DISCOURSE CONFIGURATIONAL LANGUAGES
OXFORD STUDIES IN COMPARATIVE SYNTAX Richard Kayne , General Edito r PRINCIPLES A...
42 downloads
987 Views
23MB 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
DISCOURSE CONFIGURATIONAL LANGUAGES
OXFORD STUDIES IN COMPARATIVE SYNTAX Richard Kayne , General Edito r PRINCIPLES AND PARAMETER S OF SYNTACTIC SATURATION Gert Webelhut h VERB MOVEMENT AND EXPLETIV E SUBJECTS IN THE GERMANIC LANGUAGES Sten Vikner PARAMETERS AND FUNCTIONAL HEADS: ESSAYS IN COMPARATIVE SYNTAX Edited by Adriana Bellett i and Luizi Rizz i DISCOURSE CONFIGURATIONS LANGUAGES Edited by Katalin E. Kiss
DISCOURSE CONFIGURATIONAL LANGUAGES Edited by KATALIN E. KISS
New Yor k Oxfor d OXFORD UNIVERSITY PRESS 1995
Oxford Universit y Pres s Oxford Ne w York Toront o Delhi Bomba y Calcutt a Madra s Karach i Kuala Lumpur Singapor e Hon g Kon g Toky o Nairobi Da r es Salaam Cap e Town Melbourne Aucklan d Madri d and associate d companies in Berlin Ibada n
Copyright © 199 5 by Oxford Universit y Press, Inc . Published b y Oxford University Press, Inc . 200 Madison Avenue, New York, New York 1001 6 Oxford i s a registered trademar k of Oxford University Press, Inc . All rights reserved . N o part of this publication may be reproduced , stored in a retrieval system , o r transmitted, i n any form or by any means , electronic, mechanical , photocopying, recording o r otherwise , without the prior permission o f Oxford University Press. Library o f Congress Cataloging-in-Publication Data Discourse configurational languages / edited b y Katalin 6. Kiss . p. cm . — (Oxfor d studies in comparative syntax) ISBN 0-19-508833-6. — ISB N 0-19-508834-4 (pbk. ) 1. Grammar, Comparative and general—Topic and comment. 2. Generative grammar . I. Kiss, Katali n fi . II . Series. P298.D57 199 4 415—dc2 0 93-4103 7
246897531 Printed in the United State s o f Americ a on acid-free paper
Contents 1. KATALI N E. Kis s Introduction, 3 2. JULI A HORVAT H Structural Focus, Structural Case, and the Notion of Feature-Assignment, 28 3. MARC O SVOLACCHIA , LUNELLA MEREU AN D ANNARITA PUGLBELLI Aspects of Discourse Configuationality i n Somali, 65 4. JO
N ORTIZ DE URBINA Residual Verb Second and Verb First in Basque, 99
5. ENRI C VALLouvf Structural Properties of Information Packaging in Catalan, 122 6. JUA N URIAGEREKA An F Position in Western Romance, 153 7. IANTH I MARIA TSIMPL I Focusing in Modern Gret'.k, 176 8. KATALI N E. Kiss NP Movement, Operator Movement, and Scrambling in Hungarian, 207 9. MARI A VILKUN A Discourse Configuationality i n Finnish, 244
10. HYO N SOOK CHOE
Focus and Topic Movement in Korean and Licensing, 269
11. Mi-JEUNGJ o The Theory of Syntactic Focalization Based on a Subcategorization Feature of Verbs, 335 12. PlETER MUYSKE N
Focus in Quechua, 375
This page intentionally left blank
DISCOURSE CONFIGURATIONAL LANGUAGES
This page intentionally left blank
1 Discourse Configurational Languages Introduction* KATALIN E . KIS S
1. Goals The purpos e o f thi s volum e i s to describ e i n th e generativ e framewor k th e lan guage type in which primary sentence articulatio n is motivated b y discourse-se mantic, rather than theta role o r case, considerations . Unti l recently, the existenc e of suc h a languag e typ e ha s no t bee n acknowledge d i n mainstrea m generativ e research. A s a consequence o f the fac t tha t generativ e researc h focuse d o n En glish fo r a long time, an d as a consequence o f the assumptio n that the grammar s of al l language s ar e instantiation s o f on e an d th e sam e Universa l Grammar , i t has been hypothesized that the phrase structure identified in the English sentence , involving a grammatical subject—V P dichotomy , c-commande d b y a single op erator position reserved for a WH operator, constitute s the core o f sentence struc ture i n ever y language . Th e possibilit y tha t thi s hypothesi s ma y no t b e correc t has bee n raised , amon g others , by th e studie s o f th e volum e Subject and Topic (Li an d Thompson (1976)) . A s this volume demonstrate d o n empirical materia l from variou s languages, th e structura l rol e tha t th e grammatical subjec t play s i n the English sentenc e ma y be fulfille d b y a constituent no t restricted wit h respec t to grammatica l functio n o r cas e i n othe r languages ; or , in generativ e terminol ogy: th e structura l relation [NP , S ] can be used t o express no t only th e function 'grammatical subject, ' associate d wit h th e most prominen t thet a role , but , alter natively, th e discourse-semanti c functio n 'topic, ' a s well . Actually , the volum e * Th e ide a o f compilin g thi s boo k wa s initiate d b y th e EUROTY P project o f th e Euro pean Scienc e Foundation, i n the Word Orde r Workgrou p o f which the editor was assigned the task o f studyin g discourse configurationality . Som e o f the editing , an d th e writin g of the Introductio n an d th e editor' s chapte r o n Hungaria n wa s conducte d whil e sh e wa s a Fellow a t th e Cente r fo r Advance d Stud y i n th e Behaviora l Science s a t Stanford . Sh e gratefully acknowledge s th e financia l suppor t o f th e Andrew W . Mellon Foundation . 3
4
Discourse Configurational Languages
Subject and Topic has not been the first one to put forth this idea; it emerge d as early a s Brassa i (1860 , 1863-1865) , an d ha s bee n i n th e cente r o f th e wor k o f the Pragu e Schoo l fo r decade s (see , fo r example , Sgal l an d HajiCov a (1973) , HajiCova (1983) , and Sgal l (1984)) . Within th e generative framework, firs t i t appeared that the 'leas t costly' way of derivin g th e sentenc e structur e o f a topic-prominen t languag e i s t o appl y Topicalization, a transformation available in the grammar of English and simila r languages, as well, to a D-structure with a grammatical subject—VP articulation (see, fo r example, Horvat h (1981)). I n this approach, however , topic-prominen t and subject-prominen t language s diffe r merel y i n ho w frequentl y the y utiliz e Topicalization, which , o n th e on e hand , blur s th e differenc e betwee n th e tw o language types , and , o n th e othe r hand , doe s no t reflec t th e intuitio n tha t in a topic-prominent language , the topi c is , in a way, an alternative t o the subjec t a s the VP-external argument. Ther e have also been attempts t o generate topic-prominent sentenc e structur e b y applyin g Topicalizatio n t o a fla t S(entence)—se e Farmer (1980) , an d E, Kis s (1976, 1981) . Thes e approache s coul d no t captur e the parallellism between the subject-predicat e and topic-predicate structures , either. Apparently , the generative framework became flexible enough t o represen t both the differences an d the similarities of the two language types after the emergence o f the 'principle s and parameters* theory of Universal Grammar (Chomsk y (1981)). Particula r innovation s in generativ e theory tha t hav e provide d mean s of representin g topic-prominen t languages as a typ e different from , bu t paralle l to, th e subject-prominen t typ e includ e th e base-generation o f the subjec t insid e the V-projectio n i n ever y languag e (see , fo r instance , Koopma n an d Sportich e (1990)), or the assumption of Head (Verb) Movement (se e Chomsk y (1986)). Th e base-generation o f the subjec t inside the VP has lead t o the recognition o f Sub ject Movemen t a s a transformation simila r t o Topicalization; an d as a result of this, subject-prominen t an d topic-prominen t language s ca n b e viewe d no w a s differing i n whether their sentence structur e is derived by externalizing th e gram matical subject, or by externalizing an arbitrary argument. (E. Kiss (thi s volume) , in fact , propose s to analyze Topicalization, together with Subject Movement, as NP Movement , redefining NP Movement as a transformation creating a syntactic predicatio n relation. ) Naturally , thi s approac h require s a n explanation (per haps i n term s o f case theory ) o f why argument-externalization mus t involv e th e grammatical subjec t in some languages , and can involve any argument in others . Alternatively, th e subject i n subject-prominent language s an d the topic in topicprominent language s ca n be assigned t o different VP-externa l positions , an d the structurally paralle l subject-predicate an d topic-predicate relations can be established b y Verb Movement int o the head position o f the maximal projection domi nating the subject, or into the head position of the maximal projection dominating the topic , respectively (see , fo r instance , Ortiz de Urbina (1991)) . The fac t that many languages have a designated structura l positio n fo r focus , akin to the WH-position o f the English sentence, ha s become recognize d i n gen erative theory in the second hal f of the seventies—owing to the work of Schachter (1973) on Akan, Hausa, and Ilonggo; Horvath (1976), (1981), and E. Kiss (1977), (1981) o n Hungarian ; Givon (1975 ) o n variou s Bant u languages ; Awobuluyi
Introduction 5
(1978) o n Yoruba ; Watter s (1979 ) o n Aghem ; an d d e Rij k (1978 ) o n Basque , among others. Wherea s th e early studie s focuse d o n where t o assum e th e focu s position i n phrase structure , an d how to analyz e the Focusing transformation , in the eightie s an d earl y ninetie s th e comparativ e perspectiv e o f th e Governmen t and Bindin g theor y an d th e principle s an d parameter s approac h ha s le d t o th e question wh y over t focu s movemen t i s triggere d i n som e language s bu t no t i n others. I s i t th e cas e tha t language s may diffe r i n whethe r th e focu s operato r must occup y a scope positio n a t S-structure or at LF (cf. E . Kiss (1987))? O r is a designated structura l focus positio n licensed in languages in which the syntac tic feature [+Focus]—assumed to trigger Focus interpretation both at LF and PF— is limited t o being assigne d by a particular X° category , namely V (see Horvat h (1981), (1986)) ? Wit h the emergence of functional projections , furthe r possibili ties o f description hav e arisen: language s may have a particular functional projection wit h a head marke d [-fFocus] , an d focu s movemen t may be triggere d b y the necessity of spec-head agreement , o r by the requirement o f feature licensin g (see Choe (1989, thi s volume) and Brody (1990)). I n this framework, languages with structura l focus and languages with focus-in-situ can, again , be viewe d ei ther a s differing i n the leve l o f representation a t which the spec-hea d agreemen t or feature licensin g must take place, or a s differing i n the feature s of thei r func tional heads . As generativ e analysi s ha s bee n extende d t o mor e an d mor e languages , i t has becom e obviou s tha t language s in whic h topi c an d focu s for m ke y consti tuents o f sentenc e structure , i.e. , language s i n whic h primar y sentenc e arti culation serve s t o expres s discourse-semanti c functions, represent a type whic h is presumabl y a s commo n a s th e languag e typ e represente d b y English ; i t spreads acros s continent s an d acros s languag e families . I n Europe , Basqu e (de Rij k (1978) , Orti z d e Urbin a (1991) , Eguzkitz a (1987) , Rebusch i (1989)) , Catalan (Vallduv i (1990 , 1991)) , Hungaria n (Horvat h (1986) , E . Kis s (1987)) , Bulgarian (Rudi n (1986)) , Russia n (Kin g (1993)) , Gree k (Tsimpl i (1990) , Agouraki (1990)) , Finnis h (Vilkun a (1989)) , Romania n (Primus (1992) , Ulric h (1985)), Turkis h (Erguvanl i (1984)), an d severa l o f th e Caucasia n languages , among the m Armenian (Comri e (1984) ) an d Circassia n (Ouhall a (1991) ) hav e been identifie d a s discourse-configurationa l languages . Fro m Asia , Nepal i (Wallace (1985)), Hindi (Gambhir (1981)), and Korean (Choe (1989), J o (1986)) have bee n reporte d t o b e discourse-configurational , wherea s Japanese , an d in some accounts, Chines e ar e known to be topic-prominent. Africa n discourse configurational language s include , amon g others , Somal i (Lecarm e (1992) ; Svolacchia, Mereu , an d Pugliell i (thi s volume)) , th e Chadi c language s (Tulle r (1992), Horvat h (thi s volume)) , the Bant u Aghem an d Kikuy u (Watter s (1979) , Clements (1984)) , Yorub a (Awobuluy i (1978 ) an d Berbe r (Calabres e (1987) , Ouhalla (1990)) . O f th e America n India n languages , Haid a (Enric o 1986 , in preparation) , Omah a (Rudi n (1992)) , th e Maya n languages (Englan d (1991) , Aissen (1991)) , an d Quetchu a (Muyske n (thi s volume) ) ar e know n t o b e discourse-configurational. Accordin g t o evidenc e presente d i n Schachte r (1973), th e Austronesia n Ilonggo ma y als o belon g t o th e languag e typ e i n question.
6
Discourse Configurational Languages
The eleven paper s in this volume analyze discourse-configurational languages from fou r continents : Chadic , Somali , Basque , Catalan , Ol d Romance , Greek , Hungarian, Finnish , Korean , an d Quetchua . (Th e orderin g o f th e paper s i n th e volume corresponds t o the "areal" order of the languages that they discuss.) Thes e papers serve , o n the one hand, to establish empiricall y th e claim tha t discourse configurational language s represent a widespread languag e type that is to be reckoned with when formulating hypotheses about Universal Grammar. O n the other hand, the y ai m t o explor e way s o f representin g discourse-configurationa l lan guages in the generative framework; they propose phrase structures for such languages, an d attemp t to identif y th e parameter s o f Universa l Gramma r tha t ar e responsible fo r topic-prominenc e an d fo r structura l focus. Althoug h the analy ses proposed her e converge t o a considerable extent , n o unified solution i s pro vided; rather, alternative hypotheses are put forth, the viability of which will have to be teste d b y furthe r research .
2. What is Discourse Configurationality? The properties o n the basis of which a language is categorized a s discourse con figurational are , i n intuitiv e terms, th e following : A. Th e (discourse-)semanti c function 'topic, ' servin g t o foregroun d a specifi c individual tha t somethin g wil l b e predicate d abou t (no t necessaril y identica l with th e grammatica l subject) , i s expresse d throug h a particula r structura l relation (in other words, it is associated with a particular structural position). B. Th e (discourse-)semanti c functio n 'focus, ' expressin g identification , i s realized throug h a particular structura l relation (tha t is , b y movemen t int o a particular structural position). While these two properties ofte n co-occur , they are independent of each other. Most o f th e language s studied in thi s volum e share both , i.e. , the y ar e typ e AB discourse-configurational languages , but , fo r instance , Aghe m i s show n (b y Horvath) onl y t o hav e property B (typ e B discourse-configurationality) . Ther e are als o language s whic h onl y displa y propert y A (typ e A discourse configurationality); th e best-known example i s Japanese. Properties A and B ar e sometime s realize d i n modifie d forms . Fo r instance , in Finnis h th e semanti c functio n o f contras t ma y cu t across , an d supersed e th e functions o f topic and focus: ther e is a sentence initial position (i n addition t o a topic position ) tha t ca n b e occupie d b y a contraste d elemen t o f eithe r topi c o r focus function . Below, I wil l giv e a brief semanti c an d syntacti c characterizatio n o f proper ties A and B, and their key notions , the topic an d the focus—partl y o n the basi s of previou s literature, partly o n th e basi s o f th e studie s i n thi s volume . I wil l compare th e syntacti c realizations of topi c and focus acros s languages , pointing
Introduction 7
out th e invarian t elements i n the language-specifi c descriptions, an d attemptin g to identif y th e parameter s of variation. 3. Topic-Prominence Topic a s a (discourse-)semantic ter m denote s th e functio n of the constituen t tha t the sentence i s about. Unde r a particular, semantic , o r 'notional, ' interpretatio n of the subject-predicate relatio n (se e Rothstein (1983)) , the topic is identical with the subjec t o f predication, o r in othe r words , wit h th e 'notiona l subject. ' Although th e topic , o r notiona l subject , ver y ofte n coincide s wit h th e gram matical subject, that is, with the constituent expressing the most prominen t thet a role, and bearing nominative case and agreeing with the finite ver b i n many languages, thi s i s no t necessarily so . Sentence s which have a grammatical subjec t can b e subjectles s semantically . Th e distinctio n between proposition s contain ing a notiona l subject, an d proposition s no t containin g any (categorica l versu s thetic judgements in the logica l theor y of Marty (1918), (1965)) has been intro duced int o generative literature by Kuroda (1972-73). A s he puts it, categorica l judgements, displaying a notional subject-notional predicat e structure , consist of two acts : th e act of recognition o f that which is to be made the notional subject , and the ac t of affirmin g o r denying what is expressed b y the predicat e abou t th e notional subject . I n accordanc e wit h thei r rol e o f fore-groundin g a particula r individual a s th e subjec t o f predication , notional subjects/topic s ar e [+specific ] referring expressions . (Mos t type s of generics , fo r instance , definit e generics , can als o functio n a s topics—as expecte d o n th e basi s o f analyse s treatin g them as referentia l expressions, representing names o f kinds—se e Carlso n (1978 ) o r Heyer (1985)) . Her e are two categorical judgements, displaying a notional subject—notional predicat e structure: (1) a . Fid o i s chewing a bone . b. Th e do g is a domestic animal . A theti c judgement, containin g a mere notiona l predicate, o n the othe r hand , consists o f a singl e act : th e ac t o f th e recognitio n o f th e materia l o f a judgement. Th e linguisti c realizations of thetic judgements include impersonal , exis tential, presentative , an d universal sentences: (2) a . I t i s raining. b. Ther e i s a dog i n th e room . c. A do g cam e int o the room. d. Al l dog s lik e bones. Although i t i s no t predicte d b y Marty' s theor y tha t a categorica l judgemen t can contai n more tha n on e notiona l subject, this seem s t o b e th e cas e i n many natural language s (in the majorit y o f those discussed in this volume) . Conside r the Catala n sentence s in (3):
8
Discourse Configurational Languages
(3) a . L'Ann a e l caf e [e l v a fe r ahir ] Anna th e coffee i t mad e yesterda y 'The coffe e Anna made yesterday. ' b. E l caf e 1'Ann a [e l va fe r ahir ] The two sentences mea n exactly the same, whic h argues against a recursive, two level notiona l subject-notional predicat e structure , such that, for example, i n (3a ) el cafe el va fer ahir is predicated o f I'Anna, an d withi n th e predicate , el va fer ahir is predicated o f el cafe. I t seems mor e appropriat e to say that the sentence s in (3a ) an d (3b ) mak e statement s abou t two participants o f the given event , tha t is, the y predicate abou t two notiona l subjects . The notion of topic, o r notional subject , has also been reinterpreted i n dynami c semantic frameworks. Accordin g to Vallduvi (1990, thi s volume), the role of the topic (in his terminology, 'link') is to indicate under which address in the hearer's knowledge store th e new information carried by the sentence i s to be entered. A language is identified as topic-prominent, more precisely, a s a discourse configu rational language with property A, if it realizes categorica l an d thetic judgements in differen t syntacti c structures . Englis h i s obviousl y no t o f thi s type , a s sen tences (la , b) , expressing categorical judgements, and sentences (2a-d) , express ing theti c judgements , hav e th e sam e syntacti c structure . No w compar e thei r Hungarian equivalents : (4) a . [
T
Fido] [ vp rag eg y csontot] Fido chew s a bon e 'Fido i s chewin g a bone.' b. [ T A kutya ] [vp haziallat] the do g domestic-anima l 'The do g is a domestic animal. '
(5) a . [
vp
Esik a z eso ] falls th e rai n 'It is raining.' b. [ vp Van egy kutya a szobaban ] is a do g th e room-i n 'There is a dog in th e room.' c. [ vp Bejott eg y kutya a szobaba ] came a do g th e room-into 'A do g has com e int o the room.'
