Fava and Benacchio, 1991; Fava, 1998). In (6), for instance, the co-occurrence of the perfective and imperfective imperative highlights their different functions: (6) S"es' kusocek torta; es', es'! Eat [+ perfective] the cake; eat [+ imperfective], eat [+ imperfective] The speaker, approaching with the cake, initially addresses the interlocutor with the perfective form, and then, in the face of distraction or hesitation on the part of the latter, repeats the request for action using the imperfective form. The sequence of these two imperative forms is not free, since the second, the imperfective, defines and specifies the perfective, which implies a certain subjective distance of the speaker from the action requested, that is from its concrete execution. It is therefore the more neutral of the two forms, or the more formal, and thus it may seem the more polite form, hi this way, aspectual distinctions are interpreted as politeness phenomena. They are an inherent part of requests for actions, not on the level of formal etiquette, but on that of human participation, expressing encouragement, polite invitation, or even permission to begin or continue the action. In dealing with the relationship between illocutionary force device and acts, this complexity of features pertaining to the structure of the imperative in Slavic languages, concerning both the aspectual system and the inflectional paradigms, proves not only Benveniste's, but any proposal limiting the illocutionary force correlation to one feature - one force, to be descriptively inadequate.

270 Langue and Parole in Synchrony and Diachrony The possible scope of variation of forms and functions in the specification of the type of act can lead to an analysis of the relationship of the different features involved, clarifying their role in different languages.

4. THE DESCRIPTIVE INADEQUACY OF ONE FEATURE - ONE FORCE PROPOSALS The tendency to reduce the correlation between grammatical structure and illocutionary force to a single feature, concentrating on mood, and neglecting the possible co-existence of more than one feature, and the scope of variation that this implies, characterises even the most recent literature. In the attempts to identify the cluster of criteria that permit the characterisation of an utterance as an act of a certain type, while some attention has been devoted to the phenomena of lexical organisation and the illocutionary verbs, the importance given to morphological and syntactic features has so far been very limited. Analyses concerning the set of morphosyntactic phenomena are rather often rhapsodic and very often oversimplified. A wide variety of solutions have been proposed which show a strong tendency to reduce the correlations of grammatical structure and illocutionary force either to systematic variations of a single feature ('mood') or to a jumble of syntactic features corresponding to some kind of act ('types' or 'moods'), excluding any phenomenon of variation. In some cases the two notions, mood, intended in its restricted sense, as a linguistic category linked to the verb (indicative, imperative, etc.), and syntactic type, conflate. This seems to be the case for Davidson (1979, p.9): we have on the one hand the syntactic, and presumably semantic, distinction among moods (such as: indicative, imperative, optative, interrogative) and on the other hand the distinction among uses of sentences (such as: to make assertions, to give orders, to express wishes, to ask questions). The moods classify sentences, while uses classify utterances. The indicative, optative, imperative, together with the subjunctive, are categories well represented in the Ancient Greek mood system, and, more in general beyond any reasonable doubt, in the late Indo-European verb. However, these modal distinctions inherited by Greek grammatical distinctions have to be reconsidered in a language-specific approach. For instance, although the relations between form and function for the optative are easily recognizable in Ancient Greek, it is far from clear how this notion should apply for modern Germanic or Romance languages: the nature of the interrogative or optative mood in English or Italian is rather puzzling. They seem rather to conflate two different linguistic categories, moods and syntactic types.

4.1. The imperative and the other moods in Apollonius Dyscolus: a non-isomorphic approach A rather different, positive contribution to the debate on the complex relationship between lexical organisation and morphosyntactic structure, opening up a debate which I believe could pave the way for major developments is offered by Apollonius Dyscolus' theory of moods,

Langue and Parole in Speech Act Theories 271 expounded in several passages of his Syntax. His discussion of the imperative and the other moods in relation to other grammatical features is very perceptive and surprisingly modern. The modal distinctions inherited by Western grammatical tradition from Apollonius Dyscolus' theory of moods were built up on the notion of n/uûcr) 8icc0emc; (psuchike diathesis), to the point that this has even been translated as 'illocutionary force1 (Householder, 1981; Robins, 1988; Fava, 1995). In dealing with the criteria, explicit or implicit, used to describe the relationship between moods and illocutionary force devices, Apollonius represents the opposite position from that of Benveniste: the relationship rejected by the latter is discussed in a complex way by the former. While Benveniste restricts illocutionary force to the criteria of lexical organisation (the explicit performatives), for Apollonius some moods are defined as a morphological expression of particular types of speech acts, and can also be defined in relation to illocutionary force. This relation, however, has not to be interpreted univocally (Apollonius 1910: III, pp. 54-146, 319394 (editionis bekkerianae XII-XXX, pp. 226-276b); cf. Ill, 21, pp. 286-288 (ed. bekk. VI, pp. 204-206) and I, 51, pp. 44-45 (ed. bekk. VIII, pp. 31-32). There are two terms relating to 'mood' and 'modality' in Apollonius' system of grammatical nomenclature: e/Kîcng (enklisis, inclination, mood) and psuchike diathesis (mental disposition), which is believed by modern scholars to be an anticipation of Austin's notion of illocutionary force. Not all the enkliseis can demonstrate or possess psuchike diathesis. The moods that possess diathesis, that is the indicative (horistike or aphophatike), the optative (euktike), the imperative (prostaktike), are analysed as being like a syntactic construction containing two verbs, of which the first term transmits the meaning of the mood, indicating it together with person distinctions, and the second transmits the lexical meaning of the verb. Each mood (enklisis) is transformed into nothing but an infinitive plus a word (lexis) conveying the meaning of the enklisis: thus, for Apollonius, the imperative ypacps equals ypaysiv aoi npoamaaco (I order you to write), the optative nspinatoiriq eu%o/nai as nspmarsiv (I hope that you walk), while the indicative ypdcpsig can be equated to o'pi£o/jai (Novg.Urk. 1296-1301) 'wir kiissen ein Kreuz darauf ( d. h. 'wir schwb'ren darauf) a ce xo6e noeedbieaeMb (Rig. Urk. 1298) 'und das teilen wir dir mit'

Die Prasensformen imperfektiver Verben weisen die Bedeutung des Futurs in futurischen Kontexten auf: (5)

(6)

ame JIH 6e3i> rpaMOTti npULnyiL... aa depsKUMb H xpauuMb (Laur.-Chr. 1377) 'wenn sie aber ohne ein Schreiben kommen...,werden wir sie festnehmen und gefangen halten' BnaHMMcg neMentroM. aa Koro oncuesmi\>. KOFO .nH yMepxBgTb (Laur.-Chr. 1377) 'ergeben wir uns den Petschenegen, so werden sie einige am Leben lassen, andere vielleicht toten'

Im Beispiel (5) bildet die Verbform im pf. A. npmryTb 'sie kommen' den Kontext des Futurs, in dem die letzten beiden Verben, die imperfektiv sind, die Bedeutung des Futurs zum Ausdruck bringen. Im Beispiel (6) haben die perfektiven Verbformen BflaHMtica 'wir werden uns ergeben' und yMepxBHTb 'sie werden toten1 die Bedeutung des Futurs. Die imperfektive Verbform JKHBHTB 'sie werden am Leben lassen' wird in diesem Kontext als Futur aufgefaBt.

Futurparadigmen im Russischen 377

1.1.3 Das Futur der aspektindifferenten Verben. Die meisten Verben batten im Altrussischen eine bestimmte aspektuelle Bedeutung des pf. oder des ipf. A. Jedoch blieben zahlreiche Verben noch aspektindifferent und konnten sowohl perfektiv als auch imperfektiv gebraucht werden. Die aterminativen verba simplicia vom Typ mrrH 'gehen1, ixaxH 'fahren', HCCTH 'tragen', BHJTBXH 'sehen1, cjiBiniaxH 'horeri, cyflHXH 'richten', rpinraxH 'siindigen', die in der russischen Sprache der Gegenwart imperfektiv sind, verhalten sich im Altrussischen noch unentschieden zum Aspekt, vgl. die perfektive (bzw. futurale) Deutung solcher Verben im Beispiel (7) und die imperfektive (bzw. prasentische) Interpretation dieser Verben im Beispiel (8). (7)

He IIIJIH oxpoica tdy [= npne^y] caMt H JBC rpuBHbi eesy [= npHBeay] (Birkenrinden-Inschriften des 13. Jh's aus Stara Russa) 'Schicke nicht den Diener, ich werde selbst fahren und werde zwei Grivnas mitnehmen1

(8)

Hropt kdem-b no BopHqesy (Igorlied: 12. Jh.) 'Igor fahrt durch Boritschev '

Die terminativen verba simplicia vom Typ jcaxH 'geben', nacxn 'fallen1, KynHTH 'kaufen', JIHUIHXH 'wegnehmen', cicxw 'sich (hin)setzen', JICHH 'sich (hin)legen', die in der russischen Sprache der Gegenwart perfektiv sind, waren im Altrussischen nicht auf den perfektiven Aspekt (Beispiel (9)) beschrankt, vgl. die Verwendung des Verbs Kyrarra 'kaufen' im imperfektiven Kontext (mil dem Phasenverb) im Beispiel (10): (9)

yHCHHim 60 ero OIIIBJIH 6iaxy BT> rpafli>, aa 6panibHO Kynsimb (Ostr.-Ev. 1056/7) 'seine Schiller waren in die Stadt gegangen, um Nahrung zu kaufen'

(10)

nonaxoM-B Kynwnu xjie6i> (Novg. Chr.: 13.-14. Jh.) 'wir haben angefangen, Brot zu kaufen1 (d. h. 'wir sind Brot kaufen gegangen')

Die entsprechenden Imperfektiva vom Typ flanxH oder aasaxH 'geben' waren zwar schon in den altesten Denkmalern des Russischen vorhanden, aber sie wurden selten verwendet. Prof. P. S. Kuznecov, ein aufmerksamer Forscher des Altrussischen, registriert einige Beispiele mit dem imperfektiven Gebrauch des Verbs nacxH '(hin)fallen' bis zur Halfte des 19. Jh. Solche Beispiele wurden sogar bei A. S. Puskin gefunden (KyaneuoB 1953: 245-246). Die prafigierten Verben vom Typ Btxe^H 'munden, sich ergieBen', die in der russischen Sprache der Gegenwart perfektiv sind und deshalb keine Prasensbedeutung aufweisen konnen, waren im Altrussischen aspektindifferent und konnten in Kontexten auftreten, die als imperfektiv bzw. prasentisch gelten: (11)

A flninpt emenemb [=xeiiexB ] B HonxBCKoe Mope (Laur.-Chr. 1377) 'Aber der Dnepr miindet ins Schwarze Meer'

378 Langue and Parole in Synchrony and Diachrony Die spatere Akademische Redaktion und die Radzivil-Chronik geben in diesem Beispiel eine eindeutig imperfektive prasentische Form Tenexb an. Die Form noxeqe im Beispiel (12) aus der Laurentius-Chronik (1377) wird durch die eindeutig imperfektive prasentische Form revert im Beispiel (13) in der Hypatius-Chronik (urn 1425) ersetzt. (12) flHtnp-b 60 nomene(mb) H3 OKOBbCicaro Jitca (Laur.-Chr. 1377) 'Der Dnepr aber entspringt aus dem Okov-Wald1. (13) flfffenpt 60 menemb H3 Boji(o)KOBbCKar(o) Jiica (Hypat.-Chr. um 1425) 'Der Dnepr aber entspringt aus dem Vol(o)kov-Wald'. Die Prafixe solcher Verben treten im Altrussischen nicht als Marker der Perfektivitat, sondern als Marker der Terminativitat (oder der Grenzbezogenheit) auf (vgl. KapasaHOB 1992: 106107).

1.2 Das erste zusammengesetzte Futur 1.2.0 Allgemeine Bemerkungen. Das erste zusammengesetzte Futur wurde durch die Verbindung eines Hilfsverbs mit dem Infinitiv des zu konjugierenden Verbs gebildet. Die Hilfsverben xo^ro 'ich will', HMaMb 'ich habe', HMy 'ich nehme1, HantHy und no^bHy 'ich beginne' weisen vielfach zugleich modale Bedeutungen des Wunsches, der Notwendigkeit, der Moglichkeit und des Anfangs auf. Die kunftig gedachte Handlung wurde als wunschenswert, unausweichlich oder moglich dargestellt. Alle diese Umschreibungen mit verschiedenen Modalverben kann man nicht als paradigmatische Formen des betreffenden Verbs ansehen, da sie auBer der reinen Tempusbedeutung auch noch andere Bedeutungsnuancen ausdrucken. In Abhangigkeit vom Kontext wird entweder eine rein grammatische Bedeutung des Futurs oder eine modale Bedeutung hervorgehoben. Es gibt einen genetischen Zusammenhang zwischen der rein temporalen Futurbedeutung und modalen Bedeutungen. Das Futur ist ein irreales Tempus, das ursprunglich mit den verschiedenen modalen Bedeutungen verbunden war. Futurmarker haben deshalb als lexikalische Quelle die verba desiderativa,necessitativa, possibilitativa et ingressiva. In einigen Sprachen sind bis heute Spuren des modalen Charakters des Futurs enthalten, vgl. z. B. die Futurparadigmen mit den Hilfsverben will und shall im Englischen und mit dem Hilfsverb imu im Ukrainischen: pysatymu 'ich werde schreiben1 CXOX-BXH. Der Infinitiv konnte in dieser Konstruktion als imperfektiv, perfektiv oder aspektindifferent auftreten. Die Form des Infmitivs war folglich von der aspektuellen Bedeutung des Stammes nicht abhangig. Das war auch ein Kennzeichen einer freien Wortverbindung. Im Altrussischen treten die Umschreibungen mit xonro haufiger auf als die Periphrasen mit HMaMb/HMy. Die Wortverbindungen mit na^bHy/ no^bny waren weniger gebrauchlich als die ersten beiden Konstruktionen.


Im folgenden warden lexikalische und grammatische Bedeutungen dieser Umschreibungen im einzelnen ermittelt. 1.2.1 Xoqro + Infinitiv. Die Wortverbindungen mit dem Verb xorb™ 'wollen' behalten meistens dessen modale Bedeutungen. Deshalb kann man in solchen Fallen von zusammengesetzten Pradikaten, nicht aber von grammatischen Formen sprechen. Dieses Verb hat mehrere modale Bedeutungen. Es ist in dieser Hinsicht stark markiert. Die folgenden modalen Bedeutungen wurden registriert: die desiderative (Beispiele (14) und (15)), die nezessitative (Beispiele (16) und (17)), die possibilitative ( Beispiel (18)), die Bedeutung der Wiedergabe von Vermutungen (Beispiel (19)). (14) (15) (16)

(17)

(18)

(19)

ast nafli peicy eaM H^jiOBinfl JKHTHH omumu xouimio (Izbor.Svjat. 1076) 'ich als Gotteskind sage euch: ich will das menschliche Leben verlassen1 XOHTO Bbi noHmumu nayrpHH npeflij JHO^LMH CBOHMH (Laur.-Chr. 1377) 'ich will euch am nachsten Morgen vor meinen Leuten ehren' Ce y?Ke XOHCMT, noMepemu o(Tb) rjiaaa, a o(n>) KHHSJI IIOMOHH ntxy (Laur.-Chr. 1377) 'Jetzt werden wir Hungers sterben (miissen), und vom Fiirsten gibt es keine Hilfe' HeroflHO Hbint BCCITB HTH, XOHCM nozyffumu cMep^bi H pojitro HXT> (Laur.-Chr. 1377) 'Es ist ungiinstig, jetzt im Friihjahr ins Feld zu ziehen, denn wir werden die Bauern und ihre Felder unweigerlich vernichten' KaKO a3i> XOHTO HHT> 3aKom> npusimu eflHirt. a apyacnHa ceMy CM^HTHCH na^Hyrt (Laur.-Chr. 1377) 'wie kann ich das Gesetz allein verabschieden, die Drushina aber wird dariiber lachen' JiroTt ce My5Kb xonemb ffumu, HKO HM%HBH ne 6pe»ceTB, a opy^cte eMJiert (Laur.Chr. 1377) 'das wird (im Sinne das muB) ein grausamer Mann sein, der (wertvolle) Geschenke verachtet, aber Waffen annimmt'

Die grammatische Bedeutung des Futurs, die bei diesem Verb relativ selten ist, tritt am deutlichsten zutage, wenn das Subjekt des Satzes unbelebt ist (Beispiele (20) und (21)). Es gibt aber auch einige unumstrittene Falle, wo die futurische Bedeutung bei belebtem Subjekt auftritt (Beispiel (22)). (20) (21) (22)

H pene eMy: TO etcH JIH ÎTO yxpo xomemb 6umu (Laur.-Chr. 1377) 'Und er sagte ihm: weiB du, daB der Morgen bald anbrechen wird' norbiGejit xomsaue ffbimu eMy, HCKyci. xow,emb ecmamu na TH (Russk. Pravda 1282) ' fur ihn wird der Untergang sein, die Versuchung wird iiber dich kommen1 aace ne orjio^KHinb jranraero atjia H BCHKoe nenpaBflbi MM XOHCMT* 6ory 3KSuioeamucsi H TeMi> KTO npaB^y Jiio6HTb (Rig. Urk. um 1300) 'wenn du unnotige Sachen und verschiedene Unwahrheiten nicht beiseite legen wirst, werden wir uns bei Gott und bei denen, die die Wahrheit lieben, beschweren'

In manchen Fallen ist es nicht klar, ob es sich um eine rein grammatische oder um eine lexikalische Lesart handelt. Im Beispiel (23) gibt es zwei Moglichkeiten, die Worte von Olga

3 80 Langue and Parole in Synchrony and Diachrony zu verstehen: Sie wird (bestimmt) nicht Rache nehmen oder sie hat nicht die Absicht, Rache zu iiben. (23)

a y2Ke ue xomio Mbmamu, HO xoiqio ,naHb uMamu no Majiy (Laur.-Chr. 1377) (a) 'ich werde nicht Rache nehmen, sondern einen Tribut erheben' (b) 'ich habe nicht die Absicht, Rache zu iiben, sondern will einen Tribut erheben'

1.2.2 MMaMi. und HMV + hifmitiv. In altrussischen Texten kommen zwei Hilfsverben vor, namlich das altslavische imperfektive Hilfsverb HMaMt 'ich habe' (< HMfrra 'haben') und das altrussische perfektive Hilfsverb HMy 'ich nehme' (< JTTH 'nehmen'). In vielen Fallen konnen die Umschreibungen mit HMaMt/HMy rein modal gedeutet werden. Die nezessitative Bedeutung ist gut erkennbar in den Beispielen (24) und (25). Die possibilitative Bedeutung und die Bedeutung der Wiedergabe von Vermutungen illustrieren entsprechend die Beispiele (26) und (27). Die ingressive Bedeutung taucht in Beispiel (28) auf. (24)

(25) (26)

(27)

(28)

ame He noflcrynHxe aayxpa, npedamu ca uMaMb ntêHeroMT> (Laur.-Chr. 1377) 'wenn ihr morgen nicht zu Hilfe kommt, werden wir uns den Petschenegen ergeben (mussen)' xonia THH jnoflH xaKO 5Ke uMeiomb meepdwnu ropoa CMOTPHHT, (Urk. 1375) 'dann mussen diese Leute auch die Stadt Smotritsch bauen' Cpe6poM H sjiaxoM ne UMUM Hcuitcmu .npy^KHHti, a apy^cHHOK) Hajitey cpe6po H sjiaxo (Laur.-Chr. 1377) 'Mit Hilfe von Silber und Gold kann ich keine Drushina finden, aber durch eine Drushina werde ich Silber und Gold finden' H pene HryMent eoflocHH, HKO ce UManu> 6bimu OT 6tcoBCKoro atcTBa (Laur.-Chr. 1377) 'und es sagte der Abt Feodossius, da/3 das von der Teufelsversuchung sein kann' A XTO MOHXI 6oHpT> uMemb cjiy^cumu y Moee KHHFHHH ...(Urk. 1353) 'Wer von meinen Bojaren in den Dienst meiner furstlichen Gemahlin treten sollte...'

Die Bedeutung des Futurs tritt in den Beispielen (29) und (30) auf. (29)

(30)

HTO uMymb MOJieumu noctJio(Be) BejiHKort KHHSH..., TOMy eepy HMHTB (= HMHTB) (Novg. Urk. 1294) 'was die Abgesandten des GroBfursten sagen werden..., dem sollt ihr Glauben schenken' A3t yK.e TH orpuHy H ue UMaMb TH noMUJioeamu naKH (Laur.-Chr. 1377) 'Ich werde dich zuruckweisen und werde dich nicht nochmals begnadigen'

Das Beispiel (31) illustriert die Ambiguitat der Lesarten.


(31)

Ame 60 BtSMexe part MCJKIO co6oro, noraHHH UMymb padoeamu csi (Laur.-Chr. 1377) (a) 'Denn wenn ihr miteinander eine Fehde beginnt, warden Heiden ihre Freude daran haben' (b) 'Denn wenn ihr miteinander eine Fehde beginnt, beginnen Heiden ihre Freude daran zu haben'

1.2.3 HaqbHy und noqbny +Infinitiv. Die Wortverbindungen mil den Hilfsverben Haqtny und noHbHy weisen in unseren Beispielen nur ingressive Bedeutung auf: Beispiele (32) und (33). Sie sind in dieser Hinsicht schwach markiert. (32) flajKe KoroptiH KHH3b no MOCMt KHHJKeHHH noHHemb xomfynu O(T)HTH y cx(o)ro reopraH...(Mstisl.Urk. urn 1130) 'wenn aber ein Furst nach meiner Regierungszeit dem St. Georgkloster (die Schenkung) wegnehmen wollen wird...' (33) TOJIH HC 6y,neTb MCJKIO naMH MHpa. CJIHKO KaMent Hcmuemb ruiaeamu. a XMCJIL noHHe(mb) monymu (Laur.-Chr. 1377) 'es wird zwischen uns solange keine Eintracht sein, bis der Stein zu schwimmen und der Hopfen zu versinken beginnt1 Die grammatische Bedeutung des Futurs ist in den Beispielen (34) und (35) deutlich ausgedriickt: (34) (35)

a5Ke HdHHemb HC 3namu y KOFO KynHJii>...(Russk. Pravda 1282) 'sollte er nicht wissen, bei wem er (das Diebsgut) gekauft hat...' amerbi padoeamucsi HdHHeiub, npyiKima cM&tmucx HaHHymb (Laur.-Chr. 1377) 'wenn du dich freuen wirst, wird die Drushina lachen'

Im Beispiel (36) ist eine futurische und eine ingressive Lesung moglich. (36)

H HdHHemb 6pan> 6para saKcuiamu (Laur.-Chr. 1377) (a) 'und es wird ein Bruder einen anderen ermorden' (b) 'und es wird ein Bruder einen anderen zu morden beginnen'

1.3 Das zweite zusammengesetzte Futur Das Futur II wurde mit Hilfe des Part.praet.act.auf -JIT. des zu konjugierenden Verbs und des Hilfsverbs 6yay gebildet. Diese Konstruktion tritt im Altrussischen als freie Wortverbindung mit der Bedeutung der Vergangenheit in der Zukunft (praeteritum futuri) auf. Die beiden Komponenten dieser Konstruktion waren syntaktisch unabhangig voneinander. Das Verb 6y;ry hatte in Konditionalsatzen die modale Bedeutung der moglichen Ubertragung einer Handlung in die Zukunft (wenn es sich in Zukunft herausstellen sollte/wird). Die Handlung selbst, die durch das 1-Partizip ausgedriickt war, war in der Vergangenheit lokalisiert.


(38)

(39)

Ame 6ydy 6(or)y yzodmrb. H npUHiun> MJI 6ydemb 6(or)i>. TO no MOCMT, (oT)inecT BHH MOHacrtipt Ha^HCTb crpOHTH H npH6i>iBaTH B He (Laur.-Chr. 1377) 'Wenn es sich (in Zukunft) herausstellen wird / sollte, daB ich Gott gefallig war und er mich angenommen hat, dann wird man nach meinem Tode ein Kloster bauen und darin weilen' O5Ke CH nrb 6ydy onuccun, HJIH nepenuccun, HJIH ne donuccurb, HTHTC HcnpaBJiHsafl Bora fltjiH, a HC KJICHHTC (Laur.-Chr. 1377) 'wenn ich mich irgendwo verschrieben oder zuviel oder zuwenig geschrieben haben sollte, lest und verbessert, um Gottes willen, schimpft aber nicht' Koro 6ydy npuicyniurb HJIH XTO MH csi dydemb B BHHi docmam, HJIH XTO ca 6ydem y Ttixt jiroflHH oncemurb, sctMt rfeMt jnojaeMt ,najn> ecMt BOJIIO (Mosk. Urk.1353) 'wenn immer ich gekauft haben werde, wer mir durch das Gericht zugefallen sein wird oder wer von diesen Leuten geheiratet haben wird, alien diesen Leuten habe ich hiermit die Freiheit gegeben'

Die Unanfechtbarkeit des Lebens des Sprechers im Beispiel (37) ist zum Zeitpunkt des Schreibens des Textes ohne Zweifel. Die Moglichkeit der Festellung dieser Tatsache ist auf die Zukunft bezogen. Im Nachwort zur Laurentius-Chronik 1377 (Beispiel (38)) verwendet ein Abschreiber diese Konstruktion nicht fur die Bezeichnung der zukiinftigen Handlungen (die Chronik ist zu diesem Zeitpunkt schon geschrieben worden). Auf die Zukunft ist nur die Possibility des Auftretens der Handlungen in der Vergangenheit bezogen. Im Beispiel (39) aus dem Testament Simeons des Stolzen sind die Handlungen im Futur II auf die Zukunft bezogen. Diese Handlungen sind aber in der Vergangenheit in bezug auf den (noch bevorstehenden) Zeitpunkt des Todes des Erblassers lokalisiert. Die Selbstandigkeit des 1-Partizips wird durch die Erganzungen unterstrichen, die auf die Vergangenheit referieren: (40)

(41)

Eydemb JIH aoiojii Gjisum. rp(n)BHa cepe6pa sa copoMT> (Smolensk. Urk. 1229) 'Wenn es sich (in Zukunft) herausstellen sollte, daB sie bis dahin einen Ehebruch begangen hat, soil sie eine Griwna Silber fur die Schande bekommen1 a KTO dydernb Goapij H cjiyrb...or MCHC orm>ixcun> pp cero floKOHqankfl, no CCMB ÔKOHqaHbH...npH'bfleTb (Urk. 1367) 'wer von den Bojaren und Dienern bis zu dieser Vereinbarung von mir weggefahren sein sollte, kommt nach dieser Vereinbarung zuriick'

Die adverbielle Gruppe flo ceroflOKOHHanbJi'bis zu dieser Vereinbarung1 im Beispiel (41) zeigt, daB die Handlung des Wegfahrens auf die Vergangenheit referiert. P. S. Kuznecov (1959: 252) meint, daB sich die ganze Konstruktion klar auf die Vergangenheit bezieht. A. V. Issatschenko (1983: 392) sieht in dieser Konstruktion die Bedeutung des Futurs II. Meiner Meinung nach sind hier die beiden Tempora (Futur und Prateritum) vereint. Auf die Vergangenheit referiert auch das Part.praet.pass. auf -HT>, das im Beispiel (42) parallel mit dem Part.praet.act. auf- JIT> auftritt:


(42)

Ame KTO no3HaeTb ceoe MTO Gydemb nozy6tun> HJIH yjcpadeno y Hero. noHflH Ha CBOjfc (Russk. Pravda 1282) 'Wenn jemand erkennt das Seine, das er verloren hat oder bei ihm gestohlen wurde, soil er in die Versammlung geheri

Diese Konstruktion bezeichnete auch eine Handlung in der "Vorzukunft", d. h. eine zukiinftige Handlung, die vor einer anderen zukiinftigen Handlung stattfmden wird oder soil. Diese Bedeutung des Futurum exactum ist in den Beispielen (43), (44) und (45) ausgedriickt. (43)

(44) (45)

ame H rptxbi 6ydemt> KT>TO cbmeopmrb. ast HMaivi-b o TOMB npt^t 6t(ro)Mt QTBbmaTH (Lebensbeschr. von Feodosij Pecerski: 12. Jh.) 'wenn jemand gesiindigt haben wird, werde ich dariiber vor Gott verantworten' a KTO ffydemb Kynmn> cejia...a TO noHfleTb B HoBropoay (Novg.Urk. 1325-1327) 'wer Dorfer gekauft haben wird..., derjenige wird nach Novgorod gehen' aa BOSbMerb CBOC n^ce KTO 6ydemb nomepsurb (Russk. Pravda 1282) 'es soil das Seine zuriickbekommen, wer [etwas durch Diebstahl] verloren haben wird1

2 Entwicklung der Futurparadigmen im Mittel- und Neurussischen 2.1 Das einfache (= synthetische) Futur 2.1.1 Der Schwund der aspektindifferenten Verben. In der mittelrussischen Periode (15.-17. Jh.) werden die Reste der aspektuellen Indifferenz bei prafigierten Verben und bei verba simplicia (vgl. Abschnitt 1.1.3) beseitigt. Die neutrale aspektuelle Bedeutung der prafigierten Verben vom Typ BtTeqn 'munden, sich ergiefien' wird mit der Grammatikalisierung der Prafixe als perfektiv umgedeutet. Die prasentischen Formen solcher Verben vom Typ B-bTCHCTb erhalten damit die Bedeutung des Futurs (= 'wird munden'). Die aterminativen verba simplicia vom Typ TCHH 'fliefien' bekommen im Kontrast zu den prafigierten Verben vom Typ B^TCHM 'miinden' die Bedeutung des ipf. A. Die prasentischen Formen vom Typ TeneTb werden als Prasens aufgefafit (= 'fliefit'). Die terminativen verba simplicia vom Typ flara 'geben' werden als Perfektiva gebraucht. Einige von ihnen werden durch prafigierte Ableitungen vom Typ nacTH > ynacra 'fallen1 ersetzt. Der Aspekt umfafit von da an alle Verben ohne Ausnahme. Es vollzieht sich ein UbergangsprozeB zur Obligatheit der Opposition pf. A./ipf. A. Der Aspekt wird vollstandig grammatikalisiert. Die voile Ausbildung der Aspekt-Paradigmen war im wesentlichen zum Anfang des 19. Jh. abgeschlossen. 2.1.2 Stabilisierung des perfektiven Futurs. Die endgiiltige Ausdifferenzierung der Tempora im nichtprateritalen Bereich wurde durch die Auspragung der Aspekte ausgelost. Die Klasse perfektiver Prasentia mit der Bedeutung des Futurs wird urn die urspriinglich aspektindifferenten Verben vom Typ ,n,aM 'ich werde geben1 und noTeêT 'wird fliefien' erweitert. 2.1.3 Veranderungen im Gebrauch der imperfektiven Prasentia. Es vollzieht sich die endgiiltige Trennung zwischen dem Prasens und dem Futur. Die Futurbedeutung der imperfektiven

384 Langue and Parole in Synchrony and Diachrony Prasentia ist kontextabhangig (vgl. Beispiel (46) (b)). Das Auftreten imperfektiver Verben im futurischen Kontext mit den anderen Verben im Futur (vgl. Beispiele (5) und (6)) wird nicht mehr moglich sein (vgl. Beispiele (46) (c) und (46) (d)). (46)

(a) (b) (c) (d)

fl

H noedy (pf. A.) aasxpa B GnGjiHoreKy H eosbMy (pf. A.) 3xy KHHry edy (ipf. A.) sasxpa B 6H6jiHoxeKy H 6epy (ipf. A.) sry KHHry *% noedy (pf. A.) aasxpa B GnGjiHoreKy H 6epy (ipf. A.) 3ry KHHry *5I edy (ipf. A.) aasrpa B GnGjiHoxeKy H eo3bMy (pf. A.) 3ry KHHry 'Ich fahre morgen in die Bibliothek und hole dieses Buch'

2.2 Das erste zusammengesetzte Futur 2.2.0 Veranderungen im Gebrauch des Futurs I. Das Futur I hat im Mittelrussischen einige Veranderungen erfahren. Das Gewicht einzelner Hilfsverben wird im Laufe der Entwicklung verandert. Die Hilfsverben XOHK> und HMaMt verschwinden. Die neuen Hilfsverben crany und ynny kommen zusatzlich noch dazu. Am haufigsten treten die Hilfsverben HMy in der Literatursprache und y^ny in der Volkssprache (besonders in der Geschaftssprache) auf. Das Futur II geht im 17. Jh. endgiiltig verloren. Zum Ende der mittelrussischen Periode und zum Anfang der neurussischen Epoche (das 18. Jh.) bildet sich die analytische Form des imperfektiven Futurs mit dem Hilfsverb 6y,ny. 2.2.1 Xoqy + Infinitiv. Die Konstruktion xo*iy + Infinitiv, die aus dem Altslavischen vererbt wurde, wird schon am Anfang der mittelrussischen Epoche nicht mehr verwendet. Einzelne Beispiele (vgl. Beispiel (47)) sind noch im 16. Jh. registriert worden. Diese Konstruktion verliert ihre Futurbedeutung und wird nur in der voluntativen Bedeutung verwendet. (47)

eraa ao»cfli> xoiuemb 6bimu, Torâ ropti... KypnxcH (Naziratel1: 16. Jh.) 'wenn es regnen wird, dann rauchen die Berge'

2.2.2 HMV + Infinitiv. Im Zusammenhang mit dem Schwund der athematischen Konjugation werden die Umschreibungen mit dem Hilfsverb HMaMB nicht mehr verwendet. Die Konstruktion mit HMy war im 15. Jh. in der Kanzleisprache in Zentralrufiland verbreitet (vgl. Beispiel (48)). Im 16. und im 17. Jh. war sie fur die Literatursprache sehr typisch (vgl. Beispiel (49)). In der Kanzleisprache wird die Konstruktion y^ny + Infinitiv mehr verwendet. (48)

(49)

a 6oapo(M) HJIH fltxe(M) 6onpcKH(M)...c cyao(M) c 6oapCKH(M) uMy(m) cydumu (Sudebnik 1497) 'iiber Bojaren und Kinder von Bojaren wird ein Bojarengericht entscheiden' *rro6Bi B oropo(fl) co6aKH...63oumu ne uMyrm> (Domostroj: 16. Jh.) 'damit die Hunde nicht in den Gemiisegarten kommen konnen / werden1

Heute werden die Periphrasen mit HMy in der Futurbedeutung in nordrussischen Dialekten gebraucht, vgl. HMy po6nTH 'ich werde arbeiten', HMCM Hrpart 'wir werden spielen', qxo HMCM aejiart ? 'was werden wir machen?'