Although a universally quantified subject is in preverbal position , it s position i s different fro m tha t of a nonquantified specifi c subject . Thi s become s clea r if fo r example a sentence adverbial , whic h must be external t o the VP, is inserted int o both sentenc e types :
Introduction 9
(6) [
Fido] szerinte m [ ] vpszereti a csontot Fido according-to-m e like s th e bon e 'Fido, accordin g t o me, likes bones. ' T
(7) a . *[
Minden kutya ] szerinte m [ ] VP szereti a csontot every do g according-to-m e like s th e bon e 'According to me, all dogs lik e bones.' b. Szerinte m [ vp minden kutya szereti a csontot ] T
In the language type which assign s to categorial an d thetic judgements differ ent syntactic structures, the notional subject , o r topic, doe s no t have to coincid e with the grammatical subject at all. Thu s sentences (5a-c) , or (7b), in which the grammatical subject , not being referentia l and specific , i s not suitabl e fo r a no tional subjec t role , ca n b e reformulate d a s sentence s wit h a notiona l subject notional predicate articulation, predicating about the object or the goal argument: (8) a . [
A szobaba ] [ vpbejott eg y kutya ] the room-int o cam e a do g 'Into the room came a dog.' b. [ T A csontot ] szerinte m[ vpminden kuty a szereti ] the bone-AC C according-to-m e ever y do g like s 'In my opinion, bones are liked by every dog. ' T
The question whether in a particular language categorical an d thetic judgements are expressed by identical or different syntacti c structures i s actually no t a trivial one. A s Vallduvi (1993) showed in connection with Catalan, a T(opic) V P structure ca n be easil y misdiagnose d as a S(ubject ) VP structure—owing t o th e fac t that i n the case o f an agentive predicate , the agent is the most unmarked carrie r of bot h th e topi c rol e an d the grammatica l subjec t role . T o decide th e issue , i t has to be carefully examined whether or not specific and non-specific (including universally quantified ) grammatical subject s shar e th e sam e structura l positio n (if the y do, the language doe s no t distinguish categorical an d thetic judgements syntactically; hence i t is not topic-prominent). I t is, naturally, also t o be teste d if a sentence-initia l grammatica l subjec t occupie s th e sam e positio n as , say , a proposed object ; o r i n cas e a sentenc e contains a proposed grammatica l subject and a proposed object , whethe r their relativ e orde r i s free . I f i t is , i t i s indica tive of property A (topic-prominence). (Recal l that in the great majorit y o f dis course-configurational language s with property A, more than one topic is possible , and thei r orde r i s free. ) To make the classification of languages on the basis o f criterion A even mor e difficult, ther e als o exis t comple x borderlin e cases . I n Finnish , accordin g t o Vilkuna (thi s volume) , ther e i s a syntacti c positio n whic h i s to be fille d b y th e topic in the case o f categorical judgements. I n the case o f thetic judgements, on the othe r hand , th e sam e positio n ha s t o b e fille d b y th e grammatica l subject. Vilkuna calls the latter case default topicalization. In another language type, represented, fo r example , b y Dutc h an d Icelandic , o r b y Germa n i n embedde d
10
Discourse Configurational Languages
contexts (see Diesin g (1992)), the grammatical subjec t occupie s the VP-external subject-of-predication positio n not onl y if it i s referential an d specific , but als o if it is quantified (accordin g to Diesing, if it is quantified and specific) . I n Hungarian (se e E . Kiss (199la) ) or in Mayan (Judith Aissen p.c.) quantified phrases are moved into an A-bar position distinc t from the topic position . Apparently , in Dutch, Icelandic , an d i n Germa n embedded clauses, th e topi c position an d th e quantifier positio n are not distinguished; or, in these languages, the condition of complement externalizatio n is tha t th e complemen t be [+specific] , rathe r tha n referential and [+specific] (see Diesing (1992), chapters 3.3 and 3.4). I n any case, neither Finnish, no r Dutch, Icelandic, o r German are topic-prominent accordin g to our criterion, as categorical an d thetic judgements are not consistently distinguished i n either o f them. To illustrate the difficultie s o f distinguishing between T VP an d S VP structures, let us refer to a case in Basque. A s Ortiz de Urbina (this volume) explains , in a Basqu e sentenc e o f th e orde r [X P YP V], X P function s a s topic , an d Y P functions a s focus—unles s X P i s subject , an d Y P i s object , i n whic h case th e sentence can also be communicatively neutral. T o account for its neutrality, Ortiz de Urbina assumes SOV to be the initial order, and derives TFV by Topicalization, Focusing, an d V Movement (which is a necessary correlat e o f Focusing), as follows: [Ty j Fy j V k [t j t j tk]]. Thi s derivation would predict th e possibilit y o f [T j [S t; V]] sentences; however , suc h sentences d o not exist; i n an OSV sentence, S is necessarily focus. Th e description faces a dilemma: i t either needs an ad ho c filter t o rule out [T { [S t; V]] sentences, or it assimilates th e neutral SOV order t o the TFV pattern, an d allow s objec t foc i no t t o b e interpreted a s identifyin g operators. The identification o f topic ma y be problematic no t only because of the close correspondence between the topic an d the grammatica l subject. Th e topi c ma y also be difficult t o distinguish from non-topi c lef t periphera l elements . I n Hungarian, fo r instance , a lef t periphera l elemen t ca n b e eithe r CP-interna l o r CP external, an d i t function s a s a notiona l subject onl y i n th e forme r cas e (cf . E . Kiss (1987 , ch . 2) , an d (1991b)) . A CP-internal left periphera l elemen t (i.e. , a topic) i s alway s referential an d specific ; a CP-externa l lef t periphera l element, on the other hand, can also be a quantifier o r a predicative nominal , eve n a verbal prefix. Fortunately , in Hungarian there ar e als o som e forma l (syntacti c an d phonological) criteri a fo r distinguishin g between th e tw o type s o f constituents. The CP-internal topi c i s proposed from the VP by a movement rule , th e equivalent o f Englis h Topicalization , henc e th e topic—ga p relatio n i s subjec t t o Subjacency. Th e CP-externa l lef t periphera l elemen t ha s bee n identifie d a s a n instance of Left Dislocation: it can be associated wit h a resumptive pronoun, its relation t o the corresponding gap does no t have to observe Subjacency , etc. This naturall y does not mean that it is the syntactic rul e called Topicalizatio n that creates constituents functioning a s topics/notional subjects across languages; or that a constituent derived in syntax by Left Dislocation i s never topi c seman tically. O n th e contrary , the topi c is analyze d a s the outpu t of Lef t Dislocatio n in many languages—especially in those in which it is associated with a resumptive pronoun, a s i n Haid a (Enrico (1986, an d i n preparation)) , Jakaltek an d Tzotzi l
Intr
\
(Aissen (1991)) , or Somal i (Svolacchi a e t al . (thi s volume)) . Th e stud y o n So mali distinguishe s betwee n topic s create d b y Cliti c Lef t Dislocation , an d topics created b y Hanging Topic Lef t Dislocatio n (cf . Cinque (1990)), traditionally calle d nominativus pendens. Th e latter are more loosely related t o the subsequent predi cate: the y hav e nominativ e case , an d the y ar e no t argument s o f th e predicat e (therefore, the y d o not actuall y qualify a s subject s o f predication , o r topics, ac cording t o the criteria t o be presented below) . Wherea s sentenc e adverbial s of ten shar e th e left-periphera l positio n of topic constituents, they are not analyze d as suc h i n th e paper s o f thi s volume ; onl y argument s ar e claime d t o hav e th e ability o f denotin g that which th e sentenc e i s about . (I n E . Kis s (thi s volume) , locatives an d temporals ar e allowed optionall y t o have th e statu s o f non-obliga tory arguments ; henc e the y ca n be analyze d a s topics. ) It is importan t to emphasize that the answer to the question o f whether a lan guage display s propert y A—o r propert y B—o f discours e configurationalit y i s independent o f th e traditiona l configurationalit y issue, i.e. , whethe r o r no t th e grammatical function s 'subject ' an d 'object ' ar e configurationally distinguishe d in th e give n language . Th e existenc e o f a notional subjec t positio n ha s nothin g to d o wit h whethe r th e grammatica l subjec t an d th e objec t occup y structurall y parallel o r non-paralle l position s inside th e notional predicate . Tw o of th e lan guages claime d t o b e discourse-configurational : Haid a (Enric o (1986 , an d i n preparation)), and Hungarian in the analysis of E. Kiss (thi s volume), ar e assume d to have a flat VP , which also include s the subject . I n th e other case s it is eithe r explicitly claime d o r implicitl y assume d tha t th e base-generated positio n o f th e subject i s structurall y more prominen t tha n tha t o f th e object . The forma l propertie s o f th e topi c constituent s o f the language s examine d i n this volume can be summarized as follows: Th e topic has a morphological marke r in two of the languages (Korean and Quetchua); apparently topic marker s ar e less frequent tha n focu s markers . Th e studie s tha t als o conside r fact s o f phonolog y claim tha t th e topi c canno t bea r th e main stres s o f the sentence . Syntactically, the topic i s external t o the notional predicate—eve n thoug h th e syntactic categor y assigne d t o th e notiona l predicat e i s no t th e sam e i n ever y language (i t i s unclear to wha t extent thi s i s du e t o factua l difference s betwee n the languages , an d to what extent t o differences in notation). I n Finnish, an d in Hungarian (accordin g t o E. Kiss) , i t is sai d to be external t o the VP. I n Basque , Catalan, Korean , Somali , an d i n Hungaria n i n th e analysi s o f Horvath , i t i s ex ternal t o th e IP . I n Gree k i t i s externa l t o T(ense)P . I n Quetchua , wher e I P i s claimed t o be dominate d b y a focus projectio n calle d Ev(indential ) Phrase , i t i s external to EvP . The precise : location of the topi c als o depend s on whethe r or not th e give n languag e allows multipl e topics . I n case it does, the author s usu ally assig n th e topics t o adjoined positions. Thu s i n Catalan i t is adjoined t o IP; in Greek , to TP; i n Quetchua , t o EvP. (Th e adjunctio n analysi s correspond s t o the standar d G B analysi s o f Topicalization a s adjunctio n to IP—se e Lasni k an d Saito (1992). ) I n Somali , the topic i s placed i n [Spec , CP] , an d multiple topic s are derived by the recursion o f the C projection. E . Kiss derives multipl e topic s in Hungaria n b y moving one into [Spec , T(ense)P], an d adjoining th e rest t o TP.
12
Discourse Configurational Languages
In Korean, which allows a single topic per clause, the topic is placed i n [Spec , CP] by Jo. Cho e analyzes Korean topic constructions similar to relative clauses : the topicalize d N P is external t o a Topic Phrase , th e specifie r o f which i s occu pied b y a n empty operator . Topics typicall y precede WH-phrases. Som e author s take thi s as evidence of their being adjoined to CP. A t the same time, however, topics in embedded clause s tend t o follo w th e complementizer . I n fact , Bulgaria n is th e onl y languag e de scribed i n th e literatur e surveye d i n thi s introductio n i n whic h th e topi c o f a n embedded sentenc e ca n preced e th e complementize r (se e Rudi n (1986)) . I n Catalan, Hungarian, or Korean no contradiction arises i n the relative orderin g of the complementizer , th e topic , an d th e WH-phras e becaus e th e WH-phras e i s assigned t o th e specifie r position o f a projection below th e complementize r (IP , VP, and FP, respectively). Anothe r way of resolving this contradiction is assuming CP-recursio n (se e fo r instance, Aissen (1991)). I n Somali , th e strin g of top ics can be not only followed, but also interrupted, or preceded b y the focus, which is derive d b y placin g both topic s an d the focu s into the specifie r positio n o f re curring C P projection s (se e Svolacci a e t al . (thi s volume)) . Some languages , for instance Somali, Catalan, and Greek, also have after-topics, whic h shar e th e phonologica l propertie s an d the morphologica l marke r (if any) o f pre-topics. After-topic s expres s ol d information ; it i s unclear, however , if the y hav e an y other semanti c function. Vallduv i (1993 ) show s tha t a Catala n post-topic doe s no t functio n as a notiona l subject , o r i n hi s dynami c semantic s framework, a s an address under which the information conveyed b y the sentenc e is t o b e entered. The topi c i s often associate d with a resumptive pronoun or clitic (mor e ofte n than th e focus) . Ther e i s a n obligator y resumptiv e pronoun i n Somali , Greek , and Catalan , an d a n optiona l resumptiv e pronoun i n Korean . Othe r language s reported in the literature in which the topic co-occur s wit h a resumptive pronoun include Haida (Enrico (1986, and in preparation)), Bulgarian (Rudin (1986)), and Jakaltek an d Tzotzil (Aisse n (1991)). There i s n o genera l agreemen t amon g th e author s o n whethe r topic s ar e ex tracted fro m th e predicate by movement, or base-generated i n their surfac e position and associated with the predicate-internal gap by coindexation. Th e presenc e of a resumptiv e pronou n i n Somali , Greek , an d Korea n (a s wel l a s i n Haid a (Enrico (i n preparation) ) an d Jakalte k and Tzotzi l (Aisse n (1991)) ) i s take n t o be evidence of base-generation. I n the case of Korean (and Haida, Jakaltek, and Tzotzil), th e assumptio n of base-generation is als o supporte d by th e limited oc currence o f topic in embedded contexts , and by a lack o f Subjacency effects . I n Greek, th e topic can not only bear a case corresponding t o the predicate-internal gap, a s expecte d i n th e cas e o f movement , but , alternatively , i t ca n als o bea r nominative. A t th e sam e time , th e topic-ga p relatio n observe s Subjacency . Tsimpli resolves th e contradiction betwee n the presence o f a resumptive pronoun, indicative o f bas e generation , an d th e observanc e o f islan d constraints , indica tive of movement, by base-generating the topicalized lexical element i n situ, and moving a n invisibl e pro, generate d i n argumen t position a s a siste r t o th e resumptive clitic . A similar derivation is propose d by Choe .
Introduction 1
3
The structural relation holding between th e topic constituen t an d the predicat e phrase (whether its syntactic category be VP, IP, TP, FP, or EvP in the give n lan guage) appears to be a version of th e syntacti c predication relatio n identifie d in Williams (1980 ) an d Rothstein (1983) . (I n Somali , th e situatio n ma y be mor e complex, owing to the mingling o f the topics and the focus.) Thi s syntactic predication relatio n is , i n essence , a strictl y loca l relatio n (marke d b y coindexation ) between tw o maxima l projections—one o f which , functionin g as th e subjec t o f predication, c-command s th e other , functionin g as th e predicate , an d bind s a n argument positio n i n it. Th e locality relatio n betwee n th e subject o f predicatio n and th e predicate i s hard to give an invariant formulation, give n tha t the techni cal details of description in the different language s do not coincide; for instance , the precis e formulatio n is affecte d b y whethe r the subjec t o f predicatio n occu pies specifie r position , adjoine d position , o r ca n occup y either . Th e localit y condition on predication was identified i n Williams (1980) as mutual c-command. With the emergence o f Infl, it had to be reformulated as mutual m-command (se e Rothstein (1983)) . If , however, Topicalization involves adjunction, as propose d by mos t of th e authors of this volume, then eve n m-command is to o restrictive , because i n th e framewor k of Chomsky (1986) , th e complemen t o f a hea d doe s not m-command th e constituents adjoine d t o the maximal projectio n o f the hea d (thus e.g. th e IP complement o f C does not m-command the NPs adjoined to CP) . A locality condition that would adequately constrain in most languages described in thi s volum e the relatio n between th e subjec t of predication an d th e predicat e would b e the requiremen t tha t ther e b e n o intervening maxima l projectio n tha t c-commands th e predicate, an d does no t c-command th e subjec t o f predication. (Alternatively, predicates could be analyzed as being o f the category X', and the subject o f predication could be identifie d wit h the specifie r of XP, as in Stowel l (1981). Thi s view , however, would no t b e compatibl e wit h analyse s i n whic h the topic phrase i s adjoined to the predicate phrase , for instance, wit h the analy ses of Catalan , Greek , o r Quetchua proposed i n thi s volume. ) If th e structural relation holdin g between the topic an d the predicate in topic prominent languages is, indeed, the same syntactic predication relatio n tha t hold s between th e subjec t an d the predicat e i n subject-prominen t languages, the n th e main differenc e between th e two language type s i s tha t i n topic-prominent lan guages th e (primary ) syntacti c predication structur e i s alway s directl y mappe d on a notional predication relation, whereas in subject-prominen t language s thi s is not the case . It is reasonabl e to assum e that th e notional predication structure (tha t is , th e categorical o r the thetic nature) of a sentence is part of its meaning, and it is 'rea d off th e structur e entering semanti c interpretation i n every language . The n th e subject-prominent Englis h an d th e topic-prominen t languages described i n thi s volume diffe r i n th e leve l o f representation at whic h they structurall y represent predication. Wherea s i n topic-prominent language s i t i s represented a t S-struc ture already , in English, th e predicatio n structur e ma y be identifie d a t LF, afte r operator movemen t has taken place (se e Guero n (1980) , an d Rothstein (1983)) . Hence a quantifier , o r a non-specific, obligatoril y stressed , focu s phras e i s re -
14
Discourse Configurational Languages
moved fro m subject positio n t o operator positio n prio r t o predication interpreta tion, s o it is not interprete d a s a notional subject . In vie w of this , th e criterio n o f topic-prominence ca n be reformulate d a s fol lows: a language has property A of discourse configurationalit y i f an d onl y i f in that languag e ther e i s a one-to-on e correspondenc e betwee n th e syntacti c an d notional predicatio n structures . The possibility of representing the notional predication structur e o f a sentenc e structurally eithe r visibly , a t S-structure , o r invisibly , a t LF , might b e regarde d as a parametri c variatio n allowe d fo r b y Universa l Grammar . I t als o hold s o f other semantically significant relations, fo r instance, of various operator-variabl e relations, that they can be structurally encoded eithe r at S-structure o r at LF across languages. However , th e assumptio n tha t language s ca n choos e a t rando m be tween representin g th e categorica l versu s theti c semanti c structur e o f sentence s at S-structure , an d representin g i t a t LF , would b e contrar y t o th e spiri t o f th e Principle o f Least Effort o f Chomsky (1991), o r the Earliness Principle o f Pesetsky (1989). Englis h i s presumably force d t o externaliz e th e grammatica l subjec t i n every case, an d subsequently remove it from subjec t position i f it is not a [^specific] referrin g expression , an d canno t be interprete d a s a notiona l subject , be cause i n Englis h th e grammatica l subjec t canno t receive nominativ e cas e fro m Inflection insid e the VP. Onc e the condition s of nominative assignmen t are sat isfied, argumen t externalization is free—via Topicalizatio n an d PP Preposing. I f this assumptio n i s right , the n topic-prominenc e an d subject-prominenc e deriv e from a paramete r determinin g th e wa y o f nominativ e assignmen t i n a give n language. The language type represented b y Dutch, Icelandic, or German embedde d sen tences does no t confor m eithe r t o th e subject-prominen t o r t o th e topic-promi nent type. I t is beyond the scope of this introduction and this volume to speculat e about whic h parameter s interac t i n triggerin g an d constrainin g complemen t externalization i n these languages. The variatio n observed amon g topic-prominent languages ca n be trace d bac k to a fe w mino r parameters . I n th e syntacti c realization o f th e topic-predicat e structure, th e syntacti c category o f th e predicat e can var y (i t ca n b e VP , IP, FP, etc.) I t i s als o subjec t t o parametri c variation whethe r o r no t th e topi c i s mor phologically marked , an d whether or not it bears the case assigne d t o the gap in the predicate . Th e ga p i n th e predicat e ca n b e eithe r empty , o r fille d wit h a resumptive pronoun/clitic . Th e presenc e o r absenc e o f resumptiv e pronoun s may be a secondary effect, dependin g on whethe r the topi c i s related t o the ga p by movemen t o r merel y b y construal . I t varie s whethe r a language allow s on e or more topi c pe r clause; this, again , may be a consequence o f whether the topi c argument i s externalize d vi a substitution , o r b y som e othe r means . Or , i f multiple topicalizatio n i s substitutio n into th e specifie r position s dominate d b y an iterate d maxima l projection, the n th e possibilit y o f multipl e topicalizatio n depends o n whethe r or no t th e maxima l projectio n immediatel y dominating the topic ca n b e iterated.
Introduction 1
5
4. Focus-Prominence The term focus is used in linguistic literature in at least two different senses : i t can denot e the sentenc e par t carrying new information , and i t can als o mea n an operator expressing identification. Th e two meanings are often referre d to as wide focus an d narrow focus, respectively . I n generative synta x the latter notion, that of th e focus operator , has gaine d significance . Althoug h th e focus operato r has neither a morphologica l marker , no r a n invarian t positio n i n English , it s pres ence i n a sentenc e ha s visibl e consequences ; thu s i t trigger s Wea k Crossove r effects, simila r t o quantifiers—se e Chomsk y (1976) . Consider , fo r instance: (9) a . *HiS ; mother love s JOHN; . b. His ; mother loves John ;. Whereas i n (9b) John an d the pronoun his are co-referent, i n (9a) they ar e not— obviously becaus e i n (9a ) the focused JOH N i s an operator, which , incapable o f referring, ca n only bind a pronominal, provided i t c-commands th e pronominal . Although at LF the focus operator is moved into an A-bar position from which it c-commands his, the c-command conditio n o f binding shoul d be satisfied b y the variable lef t behin d i n its S-structur e position . It is the focus operator tha t is associate d wit h a particular structura l positio n in mos t discourse-configurationa l languages , as well . As fo r it s semanti c role, th e focu s operator operates o n a se t o f contextually relevant entitie s present i n th e domai n o f discourse , and identifie s al l an d onl y the elements o f this se t of which the predicate holds . Consider , fo r example, th e interpretation o f th e following Hungarian sentence : (10) JANO S kapott jelest. John go t A + *It was Joh n wh o go t A+.' We can use (10) in a context or situation which involves a previously establishe d set of persons: fo r instance, the members o f a class. Th e focusing of Jdnos mean s that of the members o f this set John is the only one of whom it is true that he got A+. The intuitiv e content of focusing ha s been formulated semantically in various ways. Szabolcs i (1981) an d (1983 ) describe d th e semanti c function o f focu s i n terms o f firs t orde r predicat e logic , showin g that th e propose d interpretatio n i s equivalent to a Montagovian higher order representation. I n her formulation, th e meaning of (10 ) can be paraphrase d as follows : (11) Fo r ever y x, x go t A+ if an d only i f jc=John. The significanc e of Szabolcsi's wor k o n focus consisted, among other things , in arguing against the views that focus is a stylistic or pragmatic phenomenon (see Rochemont (1978)) , o r tha t it s semanti c contribution ca n b e analyze d as a mer e
16
Discourse Configurational Languages
conversational implicature (see Horn (1981)). A s Szabolcsi (1981 ) demonstrated on Hungaria n material , a focu s operator changes the truth condition s of a sen tence; fo r instance , it change s th e logica l consequence s o f the sentence . Com pare (12a, b) with their counterparts involving focusing in (13a, b). Wherea s (12b ) is a logical consequenc e o f (12a) , (13b ) i s not a logical consequenc e o f (13a): (12) a . Jane s e s Mar i jeles t kapott , John an d Mar y A + go t 'John an d Mary got A+.' b. Jano s jeles t kapott . John A + go t 'John go t A+.* (13) a . JANO S ^S MAR I kapot t jelest . John an d Mar y go t Af 'It was John an d Mary wh o got A+.' b. JANO S kapot t jelest . John go tA + 'It was John wh o got A+.' In accordanc e with th e fac t tha t (13b) i s no t a logica l consequenc e o f (13a) , either sentence can be conjoined with the negation of the other, without any con tradiction arising. I f the two subjects were not focused, such a conjunction would be impossible . (14) Ne m JANOS kapott jelest, hanem JANOS ES MAR I (kapott jelest). not Joh n go t A + bu t Joh n an d Mary go t A + 'It was not John who got A+ but it was John and Mary (who got A+).' Kenesei (1986 , 1993) , whil e maintaining the quantificational approach t o fo cus, argue d agains t Szabolcsi' s formul a in (11), whic h treats focu s a s an opera tor expressin g exhaustive listing, and proposed t o analyz e focus a s a n operato r expressing identification , or , when contrastive, expressing exclusio n b y identifi cation wit h respect t o som e domai n of discours e D. I t i s a n appealin g property of th e quantificationa l approac h t o focu s that i t ca n easil y accoun t fo r th e fac t that th e semanti c operatio n performe d b y th e focu s ha s tw o versions : i t ca n express contras t (tha t is , identificatio n with exclusion) , o r identificatio n only. When th e focus operato r quantifies o n a closed set of individuals, the identifica tion of the subset of which the predicate holds also creates a complement set , of which th e predicate doe s not hold—henc e identification goe s togethe r wit h ex clusion (fo r slightl y differen t view s o n this , se e Szabolcs i (1992 ) an d E . Kis s (1993, n.7)) . I f th e relevan t set o f entitie s quantifie d ove r i s no t a closed set , naturally, complemen t formation cannot take place ; henc e th e subse t identifie d as suc h o f whic h th e predicat e holds cannot be contraste d with a complement subset o f which the predicate doe s no t hold. For example :
Introduction 1
7
(15) A Habor u 6s beke t TOLSZTO J frta . the Wa r an d Peace-AC C Tosto y wrot e 'War an d Peace wa s writte n by TOLSTOY.' In the unmarked case, (15) does not presuppose a closed se t of persons who might have written War and Peace. Consequently , th e focu s operato r identifie s th e writer of War and Peace withou t also excluding particular, contextually relevan t individuals a s it s author . Constituent s no t denoting individuals , fo r instance , adverbials of manner, do not expres s identificatio n by exclusion, either , since — as Szabolcsi (1983) an d (1992) argues—complemen t formation (similar t o other Boolean operations ) i s only meaningful when applied t o individuals (se e (16a)) . The context, however, can individuate predicative constituents, as well—by list ing them . Th e lis t als o provides a closed se t of them ; henc e i n suc h case s (fo r example, i n (16b)) both condition s of 'identificatio n b y exclusion' ar e satisfied . (16) a . Jano s FIGYELMESE N toltott e k i a kerdofvet . John carefull y fille d i n th e questionnair e 'John fille d i n the questionnaire CAREFULLY. ' b. Jano s ne m FIGYELMESEN , haner n LASSA N toltott e k i a John no t carefull y bu t slowl y fille d i n th e kerdofvet. questionnaire. 'John filled in the questionnaire not CAREFULLY but SLOWLY* In the influential work of Rooth (1985), the focus is assumed t o generate a set of alternatives . Thu s the sentenc e JOHN got A+, whe n use d i n a situatio n in volving, say , a clas s consistin g o f fiv e member s (John , Mary , Eve , Susan , an d Peter), expresses that the alternatives exist that John got A+, Mary got A+, Eve gotA+, Susan got A+, and Peter got A+. A problematic feature of Rooth's theor y is tha t i t doe s no t represen t th e exhaustivenes s involve d i n man y instance s o f focusing; unless the sentence contains an explicit only, Rooth' s representatio n will not impl y that of the set of alternatives only th e alternative John gotA+ i s true . The non-quantificationa ! approach to Focusing, treating focus a s a n operato r generating a se t o f alternativ e propositions, argue s agains t th e quantificationa l view, amon g others , b y pointin g ou t tha t th e quantificationa l approac h implie s the movemen t of focus into A-bar position , eve n thoug h Focusin g doe s no t ob serve Subjacency, th e criterion of movement transformations. Thu s in th e sen tence They are looking for an ex-convict with a RED shirt, th e focu s i s th e adjective red, whic h canno t b e moved ou t o f its NP . I n Rooth's approach, th e identification o f focu s doe s no t ru n int o thi s problem . Alternative s o f th e fol lowing type are generated t o th e proposition expresse d i n the sentence: They are looking for an ex-convict with a BLUE shirt; They are looking for an ex-convict with a WHITE shirt; an d the focu s i s th e constituen t (whethe r majo r or minor ) that has different values i n the alternatives. However , on the one hand, it i s not obvious that L F movement i:; subject to Subjacency. O n the othe r hand, in Ian-
18
Discourse Configurational Languages
guages with overt focus movement, Focusing doe s observe Subjacency : onl y the equivalent of the whole constituent an ex-convict with a red shirt ca n be focused , whether o r not red is given contrastiv e stress withi n it. Interpret:atio n must obviously b e derive d th e sam e wa y a s th e interpretatio n o f interrogativ e phrase s with a n embedded WH-word i s derived (for instance , in An ex-convict with what shirt are they looking for?). I t woul d no t b e counter-intuitiv e at al l t o assum e that in suc h cases , the focus/interrogative operato r quantifie s ove r th e individu als bein g looke d for , instea d o f colors , an d t o refe r t o pragmatic s th e fac t tha t the individual s looked for in the domain of discourse are all ex-convicts wearing a shirt . Notic e that th e answe r require d b y a questio n involvin g a n embedded WH wor d cannot be o f the categor y of the WH-word; it must be o f the category of th e whol e WH-move d phrase : (17) a . Milye n szm u inge s fegyenc-e t keresnek ? what colore d shirte d convict-AC C search-the y 'A convict wit h what shir t are they lookin g for? ' b. *Piros . /*Piros-at. red / red-AC C c. Piro s szfn u inge s fegyenc-et . red colore d shirte d convict-AC C 'A convic t with a red shirt. ' d. Piro s szfn u inge s pr o -et . red colore d shirte d -AC C 'One wit h a red shirt.' The fac t tha t th e answe r must be heade d by fegyenc 'convict ' (o r b y a pro co indexed wit h fegyencc in the preceding question) indicates that interrogation op- operates ove r convict s and not colors . In anothe r semanti c approach—elaborated , amon g others , b y vo n Stecho w (1981, 1991), Jacobs (1983) , von Stechow an d Uhmann (1986), and Krifka (1991), (1992)—called the structured meaning approach , the focus feature of a constituent induces th e partitioning of the semantic representation of the sentence into a focus par t an d a background/presupposition part . Fo r instance , the focu s struc ture in (18a ) determine s th e structure d meanin g (18b) : (18) a . [ s John [ vp introduced [ F Bill] t o Sue]] b. (18b) expresses tha t th e individua l wh o has th e propert y o f havin g bee n intro duced t o Sue by John i s Bill. As has been observed , the interpretation of a large se t of operators (only, even, must, not, always, the generic operator , or the superlative) is sensitive to the focusbackground structur e o f it s sentence : th e backgroun d i s understoo d a s th e restrictor of the operator, an d the focus is understood a s its nuclear scope. Con sider the logical paraphrasis of (19), involvin g the universal adverbial quantifier always:
Introduction 1
9
(19) Joh n alway s goes o n vacatio n wit h MARY. always c , 3x(John goe s o n vacation wit h x in c) , John goe s o n vacatio n with Mary In (19 ) always quantifie s over case s i n whic h there i s someon e wh o Joh n goe s on vacatio n with , an d th e sentenc e mean s tha t i n eac h suc h cas e it i s Mar y tha t John goe s o n vacatio n with . The 'structure d meaning' approac h t o focus has led to a non-quantificational , relational vie w of focus. Accordin g to this, the focused constituent itself is never an operator ; i t alway s represents th e nuclea r scop e o f a n operator . I f th e sen tence contains no overt operator (othe r than the focus), an invisible illocutionar y operator i s assumed . Whe n th e focu s appear s t o hav e scop e ove r a n operator , for instance , over a universal quantifier, as in JOHN met everybody, i t is, in fact, the illocutionar y operator tha t has wid e scope . The relational view of focus has been elaborated o n the basis o f languages not containing a structura l focu s position . Fo r speaker s o f language s tha t hav e a particular A-bar position fo r focus, this view is somewhat counterintuitive; in such languages, fo r exampl e i n Hungarian , the focu s slot i s syntacticall y clearl y on e among th e operato r positions . I n an y case , i t i s necessar y tha t th e empirica l material o n whic h the variou s semanti c analyse s of focu s ar e based b e derive d from th e languag e type discusse d i n this volume , a s well . The 'pragmatic ' vie w of focus , claimin g that the focus , a s wel l a s th e topic , are no t par t o f th e truth-conditional , logico-semantic interpretatio n o f th e sen tence, bu t merely expres s the informational value of its logico-semantic content , has also live d on; it has been formulate d in an explicit, partiall y formalize d way, for example , i n the work of Vallduvi (1992). Vallduv i assumes that the interpre tive componen t o f gramma r also contain s a special, non-truth-conditiona l mod ule o f sentenc e interpretation , calle d informatio n packaging . Th e inpu t t o information packaging is the so-called informatio n structure of the sentence, which is non-distinc t fro m it s S-structur e i n discours e configurationa l languages . (I n the cas e o f language s lik e English , th e mappin g o f S-structur e o n informatio n structure i s mor e complex ; i t als o take s th e structur e o f pitch accent s int o con sideration.) Vallduv i describes th e informational role of focus in the framework of Fil e Chang e Semantics : wid e focus is th e par t o f th e sentenc e tha t i s t o b e entered int o the hearer's knowledg e store . A narrow focus may involve a mor e complex operation; for instance, the replacement of an entity in a previously stored proposition (se e als o Huc k an d Na (1990)) . As fo r it s forma l properties , th e focu s operato r i s ofte n morphologicall y marke ; it bears a focus marker in Kikuyu (see (Horvat h (this volume))), Somali, and Q etchua, as well as in Haida (Enric o (1986), an d in preparation), an d ther e are variou s optiona l focu s marker s i n Ol d Romance , too . Korea n onl y mark s contrastive focu s morphologically . I n Somali , ever y matri x clause obligatoril y has a focus. I n Somal i an d Quetchua , an d certai n Haid a dialects , th e focu s i s restricted t o matrix—o r tensed—clauses . I n Greek , ther e i s a somewha t differ ent restriction : th e focu s ca n als o b e locate d i n a n embedde d claus e (provide d the matrix clause has none), but it must have matrix scope. Mor e than one focus
20
Discourse Configurational Languages
per claus e seem s t o be impossible in every languag e except in Korean—conse quently most studie s plac e th e focus in a specifier position . I n Korea n multipl e focus construction s analyze d b y Choe, on e focus move s to [Spec , FP], an d th e rest ar e adjoine d t o it . I n Somali , a n objec t focus—unlik e a subjec t focus—i s associated wit h a resumptive pronoun . In mos t o f th e language s know n t o hav e a structura l focu s position , fo r in stance in Aghem, most Chadi c languages (Horvat h (this volume)), in Basque, Old Romance, Hungarian , Greek , Korea n (i n th e constructio n studie d b y Jo) , Quetchua, and Bulgarian (Rudin (1986)), the focus is located next to the inflected V. Th e exception s includ e Catala n an d Haid a (Enric o (i n preparation)) , wher e the possibility o f non-adjacenc y betwee n th e focu s an d th e V ma y be a consequence o f free scramblin g i n the post-focus sentenc e part , Kikuyu (Horvath (thi s volume)), wher e the focu s is nex t to the complementizer , Somali , and the Korean focu s constructio n described b y Choe. I n Finnish, Greek , an d Korean, th e focus constituen t ca n also remain optionall y i n situ . The close surface relationship betwee n th e focus and the V led Horvath (1981, 1986) t o th e conclusio n tha t th e sourc e o f th e [+F(ocus) ] featur e tha t focuse d constituents assume as a result of Focus Movemen t i s the V. She claimed tha t in languages wit h structural focus th e featur e [+F] i s part o f the featur e matrix o f the V, and can be assigne d b y th e V to a constituent tha t it govern s an d is adja cent to, that is, if the general conditions o f feature assignment ar e observed. Th e focus paramete r o f Universa l Gramma r also provide s a n alternativ e optio n o f association wit h the feature [+F]: language s can choose to associate [+F ] freel y with an y category . Thi s i s wha t happens i n language s wit h focus-in-situ , fo r instance, i n English . Tuller (1992 ) noticed tha t in certain Chadic languages, for instance, i n Tangale, Ngizim, o r Kanakuru , the focu s ca n no t onl y occup y a verb-adjacen t position ; alternatively i t ca n als o sho w u p i n a periphera l position . Sh e conclude d tha t the sourc e o f the [+F ] featur e is inflection — provided i t is verbal , tha t is , i f Vto-Infl movemen t ha s take n place . The n Inf l ca n assig n [+F ] t o a constituen t adjoined t o VP , which i t govern s an d t o whic h i t i s adjacent , or , alternatively, Infl can move up to C in LF, and can assign [+F ] to the constituent in [Spec , CP ] under spec-hea d agreement . In a n approac h becomin g mor e an d mor e wide-spread , th e focu s operato r i s associated with a functional projection o f its own; it occupies th e specifier posi tion o f the projectio n of a focus head . Th e firs t proposal s alon g these line s ap peared i n Uriagerek a (1988 ) (i n whose work , includin g Uriagerek a (1992 ) an d (forthcoming a and b), F is a head associate d no t only with focus , bu t wit h an y operator expressin g a poin t o f view) , i n Cho e (1989) , an d i n Lak a (1990 ) (i n whose theor y I P i s a complement t o a sigm a hea d includin g negation an d em phatic affirmation) . A n influential versio n of this theory was put fort h i n Brody (1990). Brod y (1990 ) claims—followin g Horvat h (1981, 1986)—tha t focu s in terpretation i s du e t o a [+F(ocus) ] featur e assigne d b y th e V . Th e V i s raise d into F, the head position of a functional focu s projection, where it assigns its [+F ] feature t o th e constituen t move d into [Spec , FP] . V movement into F i s forced
Introduction 2
1
by par t A o f th e followin g Focu s Criterion , analogou s t o th e WH-Criterio n proposed b y Ma y (1985 ) an d Rizzi (1990) : (20) A . Th e Spe c o f a n FP mus t contain a [+F]-phrase . B. Al l [+F ] phrase s mus t b e i n a n FP. The focu s projectio n i s claime d t o b e presen t bot h i n th e sentenc e structur e o f languages wit h structura l focus , an d i n tha t o f language s wit h focus-in-situ ; th e difference betwee n th e two types is derived fro m the parametrization o f the level of representatio n a t whic h th e Focu s Criterio n applies . I f i n a language condi tion A of th e Focus Criterio n mus t be observe d a t S-structure, the languag e will have structura l focus; i f i t i s only checke d a t LF , the languag e wil l hav e focus in-situ. Horvath (thi s volume ) give s a criticism o f the view that th e 'structura l focus ' versus 'focus-in-situ ' variatio n derive s fro m a parametric variatio n i n th e leve l of applicatio n o f th e Focu s Criterion , showing o n th e basi s o f comparativ e evi dence (taken , in part, from Tulle r (1992) ) that it canno t account fo r th e rang e of variation of S- structure focus positions attested within th e set of designated Focu s languages. Sh e argues that under the minimal assumption that there is a syntac tic featur e [+F ] whic h constituent s mus t receiv e t o b e interprete d a s identificational focus , the observed range of S-structure variation with respect t o focus position falls out o f independently motivated parameters o f feature assignment. Thes e parameters allo w th e followin g mode s o f associatio n wit h th e fea ture [+F] : [+F ] ca n b e assigne d b y a functiona l head, th e categor y o f whic h (typically Inf l o r C ) may var y acros s languages . Th e featur e [+F ] ca n be trans mitted b y Infl/C int o a position governe d b y Infl/C, o r it can be assigned b y Infl / C to [Spec , IP/CP ] under spec-hea d agreement . I n language s with focus-in-sit u no functional head carries th e featur e [+F] ; association with [+F] is free. Horvat h predicts, an d demonstrates, a ful l parallelis m between th e kind s of variatio n ob servable with respect t o case assignment , and [+F ] assignmen t acros s languages . In thi s volume , Choe , Orti z d e Urbina , Tsimpli , Uriagereka , an d Muysken s present focu s theories involvin g a separat e functiona l focus projection . In Choe' s approach , th e feature [+F ] i s assigne d i n D-structure, an d it can b e assigned (a t least i n Korean) to more tha n one constituent per clause . [+F ] con stituents ar e licensed i f the y agre e with a [+F ] head ; therefore , the y mus t mov e to [Spec , FP] , o r be adjoine d to it—eithe r i n synta x or in LF. The featur e [+F ] i s carrie d b y Inf l i n the theor y o f Orti z d e Urbina (thi s vol ume), too . Jus t a s interrogative clause s licens e a [+WH ] Infl , emphatic clause s license a [+F ] Infl . Th e Operato r Criterio n ( a more genera l versio n o f the WH Criterion) require s tha t a n operator occupyin g scope positio n b e i n a spec-hea d relation wit h a head marke d for the relevant feature. Sinc e i n Basque ther e is V Movement to Infl , an d (V+Infl ) Movemen t t o C , the focu s lands i n [Spec , CP] . Motivated by th e theor y o f parametrization of Chomsky (1991) , in whic h pa rameters ar e associated with functional categorie s rather than levels o f represen tation, Tsimpli (thi s volume) modifies Brody's theory (1990) alon g the following lines. Language s wit h obligatory Focu s Movement , lik e Hungarian , languages
22
Discourse Configurational Languages
with optional Focu s Movement , lik e Greek , an d presumably als o languages wit h no syntacti c Focu s Movement , lik e English , al l hav e a n F projectio n i n thei r sentence structures ; the y onl y diffe r i n th e valu e o f the F featur e carried b y th e head o f their FP projection. I n Hungarian the head of FP is always marke d [+F] ; in English, it is presumably always [-F], wherea s in Greek it can have either value . In thi s theory , th e [+F ] featur e i s freel y assigne d i n D-structur e i n al l languag e types. Th e over t Focu s Movemen t o f a constituent marked [+F ] i s triggere d i n case th e hea d o f F P i s als o marke d [+F]—b y a Focus Criterio n operativ e a t S structure, which requires that a [+F] head be in spec-head agreemen t wit h a [+F ] operator. The theorie s o f [+F ] assignmen t surveye d abov e ca n captur e th e proces s o f Focusing i n most o f the language s described i n the volume : i n Chadic, Somali , Basque, Ol d Romance , Hungarian , Greek , an d Quetchua , an d the y ca n als o ac count for Haida (Enrico (in preparation)), Bulgarian (Rudin (1986), or the Mayan languages (Aisse n (1991)) . I n thes e languages , th e constituen t functionin g as a focus operato r occupies a designated A' position, wher e it c-commands it s scope. In thre e o f th e language s examined: i n Finnish , Catalan , an d i n a particular construction i n Korean , something els e i s goin g on . Finnish actuall y can b e see n a s a parametric varian t of th e canonica l typ e o f discourse-configurational language s with propert y B . I n Finnish , a s show n b y Vilkuna, focu s constituent s ca n eithe r b e move d int o a designate d A ' position : [Spec, CP] , o r they ca n b e lef t i n situ . Thi s A ' position , however , i s associate d not wit h th e featur e [+F] , bu t wit h th e featur e [+Contrastive] ; henc e i t i s ope n not onl y fo r a focu s bu t als o fo r a contrastiv e topic . I f functiona l head s can , indeed, conve y suc h affectiv e features a s [+WH ] o r [+F] , the n i t is plausibl e t o assume tha t th e se t o f feature s tha t ca n b e associate d wit h C universall y als o includes [+Contrastive] . I n Finnish, th e [+Contrastive ] positio n ([Spec , CP] ) i s outside th e topic slo t ([Spec , IP]). Thi s property ma y be related t o the fac t tha t the Contrastiv e position ca n als o host a topicalized constituent. Catalan seem s t o represen t a radicall y differen t case . A s Vallduv i argues , Focusing i n Catala n does no t involv e any focu s movement o r an y [+F ] assign ment. I n th e Catala n sentence , th e non-focu s constituent s ar e moved , an d th e focus constituent s are lef t i n situ . Everythin g that is not ne w information i s dis located fro m th e IP . I f a singl e constituen t remains , i t i s t o b e interprete d a s narrow focus , i.e. , a s a focus operator . The Korean construction tha t Jo (this volume) describes, whil e different from the canonica l cas e o f Focusing , does no t represen t a n isolate d pattern . I n th e Korean constructio n i n question , a s wel l a s i n Turkish , Armenia n (se e Comri e (1984)), or Hungarian, th e focu s is in complementary distributio n wit h the con stituent tha t i s closest t o the V in the VP . Thi s constituen t is non-referential ; i t has n o theta-role; it form s a complex predicat e wit h the V. I n Korea n an d Hun garian, eve n it s syntacti c movemen t possibilitie s ar e limited ; tha t is , i t display s several propertie s typical of incorporate d constituents. J o argue s tha t th e posi tion o f this constituent, a V-adjacent, V'-dominate d A' position, when vacated, is allowed by principle s of Universal Grammar to host an operator. I n Hungarian, where the incorporated constituent is at the lef t edg e of the VP, alternative analy-
Introduction 2
3
ses are als o possibl e (se e Horvat h and E. Kis s i n this volume) , fo r instance, th e focus an d the incorporated constituen t ca n be assigne d to separate , adjacen t positions (on e outsid e th e VP , the othe r insid e th e VP) , an d th e simultaneou s fill ing of the two positions can be ; blocked by some principle or filter. In language s where this non-argumen t position i s neste d i n the VP, the possibility o f alterna tive explanation s i s les s obvious . I n additio n t o Korean , Aghem, i n whic h th e focus positio n i s between th e V and the direc t object , migh t als o represen t thi s type. I n the focus construction described by Jo the focus operator i s not in scope position, s o a t LF i t must undergo Operator Movement . Jo's theor y raises an interesting theoretical possibility. Namely , Focusin g an d incorporation shar e importan t properties acros s languages . Th e adjacenc y be tween th e focu s an d th e inflecte d V i s stricte r i n mos t language s tha n an y ver sion o f featur e assignment—whethe r unde r government , o r unde r spec-hea d agreement—would justify. I t appears that the strictness of this adjacenc y requirement i s onl y parallele d b y the ; adjacenc y conditio n o f incorporation . I t i s als o remarkable tha t bot h Focusin g an d incorporatio n ar e typicall y license d b y th e same category : th e inflected V. Focu s Movemen t involves incorporation o n th e phonological leve l i n man y languages ; the focuse d constituen t doe s no t simpl y assume heavy stress corresponding to its semantic weight; it also deletes th e stres s of the adjacen t inflected V; that is, the focus an d the V form a single phonologi cal word . Th e questio n tha t arises is naturally whether the similaritie s betwee n Focusing an d incorporatio n ar e accidental , or Focusin g i s incorporation . I f th e feature [+F ] i s inherently associated wit h the (inflected ) V, then it is not implau sible t o assum e tha t featur e sharing can als o tak e plac e vi a incorporation . It i s generall y believed , an d in Horvath (1986 ) i t is als o state d i n the for m of a universal principle, that interrogative WH-phrases shar e the syntactic behavio r of foci . Thi s assumption is basically confirme d i n canonical discourse-configu rational language s wit h propert y B , i n whic h th e focu s occupie s a scop e posi tion. Interrogativ e WH-phrases mus t land in the focus position i n Somali, Chadic , Aghem, Basque, Hungarian, Haida, Omaha, Quetchua, Korean, and in Greek main clauses. WH-phrase s an d foci giv e the illusion o f sharing the sam e landin g sit e in Bulgarian, too, eve n though , as Rudin (1986 ) argues , WH-phrases i n fact oc cupy [Spec , CP], while foci ar e adjoine d to IP. WH-phrases , whethe r interroga tive or relative, land in the position associate d wit h the feature [+Contrastive ] i n Finnish. In Catalan , wher e the focus > operato r does no t occup y a scop e position, WH phrases nevertheles s d o undergo WH-Movement t o scop e position . Summarizing th e informatio n provide d b y th e paper s o f thi s volum e o n typ e B discours e configurationality : i n languages with property B the focus operato r occupies—or, a t least , ma y occupy—a n A-ba r positio n a t S-Structure . I t typi cally occupie s a n invarian t A-bar position ; bu t thi s i s no t necessaril y s o (cf . Tangale, Ngizim). Th e exact locatio n of the landing site o f Focus Movemen t i s subject to parametric variation: i t can be [Spec, VP], [Spec, IP], [Spec, FP], [Spec , CP], a VP-adjoined position , a n FP-adjoined position , or even a n A-bar position under V .