2.2.3 HaqHy/nogny/yMHy + Infinitiv. Seit dem Anfang des 15. Jh's taucht in der Futurbedeutung das Verb y^Hy auf, vgl. Beispiele (50) und (51). Die Hilfsverben nanny und noHHy (vgl. Beispiel (52)) werden weiterhin benutzt. In der Geschaftssprache des 17. Jh's wird die Periphrase mit ynny am haufigsten gebraucht, vgl. Beispiele (53) und (54) aus Ulozenie 1649. (50) (51)

(52) (53) (54)

a XTO y HHXT> ynnenrb oacumu JiroaeH (Urk. 1486) 'und wer von ihnen wird mit den Menschen leben' ynuembOTHKcy^HOH CFIHCOK na6ejio nucam...vi e\iy nHcneso êceT(b) aener (Sudebnik 1589) 'wird ein Beamter eine Gerichtsliste ins reine schreiben, (dann) soil er flir das Schreiben 10 Stuck Geld bekommen' a KTO noHHem-b na ry SCMJIIO uacmynamb (Dvinsk. Urk.: 15. Jh.) 'und wer dieses Land angreifen wird' KTO 5Ke Koro ynuem-b doeodumu rocyaapeBO flijio (Ulozenie 1649) 'wer gegen jemand einen Staatsprozefi ftihren wird' KTO yHHenrb y KOFO Kpacmu (Ulozenie 1649) 'wer bei jemandem stehlen wird'

In der reprasentativen Sammlung von Dokumenten aus der ersten Halfte des 17. Jahrhunderts betragt der Anteil der Konstruktion mit yqny etwa 74 % aller Futurformen, wahrend die Konstruktionen mit CTany und 6ypy entsprechend etwa 22 % und 4% ausmachen (FopniKOBa/XaGypraeB 1981: 323). Am Anfang des 18. Jh's verschwindet die Konstruktion mit ynny aus dem Sprachgebrauch. 2.2.4 CTany + Infinitiv. Die Periphrasen mit crany in Futurbedeutung sind seit dem 16. Jh. registriert worden (vgl. Beispiel (55)). Diese Konstruktion war im 17. Jh. besonders in der Geschaftssprache (d. h. in der Sprache der Dokumentation, der privaten Korrespondenz usw.) verbreitet, vgl. Beispiele (56) und (57) aus der Lebensbeschreibung von Awakum. (55)

(56) (57)

KOTOpbie JiroflH.HHoropOflUbi HJIH npnuuieiibi.cmaHvfmj ffumu HCJIOM...VI TCM JIK) A6M...cya aasaTH (Sudebnik 1589) 'den auswartigen oder zugereisten Leuten, die um ... ersuchen werden, diesen Leuten ... ist das Recht auf Gerichtbarkeit zu geben' ne cmaneiub nucamb, TSK a ocepjKyct (Lebensbeschr. von Awakum: 17. Jh.) 'wenn du nicht schreiben wirst, dann werde ich mich argern' a MH cmaueM 6ora Monwnb (Lebensbeschr. von Awakum: 17. Jh.) 'und wir werden Gott anflehen'

Das Hilfsverb CTany wird noch heute als Bezeichnung des Futurs gebraucht, vgl. Beispiel (58) aus Lermontovs beriihmten Wiegenlied. Im modernen Russischen wird dieses Verb mit starker modaler Farbung gebraucht, vgl. Beispiel (59). Besonders deutlich kommt die voluntative Bedeutung zur Geltung, wenn CTany mit der Negation verwendet wird, vgl. Beispiel (60). (58)

cmany cKa3bteamb % cKaaKy, neceHKy cnoro (M. Ju. Lermontow: 19. Jh.) 'ich werde ein Marchen erzahlen, ein Liedchen singen'

3 86 Langue and Parole in Synchrony and Diachrony (59) (60)

fl

cmany usynamb aHTjiHHCKHH H3HK 'Ich beabsichtige, Englisch zu lernen' H HC cmany ero yzoeapueamb 'Ich denke gar nicht daran, ihn zu iiberreden'

2.2.5 Eyjy + Infmitiv. Das zu GUTH 'sein' gehorende 6yay (mit dem Stamm 6ya- < *b(h)u-) war urspriinglich ein perfektives Prasens mit ingressivem Infix -n- (6yay < bpdo < *bondom), das dem imperfektiven Prasens ecMb (mit dem Stamm ec- < *es-) gegenuberstand. Daher komnit die Futurbedeutung dieses Stammes. Im Zusammenhang mit dem Schwund der nasalen Segmente im Ostslavischen geht auch die ingressive Bedeutung dieses Stammes verloren. Erhalten geblieben ist aber die andere modale Bedeutung, namlich die nezessitative. Diese Bedeutung tritt in unpersonlichen Konditionalsatzen auf, vgl. Beispiele (61) und (62). Der Stamm 6ya- konnte auch in der Vergangenheit verwendet werden, vgl. das Beispiel (63), wo die Verbform 6yflHine die 3. P. Sg. des perfektiven Imperfekts vom Verb GHTH 'sein' darstellt. (61)

(62)

(63)

AJKC dydemb pycnny roeapt uMamu Ha HCMHHHe... a pyGe^ca He fliflTH (Smolensk. Urk. 1229) 'Wenn es dazu kommt, daB ein Russe in Deutschland eine Ware kaufen muB, soil man keine Hindernisse [ihm] in den Weg legen1 A êro Gydemb ucKamu Mnt ...y HOBT>ropo<m>ueB'b...BceMy cy^t aaxn 6e3t nepeBoaa (Novg. Urk. 1296-1301) 'Wenn es dazu kommt, daB ich etwas von Novgorodern verlangen werde, dann soil es fur das alles unverzuglich ein Gericht geben' Ame nofacamu Sydsaue O6tpHHy, He aaflHnie BtnpHMH KOHH HH BOJia (Laur.-Chr. 1377) 'Wenn es so kam, daB ein Avare fahren mufite, liefi man ihn weder ein Pferd noch einen Ochse anspannen'

Der Ubergang von der unpersonlichen Konstruktion vom Typ KaKi> 6ydena> ce eMy d&aamu 'wie soil er das machen' zur personlichen Konstruktion vom Typ KaKt 6ydemb cejmi d&namu 'wie wird er das machen' konnte relativ einfach gewesen sein. Sehr wahrscheinlich erscheint auch der Weg der Kontamination der Konstruktion vom Typ 6yjiy nucajit 'ich werde geschrieben haben' (= Futur II) mit der Konstruktion vom Typ HMy nHcara 'ich werde schreiben' (= Futur I). Diese Moglichkeit wird spater nach der Betrachtung der Evolution des Futurs II erortert. Der Gebrauch der Konstruktion 6yay + Infinitiv in der rein grammatischen Bedeutung des Futurs ist seit dem Ende des 14. Jahrhunderts belegt. In der westrussischen Urkunde von Korbut (1388) ist das Beispiel SyaeMt flepaca(T) 'wir werden halten' registriert worden. Diese Konstruktion war zuerst im Nordwesten des russischen Sprachgebiets heimisch. Sie ist in den Urkunden von Polozk reichlich belegt. Diese Konstruktion wird haufig verwendet von Peresvetov (16. Jh.), der mehrere Jahre in Tschechien und in Polen verbracht hat und aus Litauen zugewandert war, sowie in den Werken von Fiirst Kurbski, der im Jahre 1564 nach Litauen floh. Die Fibel von Ivan Fedorov (1574) hat schon diese Form: Gyêuib HMtTH Haae^cfly 'du wirst Hoffhung haben', paflOBaracH Gyêrt cepaue MOC 'mein Herz wird sich freuen'. In der Kanzleisprache von Moskau tritt diese Konstruktion erst ab 17. Jh. auf. Das alles


kann auf einen westslavischen Einflufl zuriickgefuhrt werden. V. Kiparski (1967: 235) spricht iiber die polnische Herkunft dieser Konstruktion, die im Polnischen schon in der zweiten Halfte des 14. Jahrhunderts voll entwickelt war. Die Belege aus polnisch-russischen diplomatischen Verhandlungen der Jahre 1487-1560 werden als Beweis fur diese Behauptung angefuhrt. A. V. Issatschenko (1983: 390) weist auf die Korrespondenz des Michail Fedorovic Romanov (16131638) hin, die viele Polonismen (auch die Futurformen mit 6yay) enthalt. Man kann trotzdem kaum von einem "Literaturimport" (Vlasto 1986: 164) sprechen, weil diese Konstruktion im ganzen Sprachgebiet von Rufiland verbreitet war. Im 16. und 17. Jh. war diese Konstruktion fur die lebendige Volkssprache typisch. In der russischen Grammatik von G. Ludolf (Oxford 1696) werden die Konstruktionen 6yay und crany + Infinitiv als Ausdruck des Futurs fur die Umgangssprache genannt. Diese Form wird um diese Zeit auch in der Geschaftssprache benutzt. Die russische Literatursprache stand unter starkem EinfluB des Altslavischen. Die Futurformen mit 6yjry waren dem Altslavischen nicht bekannt. Der AnstoB zur Ubernahme dieser Form aus der Umgangssprache in die Literatursprache konnte vom Polnischen ausgegangen sein. Erst ab Mitte des 18. Jahrhunderts (vgl. die Grammatik von M. V. Lomonossov 1755) hat sich die Konstruktion 6yjry + Infinitiv als analytische Form des imperfektiven Futurs endgiiltig etabliert. Die anderen Hilfsverben , die eigene lexikalische Bedeutung hatten, waren aus dieser Funktion verdrangt. Sie sind in einigen nordrussischen Dialekten geblieben, vgl. HMy poSHTb 'ich werde machen', y^ny peseTt 'ich werde weinen', crany cnart 'ich werde schlafen'. Aufierdem waren die Verben xorfcra, HM^TH und craxH nur mit belebtem Subjekt kompatibel. Das Verb 6yay war geeignet fur die Grammatikalisierung in der Bedeutung der Zukunft. Es hatte eine vollig verblaBte lexikalische Bedeutung und konnte mit belebtem und unbelebtem Subjekt auftreten. Es war auch die einzige analytische Form des Aktivs im Indikativ, vgl. H 6yay ^Hxart 'ich werde lesen' und *H 6tui HHTaxt wortlich '*ich war lesen' (vgl. aber H craji HHTart 'ich fing an zu lesen'). Das Hilfsverb 6y,zry wird zur Bildung der Passiv-Paradigmen benutzt, vgl. 6yaex HanncaHO 'es wird geschrieben werden', aber *cxaHeT HarmcaHO. Die beiden Komponenten dieser Konstruktion weisen den imperfektiven Aspekt auf: Sie sind aspektuell homogen.

2.3 Das zvveite zusammengesetzte Futur Der Zerfall der Konstruktion 6y^y + Part.praet.act. auf -JIT> war schon im Altrussischen vorprogrammiert, indem die beiden Komponenten selbstandige Funktionen hatten. Das Hilfsverb 6y,ny bezeichnete die Projektion einer moglichen Handlung in die Zukunft, wahrend das 1-Partizip die Bedeutung einer vergangenen Handlung ausdriickte. Das Hilfsverb 6yay konnte weggelassen werden, vgl. die parallelen SStze (64) und (65) aus Russkaja Pravda 1282. (64)

(65)

PfeiTia OHt CBoe BT.3MeTt, a HTO Gydemb c HHML nozbiffjio, TO JKC eiviy naHHCTb njiaTHTH (Russk. Pravda 1282) 'Dann wird er das Seine nehmen, und was bei ihm verlorengehen sollte, das wird man ihm bezahlen1 A MTO c HHMB nozuGjio, a xoro eMy ^cejitTH (Russk. Pravda 1282) 'Und was bei ihm verlorenging, deswegen soil man ihn bedauern1

388 Langue and Parole in Synchrony and Diachrony Das 1-Partizip funktioniert im Beispiel (65) als Form des Prateritums. Die zweite Komponente dieser Konstruktion, namlich das Hilfsverb 6yay, entwickelt sich gewissermafien getrennt vom 1-Partizip. Das auBert sich in der Verletzung der Kongruenz zwischen dem Hilfsverb 6yay und dem 1-Partizip, vgl. Beispiel (66). (66)

a HCBO Sydenrb [= 6yay] (a) saGbuia HanHcara H BT> TOMT> Btflaert Bor (Urk. 1579) 'Und was ich zu schreiben vergessen haben sollte, das weiB der liebe Gott1

Im 17. Jh. verliert sich der Zusammenhalt mit dem 1-Partizip vollig. Auf diese Weise bekomnit die Form der 3. P. Sg. 6yaeTb die Bedeutung der konditionalen Konjunktion ecjra 'wenn, falls', vgl. Beispiel (67). Die Anderung der Bedeutung des Hilfsverb 6yay wird durch die Verbindung des Verbs 6yjiy mit einer finiten Form gekennzeichnet, vgl. Beispiele (68) und (69). (67)

(68) (69)

H KTO 6yderm> [= CCJIH] 6bun> Ha paaGot H ynHHHJi y6HHCTBO, a xaKHx ntrraioT (Kotoschichin: Mitte des 17. Jahrhunderts) 'Wenn aber jemand an einem Raububerfall teilgenommen hat und einen Raubmord begangen hat, so werden solche Leute gefoltert' A 6yderm> [= CCJIH] KOpojit ue Gydenrb Ha TOT cpoK (Urk. 1556) 'Wenn aber der Konig zu diesem Zeitpunkt nicht eintrifft' a Gydynrb [= CCJIH] xe6e yHHymb 3BaTH H TBI 6 xoro ne fltjian (Urk. 1572) 'Und wenn man dich auch auffordern wird, sollst du das nicht tun'

In der zweiten Halfte des 17. Jahrhunderts wird 6yjieTb zu 6yfle gekiirzt. Diese heute archaische Konjunktion war im 17., 18. und im 19. Jh. verbreitet. Der Gebrauch der Konstruktion 6yay + Part.praet.act. auf -JIT> geht im 15. und im 16. Jh. zuriick. Im 17. Jh. verschwindet diese Konstruktion vollig.

3 Natiirlicher grammatischer Wandel 3.1 Interaktion von Tempus und Aspekt in den Futurparadigmen des Altrussischen Die Kategorie des Tempus und die Kategorie des Aspekts interagieren im Altrussischen unterschiedlich im Bereich der Prasensparadigmen (= einfaches Futur) und im Bereich der zusammengesetzten Futura. In den Prasensparadigmen bestimmt die aspektuelle Kennzeichnung des Verbs seine temporale Interpretation. Die Prasensparadigmen der imperfektiven Verben vom Typ nnuiy 'ich schreibe' weisen die Bedeutung des Prasens (vgl. Beispiele (3) und (4)), die der perfektiven Verben vom Typ nanHniy 'ich werde schreiben1 die Bedeutung des Futurs (vgl. Beispiele (1) und (2)) auf. Abweichungen von dieser Tempuszuweisung sind kontextuell bedingt. Die Prasensformen imperfektiver Verben werden als Futur aufgefafit, wenn sie zusammen mit anderen Verben im Futur (vgl. Beispiele (5) und (6)) auftreten. Die Prasensformen perfektiver Verben werden als nichtaktuelles Prasens im generischen Kontext interpretiert:


(70)

(71)

r(ocnofl)b/?&«Mmb (pf. A.) OKOBaHHtin. r(ocnofl)t yMyffpfleTb (ipf. A.) cjiinua... (Simon. Psalter bis 1280) 'Der Herr entfesselt die Gefesselten, der Herr gibt Vernunft dem Blinden1 CtBepmeHaH 60 jiroGti BT>HT> u3Meiu,emb (pf. A.) 6ofl3Ht (Sage von Boris und Gleb: 14. Jh.) 'Ideale Liebe vertreibt die Angst'

Die zahlreichen aspektindifferenten Verben (sowohl Simplicia vom Typ HJTTH 'gehen' als auch prafigierte Verben vom Typ noxenH 'fliefien') konnten in Abhangigkeit vom Kontext als Prasens oder als Futur aufgefaBt werden (vgl. Beispiele (7) und (8)). Die Prasensparadigmen solcher Verben weisen eine Nicht-Differenziertheit des Funktionsbereichs der NichtVergangenheit in Prasens und in Futur auf. In den eigentlichen Futurparadigmen hangt die temporale Deutung des Verbs nicht von der aspektuellen Kennzeichnung des Verbs ab, d. h. die Kategorie des Tempus und die Kategorie des Aspekts funktionieren hier relativ autonom. Das Futur I und das Futur II werden von imperfektiven und von perfektiven Stammen gebildet: xony (Ha)nHcaxH 'ich werde schreiben', 6yay (Ha)riHcajn> 'ich werde geschrieben haben'. Das Futur I wird mit den Hilfsverben des pf. A. HMy und HaHLHy/no^bHy und des ipf. A. xo^ro, HMaMt (vgl. auch die perfektiven Entsprechungen zu xony noxoqy, Btcxony) gebildet. Es existiert allerdings eine Affinitat zwischen dem Tempus und dem Aspekt solcher Art, daB das Futur II hauptsachlich von perfektiven Verben gebildet wird.

3.2 Abbau der Markiertheit und Umstrukturierung der Futurparadigmen Die grammatischen Veranderungen im Sprachsystem verlaufen nach generellen Entwicklungstendenzen, die durch die Wirkung der Markierheitsprinzipien determiniert sind. Die Markiertheitstheorie ist ein unentbehrlicher Bestandteil des Konzepts des natiirlichen grammatischen Wandels, d. h. der Theorie des Sprachwandels, die den grammatisch initiierten Wandel erklaren will (vgl. Wurzel 1994: 32-39). Die grundlegende Annahme des Markiertheitskonzepts zum Sprachwandel ist als Prinzip des natiirlichen grammatischen Wandels formuliert. Diesem Prinzip zufolge verlauft grammatisch initiierter Wandel in Richtung der Beseitigung der markierten Fragmente des Sprachsystems. Das Wesen jedes natiirlichen grammatischen Wandels ist es, die Markiertheit grammatischer Erscheinungen im Sprachsystem abzubauen. Die Auspragung der Aspekte und die immer starker werdende Einbeziehung der Aspekte in das gesamte Verbalsystem haben entscheidend zur tiefgreifenden Umgestaltung des Tempussystems beigetragen. Der Funktionsbereich der Nicht-Vergangenheit wird jetzt endgiiltig in Prasens und in Futur aufgelost. Der natiirliche grammatische Wandel verlauft in Richtung der Herausbildung von zwei aspektsensitiven Futurparadigmen (des imperfektiven und des perfektiven Futurs) und der Beseitigung der alten markierten aspektindifferenten Paradigmen des Futurs I und des Futurs II. Das diese Entwicklung steuernde Prinzip des stufenweisen (oder geordneten) Markiertheitsabbaus (Vennemann 1988: 2-3) legt die Reihenfolge der grammatischen Verande- rungen fest. Zuerst geht die Form des Futurs II

390 Langue and Parole in Synchrony and Diachrony verloren, die gegeniiber dem Futur I markiert war. Danach wird die Form des Futurs I eliminiert, die dem neuen Futur gegeniiber markiert war.

3.3 Konkurrenz zwischen zusammengesetztem Futur I und zusammengesetztem Futur II. Schwund des Futurs II Das Futur II vom Typ 6y,ny (Ha)nncajit 'ich werde geschrieben haberi war gegentiber dem Futur I vom Typ HMaMt nariHcaTH 'ich werde schreiberi markiert: Es driickte eine zusatzliche Bedeutung der Handlung aus, die einer anderen Handlung vorangeht (Beispiele (43)- (45)). Im Laufe der Sprachentwicklung verliert das Futur II seine Bedeutung einer Handlung, die einer anderen Handlung in der Zukunft vorangeht, und wird in der gleichen Funktion wie Futur I gebraucht, namlich als Bezeichnung des absoluten Futurs. (72)

Gydeuib OTBirL uM^jib [= HM^TH], KHJDKC, o?Ke TH MOHacrupt paaT.rpaGirrb (Hypatius-Chronik 1425) 'Du, Ftirst, wirst eine Antwort haben, wenn man dein Kloster ausgepliindert haben wird'

Im Beispiel (72) driickt das Futur II im Hauptsatz die Bedeutung der Handlung aus, die der Handlung im Nebensatz folgt, d. h. es wird in der Bedeutung des Futurs I verwendet. Solche Entwicklung ist in den belorussischen und in den ukrainischen Sprachdenkmalern seit dem 14. Jh. nachgewiesen. Im Polnischen ist das zweite Futur mit der Bedeutung des ersten Futurs noch heute eine gebrauchliche Form: bedg pisat- neben b$d$ pisac 'ich werde schreiben'. Dieses relative Tempus verschwindet als markierte Form genauso wie das Plusquamperfekt im Subsystem der Vergangenheitstempora, das dem Perfekt gegeniiber markiert war. Im 15. und 16. Jh. schwindet der Gebrauch des Futurs II deutlich. Im 17. Jh. geht diese Form verloren.

3.4 Herausbildung der neuen Form des imperfektiven Futurs Die beiden Komponenten des Futurs II, namlich das Hilfsverb 6y;ry und das Part, praet. act. auf -jn>, funktionierten relativ selbstandig (vgl. Abschnitt 1.3). Bei solcher Konstellation der zusammenwirkenden Konstruktionen des Futurs II und des Futurs I konnte eine Verschmelzung folgender Art stattfinden. Das 1-Partizip im Futur II wurde durch den Infmitiv verdrangt. Andererseits wurden die Hilfsverben vom Typ HMy beim Futur I durch das Hilfsverb 6yay ersetzt. Die moderne Form des imperfektiven Futurs, das sich aus der konjugierten Form des Verbs 6y,ny und dem Infmitiv des imperfektiven Verbs zusammensetzt, konnte folglich als Kontamination der Konstruktion 6yfly + 1-Partizip und der Konstruktion HMy + Infmitiv zustandekommen (JIoMxeB 1961: 310-312). Das Hilfsverb 6ysy hatte keine eigene lexikalische Bedeutung und keine zusatzlichen modalen Bedeutungen. Die neu entstandene analytische Form des imperfektiven Futurs vom Typ Syjiy iracaTb 'ich werde schreiben1 tritt folglich als Parallele zur semantisch neutralen Form des perfektiven Futurs vom Typ Harraiiry 'ich werde schreiben' auf. Von imperfektiven Verben des Typs rmcaTH 'schreiben1 wird das analytische Futur des Typs 6y,ny rmcaTL 'ich werde schreiben' gebildet. Perfektive Verben vom Typ HanHcaxH 'schreiben' weisen das einfache Futur des Typs Harrauiy 'ich werde schreiben' auf.


3.5 Konkurrenz zwischen neuem Futur und zusammengesetztem Futur I. Verlust des Futurs I. Im 17. Jh., als das Futur II verschwunden war, waren die Konstruktionen mit dem Hilfsverb y^Hy (in der Kanzleisprache), mit dem Hilfsverb HMy (in der Literatursprache) und mit dem Hilfsverb crany (in der Umgangssprache) sehr verbreitet. Im Laufe des 18. Jh. werden die Formen des Futurs I durch die parallelen Formen des perfektiven und des imperfektiven Futurs verdrangt. Schon in den fruheren Denkmalern des Russischen werden das perfektive Futur und das Futur I parallel gebraucht: (73)

ame JIH ne n(o)zy6uuib. uMamb nozyGumu scero cionra (Laur.-Chr. 1377) 'wenn du [ihn] nicht vernichten wirst, wird er das ganze Agypten vernichten1

In der Gegeniiberstellung des Futurs I und des perfektiven oder des imperfektiven Futurs war das Futur I markiert: Es hatte unterschiedliche Hilfsverben (vgl. die Beispiele (73) und (74) mit HMtxH bzw. xorbTH) mit zusatzlichen modalen Bedeutungen, und es war aspektuell nicht festgelegt (das betrifft das Hilfsverb und den Infinitiv gleichermaBen). (74)

B CH JKQ speMHHa pOflHCH MOHctH Bt ^cHfltxi.. H3KC xomemb nozySwnu eionert (Laur.-Chr. 1377) 'in dieser Zeit wurde Moses in Israel geboren, der Agypten vernichten wird1

Die neuen Formen des Futurs hatten keine modalen Schattierungen und waren aspektuell eindeutig. In dieser Konkurrenz setzen sich die unmarkierten Formen des neuen Futurs durch.

3.6 Tempus und Aspekt in den Futurparadigmen des modernen Russischen Im Fruhaltrussischen funktionierten in den Futurparadigmen die Kategorie des Tempus und die Kategorie des Aspekts relativ unabhangig voneinander. Die temporale Interpretation einer Verform hangt grundsatzlich nicht von ihrer aspektuellen Kennzeichnung. Mit der Etablierung der neuen Futurparadigmen bestimmt im Subsystem der Nicht-Vergangenheit die aspektuelle Kennzeichnung des Verbs die Auswahl und die temporale Deutung seiner Formen. Imperfektive Verben haben das Prasensparadigma vom Typ nmiry 'ich schreibe1 und das Futurparadigma vom Typ 6yay rmcaTt 'ich werde schreiben1. Perfektive Verben weisen nur das Futurparadigma vom Typ Harmmy 'ich werde schreiben' auf. Imperfektive Prasentia vom Typ runny 'ich schreibe' bezeichnen im markierten Gebrauch das Futur, vgl. Beispiel (46) (b). Perfektive Prasentia vom Typ nannniy 'ich werde schreiben' treten kontextbedingt als nichtaktuelles Prasens auf, vgl. die Beispiele (75) und (76). (75) (76)

PaayM cmiy noGedwn 'Die Vernunft wird die Gewalt besiegen' OH Bceraa noMooicem B 6e.n;e 'Er wird immer helfen in der Not'

392 Langue and Parole in Synchrony and Diachrony Literatur FopniKOBa, K. B. H T. A. XaSypraeB (1981). HcmopunecKasi zpoMMamuKa pyccicozo xsbuca. MOCKBE Issatschenko, A. (1983). Geschichte der russischen Sprache. 2. Band. Heidelberg. KapaeaHOB, A. A. (1992). K sonpocy o xapaKrepe KareropHH npeaejibHOCTH apeBHepyccKOro rjiarojia. In: Bonpocu S&UKOSHOHUSI 6, 103-112. Kiparsky, V. (1967). Russische historische Grammatik. Band II. Heidelberg. KyaneiioB, IT. C. (1953): K sonpocy o reneaHce Btmo-BpeMCHHtix OTHOineHHH apeBHepycCKoro H3tiKa. In: Tpydbi HHcmumyma si3biK03Hanusi AH CCCPII. MocKBa, 220- 252. Ky3HeuoB, H. C. (1959): Onepicu ucmopunecKou Mopopojiozuu pycacoeo si3biKa. MocKBa. JIoMTeB, T. H. (1961). CpaenumejibHo-ucmopuHecKasi zpaMMamuica 513UKO8 (MopOpOJIOZUSL).

eocmoHHocjiaesiHCKux

MoCKBa.

Vennemann, Th. (1988). Preference Laws for Syllable Structure and the Explanation of Sound Change. Berlin/New York/Amsterdam. Vlasto, A. P. (1986). A Liguistic History of Russia to the End of the Eighteenth Century. Oxford. Wurzel, W. U. (1994). Grammatisch initiierter Wandel. (Bochum-Essener Beitrage zur Sprachwandelforschung, Bd. XXIII). Brockmeyer, Bochum.

29 POSSESSIVAUSSAGENIM DEUTSCHEN UND DIE AUXILIARISIERUNG VON HABEN1

Michail L. Kotin, State Pedagogical University, Moscow, RF

Abstract Die der deutschen Perfekt- bzw. Plusquamperfektform zugrunde liegende 'haben' Perspektive lafit sich aus der kognitiv erfafibaren Possessivrelation ableiten, welche ihrerseits den syntaktischen Archetyp der 'haben' - Periphrasen pragt. Dieser weist eine weitgehende Ambiguitat beztiglich der semantischen Deutung der Possessor-Funktion auf: Possessor kann namlich sowohl mit dem Agens der 'haben'-Perspektive zusammenfallen als auch damit divergent sein. Die Dominanz ersterer Lesart fuhrt die Irrelevanz des Merkmals der Transitivitat bei den funktionalisierten 'haben' - Aussagen herbei, was zusammen mit stufenweisem Abbau der resultativen Aktionalitat der einstigen 'haben' + Part. Prat. - Konstruktion die Auxiliarisierung des Verbums finitum und Grammatikalisierung der Gesamtfugung fbrdert.

ZUM GEGENSTAND Seit Brinkmanns 1959 Abhandlung iiber die haben-Perspekiive im Deutschen ist die Perspektivierung der Verbalsysteme auf der Grundlage der jeweiligen Konzeptverben fur das Deutsche und mutatis mutandis andere Sprachen mit analytischen Sprachbauelementen ein durchaus ublicher theoretischer Ansatz geworden (vgl. z.B.: Rupp 1967; Wierzbicka 1988; Kotin 1995). Allerdings fallen unter den Begriff ,,Perspektive" haufig unterschiedliche Potenzen der Verben vom Typ haben, sein, werden, bleiben u.dgl (vgl.: Rupp 1967,154 f., Lenz 1993u.a.). Offenkundig kann von einer ,,Perspektive" erst dann gesprochen werden, wenn das betreffende Verbum finitum in seiner Funktionsgeltung par excellence sowie als Fiigungs- bzw. Auxiliarverb wesentliche kognitiv bzw. konzeptuell erfafibare Bereiche abdeckt, die in letzter Instanz auf eine Sonderstellung dieses Verbs im Sprachsystem und - was noch wichtiger ist im Sprachbau schlieBen lassen. 1 Im vorliegenden Beitrag werden erste Ergebnisse des Forschungsprojektes zur Diachronie des deutschen Verbalsystems zusammengefaBt. Das Projekt im Ganzen sowie die Teilnahme des Autors an der 31.Jahrestagung der SLE wurden von der Alexander von Humboldt- Stiftung finanziert.

393

394 Langue and Parole in Synchrony and Diachrony Zweifelsohne sind es fur das Deutsche drei Verben - haben, sein und werden, welche den genannten Kriterien in jeder Hinsicht geniigen - sie bilden namlich konzeptuell wichtige Grundaussagen (Elementarpropositionen) mit resp. possessiver, existentialer und mutativer Funktion und dariiber hinaus (in Verbindung mit infiniten Verbalformen) Tempus-, Modusund Genusperiphrasen, die zum analytischen Sektor des Verbalparadigmas gehoren. Das Verb haben ist hierbei auf den Tempusbereich beschrankt, wahrend sein sowohl Tempus(Perfekt) als auch Genusfunktion (,,Stativ") ausiibt und werden in alien drei Spharen des grofien Verbalparadigmas angesiedelt ist (Futur, Passiv bzw. Konditionalisformen bildet). Im vorliegenden Beitrag wird nun versucht, die Entfaltung der haben-Perspoktive im Deutschen von der einfachen Possessivrelation bis bin zum Auxiliarverb mit Tempusfunktion nachzuvollziehen.

DAS FUNKTIONSRASTER DES VERBS HABEN IM HEUTIGEN DEUTSCH Die eingangs skizzierte Fragestellung setzt grosso modo eine diachrone Herangehensweise voraus. Dennoch bietet sich zunachst ein kurzer Uberblick iiber die Kernfunktionen des Verbs haben in dem synchronen Querschnitt der Gegenwartssprache als durchaus angebracht an, zumal dieser eine handfeste Basis fur spatere retrospektiv erworbene Evidenzen bildet. Sieht man zunachst vom jeweiligen kategorialgrammatischen Status des Verbs haben ab, konnen im allgemeinen folgende wesentliche Relationen festgestellt werden, die mittels haben versprachlicht werden: I. Possessivrelationen verschiedener Art 1. Besitzen von diversen Objekten (Eigentums- bzw. Verfugungsrelation). la)Erhat ein Auto. Ib) Die Bibliothek hat iiber 500000 Titel in ihrem Bestand. Ic) Er hat eine Professur in Prag. 2. Aktionsgerichtete Possessivrelation (,,haben" als Voraussetzung fur eine Handlung). 2a) Zum Fruhstuck haben wir heute Kaffee mit Schinkenbaguettes. 2b) Ich habe einen Auftrag. 2c) Kollege Muller hat wieder Probleme. II. Eigenschaftsrelationen verschiedener Art 3. Konstante Wesensmerkmale. 3 a) Der Mensch hat zwei Augen.

Possessiwelation und A uxiliarisierung von haben 395 3b) Der Tag hat 24 Stunden. 3c) Die Volga hat zwei grofie Nebenfliisse. 4. Individuelle konstante bzw. variable Merkmale 4a) Sie hat blaue Augen (graue Haare) / starke Nerven. 4b) Dieses Gebaude hat zwei Terrassen. 4c) Das Auto hat einen Katalysator. 5. Grundeigenschaften abstrakter Grofien. 5a) Dieses Problem hat nur eine Losung. 5b) Diese Krankheit hat in der Regel (einen) akuten Verlauf. 5c) iibertr. Liigen haben kurze Beine. 6. Temporare Merkmale (Eigenschaften bzw. Zustande von bestimmter Dauer). 6a) Er hat Grippe. 6b) Wir haben jetzt Ferien. 6c) Ich habe Mitleid mit dir. 7. Folgemerkmale absrakter Grofien. 7a) Grippe hat haufig schwere Komplikationen. 7b) Dein Benehmen kann unberechenbare Folgen haben. 7c) Der Film hatte (einen) groBen Erfolg. 8. konstitutive Merkmale, die einer abstrakten Grofle vorausgehen. 8a) Der I.Weltkrieg hatte viele Ursachen. 8b) Sein Benehmen scheint keinen Grund zu haben. 8c) Die Bewerbung um diese Stelle hat eine Reihe von Voraussetzungen. III. Aktionsgerichtete Relation 9a) Ich habe heute noch zwei Briefe zu schreiben. 9b) Er hat viel zu tun. 9c) Demnachst haben wir auch andere Probleme zu losen. IV. Temporale Relation in Verbindung mit dem Partizip II lOa) Er hat gestern zwei Briefe geschrieben. lOb) Er hat Gothes ,,Faust" gelesen [und kann nun die Prufung in Literaturgeschichte ablegen]. 10b-l) Er hat gestern zwei Stunden lang Goethes ,,Faust" gelesen.

396 Langue and Parole in Synchrony and Diachrony lOc) Er hat endlich seine Brille gefunden. 10c-l) Er hat gestern seine Brille lange gesucht. 1 Od) Wir haben uns fur morgen verabredet. lOe) Das kleine Madchen hat mitten in der Strafie gestanden und geweint. Affinitaten zwischen den aufgefiihrten Aussagetypen lassen sich - bis auf die Relation IV, bei der haben als Auxiliarverb fungiert, - relativ unproblematisch feststellen. Aber auch bei den Belegen 10a),10b) und lOc) (Satze mit resultativer Lesart) raumt selbst die idiosynkratische Periphrase gewisse Deutungsmoglichkeiten ein, die - wenngleich indirekt - auf eine inharente ,,Possessivitat" hinweisen. Lediglich die Belege 10b-l), 10c-l), lOd) und lOe) schlieflen jegilche ,,possessive" Interpretation von haben aus. Es handelt sich hierbei um Aussagen, in denen das Auxiliar haben entweder mit den Transitiva in durativer Lesart (10b-l) bzw. Durativa tantum (10c-l) oder mit den Reflexiva bzw. Intransitiva (10d,e) verwendet wird. Es ist nun anzunehmen, daB die primare possessive Funktion des Verbs haben bei dessen Auxiliarisierung zunachst gerade in Vebindung mit den transitiven Resultativa zum Tragen kommt und erst spater bei transitiven Durativa sowie zuletzt bei den Intransitiva neutralisiert wird.

DER SYNTAKTiscHE ARCHETYP DERPERFEKT- BZW. PLUSQUAMPERFEKTPERIPHRASE UND DESSEN BESONDERHEITEN Althochdeutsche Possessivsatze mit dem Partizip II als Attribut Die oben aufgefuhrte Funktionspalette des Verbs haben in Aussagen verschiedenen Typs erweckt zunachst den Eindruck eines nicht nur gerasterten, sondern auch graduierten Schemas mit stufenartiger Kategorisierung des mit jeder Stufe abstrakter wirkenden Verbums finitum. Dieser erste Eindruck darf jedoch auf keinen Fall tauschen, derm die unter IV angefuhrte und mit den Beispielen lOa-c belegte Relation ist historisch gesehen nicht etwa das Ergebnis einer stufenweisen Abschwachung der Eigensemantik des Verbs haben, sondern entspringt unmittelbar der ursriinglichen reinen Possessivaussage. Die Eigenschaftsrelationen sind durch andere Mechanismen in die Wege geleitet worden und brauchen bei der diachronen Interpretation der temporalen haben-Umdeutung zunachst nicht bemiiht zu werden. Im ahd. Tatian (um 830) finden sich neben einfachen Possessivsatzen mit haben drei Belege, in denen haben - wohlgemerkt als Vollverb - ein Objekt ftihrt, welches seinerseits ein Partizipialattribut bei sich hat. Zwei davon (11,12) sind eine wortliche Ubersetzung der entsprechenden lateinischen Konstruktion, einer (13) stellt hingegen eine Originalaussage dar, der in der lateinischen Vorlage ein einfaches Verbum finitum entspricht, was auf die

Possessivrelation und Auxiliarisierung von haben 397 Eigenstandigkeit der Possessivkonstruktion mil Partizipialattribut im Althochdeutschen schlieflen laBt, vgl.: 11) Tat. 102,2: Phigboum habeta sum giflanzotan in sinemo uuingarten. Lk.13,6: Arborem fici habebat quidam plantatam in vinea sua. 12) Tat. 151,7: senu thin mna, thia ih habeta gihaltana in sueizduohhe. Lk. 19,21: esse mna tua, quam habui repositam in sudario. 13)Tat.28,l: iogiuuelih, thie thar gisihit uuib sie zi geronne, iu habet sia forlegana in sinemo herzen. Mt. 5,28: omnis, qui viderit mulierem ad concupiscendum earn, iam moechatus earn in corde suo. Zunachst ist das Partizip II in seiner Funktion weitgehend einem Adjektiv ahnlich: Statt giflanzotan hatten in (11) theoretisch Adjektive wie z.B. grozan, sconan o.a. stehen konnen. Dennoch weist das Partizip eine unverkennbare Besonderheit auf, welche es grundsatzlich von einem Adjektiv unterscheidet, selbst wenn dieser Unterschied hin und wieder in den Hintergrund riickt. Es handelt sich um seinen inharenten, vom jeweiligen Verb vererbten ,,Handlungsbezug". Das Partizip II beschreibt namlich stets ein Merkmal, welches durch die damit verbundene Handlung vermittelt und darin aufgelost ist. Beim Handlungsbezug, der auch als ,,inharente Verbalitat" eingeordnet werden kann, sind nun aber solche Eigenschaften, wie Generitat, Aktionalitat, aber auch das Vorhandensein eines Handlungstragers unabdingbar. Wenn jemand in seinem Weingarten einen schonen oder grofien Feigenbaum hat, existiert das Merkmal unvermittelt; wenn aber der Feigenbaum gepflanzt ist, so fragt sich, \ver ihn gepflanzt hatte. In der Possessivaussage gibt es dabei grundsatzlich zwei Lesarten: 1) Possessor ist gleichzeitig Handlungstrager der involvierten Proposition und 2) Der Handlungstrager der involvierten Proposition ist nicht identisch mit dem Possessor. Die Tatian-Belege verhalten sich hinsichtlich dieses Kriteriums unterschiedlich: bei (12) und (13) ist die Identitat von Possessor und Agens evident, bei (11) hingegen ist die Lesart ambig.

398 Langue and Parole in Synchrony and Diachrony Die Auxiliarisierung von haben als ,,Funktionsaustausch" Die Identitat von Handlungstrager des passiven Partizips und Possessor der haben-Relaiion 1st fur die Auxiliarisierung des Verbs haben und die Herausbildung einer grammatikalisierten Verbalperiphrase eine conditio sine qua non. Bin indirekter Beweis dafur ist das Aufkommen der ,,haben"-Fiigungen vom Typ (14) Nun hat er endlich die email-Adresse von Dr.Herrmann in seinem Rechner eingeblendet mit der Mafigabe, dali die Handlung von einer anderen Person, als der Possessor, ausgefuhrt worden ist. Derartige Satze werden in der Literatur zu Recht als passivische Aussagen gedeutet (vgl. z.B. Eroms 1997, 30). Wichtig fur unsere Fragestellung ist hier aber, dafl das Verb haben dabei seine Eigenbedeutung vollig beibehalt und die Aussage im Ganzen eine einfache Possessivrelation darstellt. Wird aber der Possessor als Handlungstrager der involvierten Proposition mit dem Partizip II gedeutet, kommt eine Situation zustande, die als ein ,,Funktionsaustausch" sui generis bezeichnet werden kann: die Possessorfunktion des grammatischen Subjekts beim Verbum fmitum haben wird zugunsten der Agensfunktion beim Verbum infinitum (Partizip II) zuruckgestellt, und die passivische Handlungsausrichtung der ,,Partizip II - Perspektive" wird zugunsten der aktivischen Ausrichtung der Possessivrelation venvorfen. Also statt ,,Ich habe etwas, was (von mir oder von jemand anderem) gemacht worden ist" - ,,Ich habe etwas gemacht". Darauffolgende Prozesse - Verlust der Eigenbedeutung von haben, Idiomatisierung der Fiigung im Ganzen sowie Verwandlung des Partizips II aus dem Attribut ins nonfinite Teil der analytischen Verbalform - sind nur ein unmittelbares Ergebnis dieses ,,Funktionsaustauschs".

Die Aktionsart-Funktion der haben-Periphrase Die haben-Relaiion verhalt sich hinsichtlich der Aktionsart- bzw. Aspektfunktion ambig, d.h. einzelne Lesarten konnen abhangig von dem jeweiligen Kontext sowohl durativ bzw. immergultig als auch resultativ sein, vgl.: 15) Er hat zwei Autos, versus 16) Endlich habe ich nun ein Auto. Im Falle einer erweiterten Possessivaussage mit attributivem Partizip II liegt die AktionsartBedeutung eher bei dem Partizip. Nun ist aber auch das Partizip II im Deutschen aktional ambig - es kann abhangig von der jeweiligen Verbsemantik bzw. dem Kontext sowohl durative als auch resultative Lesarten haben (wie z.B. bei gesucht - gefunderi). Die ursprunglich wohl resultative Lesart der gesamten Fiigung haben+Pari.ll ergibt sich offenkundig aus der aktionalen Deutung von zwei aufeinander bezogenen verbalen Merkmalen

Possessorelation und A uxiliarisierung von haben 3 99 - dem Merkmal ,,besitzen" und dem Merkmal ,,tun", die in ihrer Kombination als Folge und Ursache auftreten: Jemand hat einen Feigenbaum, weil er gepflanzt worden ist. Die Ergebnisbezogenheit (Resultativitat) ist dabei ipso facto gegeben. Die resultative Lesart wird aber fakultativ, sobald der oben beschriebene ,,Funktionsaustausch" stattgefunden hat, denn das Verbum fmitum verliert damit seine Eigenbedeutung, wodurch nun die inharente aktionale Ambiguitat des Partizips II freigesetzt wird, da es nicht mehr an die Eigensemantik von haben gebunden ist. DaB dieser Wandel nicht nur das Verb haben betrifft, sondern schlechthin die (einstige) Possessivrelation, kann u.a. an der Temporalisierung der and. Fugung e/gaw+Part.II nachvollzogen werden, die zunachst als Konkurrenzperiphrase zur Fugung /zaien+Part.auftritt und spater zugunsten letzterer aufgegeben wird, vgl. die Belege aus Otfrid und Notker: 17) Eigun sie iz bithenkit (O.I,1,21); 18) Uuaz tu ... guotez ketan eigist. tes habest tu luzzel gesaget (N.I,26). Dariiber hinaus wird auch das urspriinglich obligatorische Merkmal der Transitivitat des jeweiligen Vollverbs zuriickgestellt, denn diese war ebenfalls nur durch die Eigensemantik des Verbs haben vorgegeben. Das Objekt des Besitzens in einer Possessivrelation verwandelt sich nun in ein Objekt der Handlung in einer temporal markierten Aktions-Relation; eine Handlung kann aber - im Unterschied zum Besitzen - auch ohne Objekt gedacht werden, vgl.: 19) dinen zorn furhtendo habo ih giuueinot (N. 1,26). 20) Wie heten so geworben die edelen bruoder min,/ daz ih eigen mannes wine solde sin (Nib. 14,822-823).

FAZIT Althochdeutsche Possessivkonstruktionen mit haben und eigan involvieren u.a. Aussagen mit dem Partizip II vt, das urspriinglich die Funktion eines Attributs zum direkten Objekt ausubte. Da die Possessivrelation ein gmndlegendes Charakteristikum von Propositionen dieser Art gewesen ist, konnen wir diese als kognitive Basis der Perfekt- und Plusquamperfektformen des Deutschen behandeln, welche sich gegen Anfang mittelhochdeutscher Zeit als ParadigmaBestandteil etablert haben. Die originare Possessivausage weist keine deutliche Evidenz beziiglich des Agens der jeweiligen Verbalhandlung auf: es kann namlich mit dem Agens der /za&ew-Perspektive zusammenfallen, aber auch damit divergent sein. Die Auxiliarisierung von haben innerhalb der behandelten Konstruktion setzt die Identifizierung des Possessors mit dem Agens der durch das Partizip II genannten Verbalhandlung voraus. Die Auxiliarisierung des Verbums fmitum fuhrt ihrerseits die Irrelevanz des Merkmals ,,Transitivitat" herbei, so daB intransitive Verben ebenfalls in die sich herausbildende

400 Langue and Parole in Synchrony and Diachrony analytische Tempusform zunehmend involviert werden und somit den Endpunkt ihrer Grammatikalisiemng manifestieren. Mit der Auxiliarisierung von haben und Paradigmatisierung der gesamten Fiigung haben+Partizip II geht weiterhin auch der stufenweise Abbau der resultativen Aktionalitat der einstigen possessiven /za&en+Part.II-Konstruktion einher.

LITERATUR Brinkmann, Henning (1959). Die haben-Perspektive im Deutschen. In: Sprache - Schlusselzur Welt. Festschrift fur Leo Weisgerber (H. Gipper, Hg.), S. 176-194. Schwann, Dusseldorf. Eroms, Hans-Werner (1997). Verbale Paarigkeit im Althochdeutschen und das ,,Tempussystem" im 'Isidor'. Zeitschrift fur deutsches Altertum und deutsche Literatur, 126,1, 1-31. Kotin, Michail L. (1995), Das Deutsche als ,,werden"-Sprache.Synchronie und Diachronie der werJew-Perspektive im deutschen Verbalsystem. In: Das Wort, Germanistisches Jahrbuch Rutland- Deutschland (A.EUeri, Hg.), S. 12-27. DAAD, Moskau. Lenz, Barabara (1993). Ausgewahlte Aspekte von werden, sein und bleiben. In: Theorie des Lexikons, Arbeiten des Sonderforschungsbereichs, 282,51,1-30. Rupp, Heinz (1967).Zum deutschen Verbalsystem. In:. Satz und Wort im heutigen Deutsch. Jahrbuch des Institutsfur deutsche Sprache 1965/66.(H. Moser, Hg.). Schwann, Dusseldorf, 148-164. Wierzbicka, Anna (1988). Why can you have a drink when you can't *have an eatl In:, The Semantics of Grammar, by Anna Wierzbicka. Chapter 5. Univ.Press, Amsterdam (Philadelphia).

QUELLEN Otfrids Evangelienbuch, hg. von Oskar Erdmann. Halle (S.) 1934. - O. Tatian, hg. von Eduard Sievers. Paderborn 1966. - Tat. Die Werke Notkers des Deutschen, I-X, hg. von E.H.Sehrt, Taylor Starck, J.C.King und P.W.Tax (Altdeutsche Textbibliothek 73). Tubingen 1972-1981. - N. Das Nibelungenlied, nach der Ausgabe von K.Bartsch, hg. von Helmut de Boor. 22, revidierte und von Roswitha Wisniewski erganzte Auflage. Mannheim 1988. - Nib.

30

DlCHOTOMISCHE ZUGRIFFE IM BEREICH DER AKTIONALITAT DES WESTGERMANISCHEN (IM VERGLEICH ZUM SLAWiscHEN)1 Michail L. Kotin, State Pedagogical University, Moscow, RF

Abstract Das germanische Verbalsystem weist seit der Fruhzeit seiner Geschichte unverkennbare aspektuale Zugriffe auf. Bereits im Friihgermanischen kann von der grammatischen Kategorie des Verbalaspekts gesprochen werden mit der Mafigabe, dafi unter Verbalaspekt nicht die hauptsachlich aus dem Slavischen stammende Opposition von kategorialen Merkmalen perfektiv - imperfektiv verstanden wird, sondern vielmehr eine in der jeweiligen Sprache paradigmatisierte, d. h. konsequent dichotomisch organisierte und strukturierte Aktionsartfunktion. Das Germanische bemiiht bei der Herausbildung des Verbalaspekts wohl andere kategoriale Merkmale als das Slavische, namlich die Opposition von mutativer und nichtmutativer (stataler) Funktion, welche aber spater im Westgermanischen zugunsten anderer Oppositionen aspektualen bzw. nichtaspektualen Charakters verworfen wird.

FORSCHUNGSLAGE UND FRAGESTELLUNG Ich fange mit folgenden - wohlgemerkt ostgermanischen!- Belegen an:

got.: (1) Joh.10,20: qel>unuh managai ize: unhull>on habail> jah dwalmol>; hua £amma hauseiP7 ,,Und viele von ihnen sagten: Er hat einen bosen Geist und rast: was hort ihr Ihn?" (2) Mt.l l,2:il> Johannes gahausjands in karkarai waurstwa Xristaus, insandjands bi siponjam seinaim qal> du imma... ,,Als Johannes im Kerker von Christi Taten gehort hatte, schickte er seinen Schuler, und (dieser) fragte Ihn..."

1 Der Autor bedankt sich bei der Alexander von Humboldt-Stiftung fur die finanzielle Unterstiitzung seines Forschungsprojekts und seiner Teilnahme an der Jahrestagung der Societas Linguistica Europeae in St Andrews, wo dieser Beitrag im Workshop grammatical Categories" vorgestellt wurde.

401

402 Langue and Parole in Synchrony and Diachrony (3) Mt.l 1,4-6:... gaggandans gateihiP lohanne I>atei gahauseiPjah gasaihuiP: blindai ussaihuand, jah haltai gaggand, I>rutsfillai hranjai wairPand, jah baudai gahausjand, jah dau£ai urreisand, jah unledai wailamerjanda: jah audags 1st huazuh saei ni gamarzjada in mis ,,geht und sagt Johannes, was ihr gehort und gesehen habt: Blinde werden sehend,und Lahme gehen, Aussatzige werden rein,und Taube bekommen Gehor,und Tote erstehen auf, und Arme verkunden, und selig ist, wer nicht Argernis nimmt an Mir". Die funktionale Ambiguitat des Prefixes ga- ist selbst bei einer oberflachlichen Belegmusterung unverkennbar. Offenkundig standen die gotischen ,,ga-Composita" in Opposition zu den entsprechenden Simplizia, aber diese Gegeniiberstellung scheint sich jeglicher Definition zu entziehen. Die ,,klassische" aspektuale Dichotomic von Perfektivitat und Imperfektivitat, wie wir sie aus dem Slawischen kennen, kann dabei nur tauschen - und die Polemik mit Streitbergs 1891 Aspekt-These hat dies deutlich gezeigt (vgl.: Mourek 1890; Maslov 1984; Makovskij 1955; Kacnel'son 1960; Smirnickaja 1966; Kotin 1989 u.a.). Seit Streitbergs Abhandlung zu dem ,,Aspekt" im Gotischen wurde die Aspektproblematik vorrangig gerade bei dem Sprachvergleich - zum Mittelpunkt lebhafter Diskussionen. In der Germanistik hat sich zunachst die Auffassung durchgesetzt, im Deutschen sei kein Aspekt vorhanden, welcher mit dem slawischen Verbalaspekt vergleichbar ware - auch sprachhistorisch wurden aspektuale Zugriffe weitgehend vernachlassigt. Einen Wandel dieses theoretischen Imperativs hatte die These von J.Kurylowicz 1964 einbringen konnen, dafl die Kategorie des Verbalaspekts zwar eine universaltypologische Dimension hat, aber in ihrer konkreten Beschaffenheit eher idioethnischer Natur ist und daher in verschiedenen Sprachen unterschiedlich strukturiert und manifestiert werden kann. Dieses Herangehen an den Verbalaspekt ist aber in der Germanistik weitgehend ohne Belang geblieben, und die Aspektualitat wird auch nach wie vor auf dem Hintergrund der anerkannten Aspektsprachen wie Russisch oder Polnisch gesehen. Was nun die kategoriale Funktion des gotischen Prafixes ga- betrifft, ist zunachst evident, daB es nicht verglichen werden darf z.B. mit der Funktion der russischen Vorsilbe s-(so-) als morphologisches Perfektivum-Marker bei den kategorialen Verbaloppositionen vom Typ igrat'-sygrat', delat'-sdelat' u.dgl. Andererseits ist aber auch nicht zu bezweifeln, daB ein Unterschied zwischen gotischen Simplizia und ga-Komposita besteht, welcher nicht nur - und sogar nicht vorrangig lexikalischer Natur ist, sondern gewisse dichotomische Zugriffe zum Ausdruck bringt, welche einer eingehenden Untersuchung bedurfen und auf jeden Fall komplizierter sind als die Gegenuberstellung der aspektualen Merkmale durativ - resultativ, perfektiv-imperfektiv oder telisch-atelisch, welche in einschlagiger Literatur als kategoriale Aspektbedeutungen fur das Slawische und mutatis mutandis andere Sprachen gelten. H.-W.Eroms 1997 schreibt zu Recht von dichotomischen Zugriffen im Bereich der Verbalbedeutungen, die gerade die aspektuale Sphare des Germanischen betreffen, wobei die

Dichotomische Zugriffe im Bereich der Aktionalitat 403 hier seit altester Zeit vorhandene ,,verbale Paarigkeit" welter gefafite Oppositionen widerspiegelt, als ,,lexikalisch verfestigte Handlungsperspektivierung" (vgl.: Eroms 1997,26).

ZUR DEFINITION DER GRAMMATISCHEN KATEGORIE DES VERBALASPEKTS Wie laflt sich nun diese dichotomische Paarigkeit in Termini der kategorialen Grammatik beschreiben? Dafur ist zunachst eine entscheidende Trennung von dem hauptsachlich aus dem Slavischen stammenden Aspektmodell unabdingbar. Und dies hat auch zu bedeuten, dafi wir auf jeden Fall einer von dem heutigen terminologischen Usus abweichenden Aspektdefinition bediirfen. Ich folge dem grundlegenden Ansatz von J.Kurylowicz (s.o.)iiber idioethnische Natur der Kategorie des Verbalaspekts, wenn ich den Verbalaspekt als eine in der betreffenden Sprache paradigmatisierte, d.h. konsequent dichotomisch organisierte und strukturierte Aktionsartbedeutung definiere. Welche konkreten aktionalen Komponenten dabei in ihrem Status ,,erhoht" werden, hangt von der jeweiligen Sprache ab. Aspektuale Gegensatze konnen - wie im Russischen Unterschiede zwischen Abgeschlossenheit (Perfektivitat) und Nichtabgeschlossenheit (Imperfektivitat) der Handlung ausdriicken; sie konnen aber auch - wie im Englischen - die Dichotomic von ,,durativ" und ,,immergultig" bzw. ,,indefmit" bezeichnen; wichtig ist nur, daft die an die jeweilige Dichotomic gebundene Kategorie tatsachlich grammatisch-morphologische und nicht etwa kategorialsemantische Gegensatze in der verbalen Handlungsperspektivierung manifestiert.

ZUM ,,VERBALASPEKT" IM GERMANISCHEN Das Gotische Aus den o.a. gotischen Belegen ist sehr gut ersichtlich, daB dem Prafix ga- eine gewisse kategorialgrammatische Funktion zukommt, welche die Ableitungen mit ga- von den entsprechenden Simplizia abhebt. Es ist aber zugleich auBerst schwer, diese Funktion auf einen kategorialgrammatischen Nenner zu bringen. Die Streitbersche Tradition sieht darin einen nahezu ,,klassischen" Fall der Perfektivitat, was von vielen Forschern - von Mourek bis Maslov und Makovskij - zu Recht angefochten wird. Kacnel'son, Smirnickaja u.a. versuchen die kategoriale Funktion des got. ga- (und seiner Entsprechungen in westgermanischen Sprachen) als ,,sukzessiv" zu beschreiben, worunter ein ,,besonders erfolgreicher bzw. vollkommener AbschluB der Verbalhandlung" verstanden wird (vgl.: Kacnel'son 1960,338 ff.; Smirnickaja

404 Langue and Parole in Synchrony and Diachrony 1966,3ff.). Diese Auffassung hat den Vorteil, daB darnit die urspriingliche soziative Bedeutung von ga- (wie bei got. qiman ,,kommen" - gaqiman ,,zusammenkommen; sich versammeln") zumindest als motivierender Archetyp - mit erfaBt wird. Die Belegmusterung zeigt jedoch, daB got. ga- kategorialgrammatisch gesehen eher den Zustandswechsel ausdriickte und somit als mutativ einzuordnen ist - zum Terminus und Wesen der Opposition von Zustand (Merkmal 'statal') und Zustandswechsel (Merkmal 'mutativ') vgl. z.B.meine Ausfuhrungen in Kotin 1998,89ff. Im Beleg (3) tritt diese kategoriale Bedeutung besonders deutlich in Erscheinung, werden hier doch neben ga- seine vollen funktionalen Synonyme mit eindeutig mutativer Funktion der Verbalprafixe und insbesondere die syntaktische Ftigung mit dem klassischen mutativen Verb wairPan „werden" verwendet -... blindai ussaihuand,... Prutsfillai hrainjai wairPand,... baudai gahausfand,.. .dauPai urreisand... . Freilich gibt es daneben Belege, wo ga- auch in anderen Funktionen - vor allem als Marker der Perfektivitat - auftritt -vgl. die Verwendung von gain gateihiP, gahauseiP, gasaihuiP (3); gahausjands (2) vs. hauseiP. Unverkennbar scheint also eine gewisse ,,aspektuale Dualitat" durchzuschlagen, welche auf dem Hintergrund der allgemeinen Entwicklungsperspektive der germanischen aspektualen Opposition eher auf deren spateren Verfall als auf Stabilisierung hindeutet.

Das Althochdeutsche In meinem Artikel iiber die Rolle der aktionalen Bedeutungen bei der Herausbildung des deutschen Verbalparadigmas (Kotin 1995) habe ich versucht, nachzuweisen, daB auch im Althochdeutschen das Prafix gi- vorrangig mutative Aktionalitat kategorisiert - wie dies neben vielen anderen Belegen - z.B. aus dem folgenden Vers aus dem Hildebrandslied ersichtlich ist: (4) Hiltibrant grniahalta (Heribrantes sunu): her uuas heroro man / ferahes frotoro; her fragen g/stuont / fohem uuortum, hwer sin fater wari / fireo in folche... (Hilt.7-10). Offensichtlich bedeutet hier gi- in gimahalta nicht ,,sagte" oder ,,sprach" - wie dieses Verb haufig ins Neuhochdeutsche iibersetzt wird - sondern ,,fing an zu sprechen", ist also ingressiv bzw. inchoativ zu deuten (was ohne Zweifel eine Abart der mutativen Bedeutung ist). Fur diese Lesart spricht u.a. die Tatsache, daB das Objekt der Aussage weiter nicht genannt wird: man kann ja nicht ,,gesagt bzw. gesprochen haben", ohne zu erwahnen, was gesagt bzw. woriiber gesprochen worden ist(naheres vgl.: Kotin 1995, 38). Auch die Verwendung \ongi- in gistuont ,,begann, fing an" weist - wenngleich indirekt - auf die Pradestiniertheit des Prefixes gi- zur Markierung inchoativer bzw. mutativer Bedeutungen, zumal es in Verbindung mit dem Verb stantan auftritt, welches hier auch als Simplex (in der Bedeutung ,,anfangen") inchoativ ist. Wie im Gotischen, gibt es auch im Althochdeutschen neben gi- andere Mittel des Ausdrucks der kategorialen Dichotomic mutativ:statal. Freilich konnen auch andere Verbalprafixe - wie

Dichotomische Zugriffe im Bereich der Aktionalitat 405 z.B. fir- oder ir- - aspektahnliche Funktionen ubernehmen. Allerdings hat es dabei nie eine deutliche und widerspruchsfreie Aspektkennzeichnung gegeben, da die Verbalprafixe darunter ubrigens zunehemend auch das Prafix gi- - stets neben aspektualen auch andere, eher in der lexikalischen Sphare der Verbalbedeutungen angesiedelte Funktionen ausubten. Hinzu kommt die dem Gotischen ahnliche aspektuale Ambiguitat des Prafixes gi-, das neben mutativen auch perfektive bzw. resultative Funktionen iibernehmen konnte. Dies hat jedoch nicht zu bedeuten, daB damit jegliche aspektuale Zugriffe im Bereich des ahd. Verbalsystems schon sehr friih verworfen wurden. Es ist Eroms Recht zu geben, wenn er bei der Beschreibung der Verbalkategorien von der Verbalpaarigkeit auf alien Ebenen des Sprachsystems ausgeht, d.h. z.B. bei der konkreten Fragestellung nach den Aspektbedeutungen neben Funktionen der Prafixe auch die Funktionsgeltung syntaktischer Konstruktionen hinterfragt, falls diese ihrerseits eine dichotomisch zu deutende Paarigkeit aufweisen (vgl.: Eroms 1997,4 £). Fiir das Deutsche - und in gewisser Hinsicht generell fur das Germanische kommen dabei augenscheinlich Verbalperiphrasen mit sein, werden und - im Westgermanischen - haben in Frage. Sieht man von einigen - zugegeben haufig nicht unwichtigen! - Spezifika dieser syntaktischen Konstruktionen ab, kann festgestellt werden, da!5 sie ebenfalls an aspektual relevanten Oppositionen teilnehmen, die mit den oben beschriebenen dichotomischen Relationen von Simplizia und Komposita vergleichbar sind. Die Parallelitat ist hier dermaBen auffallig, daB sie sich bis in die ,,Details" verfolgen laflt. Einerseits ist der aspektuale Zugriff bei der Opposition von uuerdan+Part.I bzw.Part.II versus uuesan+Part.I bzw.Part.II als Gegenuberstellung von kategorialen Merkmalen mutativ:statctl unverkennbar; andererseits manifestiert die Konstruktion haben/eigan+P&ri.ll in deren Opposition zur Verwendung von Simplizia im Prasens und Praterit die ebenfalls aspektual aufzufassende Dichotomie von den Merkmalenperfektiv : imperfektiv, vgl. folgende Belege: (i) uuerdan+Part.I/II vs. uuesan+Part.I/Il [mutativ vs. statal] Prasens: (5) Tat. 2,9: Inti nu uuirdist thu suigenti inti ni maht sprehhan unzan then tag, in themo thisu uuerdent, bithiu uuanta thu ni giloubtus minen uuortun, thiu thar gifultu uuerdent in iro ziti. (6) Is. 14,9: Gotes gheist ist sprehhendi dhurah mih endi siin uuort ferit dhurah mine zungun. (6a) Tat. 164,6: ... therde mih minnot ther ist giminnot fon minemo fater. Praterit: (7)0.1,9,27: tho ward mund siner sar sprechanter. (7a) Tat. 3,1: In themo sehsten manude gisentit uuard engil Gabriel fon gote in thie burg Galilee. (8)Tat. 2,10: Inti uuas thaz folk beitonti Zachariam, inti wuntoroton thaz her lazzeta in templo. (8a) Tat.230,1: ...iz spato uuas thes selben tages ... inti duri uuarun bislozzan.

406 Langue and Parole in Synchrony and Diachrony (ii)haben/eigan+Part.ll vs.Simplizia [perfektiv vs. imperf.] Prasens: (9) Tat. 28,1: iogiuuelih, thie thar gisihit uuib sie zi geronne, iu habet sia forlegana in sinemo herzen. (10)0.1,25,11: Laz iz sus thuruhgan, so uuir eigun nu gisprochan. (1 l)Tat.!43,6: Thiu ih sprihu, soso mir ther fater quad, so sprihu. Praterit (12) Tat. 151,7: senu thin mna, thia ih habeta gihaltana in sueizduohhe. (13) Musp.35-36: dar seal er vora demo rihhe daz rahhu stantan,/ pi daz er in uuerolti eo kiuuerkot hapeta. (14) Musp.30:... uuanta hiar in uuerolti after ni uuerkota. Bei der Gegeniiberstellung von den Merkmalen mutativ - statal versucht das Althochdeutsche des ofteren - in deutlichem Gegensatz zum Lateinischen - aspektahnliche Oppositionen selbst dort zum Ausdruck zu bringen, wo sie in der lateinischen Vorlage gar nicht bemuht werden, vgl. in jeder Hinsicht kennzeichnende Stelle aus dem Anfang des Johannes-Evangeliums: (15) Tat. 1,2: Alliu thuruh thaz [uuort] wurdun gitan inti uzzan sin ni uuas uuiht gitanes thaz thar gitan uuas. Joh. 1,3: Ornnia per ipsum facta sunt et sine ipso factum est nihil quod factum est.

DIE WEITERENTWICKLUNG IM WESTGERMANISCHEN Die Opposition der aspektualen Merkmale mutativ vs. statal im Deutschen wird bereits zum Ausgang althochdeutscher - Anfang mittelhochdeutscher Zeit verworfen. Bereits seit Notker und erst recht seit fruhmhd. Periode wird der Akzent bei werden in Verbindung mit Partizipien deutlich auf den Vorgangsbezug verlagert: Das ,,Werden" wird namlich nicht mehr als sprunghafter Ubergang von Status A zum Status B, sondern zunehmend als allmahliche Entwicklung gedeutet, wodurch an Stelle der mutativen die prozessuelle Komponente tritt. Die aspektuale Markiertheit eriibrigt sich damit durch Uberlappungen mit der einst kontraren Fiigung uuesan+Pait.l/ll. Divergente aktionale Lesarten sind seither nur durch aktionale Semantik der jeweiligen Partizipien bzw. den Kontext moglich, vgl.:

Dichotomische Zugriffe im Bereich der Aktionalitat 407 (16) N. 1,75,30-76,4 (2,18): Ter uudsfilo hdrto geeret. temo ddz lob uudrtpeuolen zetiionne.dlde der scdz zespendonne. Ter sigo habet namen fone dien signis. So dien hostibus uuerdent tie signa genomen . taz heizet sigo nemen. ,,Derjenige war sehr stark geehrt, dem es befohlen wurde, lobzupreisen oder die (erbeuteten) Schatze zu spenden. Der Sieg hat seinen Namen von den signis. Wenn den Feinden die signa genommen werden, heifit es Sieg nehmen". Die Inaktiv-Periphrase mil uuesan/sm riickt nun zunehmend in den Hintergrund des sich herausbildenden analytischen Passivs und erhalt in der Peripherie der neuhochdeutschen Passiv-Diathese einen neuen Status - sie ist namlich aktional (resultativ) markiert, hat aber keine Entsprechung im Aktiv-Bereich und kann deshalb wohl kaum als Aspekt im eigentlichen Sinn interpretiert werden, vgl.: (17) Die Arbeit ist/war abgeschlossen; (18) Das Problem ist/war gelost; (19) Das Konto ist/war eroffnet. Auch die - schon im Althochdeutschen periphere - Gegenuberstellung von aspektualen Merkmalen perfektiv - imperfektiv verschwindet allmahlich durch ,,Temporalisierung" der haben-Periphrase, was ihre aspektualen Ambiguitat herbeiruhrt, vgl.: (20) Ich habe gestern zwei Stunden lang ferngesehen; (21) Er hat sie den ganzen Tag gesucht versus (22) Ich habe den Fernseher ausgeschaltet; (23) Er hat sie endlich gefunden. Die Dichotomisierung von Simplizia vs. gi-/ge- -Prafigierungen nach demselben aspektualen Modell (mutativ vs. statal) wird ebenfalls bereits im Fruhmhd. aufgegeben; die aspektahnliche Opposition perfektiv vs. imperfektiv lebt nicht viel langer fort - die restlichen Verbpaare unterscheiden sich im heutigen Deutsch lediglich rein semantisch (wie horen und gehoren; horchen und gehorchen; fallen und gefalien oder stilistisch wie leben und geleben; pflegen und gepflegen). Mit dem Verlust des dichotomisch orgam'sierten Oppositionssystems im Bereich der Verbalperiphrasen mil dem Partizip II verschwinden logischerweise auch die Verbindungen von ,,sein" und ,,werden" mit dem Partizip I, derm in der Aktiv-Diathese wurden sie in ihrer aktionalen Funktion einfachen Simplizia ahnlich. Im Englischen aber, wo werden-Konstruktionen bereits im Hohen Mittelalter verworfen werden, kann die Fugung 6e+Part.I nicht nur erhalten bleiben, sondern auch ein voiles Aspekt-

408 Langue and Parole in Synchrony and Diachrony Paradigma mit dem kategorialen Grundmerkmal ,,durativ" ausbilden, welches dem indefinite!! Aspekt gegeniibersteht: (24) He is/was reading the ,,Times". versus (25) He reads/read the ,,Times". Aber auch die Konstruktion have+Part.II behalt im Englischen - im Unterschied zum Deutschen - wesentliche aspektuale Merkmale bei, vgl.: (26) He has already written both letters. Doch nicht: (27) * Yesterday afternoon he has walked for two hours.

LlTERATUR Eroms, Hans-Werner (1997), Verbale Paarigkeit im Althochdeutschen und das ,,Tempussystem" im 'Isidor'. Zeitschrift fur deutsches Altertum unddeutsche Literatur 126/1, 1-31. Kacnel'son, Solomon D. (1960), Drevneislandskije sukzessivnyje casticy of\ um. In Voprosy grammatiki (Viktorija N. Jarceva - Vladimir G.Admoni, Hg.), S. 331-345. Nauka, MoskauLeningrad. Kotin, Michail L. (1989), Wortbildung und Bedeutungsentwicklung (am Beispiel einer diachronen Untersuchung von Verben mit dem Prafix ge-). In: Das Wort. Germanistisches Jahrbuch UdSSR-DDR, S. 18-24 Moskau. Kotin, Michail L. (1995). O roli akcional'nych znacenij v formirovanii nemeckoj glagol'noj paradigmy. Izvestija Akademii naukRF, Serija literatury ijazyka 54/4, 36-45. Kotin, Michail L. (1998). Die Herausbildung der grammatischen Kategorie des Genus Verbi im Deutschen. Eine historische Studie zu den Vorstufen und zur Enstehung des deutschen Passiv-Paradigmas. Buske, Hamburg. Kurylowicz, Jerzy (1964). The Inflexional Categories in Indo-European. Winter, Heidelberg. Makovskij, Mark M. (1955). Funkcii i znacenija glagol'nogo prefiksaga- v gotskom jazyke. Autoreferat der Dissertation. Akademija nauk SSSR, Moskva. Maslov, Jurij S. (1984). Ocerki po aspektologii. Nauka, Leningrad Mourek, V.E. (1890). Sintaksis Gotskich predlozek. Prag. Smirnickaja, Svetlana V. (1966). Funkcional'noje razvitije prefiksage- v nemeckom jazyke. Autoreferat der Dissertation. LGU, Leningrad. Streitberg, Wilhelm (1891). Perfektive und imperfektive Actionsart im Germanischen. Beitrage zur Geschichte der deutschen Sprache und Literatur (PEE) 15, 70-177.