24
Discourse Configurational Languages
A focu s operato r move d int o VP-externa l positio n c-command s it s scope ; a focus operato r occupyin g an A-bar positio n unde r V, o n th e othe r hand , obvi ously does not c-command it ; it needs further LF-movement . I n some languages , the focu s operato r i s require d t o hav e matri x scope; i n others , i t ca n als o hav e embedded scope . Th e presenc e o f a focus operato r i s typicall y optional , bu t i t may als o b e a n obligatory elemen t o f sentenc e structure . In the canonical typ e of languages with property B , a constituent assumes th e feature [+F ] b y movement int o the designated A-bar position. Th e source , or , at least, licenser , o f th e featur e [+F ] i s presumabl y the hea d I , V+I, F , or C . Th e [+F] feature of a focused phrase is assigned—or licensed—under spec-head agree ment, o r i t is assigne d under government and adjacency . Th e featur e [+F ] usu ally has no morphological reality (but it can also be spelled ou t in the form of an inflectional morpheme , a s is the case i n Berber accordin g t o Tsimpli). Th e rel evant featur e can be , instea d o f [+Focus] , [+Contrastive] . In languages displaying S-structure Focus Movemen t int o scop e position , interrogative WH-phrases shar e the landing site o f foci. I t still remain s a questio n whether interrogativ e WH-phrases als o share th e feature [+F]—give n that many languages allo w mor e tha n on e WH-phrase bu t onl y on e focu s pe r claus e (Ko rean, o n the othe r hand , allows the overt, syntacti c movement o f mor e tha n on e focus bu t onl y a singl e WH-phrase per clause).
References Agouraki, G . (1990 ) "O n th e Projectio n of Maxima l Categories : th e Cas e o f C P an d F P in Moder n Greek, " UCL Working Papers in Linguistics 2, 183-200 , Universit y College London . Aissen, J . (1991 ) "Topi c an d Focus i n Mayan, " Language 68 , 43-80. Awobuluyi, O. (1978) "Focu s Constructions a s Noun Phrases," Linguistic Analysis 4 , 93 114. Brassai, S . (1860, 1863-65 ) "A magya r mondat," Magyar Akademiai Ertesitff. A Nyelves Szeptudomdnyi Osztdly Kozldnye 1 , 179-399; 3 , 3-128, 173-409 . Brody, M. (1990 ) "Som e Remarks on the Focus Fiel d i n Hungarian, " UCL Working Papers in Linguistics 2, 201-226, Universit y Colleg e London . Calabrese, A. (1987 ) "Focu s Structur e in Berber : A Comparative Analysis with Italian, " in M. Guerssel and K. Hale, eds. , Studies in Berber Syntax. Lexicon Project Working Papers 14. Cente r fo r Cognitiv e Science , MIT , Cambridge , Massachusetts . Carlson, G . (1978 ) Reference to Kinds in English, Garland, Ne w York. Choe, H . S . (1989 ) "Restructurin g Parameter s an d Scramblin g i n Korea n an d Hungar ian," i n L. Marac z an d P . Muysken, eds., Configurationality,, Foris, Dordrecht. Chomsky, N. (1976) "Condition s o n Rules of Grammar," Linguistic Analysis 2, 303-352. Chomsky, N . (1981) Lectures on Government and Binding, Foris, Dordrecht . Chomsky, N . (1986 ) Barriers, MI T Press, Cambridge , Massachusetts . Chomsky, N. (1991) "Some Notes o n Economy o f Derivation and Representation, " i n R. Freidin, ed. , Principles and Parameters in Comparative Grammar, MI T Press , Cambridge, Massachusetts.
Introduction 2
5
Clements, G. N. (1984) "Binding Domains in Kikuyu," Studies in the Linguistic Sciences 14, 2, 37-57. Comrie, B. (1984 ) "Som e Propertie s of Focus i n Moder n Eastern Armenian," Annual of Armenian Linguistics 5 , 1-21 . Diesing, M. (1992 ) Indefinites, MIT Press , Cambridge , Massachusetts . Eguzkitza, A. (1987 ) Topics on the Syntax of Basque and Romance, Indian a University Linguistics Club . England, N. (1991 ) "Change s i n Basic Wor d Order in Mayan Languages," International Journal of American Linguistics 57 , 446-486. Enrico, J . (1986 ) "Wor d Order , Focus , an d Topi c i n Haida, " International Journal of American Linguistics 52 , 91-123. Enrico, J . (i n preparation ) Haida. Grammar, unpublished manuscript, Queen Charlott e Islands Museum , Skidegate, British Columbia. Erguvanli, E. E . (1984 ) The Function of Word Order in Turkish Grammar, University of California Press , Berkeley. Farmer, A . K . (1980 ) On the Interaction of Morphology and Syntax, Doctora l disserta tion, MIT, Cambridge , Massachusetts. Gambhir, V . (1981) Syntactic Restrictions and Discourse Functions of Word Order in Standard Hindi, Doctora l dissertation , University of Pennsylvania. Giv6n, T. (1975) "Focus an d th e Scope o f Assertion. Some Bantu Evidence," Studies in African Linguistics 6, 185-205 . Hajic'ova', E . (1983 ) "Topi c an d Focus," Theoretical Linguistics 10 , 268-276. Heyer, G. (1985) "Generi c Descriptions, Default Reasoning, and Typicality," Theoretical Linguistics 12 , 33-72. Horn, L. (1981 ) "Exhaustivenes s an d the Semantic s of Clefts, " Proceedings of the llth Annual Meeting of NELS, MIT , Cambridge, Massachusetts, 125-142 . Horvath, J. (1976 ) "Focu s i n Hungarian and the X-bar Notation," Linguistic Analysis 2, 175-197. Horvath, J . (1981 ) Aspects of Hungarian Syntax and the Theory of Grammar. Doctora l dissertation, UCLA. Horvath, J. (1986) FOCUS in the Theory of Grammar and the Syntax of Hungarian, Foris , Dordrecht. Huck, G . J., an d Y. Na (1990) "Extrapositio n and Focus," Language 66 , 51-77. Jacobs, J . (1983 ) Fokus und Skalen. Zur Syntax und Semantik von Gradpartikeln im Deutschen. Niemeyer, Tubingen. Jo, M.-J . (1991) Fixed Word Order and the Theory of the Pre-Verbal Focus Position in Korean, Doctora l dissertation, University o f Washington, Seattle, Washington. Kenesei, I. (1986 ) "O n th e Logi c of Focus," i n W. Abraham and S . d e Meij, eds. , Topic, Focus and Configurationality, Joh n Benjamins, Amsterdam, Kenesei, I . (1993 ) "O n th e Synta x of Focus, " unpublishe d manuscript, Universit y of Delaware, Newark, and Universit y of Szeged , Hungary. King, T. H. (1993) "VP Internal Subjects in Russian," paper presented o n the 2n d workshop o n Forma l Approache s to Slavi c Linguistics , MIT , Cambridge , Massachu setts, May 1993 . E. Kiss, K . (1977) "Topic and Focus i n Hungarian Syntax," Montreal Working Papers in Linguistics 8 , 1-42. E. Kiss, K. (1981) "Structura l Relations in Hungarian, a "Free" Word Order Language," Linguistic Inquiry 12 , 18'i-215 . E. Kiss , K. (1987 ) Configurationality in Hungarian, Reidel , Dordrecht .
26
Discourse Configurational Languages
E. Kiss (199la) "Logical Structur e in Syntactic Structure: the Case of Hungarian," in C.T. J. Huang and R. May, eds., Logical Structure and Linguistic Structure, Kluwer , Dordrecht. E. Kiss , K . (1991b ) "A n Argumen t for Movement, " i n H . Haide r an d K . Netter , eds. , Representation and Derivation in the Theory of Grammar, Kluwer , Dordrecht . E. Kiss , K . (1993 ) "Genericity , Predication , an d Focus," unpublishe d manuscript , Linguistic Institut e of the Hungarian Academy of Sciences , Budapest . Koopman, H. , an d D. Sportiche (1990 ) "Subjects, " unpublished manuscript, UCLA. Krifka, M . (1991) " A Compositional Semantic s for Multiple Focus Constructions, " Proceedings of Semantics and Linguistic Theory (SALT) I, Cornell Working Papers 11. Krifka, M . (1992 ) " A Framewor k for Focus-Sensitiv e Quantification," in C . Barke r and D. Dowty , eds., SALT II. Proceedings from the Second Conference on Semantics and Linguistic Theory. Working Papers in Linguistics 40, Ohi o Stat e University, Columbus, 215-236. Kuroda, S.-Y . (1972-73) "Th e Categorica l an d th e Theti c Judgement, " Foundations of Language 9, 153-185. Laka, I. (1990) Negation in Syntax. On the Nature of Functional Categories and Projections, Doctora l dissertation, MIT, Cambridge, Massachusetts . Lasnik, H. , an d M. Sait o (1992 ) Move Alpha, MI T Press, Cambridge , Massachusetts. Lecarme, J. (1992) "Focus en Somali: syntax et interpretation," Linguistique Africaine 1. Marty, A. (1918 ) Gesammelte Schriften. II . Band , 1 . Abteilung, Max Niemeyer , Halle . Marty, A. (1965 ) Psyche und Sprachstruktur, Verla g A. Francke, Bern . May, R . (1985 ) Logical Form: Its Structure and Derivation, MI T Press , Cambridge , Massachusetts. Ortiz d e Urbina , J . (1991 ) Parameters in the Grammar of Basque, Foris , Dordrecht . Ouhalla, J . (1991 ) "Focussin g i n Berbe r and Circassia n and th e V 2 Phenomenon, " un published manuscript , University College London. Pesetsky, D. (1989) "Language Particular Processes an d the Earliness Principle," unpublished manuscript , MIT, Cambridge , Massachusetts. Primus, B . (1992 ) "Th e Rol e o f Grammatica l Relations i n Wor d Orde r Universals, " EUROTYP Working Papers, ESF , Strasbourg. Rebuschi, G . (1989 ) "I s ther e a V P i n Basque?, " i n L . Marac z an d P . Muysken, eds., Configurationality. The Typology of Asymmetries, 85-116 , Foris, Dordrecht . Rijk, R.P.G . d e (1978 ) "Topi c Fronting , Focu s Positionin g and th e Natur e of th e Verb Phrase in Basque, " i n F. Jansen, ed., Studies on Fronting, Pete r d e Ridde r Press , Lisse. Rizzi, L . (1991 ) Relativized Minimality, MI T Press , Cambridge , Massachusetts. Rochemont, M. (1978 ) A Theory of Stylistic Rules in English, Garland Press, New York. Rooth, M. (1985) Association with Focus, Doctoral dissertation, University of Massachusetts, Amherst. Rothstein, S . (1983 ) The Syntactic Forms of Predication, Doctora l dissertation , MIT , Cambridge, Massachusetts. Rudin, C . (1986 ) Aspects of Bulgarian Syntax: Complementizers and Wh Constructions, Slavica, Columbus. Rudin, C. (1992) "Topic and Focus in Bulgarian," to appear in Acta Linguistica Academiae Scientiarum Hungaricae. Schachter, P . (1973) "Focu s an d Relativization, " Language 49 , 19-46 . Sgall, P., ed. (1984) Contributions to Functional Syntax, Semantics, and Language Comprehension, Academia, Praha.
Introduction 2
7
Sgall, P., and E. HajiCovd (1973 ) Topic, Focus, and Generative Semantics, Scriptor Verlag, Kronberg Taunus. Stechow, A. von (1981) "Topic , Focus, an d Relevance," in W. Klein and W. Levelt, eds. , Crossing the Boundaries in Linguistics, Reidel , Dordrecht . Stechow, A. von (1991 ) "Curren t Issues in the Theory o f Focus," in A. von Stecho w and D. Wunderlich , eds., Semantics. An International Handbook of Contemporary Research, D e Gruyter, Berlin. Stechow, A . von , an d S . Uhman n (1986) "Som e Remark s o n Focu s Projection, " i n W. Abraham an d S . d e Meij , eds. , Topic, Focus, and Configurationality,, John Benjamins, Amsterdam . Stowell, T . (1981) Origins of Phrase Structure, Doctora l Dissertation , MIT , Cambridge , Massachusetts. Szabolcsi, A. (1981 ) "Th e Semantic s of Topic-Focus Articulation," in J. Groenendij k et al., eds. , Formal Methods in the Study of Language 2 , 503-540 , Matematisc h Centrum, Amsterdam. Szabolcsi, A. (1983) "Focussin g Properties , o r the Trap of First Order," Theoretical Linguistics 10 , 125-145. Szabolcsi, A . (1992) "Wea k Island s and Scope," in C. Barker and D. Dowty, eds., SALT II. Proceedings from the Second Conference on Semantics and Linguistic Theory, Working Papers in Linguistics 40, Ohio Stat e University, Columbus , 407-436 . Tsimpli, I . M . (1990 ) "Th e Claus e Structur e an d Wor d Orde r o f Moder n Greek, " UCL Working Papers in Linguistics 2 , 226-258, University College London . Tuller, L . (1992 ) "Th e Synta x of Postverba l Focu s Construction s in Chadic, " Natural Language and Linguistic Theory 10 , 303-334. Ulrich, Miorit a (1985 ) Thetisch und kategorisch, Narr , Tubingen. Uriagereka, J. (1988 ) On Government, Doctora l dissertation , University of Connecticut, Storrs. Uriagereka, J . (1992 ) "Th e Synta x o f Movemen t in Basque, " in J . Lakarr a an d J . Orti z de Urbina , eds, Syntactic Theory and Basque Syntax, ASJU, Donostia . Uriagereka, J. (forthcomin g a) "Aspects o f Clitic Placement i n Western Romance," Linguistic Inquiry. Uriagereka, J. (forthcoming b) "Government Restrictions and Basque Movements," Natural Language and Linguistic Theory. Vallduvi, E. (1992 ) The Informational Component, Garland , New York. Vallduvi, E . (1993 ) "Catala n a s VOS: Evidenc e fro m Informatio n Packaging," i n W . J. Ashby e t al. , eds. , Linguistic Perspectives in the Romance Languages, Joh n Benjamins, Amsterdam. Vilkuna, M . (1989 ) Free Word Order in Finnish, Suomalaise n Kirjallisuude n Seura, Helsinki. Wallace, W. D. (1985) Subjects and Subjecthood in Nepali: An Analysis of Nepali Clause Structure and Its Challenges to Relational Grammar and Government and Binding, Doctora l dissertation , University of Illinois, Urbana-Champaign. Walters, J. (1979 ) "Focu s i n Aghem," in L. Hyman, ed., Aghem Grammatical Structure, Southern California Occasional Papers in Linguistics 7 . Williams, E . (1980 ) "Predication, " Linguistic Inquiry 11 , 203-238.
2 STRUCTURAL FOCUS, STRUCTURAL CASE, AND THE NOTION OF FEATURE ASSIGNMENT JULIA HORVATH Tel-Aviv University
1. Introduction Cross-linguistic variatio n wit h respect t o th e syntacti c position(s) occupie d b y arguments, e.g. by subjects, within clauses has traditionally been in the center of research withi n the framewor k of G B theory , an d ha s give n ris e t o importan t developments in relation to the parametrization of UG. (see e.g. Chomsky (1981), Travis (1984), Bore r (1986), Koopma n and Sportiche (1990)). Likewise , varia tion among languages with respect to the S-structure position(s) of Wh-phrases— in particular , in situ vs . move d int o [SPEC , CP]—ha s receive d considerable attention (se e e.g. Ma y (1985), Chomsk y (1986), Kuroda (1988), Rizz i (1991)). A further , fundamentall y similar, instanc e o f cross-linguisti c variation , involv ing the range of S-structure positions for focus—i.e., for constituents interprete d as the (identificational ) FOCU S of their clause—however ha s gon e largely unrecognized until recently.1 The first proposal involving restrictions an d languageparticular difference s i n th e syntacti c distributio n o f FOCU S an d thei r incorporation int o th e theor y o f GB appeared i n Horvat h (1981 , 1986) ; only a couple of recent papers take up this topic of obvious theoretical significance , such as Tuller (1988, 1992), Brody (1990), proposing modification s and additions (t o be discussed below ) for the account of the syntax of Focus i n the former works. The present study will undertake further investigatio n of the sources and theoretical implications of the cross-linguistic variation manifested b y th e syntax of Focus, more precisely, the existence and specific location(s ) of designated—i.e. syntactically/structurally limited—Focu s positions in a variety of languages. 28
Structural Focus, Structural Case, and The Notion of Feature-Assignment 2
9
Our discussio n wil l evaluate and elaborate th e hypothesis put forwar d origi nally b y Horvat h (1981, 1986 ) accordin g t o whic h (a ) there is a fundamental parallelism between the nature and origin of structural Focus positions (a s attested in languages such as Hungarian, Aghera, Basque) and of structural Caseassignment position s (as e.g. those of nominative and objective Case in languages such as English), and (b) the parallelisms observed are to be captured by the claim that UG has a syntactic feature [-fFOCUS] , whic h in terms o f its formal prop erties—namely, it s mode(s) , leve l an d condition(s ) of assignment—i s o n a pa r with (structural ) Case features . Naturally , the substance , an d consequently, th e function o f the [+FOCUS] feature postulated is distinct from, thoug h still to some extent parallel to, that of Case features. While the contribution of Case involves the licensin g o f arguments, more precisely o f A-chains, the [+FQCUS ] featur e is claimed t o turn a constituent that (unlike wh-phrases) intrinsically has no operator-element int o one that eventually ca n play the role of a variable-binding operator (on a par with wh-phrases), i.e. , i n this framework , [+FOCUS ] i s the feature tha t license s th e A-ba r chai n whic h at leas t sinc e Chomsk y (1976 ) i s commonly assume d to constitut e the LF-representation o f Focus. Strikin g sup port fo r thi s view of th e contribution of th e [+FOCUS ] feature postulated her e comes fro m th e contrasting behavior of focused vs. topicalized phrase s with respect to weak crossover, discussed recently in Lasnik an d Stowell (1991); thei r study shows that Focus phrases but not topicalized phrases ac t at LF like phrase s containing true quantifiers, even though superficially both look lik e plain, nonquantified phrase s occupying A-bar positions.2 Based on the above assumptions, Horvath (1981,1986) formulated a "FOCUSParameter" fo r UG , providin g th e followin g tw o option s fo r th e statu s o f th e [-fFOCUS] feature : (1) FOCUS-Parameter : a. [+FOCUS] : a feature associate d freel y wit h any category—derivin g the English-typ e languages, i.e. Focus in situ b. th e "grammaticalized" version of the [+FOCUS] feature: an intrinsic part of the feature-matrix o f a single category, namely , V—meant to derive the Hungarian-type, structurally limited, instantiation s of Focus This FOCUS-Parameter, in conjunction wit h an independently motivated "Locality Condition o n Feature-Assignment" requirin g that any feature-assigning categor y govern and be adjacent t o the recipient of the feature, was claimed to be abl e to account fo r th e distinctio n between the tw o type s of S-structur e realizations o f Focus (Englis h vs. Hungarian), as well as for the particular S-structure positions of Focu s attested in a couple of additiona l "designated" Focu s languages . In subsequent work, Brody (1990) proposes a modification o f the above analy sis, based crucially on the introduction of a functional categor y F, projecting into a Focu s Phras e (FP) . Whil e Brody's theor y of Focu s retain s Horvath' s (1981 , 1986) assumption s regarding (a) th e existenc e an d assignment-propertie s o f a syntactic featur e [+FOCUS], a s wel l as (b ) th e categor y V being th e sourc e of
30
Discourse Configurational Languages
the [+FOCUS ] featur e in the Hungarian-type languages, it incorporates in addi tion t o these the following pair of well-formedness conditions—paralle l to those familiar from the analysis of +WH CP's, i.e . to the so-called "Wh-Criterion " (May (1985), Rizz i (1991)) : (2) a . A t S-structur e an d LF th e Spe c o f a n FP must contai n a +f-phrase . b. A t LF all +f-phrases mus t be in an FP.3 (Brod y (1990 , (lOa, b)) Notice no w tha t th e optio n o f havin g i n ou r theor y o f Focu s som e well formedness conditio n like (2) , henceforth referred t o as the "FOCUS-Criterion," immediately raise s th e following new question: I s the cross-linguistic variatio n observed wit h respect t o th e S-structur e position s o f Focus constituent s du e t o the parametrization of the status/origin of the feature [+FOCUS]—as implied by the FOCUS-Parameter in (1)—or to the parametrizatio n o f a FOCUS-Criterion such as (2), or possibly, to a combination of these? In Brody's (1990) analysis , it is the parametrization of his FOCUS-Criterion, in particular, of condition (2a) t o apply onl y a t L F vs . t o appl y bot h a t LF an d a t S-structure , tha t is take n t o b e the sourc e o f the contras t betwee n the S-structur e of English vs . Hungaria n Focus, whil e the way the feature [+FOCUS] i s assigned t o constituents is assume d to be essentially invariant. 4 In thi s study , w e will critically compar e an d reevaluate th e various assumptions in the abov e two theories o f the synta x of Focus, wit h particular reference to th e question o f whic h typ e o f parameter is abl e t o provide a more adequate account of a fuller rang e o f cross-linguistic, a s wel l as language-internal , varia tion regardin g S-structure positions fo r Focu s element s foun d whe n w e g o be yond th e limite d Englis h vs . Hungaria n contrast. I n additio n t o evidenc e fro m Hungarian, the empirical material for broadening the typology of structurally limited Focus positions will b e drawn fro m a variety of African languages , namely from Kikuyu , Aghem (Bantu), and some of the Chadic languages studied in Tuller (1992), such as Western Bade and Kanakuru. The majo r conclusion s emergin g fro m ou r discussio n wil l b e th e following : (a) Th e mechanism of [+FOCUS] assignmen t by an X° category i s needed any way, fo r the cas e of Hungaria n in contrast t o that of e.g. English , als o i n a framework incorporatin g the (parametrized) FOCUS-Criterion (as admitte d also i n Brody 1990) . (b) Parametrizatio n o f th e FOCUS-Criterio n (a s e.g . i n Brod y 1990 ) make s incorrect prediction s wit h respect t o the rang e of variation exhibited b y S structure realizations of Focus in different designate d Focus languages. (c) Th e parametrization of the feature [+FOCUS] an d its assignment along line s matching those motivated independently by the stud y o f (structural) nominative Cas e doe s see m t o provid e a constraine d ye t sufficientl y flexibl e account o f th e S-structur e syntax of Focus .
Structural Focus, Structural Case, and The Notion of Feature-Assignment 3
1
Hence ou r proposal wil l be to capture th e S-structur e variation involvin g Focus b y limitin g parametric variatio n t o th e singl e proces s o f th e associatio n o f the [+FOCUS ] featur e wit h categories. 5
2. [+FOCUS] Assignment and/or the FOCUS-Criterion 2.1. Feature-Assignment and Focus in Hungarian Given th e recen t proliferatio n o f new functional categories headin g full-fledge d maximal projections in analyses of clause structur e (starting with Pollock (1989)), it is indeed not unreasonable to entertain the following two hypotheses, a s in fac t has bee n don e by Brod y (1990) : (a) Grammar s have a functional categor y F, projecting an FP, where F is a formative tha t corresponds t o "FOCUS." (b) Th e source of the contrast between Focus-in-situ language s (a s English) and designated Focus-positio n language s (as Hungarian) is the parametrizatio n of the FOCUS-Criterion (se e 2) , specifically, o f (2a) above , t o apply eithe r only a t LF or als o a t S-structure . These tw o proposal s indee d seem , a t firs t glance , t o accoun t fo r th e relevan t English/Hungarian contrast elegantly, on the analogy of Wh-in-situ vs . overt Wh movement languages. To illustrate the English/Hungarian contrast under discus sion, conside r th e followin g set s o f example s (wher e th e Focu s constituen t i s capitalized): (3) a . I threw awa y the newspaper. b. I threw awa y THE NEWSPAPER. (='It's th e newspaper tha t I threw away' ) (4) a . Eldobta m a z ujsagot . away-threw-I th e newspaper-ac e 'I thre w awa y the newspaper ' b. A Z UJSAGO T dobta m el . THE NEWSPAPER-AC C threw-I awa y 'It's th e newspaper that I thre w away' c. *Eldobta m A Z UJSAGO T away-threw-I TH E NEWSPAPER-AC C d. *A Z UJSAGO T eldobta m THE NEWSPAPER-AC C away-threw- I The well-known descriptiv e generalization s behin d suc h dat a ar e (a) that un like English , Hungarian does no t permit Focus i n situ a t S-structure, an d (b) that the designate d Focus positio n o f Hungaria n is left-adjacen t t o th e verb . I t i s b y now als o widel y accepted tha t this S-structur e Focus positio n i s a n A-ba r posi tion whic h c-command s the ver b an d it s argument s followin g i t (se e th e argu -
32
Discourse Configurational Languages
ments of Farkas (1986) , Horvat h (1986 , pp. 147-14 8 n. 40)). Thus, th e S-struc ture representatio n propose d b y Brod y (1990 ) fo r Hungaria n base d o n hi s F P hypothesis an d FOCUS-Criterion i s consisten t wit h these generalizations, i f w e make th e no t unmotivate d additional assumptio n tha t th e inflecte d V get s pre posed int o F. 6
While the contrast between the above S-structur e Focus configuratio n and in-situ Focus woul d indee d follo w straightforwardl y fro m th e parametrizatio n o f th e application o f (2a ) (S-structur e vs . LF) , ther e obviousl y i s mor e t o th e stor y o f the designate d Focu s positio n o f Hungarian . Note th e followin g tw o question s raised b y the descriptive generalization s characterizin g the S-structur e syntax of Focus i n Hungarian: 1. Wh y doe s th e ver b nee d t o mov e into th e (alleged ) F node ? 2. Wh y is strict adjacency required between this node an d the Focus constitu ent i n it s SPEC ? The clai m tha t th e (finite ) verb indeed need s t o prepos e i n Focu s construction s is motivate d b y th e relativ e orde r betwee n a variet y o f normall y preverba l par ticles (an d othe r incorporate d verba l modifiers) and the verb . (Se e e.g . th e con trast betwee n (4b ) an d (4d ) above , an d th e discussio n i n Brod y (1990). ) Th e distinct additiona l requiremen t o f adjacenc y betwee n th e propose d V an d th e Focus positio n i s illustrate d belo w (cf . (4b)) : (6) *A Z UJSAGO T valoszinuleg/szerintii k dobta m el . THE NEWSPAPER-AC C probably/according-to-the m threw- I awa y Crucially, neither the necessity of V-movement in Focus construction s nor the above adjacency requirement follow simpl y from th e SPEC-head relatio n betwee n the Focuse d elemen t an d F require d b y th e FOCUS-Criterion . I n regar d t o th e latter, notic e tha t the mer e fac t o f two element s bein g i n a SPEC-head configu ration doe s not impl y stric t adjacenc y betwee n th e two , a s show n b y th e grammaticality of the examples below for two of the best-known cases of SPEC head relations:
Structural Focus, Structural Case, and The Notion of Feature-Assignment 3
3
(7) Who in you r opinion should w e invit e to th e meeting ? (8) The boys / somebody certainl y could hav e arrive d b y now. 7 Thus, sinc e the FOCUS-Criterion provide s n o account for the above tw o fact s of Hungaria n Focus , w e nee d t o postulat e th e existenc e o f som e additiona l mechanism(s) i n orde r t o answe r question s 1 and 2 posed above . As fo r questio n 1 , involvin g the nee d fo r V-movement , on e ma y firs t tr y a minimal addition to the FOCUS-Criterion account by stating that F needs to be "lexicalized" i n orde r t o b e "visible, " i.e. , t o be abl e t o functio n a s F , fo r pur poses o f th e FOCUS-Criterion . Bu t clearl y suc h a n amendmen t wil l no t solv e the problem . I f th e V happene d no t t o b e i n F a t th e leve l o f S-structur e i n a particular derivation , tha t woul d just rende r th e relevant clause , namel y (2a) , of the FOCUS-Criterion voi d in that derivation at S-structure , F being "invisible. " Accordingly, n o movement int o the SPE C o f FP would (nee d to ) occu r prio r t o S-structure i n suc h a derivation. Since bot h V-raising to F an d proposing o f th e Focus X P into [SPEC,FP ] coul d i n principl e tak e plac e withi n LF , creating th e configuration fo r Focus require d by the FOCUS-Criterion, unde r this hypothesi s we woul d incorrectly predic t tha t Hungarian will i n fac t permi t in-sit u Focus a s well. Thu s w e nee d t o loo k fo r a differen t answe r t o questio n 1 , preferabl y on e that coul d als o provid e a solutio n t o the problem o f the adjacenc y requirement , raised i n questio n 2 . The followin g tw o alternativ e hypothese s eac h provid e a n answer t o bot h question s 1 and 2 ; eac h o f the m cruciall y involve s th e proces s (postulated i n Horvath (1981 , 1986) ) by means of which a n X° category assign s the syntacti c featur e [+FOCUS] : Hypothesis A Consider the proposal made in Brody (1990). Unde r his analysis, the reason wh y V must mov e into F in Hungarian Focu s sentences is that the XP occupying the [SPEC.FP] positio n a t S-structur e must receiv e th e [+FOCUS ] featur e in orde r to satisf y th e FOCUS-Criterio n a t S-structure (se e (2a) ) an d thi s featur e ca n b e assigned to non-A positions only by V. 8 That is, following the analysis in Horvath (1981, 1986) , thi s accoun t assume s th e categor y V to be abl e t o carr y th e syn tactic featur e [+FOCUS] , and , crucially, t o assig n it under the condition s famil iar fro m th e assignmen t o f othe r syntacti c feature s b y a n X ° category , suc h a s Case. Thes e condition s o n feature-assignmen t ar e (a ) governmen t by , an d (b ) adjacency t o the feature-assigning head. Consequently , w e have here a n account for wh y V need s t o prepos e i n Focu s construction s (namely , t o gover n th e [SPEC,FP] position) , an d also for why this V needs t o be strictl y adjacen t t o the [SPEC.FP] position . Hypothesis B This alternativ e proposal, whic h I would like to explore here , als o maintains that the reason fo r V-movement in Hungarian Focus construction s an d for the neces sary adjacenc y betwee n the Focused constituen t and V is the need fo r the phrase eventually interprete d as Focus t o receive the syntacti c feature [+FOCUS ] fro m an X° category, under the familia r condition s of government and adjacenc y (se e
34
Discourse Configurational Languages
also Horvath 1981,1986) . But it diverges fro m the former account—and also from Horvath (1981 , 1986)—i n claiming that V is nor the carrier and assignor of the [+FOCUS] feature. 9 I t maintains instead that V onl y helps th e particula r func tional head involved t o assign the feature, by "lexicalizing" it.10 Before turnin g to some evidenc e that may help us decide betwee n thes e tw o alternative views , notic e wha t unites thes e two accounts: the y bot h assum e th e existence o f a mechanism of syntactic feature-assignment by an X° category cru cially involve d in the syntax of Focus (a t least) i n Hungarian. Whether i t is parameters involvin g thi s feature-assignmen t proces s o r th e parametrize d FOCUS-Criterion that accounts fo r the variation observable i n the syntax of Focus i n languages wil l be th e majo r topi c o f discussio n i n th e rest of thi s study . But firs t le t us consider briefl y the issue raised b y the above tw o alternatives a s to whethe r the X ° category that i s the sourc e of th e [+FOCUS ] featur e i s V it self, o r some functiona l hea d (suc h as, e.g., a category F),
2.2. The Source of the (+FOCUS] Feature The idea that in Hungarian (and also in several other designated Focus languages) the source and assigner of the syntactic feature [+FOCUS] i s the category V was originally propose d a s part of the framework for the syntax of Focus outlined i n Horvath (1981 , 1986), and has been adopted unchanged int o Brody's (1990) account o f Focus , referre d t o above . The basi s fo r thi s clai m wa s th e systemati c relation o f governmen t and adjacenc y tha t V appeare d t o bea r wit h respec t t o Focus constituent s at the level of S-structure. However, as noted also in Ortiz de Urbina (1986) an d Tuller (1992), thi s claim raises the conceptual problem o f why it would be exactly the lexical category V, rather than any other lexica l category , that ca n ac t a s th e sourc e an d assigne r of th e featur e [+FOCUS ] i n designate d Focus languages . Moreover, on e ma y wel l wonde r why suc h a featur e woul d originate i n a lexical, rathe r than a functional, category i n the firs t place. Thus , the claim that V is the [+FOCUS] assigning categor y appear s t o be a rather un natural stipulatio n i n the abov e analyses . Proponent s o f suc h a n analysi s coul d of cours e poin t out an apparently parallel case , namely that of structural accusative Case , a s in fac t ha s bee n don e i n Horvat h (1981, 1986) . Specifically , i t could b e note d tha t in man y language s V is the category assignin g accusativ e Case unde r the purel y structura l (i.e. , non-thematic/lexical ) condition s o f gov ernment and adjacency; hence the alleged associatio n between [+FOCUS] assign ment and the category V is not as unique as it might seem a t first glance . Bu t as we wil l se e below , ther e ar e (a t least ) tw o majo r conceptua l reason s wh y thi s alleged parallelism , an d more generally, the idea of V being th e assigner o f th e [+FOCUS] featur e (i.e., Hypothesis A ) is untenable . First, notice tha t a syntactic feature tha t i s associated with, i.e. , originate s in the feature-matri x of , a lexical category may be expecte d t o potentiall y exhibit sensitivity t o particula r lexica l instantiation s of th e feature-assignin g category , as well as sensitivity to the thematic relation (o r the lack of it) between the fea ture-assigning head and the phrase which is to receive th e feature. Both of these
Structural Focus, Structural Case, and The Notion of Feature-Assignment 3
5
possibilities ar e actually substantiated in relation to Case assigne d b y V; in particular, thi s latte r feature , whe n assigne d b y V, may b e no t onl y structural (a s e.g. i n English), but also inherent, the latter type of assignment exhibiting precisely th e kin d of lexical/themati c sensitivit y referred t o above . Bu t crucially , [•fFOCUS] assignmen t neve r seem s t o manifes t thi s kin d o f sensitivity, i.e. , i t never seem s to be a lexically o r thematically governe d process . Secondly, [- f FOCUS] assignmen t is distinct i n nature even from purely structural feature assignmen t processes done by uncontroversially lexical categories, such a s fo r instanc e (structural ) accusativ e Cas e assignmen t b y V mentione d above. Recall "Burzio's generalization," which states that if a verb assigns accusative Case, then it assigns a theta-role to its subject (Burzio (1981)). This prin ciple expresse s a dependency relatio n tha t holds betwee n th e structura l featur e accusative Cas e o n V an d the theta-assignmen t propertie s o f this category . N o similar dependency relation appears to be exhibited with respect t o the presence/ absence of the feature [+FOCUS ] and thematic properties of V. Yet, if [+FOCUS ] is just another structurally assigned feature of V, one could in principle expect to find suc h phenomena, as we actually do in the case of V-assigned structural Case , Thus, [+FOCUS ] manifests no association/interaction whatsoeve r wit h the lexi cal category V alleged t o be its sourc e and assigner i n the Hungarian-typ e lan guages. The abov e dissimilaritie s betwee n Cas e feature s assigned b y V and the fea ture [+FOCUS] should not be dismissed on the grounds that they have to do with intrinsic differences between Case and Focus, sinc e there is a parallel systemati c difference holdin g within the set of Case features in the above respects, namely between Cas e assigne d b y lexical vs . Cas e assigne d b y functional categories . Contrast the behavior o f Case feature s whose source is a lexical category , suc h as V discusse d above, wit h that o f nominativ e Case , whos e sourc e i s a func tional category, namely, I (or within the framework of Pollock (1989), the functional categor y T) . No t surprisingly , nominative Case assignment—jus t lik e [-fFOCUS] assignment—ha s no thematically o r lexically governe d version ; du e to being assigne d by a functional head , it is uniformly structural, an d it manifests no dependency relation with thematic properties (comparabl e e.g. to Burzio's generalization). These observation s then strongly suggest that it is not V—or any other lexical category—tha t is the source and assigner of the [+FOCUS ] feature in designate d Focus languages , but rathe r som e functional X° category, a s as serted b y ou r Hypothesi s B in sectio n 2.1 . Empirica l evidenc e i n favo r of this conclusion fro m th e case of Hungarian is presented below . Recall tha t under Hypothesis B , the explanation for why V seems t o have t o prepose in Hungarian Focus sentences, an d to appear adjacent t o the Focus phras e has nothin g to do with the category V as such (i t has no [+FOCUS ] feature t o assign); rather the explanation for these generalization s is simply that V acts in the cases considered a s the element tha t provides lexica l content/substanc e for a functional hea d possessing the [+FOCUS] feature , thus enabling the latter to assign thi s feature . Consequently , this account makes a n interestin g prediction: i f a Hungarian Focus sentence turn s out to have no (phonologically overt) V, then some othe r overt lexical X° category will prepose t o th e Focus-adjacen t posi -
36
Discourse Configurational Languages
tion, in order t o "lexicalize" the [+FOCUS]-assigning functional head. This pre diction i n fac t turn s out t o b e correct , a s demonstrate d by th e contrast s in (9 ) and (10) : (9) a . JANOSNAL j vol t Mar i magasab b t{ JOHN-BY wa s Mar y talle r 'Mary wa s taller tha n JOHN.' b. JANOSNAL ; magasab b Mar i t{ JOHN-BY talle r Mar y 'Mary is taller tha n JOHN.' c. *JANOSNAL ; Mar i magasab b t{ JOHN-BY Mar y talle r (10) a . MAR I vol t m a faradt . MARY wa s toda y tire d 'MARY was tired today. ' b. MAR I farad t ma . MARY tire d toda y 'MARY is tired today.' c. *MAR I m a faradt . MARY toda y tire d If th e featur e [+FOCUS ] originate d i n th e lexica l categor y V , we woul d ex pect sentence s (9b) an d (lOb ) t o be impossible; bu t in reality, they ar e perfectly grammatical Focus sentences . Furthermore, if one tried to claim that in example s (9b) and (lOb) there is a phonologically null V—parallel to the past tense copula volt in (9a)-(10a)—which gets preposed, an d this is the category tha t assigns th e feature [+FOCUS] , on e would leave the contrast between the grammatical (9b)(lOb) and the ungrammatical (9c)-(10c) with no explanation; ther e woul d be no reason wh y the subject in (9c) and the adverb in (lOc) should not be able to occur before the predicate adjective, given that they do occur before it in (9a) and (lOa) respectively. Unde r our Hypothesis B, o n the othe r hand, the grammaticalit y of the (b ) sentence s an d th e ungrammaticalit y of th e (c ) sentence s woul d follo w automatically; i n the absence of a verbal element, the A heading th e small claus e must propos e in order to lexicalize the [+FOCUS]-assignin g functional head.
2.3. Focus Positions and the [SPEC,FP] Hypothesis So far we have argued (a) that even if one assumes the existence o f a category F (and FP) and a parametrized Focus-Criterion—as in (2)—one still needs a mechanism o f feature-assignmen t by a n X ° categor y involvin g the syntacti c feature [+FOCUS] in designated Focus languages , and (b ) that in th e latter type of languages, the source and assigner of the feature [+FOCUS] is not V but rather some functional X ° category. Th e nex t issu e t o addres s the n i s whethe r the variatio n exhibited by the S-structure syntax of Focus is best accounted for (a ) in terms of a universal category F, projecting an FP, in conjunction with the parametrization
Structural Focus, Structural Case, and The Notion of Feature-Assignment 3
7
of conditio n (2a ) o f the Focus-Criterion (a s propose d i n Brod y (1990)) , or (b ) in term s o f parameters involvin g the wa y in which the [+FOCUS ] featur e i s ac quired, namely , in terms of its particular source/assigner and its mode of assign ment. Whil e th e cas e o f Focu s i n Hungaria n ha s show n th e nee d fo r th e mechanism o f [+FOCUS ] assignment , i t canno t she d ligh t o n th e abov e issue , since i t turn s out to b e consisten t wit h either o f th e tw o assumption s abou t th e source of variation . Specifically, the "designated " S-structur e position of Focus phrases i n Hungarian could b e du e to th e process of [+FOCUS]-assignmen t be ing limite d t o a particular functiona l hea d (t o be identifie d belo w a s I) and tak ing plac e a s a result o f a SPEC-head relation , or , alternatively, i t may be du e t o the nee d fo r th e nod e F postulate d b y Brod y (1990 ) t o satisf y claus e (a ) o f th e Focus-Criterion (se e (2) ) alread y a t th e leve l o f S-structur e i n Hungaria n (i n contrast to , e.g., English) , i.e., to have it s SPEC fille d by a [+FOCUS] phras e a t S-structure. S o let u s se e what types o f cases o f designated Focu s coul d hel p u s decide th e issue raise d above . A theory whic h derive s th e phenomenon o f designate d S-structur e Focu s po sitions from th e requirement fo r the SPEC o f the category F P to contain a [+FOCUS] phras e a t S-structure , impose d b y th e relevan t versio n o f th e Focus-Criterion, makes a number of rather specifi c predictions, amon g them th e following: (a) I n all designated Focus languages , the position o f the ex-situ Focu s elemen t at S-structure will be identica l to its position a t LF, namely the [SPEC.FP ] position (du e to the Focus-Criterion). 11 (b) Accordin g to the Focus-Criterion, i f a clause in a designated Focus languag e has som e projectio n o f th e categor y V as it s onl y Focuse d element , thes e categories to o wil l hav e to appea r i n th e SPE C o f FP positio n a t the leve l of S-structur e (se e Brod y (1990) o n Focuse d V' s i n Hungarian). (c) Sinc e th e S-structur e positio n o f Focu s i n designate d Focu s language s i s supposed t o be the [SPEiC,FP ] position, any differences found wit h respec t to the structural position of Focus among various designated Focus languages will have to result from correspondin g differences in the position o f the F P or o f th e SPE C withi n FP in th e phras e structur e of these languages . Fur thermore, du e to the same requirement, on e would not expect t o fin d mul tiple (functionall y identical) S-structur e Focu s position s "scattered" in th e clause structure of a single designated Focus language. This is so since under any reasonabl y restrictiv e notio n o f derivin g th e hierarchica l arrangemen t of functional categories (whethe r by functional selection, or by Grimshaw's (1991) extende d projectio n mechanism) , th e categor y FP—an d henc e [SPEC,FP]—would fit into, i.e., woul d get licensed i n the sequence of func tional categories tha t make up clause-structure only at one single place, simi larly to the other, more familiar functiona l categorie s o f the clause (e.g . TP) . In th e followin g sectio n we wil l discus s empirical data, drawn from a variety of African languages , that bear on th e abov e predictions. Specifically, the case s
38
Discourse Configurational Languages
to be presented will be shown to contradict predictions (a)-(c ) of the [SPEC.FP ] hypothesis o f designate d S-structur e Focus , an d wil l cas t seriou s doub t o n th e tenability o f the accoun t of Focus formulate d in terms o f a parametrized Focus Criterion.