Dichotomische Zugriffe im Bereich der Aktionalitdt 409 QUELLEN UND DEREN AfiKURZUNGNEN Diegotische Bibel. Teil I. Hrsg. von Wilhelm Streitberg. S.Aufl. Heidelberg 1965. - got. Der Althochdeutsche Isidor, Hg. von George A. Hench. StraBburg:Karl J. Triibner 1893 [Faximile-Ausgabe des Pariser Codex nebst Chritischem Texte der Pariser und Monseer Bruchstiicke] (neuere Ausgabe von H.Eggers. Tubingen 1964) - Is. Muspilli, in Wilhelm Braune u. K.Helm (Hg.), Althochdeustches Lesebuch. lO.Aufl., 1-3, Halle(S.):Mitteldeutscher Verlag 1942. - Musp. Die Schriften Notkers und seiner Schule, hg. von Paul Piper. 1 .Bd. Schriften philosophischen Inhalts. Freiburg und Tubingen: Akad. Verlagsbuchhandlung von J.B.C. Moor (l.Lieferung:Boetius de Consolatione (1882; 2.Lieferung: Categorien; De interpretatione;Kleinere Schriften; Martianus Capella 1,11; Anhang 1882). III.Bd.Wessobrunner Psalmen,Predigten und katechitische Denkmaler,1895 (Freiburg und Leipzig:Akad. ...).-Seiten-und Zeilenzahlung nach dieser Ausgabe (neuere Ausgabe Notkers des Deutschen Werke nach den Handschriften neu herausgegeben von E.H.Sehrt, Taylor Starck, James C. King und P.N.Tax, TubingenrNiemeyer 1979)- N. Otfrids Evangelienbuch Hg. und erklart v. Oskar Erdmann. Halle a.S.:Verlag der Buchhandlung des Waisenhauses 1882 (neuere Ausgabe 1934) - O. Tatian. Lateinisch und altdeutsch mit ausfuhrlichem Glossar. Hg. von Eduard Sievers. Paderborn:Verlag von Ferdinand Schonlingh 1872 (neuere Ausgabe - 1966) - Tat.


31

THE MYSTERY OF CONSONANTAL NOMINAL STEM-BUILDING MARKERS IN ANCIENT GERMANIC Olga Ossipova, Tomsk, Russia

Abstract The consonantal stem-building markers in Indo-European served both wordformation and inflection. In ancient Germanic they preserved traces of the expression of animateness, e.g. nomina agentis and all animate things in the perception of ancient people. The primary meaning of nomina agentis was that of a possessive nature. The same stem-building markers were especially apparent in adjectives. All of them go back to deictic particles. Another of their functions is supposed to be to mark the primary active case. In this hierarchy their most recent expression is fixed as the expression of definiteness in Gothic.

1. It is known that consonantal declensions appeared earlier than vowel declensions (thematic ones) in ancient Indo-European (I-E) languages (Tronski, 1967 a : 79; Hirt, 1925 : 266; Haudry, 1988 : 56). Consonantal stem-building markers served both wordformation and inflection in declensions (in paradigms). As markers of word-formation they are fixed in nomina agentis, e.g. Goth, nuta "a fisherman", aizasmitha "a smith", gajuka "a friend" (n-stems); fathar "father", brothar 'brother" (r-stems); aims "an ambassador, messenger", barbarus " not a Greek, pagan", wairdus " a host" (u-stems); O.Icel. kappj"a warrior", bui" a neighbour" (n-stems); A.S. thearfa " a beggar, a poor man", wita " a wise man" (n-stems) etc. In such words consonantal markers are not only the suffixes of nomina agentis, but are seen all through the paradigm as stem-building markers. One thing is evident, that they were used predominantly in oblique cases. So the correlation in practice was between the oblique case and the nominative case, where they were not used at the beginning and only penetrated into it later on as for example in A.S. th-stems. Thus there is ground to suspect that their primary function included something else, which will be revealed further. 2. All consonantal stem-building suffixes served not only to mark nomina agentis but nouns signifying all animate things. This was proved by our investigations of ancient Germanic material (Ossipova, 1980). Of course, there were some attemps to connect each consonantal stem-building marker with the concrete lexical meaning of various groups of 411

412 Langue and Parole in Synchrony and Diachrony nouns, e.g. with names of relationship (I-E suffix -er-, -ter-), with names for animals (I-E suffix -bho-) (Brugmann, 1892: 430), or with paired things (Cuny, 1924 : 450) etc. In our opinion, all these attempts were doomed to failure. In the course of development any I-E language suffered great changes, and many words changed their meaning or went over from one lexical group into another. The most reliable groups are terms of relationship, parts of the body and names of living beings in general. All these groups are marked by consonantal stem-building suffixes in different I-E languages (Specht, 1947 : 5, 103, 293 etc., Ossipova, 1990 : 153-171, Shields, 1990). No doubt all these things were animate in the perception of ancient people, to which they could attribute many natural phenomena, and everything that could move or perform actions (Shields, 1988 : 228-229 ), e.g. Goth. /b/?"fire", rinno" stream", O.Icel. logi"a flame", O.H.G. aro"an eagle", Goth, fola, O.H.G. foJo, "a foal", hano" a cock", O.Icel. api'"a monkey" etc. Sometimes the meaning of the Germanic word does not fit in with this scheme. In this case etymological analysis helps a great deal to restore its earlier meaning to prove that it occurred in the consonantal declension not haphazardly. For example the Gothic word for material, cloth, is fana (n-stem, m.). In other Germanic languages it possesses nearly the same meaning: O.Icel. fana, O.Fries. fona, O.H.G. fano, O.Sax. fano "material, tissue". Such ancient languages as Latin and Greek give us one more meaning that belonged to the word of the same root: Gk. Tinvo^ (m\'v) preserved the meaning "biological tissue, spider's web" (Feist, 1920:105). All of them go back to the I-E root PANA- " spider's web" (Pick and Stokes, 1890 : 400). Thus the primary meaning of this root was associated with a living organism. It was transferred to non-living things in the same way as living organisms correlate with plants in languages of active typology, e.g. blood and juice, ear and leaf etc. (cf. Klimov, 1977 : 112-113). The Gothic word gibla "pediment" refers to n-stems, m., it is marked by a stem-building suffix like other nouns of animation. Its reference to n-stems is not accidental. The word shows its former relation to something animate, as if it possessed its own memory. And in fact if we go back and trace its origin, it will lead us to the meaning "head", which is found in the parent word in Greek KscpaA/n' "head" ( Uhlenbeck, 1896 : 61). Naturally not all the exceptions can be deciphered in this way. There are some nouns which express abstract notions or at least not concrete things that are marked by the same stem-building suffixes. What do such names and terms of relationship, parts of the body, nomina agentis and other living beings have in common? 3. The answer is connected with markers of possessivity. In fact the primary meaning of consonantal stem-building markers of nomina agentis is of a possessive nature (Whitney, 1889 : 140; Burrow, 1976 : 127). For example Skt. i/dar "otter" is interpreted as "the animal referring to water, a water animal"(Burrow, 1976 : 114-115). Even the word pitar "father" can be interpreted as "defender", cf. Skt. pa-(y) "to defend" (Burrow, 1976 : 133). Burrow is inclined to think that it is possible to restore the adjectival character of such substantives as Skt. gaus " cow", w " bird", sunu "son" (Burrow, 1976 :236-237). The possessive nature of these stem-building markers is especially apparent in adjectives, e.g. I-E suf. -wo- : Lat. parvus "small", Gr. uovo^ "lonely"; suf. -mo-: Lat.

Germanic Consonantal Stem-Building Markers

413

/br/Hz;s"warm"; suf.-no-: Lat. magnus "great", plenus "full", Skt. puma "full"; suf.-r/-ro: Lat. ruber, Skt. rudhira "red", Avest. srira "beautiful" etc. The Russian scholar A.A.Potebnya remarked that ancient Russian adjectives with such suffixes primarily expressed an attitude to the main idea in the stem of the word, e.g. the adjective HCTHH-bH-b "true" meant that it referred to truth (Potebnya, 1968 : 414). The possessive character of adjectives is well felt in Siberian languages, e.g. Khanty porchan "pea, referring to peas", jarkjan "referring to ruff", lusan "with a beard, having a beard", vatan "windy, with the wind" etc. In Hittite P.W. Brosman noticed that such adjectives were often used as animate nouns (Brosman, 1988 : 255). The same can be observed in other ancient I-E languages, e.g. A.S. segoda is used about Beowulf in the meaning of "a fine fellow", O.Icel. se rica "a powerful one" - about Hrodgar (Katsnelson, 1949 : 261), Gothic sinista "elder, syndic", dautha " a dead man", mahteiga "sovereign", andaneitha "enemy, hostile man" etc. One peculiar thing that should be noted is that the same stem-building markers often characterize abstract nouns in Germanic languages, e.g. Goth, hauhi "height", gaurei "grief, diupei "depth", braidei "breadth" etc. The explanation to this phenomenon may be found if we take into consideration the idea of expressing possessive relations, formulated by T.V.Gamkrelidze and V.V. Ivanov that, before the appearance of the genitive case, possessive relations were expressed by means of constructions with possessive pronouns: man son his = man's son (1984 : 278). It is quite clear with terms of relationship, names of parts of the body, nomina agentis, living beings, which join these suffixes because all of them have relative semantics. As M.A. Zhurinskaya states, they cannot be thought of apart from someone or something with whom or with which they are connected (1977 : 240). In the same way, nouns expressing abstract notions in modern understanding were not so abstract to ancient people. At least, such notions were not separated from the concrete things to which they were related. For example, breadth could be thought of only belonging to somebody or something : my, his, her, their etc. breadth. This idea can be supported by Siberian aboriginal languages, e.g. in Khanty (one of the Finno-Ugric languages) such names are often given with possessive suffixes: oyam "my share" (lit."share mine"), oyan "your share", oyal "his share"; qovatal "his length", qochasal "his burden", iltdl mdnds "(he) became frightened" (lit. "soul his left him") etc. It is not difficult to trace back these markers to gradually turning into exponents of definiteness. A Khanty speaker explained the usage of the word tin "price, cost" with the possessive marker as a definite one, and without such it means "some price, about which the speaker does not know anything, the unknown price". In some Finno-Ugric languages (e.g. Komi) the exponent of the possessivity of the 2nd or 3rd persons is generalized and used to form definite declensions (Sukhanova, 1962 : 84; Lytkin, 1966 : 286). It is of course not so easy to trace the same transformation of the stem-building markers from the expression of possessivity to the expression of definiteness. Still some remnants are left in I-E languages, though they are not on the surface as in Siberian languages but concealed. The same stem-building markers sometimes performed the role of the definite article, e.g. as the post-positive article in Old Armenian (ter-n "a master, owner") ( Brugmann, 1892 : 769).

414 Langue and Parole in Synchrony and Diachrony The expression of defmiteness was discovered in Gothic by comparing words of the same root and meaning belonging to different declensions. The words used with the stembuilding marker -n- express as a rule defmiteness, while the same nouns with vowel stems do not show it (Ossipova, 1988 : 14-21). For example the Gothic word for corn (seed) in the first sentence below expresses defmiteness and is declined according to n-stems. Its defmiteness is supported by the attribute referring to this word. In the second sentence the word for corn is declined according to a-stems and it does not express defmiteness, since it can refer to any corn: (1) swe KAURNO sinapis, thatei than saiada ana airtha, minnist allaize fraiwe ist thize ana airthai (Me. 4, 31). 'it is like a mustard seed, which is the smallest seed you plant in the ground' (2) habands winthiskauron in handu seinai jah gahraineith gathrask sei jah briggith KAURN in bansta seinamma, ith ahana intandeith funin unhvapnandin (L.3, 17). 'his winnowing fork is in his hand to clear his threshing floor and to gather the wheat into his barn, but he will burn up the chaff with unquenchable fire' The next pair of examples concerns the word for depth in Gothic. The meaning of defmiteness is seen only through the context in the first example (number 3), in the second example (number 4) it does not express such defmiteness: (3) 17. bauan Xristu thairh galauein in hartam izwaraim, 18. ei friathwai gawaurhtai jah gasulidai mageith gafahan mith allaim thaim weiham hva sijai braidei jah laggei jah hauhei jah DIUPEI, 19. kunnan tho ufarassau mikilonthis kunthjis frijathwa Xristaus, ei fulnaith in allai fullon gudis (E.3, 17-19). 17. 'so that Christ may dwell in your hearts through faith. And I pray that you, being rooted and established in love, 18. may have power, together with all the saints, to grasp how wide and long and high and deep is the love of Christ, 19. and to know this love that surpasses knowledge - that you may be filled to the measure of all the fullness of God'. (4) nih hauhitha nih DIUPITHA nih gaskafts anthara magi uns afskaidan af friathwai gudis thizai in Xristau lesu, frauin unsaramma (R. 8, 39) . 'neither height nor depth, nor anything else in all creation, will be able to separate us from the love of God that is in Christ Jesus our Lord'. The relative character of the noun DIUPEI in example 3 is supported by the fact that the stem-building marker -ein- is of complex nature. In fact, it contains double relativeness. Such redundancy (pleonasm) was possible because the marker -n- was not yet so strictly felt as the pronominal suffix but more like a marker of defmiteness. The relative notion was strengthened by the addition of the particle -ei-, which was used in relative formations. It goes back to the relative pronominal stem, e.g. I-E *YO-, O.Ind. YA-(Uhlenbeck, 1896: 36). The next step is to prove why stem-building markers could perform so many functions. 4. The answer lies in their pronominal origin. This has been proved and described by many scholars (Specht, 1947 : 299, 303, 307, 315, 353 etc.; Maitinskaya, 1969 : 139; Shields, 1990 : 12-17 and others). I have only to add that all their functions are logically connected, as it has already been shown.


415

5. One more logical supposition is postulated in this paper, that primarily the same stem-building markers were used to mark the exponents of the active case. This idea concerns the very early stage of the development of Proto-Indo-European, if it is taken for granted that the earlier case system was based on an opposition between active and inactive cases, where the active case was always marked (Gukhman, 1981 : 201; Savchenko, 1967 : 74; Shields, 1988 : 225). At present it is postulated that the earliest stage of I-E is more likely to be of active type than of ergative (Gamkrelidze and Ivanov, 1984 : 267-290; Klimov, 1977 : 206-209). At least what was earlier described as an ergative case in I-E languages (the s-nominative) now can be treated as a marker of the active case, since this marker was joined at first only to certain groups of nouns, referring to the active class (Tronski, 1967b : 91-94, Savchenko, 1967 : 74). I.M.Tronski writes that there was a period when names of inactive class were not changed, while there developed case differentiation in the active class (Tronski, 1967b : 92; cf. Savchenko, 1967: 74). In our opinion, I-E s-ergative (as it was considered earlier), which in time became the sign of the nominative, is of later formation in comparison with stem-building markers. They were an appropriate material to mark not only animateness but any active force from which the action was directed. This idea is exactly in line with Gamkrelidze and Ivanov's view, since they consider the I-E exponent of the nominative *-os (common with the genitive in *-os) to be a sign of active typology (1984 : 273). The idea of Gamkrelidze and Ivanov about the Hittite genitive in -an which was used with nouns of common gender (animate) while the exponent -as referred to neuter gender (inanimate), deserves special attention (Ivanov, 1963: 135; Gamkrelidze, Ivanov, 1984 : 269-270). Later on the Hittite exponent -as penetrates into the class of animate nouns (Gamkrelidze and Ivanov, 1984 : 269). It is known that the I-E exponent of the nominative and genitive case -s has originated from the demonstrative pronoun *-so (Tronski, 1967b : 94; Gukhman, 1981 : 202-203; Gamkrelidze and Ivanov, 1984 : 271-275). Its possessive nature is seen in the genitive case. The same possessive nature is typical of consonantal stem-building markers (which we have already tried to show). K.E.Maitinskaya postulates that the deictic particles, expressing direction, possessivity and space-time relations, could serve well for the formation of primary case exponents (1969 : 138-139). These primary case exponents are preserved in oblique cases. This fact is proof that the earlier I-E case system was based on the correlation between a marked active case and the unmarked one. Typologically it is reminiscent of the correlation between the ergative and absolute cases in many Caucasian languages. The reflection of the former usage of the noun in the oblique case to serve as logical subject is found in I-E constructions with the Dative/Accusative described by some linguists (Gukhman, 1967 : 58-73, Savchenko, 1967 : 88), e.g. G. Mich wundert des schwarzen Ritters; R. MHC HpaBHTca 'I like it" (Gukhman, 1967 : 64). Siberian aboriginal languages give further support to considering I-E stem-building markers as capable of expressing case relations, namely to mark the agent (the doer of the action). For example, in many Siberian languages substantives in oblique cases serve to show from where the action is directed: in some Khanty dialects the usage of the noun in the locative-instrumental case or in locative alone is equal to the subject of the sentence (Russkaya, 1962: 261; Tereshkin, 1961: 50). Often nouns in instrumental or commitative cases are used as the real agent and perceived by Khanty people as the doer

416 Langue and Parole in Synchrony and Diachrony of the action, e.g. nina put sil'a reketeta, lit. "the kettle was thrown by the woman", but Khanty people translate it as "The woman has thrown the kettle" (Balandin, 1948 : 311). Thus we come to the conclusion that the action of I-E consonantal stem-building suffixes as markers of the active case was prior to the s-nominative marker as a sign of the doer of the action. The consonantal stem-building markers are considered here to be the primary case endings, as K.E.Maitinskaya postulates. The case endings which follow the consonantal stem-building markers are of secondary character. They are defined by Jakubinski as surface endings (1953 : 169). Most such surface case endings are of adverbial character, as is shown by Gamkrelidze and Ivanov (1984 : 285-286). W.P.Lehmann noticed that they are often written in Rigveda separately (in instrumental, dative, ablative, locative pi.) (1958 : 182). 6. Typological comparison with Siberian languages provides additional support for the supposition of such many-sided functions of Germanic and in general I-E consonantal stem-building markers. If we compare the position of I-E- consonantal stem-building markers with the possessive markers in some Siberian languages it will be clear that both of them occupy not only an identical position, but in Siberian languages their possessive nature is more vividly seen. At the same time, in Siberian languages it is easier to trace the gradual transition of these markers from possessivity to definiteness. All Siberian possessive markers are of pronominal character and they always correlate with nouns of certain classes which can be defined as active ones, since they designate animate and active things and forces.

REFERENCES Balandin, A.N. (1948). Padezhy sub'ekta i obekta na sluzhbe nominativnoi i ergativnoi konstruktsii glagol'nogo predlozheniya v vakhovskom dialekte khantyiskogo yazyka . In: Uchonye zapiski. Seria vostokovedcheskikh nauk. Vyp. 2, pp. 301-315. Leningrad. Brosman, P.W. (1988). The semantics of the Hittite gender system. In: Novoe v Zarubezhnoilingvistike, vyp. 21, s. 257-261. Progress, Moskva. Brugmann, K. (1892). Grundrift der vergleichenden Grammmatik der indogermanischen Sprachen : Wortbildungslehre (Stammbildungs- und Flexionslehre), Bd. 2. Karl J. Trubner, Strassburg. Burrow, T. (1976). Sanskrit. Progress, Moskva. Cuny, A. (1924). Etudes pregrammatical sur le domains des langues indoeuropeennes et chamito-semitique. Paris. Feist, S. (1920). Etymologisches Worterbuch der Gotischen Sprache. Druck von Karras, Krober & Nietschmann in Halle (Saale). Pick, A. and W. Stokes, (1890). Vergleichendes Worterbuch der indogermanischen Sprachen. Aufl., bearb. von A.Bezzenberger. A.Fick and W.Stokes, Gottingen. Gamkrelidze, T.V. and V.V. Ivanov, (1984). Indoevropeiskii yazyk i indoevripeitsy. Chast' 1. Tbilisi University Press, Tbilisi.


417

Gukhman, M.M. (1967). Konstruktsii s datel'nym-viniternym litsa i problema ergativnogo proshlogo indoevropeiskikh yazykov. In: Ergativnaya konstruktsiya predlozheniya vyazykakh razlichnykh tipov. Pp. 58-73. Leningrad. Gukhman, M.M. (1981). Istoricheskaya tipologiya i prblema diakhronicheskikh konstant. Nauka, Moskva. Haudry, J. (1988). Indoevropeiskii yazyk. In: Novoe vzarubezhnoi lingvistike. Vyp. 21, pp.29-121. Progress, Moskva. Hirt, H. (1925). Geschichte der Deutschen Sprache. Zweite, neubearbeitete Auflage, C.H.Becksche Verlagsbuchhandlung, Miinchen. Ivanov, V.V. (1963). Khettskiiyazyk. Izdatel'stvo Vostochnoi Literatury, Moskva. Jakubinski, L.P. (1953). Istoriya drevnerusskogo yazyka. State educational pedagogical publishing house of ministry of education of RSFSR, Moscow. Katsnelson, S.D. (1949). Istoriko-grammaticheskie issledovaniya. Izdatel'stvo Akademii nauk USSR, Moskva-Leningrad. Klimov, G.A. (1977). Tipologiya yazykov aktivnogo stroya. Nauka, Moskva. Klimov, G.A. (1983). O pozitsionnykh padezhakh ergativnoi sistemy. Voprosy yazykoznaniya, N° 4, pp. 86-90. Lehmann, W.P. (1958). On earlier Stages of the Indo-European Inflection. Language, vol. 34, NO 2, pp. 179-202. Lytkin, V.I. (1966). Komi-Zyryanski yazyk. In: Yazykinarodov USSR. Nauka, Moskva, 1960, pp. 281-299. Maitinskaya, K.E. (1969). Mestoimeniya v yazykakh razlichnykh sistem. Nauka, Moskva, 1969. Ossipova, O.A. (1980). Otrazhenie kategorii odushevlonnosty/neodushevllonnosti v paradigms skloneniya v drevnegermanskikh yazykakh. Tomsk University press, Tomsk. Ossipova, O.A. (1988). Sinonimiya paradigmatiki sushchestvitel'nogo u drevnikh germantsev. In: Grammaticheskaya i leksicheskaya sinonimiya, antonimiya i omonimiya, pp. 14-21. Tomsk Pedagogical University Press, Tomsk. Ossipova, O.A. (1990). Functional variability of the old German consonant themeforming elements (collective monography). In: Yazykimira. Problemyyazykovoi variativnosti,pp. 153-171. Nauka, Moskva. Potebn'ya, A.A. (1968). Iz zapisok po ruskoi grammatike. Vol. 3. Prosveshchenie, Moskva. Russkaya, J.N. (1962). O nekotorykh osobennostyakh padezhnoi sistemy priural'skogo govora khantyiskogo yazyka. In: Voprosy finno-ugorskogo yazykoznaniya, pp. 257-274. Leningrad. Savchenko, A.N. (1967). Ergativnaya konsruktsiya predlozheniya v praindoevropeiskom yazyke. In: Ergativnaya konstruktsiya predlozheniya v yazykakh razlichnykh tipov, pp. 74-90. Leningrad. Shields, K. (1988). Some remarks concerning early Indo-European nominal inflection. In: Novoe v Zarubezhnoi lingvistike, pp. 224-261. Progress, Moskva. Shields, K. (1990). Zametki o proiskhozhdenii osnovoobrazuyushchikh formantov v indoevropeiskom. In: Voprosyyazykozhnaniya, No 5. Nauka, Moskva. Specht, F. (1947). Der Ursprung der indogermanischen Deklination. Vandenhock and Ruprecht, Gottingen.

418 Langue and Parole in Synchrony and Diachrony Sukhanova, V.S. (1962). O semantike prityazhatyel'nogo suffiksa 3 litsa edinstvennogo chisla v permskikh yazykakh. In: Lingvisticheskisbornik, pp. 80-89. Petrozavodsk. Tereshkin, N.I. (1961). Ocherkidialektovkhantyiskogoyazyka. Nauka, M.-L., 1961. Tronski, I.M. (1967a). Obshcheindoevropeiskoe yazykovoe sostoyanie. Nauka, Leningrad. Tronski, I.M. (1967b). O donominativnom proshlom indoevropeiskikh yazykov. In: Ergativnaya konstruktsiya predlozheniya v yazykakh razlichnykh tipov. Leningrad. Uhlenbeck, C.C. (1896). Kurzgefafites etymologisches Worterbuch der gotischen Sprache. Verlag von Job. Miiller, Amsterdam. Whitney, W.D. (1889). A Sanskrit Grammar. Breitkopf & Hartel, Leipzig. Zhurinskaya, M.A. (1977). Imyennye posessivnye konstruktsii i problema neottorzhimoi prinadlezhnosti. In: Kategoriya bytiya i obladaniya v yazykye, pp. 194-258. Nauka, Moskva.

32

"LANGUE" AND "PAROLE" IN PROTO-INDO-EUROPEAN RECONSTRUCTIONS Thomas V. Gamkrelidze, Tbilisi, Georgia

Abstract If we assume the notions of "langue" and "parole" to be equivalent to "linguistic system" and "realization of that system in speech", respectively, we may posit the existence of such categories in reconstructed linguistic models, as well. In particular, at the phonological level of a reconstructed linguistic proto-system the establishment of its phoneme inventory and rules of paradigmatic and syntagmatic relationships in the system come under "langue" in reconstructed models, while defining allophonic variants of such phonemic units viewed to be realizations of these entities in speech may be considered as "parole". Thus, in Proto-Indo-European reconstructions the three series of stops specified as I: glottalized (ejectives), II: voiced (aspirates), III: voiceless (aspirates), with "voiced" and "voiceless" phonemes presented in the form of two allophones - "aspirated" and "non-aspirated" - refer both to "langue" and "parole" of Proto-Indo-European, since the Proto-Indo-European model reconstructed in this form represents both the phonemic system of the proto-language, with a specification of syntagmatic and paradigmatic relationships of its units, and its supposed realizations, according to certain distributional patterns, in speech, as phonetically aspirated and unaspirated stops. On the basis of these assumptions, we have to reinterpret some of the classical syntagmatic "Phonetic Laws" of Indo-European, such as Grassmann's Law, Bartholomae's Law, etc., as operating not at the phonemic level of the Proto-language, but at its subphonemic, allophonic level, thus referring not to "langue", but to "parole" of the reconstructed linguistic system.

If we assume the notions of "langue" and "parole" to be equivalent to "linguistic system" and "realization of that system in speech", respectively, we may posit the existence of such categories in reconstructed linguistic models, too. In particular, at the phonological level of a reconstructed linguistic Proto-System the establishment of its phoneme inventory and rules of paradigmatic and syntagmatic relationships in the system refers to the category of "langue" in reconstructed models, while defining allophonic variants of such phonemic units viewed as realizations of these entities in speech may be considered as referring to "parole". Thus, in Proto-Indo-European reconstructions along the Glottalic model, the three series of stops specified as I: glottalized (ejectives), II: voiced [aspirates], III: voiceless [aspirates],

419

420 Langue and Parole in Synchrony and Diachrony with "voiced" and "voiceless" phonemes presented in the form of two allophones - "aspirated" and "non-aspirated" - refer both to "langue" and "parole" of Proto-Indo-European, since the Proto-Indo-European model reconstructed in this form represents both the phonemic system of the proto-language, with a specification of syntagmatic and paradigmatic relationships of its units, and its supposed realizations in speech according to certain distributional patterns as phonetically aspirated and unaspirated stops: The Glottalic System of PIE I II III (p') b/bh p/ph t' d/dh t/th h k' g/g k/kh

(Cf. Gamkrelidze and Ivanov 1995). On the basis of these assumptions, we have to reinterpret some of the classical syntagmatic "Phonetic Laws" of Indo-European as operating not at the phonemic level of the Proto-Language, but at its subphonemic, allophonic level, thus referring not to "langue", but to "parole" of the reconstructed linguistic system. This concerns, in the first place, a new interpretation of "Grassmann's Law". A characteristic property of the Series II (voiced) and III (voiceless) stops is their phonetic feature of aspiration. Each phoneme of these series had two allophones, aspirated and unaspirated, depending on the phonetic environment. Concrete linguistic facts tell us with a fair degree of precision in what positions the phonemes of Series II and III appeared in their aspirated and unaspirated forms. The aspirated variant is the basic allophone, since it appears in most of the phonetically independent positions which can be established for Proto-Indo-European on comparative grounds. The unaspirated variant appears in particular phonemic contexts. Consequently, the task of distributive analysis is to determine the positions in which the phonemes appeared in their unaspirated forms. The second, voiced series lends itself most precisely to this kind of analysis. In the daughter languages it has left quite clear traces which make it possible to reconstruct its distribution in Indo-European word-forms. One of the major principles determining the behavior of Series II is the fact that when two Series II phonemes appear in a single stem they must always be manifested as different allophones — one aspirated, the other unaspirated. Thus, one stem can contain only one unaspirated sound. If the initial consonant is unaspirated, the second is aspirated, and vice versa. This distributional property of Series II consonants can be clearly seen and reconstructed on the basis of Indo-Iranian and Greek evidence. These branches reflect an allophonic rule for Series II whereby the unaspirated allophone is found initially and the aspirated one non-initially before a vowel or sonant: Skt. bahuh 'arm', Gk. pekhus 'elbow' point to a PIE form with an initial unaspirated allophone and a non-initial aspirated allophone of PIE *bh and *gh (both underwent devoicing in Greek);

"Langue " and "Parole " in Proto-Indo-European 421 Skt. badhnati, later bandhati 'ties, binds', bdndhuh 'kinship, kinship by marriage, relative', Gk. pentheros 'father-in-law, wife's father' (from 'related by marriage'): PIE [*bendh-]; Skt. bahu- 'dense, thick, numerous', Gk.pakhus 'fat, thick': PIE [*bengh-]; Skt. bodhati, bodhate 'wakes up', Gk. peuthomai, punthdnomai '(I) recognize, notice, stay awake': PIE [*beudh-], [*budh-]; Skt. budhndh 'soil', Gk.pithme:n 'soil': PIE [*budh-]; Skt. ddhati 'burns', nidagd-h 'heat, summer', Gk. tephra 'ashes'; Skt. deh- 'rub, smear, spread on, anoint', dehl 'wall, dam, embankment', Gk. teikhos 'wall'. These Sanskrit and Greek cognates provide good evidence for the distribution of aspirated and unaspirated allophones in Indo-European. The closeness and indeed identity of the Sanskrit and Greek verb forms with initial reduplication, together with contemporary assumptions about the unity of the Greek-Aryan verb system, permit us to see the pattern as an areal one within Indo-European, which also confirms the Indo-European source of these phoneme alternations. They cannot be explained as being due to deaspiration and independent parallel development of unaspirated phonemes in the two branches (the usual understanding of Grassmann's Law in classical Indo-European linguistics), but must be seen as the reflex of a shared distributional rule that had areal status in Proto-Indo-European. The facts from the other Indo-European languages ~ Italic, Germanic, and others — are fully consistent with this treatment of these correspondences. Thus 'Grassmann's Law of Deaspiration' acquires a completely new sense. It can be seen as the alternation of aspirated and unaspirated sounds at the allophonic level in ProtoIndo-European, not as independent deaspiration of aspirated phonemes in Sanskrit and Greek: The virtual identity of the Sanskrit and Greek deaspiration is due to their common origin in Indo-European. It follows that this process is to be reconstructed for Indo-European as an allophonic rule referring to "parole", which later turned into a phonemic alternation when the reflexes of the aspirated and unaspirated voiced allophones were phonologized in the daughter branches. After the reflexes of Series II and III were phonologized in Sanskrit and Greek, the former allophonic rule turned into a rule for the alternation of aspirated and unaspirated phonemes within the stem, acquiring thus the status of a "langue" phenomenon.

REFERENCE Gamkrelidze, Th. V. and V.V. Ivanov (1995). Indo-European and the Indo-Europeans. A Reconstruction and Historical Analysis of a Proto-Language and Proto-Culture, 2 vols. With a preface by Roman Jakobson. Translated into English by Johanna Nichols. Mouton de Gruyter, Berlin, New York.


33

LINGUISTIC DYNAMICS IN A GERMAN AUTOBIOGRAPHY Fernande Krier, University ofRennes, France

Abstract Starting from the conception of linguistic dynamics and the principle of economy connected to it, we present a specific text in order to describe the system of German at a given point on the time axis and to find out its dynamic tendencies by comparing it to the contemporary stage of the language. The analysis is centred on grammatical and lexical realizations as well as on idiomatic phenomena. If there is evolution we shall try to explain its procedure as well as its internal and external causalities. As for the text, the autobiography of an industrial worker published in 1905 has been chosen for two reasons: 1) precisely the period of the beginning of the 20th century needs an analysis in many respects; 2) that type of documentary narrative is innovative at the given period.

i. INTRODUCTION 1.1 The Framework According to Saussure (1915/1972), there is a fundamental opposition between the concept of langue, which, for him, is essential, and that of parole, as a matter of secondary importance, while he admits, nevertheless, a close connection between these two objects, which cannot exist each without the other. However, the definition of parole as a linguistic activity realized in speech acts and of langue, as an abstract entity, an artefact, put into concrete form in the dictionary and the grammar, has to be preferred (Martinet, 1984). Saussure deduces two operating notions of his dichotomy langue-parole: the parole always precedes the langue and is at the origin of its evolution. Hence we approach his second dichotomy, the synchronic point of view, without temporal perspective and the diachronic one, with a prospective and a retrospective perspective (Saussure, 1915/1972). Now the question is whether it is possible to consider pure synchrony clearly apart from diachronic facts in linguistic description or even explanation. The answer is that the terms of innovation and survival of older structures introduce a diachronic perspective into every elaborate synchronic analysis (Martinet, 1975). That signifies that language has to be considered in its dynamics, to which, first of all, the unstable elements (Houdebine, 1985) draw attention, namely elements whose existence varies from one speaker/writer to another. These reflections lead us to the principle of linguistic economy, that is to say, the desire to transmit a 423

424 Langue and Parole in Synchrony and Diachrony maximum of information with a minimum of articulatory and phonic effort, this principle being a determining factor of evolution. Analyzing the linguistic production of one individual in a specific type of text seems to be the best means to me to get information on the evolutionary character of a given structure, without claiming, however, that this point represents a general tendency in the langue of a given stage.

1.2 The Text Autobiographies are realized as documentary narratives, interrupted by descriptions serving to explain given facts. This type of text, the function of which is to be informative, reflects the identity of the author. Its mechanism consists in reconstructing events of the past, where objective attitudes are interacting with subjective opinions. Consequently, the truth of the text depends on the faculty of memorization, but on the motivation of the author, too. Autobiographies of industrial workers have existed only since the end of the 19th century, because of the lack of interest of readers in the problems of the proletariat. The memoirs of Moritz Bromme, Lebensgeschichte eines modernen Fabrikarbeiters, written in 1903 and published in 1905, are therefore a pioneer work in this field. It is a stimulating book offering to the working class interesting socio-cultural descriptions, and contributes to its identity research and political education. Moritz Bromme was born in Leipzig in 1873 and lived near Gera in Saxony. Since his father lost his job for political reasons - he had responsibilities in the social-democratic party the author, who went to secondary school and wanted to become a teacher, was obliged to leave school and to work in a factory. Like his father he became a member of the party and the trade union. When he was thirty he wrote his autobiography in a sanatorium, where he was receiving treatment for tuberculosis. Reading this autobiography one may observe that, for the structure of the text, the syntax, to a great extent, corresponds to contemporary German, whereas the style, strangely, shows two levels, an artificial style marked by sentimentality, when the author recounts events which do not affect him closely: Ich sehe ihn noch heute mil uns Kindern aufdem Sofa sitzen, in den Pelz eingewickelt und bitterlich weinen. "Meine armen - armen Kinder!" Wie schnitten mir diese Worte in die Seele (p.68) and an objective realistic style, when he describes his own lot: Herr Beeger wies mich aber ab mit dem Bemerken, dass es in der neuen Bude noch mehr Akkord geben wtirde, als bisher. Wir wurden nun mit den Arbeitern der bisherigen Konkurrenzfirma untermischt (p. 251). On the other hand, one may immediately ask what linguistic differences can be observed over the period since 1905 to the present time. Therefore, I shall analyse the dynamic conception of the microstructure in three areas: - grammatical facts; - lexical realizations; - idiomatic phenomena.

Linguistic Dynamics in a German Autobiography 425

2. GRAMMATICAL FACTS 2.1. I first deal with morphology. As for the verbal inflection, it is known that some verbs strongly inflected in Middle High German later on took the weak inflection, like gigen. The opposite case, namely a verb that loses its weak inflection in Middle High German in order to get a strong inflection in New High German, like preisen, is rare. I quote the Germanists Frederic Hartweg and Klaus-Peter Wegera (1989), who emphasize that in the weak verbs fragen, schallen, stecken the strong inflection in New High German is reduced to the preterite tense. That phenomenon appears in my text for the verb fragen, where the forms of the preterite erfragte and erfrug are free variants with a slight emphasis onfrug: Den Oberkellner fragte die Kochin stets vorher: "Franz, was wollen Sie essen?"... Michfrug sie nie (p. 138). It would be interesting to find out by analyzing other types of texts of the same period if these fluctuations are to be explained by idiolectal, sociological or regional factors. 2.2. An economic tendency on the signifiant - level is the apocope of the dative singular -e in the strong masculines and neuters, first of all as a regional and oral movement, which did not touch the East Central German dialects. For the historical background of this feature I quote the Germanist Peter von Polenz (1994), who states that after the Reformation grammarians insisted on the conservation or the restitution of final -e in order to underline the inflection in the written language. Until the end of the 19th century full vocalic syllables of inflection would have been practised in poetry as well as in sermons or official discourse, such as im Kampfe, zum Wohle. Nowadays the dative -e serves only rhythmic stylization. In his whole book the author has a predilection for the signifiants with final -e: auf dem Bahnhofe, im Hintergrunde, an einem Sonntage, in unserem Neste and he applies it even to the feminine: vor der Ture. I am inclined to think that the conservation of the dative -e in 1905 is due first to the Saxon oral language and secondly to the obsolete written language of that period. 2.3. Concerning syntax I would lay emphasis on the genitive attribute. Its well-known function consists in identifying the nucleus, as in Autos gleicher Farbe. Its competitor is the prepositional attribute which, however, does not succeed in eliminating it, because, on the level of the text, it focuses better on its nucleus than the prepositional attribute (Eroms, 1998). I have discovered an archaism in the text, the anteposition of the genitive attribute, namely the so-called Saxon genitive. According to Polenz (1994), the prenominal genitive was a little regressive about 1700, it referred to property and persons, but also to products of human activity. Nowadays the Saxon genitive attests the "human" feature, since it is limited to proper names and names of relationship. In addition to that it attests the "known" feature, thus it refers to definite nominal groups and has a determining function: the utterance Picassos Bild corresponds to das Bild Picassos, but not to ein Bild Picassos. Otherwise the prenominal genitive appears in idioms, as in my text auf Schusters Rappen reiten. Systematically it is considered to be affected, but obsolete, too. Analyzing the examples just mentioned Picassos Bild vs. das Bild Picassos, one may wonder whether there is a difference between those variants on the level of the signifie. A French Germanist, Jean-Fran9ois Marillier (1992) who studied the syntax in Christa Wolfs Kindheitsmuster (henceforth KM) explains these subjective facts by the pragmaticcommunicative intention of the poet. He states that the signifie being in the final position has the rhematic role: Dass die Sonne fruher ganz anders schien als heutzutage, scheint belegt

426 Langue and Parole in Synchrony and Diachrony durch aller alterer Leute Erinnerung (KM 79) vs. sie darfihr Buck aufdas Ndhtischchen der Mutter legen (KM 126). In the first example, the nucleus Erinnerung is rhematic while it is the genitive phrase der Mutter in the second case. According to this convincing analysis, the signification of the Saxon genitive in contemporary German has been shifted from the semantic to the pragmatic level. Now, what is the explanation of this form of the genitive in my text of 1905? The examples which nowadays are stylistically obsolete correspond in most cases to the syntacticsemantic principles, that is to say, the nucleus and the genitive attribute are definite, the referents are relations or called by their proper names: meiner Mutter Schwester vs. eine Muhme meiner Mutter; meines Bruders Lehrzeit vs. ein Brief meines Bruders; die drei Schwestern meiner Frau vs. Heines Harzreise; near acquaintances are put in the Saxon genitive, too: des Hauswirts alter Vater, des Wins eigener Sohn vs. der Schwiegersohn des Chefs, nach Aussage seines Chefs, die Rede des Geistlichen. On the other hand one could perhaps see in the following examples put into the postnominal genitive a communicative intention of the author, namely to focus on the informative value of the genitive attribute: die Hochzeit des Voters, mit dem neuen Schwager des Voters vs. meines Voters Posten / Krankheit / Heimat; der Sterbetag / Todestag meiner Mutter vs. meiner Mutter Schwester, der Ausspruch Dr. Schuberts vs. Heines Harzreise. As far as I am concerned, I should say that it is a simple manifestation of the evolution of German in 1905, according to the principle of economy. Consequently, the genitive attribute, because of the coexistence of its signifiants and their interaction in the process of communication, is a revealing example of linguistic dynamics. In addition, there is to be mentioned that the possessive genitive attribute sometimes is in competition with the possessive dative attribute, characteristic of colloquial language:"...diese Woche werden der Heddel ihre Schuhe besohlt, dem Ernst seine sind auch wieder runter,..." (p. 241). This signiflant occurs also as a possessive genitive attribute (Polenz, 1994): als er seiner zweiten Frau ihr Geld alles verwurgt hatte, (p. 208). In fact, these are only marginal cases.

3. LEXICAL REALIZATIONS In general, grammatical facts of linguistic evolution are to be explained by internal causality, for instance linguistic economy. But how and why does this evolution happen in the lexicon? 3.1. I begin with change on the morphological level, due without doubt to the internal economy of the system, such as the elision of prefixes and suffixes in der Gestrenge > der Strenge, das Besitztum > der Besitz, meine Verheiratung > meine Heirat. 3.2. Internal causality determines also the substitution of grammatical signs on the level of the signifiant, that is to say, signs of the closed limited classes like prepositions and conjunctions; incidentally, their function in linguistic evolution has not yet been analyzed exhaustively (Munske, 1985). My example is the now obsolete preposition ob, which predominates in the text ahead of wegen: ob dieser Ehrlichkeit, ob der zu erwartenden Tracht Prugel vs. wegen seiner Beteiligung. Diachronically ob exists as a preposition with the signifie uber or wegen since the Old High German period (Grimm, 1889), whereas wegen, a simplified form of von wegen seldom appears before the period of Luther (Grimm, 1901). The causality of that

Linguistic Dynamics in a German Autobiography 427 evolutionary process is phonic degradation, a consequence of linguistic economy (Ltidtke, 1986). The monosyllabic preposition ob, which is a polyseme, because it also belongs to the class of conjunctions, is less efficacious than bisyllabic wegen. This is a compensating mechanism of the principle of economy, namely an amplification reacting against the reduced efficacy of a linguistic unit. Indeed, in this way units of high frequency and of polysemic nature, and therefore of weak informative value, are substituted by signs being more amplified and having a specific content. In the same text linguistic units coming from different chronological strata are attested, a proof of linguistic dynamics! 3.3. The principle of internal causality does not apply to signs of the open classes, such as substantives, adjectives or verbs. Here we have the well-known situation that for every new signifie must be created a new signifiant. Every new experience, be it a technical development or a political upheaval, is reflected in the lexicon, namely by innovations and eliminations of archaisms. In the text I have found technical terms like Doppelspiralbohrer and Zapfenfrasmaschine, which are neologisms in the vocabulary of an industrial worker of 1905. On the other hand, as the author is a committed social democrat, we are confronted with the agitatory vocabulary of the workers' movement, and so we find words like Parteigenosse, Agitator, agitieren, Agitation, Metallarbeiterverband, Landesparteitag, Wahlsieg, Wahlbureau, the last example proving its neological character by its written form, which is not yet adapted. Now, there are in every linguistic system units which have been stable for centuries and others which are not, being about to change. First, I quote signs whose substitution concerns only the signifiant, such as Knabe, Jungling or Jungfrau, substituted by Junge, junger Mann, junges Madchen. If used in contemporary German, Knabe and Jungling deal with ironical speech and Jungfrau refers only to the Blessed Virgin; thus they are subjected to a pragmatic-communicative change. In the same way, Kommerzienrat has been substituted by Grosskaufmann and Aftermieter by Untermieter. The dynamics of language is attested in the text by the coexistence of archaic Gevatter with Pate and archaic Muhme with the Old French borrowing Tante. Next, words disappear because their signifies are no longer used, like Sommerfrische: nowadays people can afford more than a "country holiday rest in summer", or like Marketender. in a modern war there are no merchants who accompany the troops on the battle field. Furthermore, expressions like Gesinde or sich als Knecht verdingen and in Dienst gehen are obsolete because of socio-economic upheaval. Finally, a consequence of the mawkish style used sometimes by the author are the obsolete units due to sentimentality, such as wahrlich, bitterlich, flehentlich, darben, (ver)schmachten. In conclusion of this lexical point, I quote the lexeme der/das Fehl < Old French faille, which occurs only in contemporary German in the idiomatic expression ohne Fehl; thus I come to my third and final point.


4. IDIOMATIC PHENOMENA Idioms are polylexemes characterized by a relative stability, figurativity, lexicalization and reproducibility (Fleischer, 1982; Greciano, 1983). Amongst the semantic properties of their structure I quote comparative structures (Fleischer, 1994) having the function of expressive intensification: wie ein Blitz aus heiterem Himmel and wie vom Donner geriihrt or of objective differentiation: wie ein begossener Pudel, idioms with proper names: Schmalhans ist Kuchenmeister, and gestural idioms in which parts of the body intervene, the latter being important in idiomatic creation because of anthropocentric human behaviour: von der Hand in den Mund leben or sich eins ins Faustchen lachen. The idioms of this text lay emphasis on strong emotions, and in 70% of the examples they express negative behaviour, though that is a general tendency in idioms. What is the origin of idioms? The sources are the cultural background offered by the bible and Greek mythology, but also old traditions, popular beliefs, habits and customs of the Middle Ages such as hunting or craft industry, and everyday experiences. Thus idiomatic expressions, created centuries ago, are still alive. In order to find out if they have been maintained, I have presented the 89 idioms of my text to German native speakers asking them what example was unknown to them or unused by them. None of the informants, being about thirty, understood das Hasenpanier ergreifen (to run away hastily: humorous designation for the straightened tail of the fleeing hare), wider den Stachel locken (to be obstinate, to defend oneself against someone stronger: it is the picture of the oxen under the yoke goaded by the farmer, which lashed out) nor jemandem das Fell gerben (to beat up somebody: this is an expression of tanners); 66% were unaware of wissen, wo Barthel den Most holt (to know how to extract money, to win by a trick, to be cunning: a) < slang, where barthel signifies jemmy and moos < hebrew maos altered in Most signifies money; b) Barthel, a mayor of Heilbronn living in the 13th century, was said to take for himself the wine of the cellar of the town hall) norjemanden aufden Hund bringen (to have sunk pretty low: a) related to the dogcart used by poor people of old times; b) related to the game of dice; the "dog" seemed to be the designation for the worst throw); 33% did not know nicht viel Federlesens machen (without further ado: in times past it was the designation of the flatterers who picked up the small feathers on the dresses of high-ranking persons in order to get appreciated) or sich in die Brust werfen (to boast); these are 8% of the idioms of the text. Those idioms, except the last one, which is a gestural one, have one common semantic feature, they are opaque (or unmotivated), that is to say, contrary to transparent (or motivated) idioms such as jemandem ein Dorn im Auge sein, their signification can only be explained by their etymology. Beyond opaqueness, there is a second criterion, that of low frequency, which is why the above mentioned examples were unknown to most of my informants. Indeed, they are less used than, for instance etwas aufdem Kerbholz haben (to have something on one's conscience, to be guilty: in the Middle Ages, a stick split up lengthways took the place of the account book. During a purchase, the two parts of the stick, the one belonging to the salesman and the other to the purchaser, were stacked and provided with a notch; so the debt could be easily proved), or seinen Narren an jemandem gefressen haben (to be madly in love with somebody: this expression comes from the custom of the fools who, at the court where they were in favour with the princes, could allow themselves everything. Someone is so in love with the other that he is considered to be a fool), which, of course, are opaque, but of high frequency and therefore well-known to my

Linguistic Dynamics in a German Autobiography 429 informants. Thus, idioms are much more stable than lexemes; conserved, innovations are rarely realized.

their archaisms are better

5. All things considered we come to the conclusion that a synchronic analysis of specific linguistic phenomena has to pay attention to the corresponding diachronic facts, too, otherwise it prevents comprehension of the permanent evolution of language, which is realized more or less dynamically according to its different aspects.

REFERENCES Eroms, Hans-Werner (1998). Kasus, Rollen und Kontroversen. In: Die Kasus im Deutschen, Form undlnhalt (M. Vuillaume, ed.), pp. 57-70. Stauffenburg, Tubingen. Fleischer, Wolfgang (1982). Phraseologie der deutschen Gegenwartssprache. Bibliographisches Institut, Leipzig. Fleischer, Wolfgang (1994). Phraseologismus und Sprichwort: lexikalische Einheit und Text. In: EUROPHRAS 92. Tendenzen der Phraseologieforschung (B. Sandig, ed.), pp. 155172. Brockmeyer, Bochum. Greciano, Gertrud (1983). Signification et denotation en allemand, La semantique des expressions idiomatiques. Klincksieck, Paris. Grimm, Jacob and Wilhelm Grimm (1889, 1901). Deutsches Worterbuch. Siebenter Band, Dreizehnter Band. Hirzel, Leipzig. Hartweg, Frederic and Klaus-Peter Wegera (1989). Fruneuhochdeutsch, Eine Einfuhrung in die deutsche Sprache des Spdtmittelalters und derfruhen Neuzeit. Niemeyer, Tubingen. Houdebine, Anne-Marie (1985). Pour une linguistique synchronique dynamique. La Linguistique, 21, 7-36. Liidtke, Helmut (1986). Esquisse d'une theorie du changement langagier. La Linguistique, 22, 3-46. Marillier, Jean-Fran9ois (1992). Pranominaler und postnominaler Genitiv. In: Rechts von N. Untersuchungen zur Nominalgruppe im Deutschen (P. Valentin, ed.), pp. 47-58. Narr, Tubingen. Martinet, Andre (1975). Evolution des langues et reconstruction. Presses Universitaires de France, Paris. Martinet, Andre (1984). Sprache - ergon oder energeia? Folia Linguistica, 18, 539-548. Munske, Horst Haider (1985). Lexikologie und Wortgeschichte. Germanistik - Forschungsstand und Perspehtiven (G. Stotzel, ed.), pp. 27-43. de Gruyter, Berlin - New York. Polenz, Peter von (1994). Deutsche Sprachgeschichte vom Spdtmittelalter bis zur Gegenwart. Band II. de Gruyter, Berlin - New York. Rohrich, Lutz (1973). Lexikon der sprichwortlichen Redensarten. 2 Bande. 3. Aufl. Herder, Freiburg - Basel - Wien. Saussure, Ferdinand de (1915/1972). Cours de linguistique generale, edition critique preparee par Tullio de Mauro. Payot, Paris.


34

ACCENTUAL RECONSTRUCTION OF THE PROTO-SYSTEM OF MAINLAND JAPANESE DIALECTS Akiko Matsumori, Japan Women's University, Tokyo, Japan

Abstract The oldest material suggesting the existence of a tonal system in Japanese is the early 12*-century dictionary called the Ruijumyogisho, recording a system of the Old Kyoto dialect, in which 1-mora nouns have three tonal classes, 2-mora ones five, 3-mora ones seven. Although none of the present Japanese dialects retain all the distinctions of these classes, the historical existence of these classes is also indicated by a comparison of the tonal patterns of the present-day dialects; ie., there are regular correspondences of tonal patterns between dialects for the same tonal classes of the system, indicating a strong basis for the reconstruction of the accentual system of proto-Japanese. It is now generally assumed among Japanese tonologists that the proto-system of accent for Japanese dialects is that of the Old Kyoto system mentioned above. However, certain dialects in the Shikoku area present problems concerning the validity of this assumption. An accentual system one stage prior to that of the ^""-century Kyoto dialect is proposed as a proto-system of accent for mainland Japanese dialects, explaining hitherto problematic cases of tonal merger patterns observed in these dialects.

INTRODUCTION The study of tones in the various accent systems of Japan is one of the best developed areas in Japanese linguistics. Since the early stage of descriptive studies of Japanese accent almost 70 years ago, it has been recognized that the diversity of their differences observed among current dialects in Japan provides a fruitful ground for diachronic studies of Japanese accents. Because of the regular and systemic character of accent change, and due to the regular cross-dialectal correspondences of accentual patterns, it is possible to attempt comparative reconstruction of the accentual system of proto-Japanese dialects. The purpose of this study is to question the previously-proposed idea that the accentual system of proto-Japanese is the one similar to the early twelfth century Kyoto system, and to propose a new proto-system of accent for mainland Japanese dialects. This study will first give an overall introduction to the hypothetical 'word classes' for accent, and related lists of vocabulary supposed to belong to the same classes (generally known as the 'Kindaichi

431

432 Langue and Parole in Synchrony and Diachrony vocabulary', since the vocabulary classification was originally proposed by Kindaichi), which was developed based on observations of the correspondence of tonal patterns among dialects, as well as with the early twelfth century dictionary called the Ruijumyogisho, the oldest document with specific information on the accentual system within Japan. Although the Ruijumyogisho system has traditionally been supposed by historical phonologists in Japan to represent the earliest system of Japanese dialects, an examination of data from dialects in the Shikoku area (See Map 1) poses problems concerning this traditional hypothesis. In relation to this, a system that is one stage prior to that of the Ruijumyogishosystem (i.e., the early twelfth century Kyoto dialect) is proposed, which will then be examined to determine whether this newly-proposed system can also explain hitherto problematic cases of tonal merger patterns observed in these dialects.

THE CONCEPT OF WORD CLASSES AND THE 'KINDAICHI VOCABULARY' Almost all Japanese dialects display a regular correspondence in their tonal (accentual) patterns: i.e., a class of words which shows up with a certain accent pattern in one dialect regularly matches the same group of words displaying another accent pattern in a different dialect, which suggests that a comparison of the tonal patterns of the present-day Japanese dialects could help us to reconstruct the accentual system of proto-Japanese. This may be illustrated by the simple comparison of words given in (1), which is a comparison of five twomora nouns in three Japanese dialects: Tokyo, Kyoto, and Kagoshima. (Since the distinction between the tonal patterns of each noun is maximally shown by attaching certain particles after them, hereafter a nominative case marker 'ga' in Japanese is attached after each item.) (1) Regular Correspondence of Tonal Classes between Dialects: Correspondence of Accent Patterns of Two-Mora Nouns ('ga' is a nominative case marker) TÎ

Tokyo Kyoto Kagoshima

'nose' /

/

'sound' f

N

/

N

'mountain'

i > ^ \ / \ hana ga ;/ oto ga yama ga hana ga / oto ga yama ga hana ga oto ga / vama ga hana ga / oto ga / yama ga

'boat' // / /

.' \ ^ hune ga hune ga / hune ga hune ga /

'rain' / \ \ ame ga ame ga ame ga ame ga

In the Tokyo dialect, the word 'hana' (nose) with the particle 'ga' has a LHH (low-highhigh) tonal pattern. On the other hand, the words 'oto' (sound) and 'yama' (mountain) in Tokyo both have LHL (low-high-low) patterns, while the words 'hune' (boat) and 'ame' (rain) have a HLL (high-low-low) pattern. In contrast to Tokyo, these last two words 'hune' and 'ame' have different tonal patterns in present-day Kyoto, which are LLH and LHL respectively when the particle 'ga' is attached. But even in this dialect, the words 'oto' and 'yama' show up with the same pattern, i.e. HLL. However, the distinction between the tonal patterns of 'oto' and 'yama' is clearly made in Kagoshima dialect, in which the first two words ('hana' and 'oto') merged into one tonal pattern (LHL), while the remaining three ('yama', 'hune' and 'ame') merged into the LLH pattern.

Accentual Proto-System of Mainland Japanese Dialects 433 This simple illustration with three dialects in Japan indicates that at least 5 classes of nouns must have been distinguished by accent in the proto-Japanese accent system, from which these three different dialects derived. In fact, an examination of all other dialects in Japan suggests that the two-mora nouns are supposed to have at least five different tonal categories. (The numbers of tonal patterns differ depending on grammatical categories. Generally, multinumbered tonal patterns are usually observed only when nouns are involved; the tonal pattern for verbs and adjectives is basically limited to a maximum of two in Japanese.) What is significant here is the regular correspondences of the kinds of words which show up with each tonal pattern in the dialects. For example, the group of words which show up with the 'hana' (nose) pattern in the Tokyo dialect, which is LHH, match those which show up with the 'hana' (nose) pattern in the Kyoto dialect, which is HHH. In this way, the kinds of words which appear with a certain tonal pattern in one dialect generally correspond to those with a particular accent pattern in another dialect. Such regular correspondences can be observed in dialects throughout Japan. Based on this regular correspondence, Kindaichi and Wada (1955), and Kindaichi (1974) developed the notion of word classes. They called words which showed the same tonal pattern as 'hana' Class 1 nouns, those with the 'oto' pattern Class 2, those with the 'yama' pattern Class 3, those with 'hune' pattern Class 4, and the ones with 'ame' pattern Class 5. In this way, they assumed that there are up to 5 different classes for two-mora nouns in the proto-Japanese system. In the same way, they assumed that one-mora nouns would have up to 3 classes (Class 1-3), and three-mora nouns up to 7 classes (Class 1-7); although there is some disagreement among Japanese linguists on the number of classes for two-mora and three-mora nouns, this hypothesis concerning the number of classes in the proto-Japanese system is more or less accepted by most researchers doing diachronic studies on Japanese accents. Furthermore, based on this notion of word classes, a list of classified vocabulary (commonly referred as the "Kindaichi vocabulary") was proposed by Kindaichi and Wada (1955) over 40 years ago, samples of which are shown in (2). This concept of classified vocabulary indicates classification of groups of words which are supposed to belong to the same category in protoJapanese ( i.e., they probably were pronounced with the same accent patterns at that stage). Since its establishment in the 1960s, this list of classified vocabulary has been extensively used by Japanese researchers doing field work on Japanese dialects.

434 Langue and Parole in Synchrony and Diachrony (2) Some Examples from the Kindaichi Vocabulary for Nouns [ one-mora words ] [Class 1] e (handle of an umbrella), ka (mosquito), ko (child), ci (blood), to (door) [Class 2] ha (leaf), hi (sun, day), mo (seaweed), na (name), ya (arrow) [Class 3] e (picture), hi (fire), ki (tree), me (eye), su (vinegar), ta (rice field), [ two-mora words ] [Class 1] hana (nose), ame (candy), hako (box), hasi (edge), kaze (wind), mizu (water), musi (insect), sake (wine), tori (bird), usi (cow) [Class 2] oto (sound), hasi (bridge), isi (stone), iwa (rock), kami (paper), kawa (river), mura (villlege), nasi (pear), uta (song), yuki (snow) [Class 3] yama (mountain), asi (foot), ami (net), hana (flower), inu (dog), kami (hair), kumo (cloud), kusa (herb), sima (island), uma (horse) [Class 4] hune (boat), iki (breath), hasi (chopstick), hari (needle), ita (board), ine (rice), kasa (umbrella), miso (bean paste), mugi (wheat), usu (mortar) [Class 5] ame (rain), ase (sweat), aki (autumn), haru (spring), kage (shadow), koe (voice), saru (monkey), nabe (pot), mado (window), oke (pail) [ three-mora words ] [Class 1] iwasi (sardine), kuruma(wheel, vehicle), kemuri (smoke), minato (harbour), musuko (son), sakana (fish), sakura (cherry tree), yanagi (willow) [Class 2] azuki (red bean), higasi (east), musume(daughter), tokage (lizard) [Class 3] komugi (wheat), misaki (cape, headland), cikara (power) [Class 4] atama (head), hasami (scissors), hayasi (woods), hikari (light), hukuro (bag, sack), kotoba (word, language), koyomi (calendar), tawara (bale of rice), omote (surface) [Class 5] asahi (morning sun light), abura (oil), awabi (abalone), hasira (pillar), inoci (life), karei (flatfish), kokoro (heart), makura (pillow), namida (teardrop), sugata (figure) [Class 6] dango (dumpling), hibari (lark), karasu (crow), mimizu (earthworm), nezumi (mouse), usagi (rabbit), unagi (eel), yomogi (mugwort) [Class 7] icigo (strawberry), hatake (field), kabuto (helmet), kusuri (medicine), midori (green), tamago (egg), yamai (illness) None of the present Japanese dialects retain all the distinctions of the classes shown in (2), because they all underwent mergers of some or all of these tonal patterns. However, the tonal system of the early-twelfth-century Kyoto dialect suggested by the Ruijumyogisho makes the distinction for all the tonal patterns in (2). The Ruijumyogisho-system (hereafter referred to as the Old Kyoto system) contains three tonal classes for one-mora nouns, five for two-mora ones, seven for three-mora ones, as shown in (3). (Hereafter, H stands for high tone, L for low tone, M for mid tone, F for falling tone.)

Accentual Proto-System of Mainland Japanese Dialects 435 (3) Accent System of Nouns in the Old Kyoto Dialect in the Ruijumyogisho (early 12th century Kyoto) [one-mora nouns] (vowel-lengthened) 1. HH 2. HL 3. LL

[two-mora nouns ]

[three-mora nouns ]

1. 2. 3. 4. 5.

1. 2. 3. 4. 5. 6. 7.

HH HL LL LHH(//// ga) LFL (//// ga)

HHH HHL HLL LLL LLH LHH LHL

(The numbers 1,2,3... indicate the tonal classes) (// stands for a mora) It has been generally assumed among linguists specializing in Japanese dialects, that the accent system of proto-Japanese (at least proto-mainland-Japanese, excluding the dialects in the Ryukyuan islands) is that of the above-mentioned twelfth-century Japanese shown in (3). This hypothesis is stated as follows: (4) Hypothesis : The proto-accentual system of Japanese dialects is that of the Old Kyoto dialect (3) recorded in Ruijumyogisho However, certain dialects in the Shikoku area have presented problems concerning the validity of this assumption, and they have come to be considered crucial for the discussion of the proto-system of Japanese accents. Various linguists in the past have been aware of this, but presented conflicting hypothetical analyses of the problem. This paper rejects the traditionallyaccepted hypothesis (4), and proposes a system one stage prior to (3) as a accentual system of proto-mainland-Japanese dialects.

TYPOLOGY OF TONAL CHANGE Among tonologists of Japanese (e.g., Kindaichi (1973) ) as well as investigators of other languages such as Bantu languages (e.g., Hyman and Schuh (1974)), there is an understanding that tones do not change at random but in a certain regular fashion which is common to all systems; i.e., there is a general tendency for tonal changes. Before attempting our reconstruction, it is necessary to consider what this general tendency of tonal change is. This topic has been amply investigated in the past 70 years by empirical studies done by Japanese accentologists on the Japanese pitch accent systems. The first natural tendency of tonal change delineated by tonologists such as Kindaichi (1973) and Hyman and Schuh (1974) is that tones tend to spread rightward rather than leftward. For example, as shown in (5a), the tonal pattern HLLL tends to change to HHLL, and subsequently to HHHL; i.e., the H-tone moves to the right. Also in (5b), LHHH tends to shift to LLHH and then to LLLH; i.e., the L-tone also moves rightward. (5) Rightward Shift of tone spreading (a) HLLL>HHLL>HHHL (* HHLL>HLLL) (b) LHHH>LLHH>LLLH (*LLHH>LHHH)

436 Langue and Parole in Synchrony and Diachrony This shows that both the H-tone and L-tone tend to spread from the beginning of a tonal phrase toward the end, rather than in the opposite direction. The next most common tendency of tonal change proposed in Kindaichi (1973), as well as others, is that, at the beginning of a word or a phrase, a sequence of the same tone tends to be avoided. Thus, the sequence of L tones tends to be avoided by inserting an H tone at the beginning of each phrase, as in (6a), while a sequence of H tones is avoided by inserting an L tone at the beginning, as in (6b): (6) Polarized Tone Insertion at the beginning of each tonal phrase a. LLL>HLL b. HHH>LHH Since they both operate to insert the tone opposite to the next one to the right, this process will be called "polarized tone insertion". Another natural tendency of tonal change is that, when two H tones cooccur in a single tonal phrase as in HLH, they tend to be 'simplified' by either one of the following operations: single peaking, by which HLH becomes HLL or HHL, as illustrated in (7a); plateauing, by which HLH becomes flat HHH, as illustrated in (7b); and downstepping, illutrated in (7c), by which the second H tone in HLH sequence is lowered to a mid-tone and the whole phrase goes to HHM (M here indicates a mid-tone): (7) Subsequent Shift of the Tonal Phrase with Two H Tone Peaks (a) HLH>HLL(HHL) (Single Peaking) (b) HLH>HHH (Plateauing) (c) HLH>HHM (Downstepping) These three separate changes (among them, only the first one was given by Kindaichi (1973), the other two have been given by typological studies of tonal change in Bantu tone and pitch accent languages) all operate to avoid the realization of two H tones in a single word or phrase, and further to realize a universal restriction that a word (or phrase) has only a single main peak of H-tone ( or stretch of H tones); i.e, they (especially (7a) and (7b)) operate in order to realize a universal restriction of "one H tone per word". The final natural tendency of tonal change proposed by Kindaichi (1973) is clearly due to the systemic nature of tonal change. This shows that, similar to other phonological change, a tonal change usually motivates a change in other tonal patterns within the same system, which was referred to as "parallel tone shift" by Kindaichi (1973). For example, as shown in (8a), if the tonal pattern HLLL at Stage I changes into HHLL at Stage II in the process of historical change, another tonal pattern HHLL in the same system at Stage I should receive the input of this type of change; i.e., the H-tone in this tonal pattern also spreads one mora to the right. As a result, the original HHLL changes into HHHL in Stage II, as shown in (8b). A similar event also takes place for L-tone spreading as shown in (8x) and (8y); i.e., if a tone pattern LHHH at Stage I changes into LLHH at Stage II, LLHH at Stage I also has to undergo the similar type of change; i.e., the shift of L-tone one mora to the right at Stage II should take place for another tonal pattern LLHH at Stage I and change it into LLLH. (However, in some cases, even though a certain tonal pattern, for example HLLL, changes into HHLL, there is no change of HHLL in the original system into HHHL. In such cases, these two patterns will merge into one tonal

Accentual Proto-System of Mainland Japanese Dialects 437 pattern (in this case HHLL). It is clear that once the two tonal patterns merge into one pattern, they cannot be distinguished in the same way as before.). (8) Parallel Tone Shift within the same system Stage I Stage II Stage III (a) HLLL > HHLL > HHHL (b) HHLL > HHHL > HHHH Stage I (x) LHHH (y) LLHH

> >

Stage II LLHH LLLH

> >

Stage III LLLH LLLL

Thus, for example, it is not acceptable to assume that a single tonal pattern HLLL in a system changes into HHLL and further into HHHL, disregarding all the rest of the patterns in the same system (triggering no change to another tonal pattern (e.g.HHLL ) in the original system). This indicates that tonal change is very systemic in nature. Hereafter, we will base our analyses in the following sections only on the type of change illustrated in (5) to (8) in this section.

PROBLEMS CONCERNING DIALECTS IN THE SHIKOKU AREA If we observe the typology of tonal change just illustrated, certain accentual systems in dialects in the Shikoku area in Japan present problems with regard to the traditional idea that they are developed from the Old Kyoto system (3). The dialect of Takamatsu (See Map 2) is one such system. The following is the tonal pattern of Takamatsu dialect arranged according to the hypothesis of word classes introduced earlier. Notice that in this dialect, two-mora Class 1 and Class 3 nouns with the attachment of the nominative marker 'ga' both appear with HHH tonal patterns; moreover, three-mora Class 1 and Class 4, as well as (some) Class 5 nouns all show the HHHH pattern when 'ga' is attached. The reason for the merger of two-mora Class 1 and 3 patterns into HHH and that of three-mora Class 1, 4, and 5 patterns into HHHH cannot be explained by the assumption that this dialect is a direct descendant of the Old Kyoto dialect shown in (3), in which two-mora Class 1 has HH pattern while Class 3 has LL, and three-mora Class 1 has HHH, Class 4 LLL, Class 5 LLH patterns. (9) Present-day Takamatsu Accent System [ one-mora nouns] C" ga) 1. HL,HH 2. HL 3.LL

[ two-mora nouns] (juju ga) 1. HHH 2. HLL 3. HHH, HLL 4. LLH 5. LHL

[ three-mora nouns] (nnn ga) 1. HHHH 2. LHLL-HHLL 3. HLLL 4. HHHH 5. HLLL, HHHH 6. LLLH 7. LHLL

438 Langue and Parole in Synchrony and Diachrony Let us illustrate this problem by focusing on three-mora nouns (the question marks below indicate some problematic points concerning the hypothesis that the dialect developed from the Old Kyoto system; i.e., the particular changes do not strictly observe the principle of parallel development described in (8)): (10) Old Kyoto (3) Present-day Takamatsu (9) PPP

l.HHH 2. HHL 3. HLL 4. LLL 5. LLH 6. LHH 7. LHL

>

> > > > > > >

[three-mora nouns]

VW C"A" ga)

HHH (HHHH) LHL (LHLL) ~ HHL(HHLL) HLL (HLLL) HHH (HHHH) ? HLL (HLLL), HHH (HHHH) ? LHH (LHHH) LHL (LHLL)

In order to obtain the HHH tonal pattern for Class 4 and 5 from the Old Kyoto patterns LLL and LLH respectively, by observing the natural tendency of tonal shift we previously discussed in (5), we need to assume that Class 4 pattern must have undergone some intermediate stages with a series of changes such as LLL>HLL>HHL>HHH; also Class 5 patterns must have changed from LLH> LLL>HLL>HHL>HHH. However, if such were the case, then these drastic changes must have influenced other tonal patterns in the same system through the process of change. However, notice that most of the other classes in this system were not affected by such change; it is especially incomprehensible why Class 2 (HHL) or Class 3 (HLL) in the Old Kyoto system did not undergo any change. How could it be possible that the tonal patterns of three-mora Class 4 and (some of) Class 5 nouns changed in Takamatsu, bypassing all other tonal patterns in the same system? A similar type of question is applicable to the system of Marugame (See Map 2) shown in (11). This dialect also presents problems with regard to the traditional hypothesis that it evolved from the Old Kyoto system (3). Notice here that the tonal patterns HHM is observed in two-mora Class 1 and Class 3 nouns, while HHMM is observed in three-mora Class 1, Class 4 and some Class 5 nouns: (11) Present-day Marugame Accent System [ one-mora nouns] C" ga)

1. HL 2. HL 3. LL

[ two-mora nouns] (#" ga)

1. 2. 3. 4. 5.