3. The Range of Variation 3.1. Scope and the Focus Position An accoun t of the synta x of Focus tha t is based o n the Focus-Criterion—o n th e analogy o f th e Wh-Criterion—wil l involv e parametrizatio n o f conditio n (2a) , given below as (11), to draw the distinction between in-situ Focus language s (like, e.g., English) , and designated Focu s language s (see Brod y (1990)). (11) A t (S-structur e and) LF the SPE C o f an FP mus t contai n a [+FOCUS ] phrase. Based o n whethe r the abov e claus e o f th e Focus-Criterio n mus t b e satisfie d already at S-structure , or onl y at the leve l of LF we wil l get designate d Focu s languages, or in-situ Focus languages, respectively. Under this theory of the syntax of Focu s then , a s note d i n predictio n (a ) above , th e S-structur e positio n o f a (moved) Focu s phras e i n designate d Focu s language s wil l necessaril y b e th e position i t occupie s a t LF ; namely , a t bot h level s i t wil l b e i n th e [SPEC,FP ] position. Consequently , the scop e o f Focu s i n designate d Focu s language s will correspond to the S-structure position o f the Focus phrase . Bu t thi s prediction i s contradicted b y th e cas e o f severa l designate d Focus languages , suc h as , fo r in stance, Aghem. According t o a thorough descriptive study by Walters (1979), Aghem ( a Bantu language spoken in Cameroon) i s a designated Focus language, with an SVO word order. Th e designate d syntacti c Focus positio n in Aghem, i n term s o f linea r or der, i s the position immediately following V. Thus consider the following instances of S-structure Focus, involving both interrogative wh-phrases and non-wh Focu s phrases (o n th e universa l requirement fo r interrogativ e wh-constituent s t o b e [+FOCUS], se e Horvath (1986 , sectio n 2.3): 12
(12) a . fi l a z i ki- M a n 'so m friends S M P 2 ea t fuf u i n far m 'The friend s at e fuf u i n th e farm. ' b. r t WHER E fuf u 2 ea 'Where di d th e friend s ea t fufu? ' c- ' k 5 ? friends S M P 2 ea t I N FAR M fuf u 'The friend s at e fuf u I N TH E FARM. '
Structural Focus, Structural Case, and The Notion of Feature-Assignment 3
9
(1 friends S M P 2 ea t fuf u wit h han d 'The friends at e fuf u wit h (their ) hands. ' b friends S M P 2 ea t HO W fuf u 'How di d th e friend s eat fufu? ' c.. friends S M P 2 ea t WIT H HAN D fuf u 'The friends at e fuf u WIT H (THEIR ) HANDS. ' While th e (a ) sentence s ar e unmarke d a s t o Focus , bot h th e (b ) an d th e (c ) sentences exhibit their unique Focus phrase, obligatorily, in the immediately post verbal position (se e Walters (1979, pp . 147-148)) . What is crucial t o notice her e from ou r poin t o f vie w i s tha t i n example s (b ) an d (c) , th e Focu s phras e get s interpreted a s havin g scop e ove r th e entir e sentence . Ye t these Focu s phrase s arguably occur at the level of S-structure in a position that does not have the entire sentence withi n it s c-comman d domain ; fo r instance , th e subject-N P occurrin g in the [SPECJP] position is apparently outside of the c-command domain of the Focus phrase . So , assumin g ;;cop e t o be determine d o n the basis o f c-comman d relations, th e prediction o f identity between S-structur e an d LF Focu s position s in designate d Focu s languages , note d a s prediction (a ) i n 2. 3 above , appear s t o be falsifie d b y th e fact s o f Aghem. Ther e ma y b e tw o potentia l way s t o tr y t o avoid thi s conclusion . On e coul d tr y t o clai m (a ) tha t th e immediatel y post- V Focus phrase s o f Aghem actuall y are in a [SPEC.FP ] position tha t doe s c-com mand the rest o f the clause, o r (b) that the subject-NP' s exhibite d i n our Aghem examples ar e not in an A-position, but rathe r i n som e peripheral A-bar position, and henc e fo r purpose s o f scop e interpretatio n the y migh t stil l coun t (base d o n the corresponding empt y category in the relevant A-position) a s being within the scope o f Focus . Bu t i n fac t neither o f thes e claim s seem s tenable . Startin g wit h (b), Walter s point s ou t tha t subject-NP' s i n Aghe m clause s uniforml y occu r t o the left o f I and V, unless the y are Focused, indicatin g that thi s i s the normal Aposition fo r subjects . Also th e obligator y appearanc e o f wha t Walters analyze s as an expletive element (glossed as DS="dummy subject") i n sentences wit h postV subjects, suc h as (14), argues—i f h e is righl—that the clause-initial N P in our Aghem examples i n (12)-(13) is in an A-position, namely , the SPEC of IP posi tion, an d no t i n som e left-dislocale d posilion . (14) a m o z i a-fi n b£-'ko \ DS P 2 ea t FRIEND S fuf u 'THE FRIENDS at e fufu. ' Turning no w t o clai m (a) , maintainin g tha i Ih e immediatel y posl- V Focu s posilion, i.e., Ih e allege d [SPEC,FP] , in Aghem i s hierarchically highe r tha n th e clause-initial SPEC of IP position just discussed, would imply the following kind of clause-structure:
40
Discourse Configurational Languages
While suc h a structur e woul d ge t th e scop e fact s right , derivin g fro m i t th e actual order of constituents attested would involve a number of very curious stipulations. T o get th e adjacenc y relation betwee n V an d th e Focu s phrase , firs t of all on e woul d need t o clai m tha t all intervenin g complements o f th e V ge t dis placed t o som e right-periphera l position, following [SPEC,FP] . Bu t firs t o f al l there woul d be n o wa y t o forc e suc h a n operation , moreove r Watters' s (1979 ) careful descriptio n notes no phenomenon—such as e.g. a n intonation break—that could be take n as indication of any sor t o f right dislocation i n Focus sentences . Furthermore, withi n suc h a structure , ther e seem s t o b e n o reasonabl e wa y t o explain ho w th e Focu s phras e receive s th e featur e [+FOCUS] . Notic e tha t nei ther V nor I gover n th e [SPEC,FP ] position , s o neithe r o f thes e coul d serv e a s Focus-assigners, eve n i f they wer e adjacen t to it; th e F node tha t does probabl y govern [SPEC.FP ] i s no t adjacen t t o it , du e t o th e intervenin g subjec t i n th e [SPEC,IP] position . Assumin g a modifie d version o f structur e (15 ) i n whic h F follows rathe r tha n precede s it s I P complemen t would eliminat e thi s las t prob lem, bu t i t woul d introduc e instea d a ne w undesirabl e stipulation . Unde r thi s analysis, th e interna l structure of F P i n Aghem woul d be th e exac t opposit e o f the othe r phrases—suc h a s e.g . I P and VP—i n th e language : FP , exceptionally, would have a head-final, rather than the normal head-initial, order a t the X' level, and woul d i n additio n hav e it s SPE C follow , rathe r tha n precede , th e X ' con stituent. In light o f all these problems , on e must conclude tha t a structure i n which th e S-structure post- V Focu s positio n c-command s th e whol e claus e (a s i n (15) ) i s not a tenabl e alternative . Consequently , w e ar e bac k t o ou r origina l clai m tha t the fact s o f Aghe m Focu s sentence s falsif y predictio n (a ) (sectio n 2.3 ) o f th e parametrized Focus-Criterio n accoun t of th e synta x of Focus , sinc e i n thi s lan guage th e S-structur e designated Focus positio n is apparentl y distinct from th e position o f Focu s at th e leve l of LF .
Structural Focus, Structural Case, and The Notion of Feature-Assignment 4
1
3.2. Focusing of Verbal Projections While th e prediction o f the parametrize d Focus-Criterion wit h respect t o the S structure positio n o f Focuse d V-projection s o f designate d Focu s language s i s clearly th e sam e a s it is with respect t o al l othe r categorie s (se e predictio n b in section 2.3) , whe n w e conside r th e cas e o f Focu s i n Kikuy u (a s describe d i n Clements (1984)) , w e discove r tha t th e fact s o f a t leas t thi s languag e see m t o contradict this expecte d parallelism . In Kikuy u (a Bantu language spoken in East Africa), a (single) Focused con stituent appear s i n a structura l position immediatel y following COMP (i.e. , th e head of CP), marked by a preclitic "focus particle " ne (glossed a s "FP"), accord ing t o the analysi s of Clements (1984). Kikuy u being a rigid SV O language, the Focused phrase , following the overt or non-overt C, occurs to the left o f the sub ject (unles s the subjec t itself i s the Focused element): 13 (16) a . n e KAANAK 8 KamaYt i a - 6 n-'i 'it FP SP-seeT 'It's Kanak e (that) Kamau saw. ' b. n 6 MBER8 YA MO-TE O6-RE A Kama, 'u a-5n-'ir£ Kaanakg FP FRONT O F CP-TRE E PP-DE M SP-seeT 'It's in fron t o f tha t tree that Kamau saw Kanake. ' (=Clements (1984 , (9b,c)) ) Furthermore Clement s (1984 ) establishe s (a ) tha t th e S-structur e positio n o f these Focu s phrase s is due to movement , i.e. t o thei r being extracte d ou t o f th e clause, an d (b ) that thes e phrase s sho w up precede d b y th e ne "focu s particle " only i n thi s post- C designate d Focu s position . No w considerin g firs t wha t thi s focus particl e could possibl y be, there ar e two conceivable analyses : (a) it coul d be the overt morphological manifestation of the feature [+FOCUS], being assigned to a phrase in the post-C position by some appropriatel y located functiona l head , most likely , by C, or (b) ne could be claimed t o be the head of FP occupying the F position . Thi s latte r hypothesi s howeve r immediatel y turn s ou t t o b e incom patible wit h th e ide a o f [SPEC.FP ] bein g th e designate d Focu s position , sinc e there i s n o possibl e phras e structur e we ca n postulat e fo r Kikuy u clauses (se e (16) above) within which ne and the phrase immediately followin g it could b e in a SPEC-hea d configuration. 14 S o within any theor y o f Focu s base d o n th e pa rametrized Focus-Criterion, w e would be led to assume hypothesis (a) regarding the natur e of ne; the conception o f ne as the over t realizatio n o f the structurall y assigned [+FOCUS ] featur e we have postulated woul d indeed permi t u s to claim that the Focus phrase s marke d by this particle occupy the [SPEC,FP] position— to th e lef t o f a phonologically nul l F node—i n Kikuy u clause structure , as on e would expect based o n the Focus-Criterion. Bu t notice now that if this is so, and the SPE C o f FP position i s indee d immediatel y t o th e righ t of C and to th e lef t of th e subjec t position , then a n obvious problem i s posed by th e additiona l phenomenon describe d by Clement s (1984) an d exemplifie d by th e following:
42
Discourse Configurational Languages
(17) Kama. u ne-a-dn-'i'r £ Kaanak £ FP-SP-see-T 'Kamau sa w Kanake. ' While the information given on the above construction by Clements (1984 ) i s rather incomplete , wha t we do know about such examples, base d o n the glosses , as well a s on his discussion , is that they exhibit th e sam e focu s particle that w e see on Focused NP' s i n the alleged [SPEC,FP ] position. (Clement s demonstrate s for instance , tha t thi s preverba l focu s particl e an d th e on e o n phrase s i n th e [SPEC.FP] position ar e mutually exclusive i n clauses, for non-semantic reasons. ) Since th e focu s particl e appear s i n th e sentence-typ e show n i n (17 ) attache d t o the lef t o f th e V- I complex , crucially , followin g th e subject , an d neithe r o f th e arguments of the clause is interpreted as Focus, i t is reasonable t o infer that what we have here i s a Focused verba l projection. The data presented b y Clement s i s insufficient t o determin e whether the sentenc e involve s VP/I' o r V Focusing, o r possibly bot h o f thes e options , but i n an y cas e th e fac t relevan t fro m ou r poin t of view , i.e., fro m th e poin t of view of the parametrize d Focus-Criterion, i s that under suc h a n accoun t any Focuse d verba l projectio n bearing ne woul d be pre dicted, wrongly, to appear at S-structure to the left o f the subject , namely i n the [SPEC, FP]—o r i n the cas e o f V possibly i n th e F—position o f Kikuy u clausestructure, just lik e th e Focuse d phrase s d o in (16). 15
3.3. Focus Positions, FP's, and Variations of Clause Structure Consider nex t th e phrase-structure o f som e o f the Chadic designate d Focu s lan guages—studied i n Tulle r (1988 , 1992)—fro m th e poin t o f vie w o f th e param etrized Focus-Criterion, in particular from the point of view of prediction (c) noted in sectio n 2.3 . Tuller (1992 ) discusse s severa l strictl y SVO Chadic language s whic h exhibi t a designate d Focu s positio n (both for W H an d fo r non-W H constituents ) t o th e right o f V in linea r order. I n som e o f thes e languages—a s i n Western Bad e and Podoko—the Focuse d phras e occur s immediatel y afte r V itself , precedin g th e direct objec t (i f there i s one) , an d i n others—a s i n Kanakur u and Tangale—th e Focused phras e occur s immediatel y afte r a V+N complex , whic h arise s (a s ar gued by Tuller) vi a the incorporation o f the head o f the direct object . (Thi s latte r incorporation process , induce d b y th e lac k o f Cas e transmissio n i n V-heade d chains i n th e relevan t languages , wil l mak e n o differenc e fo r purpose s o f ou r discussion.) T o illustrat e th e cas e o f suc h post- V Focu s languages , conside r examples (18 ) fro m Wester n Bade, and (19 ) fro m Kanakur u (=Tuller (1992 , (1 ) and (5))) : Western Bad e (18) a . Saak u a a bon a kajluwaan . S Inf l coo k tuw o 'Saaku wil l cook tuwo. '
Structural Focus, Structural Case, and The Notion of Feature-Assignment 4
3
b. Saak u a a bsn a K9M ? S Inf l coo k WHA T 'What wil l Saak u cook?' c. gafa- n K E viiriidgwaran ? caught WH O giant-ra t 'Who caught a giant rat? ' d. g a ba r II - T E tagda-w ? you giv e T O WH O mone y 'Who did you giv e the money to?' Kanakuru (19) a . ar e lowo i JEWO I l a lusha . bury boy-th e SLAVE-TH E i n bus h THE SLAV E buried th e boy in the bush.' b. n a dibar e garn i MANDAI ? buy ram-th e WH O 'Who will buy the ram?' c. ka a na i MANDAI ? you cal l WH O 'Who ar e you calling?' Trying to mak e sens e o f the abov e alternation s in constituen t order unde r th e parametrized Focus-Criterion hypothesis is quite problematic to begin with. This hypothesis implies that the S- structure position o f the Focus phras e i n these des ignated Focus language s must be the SPE C o f FP. This i n turn raises firs t o f al l the sam e proble m tha t we pointed ou t i n connection wit h th e cas e o f Aghem i n section 3.1 , havin g to d o with the scop e o f Focus/(WH)-phrases an d the highl y implausible syntacti c consequences involved . (See the discussion o f structure (15) , which applie s wit h equal forc e to the case o f Western Bad e an d Kanakuru.) Bu t let us ignor e the issue o f scope for the sak e of argument . I n that case , we coul d in fact sugges t a clause structur e for these language s that woul d actuall y exhibit the Focu s phrase s o f (18)-(19) i n a [SPEC.FP] position :
44
Discourse Configurational Languages
What we would need to assume in order to get the Focus dat a in (18)-(19) on the basis o f such a structure is that V (in (19 ) wit h the incorporated N ) raises t o I, presumably vi a F, in compliance wit h the ECP . Sinc e thi s V-movement t o I is a well-motivate d operation , th e structure in (20)—together with th e assumptio n of movemen t o f the phras e t o be interprete d a s Focus int o th e [SPEC,FP ] position—may indee d see m t o constitut e a n adequat e accoun t o f th e S-structur e o f Focus i n the languages unde r discussion withi n th e framewor k o f the param etrized Focus-Criterion (recall though the residual proble m of scope). But when we tak e not e o f som e additiona l facts , involvin g th e rang e o f possibl e desig nated Focus positions observable in these, and other, languages, it becomes clea r that th e parametrize d Focus-Criterion i s quit e ill-suited fo r accommodatin g th e relevant phenomena . Notice firs t o f al l the variatio n manifested wit h respec t t o the hierarchical position of Focus between individual designated Focu s languages . In particular, while e.g. i n Western Bade the designated Focus position—i.e. , the SPEC o f FP position, under the parametrized Focus-Criterion account—follow s the S-structur e subject-positio n (namely , the [SPEC.IP ] position) , i n othe r rig idly SVO languages, such as in Hausa, as well as in Kikuyu (see sectio n 3. 2 (16)), the designate d Focu s positio n i s clause-initial, i.e. , i t precedes th e S-structur e subject-position. Wha t thi s variatio n woul d impl y i n term s o f th e parametrize d Focus-Criterion accoun t of the synta x of Focu s i s tha t th e hierarchica l positio n of F P (an d consequently [SPEC,FP] ) differ s i n th e tw o language-types . Specifi cally, th e categor y F i n th e gramma r o f Haus a an d Kikuy u woul d hav e t o b e assumed to take IP a s its complement, whereas F in Western Bade (as well as in several other post-V Focus language s studied in Tuller (1992) ) would have to be specified a s taking VP as its complement, and in fact heading th e complement o f I. However , th e incorporatio n o f language-particula r distinctions involvin g th e complement-taking propertie s o f a functiona l hea d lik e F int o an y reasonabl y restrictive theor y of selectio n i s quite problematic. Notic e firs t that what we ar e dealing wit h here is not lexically determined variation in the kind of complement for differen t choice s of F; rather the alleged F selects a unique complement within each of these languages. More importantly, if one tried to maintain that the abov e cross-linguistic variatio n in the hierarchica l positio n of FP in clause-structur e i s still du e t o distinc t selectional propertie s o f F i n th e language-type s involved, one woul d fac e th e followin g problem : Th e categor y F P itself i s no t subjec t t o selection; in all th e designate d Focus language s referred t o in thi s study, the al leged F P ca n occu r i n embedde d a s wel l a s i n matri x clauses , ye t ther e ar e n o verbs, complementizers o r any other heads in these languages that select specifi cally a n FP , nor an y head s tha t prohibi t the presenc e o f F P in/as thei r comple ment. S o F would have t o be a n optionally occurring functiona l category tha t is irrelevant/invisible fo r selectiona l processes . Bu t i f so , the n on e canno t reason ably clai m tha t it stil l i s a category tha t selects it s complement . A parallel argument against the us e of selectio n fo r functiona l categories , in volving the category Neg, i s presented by Grimsha w (1991). I n her comprehen sive stud y o f functiona l categories , Grimsha w propose s a theor y o f phras e structure based on the notion of "extended projection," defined wit h reference to a "functiona l feature " specifying a n F-value for each head; this theory offers a n
Structural Focus, Structural Case, and The Notion of Feature-Assignment 4
5
alternative, non-selectional , accoun t for th e positionin g of functiona l categorie s in hierarchica l structure , an d consequentl y permit s a significantl y more restric tive theory o f selection . Within Grimshaw' s projection-based theory , th e differ ence w e foun d between th e hierarchica l positio n o f th e allege d F P i n Wester n Bade vs . i n Haus a woul d b e capture d b y attributin g differen t F-value s t o th e category F i n the two language-types . Specifically , F taking V P (bearing th e F value [FO] ) as its complement—se e Western Bade—woul d b e claime d t o b e a n [Fl] category; in contrast, the Hausa-type languages would be claimed to have F with the F-value [F2], whil e the [Fl ] categor y i n the verbal projection woul d be I, thus accounting for IP occuiring as the complement of F in this language-type. Although thi s alternativ e theory seem s t o provid e a mor e plausibl e accoun t for th e variation of Focus unde r discussion than one based o n selection would , it would also raise some problems of its own. According to Grimshaw's (1991 ) well motivated proposal , (a ) maxima l projections tha t ar e no t lexicall y selecte d ca n be license d onl y i f the y constitut e part o f a n 1-selecte d "extende d projection, " and (b ) th e F-valu e o f categorie s formin g an extended projectio n ma y diffe r b y no more tha n one. Thi s i n turn implies that an intermediate (i.e., not 1-selected ) projection lik e our alleged F P would have to be an obligatory part of all clauses , since i n its absence , th e lowe r maxima l projection s makin g u p th e claus e (e.g . IP, VP) will not be part of a well-formed extende d projection, an d consequently will not get licensed i n Grimshaw's system. But of course, thi s would be incom patible wit h the Focus- Criterio n (a s in Brody (1990)), which is formulated with reference t o the category FP: if FP were indeed obligatorily present in all clauses, then the Focus-Criterion woul d wrongly predict that universally, all clauses must have some Focused element, and furthermore that in designated Focus languages, every claus e mus t have som e elemen t appearin g in th e [SPEC.FP ] positio n al ready a t th e leve l o f S-structure . Mechanically , on e coul d o f cours e avoi d thi s conflict, b y claimin g (a ) tha t the categor y F , and henc e FP , is i n fac t presen t i n every clause , an d (b ) that the Focus-Criterion i s sensitive no t t o the presence o f F, but rathe r t o som e featur e [+/-f ] occurrin g o n th e nod e F . This woul d work , but notice that such a move would call into question the significance of the node F for the syntax of Focus, an d would create the possibility o f this new [+/-f ] fea ture being associate d wit h some X° category othe r tha n F , thus bringing us on e step close r to th e conceptio n o f Focus tha t is argue d fo r i n th e present study. Finally, a further problemati c aspect of the FP-based accoun t of th e synta x of Focus i s uncovere d whe n we conside r som e intra-languag e variatio n describe d in Tuller (1992) , involvin g the existence of more than one alternative Focus po sition i n som e o f th e Chadi c designate d Focus languages , such as i n Kanakuru, Tangale, an d Ngizim . T o illustrate the issue , compar e th e followin g Kanakur u Focus sentence s (=Tuller (1992 , (28a , c))) wit h those presente d i n (19) : (21) a . BASH A she e tup a ya . 2perf V D O 'BASHA sen t him.' b. SH I man tupa. 'We sen t HIM.'
46
Discourse Configurational Languages
In addition to having a designated Focus position in the position specified i n (20 ) above, example s (21)—an d Tuller' s discussion—revea l tha t Kanakur u als o ha s an alternative designated Focus position, whic h occurs clause-initially (in Tuller's analysis thi s i s th e SPE C o f C P position) . No w notic e tha t i n orde r t o accoun t for suc h intra-language variation regarding the designated Focus position , a theory of Focu s base d o n the Focus-Criterio n woul d see m t o requir e th e generatio n o f FP in more tha n on e (non-adjacent ) hierarchical positio n i n the clause-structur e of the same language, since both of the possible Focus-positions exhibite d would need t o be [SPEC,FP ] positions. Thi s i n turn would mea n tha t the categor y F in Kanakuru, for instance, would have to be assigne d two alternative F-values : in addition to the specification [Fl], accounting for the examples i n (19) (see struc ture (20)) , F will als o nee d a n additional , higher F-value , presumabl y (a t least ) [F3], t o generat e sentence s a s those i n (21) . Th e situatio n is even mor e trouble some, i f th e simultaneou s use o f th e tw o Focu s position s i n a singl e claus e i s indeed impossible, as implied by Tuller's description. In this case one would als o need to state in some wa y a dependency relation between th e specification o f the two F nodes for the alleged [+/-f ] feature . It has to be noted here that a potential alternative analysis for the Kanakuru-type languages—which postulates a singl e FP an d movemen t o f F t o som e highe r hea d positio n whos e SPE C woul d thu s count somehow as a [SPEC,FP] position providing the option of Kanakuru clauseinitial Focus—woul d no t work. Clause-initia l Focus sentence s manifes t n o evi dence fo r hea d movemen t tha t i s no t presen t als o i n post- V Focu s clauses . Crucially, in both types of Focus sentence s of Kanakuru, the S-structure positio n of th e raised V-complex, which would include the head F (se e structur e (20)), is not highe r tha n I , a s show n b y th e fac t tha t i t follow s th e [SPEC,IP ] (i.e. , th e subject) position ; consequently , the strictl y clause-initial, pre-[SPEC,IP ] occur rence o f Focu s i n Kanakur u could no t b e attribute d t o hea d movemen t withou t ad hoc stipulation s about the placemen t of the Specifie r o f this particular raise d head. In sum , the possession o f two F-values (i n a single language)—eve n more so than th e apparen t optionalit y o f F—i s a propert y attribute d t o F tha t i s excep tional amon g functiona l categories . A s such , it strengthen s ou r impressio n tha t an accoun t o f th e synta x o f Focu s withi n th e framewor k o f th e parametrize d Focus-Criterion ca n b e achieve d onl y a t th e cos t o f reducin g th e intrinsicall y structural notions of F and [SPEC.FP] to the status of diacritics t o designate Focus positions.
4. Parameters of Feature-Assignment and the Syntax of Focus 4.1. Functional Heads and the [+FOCUS] Feature In view of the false predictions and other problems discusse d i n sections 3.1 , 3.2 , and 3.3, it seems that the parametrized Focus-Criterion account , and specifically, its direc t consequenc e that al l designate d Focus position s atteste d i n languages are [SPEC,FP ] positions , canno t adequately accommodate th e rang e of S-struc ture Focus phenomena . Recall in addition the point we made in section 2.1 : Any
Structural Focus, Structural Case, and The Notion of Feature-Assignment 4
7
analyses of Focus, even those assuming a parametrized Focus-Criterion—as Brody (1990)—need th e (additional ) mechanism of [+FOCUS ] assignment . The [+FO CUS] featur e bein g a n instance o f th e independentl y existing clas s o f syntacti c features, fo r i t t o exhibi t variatio n in th e sam e respect s a s othe r syntacti c fea tures d o (a s e.g . wit h respec t t o sourc e an d mod e o f assignment ) i s i n fac t th e null hypothesis. Give n the inadequacies of the parametrized Focus-Criterio n ac count i n conjunctio n with th e abov e fact s regardin g the featur e [+FOCUS] , w e propose her e th e following alternative hypothesis (mentione d alread y a t the end of sectio n 1 and a t th e beginnin g o f 2.3 ) t o replac e th e Focus-Criterion-base d account o f th e S-structur e synta x of Focus : (22) Th e S-structure variatio n exhibited by languages with respect t o Focus is du e t o variatio n i n th e wa y th e featur e [+FOCUS ] i s acquire d b y categories, i.e. , i t i s du e t o parameter s involvin g the syntacti c feature [+FOCUS]. On the one hand, this hypothesis seems t o permit more flexibility with respect to possibl e S-structur e Focu s position s tha n th e parametrize d Focus-Criterio n framework did , whic h is a desirabl e consequenc e i n vie w o f ou r discussio n i n section 3 . O n th e othe r hand , i t stil l i s highl y restrictiv e an d ric h i n specifi c empirical predictions regarding the range and types o f variation expected within the synta x o f Focus ; thi s i s becaus e wha t w e ar e claimin g her e i s tha t Focu s constructions do not involve any parameters o f their own, rather, they only manifest th e effects o f (specific settings for) independently existing, general parameters o f syntactic feature s an d syntacti c feature-assignment . Belo w w e will examine and elaborate this hypothesis and its empirical consequences i n light of the variet y of Focus phenomen a introduced in th e previous sections . Before turnin g to this, notic e first tha t by adoptin g the hypothesis i n (22 ) w e automatically eliminate the claim that all designated Focus position s are S-struc ture [SPEC,FP ] positions , which was inherent t o the parametrize d Focus-Crite rion hypothesis . This i n tur n immediatel y eliminates th e proble m involvin g the scope o f Focu s discusse d i n relatio n t o Aghem i n sectio n 3.1 , sinc e th e frame work o f hypothesi s (22) , unlik e th e parametrize d Focus-Criterio n account , im plies no necessary identity between the S-structure an d the LF positions of Focus . Furthermore, th e proble m pointe d ou t i n sectio n 3.2—tha t i n Kikuyu , Focused V-projections occur in a distinct, hierarchically lower position tha n other Focuse d constituents—also gets resolved; droppin g the parametrized Focus-Criterion ac count mean s tha t w e n o longe r predic t tha t al l Focu s element s i n a particula r language wil l necessarily occu r in th e sam e positio n (namel y in [SPEC.FP] ) a t S-structure. The issue of how our alternative hypothesis (se e (22) ) is able to ac commodate cases o f intra-language and cross-linguistic variatio n wit h respect to S-structure Focus-positions—such as those i n Kikuyu, as well a s Western Bade , Hausa and Kanakuru (presented in sectio n 3.3)—wil l be taken u p in ou r discus sion below. In sections 2.1 and 2.2 we established that in designated Focus language s like Hungarian the [+FOCUS ] feature originate s in and is assigned by an X° category,
48
Discourse Configurational Languages
and tha t this X ° is a functional, rathe r tha n lexical, categor y o f the clausa l pro jection. Another syntactic feature that arises o n categories du e to the presence o f a particula r functiona l hea d o f the claus e is , o f course , Nominativ e Case. Thus , it is natural to ask whether and how parameters familia r fro m analyse s of Nomi native Cas e ma y provide a way to accoun t for the variatio n exhibited b y desig nated Focu s language s with respect t o possible Focu s positions . S o we turn first to th e discussio n o f th e source s o f variatio n i n th e synta x o f Focu s amon g an d within designate d Focu s languages , and the n retur n t o th e issu e o f wha t i n ou r framework account s for the difference between designated Focus vs . the Englishtype Focu s i n sit u phenomenon . Suppose that in designated Focus languages, the [+FOCUS ] featur e originates in th e feature-matri x o f som e functiona l X ° categor y o f th e clause , mos t obvi ously, C or I (o r som e othe r inflectiona l head , i f on e assume s Pollock' s spli t I hypothesis). W e propose tha t the identit y o f the particula r X° category serv ing a s the sourc e of the featur e [+FOCUS ] constitutes one dimension o f para metric variatio n manifested by th e synta x of Focus. Thus , conside r fo r instanc e the S-structur e positio n o f Focu s i n Wester n Bad e (example s (18) ) vs . tha t o f Kikuyu Focu s phrase s (examples (16)). The fact tha t the Focus positio n i n West ern Bad e i s t o th e righ t o f th e subjec t (=[SPEC,IP]) position , an d a s argue d b y Tuller (1992) , is an A-bar position immediately following the I+V complex, sug gests tha t in this language the source of the [+FOCUS ] featur e is the category I , which assign s thi s feature , under governmen t an d adjacency , t o a VP-adjoine d constituent. In contrast, Kikuyu Focus phrase s precede th e subject (= [SPEC,IP] ) position, an d appea r immediatel y followin g the nod e C . This differenc e ca n b e accounted fo r by claiming that the sourc e o f the [+FOCUS ] featur e i n Kikuyu is the categor y C . The condition s an d mod e o f [+FOCUS ] assignmen t i n th e tw o languages woul d then b e identical—governmen t (t o the right ) an d adjacency — and the y would diffe r minimall y in terms of I vs. C being th e sourc e o f the fea ture [+FOCUS] . Note that the postulatio n of thi s parameter is full y consisten t wit h the paral lelism implie d by ou r hypothesis between (formal ) properties o f th e [+FOCUS ] feature an d of the featur e Nominative Case. Th e variatio n of structura l Focus i n Western Bade vs. Kikuyu accounted for above is precisely mirrore d b y the variation of Nominative Case in languages like Irish, Welsh, and Standard Arabic (VSO clauses) vs . Germani c V 2 language s suc h a s West Flemish . Specifically , whil e Nominative Case assignment takes place under government and adjacency in both of these language-types , they diffe r minimall y with respect t o the identit y of th e source, i.e. , assigne r o f Nominativ e Case : i n Irish , Welsh , an d Standar d Arabi c the sourc e o f Nominative Cas e i s the categor y I , wherea s i n the Wes t Flemish type language s i t i s th e categor y C (see , e.g. , Rizz i (1991) , base d o n wor k b y McCloskey an d b y Haegeman ; Koopma n an d Sportich e (1990) ; Fass i Fehr i (1989)). Having establishe d tha t U G permit s variatio n wit h respec t t o whic h clausa l functional head—namely , C o r I—i s th e sourc e o f th e [+FOCUS ] featur e (jus t like i t does with respect to Nominative Case), consider the intra-language variation o f Focus w e demonstrated in section 3. 3 based o n Kanakuru . The existenc e
Structural Focus, Structural Case, and The Notion of Feature-Assignment 4
9
of th e tw o alternativ e Focus position s show n i n example s (19)—post-I+ V Fo cus—and (21)—clause-initial , (or possibly post-C ) Focus—no w falls naturally in place. Kanakuru (as well as Tangale and Ngizim, discussed i n Tuller (1992) ) can be construe d i n ou r framewor k to exhibi t th e conjunctio n of th e sam e tw o pos sible settings fo r Focus-assigne r tha t wer e manifested, respectively, b y Wester n Bade an d b y Kikuyu ; namely , Kanakuru would hav e bot h th e categor y I (se e examples (19) ) an d th e categor y C (se e example s (21) ) specifie d a s possibl e sources o f [+FOCUS]. Note that the case of Nominative Case again turn s out to display a parallel pattern: the option of more than one alternative assigne r withi n the sam e language i s motivated i n Rizzi (1991 ) for Italian . Specifically , Rizzi' s proposal fo r preverba l vs . postverba l subject s i n Italia n involve s tw o indepen dent assigners for Nominative Case: the highest inflectional head, AGR, and the first inflectiona l hea d abov e VP, namely T. Another dimensio n of parametri c variation involving the [+FOCUS ] feature , and consequentl y the S-structur e synta x of Focus , i s reveale d whe n w e tur n t o the questio n of how t o account for the properties o f Hungarian Focus, presente d in section s 2.1 and 2.2. What is immediately obvious is that [+FOCUS ] i n Hungarian cannot be a feature o f C, since the designated Focus position neither needs adjacency t o no r governmen t by C i n th e S-structur e o f Hungarian . Instead, a Focused phras e occur s immediatel y precedin g th e inflecte d V (se e ou r discus sion o n V-raising an d examples (4 ) and (6 ) i n 2.1). Consequently , the sourc e o f the [+FOCUS ] feature in Hungarian, within our framework of assumptions, turns out to be the categor y I, similarl y t o the case o f Western Bade discusse d above . We assum e furthermor e tha t I ca n assig n it s [+FOCUS ] featur e onl y i f i t get s "lexicalized" by V, or some other lexical head moving into it (see examples (4b,d ) and (9)-(10)). 16 No w wha t i s crucia l t o notic e her e i s tha t Hungaria n clausa l heads consistentl y precede their complements. Give n this, th e fact that the des ignated Focu s positio n occur s immediatel y to th e left o f th e V+ I comple x im plies tha t th e Focu s positio n i s th e SPEC of IP position , rathe r tha n som e VP-internal o r VP-adjoined position. S o the structur e suggested withi n Brody' s (1990) framework , given as (5) in 2.1, ca n be adapted to reflect ou r present pro posal a s follows:
50
Discourse Configurational Languages
Thus, [+FOCUS ] assignmen t b y I apparentl y take s plac e i n a SPEC-head configuration i n Hungarian . This i s i n contras t t o th e wa y w e sa w [+FOCUS ] being assigne d i n Wester n Bade ; i n th e latte r language-typ e (se e Tulle r (1992 ) and the data in (18)), I assigns the [+FOCUS ] featur e to a phrase tha t i t govern s within/adjoined t o it s complement. S o i t follow s tha t beyon d variatio n i n th e assigner o f th e [+FOCUS ] feature , another paramete r motivate d fo r th e synta x of Focus involve s the mode of feature-assignment: SPEC-hea d relatio n wit h vs. government b y th e [+FOCUS ] assigner . Remarkably , thi s paramete r to o i s on e established independently , for variation of other feature-assignment processe s (se e e.g. Koopma n an d Sportich e (1990)) . Mos t prominently , Nominativ e Cas e i s commonly claime d t o manifes t th e effect s of precisely this parameter : Nomina tive assignmen t unde r SPEC-hea d agreemen t (a s in English , French , an d SVO clauses o f Standard Arabic) vs . Nominative assignment unde r government—and adjacency—(as i n Irish, Welsh, an d VSO-clauses o f Standar d Arabic) . Notice tha t our analysis for Hungaria n implies tha t th e [SPEC,IP ] position i n this languag e is a n A-bar rathe r tha n a n A-position, du e t o th e presenc e o f th e operator-creating/licensing [+FOCUS ] featur e on I. 17 (Assumin g th e VP-inter nal subjec t hypothesis, the SPE C o f IP is open i n UG to be a n A-bar positio n a s well a s a n A-position.) Not e tha t the abov e claim cause s n o proble m i n relatio n to Nominative Case assignment , sinc e it is independently known that subjects in Hungarian—in contras t to , e.g. , English—d o not receive thei r Nominativ e Cas e in th e SPE C o f I P position ; Nominativ e Cas e is , arguably , assigne d i n a post verbal, an d consequentl y post-I , positio n ( a situatio n commonl y atteste d i n a variety o f languages). I n fact th e distinction drawn in ou r analysi s between fea tures o f I that licens e it s Specifier , namel y Cas e (as , e.g. , i n English) vs . [+FO CUS] (a s i n Hungarian) , provides a n elegan t wa y t o captur e th e often-cite d intuition tha t the structura l Focus position i n Hungaria n is i n som e wa y equiva lent to—o r i s a "discourse " counterpar t of—the subjec t positio n i n "configura tional" language s lik e English . I n ou r terms , i t i s neithe r som e notio n o f (non-)configurationality, no r th e (un)availabilit y of th e categor y I/I P tha t i s in volved here, rathe r the variation stems simpl y from th e particular feature-assign ment proces s availabl e t o I t o licens e a SPE C positio n withi n eac h o f th e tw o language-types. It is also wort h noting in this context that the pattern of [+FOCUS ] vs . Nominative Cas e assignmen t motivate d fo r Hungaria n turn s out t o b e th e mirror-im age of the pattern foun d in some of the other designated Focus language s we have considered, suc h as Western Bade (see the discussion of examples (18 ) based o n Tuller (1992)) ; i n th e latter , [SPEC.IP ] i s th e positio n o f Nominativ e Case as signment, whil e the [+FOCUS ] featur e of I gets assigne d under government (t o the right) into th e complement of I. This i n turn substantiates the availabilit y of two distinc t modes/mechanism s o f syntacti c feature-assignment—(a) b y a hea d governing (into ) it s complemen t an d (b ) b y a SPEC-hea d relation , a s wel l a s supports th e postulatio n of [+FOCUS ] a s a tru e syntactic feature. Before summarizin g the parametric variation in the syntax of Focus accounted for i n terms of our feature-based hypothesis (22), w e need to have a look a t how the Focusin g o f V (an d its projections ) could wor k i n ou r framework . The fac t
Structural Focus, Structural Case, and The Notion of Feature-Assignment 5
1
that th e [+FOCUS ] featur e i s abl e to appea r associate d wit h V at S-structur e i s demonstrated b y Hungarian sentence s suc h as (24) below , a s well a s by the dis tribution o f the Focus-marking particl e i n Kikuyu sentences suc h as (17) vs. (16a) in 3.2 , repeate d her e a s (25a, b) : (24) (Csak ) TISZTELJU K (d e ne m szeretjiik ) a tanart . (only) RESPECT-lp l (bu t no t like-lpl ) th e teacher-ace . 'We (only ) RESPEC T (bu t don't like ) th e teacher.' (25) a . n e KAANAK8 Oral' u a-5n-Tr £ FP SP-seeT 'It's Kanak e (that ) Kama u saw.' b. Kama , t i ne-a-6n-'i'r £ Kaan£k £ FP-SP-see-T 'Kamau sa w Kanake.' Obviously, neither in the Hungarian sentence (24) nor in the Kikuyu sentence (25b) i s th e V (o r an y of it s projections ) in the designate d Focus positio n moti vated abov e fo r these two languages—namely, [SPEC.IP] , and a pre-subject po sition governe d b y an d right-adjacen t t o C , respectively . Ye t our framewor k o f assumptions, unlik e th e parametrize d Focus-Criterion hypothesis , provide s a natural wa y t o accommodat e V(-projection ) Focusing . Takin g firs t th e cas e o f Hungarian, note that in this language V is assumed t o move into I; consequently, it automaticall y pick s u p th e [+FOCUS ] featur e born e b y th e categor y I . Sinc e in our framework Focus interpretatio n for a constituent depends onl y on whether it acquire s th e [+FOCUS ] featur e in th e cours e o f th e derivation , V-to-I move ment obviousl y suffice s to accoun t fo r V Focusing i n Hungarian . This cas e ex hibits an alternative to the acquisition of a feature via "feature-assignment," i.e. , via a process o f feature-transfe r from a head (i n a SPEC-hea d o r head-govern ment configuration); wha t we have here is an instance of the "absorption" o f the [+FOCUS] featur e b y a lexica l head , du e t o hea d movemen t int o th e feature bearing node . Thi s sor t o f proces s agai n seem s t o hav e a n obviou s paralle l in volving Case : th e wa y cliti c pronoun s get affiliate d wit h Case-bearin g heads . Now turnin g to th e issu e o f Kikuy u sentences lik e (25b) , i t appear s tha t th e occurrence o f th e Focu s Particl e o n th e inflecte d V a t S-structur e i s du e t o a variant o f th e proces s postulate d i n relatio n t o V-Focu s i n Hungarian . The rel evant differenc e between th e tw o is that the [+FOCUS]-bearin g head i n Kikuyu is C—as shown by the position o f Focused phrase s i n examples suc h as (25a)— and sinc e V clearly doe s no t move into C, the process by whic h the [+FOCUS ] feature reaches th e V+I complex in Kikuyu must be somewhat different fro m tha t involved in Hungarian. What we may propose her e involve s Grimshaw's (1991 ) "extended projection " theor y (which we sketched alread y in sectio n 3. 3 above) . Given thi s framework , assume that syntactic features , amon g them [+FOCUS] , may freel y spread , i.e., project, from thei r source/assigner category to an y othe r head withi n th e extende d projectio n o f that category. Sinc e th e V+I complex is i n fac t a n "extende d head " o f th e C P (i n th e sense o f Grimsha w (1991)), w e
52
Discourse Configurational Languages
have an account for the appearance o f the Focus Particl e o n the former in (25b) . Note though tha t the feature-spread proces s tentatively suggeste d here should b e limited i n som e wa y t o preven t th e incorrec t consequenc e o f turnin g th e V+ I complex in Kikuyu into a potential assigner of the [+FOCUS ] featur e it received , or similarly, for Nominative Case, t o prevent, e.g. , I or V in West Flemis h fro m acquiring th e abilit y t o assig n Nominativ e by feature-projectio n fro m C . S o i t seems tha t individua l functiona l head s woul d stil l b e designate d a s uniqu e sources/assigners of particular syntactic features (as we argued before), and the "spreading" of a feature to some othe r hea d within the same extende d projectio n would be a process disjoin t fro m th e process o f the assignment of th e featur e (i.e., from it s transfer to a category outside the extended projection o f the source) ; in other words, these processe s woul d represent tw o mutually exclusive way s of associating syntacti c feature s with thei r S-structur e recipients . Assuming thi s necessar y prohibitio n agains t th e assignmen t o f a projecte d feature (i.e. , feature-assignment b y a category other tha n its source category), i t becomes clear why the above feature-spread mechanis m ma y not be attested wit h respect to a feature such as Case. Case, i n contrast to [+FOCUS], is not relevan t for (an d may even be incompatible with) V or other non-nominal heads. Conse quently, i t canno t be born e by suc h heads at S-structure , and sinc e fo r feature s which projected/spread fro m othe r head s alon g th e extended projectio n ther e i s no alternativ e o f getting assigne d t o some othe r category , th e result will necessarily b e ill-formed .