HHM HLL HHM LLH LHL

[ three-mora nouns] (HW ga) 1. HHMM 2. LHLL-HHLL 3. HLLL 4. HHMM 5. HLLL-HHMM 6. LLLH 7. LHLL

If we adopt hypothesis (4) that system (11) developed from that of Old Kyoto (3), the development of the tonal patterns HHM for two-mora Class 3 or HHMM in three-mora Class 4 and 5 in this dialect cannot be accounted for by the natural tendency of tonal change discussed earlier. This is illustrated by two-mora and three-mora nouns in (12): we cannot explain why the sudden drastic change of two-mora Class 3 (LLL>HHM) and three-mora Class 4 (LLL>HHM), and Class 5 (LLH> HHM) took place, while other tonal patterns either stayed

Accentual Proto-System of Mainland Japanese Dialects 439 the same (e.g. two-mora Class 2, or three-mora Class 3, 6 and 7) or were subject to the minimal change (e.g. two-mora Class 4 and Class 5 or three-mora Class 2 ). (12) Old Kyoto (3) > Present-day Marugame (11) [two-mora nouns] 1. 2. 3. 4. 5. [ three-mora nouns]

fj. /j. ga HHH HLL LLL LHH LFL // ju /s

>

[two-mora and three-mora nouns]

// fj. ga

> HHM > HLL

> HHM? > LLH > LHL >

ju ju ju

(ju/jp ga)

1. HHH > HHM (HHMM) 2. HHL > LHL (LHLL)-HHL (HHLL) 3. HLL > HLL (HLLL) 4. LLL > HHM (HHMM) ? 5. LLH > HLL (HLLL), HHM (HHMM) ? 6. LHH > LHH (LHHH) 7. LHL > LHL (LHLL) Another accent system that poses problems to the traditonal hypothesis (4) is the dialect of Ibuki-jima, which is a small island located in the Inland Sea in an area close to the northeast of Shikoku (See Map 2). This dialect, reported first by Wada (1966), is well-known among Japanese linguists as the dialect which retains the greatest number of tonal patterns in Japan; in fact, this is the sole dialect that retains the 5-way distinction for two-mora nouns corresponding to the five classes established in the Word Classes previously introduced in (2). Notice in (13) that, contrary to the Takamatsu or Marugame systems, the tonal distinction between two-mora Class 1 and Class 3 nouns, and that between Class 1 and Class 4 ( and part of Class 5 ) nouns are retained in this dialect. That is, the tonal patterns HHM developed only for two-mora Class 3, and HHMM for three-mora Class 4 and some Class 5 nouns in this dialect. On the other hand, the patterns of Class 1 nouns are HHH or HHHH: (13) Present-day Ibuki-jima System [ one-mora nouns] C" ga)

1. HH-LH 2. HL

3. LL-LH

[ two-mora nouns] C">" ga) 1. HHH 2. HHL 3. HHM

4. LLH-LLL 5. LHL

[ three-mora nouns] (VW ga)

1. 2. 3. 4. 5. 6. 7.

HHHH LHHL,HHHL HHLL HHMM HLLL, HHMM LLLH-LLLL LHHL

As illustrated by two-mora and three-mora nouns in (14), observing the natural tendency of tonal change discussed in (5) to (8), it is impossible to assume that the tonal patterns HHM or HHMM developed from LLL or LLH in the Old Kyoto system (3):

440 Langue and Parole in Synchrony and Diachrony (14) Old Kyoto (3) > Present-day Ibuki-jima (13) [two-mora nouns]

// // ga l.HHH

> /j /j ga > HHH

2. HLL 3. LLL 4. LHH 5. LFL

> HHL > HHM ? > LLH > LHL

[three-mora nouns ]

n ju p 1. HHH 2. HHL 3. HLL 4. LLL 5. LLH 6. LHH 7. LHL

[two-mora and three-mora nouns]

> / / / / / / (jJju/i ga) > HHH (HHHH) > HHL>LHF (HHLL>LHHL) > HLL>HFL (HLLL>HHLL) > HHM (HHMM) ? > HLL (HLLL), HHM (HHMM) ? > LLH (LLLH) > LHL>LHF (LHLL>LHHL)

Along with Ibuki-jima, the dialect of Shishi-jima, an island also located in the Inland Sea near the northeastern part of Shikoku (See Map 2), also poses a puzzle to the traditionallyaccepted hypothesis (4). Notice that in this dialect, Class 1 nouns of all numbers of morae end with L tones, i.e., they manifest themselves with a drop in pitch from H to L, as HL for one mora Class 1 nouns, LHL for two mora Class 1 nouns, LHLL for three mora Class 1 nouns: (15) Present-day Shishi-jima Accent System [one-mora nouns]

[ two-mora nouns]

[three-mora nouns]

C" ga)

(/4W ga)

(fW

1. HL 2. HL 3. LL ~ LH

1. 2. 3. 4. 5.

1. 2. 3. 4. 5. 6. 7.

LHL HLL HHL-LHL LLL LHL

ga )

LHLL-HHLL LHLL HLLL HHLL ~ LHLL HLLL, HHLL-LHLL LLLL LHLL

If this dialect were a direct decendant of the Old Kyoto (3), we must assume that the change HH > HL, HHH > HHL, HHHH > HHLL took place for Class 1 nouns, as shown in (16). However, this type of change is an obvious violation of the tendency described earlier in (5a) for the H tone to spread to the right:

A ccentual Proto-System of Mainland Japanese Dialects 441 (16) Old Kyoto (3) > Present-day Shishi-jima (15) [one-mora nouns]

[two-mora nouns]

// ga 1. HH 2. HL 3. LL // // ga l.HHH 2. HLL 3.LLL 4. LHH 5. LFL

[three-mora nouns] / / , " / / l.HHH 2. HHL 3. HLL 4. LLL 5. LLH 6. LHH 7. LHL

> /J. ga > HL ? > HL > LL > > > > > >

// n ga HHL ? HLL HHL LLL LHL

> / / / / / / G"A« ga) > > > > > > >

HHL (HHLL) ? LHL (LHLL) HLL ( HLLL) HHL ( HHLL) HLL (HLLL), HHL (HHLL) ? LLL (LLLL) LHL (LHLL)

In sum, the dialects discussed in this section (Takamatsu, Marugame, Ibuki-jima, Shishijima) all present some challenges to the hypothesis (4) that they developed from the Old Kyoto system (3).

ACCENTUAL RECONSTRUCTION OF THE PROTO-SYSTEM OF MAINLAND JAPANESE DIALECTS Proto-Svstem of Accent of Mainland Japanese Dialects In the previous section, we have observed that the tonal patterns of Class 1 nouns varied depending on the dialects. The two-mora Class 1 nouns, for example, with the attachment of the nominative marker 'ga', manifest with the HHH pattern in the Takamatsu, Ibuki-jima systems as well as in the Old Kyoto system, with HHM pattern in the Marugame system, and with HHL pattern in the Shishi-jima system. This variation of tonal patterns of Class 1 nouns gives a clue to the solution of the problems. This fact indicates that the proto-form of Class 1 nouns should be the one from which these three tonal patterns could naturally develop. As noted earlier in (7), when two H tones cooccur in a single tonal phrase, they tend to be 'simplified' by either one of the following operations: plateauing (HLH > HHH), downstepping (HLH > HHM), and single-peaking (HLH > HLL). In order to derive all of these three different tonal patterns observed in the Shikoku area from one proto-form, the author (1993, 1997) assumed that the proto-form of Class 1 nouns for one-mora, two-mora, and three-mora nouns contained * HLH, and the various tonal patterns in these dialects evolved by the process shown in (17):

442 Langue and Parole in Synchrony and Diachrony (17) The Development of Class 1 Tonal Patterns in Shikoku Area

The HHH pattern observed in Class 1 nouns in Takamatsu (9), Ibuki-jima (13), as well as the early twelfth century Old Kyoto system (3) developed through plateauing (7b) of the HLH pattern to the level-high tonal pattern (HHH). On the other hand, the HHM pattern observed in Class 1 nouns in the Marugame system (11) developed by means of downstepping (7c). Furthermore, the HHL pattern in those in Shishi-jima developed by single-peaking (7a) as in HLH>HLL>HHL. In this way, we can obtain all the three different tonal patterns observed in Class 1 nouns in present-day dialects in Shikoku area, by means of a natural tonal change described earlier in this paper. In this way, Class 1 pattern is assumed to have once been *HLH. However, in the Old Kyoto system shown (3), the Class 1 tonal pattern (for all numbers of morae) is HHH. If we assume that Class 1 nouns once were * HLH., the tonal change * HLH > HHH (by means of plateauing) should have taken place from the proto-system to the system of twelfth-century Old Kyoto (3); this fact indicates that it is necessary to go back to the system one stage prior to the system of Old Kyoto. Based on the Old Kyoto system (3), and considering the natural tendency of tonal change described in (5) to (8), the proto-system for mainland Japanese dialects is reconstructed as shown in (18): (18) Proto-System of Accent for Mainland Japanese Dialects [one-mora nouns]

O ga) 1. *FH 2. *LL 3. *LH

[two-mora nouns]

nV

(W ga)

1. *FH(HLH) 2. * LL 3.* LH 4. *HH 5. *HL

[three-mora nouns]

// /" H 1. *HLH 2. *HLL 3. *LLL 4. *LLH 5. *LHL 6. *HHH 7. *HHL

The change from this system to that of Old Kyoto (3) is demonstrated by three-mora nouns as follows. Notice that the change from (19a) to (19b) is completed observing the common tendency of tonal change discussed earlier in (5) and (6); i.e., the changes of Class 3, 6 and 7 is explained by polarized tone insertion (6), that of Class 1 by plateauing (7b), that of Class 2 by rightward shift of the H-tone (5a), and those of Class 4 and 5 by that of the L-tone (5b):

Accentual Proto-System of Mainland Japanese Dialects 443 (19) Development of Old Kyoto (3) from the Proto-System (18) (illustrated by three-mora nouns) a. Proto-Accentual System (18)

b. Old Kyoto (3) fj. ju n

ju ju ju Class 1 Class 2 Class 3 Class 4. Class 5 Class 6 Class 7

HLH HLL LLL LLH LHL HHH HHL

> > > > > > >

HHH HHL HLL LLL LLH LHH LHL

Application of (18) to Dialects in and around Shikoku Let us see how the proto-system (18) can explain the problematic points of Shikoku dialects mentioned above; i.e., the problems arising from the assumption that they are all derived from the Old Kyoto system (3). From the proto-system (18) to the Takamatsu System (9), we will assume that the tonal change plateauing (HLH>HHH) took place for Class 1 words (for words of all morae), as well as Class 3 for two-mora words, and Class 4 and 5 for three-mora words. This is illustrated in (20): (20) From the Proto-System (18) to the Takamatsu System (9) [two-mora nouns] // // (jjju ga) Class 1 *FH(HLH) > HH (HHH) Class 3 *LH > FH(HLH) > HH (HHH) [three-moranouns]

// /j ju (////// ga) Class 1 *HLH > HHH (HHHH) Class 4 *LLH > HLH > HHH (HHHH) Class 5 *LHL > HLH > HHH (HHHH)

Thus, we are able to explain why in the Takamatsu dialect, HHH is observed for two-mora Class 3 and three-mora Class 4 and 5 nouns, since their tonal patterns (LH, LLH, LHL) can easily change to HLH (FH) tonal pattern. From the proto-system (18) to the Marugame System (11), we will simply explain the problems by assuming that downstepping (HLH>HHM) took place instead of Takamatsu's plateauing (HLH>HHH). This is illustrated by two-mora and three-mora nouns as follows:

444 Langue and Parole in Synchrony and Diachrony (21) From the Proto-System (18) to the Marugame System (11) [two-mora nouns] Class 1 Class 3

// ju (ju/j ga) *FH(HLH) > HM(HMM) *LH > FH(HLH) > HM (HMM)

Class 1 Class 4 Class 5

// // // (////// ga) *HLH > HHM(HHMM) *LLH > HLH > HHM (HHMM) *LHL > HLH > HHM (HHMM)

[three-moranouns]

As a result of this, Class 1 and 3 for two-mora words, and Class 1,4, and part of 5 naturally merged in this dialect as well as in the Takamatsu system. In the Ibuki-jima system (13), we have seen that the distinction between two-mora Class 1 and Class 3, and that between three-mora Class 1 and Class 4, (and part of Class 5) is still retained. We can explain this by assuming that in this dialect, both the process of plateauing (HLH > HHH) and that of downstepping (HLH>HHM) took place, but in a different stage of its historical development. This is illustrated in (22) by three-mora nouns. We will assume that from the proto-system (18) to the Ibuki-jima System (13), first, the tonal pattern HLH of Class 1 changed to HHH by way of plateauing (HLH>HHH) (Stage A); subsequently, Class 4 and 5 changed to HLH, then changed to HHM by way of downstepping (HLH>HHM) (Stage B): (22) From the proto-system (18) to the Ibuki-jima System (13) (illutrated by three-mora nouns)

Class 1 Class 2 Class 3 Class 4 Class 5 Class 6 Class?

ft n n *HLH > *HLL > *LLL > *LLH > *LHL > *HHH > *HHL >

[Stage A] [Stage B] p // n (nw ga) n p p (wfi ga) HHH (HHHH) HHL (HHLL) HLL (HLLL) HLH > HHM (HHMM) HLH > HHM (HHMM) LHH (LHHH) LHL(LHLL)

[Present-day / / / / / / ga n n n ga HHHH > HHHH HHLL > HHHL HLLL > HHLL HHMM > HHMM HHMM > HHMM LHHH > LLHH (LLLH) LHLL > LHHL

In this way, we may assume that in the Ibuki-jima dialect, both plateauing (7b) and downstepping (7c) took place, but in different periods, so the distinction between HHH and HHM is still maintained in this dialect. In the Shishi-jima dialect (15), we assume that Class 1 nouns for all numbers of morae were subject to the process of single-peaking (7a) instead of plateauing (7b) in the Takamatsu dialect, or downstepping (7c) in the Marugame dialect. As a result, in this dialect, too, the tonal patterns of the two-mora Class 1 and Class 3 and the three-mora Class 1, Class 4 and part of Class 5 merged.

Accentual Proto-System of Mainland Japanese Dialects 445 (23) From the proto-system (18) to the Shishi-jima System (15) [one-mora nouns]

// // (//// ga)

// //

(In this dialect, vowels of one-mora nouns are lengthened to become two-morae.) Class 1 *FH(HLH) > HL Class 2 *LL > HL

Class 3 [two-moranouns]

*LH

>

LL

Class 1 Class 2 Class 3 Class 4 Class 5

^ // (^ ga) *FH(HLH) > *LL > *LH > *HH > *HL >

p ju ga HHL(LHL) HLL HHL(LHL) LHH (>LLH>LLL) LHL

Class 1

// // p. *HLH

>

// /j. // (jujuju HHL (HHLL)

Class 2

*HLL

>

HHL (HHLL)

Class 3 Class 4 Class 5 Class 6

*LLL > *LLH > HLH > *LHL > HLH > *HHH >

HLL(HLLL) HHL (HHLL) HHL (HHLL) LHH (LHHH) (> LLHH)

Class 7

*HHL

LHL (LHLL)

[three-mora nouns]

>

ga)

In this way, the newly proposed proto-system (18) can solve the problems posed by the traditionally-accepted hypothesis (4) with regard to some of the Shikoku dialects focused on in this paper.

CONCLUSION Past studies of accents in Japanese dialects have revealed that there is a cross-dialectal regular correspondence of tonal patterns. We have seen that thanks to this regular correspondence of tonal patterns between present-day Japanese dialects, research is now at a stage at which we have a great opportunity to attempt reconstruction of the proto-accentual system of Japanese. Based on the assumption of word classes proposed by Kindaichi and Wada (1955) and Kindaichi (1974), and using the very systematic nature of accent change described in (5) to (8), this study has attempted a reconstruction of the proto-accentual system of mainland Japanese dialects. The present study has returned one stage prior to the early-twelfth-century Old Kyoto system (3), and presented a reconstructed system of proto-mainland-Japanese (18). The study has shown how the newly-proposed system can explain the problems arising from some of the Shikoku dialects, hitherto considered to be systems presenting problems for the traditionallybelieved idea that they had evolved from the Old Kyoto system (3).


REFERENCES Hyman snd Schuh (1974). Universals of Tone Rules: Evidence from West Africa. Linguistic Inquiry , 5, 81-115. Kindaichi, Haruhiko (1973). Hikaku-hougengaku to Hougen-chirigaku. Kokugo to Kokubungaku , 50-6, 1-19. Kindaichi, Haruhiko (1974). Kokugo-akusento no Shiteki-kenkyu: Genri to Houhou. Hanawa Shobo, Tokyo. Kindaichi, Haruhiko and Minoru Wada (1955). Kokugo Akusento Ruibetsu Goi Hyou. Kokugogaku-jiten, 994-8. Matsumori, Akiko (1993). Nihongo Akusento no Sotaikei Saiken no Kokoromi: Iwayuru "kakou-shiki akusento" no Seiritsu ni kansuru Kousatsu o motoni shite. Gengo-kenkyu, 103: 37-91 (1997). Tokushima-ken Wakimachi, Mikamochou no Akusento to Hondo-sogo no Akusento Taikei. Kokugogaku , 189: 15-28. Shibatani, Masayoshi (1990). The Languages of Japan. Cambridge University Press, Cambridge. Wada, Minoru (1966). Daiichiji akusento no hakken: Ibuki-jima. Kokugo-Kenkyu, 22:24-

Accentual Proto-System of Mainland Japanese Dialects 447

[Map 1] Accent in Japanese Dialects

[Map 2] Map of Eastern Shikoku


X

GENERATIVE GRAMMAR


35

TEMPORAL REASONING IN ITERATIVE AND HABITUAL CONTEXTS Florian Panitz, University of Oldenburg, Germany

Abstract Iterative and habitual temporal patterns as they are expressed in an authentic text sample are analyzed to show how default principles govern the interpretation of narrative texts. A new pragmatic framework is outlined which explains the logical relations amongst such principles in a reference time oriented model. Conclusions are drawn as to where temporal interpretation becomes a performance phenomenon, i.e. where the reconstruction of temporal patterns might vary from individual to individual, and how the framework presented here can incorporate such performance aspects of temporal reasoning.

i INTRODUCTION It is often revealing to describe the meaning content of temporal adverbials, tenses and other temporal elements in truth conditional semantic frameworks. Thus, a sentence is usually true if the eventuality described holds or takes place at a certain time and relative to one or more other events or times, whereby the exact nature of these temporal relations is given by the tenses, adverbials, or the lexical content of eventuality descriptions (cf. e.g. Binnick 1991: 175 ff., Nerbonne 1984, Comrie 1989). The truth conditional type of analysis leads to problems when dealing with texts (see e.g. Hamann 1991, Nerbonne 1984: 23, Partee 1984). For example, the succession of eventuality descriptions in [1] Jane mailed and [2] [Jane] wrote the letter is probably false, although the component parts [1] and [2] may be true. Instead of assuming strict rules by which the arrangement of eventualities can be compositionally derived from the invariant semantic meanings of the relevant components of utterances or texts, some frameworks are based at least in part on the assumption that the temporal interpretation on the text level follows pragmatic principles which are defeasible (Ehrich 1992, Lascarides 1992, Fleischman 1991). It might, for instance, be appropriate to say Jane mailed and wrote the letter when we consider additional contextual information to the effect that we do not want to relate the events in chronological succession, but rather talk about the things that Jane did with the

451

452 Langue and Parole in Synchrony and Diachrony letter besides mailing it. The 'normal' assumption that the succession of [1] and [2] is false and hence inappropriate would then be suspended - it is defeasible. Defeasible assumptions about temporal arrangements may also just be modified or specified in light of a larger context. Can the interaction of such defeasible inferences in text interpretation be grasped in a logically consistent framework? And can such a framework do justice to the complexity of textual reality? To what extent might the inferences vary from individual to individual and thus become a performance phenomenon? To answer these questions I shall first take a look at a rather complex temporal phenomenon, namely the conceptualization of iterative and habitual perspectivization (section 2). I shall then demonstrate the role of such conceptual patterns in an authentic text sample (section 3). The main portion of this paper deals with a newly-developed pragmatic system of defeasible inferences, as I have outlined in Panitz (1998) (section 4). The framework helps to explain how we reconstruct temporal arrangements expressed in narrative texts by applying default inferences. This system is applied to the authentic text sample (section 5). Finally, some suggestions are made how the framework which is applied here can incorporate competence-performance interfaces (section 6).

2 ITERATIVE AND HABITUAL PERSPECTIVIZATION A habitual situation is either characterized by an eventuality that occurs repeatedly (as a habit) or by the existence of a stative situation that holds throughout a protracted period of time, and whose existence characterizes this period (see e.g. Comrie 1976: 27, 28). Habitual predications may not only apply to individuals, but also to things, as e.g. in connection with used to; see e.g. (3) below (also cf. Smith 1991: 39). Iterative situations are made up of eventualities that happen repeatedly. Iterative and habitual situations show close conceptual affinity, because often referring to the repeated occurrence of something entails that the situation re-occurs as a habit throughout a rather protracted period of time, as in: (1) Jane always used to arrive late. However, iterative situations do not necessarily have to be habitual on the one hand: (2) John sneezed (/*used to sneeze) several times and stood up. Because the time period referred to in (2) cannot really be protracted, the situation is not habitual. On the other hand, habitual situations do not necessarily have to involve iterativity (cf. Comrie 1976), as in: (3) The temple of Diana used to stand at Ephesus.

Temporal Reasoning 453 We should note that habituality is not a very concrete notion, while iterativity represents a clear temporal pattern that e.g. in connection with the simple-past tense roughly corresponds to the following arrangement (for similar representations see e.g. Panitz 1994, Schopf 1984):

In the representation used here event times are the times at which states hold or terminative events take place. Reference times are the times we actually talk about and from which the events are seen and perspectivized in addition to the speech time: in We used to have dinner at 8:00p.m., for example, the reference time is '8 o'clock1, but the event time (the time interval during which we actually ate) may typically be e.g. '8:00 - 8:45' and vary from occurrence to occurrence. Event times can be points or intervals, reference times are points only. Thus, in J. walked to the station it is the reference point when J. reached the station that is relevant for the succeeding text portion (see e.g. Krifka 1989: 75 ff. and Panitz 1998: 44 ff. for discussion). According to (4) an eventuality takes place repeatedly, i.e. at each event time el, .... en. Each occurrence may be seen as an unanalyzable whole ('blob'-like) from the corresponding reference times rl, ... rn. Taken together, the repeating occurrences of eventualities constitute a stative situation (the "state" of something happening repeatedly) which holds at "super" event time e'. This stative situation is called eventuality Y here. Eventuality Y may also be seen perfectively from a reference time. Assigning a reference time to one of the events at el, ...., en means that we automatically also assign a reference time to the super-eventuality Y, i.e. eventuality Y is then seen from the particular perspective of that reference time. Instead of having one event at each event time el, ...., en, also a sequence of events may take place at each of these times. In this case iterativity involves the repeated occurrence of a sequence of events.


3 AN AUTHENTIC TEXT SAMPLE The following example involves iterativity and habituality, but entails much more than the temporal structures mentioned above: (5) [1] Here at the age of thirty-nine I began to be old. [2] I felt stiff and weary in the evenings [3] and [I felt] reluctant to go out of camp; [4] I developed proprietary claims to certain chairs and newspapers; [5] I regularly drank three glasses of gin before dinner, never more or less, [6] and went to bed immediately after the nine o'clock news. [7] I was always awake and fretful an hour before reveille. (E. Waugh (1983 /1945). Brideshead Revisited, p. 14. Harmondsworth, Penguin)

The most prominent feature of the temporal arrangement of eventualities expressed in (5) is that eventuality description [1] Here at the age of thirty-nine I began to be old sets a temporal and thematic frame for the rest of the excerpt; the whole excerpt is seen under the thematic aspect of [1]. [1] is inchoative, i.e. a situation of the type 'be-/feel- old' is gradually brought about (cf. e.g. Smith 1991: 44). Utterance [1] is also interpreted similarly to state descriptions (it can be interpreted as 'bein what we normally consider the early stages of being old'); the gradual process that leads to 'be- old1 can be protracted and thus becomes statelike (i.e. a state-of-affairs holding for a period of time). By the same token, the eventuality related in [1] can be seen as a habitual situation along the lines of (3) (habitual, not iterative). Because [1] sets a thematic frame for the excerpt, the whole text excerpt becomes habitual ([1] has temporal and thematic scope over the rest of the excerpt). [2] I felt stiff and weary ... and [3] ... and [I felt] reluctant to go out of camp are habitual and iterative (repeated stative situations). There is causality involved in [2] and [3]: 'I feelstiff and weary1 is the reason for 'I feel- reluctant to go out of camp'. [4] / developed proprietary claims to certain chairs and newspapers is inchoative like [1]; it is also habitual (it denotes a protracted situation, namely the gradual process of developing proprietary claims to something). Arguably, there is also iterativity involved here: the gradual acquisition of proprietary claims would then be seen as manifesting itself in certain ways of behaving that show up each evening. [5] / regularly drank three glasses of gin ... and [6] ... and went to bed immediately after the nine o'clock news refer to two events in chronological order; the sequence 'I drinkthree glasses of gin1 — 'I go- to bed1 happens repeatedly throughout the period of time referred to by at the age of 39 in [1]; [5] and [6] are habitual and iterative). In [7] / was always awake and fretful an hour before reveille a repeated stative situation is described; this situation typically follows 'I go- to bed' referred to in [6], but there is

Temporal Reasoning 455 a temporal gap between the eventuality of [6] and that of [7]; yet a strong thematic connection exists between the two. [7] is both habitual and iterative. In the next section we shall see how a pragmatic approach can deal with the rather complex temporal relations that underlie even a seemingly simple text excerpt like (5).

4 THE FRAMEWORK 4.1 General Considerations As in other types of default reasoning, the application of temporal default principles in text interpretation follows a general logical pattern of defeasible inferences which is called the 'Penguin Principle' (for the logical foundations of default reasoning along these lines see e.g. Brewka 1991, Lascarides & Oberlander 1993, Matsumoto 1995): (6)

The Penguin Principle

(6) shows the inferences that are connected with the statement Tweety is a penguin. Our world knowledge about the fact that penguins are birds leads us to two possible conclusions, namely Tweety is a penguin and a bird' in 2 and the simpler conclusion Tweety is a bird' in 3. 3 can be deduced from 2 by simplification: if two statements are true it follows that also only one of them is true. From 3 (T. is a bird1) we can now deduce Tweety can fly' in 5 on the basis of our knowledge about the fact that birds fly. From the more informative information base in 2 (T. is a bird and T. is a penguin'), however, we derive the contrary statement 4 Tweety cannot fly' by our knowledge about the fact that penguins don't fly even though they are birds. The last inference from 2 to 4 suspends the one from 3 to 5, because 4 is based on the more extensive information base in 2 (compared to 3). We can therefore say that the inference from 2 to 4 is more specific than the one from 3 to 5. More specific inferences either have a more extensive knowledge base than more general ones as their starting point, or they contain more

456 Langue and Parole in Synchrony and Diachrony inference steps. It might, for example, be possible to have intermediate steps in an inference as from 2 to 4. More specific principles suspend more general ones, if the results of the inferences are contradictory. Specific inferences modify and add information to the results reached by general ones, if there is no contradiction. More general inferences are shown to the left in the diagram, more specific ones on the right. Default reasoning follows Grice's quantity maxims "... make your contribution as informative as is required ...!" and "... do not make your contribution more informative than is required ...!" (cf. Grice 1975): we can proceed on the assumption that an author who suggests a complicated inference step considers this step relevant to the reconstruction of the context.

4.2 The Temporal Penguin Principle of Lascarides & Oberlander (1993) Lascarides & Oberlander (1993) applied the above inference pattern to temporal reasoning. The inferences connected with the sentence sequence Jack fell. John pushed him would roughly be represented in the following form:

Temporal Reasoning 457 (7) A Temporal Penguin Principle (cf. Lascarides & Oberlander 1993) Jack fell. John pushed him.

Again, inferences to the right are more specific, those to the left more general. The sentence sequence here provides us with specific contextual information that is shown in 1: means that we are confronted with two propositions (atomic or tenseless sentences) that are expressed by two adjacent enventuality descriptions (i.e. references to propositions). The two labels pi and p2 for the propositions are arranged in an ordered pair, because we are confronted with the eventuality description that expresses pi first, and then with the one that expresses p2. 'Jack fall-'=pl means for example that pi is of the type 'Jack fall-'; in other words, eventuality description pi refers to the event 'Jack fall-'. Finally, the more complicated ordered pair is used to show that the truth value of pi is determined relative to event time el, i.e. the event expressed by pi 'Jack fall-' takes place at el, and p2's event at e2. Following a pattern similar to that shown in in (6) we can now take 2 (which is identical to 1 to maintain logical consistency of the system) and deduce stage 3 from it (simplification), i.e. here we consider the fact that we are confronted with two adjacent eventuality descriptions only, notwithstanding their specific character. On the basis of 3 we would apply a very general 'Narrative Chronology Principle': the event time connected with pi is located anterior to the event time connected with p2; in other words, the order of eventuality descriptions corresponds

458 Langue and Parole in Synchrony and Diachrony to the order of the events described by them. Despite this, in (6) a more specific principle comes into play, due to the specific character of the propositions pi and p2. When the first event is of the type 'Jack fall-1 and the second one of the type 'John push- Jack', we are likely to apply knowledge about causality between "push-1 and 'fall-1: we know that normally 'push-1 is a prerequisite for 'fall-', here represented by Prer(push-, fall-). By applying this knowledge, we arrive at the arrangement of event times in 4: e2<el , i.e. the second event time (of'John pushJack') precedes el (of'Jack fall-'). Since the second inference step from 2 to 4 is based on more specific information than the one from 3 to 5 (2 is more specific than 3 because it considers a larger amount of information), the inference from 2 to 4 suspends the more general inference from 3 to 5.

4.3 A Modified Framework 4.3.1 Different Text Levels. The above pattern only accounts for the introduction of an eventuality description relative to the immediately preceding one. It is very likely, however, that considering two adjacent eventuality descriptions alone is not enough to gain an unambiguous temporal pattern. Thus, the interpretation of eventuality description (5) [7] I was always awake and fretful an hour before reveille requires one not only to consider the immediately preceding one [6]/7/ went to bed immediately after the nine o 'clock news, but also at least [1] Here at the age of thirty-nine I began to be old in order to grasp the contextual embedding of [7]. Of course, the temporal interpretation of [7] relative to [6] is still relevant. It takes place on the local text level of two adjacent eventuality descriptions. By contrast, interpreting [7] in light of [1] marks an inference step that is applied on the thematic text level, i.e. the inferences involved in the interpretation here contribute to identify a thematic text unit. By a "thematic text unit" we mean a coherent sequence of eventuality descriptions that is seen under a common thematic aspect. Text excerpt (5) represents a thematic text unit where the common thematic aspect and a common temporal frame is referred to in [1]. Possibly there might be hierarchical arrangements of several embedded thematic units in a text. There is also a global text level where, for example, relations that hold among the different chapters or even larger units of a book come into play. A system like (7) should at least explain the relationship between the relevant local' and 'thematic' inference steps that are involved in the immediate contextual embedding of utterances and distinguish between different inference patterns on these two levels. 4.3.2 Reference Time Orientation. A second necessary modification of (7) concerns the evaluation of a proposition relative to the event time only: in order to get a realistic picture of temporal conceptualization we have to introduce reference times, as already mentioned in

Temporal Reasoning 459 connection with (4). Instead of having as in (7), we should now have instead. This means that each proposition is not evaluated relative to its event time, but to an ordered pah- that consists of the reference time and the event time. The two times are arranged in a specific order here (hence the embedded ordered pair ), because we normally identify the reference time (the moment we talk about) first and then make inferences regarding the location of the event time relative to this point of orientation should they become relevant (cf. Schopf 1984 for discussion).

4.3.3 Thematic Principles. We shall focus on the principles that apply on the thematic text level first. The most typical relations between reference times and event times in narrative contexts can be subsumed under a general principle of narrative text constitution that we shall call 'Extended Reichenbach Pragmatics' or ERP: (8) Extended Reichenbach Pragmatics (ERP) Given a temporally coherent text unit <Si, 82, 83, ..., Sn> the following relation holds: r(Si)~1> r(Si+k), whereby 1- i - n and k - 1 and each reference point r stands in one of the following relations to the corresponding event interval: (I) r C e or (II) Rel(r,e). ^Si)"1 > r(Si+k) means that any reference time r(Si) should not be located posterior to any other reference time r(Si+k) that is introduced later in the text than r(Si). Reference times may be identical. In the marked context (7) the ERP would be suspended, but normally narrative texts follow the above principle, which also tells us something about the event times: r C e means that a reference time might be a part of the corresponding event interval or identical with the event interval (in the case of punctual events). In some cases there is no apparent concrete temporal relation between the reference time and the event time. In this case we must follow Rel(r,e), i.e. the event time is of particular relevance at the reference time. "Relevance" here means that either the event is seen perfectively as a single blob from the reference time, or that a more abstract relation exists between the event time and the reference time. Specific lexical (Aktionsart-) relations might suspend the ERP in this respect. In addition to the temporal arrangements expressed by the ERP, coherent thematic text units like (5) normally follow a principle which I shall call the Thematic Frame Principle:

460 Langue and Parole in Synchrony and Diachrony (9) Thematic Frame Principle In a succession of temporally coherent eventuality descriptions <S], 82,... Sn> the following relations hold: Etframe : (r(Sl), r(S2> - r(Sn)) C tframe & Assoc[(e(Sl), e(S2> •-., e(Sn)), tframeL whereby in the special case (i) tframe C (e(S1), e(S2>-, e(Sn)}, tframe C TADV and TADV represents the set of the time intervals established by temporal adverbials in Si - Sn, or in the normal case (ii) Assoc(tframe, t'), whereby t' C (e(Sl), e(S2> -, e(Sn)} or t' C TADV or t' can plausibly be deduced from the context in some other way. For the relations amongst the event times and the reference times of each pair in {, , ...,} principles (I) and (II) of the ERP hold. (9) simply states that all the event times of a thematic text unit are associated with a common temporal frame. The frame is normally not identical with one of the times explicitly referred to in the thematic unit, but rather must be deduced from one or more of these times. This is the case in (5): the temporal frame is somehow related to the event time of the gradual process 'begin- to be old' and the time mentioned in here at the age of thirty nine. The frame is not identical with either of these times, however. To generalize this, in (9) Etframe : (r(Sl> r(S2), ... r(Sn)) C tframe & Assoc[(e(Sl), e(S2> -, e(Sn)), tframe] means that there exists at least one temporal frame (EtframeX the reference times from r(Sl) (-reference time of eventuality description SI) to r(Sn) are included in the frame, and the event times of the text units are associated with the frame. This marks a special case, where the frame interval is one of the event times or one of the times mentioned by the adverbials. In the normal case, however, the frame interval is not directly given, but rather an interval t' which is associated with one of the event times or one of the times referred to by adverbials, as shown in (ii). Being able to establish a temporal frame for a text unit in narrative texts normally means that the event times and reference times are arranged according to the ERP in (8). For example, none of the relations expressed in (5) (which all follow principle 9) contradict the ERP. Therefore, the content of the ERP is also included in (9): the Thematic Frame Principle automatically entails compliance with the ERP. There is no contradiction between the ERP and the Thematic Frame Principle now, but we indicate that in narrative contexts the latter specifies the temporal relations suggested by the ERP. In narrative texts there must be a strict temporal sequence of reference times and event times at least somewhere in a thematic text unit (otherwise there would not be a main story line). It is therefore realistic to have a Thematic Chronology Principle in addition to the Frame Principle and the ERP:

Temporal Reasoning 461 (10) Thematic Chronology Principle In a succession of temporally coherent eventuality descriptions <Si, 82,...., Sn> the following relations hold: (a) r(Si) < r(S2), ... < r(Sn), as well as (b) Etframe: •••• (content of the Thematic Frame Principle) and (c) For the relations amongst the event times and the reference times of each pair in (,, ...,} (I) and (II) of the ERP hold. (a) says that the reference times occur in chronological succession; the reference time chronology corresponds to the order in which the times are referred to in the text. Following the chronological order also entails the content of the Frame Principle and consequently the relations between reference times and event times of the ERP, as shown in (c), because the Thematic Frame Principle shows no contradiction to the other two principles. It only specifies the relations given by the other two and it tells us how to specifically arrange the reference times when the other two principles hold. Amongst the thematic principles, (10) is therefore the most specific one; it is most likely to be suspended somewhere in the text (cf. e.g. Partee 1984), e.g. in connection with most state-of-being descriptions like The weather was nice and the birds were singing where the event times overlap and the reference times are probably identical (i.e. here background information supporting the main story line is given).

4.3.4 Local Principles. Let us for a moment consider two authentic utterances, namely [1] A date was fixed for Brideshead's wedding, early in the Christmas holidays, so that his future step-children might take part. [2] One afternoon in November Julia and I stood at a window in the drawing-room ... (E. Waugh, Brideshead Revisited, p. 341). We tend to see the events referred to in the two main clauses in chronological succession regardless of the thematic setup of the text: the event mentioned in the underlined portion of [1] most likely precedes that mentioned in [2]. Utterance [2] might e.g. introduce a completely new thematic text unit, but on the basis of [1] and [2] alone we do not know this for certain. Context [1] and [2] can thus be said to be 'underdetermined1 with regard to the thematic set-up of the text. Yet we are able to establish clear temporal relations between [1] and [2]. In addition to the thematic principles it seems therefore reasonable to assume principles that are at work on just these local levels. By 'local level1 we mean two adjacent eventuality descriptions in a text. The most general principle that is at work on the local level in narrative contexts is the Local Chronology Principle. Eventualities referred to by two consecutive utterances in narratives are usually arranged in the same order as they are mentioned in the text, regardless whether they belong to the same or different thematic text units:

462 Langue and Parole in Synchrony and Diachrony (11) Local Chronology Principle For two successive eventuality descriptions SI and S2 the following relations hold, provided that SI expresses the proposition pi and S2 proposition p2, whereby pl() and p2(): rl < r2 and el < e2. In (5) many eventuality descriptions, e.g. [3] relative to [2], [6] relative to [5] and [7] relative to [6], clearly have to be interpreted according to the Local Chronology Principle, regardless of their thematic embedding in the iterative and habitual temporal pattern that has scope over the entire thematic text unit. When we interpret (5) [2] I felt stiff and -weary ... relative to [1] Here ...I began to be old ..., another principle is in force. [1] sets a temporal frame for the localization of the eventualities described in [2], following something like principle (12) below: (12) Local Frame Principle For two successive eventuality descriptions S1 and S2, the following relations hold, provided that SI expresses the proposition pi and S2 proposition p2, whereby pl() and p2(): e2 C el and rl=r2, given that el plausibly represents a frame for the localization of e2. e2Cel means that the second event time is included in the first and rl=r2 tells us that the two reference times are identical, el provides a frame for the localization of r2. Note that (12) only applies to (5) [2] relative to [1]. For example, [6] relative to [5] would follow the Local Chronology Principle (cf. 3), despite the fact that [6] is also connected with [1] according to the Thematic Chronology Principle.

5 APPLICATION OF PRINCIPLES TO (5) [2] With regard to the interpretation of (5) [2] relative to [1] we can say that the Local Frame Principle suspends the more general Local Chronology Principle. At the same time we make assumptions concerning the thematic context: probably [2] will be in compliance with the ERP and the Thematic Frame Principle, with the Thematic Chronology Principle being suspended. I shall now incorporate the relevant principles in an inference pattern along the lines discussed above. The Local Frame Principle is still somehow incomplete. It only tells us that el has to plausibly represent a frame for the localization of e2 in order for this principle to apply. It does not tell us when el might represent such a frame. As opposed to the Local Chronology

Temporal Reasoning 463 Principle, we cannot couch this in general terms. Rather, we have to assume that in (5) the Local Frame Principle applies for the simple reason that [1] refers to a particular eventuality of the type 'I be- old' that is predominantly habitual and stative, albeit somehow inchoative (rather: I bein the early stages of being old; cf. 3 above), and [2] expresses 'I feelstiff and weary in the evenings' which is iterative (i.e. a repeating state). We then have to formulate (12) as a specific principle that applies in (5) only: (13) Local Frame in (5) For two successive eventuality descriptions S1 and S2 the following relations hold, provided that SI expresses the proposition pi and S2 proposition p2, whereby pl() and p2(): e2 C el and rl=r2, whereby e2 C el and rl=r2 iff: here(at the age of 39(Hab(I be- old))) (el) and state(), and Iter(in the evening(I feelstiff & weary)) (e2) and state(). (13) is a truly specific principle. Its relation e2Cel suspends the Local Chronology's el < e2, and rl=r2 suspends rl < r2 because (13) contrary to (11) can only be suggested by an information stage which is specific about the idiosyncratic features of eventualities, about their Aktionsart-features, such as their stative or non-stative character (state vs. event), as well as their iterative or habitual perspectivization (Iter and Hab). The overall pattern of inferences involved in the interpretation of (5) [2] on the basis of (13) will have the following form:

Temporal Reasoning 465 (14) shows the local inferences that are connected with the interpretation of (5)[1] on the right of the dividing line and the thematic inferences that will be drawn on the basis of [1] and [2] on the left. 1 represents the initial information stage when processing (5) [2] relative to (5)[1]. Following our approach outlined in 4.3.1, always entails that pi is true relative to rl and el and that p2 is true relative to r2 and e2. In our example, here(at the age of 39(1 begin- to be old)) is true at rl and el and in the evenings(I feelstiff and weary)) is true at r2 and e2, as indicated in 1. 2 lists the information that we gain on the assumption that the eventuality descriptions which express pi and p2 belong to a thematic text unit that consists of the utterances SI, ..., Sn (i.e. <S1, S2, ..., Sn>). The inference step can be called 'Local Context Embedding' which will most likely precede any other inference step in narrative contexts and that leads us to 2. The most specific inference that can be based on 2 is one where the information given by the adverbs and the particular lexical content of SI and S2 is considered. This leads to 3, from where an inference step following principle (13) must be applied. The latter brings us to the conclusion in 11 that e2 is included in el and that the reference times are identical. From 3 we can also deduce 4 via simplification. 4 considers the most likely position of the utterances that refer to pi and p2 in the thematic text unit only, disregarding specific lexical content and adverbial information. In this case the Local Chronology Principle would seem appropriate with its results shown in 12, but the inference step is suspended by the more specific one based on 3, because e2Cel and rl=r2 contradict rl Configurational Case. GEN-Q is treated as a purely structural case, on a par with nominative and accusative. This hierarchy then explains the distinction between heterogeneous and homogeneous case constructions:5 NOM/ ACC cannot percolate from NP down to N', since N' is already assigned GEN by a quantifier,

In this example heads X and Y are of a type that have Case assigning potential. They simultaneously assign Case to their sister phrasal categories, YP and ZP, respectively. After assigning Case, these heads are free to receive eventually Case by percolation from the dominating phrasal category, if that dominating category was assigned Case. In our example, Case was assigned to YP and ZP and must percolate to their heads, Y and Z, respectively. So, in this example Y is a head which both assigns and receives Case, but still there is no clash of Case features, since Case assignment and Case percolation are not simultaneous, but subsequent processes. Therefore, nothing prevents Y from receiving another Case by percolation, after it discharges its own Case. 4 This proposal is inconsistent with standard theoretical assumptions about case assignment. Case is not assigned by phrasal categories (QP in Babby's system), but rather by heads, and it is not assigned to intermediate projections (N' in Babby's proposal) but rather to maximal XPs. ^Babby's terminology used to distinguish case assignment in direct vs. oblique contexts.

494 Langue and Parole in Synchrony and Diachrony but an oblique case can, since it is assigned earlier than structural cases (including GEN-Q).6 Babby's Case Hierarchy is based on the assumption that inherent case is assigned at D-structure and structural case at S-structure (cf. Chomsky (1981, 1986)). Franks (1994) adopts Babby's ideas as essentially correct, but he tries to show that the proposal about the quantificational genitive as a structural case does not extend to Serbo-Croatian where it must be analysed as inherent.

THE CASE ASSIGNMENT TO QUANTIFIED PHRASES IN BOSNIAN The examples in (2) - (5) translated into Bosnian are identical (in terms of cases realized in quantified phrases) to Russian, except for (4c) and (5c). The Bosnian examples in (7), corresponding to Russian (4c) and (5c) are ungrammatical: (7)

a.

* Ivan upravlja Ivan manages

petima fabrikama. five factory-INST PL

b.

*o petima about five

knjigama book-LOC PL

I have pointed out that the lexical item 'five' in Russian is a quantifier in direct cases assigning genitive to its complement, and an adjective in oblique cases agreeing in case with the head noun. In Bosnian, on the other hand, there is no adjectival 'five', and therefore both (7a) and (7b) are unacceptable. The putative adjectival form petima simply doesn't exist, and the structure in (7a) may be rescued only by inserting the instrumental assigning preposition sa 'with' which serves as a flag to indicate that instrumental is being assigned to the quantified phrase, as in (8a): (8)

a.

Ivan upravlja sa Ivan manages with

b.

o pet about five

pet five

fabrika. factory-GEN PL

knjiga book-GENPL

I will claim, however, that the verb assigns case in (8a). hi (8b), on the other hand, the preposition o 'about' assigns locative and signals that this case has been assigned, although there is no morphologically realized locative ending on pet. From examples similar to these, Franks (1994) concludes that the quantified phrases following prepositions are actually cased, even if no element within quantified phrases shows instrumental or locative case morphologically, as in (8). The implication is that these phrases are NPs. In my approach QPs can also be cased, and therefore there is no reason to postulate NP categorial status of these quantified phrases. ^This approach assumes that in oblique contexts there is still a potential GEN-Q assignor, but it is not able to assign this case to N', because an oblique case assigned earlier blocks GEN-Q. I want to point out that in oblique contexts there is no potential GEN-Q assignor at all, the numeral appearing in its adjectival form, and as such not having case assigning potential.

Categorial Status of Quantified Phrases in Slavic 495 Franks also postulates parametric difference between Russian and Serbo-Croatian claiming that GEN-Q is a structural case in Russian but inherent in Serbo-Croatian, and that this difference constitutes a fundamental source of variation between quantificational structures in two languages. He argues that quantifiers in Serbo-Croatian must inherently assign GEN-Q to their sister N', otherwise it would be overridden by percolation of the inherent case assigned to NP. If this proposal is correct, than the example in (9a) shouldn't be possible, because Franks' solution allows only (9b): (9)

a.

o trima about three-LOC

knjigama book-LOC PL

b.

o tri knjige about three book-GEN PAUC7

^The form knjige looks like genitive singular, but it is not a singular form, but rather paucal. Consider the following examples: (i) Vidio sam onog studenta o kojem si mi govorio. (I) saw that-ACC SG student-ACC SG about whom-SG (to) me (you) talked 'I saw that student that you talked to me about.' (ii) Vidio sam oba ona studenta o kojima si mi govorio. (I) saw both that student about whom-PL (to) me (you) talked 'I saw both of those students that you talked to me about.' (iii) *Vidio sam oba ona studenta o kojem si mi govorio. whom-SG (iv) Vidio sam pet onih studenata o kojima si mi govorio. five those-GEN PL students-GEN PL about whom-PL 'I saw those five students that you talked me about.' Notice three forms of the demonstrative: onog in (i) is clearly singular, onih in (iv) is clearly plural, whereas ona in (ii) differs in form from both and must be something else, namely paucal. Therefore, the form studenta in (ii) should be also treated as paucal, although its form is identical to singular studenta in (i). In this paper, I will treat forms following numerals oba, dva, tri, cetiri as genitive paucal. The form of the demonstrative ona in (ii) resembles in form the genitive singular of indefinite adjectives, e.g. zelena-lNDEF vs. zelenog-DEF 'green'. (Notice also the similarity in form onog vs. zelenog). The form of the demonstrative ona-MASC in (ii) is never used in another context except after numerals oba, dva, tri, cetiri. Consider now the following examples: (v) Vidio sam dva zelena studenta. 'I saw two green-INDEF students.' (vi) *Vidio sam dva zelenog studenta. green-DEF Definite form adjectives have a separate genitive paucal form used exclusively after numerals oba, dva, tri, cetiri: (vii) Vidio sam dva zelena studenta. green-INDEF (viii) Vidio sam dva zelena studenta. green-DEF Notice the difference in the position of stress in zelena. In addition to demonstratives, possessives also have a specialised paucal genitive form not used in other contexts: (ix) Vidio sam dva nasa studenta. our 'I saw two students of ours.' (x) *Vidio sam nasa studenta.

496 Langue and Parole in Synchrony and Diachrony Both of these examples are acceptable, contrary to the predictions of Franks' theory: a quantifier would always be able to assign inherent GEN-Q, not allowing it to be overridden by another inherent case, as in (9a). In my system all case assignment is done simultaneously, that is at the same level of structure. Paucal numerals (dva 'two', tri 'three1, cetiri Tour') have adjectival counterparts, so that both examples in (10) are acceptable: (10)

a.

Ivan Ivan-NOM

upravlja trima fabrikama. manages three-INST factory-INST PL

b.

Ivan upravlja sa Ivan manages with

tri three

fabrike. factory-GEN PAUC

The example (lOa) being grammatical, one would not expect that the option (lOb) is available also, and actually preferred in colloquial speech. This only shows that Bosnian is probably in the process of losing adjectival numerals, which still exist only among paucals. However, both of the following examples are unacceptable: (11)

a.

* Ivan upravlj a pet fabrika. Ivan manages five factory-GEN PL

b.

* Ivan upravlja tri fabrike. Ivan manages three factory-GEN PAUC

The categorial status of pet and tri in (11) is Q, and since Qs lack oblique cases, they cannot be used after verbs which assign oblique cases, like the verb upravljati 'to manage' in (11). The example (1 la) is unacceptable and a preposition must be inserted in order to indicate which case is assigned to the QP, as in (8a). The oblique case assigned by the verb in (lOa) may percolate all the way down to the noun head which has only one source to receive the case from, that is the verb, since a quantifier is absent in this structure, the numeral 'three' appearing in its adjectival form trima and agreeing in case (and other features) with the head noun.

THE CATEGORIAL STATUS OF QUANTIFIED PHRASES IN BOSNIAN Franks claims that only NPs may appear in oblique positions, and since quantified phrases do appear in oblique positions in Serbo-Croatian he concludes that they must be NPs, not QPs. This conclusion is based on the assumption that QPs cannot receive oblique cases. Being NPs, they should potentially be able to bear any oblique case, although this need not be reflected morphologically, since percolation of the oblique case down from NP is blocked by GEN-Q.

(xi)

Vidio sam naseg studenta. Franks (1994: 606) observes that "some adjectives distinguish the paucal form from the indefinite genitive singular: in dva crna ovna 'two black rams', with long rising, is indefinite, and dva crna ovna, with long falling, is definite". He adds: "If the SC paucal numerals are heads, then one can claim that [+PAUC, +Q] heads take paucal complements".

Categorial Status of Quantified Phrases in Slavic 497 However, Franks (1994: 622) points out in a footnote that the dative is strongly disfavoured, giving the following examples: (12)

a.

*Dao gave-MASCSG

je AUX-3SG

b.

?*Koracao je stepped-MASC SG AUX-3SG

knjigu book-ACC prema towards

petljudi. five people-GEN PL petljudi. five people-GEN PL

Franks points out that (12a) is unacceptable because dative case on the indirect object pet ljudi 'five people' is not morphologically realized. He claims that (12b) with the phrase pet ljudi governed by dative assigning preposition prema 'towards' is marginal. However, examples like (12b) are fully acceptable and we will discuss this and similar examples with dative and other oblique cases next. Because not only dative is disfavoured, as pointed out by Franks, but other oblique cases as well, for example instrumental in (13):

(13)

a.

*Kopa pet lopata. (he) digs (with) five shovel-GEN PL

b.

*Kopa dvije (he) digs (with) two-FEM

lopate. shovel-GEN PAUC FEM

The instrumental case assigned by the verb kopati 'to dig' in (13) cannot be realized on the quantifier, since quantifiers have incomplete declension with only direct cases. On the other hand, the oblique case cannot percolate further down to the lower N lopata 'shovel', because it is already assigned genitive by the quantifier pet 'five1/ dvije 'two'. Two cases cannot be assigned to a single element and therefore sentences in (13) are ungrammatical. Actually, two factors jointly contribute to the ungrammaticality of (13): quantifiers lack oblique cases and XPs cannot be assigned two cases. Let us now analyse examples found in other case contexts. First accusative: the example in (14a) shows that verbs like procitati 'to read' assign accusative. This case may be realized on a quantifier and we conclude that the accusative assigned by V to the QP in (14b) actually percolates to the Q head and is realized on pet. (14)

a.

Procitaoje (he) read

knjigu. book-ACC SG

b.

Procitaoje (he) read

pet five-ACC

knjiga. book-GEN PL

The following examples illustrate inherent case assignment by verbs like kopati 'to dig':

(15)

a.

Kopa lopatom. (he) digs (with) shovel-INST SG

498 Langue and Parole in Synchrony and Diachrony b.

Kopa (he) digs

sa with

petlopata. five shovel-GEN PL

c.

Kopa (he) digs

sa with

drugom. friend-INST SG

d.

Kopa dvjema (he) digs (with) two-INST

e.

Kopa (he) digs

sa with

lopatama. shovel-INST PL

dvije lopate. two shovel-GEN PAUC

The noun in (15a) appears in its instrumental form. However, when the same verb takes QP as its complement, the structure is ungrammatical: instrumental cannot be realized on a quantifier, as in (13). There is, however, a rescue strategy - vacuous insertion of a preposition sa 'with'. The examples like (15b) are widely used, although frowned upon by prescriptive grammarians. The reason being that the preposition sa is used to imply company, as in (15c), so that (15b) would imply that he/ she is digging together with shovels, rather than using shovels as the tool to perform the job. Therefore, two structures in (15a) and (15c) differ sharply. In (15a), the verb takes DP as its complement and assigns case to it. In (15c) instrumental case on the noun is assigned by P, not by V. In everyday speech sentences like (15b) are widely used, but not in this 'company' meaning. Therefore, I claim that (15b) does not have the structure corresponding to (15c) in which sa assigns the case, but rather one corresponding to (15a) in which the verb assigns the case. Consequently, in (15b) the case is assigned by a verb, and the preposition has the function to license the structure, or possibly, carry the case assigned by V in the absence of any morphological ending on a quantifier which would be able to indicate that the oblique case was assigned. Therefore, in my proposal, the preposition sa in constructions of this type is deprived of its case-assigning potential, and does not occupy P° position. Instead, I assume that it occupies F° position, as in the following partial structure of VP in (15b):

Notice, however, that paucal numerals, in contrast to higher numerals like pet, do have adjectival counterparts, so that (15d) is grammatical. The oblique case assigned by the verb in

Categorial Status of Quantified Phrases in Slavic 499 (15d) may percolate all the way down to the noun head which has only one source to receive the case from, that is the verb, since a quantifier is absent in this structure, the numeral 'two' appearing in its adjectival form and agreeing in case (and other features) with the head noun. Interestingly, (15e) is also acceptable with the quantifier 'two' preceded by sa, comparable to (15b). This is probably just an extension of a rule that is available for quantifiers that do not have adjectival counterparts, like 'five'. The examples in (17) illustrate dative assignment by the verb: (17)

a.

Prisaoje (he) approached

zeni. woman-DAT SG

b.

Prisaoje (he) approached

dvjema two-DAT

c.

* Prisaoje (he) approached

d.

*Prisaoje (he) approached

zenama. woman-DAT PL

pet zena. five woman-GEN PL dvijezene. two woman-GEN PAUC

The ungrammaticality of (17c) and (17d) may be explained in the same way as ungrammaticality of (13). Notice that no preposition insertion rescue strategy is available in these cases. Therefore, the only option is one in (17b) with paucal numerals, since they have adjectival counterparts. However, this option is not available to higher numerals which lack adjectival counterparts. Finally, let us consider examples with verbs that inherently assign genitive. These are mainly reflexive verbs like bojati se 'to fear, to be afraid of, as in (18): (18)

a.

Boji se (he) is afraid (of)

zena. woman-GEN PL

b.

Boji se (he) is afraid (of)

dviju two-GEN

c.

?Boji se (he) is afraid (of)

dvije zene. two woman-GEN PAUC

d.

?Boji se (he) is afraid (of)

pet five

zena. woman-GEN PL

zena. woman-GEN PL

For many speakers examples in (18c) and (18d) are not even questionable, but completely acceptable. Why should QPs, of all oblique contexts, be acceptable only in genitive? What is special about genitive? Does it mean that numerals which are Qs actually have only one syncretic form, that is NOM/ACC/GEN? This is certainly a possibility, but I am rather inclined to offer another explanation. The fact that genitive case is morphologically realised on a noun inside a QP after the verbs which assign genitive is enough to license the

500 Langue and Parole in Synchrony and Diachrony structure, although the genitive is realized on the "wrong" place, namely on a noun rather than on a quantifier. The contrast between (18b) and (18c) clearly shows that two different categories of numerals are involved: in (18c) dvije is a quantifier assigning genitive case to its paucal complement, whereas in (18b) dviju is an adjective agreeing in case with the plural head noun which is assigned genitive by the verb. In (18c) the verb assigns genitive to the QP, which cannot be realized on the Q head, since it lacks oblique cases, but the very presence of the genitive case in the structure, namely on the noun, although it is not assigned to it by the verb but by the quantifier, licenses the structure. Next we should consider what is the behaviour of quantified phrases as complements of oblique case assigning prepositions. We will consider here only assignment of the locative case which is exclusively assigned by prepositions, as in (19), but we find the same behaviour in other oblique cases assigned by prepositions: (19)

a.

Govoriojeo (he) talked about

zeni. woman-LOC

b.


petzena. five woman-GEN PL

c.


dvije zene. two woman-GEN PAUC

d.


dvjema two-LOC

zenama. woman-LOC PL

All examples in (19) are grammatical which could mean that oblique assigning prepositions, in contrast to oblique assigning verbs, are able to assign all oblique cases to Q heads, and that would mean that Qs have one syncretic form for all cases. Since this putative syncretic form cannot be used when the oblique case is assigned by the verb, this proposal cannot be maintained. The explanation for the different behaviour of quantified phrases as complements of verbs and prepositions may be the following: Oblique case-assigning verbs, that is verbs which inherently assign dative and instrumental, cannot assign their case to the quantifiers, which lack morphological endings to indicate which case was assigned. On the other hand, prepositions not only assign case, but also serve as a replacement for the morphological ending absent on the quantifier. Notice that in case of paucal numerals, an alternative is allowed, as for example in (19c,d). Paucal numerals in Bosnian are parallel to their Russian counterparts in having adjectival and quantifier forms. In contrast to Russian, which allows only adjectival forms of paucals after prepositions which assign oblique cases, both forms, adjectival and quantifier, may be used in Bosnian, as in (19c) and (19d). The explanation for this may be in different properties of prepositions assigning oblique cases in the two languages. The prepositions in Bosnian not only assign case, but also may serve as a flag to indicate which case has been assigned. Therefore, when lexical items which lack oblique cases appear in oblique case contexts, they are licensed by the preposition itself which assumes the role of the

Categorial Status of Quantified Phrases in Slavic 501 morphological case ending. The Russian prepositions obviously lack this property, and therefore numerals which are of the category Q cannot appear in oblique case contexts after prepositions. Ambiguity may occur in Bosnian examples with prepositions which may assign more than one case, since there is no case ending on the quantifier to morphologically indicate which case was assigned. For example, u 'in' is a preposition which may assign both accusative and locative, as in (20): (20)

a.

Guraju namjestaj u sobama. push-3PL furniture in room-LOC PL 'They are pushing furniture in the rooms.'

b.

Guraju namjestaj usobe. push-3PL furniture in room-ACC PL 'They are pushing furniture into the rooms.'

c.

Guraju push-3PL

namjestaj furniture

upet soba. in five room-GEN PL

The sentence in (20c) is acceptable, but it is ambiguous between locative and accusative meaning, since there is no morphological ending on the Q head to indicate whether accusative or locative was assigned. Franks (1994: 622) discusses subject-verb agreement patterns in Serbo-Croatian claiming that in principle both singular and plural agreement is possible, illustrating this with the examples in (21): (21)

a.

Dva muskarca su dosla/ ?je doslo/ ?su dosli. two men AUX-3PL came-N PL/ AUX-3SG came-N SG/ AUX-3PL came-PL

b.

Pet muskaraca *su dosla/ je doslo/ ?su dosli. five men-GEN AUX-3PL came-N PL/ AUX-3SG came-N SG/ AUX-3PL came

He points out that the neuter singular form of the verb (doslo in (21b)) is considered standard with higher numerals, and neuter plural (dosla in (2la)) or feminine plural (dosle) with paucals. He considers the singular with the nonpaucal numerals to reflect true syntactic agreement and the plural to instantiate semantic agreement. Since Franks postulates difference in syntactic agreement forms in Russian and Serbo-Croatian, he explains it in the following way. The fact that N' is GEN-Q in Serbo-Croatian inhibits percolation of pronominal features from N head up to NP. Consequently, the pronominal features of the NP are set as neuter singular in the absence of any further specification. In Russian, on the other hand, GEN-Q is not assigned until S-structure, so it does not block percolation, which is induced at D-structure. Franks concludes that the plural option in Serbo-Croatian is thus a marked variant, in which the verb appears to exhibit semantic agreement. This might be understood as agreement with the head N, rather than with the NP itself, so that the verb shows number (and gender) features of the subject noun.

502 Langue and Parole in Synchrony and Diachrony Since the third person neuter singular verb form is the norm in Serbo-Croatian (with the nonpaucal numerals), it implies that the quantified NP phrase is neuter singular, concludes Franks, and raises an interesting question: why quantified NPs are neuter singular in SerboCroatian, but plural in Russian. My answer to this question is that quantified phrases are neuter singular in both languages, but in Russian semantic agreement wins over syntactic agreement. In both languages syntactic and semantic agreement are available options. But whereas in Bosnian syntactic agreement is stronger, and therefore preferred, it is the other way round in Russian. I assume that the neuter singular form of the verb represents syntactic agreement in both languages, in contrast to Franks who claims that the plural form represents syntactic agreement in Russian. In my approach, quantified subject phrases in both languages are QPs, with a Q head which bears third neuter singular features (with the nonpaucal numerals), either inherently or filled in as such by default. I also do not agree with Franks' analysis of (2la). He regards Serbo-Croatian paucal numerals as pure modifiers of a nominative head noun in a special paucal, rather than the expected plural form. The verb takes plural syntactic agreement form, since there is no paucal auxiliary, so that the plural form su must be used instead. He goes on and says that nominative paucal forms such as muskarca in (2la) are easily taken as genitive singular (having identical form), leading to the je doslo option found with the nonpaucal numerals!? There is no explanation why would the form je doslo be used with nonpaucals, instead with paucals.8 He also points out that masculine nouns typically induce a special agreement option that resembles the neuter plural ending (dosla in (2la)). For me, however, this form of the predicate doesn't resemble, but actually is the neuter plural form. To recapitulate Franks' analysis of (2la): dva is an adjective modifying the head paucal noun muskarca and agreeing in case (nominative) with it. Consequently, the predicate should agree with the head of the subject phrase, namely muskarca. Since there is no paucal form of the auxiliary, the plural agreement form must be used in the predicate. My analysis differs radically. But before elaborating, we should consider additional examples, involving feminine and neuter nouns: (22)

a.

Zene women-NOM PL

su AUX-3PL

dosle. came-FEM PL

b.

Dvije two-FEM

zene su women AUX-3PL

dosle. came-FEM PL

c.

Petzena je five women-GEN PL AUX-3SG

doslo. came-NEUT SG

d.

Goveda cattle-NOM PL

su AUX-3PL

dosla. came-NEUT PL

e.

Dvagoveda two cattle-GEN SG

su AUX-3PL

dosla. came-NEUT PL

8j should point out here that the optionee doslo is strongly deviant with paucals, and no native speaker could possibly accept it. It is surprising that this (non-existent) option (with paucals) is marked with ? , rather than with *, in Franks' example given in the footnote 27, page 622, and reproduced here as (2la).

Categorial Status of Quantified Phrases in Slavic 503 f.

Petgoveda je five cattle-GEN PL AUX-3SG

doslo. came-NEUT SG

Comparing these examples, we may conclude that pet is a quantifier which selects a plural noun of any gender and assigns genitive to it. The neuter singular form of the predicate (je doslo in (22c,f)) may be explained either as a default agreement form (neuter and singular being the least marked) if we assume that pet actually lacks inherent gender and number features. Or we may assume that quantifiers also have inherent phi-features, pet being neuter singular, so that the predicate manifests syntactic agreement with it. The paucal numerals in (22b,e) are also quantifiers, not adjectives as in Franks' proposal. Dva is a quantifier which selects a paucal noun of masculine or neuter gender and assigns genitive to it (which happens to be identical in form to genitive singular). We may say that ova also has inherent phi-features, being neuter plural, and the predicate manifests syntactic agreement in these features with it (su dosla in (22e). Finally, I assume that dvije is a quantifier which selects a paucal noun of feminine gender and assigns genitive to it (which happens to be identical in form to nominative plural). This quantifier is of feminine gender and plural number and the predicate syntactically agrees with it in these features (su dosle in (22b)).

THE CATEGORIAL STATUS OF QUANTIFIED PHRASES IN RUSSIAN Pesetsky (1982) argues that a quantified expression such as pjat' rublej 'five rubles' in Russian is ambiguous and may be either a QP or an NP. This claim is supported by the fact that there are two possible subject-verb agreement patterns with quantified subjects in Russian, illustrated by the following examples: (23)

a.

Pjat' five

krasivyx beautiful-GEN PL

b.

Prislo arrived-NEUT SG

devusek prisli. girls-GEN PL arrived-PL

pjat' krasivyx devusek. five beautiful-GEN PL girls-GEN PL

These two agreement possibilities make Pesetsky (1982) postulate two categories for quantified phrases: they are subject NPs when plural agreement of the verb obtains, as in (23a), and QPs when the default neuter singular form appears, as in (23b). In the second case the quantified phrase is internal to the verb phrase. When it is an NP, it must move to subject position to receive case, but when it is a QP, which does not require case, it remains in situ within VP. He claims that, with a few exceptions, QPs cannot be true subjects. Franks shows that QPs can appear in subject position in Russian, giving examples with neuter singular forms of the verb after quantified phrases: (24)

Neskol'ko studentov several students-GEN PL

procitali/ procitalo etu knjigu. read-PL/ NEUT SG this book-ACC

At the same time Franks assumes that both NPs and QPs can appear in structural case positions. However, he claims that QPs are not cased in these positions, in contrast to NPs.

504 Langue and Parole in Synchrony and Diachrony Structural case need not be assigned, in contrast to oblique cases, and therefore only NPs can appear in oblique positions: (25)

Jadumaju o pjati I am thinking about five-LOC

knigax. books-LOC

In my approach, quantified subject phrases in Russian are QPs, in the same way as in Bosnian. In contrast to Franks, who claims that the plural form of the verb, as in (23a), represents syntactic agreement, I assume that neuter singular form represents syntactic agreement. Both syntactic and semantic agreement are available options, but semantic agreement appears to be stronger and wins over syntactic agreement. The neuter singular form of the verb in (23b) represents syntactic agreement with the empty expletive in subject position, which is neuter singular. The plural form of the verb procitali in (24) represents semantic agreement, whereas the singular form procitalo, which is acceptable but rarely used, represents syntactic agreement. This example is yet another proof that semantic agreement in Russian is stronger. As far as the example (25) is concerned, I argue that the complement of the preposition o 'about' is a QP, but with an empty Q head. Therefore, the locative case assigned by the preposition percolates all the way down to the noun and then spreads by agreement to the adjectival numeral.

CASE PROPERTIES OF QUANTIFIERS Franks distinguishes two genitives, quantificational and non-quantificational, which share all case features except [oblique]. According to him, genitive of quantification (GEN-Q) is [-oblique] and assigned (in Russian) at S-structure, whereas the regular genitive is [+oblique] and assigned at D-structure. In order to illustrate this difference he gives the following examples: (26)

(27)

a.

opisanie description

trex three-GEN

gorodov cities-GEN PL

b.

Jaizbegaju trex I-NOM avoid three-GEN

ljudej. people-GEN PL

a.

*opisanie description

trex three-GEN

goroda city-GEN SG

b.

*Ja I-NOM

izbegaju avoid

trex three-GEN

celoveka. person-GEN SG

Franks says that the regular genitive assigned to the quantified NP overrides the quantificational genitive, and therefore (27) with quantificational genitive on goroda and celoveka is ungrammatical. My explanation for this is different. The adnominal genitive assigned by nouns as in (26a), and inherent genitive assigned by verbs, as in (26b), cannot be realised on quantifiers in Russian, since they lack oblique cases, and therefore genitive

Categorial Status of Quantified Phrases in Slavic 505 percolates all the way down to the noun, and then spreads by agreement to the adjectival numeral (trex). Consequently, quantificational genitive cannot be assigned at all, because there is no Q head in the structure to assign it. The numeral appears in its adjectival form and agrees in case (genitive) with the head noun gorodov and ljudej, respectively. Also, ungrammatical sentences in (27) are not relevant for Franks' point, because quantifiers never occur in oblique forms. If Russian allows genitive-assigning nouns and verbs to assign this case to quantifiers, then the structures would contain the indeclinable quantifier tri, as in (28), and not the adjectival declined form trex, as in (27): (28)

a.

*opisanie tri goroda

b.

* Ja izbegaju tri celoveka.

Bosnian allows quantifiers to appear in structures comparable to (28), so that examples in (29) are grammatical: (29)

a.

opis description

tri grada three city-GEN PAUC

b.

Ja se I-NOMrefl

bojim tri muskarca. fear three man-GENPAUC

Although the genitive assigned by the noun opis 'description' and the verb bojati se 'to fear' cannot be realized on the quantifier, the presence of the morphologically realised genitive on the nouns, grada 'city' and muskarca 'man', is enough to license structures in (29). hi addition to grammatical examples in (29), there is also a possibility of having quantified phrases with adjectival numerals, if numerals have adjectival forms (as in the case of paucals), so that examples in (30), corresponding to those in (29), are also grammatical: (30)

a.

opis description

triju three-GEN

gradova city-GEN PL

b.

Ja se bojim I refl fear

triju three-GEN

muskaraca. man-GEN PL

Franks also considers other cases in Russian, and contrasts the adjunct instrumental, as in (3 la), with true complement instrumentals, as in (31b), and their interaction with GEN-Q, as in (32): (31)

a.

Ivan el ikru lozkoj. Ivan-NOM ate caviar-ACC spoon-INST 'Ivan was eating caviar with a spoon.'

b.

Ivan Ivan

upravljaet manages

fabrikoj. factory-INST


a.

Ivan el ikru dvumja Ivan ate caviar-ACC two-INST

b.

Ivan upravljaet dvumja Ivan manages two-INST

lozkami/ spoon-INST PL/

fabrikami/ factory-INST PL/

*lozki. GEN SG

*fabriki. GEN SG

Examples in (32) illustrate Franks' point that neither type of instrumental can be overridden by GEN-Q. His conclusion is that an oblique case assigned to a complement because it is lexically required by a particular verb, as in (26b) and (32b), behaves the same with respect to the genitive of quantification as the oblique case that is assigned to an adjunct on purely configurational grounds, as in (26a) and (32a): that is, both types of case override genitive of quantification. Again, my claim is that in Russian oblique case assigners simply cannot assign their case to quantifiers, which lack oblique cases, and therefore GEN-Q is not overridden, because there is simply no GEN-Q assigners in oblique contexts, a numeral always behaving as an adjective in such a context, and not as a quantifier (which would be a potential GEN-Q assigner). In Bosnian, instrumental assigning verbs cannot assign this case to quantifiers, and therefore, only structures with adjectival numerals (of paucals) are allowed (33a,c), as in Russian. However, there is also an option of inserting the preposition sa 'with' after the verb (33b,d), which licenses a quantifier, so that in such a case genitive may be assigned by a quantifier to its complement, as in (33), corresponding to Russian examples in (32): (33)

a.

Ivan Ivan

je jeo ikru dvjema AUX eat caviar-ACC two-INST

b.

Ivanje Ivan AUX

c.

Ivanupravlja dvjema Ivan manages two-INST

d.

Ivanupravlja sa Ivan manages with

jeo ikru eat caviar-ACC

sa with

kasikama. spoon-INST PL dvije kasike. two spoon-GEN PAUC

fabrikama. factory-INST PL

dvije fabrike. two factory-GEN PAUC

Strictly speaking, (33b) and (33d) would be treated as ungrammatical in standard literary (prescriptive) language, since the preposition sa denotes company: e.g. Ivan je jeo sa drugom. 'Ivan was eating (together) with a friend.' vs. *Ivan je jeo drugom. However, in colloquial speech both (33b) and (33d) would be acceptable.

CONCLUSION The analysis of quantified expressions in Russian and Bosnian proposed in this paper supports the view that three groups of these expressions may be distinguished:

Categorial Status of Quantified Phrases in Slavic 507 i. ii. iii.

those that behave as quantifiers (and assign case) in direct case contexts; those that are adjectival in all contexts, or only in oblique contexts; those that are nominal in all contexts. It was argued also that a quantified expression is always a QP, but its head may be empty in cases when a quantificational lexical item behaves as a noun or an adjective. When the Q head is filled, its case is not morphologically realized, Q being a category with a single form. However, it is claimed that Q is not caseless, but rather has a syncretic nominative/ accusative form. Therefore, quantifiers, having a deficient declension, cannot appear in oblique case contexts, unless a preposition is inserted to indicate and carry the respective oblique case, as in Bosnian examples. Some numerals have adjectival counterparts, but only in oblique cases. Having a deficient adjectival declension without direct cases, these numerals cannot appear in their adjectival form in direct cases. Therefore, in a direct case context a quantifier form must be used. Consequently, there is no agreement of this form with the following noun. Instead, a noun appears in its genitive form, assigned by a quantifier. This analysis supports the view that there is no reason to postulate any parametric differences between two languages with respect to quantified structures. The only contrast is in different categorial organisation of quantified expressions. Bosnian is probably in the process of losing adjectival numerals completely, which still exist only among paucals. On the other hand, Russian preserved adjectival declension of most of its numerals.

REFERENCES Abney, Steven (1987). The English Noun Phrase in its Sentential Aspect. Ph.D. diss., MIT. Babby, Leonard (1986). The Locus of Case Assignment and the Direction of Percolation. In: Case in Slavic (R. Brecht and J. Levine, eds.), pp. 170-219. Slavica Publishers, Columbus. Babby, Leonard (1987). Case, Pre-Quantifiers, and Discontinuous Agreement in Russian. Natural Language and Linguistic Theory, 5, 91-138. Borras, F. M. and R. F. Christian (1971). Russian Syntax.. Aspects of Modern Russian Syntax and Vocabulary, 2nd edition. Clarendon Press, Oxford. Browne, Wayles (1993). Serbo-Croat. In: The Slavonic Languages (B. Comrie and G. Corbett, eds.), pp. 306-387. Routledge, London. Chomsky, Noam (1981). Lectures on Government and Binding. Foris, Dordrecht. Chomsky, Noam (1986). Knowledge of Language: Its Nature, Origin, and Use. Praeger, New York. Corbett, Greville (1983). Hierarchies, Targets, and Controllers: Agreement Patterns in Slavic. Pennsylvania University Press, University Park. Franks, Steven (1986). Case and the Structure of NP. In: Case in Slavic (R. Brecht and J. Levine, eds.), pp. 220-243. Slavica Publishers, Columbus. Franks, Steven (1994). Parametric Properties of Numeral Phrases in Slavic. Natural Language and Linguistic Theory, 12, 597-674. Franks, Steven (1995). Parameters of Slavic Morphosyntax. Oxford University Press, Oxford

508 Langue and Parole in Synchrony and Diachrony Freidin, Robert and Rex Sprouse (1991). Lexical Case Phenomena. In: Principles and Parameters in Comparative Grammar (R. Freidin, ed.), pp. 392-416. MIT Press, Cambridge. Giusti, Giuliana (1991). The Categorial Status of Quantified Nominals. Linguistische Berichte 136, 438-454. Giusti, Giuliana and Nedzad Leko (1995). On the syntax of quantity expressions in Bosnian. University of Venice Working Papers in Linguistics, 5, no.2, 23-47. Giusti, Giuliana and Nedzad Leko (1996). Definite and indefinite quantity expressions in Bosnian. In: Determinatezza e indeterminatezza nelle lingue Slave. Atti del Convegno svoltosi a Firenze (Ottobre 1995) (R. Benacchio, F. Fici and L. Gebert, eds.), pp. 147162. Unipress, Padova. Leko, Nedzad (1995). Syntactic behaviour of numerals 1, 2, 3,4, 5 in Bosnian and Slovene. Quaderni del Dipartimento di Linguistica, Universita di Firenze 6, 213-228. Neidle, Carol (1988). The Role of Case in Russian Syntax. Kluwer, Dordrecht. Offord, Derek (1995). Modern Russian. An Advanced Grammar Course. Bristol Classical Press, Bristol. Pesetsky, David (1982). Paths and Categories. Ph.D. dissertation, MIT. Shlonsky, Ur (1991). Quantifiers as Functional Heads: A study of Quantifier Float in Hebrew. Lingua, 84, 159-180. Speas, Margaret (1990). Phrase Structure in Natural Language. Kluwer, Dordrecht. Timberlake, Alan (1993). Russian. In: The Slavonic Languages (B. Comrie and G. Corbett, eds.), pp. 827-887. Routledge, London. Wade, Terence (1992). A Comprehensive Russian Grammar. Blackwell, Oxford.

ADDRESSES OF AUTHORS Prof. Sven-Gunnar Andersson, University of Gothenburg, Intitutionen for tyska och nederlandska, S-41298 Goteborg, Sweden e-mail: [email protected]

Dr Isabella Buniyatova, Kiev State Linguistic University, 73 Chervonoarmiyska St., GSP 5, 252650 Kiev, Ukraine e-mail:[email protected]

Dr Christopher Beedham, Department of German, School of Modern Languages, University of St Andrews, Buchanan Building, St Andrews, FifeKY169PH, Scotland/UK e-mail: [email protected]

Kathleen Connors, Universite de Montreal, Departement de linguistique et de traduction, CP6128 succursale Centre - ville, Montreal, Quebec, Canada H3C 3J7 e-mail: [email protected] Prof. Jozef Darski, Adam Mickiewicz University, Institute of German Philology, Al. Niepodleglosci 4, PL-61-874Poznan, Poland e-mail: [email protected]

Dr Bozena Bednafikova, Palacky University, Philosophical Faculty, Dept. of Czech Studies, Krizkovskeho 10, 771 80 Olomouc, Czech Republic e-mail: [email protected] Alexander Bergs, M.A. Heinrich-Heine Universitat Diisseldorf UniversitatsstraBe 1 AnglistikIII-Geb23.21 40225 Dusseldorf Germany e-mail: [email protected] Dr Igor Boguslavsky, Head, Computational Linguistics Laboratory, Institute for Information Transmission Problems, Russian Academy of Sciences, Bol'shoj Karetnyj per. 19, GSP-4, 101447 Moscow, Russia, e-mail: [email protected]

Dr Gabriele Diewald, Institut fur Germanistik, Universitat Erlangen, Bismarckstr. 1, D-91054 Erlangen, Germany e-mail: [email protected] Prof. Dmitrij O. DobrovoFskij, Russian Academy of Sciences, Institute of Russian language, Volkhonka 18/2, 121019 Moscow, Russia e-mail: [email protected] or [email protected]

509

510 Langue and Parole in Synchrony and Diachrony Prof. Elisabetta Fava, Dipartimento di Linguistica, Universita di Padova, Palazzo Maldura, via beato Pellegrino 1, 1-3 5100 Padova, Italy e-mail: [email protected] Dr Isabel Forbes, 237 Perth Road, Dundee DD2 1EL, Scotland/UK e-mail: [email protected] Prof. Thomas V. Gamkrelidze , Tsereteli Institute of Oriental Studies, Georgian Academy of Sciences, Acad. Tsereteli St. 3, 380062 Tbilisi, Georgia Dr Christopher Gledhill, Department of French, School of Modern Languages, University of St Andrews, Buchanan Building, St Andrews, Fife KY16 9PH, Scotland/UK e-mail: [email protected] Natalia Guermanova 81-4-649 Svobody Street 123481 Moscow, Russia Dr Ralph Hartmann, BahnhofstraUe 3, D-88299 Leutkirch, Germany e-mail: [email protected]

Prof. Dr. Roland Harweg, Ruhr-Universitat Bochum, Fakultat fur Philologie, Germanistisches Institut, UniversitatsstraBe 150, Gebaude GB, D-44780 Bochum, Germany e-mail: [email protected] Prof DrMati Hint, Head of the Department of Estonian Language, Tallinn Pedagogical University, Narva mnt 25, Tallinn EE-10120 Estonia e-mail: [email protected] Magdalena Jurewicz, Katedra Glottodydaktyki i Translatoryki UAM, Poznan ul 28 Czerwca 1956 nr 198, 6 1-4 85 Poznan, Poland Dr Wladimir Klimonow Zentrum fur Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, JagerstraBe 10/11 10117 Berlin Germany e-mail: [email protected] Prof. Dr. Dr. h.c. E.F.K. Koerner, Department of Linguistics, University of Ottawa, Ottawa, Ont, Canada KIN 6N5 e-mail: [email protected] Prof. Dr. Mikhail L. Kotin, Nagatinskaja nab. 18-152, RU 115533 Moscow Russia

Addresses of Authors 511 Prof. Dr. Fernanda Krier, section d'allemand, Universite Rennes II Haute Bretagne, 6, avenue Gaston Berger, F - 35043 Rennes, France Dr Nedzad Leko, Universitetet i Oslo, Institutt for lingvistiske fag, Tekstlaboratoriet, P.A. Munchs hus, Boks 1102Blindern, 0317 Oslo, Norway e-mail: [email protected] Akiko Matsumori, Japan Women's University, School of Humanities, 2-8-1 Mejiro-dai, Bunkyo-ku, Tokyo, 1128681 Japan e-mail: [email protected] Dr Andreas Musolff, University of Durham, Department of German, School of Modem Languages, Elvet Riverside, New Elvet, Durham DH1 3JT, England e-mail: [email protected] Dr Larissa Naiditch, Dept. of Linguistics, Faculty of Humanities, The Hebrew University of Jerusalem, 91905 Mount Scopus, Jerusalem, Israel e-mail: [email protected]

Prof. Olga A. Ossipova, Honorary Professor of Tomsk State Pedagogical University, Dept of English Philology, Komsomol'ski av. 75, 634041 Tomsk, Russia e-mail: [email protected] Dr Florian Panitz, Katharinenstrafie 6a, D-26121 Oldenburg, Germany e-mail: panitz@hpux3 3 .hrz.uni-oldenburg.de Prof Dr Youri Poupynin, Dept of Grammar Theory, Russian Academy of Sciences, Tuchkov per 9, 199053 St Petersburg, Russia e-mail: [email protected] Prof. R.H. Robins, 65 Dome Hill, Caterham, Surrey CR3 6EF, England Dr Ivanka P. Schick, Dept. of Slavic Languages, University of Potsdam, Am Neuen Palais 10, D-14469 Potsdam, Germany e-mail: [email protected] Senta Setinc, Institute of Information Science - IZUM, 2000 Maribor, Presernova 17, Slovenia e-mail: [email protected]

512 Langue and Parole in Synchrony and Diachrony Prof. Pieter A.M. Seuren, Faculty of Letters, Nijmegen University, P.O. Box 9103, 6500 HD Nijmegen, Holland e-mail: [email protected] Prof. Dr Radmila B. Sevic English Dept, University of Novi Sad Stevana Musica 24 YU 21000 Novi Sad Yugoslavia e-mail: [email protected]

Prof. Tadao Shimomiya, Gakushuin University, Mejiro 1-5-1, Tokyo, Japan -171 - 0031 Prof. Yishai Tobin, Ben-Gurion University of the Negev, Department of Foreign Literatures and Linguistics, PO Box 653, 84 105 Be'er Sheva Israel e-mail: [email protected]

NAME INDEX Alexander the Great, 50, 51 Apresjan, 105, 109, 111,117 Aristotle, 47-61 Austin, J.L., 265-8, 271, 277, 278 Ayer, A.J., 66, 67, 70, 71 Bally, C., 24, 25, 31,33 Berkeley, G., 67 Betz,W., 321,324, 329, 332 Bloch, B., 67, 70 Bloomfield, L., 21, 22, 31, 64, 66-70 Bogatyrev, P., 99-101 Bolinger,D., 169, 179 Bondarko, 106, 117 Brogyanyi, B.,22, 23, 31 Brugmann, K., 26, 65, 71 Biihler,K., 105, 117 Buyssens,E., 19,20,31,33 Carrol, L., 175, 180 Chomsky, N., 64, 66-71 Clark, H., 105, 117 Collinge, N.E., 19, 31 Comrie,B., 105, 106,117 Comte, A., 25 Contini-Morava, E., 170, 179 Courtenay, B. de, 25 Curtius, G., 26 Davidson, D., 270, 279 Davies, A.M., 65, 70 Davies, N., 64, 70 Decsy, G., 100, 101 Demosthenes, 48, 50, 51 Descartes, R., 67 Diver, W., 170, 179 Duckworth, D., 321, 324, 329, 332 Emeneau, M., 67 Engler,R., 19-21,32,33 Eubulides, 47-61 Fillmore, C.G., 110,117 Firth, J.R., 31,68 Fishman, J.A.,24, 31 Frege, G., 47, 56, 57, 59, 61 Garcia, E., 170, 179 Gleason, H.A., 68, 71

Godel,R., 19,20,30,31,33 Grammont, M., 21, 22,26, 31 Halle, M., 66, 70 Harris, R., 19,21,30, 31,33 Harris, Z.S., 67-71 Haugen, E., 22, 31 Hewson,J.,23,29,31 Hill, A., 68, 71 Hjelmslev, L., 22,31 Hockett, C.F., 63-8, 71 Holdcroft,D.,21,31 Huffman, A., 170,180 Jakobson,R.,65,71,99-101 Jones, W., 64, 65, 71 Juhasz, J., 324, 328, 332 Katz, J., 66, 68, 71 Keenan,E., 107, 117 Kiparsky, V., 321, 324, 329, 332 Kirsner,R., 170, 180 Kozinskij, I.Sh., 107, 118 Leiss, E., 105, 118 Lepschy, G.C., 23, 32 Leskien, A., 65, 71 Levinson, S., 105,110,118 Littre, E., 23, 25, 32 Lyons, J., 273-5, 280 Macaulay, T., 64, 71 Malmberg, B., 64, 71 Martinet, A., 24 Maslov,J.S., 109, 113, 118 Matthews, P.H., 67, 70, 71 Meillet, A., 21-3, 25-30, 32, 34 Mey, 24, 25, 33 Miklosich, F., 321, 325, 327, 333 Milewski, T., 99, 101 Milne, A.A., 177,180 Moulton,W.G.,21,22,33 Mounin, G., 20, 23, 29, 32, 33 Mulder, J.W.F., 24, 33 Newmeyer, F., 69, 71 Osthoff, H., 26, 29, 65, 71 Paducheva, E.V., 105, 118 Paul, H., 321, 322, 324, 333 513

514 Langue and Parole in Synchrony and Diachrony Peelers, B., 23, 33 Pike, K.L., 67, 71 Plato, 49-54 Polenz, P. von, 321,329 Popper, K., 70, 71 Postal, P., 66, 68, 71 Quinton, A., 67, 71 Redard, G., 19, 20, 29, 33 Reid,W., 170, 180 Rudy, S., 169, 180 Saint-Exupery, A. de, 178, 180 Saussure, F. de, 19-33, 35-45, 64-71, 169, 180 Schlegel, August, 65 Schlegel, Franz, 65 Schuchardt, H., 321, 327, 332, 333 Searle, J.R., 35, 36, 40, 42-5 Sebeok, T., 22, 32, 33 Sechehaye, A., 19, 20, 25, 33

Seuren, P.A., 69, 71 Smith, H.L., 67, 71 Socrates, 49-51,54, 61 Sorites, 47, 54, 57, 58 Streitberg, W., 30, 33 Striedter-Temps, H., 321, 325, 326, 333 Sweet, H., 65, 71 Szemerenyi, O., 22, 23, 33 Toman, J., 23, 26, 34 Trager, G., 67, 71 Trubetzkoy, N.S., 22, 24, 25, 34, 65, 70-2 Vachek, J., 66, 70, 72 Vendryes, J., 22, 26, 34 Verner, K., 64, 65, 68, 72 Wartburg, W. von, 23, 34 Waugh,L., 169, 180 Wittgenstein, L., 35, 36, 39, 41, 42, 44, 45

SUBJECT INDEX Ableitungen, 329 accent, 431-9, 445 activa tantum, 107 active case, 411, 415, 416 active, 105, 107 adjectival form, 75, 80 adjectival numerals, 492 adverbial, 364, 368-73 agentive by-phrase, 11,12 Akronyme, 321,331 aktionale Opposition, 403, 404, 406, 407 Aktionsart, 105, 106,108,111 alternation, 77-80 analogy, 81,83, 86 analytic, 81, 84-6 anaphoric, 120-2, 127 Ancient Greek, 265, 270 animate, 411-16 annals, 19,20,64 archaisms 425,427,429 Archetyp, syntaktischer, 393 argument, 283, 285, 286, 289-93 artefact, 11,12 aspect, 12-13, 105-14, 117,451,454,458,467 aspektuale Opposition, 401, 402, 404-8 aspirated consonants in Proto-Indo-European, 419-21 atelisch, 402 autobiography, 423, 424 Auxiliarisierung, 393, 396, 398-400 Auxiliary + Participle, 12 Balkan-Slavic languages, 473, 474, 476, 478, 482 Baltic, 367 Bartholomae's Law, 419 biuniqueness, 67 Bivalence, Principle of, 48, 52, 54 Bosnian, 489,494,496, 500-8 Bulgarian, 269, 469-88 case assignment, 490, 493, 496,497 cassatio, 55 cataphoric, 122, 127 categorial status, 489-491, 494, 496

classifiers, 133-47 clitic doubling, 469-71, 474-7,485 cognitive theory of metaphor, 203, 208, 211, 214 cognitivist theory, 283, 287, 288 collocation, 221-36 colloquial, 82 colour categorization, 183 colour semantics, 181,183, 187 Columbia School, 170 comparative, 253, 254, 262 complements, 363, 368, 369, 372, 489-500, 504-6 completeness, 81 completive, 110, 111, 113 complex complementation, 363 complex sentence, 363 compositional aspect, 12 conceptual metaphor, 203, 204, 208, 214, 283, 287-93 conditional, 275, 276 connecteme, 76 consonantal stem-building markers, 411,412, 415,416 contact, languages in, 192 contextual meaning, 100 Contradiction, Principle of, 48, 52, 54 contrastive idiom analysis, 203-5, 217 coordinate, 366, 367 corpora, 337-46 corpus linguistics, 8, 11-13, 227, 234 counteragent, 116 cross-linguistic equivalence of idioms, 203, 206,210,212 cultural component, 215 culture-specific idiom, 204, 215 dative singular -e, 425 declarative, 267, 272-5 declension, 133, 139-41, 145-8 deduction, 69 default reasoning, 455 definiteness, 411, 413, 414, 416, 471-7, 481-4 deictic centre, 105-7

515

516 Langue and Parole in Synchrony and Diachrony deixis, 105-14, 117, 119-29 deontic, 119, 126 description, 75 descriptive present, 84, 86 diachronic, 20, 24, 30, 349, 350, 352, 360 dialect, 20, 21,33 dialectical, 11,13 Dialekt,321,325,328,333 diathesis, 208, 209 dichotomische Paarigkeit, 401-3, 405,407 dinosaur, 283-93 direct case, 490-4, 497, 507 discourse analysis, 222 discourse, 359, 362 displacement in space, 113-17 displacement, 121, 122, 126 distinctive features, 170, 171, 179, 180 distributionalism, 69 dots-and-lines-model, 350 DP, 489,490, 492,498 diirfen, 119, 126, 127 dynamics, 423, 424, 426, 427, 429 economy, principle of, 423, 425-7 Electra, 47, 53, 55-7 elicitation test, 241, 242, 245-50 ending, 77-80 enklisis, 271 Entlehnungen, 323-9 epistemic, 119,120,128,129 Estonian, 82-7 Excluded Middle, Principle of, 52, 57 Excluded Third, Principle of, 52 exponent, 156, 158, 159, 161 extralinguistic application of langue and parole, 100 factuality, 119-27 fiktionale Texte, 92, 96, 97 final clause, 241, 243, 246-50 finite clause, 363-72 Finnish, 82-4, 87 Flexionsstamm, 149,158-62 focus, 256, 257 formal morphology, 76, 78 Fremdworter, 321, 322, 325-33 Futur I, 375, 376, 381, 382, 384, 386, 389-91 Futur II, 375, 384, 389, 390

Futurparadigmen, 375, 376, 378, 383, 388, 389,391 fuzzy logic, 58 ga-Komposita, 402 Gebete, 94, 95 Gebrauch, 37, 39,40-2 Geltung, 37,40,41,89, 91-5 gender, 133-47 generative grammar, 66-70 generative semantics, 68 Geneva lectures, 19, 21, 24, 25, 28, 31, 32 genitive of quantification, 489, 490, 504, 506 genitive, 489-507 Gepflogenheit, 42 Gesetze, 93, 96 Gesprache, 295-8, 300, 303 Gothic, 366, 369 grammaticality judgements, 13 grammaticalization, 119 Grammatikalisierung, 393,400 Grammatikersatze, 93-5 grammeme, 106 Grassmann's Law, 419,421 Greek, 366, 369,420, 421 'haben'-Perspektive, 393 habituality,451-5,462,463 heads, 489-507 history of linguistics, 64, 65 homonymy, 75, 78, 79 horns, 47, 53, 58, 59 hypotaxis, 363, 366 I-Here-Now, 105-8 iconicity, 171 idiom pragmatics, 203, 204, 206, 213, 214, 217 idiom semantics, 203-17 idiom syntactics, 203-9, 213, 214, 217 idiomatic expressions, 423, 424, 427, 428 idioms, 174,221-8,231,234-6 illocutionary force devices, 263-78 illocutionary function, 215-17 image component, 204, 205, 208, 210, 213-16 imagination-oriented deixis, 120-2, 128 immediate constituent analysis, 66 Imperativ, 150, 151 imperative, 82-4, 266-75, 280

Subject Index 517 imperfective, 105-12, 117 imperfektiv, 401, 402, 405, 407, 408 importierte Worter, 326, 330, 331 indicative, 119-24, 128,270-6 Indo-European, 19, 20, 22, 23, 26, 28-31, 86, 363-5,373,419-21 induction, 69 infinitive, 363, 364, 368, 369, 372, 373 inflection, 76-9 information structure, 469, 472, 475, 477, 480, 484, 485 ingressive, 110, 111, 113 inherent case, 489, 490, 494-7 innovation, 423, 427, 429 insolubilia, 54 Institution, 41 intensional contexts, 47, 56, 57 Interferenz, 321, 322, 324, 327, 328, 330, 332 internal causality of evolution, 423, 426, 427 interrogative, 264-79 invented sentences, 13 involvement vs. detachment, 250, 251 irony, 284, 285, 288-92 irregularity, 6, 11, 81 isomorphism, 171 isomorphy, 75, 77 Italian, 264, 266, 267, 270, 272, 274, 275, 279, 280 iterativity, 452-5, 462, 463 iussive, 267, 269, 275 Japanese, 431-45 Junggrammatiker (see also Neogrammarians), 65 Kompetenzlangue, 91, 95 konnen, 119, 124-7 Konsekutivdolmetschen, 295-6, 298, 302, 304 Konsonantenwechsel, 156, 160 Konvention, 40, 43, 44 langage, 36-9, 41 language norms, 351 language perception, 307-16 langue, 1,2,4, 10-13, 19, 20, 22, 23, 25-8, 30-3, 35-9, 81, 83, 84, 87, 89-97, 99-101, 263, 265, 267, 273, 276, 278, 281, 321, 325, 327, 328, 423, 424, 429 Lehnworter, 321-33

Leipzig circle, 26 lexical meaning, 100 lexicon, 135, 136 Liar, 47, 53, 54, 55 linguistique externe, 20 linguistique interne, 20 literary language, 81, 82, 84 Lithuanian, 19 Lugen, 94, 95 Macedonian, 269, 471, 474-7, 480 mag, 126 markedness, 169-80 marker, 158-60 maxim of quantity, 257, 258, 262 Memoire, 19, 20, 23, 25, 27-30, 32, 33 mental image, 204, 207, 210-14 Metalangue, 90 metalinguistic awareness, 307 Metaparole, 90, 91 metaphor, 283-94 Metasprache, 295, 296, 304 Metatext, 295, 296, 298, 299, 302, 303 Middle English dialectology, 337-46 Middle English, 349-62 modal verbs, 119-29 modality, 253, 255, 256 mogen, 119, 126, 127 mood, 105, 106, 119-22, 124, 127, 128, 241-50,265-80 morpheme, 76, 77 morphemic analysis, 76 morphological basis, 76, 77, 79 morphological case, 478 morphology, 133, 134, 139-44 motion, verbs of 108, 113 miissen, 119, 125-8 Musterbriefe, 91, 92 mutativ, 401,404-7 natiirlicher grammatischer Wandel, 388 negation, 255, 258-62 negative Langue, 96 Neogrammarians (see also Junggrammatiker), 65, 68, 69 network strength scale, 350, 352, 354, 355, 358, 360 newspaper language, 84

518 Langue and Parole in Synchrony and Diachrony non-finite clause, 363, 365, 367, 368, 369, 372 noun, 133-47 numerals, 489-508 oblique case, 490-500, 504, 506, 507 observation post, 109,117 Old English, 353, 356, 362, 364, 372,373 Old Germanic, 363, 365, 372 Old Icelandic, 351,367 Old Russian, 363,367 optative, 268,270-2,275-7 organicist, 25, 30 ou tout se tient, 21-34 paradigmatic, 31 paradigms, 76, 81-7 paradox, 47,48, 53-61 parameters of idiom comparison, 205,206, 209,217 parole, 1, 2, 4,10-13,20, 35, 36, 38, 39, 89-97, 99-101,263,265,267,273,276, 278,290, 321, 325, 327, 328,423 participle, 84-6, 363-72 passive, 11-13,105,107,108 passivizability, 12 paucal numerals, 496, 500 perceiver, 107-17 perfect, 2,4,12,13 perfective, 105,106, 111, 113,117 performance, 451, 452,465 Performanzlangue, 91, 92,95 person, 105-16 phase verbs, 111-13 Phonem, 155,156 phonology, 66, 68,72 phoric, 122-8 phraseology, 221,222,230-4,236 political .discourse, 283,288,290,291 Possessivrelation, 393, 394, 398,399 Possessor, 393, 397-9 possessor-relation, 470,477,481,482,484, 487 possibility, 253-62 pragmatics, 451,455,467 Prague School, 65,170,180 prescription, 75 present tense, 84, 85 presupposition, 47, 58-60

pri-, 108,109 principle of uniformity, 349 private documents, 337-46 processual, 109-17 Profit, 321, 330 progress in linguistics, 11 progressive, 12 pronominal clitics, 469-85 Proto-Indo-European, 20,419-21 proto-system of accent, 431,435,442-5 Psychomechanics School of Language, 170 QP, 489-504, 507 quantified phrases, 489-96, 500, 502, 503, 505 quantifiers, 491-507 Quedarative, 212 quotative, 122,124,126,128 rationalism, 69 Realisierungsvariante, 160-2 reanalysis, 363, 365 reconstruction, 431,432, 435, 442,445 reduced clause, 363, 368 redundant forms, 83 Referenz, 89, 91-6 Regel, 39-44 relative meaning, 113-17 relevance theory, 292 restrictio, 54, 55 revolution, 1,13 root, 76,79,149,150 Russian language abroad, 189,192 Sanskrit, 421 Saxon genitive, 425,426 scenario, 285,287,289,290,292,293 schema, 285-9,293 Scherze, 94,95 science writing, 226,229-35 science, 11,12,13,19,20,25, 30,32, 63-71 scientism, 11-13 self-contradiction, 11 Semantic Integrality, 169-72,178 semantic prosody, 228,234 semantics, 67, 68,189,191,192,195,196 sentence-grammar, 13 Serbo-Croation, 489,490,494-6, 501, 502, 507

Subject Index Shikoku dialects, 431, 432,435, 437, 439-43, 445 signifie, 169, 176, 179 Slavonic, 365, 367 Slowenisch, 321-33 social networks, 349, 360, 361, 362 sollen, 119, 125, 126 Sophists, 48, 49, 51,53, 55 specificity, 473, 479 speech acts, 11, 13,263-80 spoken language, 20 spoken vs. written language, 241, 247-51 Sprachgemeinschaft, 37, 39-41,44 Sprachnorm, 321, 324, 327, 328 Sprachspiel, 39 Sprachsystem, 37, 39-45 Sprachwandel, 37, 40 Sprechakt, 38, 40, 42-5 Stamm, 149-63 standard, 351,353, 354, 360 standardisation, 307, 312 Standardsprache, 321, 328 statal, 401,404-7 stem formant, 76 stem suffix, 75-80 stem, 149, 150, 152, 164 stops in Proto-Indo-European, 419,420 structural case, 478-80,489, 490, 493,495, 503 structuralism, 1, 2, 13, 65-70 structure, 19, 23, 25, 29, 30, 75-9 style, 353, 354, 356, 359 stylistic, 85, 86 subjective, 119, 120, 128 subjunction, 365 subjunctive, subjunctive, 119-28, 270, 271, 274-6 submorpheme, 76 subordinate, 363-72 Substitutivity, Principle of, 53, 56 symmetric, 81 synchronic, 19, 20, 24, 30, 31, 349, 350, 352, 355,360 Synchronie, 321,328

519

syntactic synonym, 372 syntagmatic, 31 system, 19-21, 23, 25, 26, 28-31, 75-80 systeme, 19-31,33 tagmemics, 68 temporal-aspectual form, 106 tense, 105, 106, 241-50, 451, 453, 457,467 text grammar, 11, 12 textbooks, 63, 64, 67, 68 Texte, 89, 92-4, 96, 97,295-303 theme-rheme, 11-13 theory of truth, 47, 53, 54, 56, 57, 61 topic, 469, 470, 472, 474, 476, 478, 480-5 traditional grammar, 11 transformational grammar, 68 translation and idioms, 205, 206, 208, 209, 211,213,217,219 translation equivalence, 205 truth conditional semantics, 451 typology of cross-linguistic differences, 203, 206,207,217 Ubereinkunft, 40 unaspirated consonants in Proto-Indo-European, 419-21 Untersuchungsobjekt, 1,11 usage, 82, 83, 85 vantage theory, 183, 184 variation, 241, 242, 247, 337-46 verbal form, 75-8 Verlust, 321,322, 330 vernacular, 351, 353, 354, 360 voice, 11, 12, 105, 107 Vokalwechsel, 154,160, 162 'vulgata'text, 20, 21,30 Whimperative, 272 wo/few, 119, 125, 126, 128 word association, 189, 191, 192, 196 word form, 75, 76 word order, 365, 367, 368, 370, 371, 373 Wurzel, 150, 151, 163 Zeigfeld, 105 zero ending, 78, 79 zero morpheme, 77

Langue and Parole in Synchronic and Diachronic Perspective

Preaspiration in the Nordic Languages: Synchronic and diachronic aspects

Interdependence of Diachronic and Synchronic Analyses (Studies in Language Companion)

Modality and Ellipsis: Diachronic and Synchronic Evidence (Trends in Linguistics. Studies and Monographs)

Quantity and Prosodic Asymmetries in Alemannic: Synchronic and Diachronic Perspectives (Phonology and Phonetics, 5)

Diachronic Clues to Synchronic Grammar (Linguistik Aktuell Linguistics Today)

Diachronic Clues to Synchronic Grammar (Linguistik Aktuell Linguistics Today)

Parole

Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives

Quotative Indexes in African Languages. A Synchronic and Diachronic Survey (Empirical Approaches to Language Topology)

Japanese Discourse Markers: Synchronic And Diachronic Discourse Analysis (Pragmatics and Beyond New Series)

Language and Rhythm in Plautus: Synchronic and Diachronic Studies (Sozomena Studies in the Recovery of Ancient Texts)

Current Trends in Diachronic Semantics and Pragmatics

Language and Rhythm in Plautus: Synchronic and Diachronic Studies (Sozomena Studies in the Recovery of Ancient Texts - Vol. 3)

Parole Passion

Parole passion

Synchronic and Diachronic Perspectives on Contact Languages (Creole Language Library, Volume 32)

Synchronic and Diachronic Perspectives on Contact Languages (Creole Language Library, Volume 32)

Parole Emil

Parole Emil

The Probation and Parole Treatment Planner

Diachronic Syntax

Parole Passion

La Parole malheureuse

The Probation and Parole Treatment Planner

Germanic linguistics: syntactic and diachronic (Current Issues in Linguistic Theory)

Nominalization in Asian Languages: Diachronic and typological perspectives

Parole Sulla Sabbia

Imaginative Programming in Probation & Parole

Infinity and Perspective

Infinity and perspective

Langue and Parole in Synchronic and Diachronic Perspective

Preaspiration in the Nordic Languages: Synchronic and diachronic aspects

Interdependence of Diachronic and Synchronic Analyses (Studies in Language Companion)

Modality and Ellipsis: Diachronic and Synchronic Evidence (Trends in Linguistics. Studies and Monographs)

Quantity and Prosodic Asymmetries in Alemannic: Synchronic and Diachronic Perspectives (Phonology and Phonetics, 5)

Diachronic Clues to Synchronic Grammar (Linguistik Aktuell Linguistics Today)

Diachronic Clues to Synchronic Grammar (Linguistik Aktuell Linguistics Today)

Parole

Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives

Quotative Indexes in African Languages. A Synchronic and Diachronic Survey (Empirical Approaches to Language Topology)

Japanese Discourse Markers: Synchronic And Diachronic Discourse Analysis (Pragmatics and Beyond New Series)

Language and Rhythm in Plautus: Synchronic and Diachronic Studies (Sozomena Studies in the Recovery of Ancient Texts)

Current Trends in Diachronic Semantics and Pragmatics

Language and Rhythm in Plautus: Synchronic and Diachronic Studies (Sozomena Studies in the Recovery of Ancient Texts - Vol. 3)

Parole Passion

Parole passion

Synchronic and Diachronic Perspectives on Contact Languages (Creole Language Library, Volume 32)

Synchronic and Diachronic Perspectives on Contact Languages (Creole Language Library, Volume 32)

Parole Emil

Parole Emil

The Probation and Parole Treatment Planner

Diachronic Syntax

Parole Passion

La Parole malheureuse

The Probation and Parole Treatment Planner

Germanic linguistics: syntactic and diachronic (Current Issues in Linguistic Theory)

Nominalization in Asian Languages: Diachronic and typological perspectives

Parole Sulla Sabbia

Imaginative Programming in Probation & Parole

Infinity and Perspective

Infinity and perspective

Recommend Documents