4,2, The Sources of Variation: [+FOCUS] and Nominative Case Below we present a summary of the possible source s of variation i n the S-struc ture syntax of Focus predicte d b y our hypothesis (22) , namel y a summary of the range an d types o f parameters involvin g th e wa y syntactic feature s ge t associ ated wit h th e categorie s whic h bear them at the level o f S-structure . First , how ever, w e nee d t o retur n t o th e cas e o f Focu s i n sit u a s i n th e English-typ e languages. In ou r terms, what distinguishes Focus i n sit u fro m th e phenomenon we referre d t o i n th e abov e discussio n a s designate d Focu s i s tha t th e forme r situation arise s fro m th e optio n o f th e fre e generation/occurrenc e o f th e [+FO CUS] featur e on categories, in contrast t o the latter , whic h involve s th e limita tion of [+FOCUS] to originate in and be assigned by some particular functiona l X° category o f th e clausa l projection. Late r i n thi s subsection w e wil l retur n t o the question o f the level of representation a t which this free association of [+FOCUS] wit h categories migh t tak e place. Fo r now, let us just not e that the abov e conception o f Focus i n situ i s no t unprecedented wit h respect t o other syntacti c features. Thus , eve n though Cas e i s usually thought o f as involving a particula r X° category that assigns it, there have been proposals in the literature under which a Case feature ma y arise on noun phrase s without th e mediation of an assigner. Besides so-calle d "default " Case-marking , the cas e tha t come s t o min d here i s the situatio n with the Case-particl e ga i n Japanese. Base d o n th e stud y of a va riety of asymmetries between the behavior of Nominative/ga-marked phrases an d Accusative/o-marked phrases , Sait o (1985 ) propose s tha t Nominativ e Cas e i n
Structural Focus, Structural Case, and The Notion of Feature-Assignment 5
3
Japanese is "inherent" in that it is not assigne d b y an y element (unlik e Accusative whic h is assigned by V under government) . The conclusions we have reached regarding the expected range of variation in the syntax o f Focus , derivin g fro m genera l parameter s o f syntacti c feature-as signment affectin g th e featur e [+FOCUS ] (henc e parallel , e.g. , t o th e rang e of variation involvin g Nominative Case), ar e summarized below:
(26) 1. the nature of th e feature: (i) freely occurring, i.e., vs . (ii ) assigne d by a specific X° not transferred from category ; e.g.: Nominativ e another category ; e.g. : Cas e i n English , Welsh , West Nominative Cas e (ga) i n Flemish ; [+FOCUS ] i n Japanese; [+FOCUS ] i n Hungarian , Western Bade , English Kanakur u 2. wha t X° functional category o f the clause i s the assignor , i.e. th e source o f the feature (e.g. I vs. C) 3. whethe 4. th
r the feature-assignin g category need s to be "lexicalized"
e mode/natur e of th e process o f feature-assignment: (i) feature-transfer , subjec t t o the Locality Condition s o f Horvath (1981, 1986) , namel y government and adjacency (ii) SPEC-hea d "agreement'Vrelation (se e discussio n i n 4.3. )
In view of the parallel forma l properties o f the assignment o f [+FQCUS ] an d of (structural ) Cas e feature s establishe d i n thi s study , le t u s retur n no w t o th e parallel yet distinct function o f Case vs. [+FOCUS] in our framework which we remarked o n briefly i n sectio n 1 . There we noted tha t [+FOCUS ] is a counterpart o f Case in the sens e that the functio n o f Case is t o license A-chains and th e function o f [+FOCUS] is to license a (particular kind of) A-bar chain. Notice now that i n th e cas e o f A-chains, their licensing by Case is achieve d via assignment of the Case-featur e to the head o f the A-chain. Thus , if the parallel wit h [-(-FO CUS] is taken seriously, we should hypothesize that the licensing of A-bar chains by th e [+FOCUS ] feature is also done, crucially, by assigning th e feature to th e head of the A-bar chain at S-structure. Remarkably, this hypothesis i n fact turns out to account for two otherwise puzzling generalizations that seem to be involved in the synta x o f Focus: (1) I n designated Focus languages, no phrase seems to receive the [+FOCUS] feature a t S-structur e withou t undergoing (pre-LF) movement . Even if th e [+FOCUS]-assigning hea d governs and is adjacen t to a phrase i n its basegenerated position at S-strueture, the assignment still can take place only if the phrase gets moved, as argued e.g. in relation to post-V Focus languages in Tulle r (1992) ; furthermor e the movemen t must b e t o a n (appropriatel y located) adjoine d o r other A-bar position.
54
Discourse Configurational Languages
(2) Designate d Focu s language s appea r no t t o permi t a phrase t o becom e Fo cus by means of its trace being in a position o f [+FOCUS ] assignment , i.e. , via transmissio n o f [+FOCUS ] withi n it s chain . Thi s i s illustrate d fo r in stance b y th e cas e o f Wh-extraction s i n Hungarian . A s i s commonl y as sumed, interrogativ e Wh-phrase s mus t b e [+FOCUS ] universall y (se e Horvath (1981) , (1986)) . Bu t a n extracte d interrogativ e wh-phras e seem s unable to receive th e necessary [+FOCUS ] featur e based on the position of any o f its traces even whe n the latter satisfies all conditions fo r [+FOCUS ] assignment; that is why the wh-phrase itself must be in a position o f [+FO CUS] assignmen t a t S-structure , bot h i n matri x an d i n embedde d clause s (as argue d i n detai l i n Horvat h (1981) , (1986)) . The above observation s fal l ou t naturally from our claim tha t parallel t o Case in A-chains, the [+FOCUS ] featur e in A-bar chain s i s assigne d a t S-structur e to the hea d o f the chain. This hypothesi s make s th e furthe r predictio n tha t th e as signment o f Cas e an d [+FOCUS ] t o an y particula r phrase i n th e sam e positio n will no t be atteste d i n languages , given that no phrase ca n simultaneousl y hea d both a n A-chai n an d a n A-ba r chain . I n vie w o f th e stric t complementarit y o f positions o f Nominative Case assignmen t by I and of [+FOCUS]-assignmen t b y I within , e.g. , Hungaria n or Western Bade pointed out in section 4.1 , thi s conse quence o f ou r hypothesi s appear s to b e a plausible one. Finally, notice what this conception of the function o f [+FOCUS ] implie s with respect t o the analysi s of Focus i n situ languages (like English). I n (26 ) w e pos ited th e fre e occurrence/generatio n of the [+FOCUS ] featur e on an y constituen t in such languages. But if this free acquisitio n of the [+FOCUS ] featur e by in situ phrases took place at S-structure, this would conflict with the hypothesis according to whic h th e S-structur e functio n o f [+FOCUS]-assignmen t i s th e licensin g o f A-bar chain s via their heads: in this case the [+FOCUS ] featur e would be place d on phrases a t S-structure that are not heads o f A-bar chain s until the level of L F is reached . Thi s conflic t suggests a n alternative view o f in sit u Focus according to whic h in the English-typ e languages constituents are generate d wit h the (op tional) feature [+FOCUS] alread y at th e leve l o f D-structure. Since these phrase s would carry with them the [+FOCUS ] featur e under movement, as they carry all of thei r othe r inherent features , it i s no t surprisin g tha t i n sit u Focu s languages exhibit [+FOCUS ] constituent s not only in their base-position bu t also in derived positions, a s e.g. Englis h interrogativ e wh-phrase s i n th e SPE C o f CP. 4.3. Adjacency, Feature-Transfer,
and SPEC-Head Relations
The las t issu e t o be discusse d her e ha s t o d o with th e notio n o f feature-assign ment unde r SPEC-hea d relation . I n (26 ) abov e w e state d tha t on e o f th e tw o possible alternativ e mode s availabl e for syntacti c feature-assignmen t withi n UG involves SPEC-hea d configurations ; unde r thi s strategy , a constituen t i n th e SPEC positio n o f a particular phrase receive s som e syntacti c feature o f th e X ° head o f th e phrase . Thi s i s th e wa y Nominativ e Case i s commonl y assume d t o be assigne d i n language s like e.g . Englis h (se e e.g . Koopma n an d Sportich e
Structural Focus, Structural Case, and The Notion of Feature-Assignment 5
5
(1990)) an d thi s i s th e wa y w e hav e argue d [+FOCUS ] t o b e assigned—i n th e SPEC o f IP position—in Hungarian . Now recal l tha t the othe r possibl e mode o f feature-assignment, namel y value (i) of parameter 4 in (26), cruciall y involve s a requirement o f government by and adjacency to the feature-assigning head . Thi s requirement ha s bee n observe d t o hol d wit h respec t t o [+FOCUS ] assignmen t exactly a s with respect t o structura l Case assignmen t affectin g th e complement , or a constituent withi n th e complement, o f the feature-assigning hea d (see , e.g. , Focus i n Western Bade , Kikuyu , and Accusative Cas e in English, o r Nominative Case in Iris h an d Welsh). Give n this , i t i s natural to as k (a ) whethe r an y condi tions of adjacency and/or government hold with respect t o feature assignment i n SPEC-head configuration s (i.e., fo r valu e (ii ) o f paramete r 4) , an d (b ) whethe r the answe r t o (a ) hold s th e sam e wa y fo r Cas e feature s an d fo r th e [+FOCUS ] feature. Th e issu e o f whethe r o r not a head govern s it s specifie r i s a controver sial, highly theory-dependent issue; we will assume here, followin g Koopman and Sportiche (1990) , tha t it does not . The interesting cas e i n relation t o the empiri cal materia l w e hav e analyzed abov e involve s th e issue of adjacency. Notice tha t a t least a t first glance, ther e indee d appear s t o be a curious differ ence between SPEC-hea d relation-base d assignmen t of Case an d SPEC-hea d re lation-based assignmen t o f [+FOCUS] : th e forme r (a s e.g . i n Nominativ e assignment in English) seem s to manifest no adjacency requirement, wherea s th e latter (a s e.g. [+FOCUS ] assignmen t in Hungarian ) does manifes t a n adjacenc y requirement. If this apparent discrepancy in fact turn s out to be real, it obviousl y calls fo r som e elaboratio n o r modificatio n of ou r assumption s abou t th e mode of-assignment paramete r give n in (26) . Belo w I wil l sketc h tw o alternativ e di rections t o resolv e thi s apparentl y problemati c lac k o f parallelis m betwee n th e assignment o f Cas e vs . [+FOCUS ] i n SPEC-hea d configurations . First, not e that the examples tha t seem t o constitute the main type of evidenc e for allege d non-adjacency betv/een the Nominative-assigning head (say I, or AGR) and th e subject-N P occupyin g its SPE C positio n ar e o f th e followin g kind (se e also (8 ) in sectio n 2.1) : (27) a . Joh n probably/unfortunatel y has alread y talke d t o Mary . b. Ther e obviousl y have been som e soldier s here . c. I t unfortunately may see m tha t w e are tryin g to cheat . The elements intervenin g between the Nominative-marked subject-N P and the Case-assigning—and agreement-bearing—head is usually a sentence adverb. Th e question i s whethe r suc h example s indee d exhibi t th e subjec t an d th e Case-as signing hea d i n a SPEC-head configuratio n at S-structure with the adver b inter vening between them , o r there i s a plausible alternative analysis . Belletti (1990 , 2.2) suggests—base d on some comparative English/Italian/French evidence—that subject-NP's tha t precede sentenc e adverb s in English an d Italian ar e in fact in a topicalized position , not in a SPEC position, and the adverbs ar e always adjoine d to the highest functiona l projectio n o f the clause. S o under this analysis, the as signment of Case in a SPEC-head configuratio n woul d take place under adjacency, just like th e assignment of [+FOCUS ] was shown to in Hungarian. Yet when w e
56
Discourse Configumtional Languages
consider dat a suc h a s (27b , c) , involvin g expletiv e subjects , i t become s quit e implausible t o attribut e the possibilit y o f intervenin g adverb s i n thes e case s t o having topicalized subjects. An alternative proposal which would permit th e elimination o f the apparen t non-adjacenc y involved i n example s lik e (27a-c) i s pre sented i n Kayn e (1989) . Accordin g t o thi s analysi s th e auxiliar y exhibitin g agreement—at leas t i n English—doe s no t mov e t o th e highes t functiona l hea d position (contrar y t o Polloc k (1989)) , whil e the subject-N P occupie s th e SPE C of the highest inflectional head; crucially , in Kayne's framework, no adverbs ar e generated betwee n th e highest functiona l head an d its SPEC , sinc e he limits ad verbs t o be adjoine d only t o maximal projections. Thi s proposa l woul d indee d eliminate the apparen t lack o f adjacency for SPEC-head relation-base d Nomina tive assignmen t i n English ; but , a s pointe d ou t b y Bellett i (1990) , thi s analysi s would no t b e plausibl e fo r Italian , a languag e in whic h verba l element s hav e a distribution indicativ e of raisin g t o th e highes t clausa l head , simila r t o Frenc h rather than t o English, an d which stil l exhibit s dat a o f the kin d give n i n (27). 18 At an y rate , i f on e o f th e abov e proposal s turne d ou t t o b e tenable , nothin g further woul d need to be said; the uniformity between SPEC-head relation-base d assignment of Case an d [+FOCUS ] woul d be preserved. Bu t let us consider no w the possibilit y tha t Case , i n contras t t o [+FOCUS] , actuall y ma y be assigne d i n a SPEC-head configuratio n without adjacency holding between th e two element s involved. Thi s i s i n fac t th e standar d vie w regardin g Cas e i n SPEC-hea d con figurations (see , e.g. , Rizz i 1991) , give n that it is conceived of , appropriately, a s a specific instanc e of SPEC-head agreement; since by definition, this agreemen t relation hold s simpl y by virtue of the two elements involve d bein g specifie r an d head o f a particular phrase, respectively, ther e i s n o reason t o expec t adjacenc y to pla y any role . I f so , how ca n w e accoun t for th e distinc t behavior o f the fea ture [+FOCUS ] (a s atteste d in Hungarian) , namely fo r th e fac t tha t [+FOCUSJ assignment require s adjacenc y eve n whe n i t take s plac e withi n a SPEC-hea d configuration? (Notic e tha t thi s questio n woul d hav e t o b e face d als o withi n Brody's (1990 ) Focus - Criterion-base d analysis. ) One approac h tha t w e migh t tr y i s th e following : Contrar y t o th e wa y adja cency appear s i n (26)—a s a conditio n o n valu e (i ) o f th e mode-of-assignmen t parameter—assume that it cuts across the two values of parameter 4 , so that under specifiable conditions , i t i s relevan t no t onl y fo r feature-assignmen t to/int o a complement (unde r government) but also for feature-assignment in a SPEC-hea d configuration. Specifically , suppose tha t an y proces s o f feature-transfe r fro m one category to another—as distinct from agreement—is subject to an adjacenc y condition. No w th e distinctio n wit h respec t t o adjacenc y w e observe d betwee n Case i n SPEC-hea d configuration s vs. [+FOCUS ] i n SPEC-hea d configuration s can b e attribute d to th e presenc e o f agreemen t feature s (i.e. , phi-features , "car rying" Case ) o n th e Case-assignin g hea d i n language s like Englis h vs . th e ab sence o f agreement feature s (involving th e element i n the SPE C position ) o n the [+FOCUS]-assigning head . Th e presenc e o f agreemen t feature s permit s Case assignment unde r agreement, henc e n o feature-transfe r is needed , an d conse quently ther e i s n o adjacency-effect . I n contrast , th e lac k o f agreement-features relevant fo r th e Focu s constituen t (at least i n Hungarian ) leaves onl y the option
Structural Focus, Structural Case, and The Notion of Feature-Assignment 5
7
of feature-transfer fo r the assignment of [+FOCUS ] i n a SPEC-head configuration, henc e th e necessar y adjacenc y betwee n th e feature-assignin g hea d an d it s specifier. This accoun t presupposes tha t the fact that two elements ar e in a SPEC head relatio n doe s no t impl y tha t the y automaticall y com e t o shar e al l features : thus i n contrast t o e.g. phi-features , th e [+FOCUS ] featur e i s crucially assume d here not to be shared . A case o f another feature, providing independent evidenc e for th e non-automatic nature of "feature-spreading" i n SPEC-head configurations , is th e cas e o f th e +W H feature . Assumin g th e framewor k o f th e Wh-Criterio n developed i n Rizzi (1991) , notice that in English, the presence o f a wh-operato r in th e [SPEC,CP ] positio n i s unable to provid e C wit h the featur e +WH , i.e. , i t manifests n o feature-assignmen t unde r SPEC-hea d agreement . Thi s i s wh y th e matrix I carryin g th e +W H feature need s t o mov e int o C for satisfactio n o f th e Wh-Criterion. I n contrast, exactl y this SPEC-hea d agreemen t process fo r the +WH feature (referre d to by Rizzi as "dynamic agreement" ) doe s exist i n French. Thu s the cas e o f +WH in Englis h matri x CP's demonstrate s tha t feature s d o not nec essarily "spread " automatically withi n SPEC-hea d configurations . As for feature-assignment unde r governmen t (i.e. , to/into a complement), i t can onl y b e a process o f feature-transfe r sinc e n o "head-complement " agree ment mechanis m exist s i n th e theory ; consequentl y thi s typ e o f feature-assign ment configuratio n is correctl y predicted i n our framewor k to alway s exhibit a n adjacency-effect. To summarize now th e implication s of the abov e proposal fo r paramete r 4 in (26), i t actually would spli t into two separat e parameters , call them 4 and 5, respectively: (28) 4 . th e configuratio n of assignment : (i) unde r governmen t (ii) unde r SPEC-hea d relatio n 5. th e mod e o f assignment : (i) feature-transfer , subject t o a n adjacenc y requiremen t (ii) (dynamic ) agreemen t [n o adjacency relevant; availabl e only fo r assignment s under SPEC-head relation ] In thi s framework , [+FOCUS] i n Hungaria n would involv e valu e (ii ) o f th e revised paramete r 4 an d valu e (i ) o f paramete r 5 . Nominativ e Cas e i n Englis h would als o involv e valu e (ii ) o f 4 , bu t i t woul d hav e valu e (ii ) o f paramete r 5 . This differenc e i n terms o f 5 would be the sourc e of the presence vs . absenc e of adjacency-effects. Th e abov e conceptio n o f feature-assignmen t an d th e SPEC head relatio n predict s tha t i n principl e ther e ma y als o b e instance s o f Nomina tive Cas e assignmen t whic h ar e lik e [+FOCUS ] i n Hungaria n wit h respec t t o parameters 4 and 5, namely, which tak e place i n a SPEC-head configuratio n ye t involve n o agreemen t bu t instea d feature-transfe r (see valu e (i ) o f paramete r 5 in (28)) . Such a process o f Nominative assignment is not easy t o come across ; it is unlike the familiar case s of Nominative assigned by SPEC-hea d agreement , as in English , and i t i s als o unlik e e.g. th e cas e o f Standar d Arabic, wher e th e ab sence o f agreemen t o n I automaticall y implies Nominativ e assignmen t under
58
Discourse Configurational Languages
government (i.e . int o th e complemen t o f I) . Ye t there i n fac t i s a n exampl e o f Nominative Cas e assignmen t discusse d i n th e literatur e tha t ha s precisel y th e relevant properties. Koopma n (1983 , Ch . 5) provides a n analysis o f Nominativ e Case assignmen t i n Vata arguing (a ) that Nominativ e i s assigne d b y a (verbal ) tensed I node in a SPEC-head configuration , (b) that I in Vata has no AGR ele ment, and it exhibits n o agreement, an d (c) that no adverb or any other elemen t may intervene between a lexical subjec t and the Case assignin g I node, i.e., stric t adjacency i s required between the two. This case of Nominative assignment pro vides strikin g confirmatio n for th e vie w o f configuratio n an d mod e o f feature assignment proposed in (28) , a s well as for our analysi s of structural Focu s an d its range of variation in terms of a syntactic feature, predicting parallelism s wit h the forma l properties an d rang e of variatio n o f (structural ) Case .
5. Concluding Remark s In the foregoing discussion w e have explored wha t kind of account may be bes t suited t o captur e th e substantia l rang e o f variatio n atteste d i n language s with respect t o the S-structure syntax of identificational Focus. Base d on comparativ e evidence involvin g the location s o f S-structur e focus position s (i.e . th e variou s attested landin g site s fo r "focus-movement" ) withi n th e se t o f language s wit h "designated" Focus positions, it has been argued that a simple two-valued S-structure vs. LF parametrization involving the SPEC position o f a category F P (suc h as e.g. Brody's (1990) parametrize d FOCUS-Criterion proposal ) canno t account for th e observed range of phenomena, at least no t within any reasonably restric tive framework of assumptions. Furthermore, w e have presented evidenc e dem onstrating tha t once w e make the minima l assumption—adopte d als o b y Brod y (1990), a s well as by other previou s accounts , from Horvat h (1981; 1986)—tha t there i s a syntactic feature [+FOCUS], withou t th e presenc e o f whic h n o identificational Focu s interpretatio n i s possible, th e observe d phenomen a o f S structure Focus fal l ou t of independently motivated properties an d parameters of syntactic feature assignment. This claim has been supporte d empirically by demonstrating the existence of a point-by-point parallelism across language s between the formal options/dimensions of variation found wit h respect to the way our hypothesized [+FOCUS ] feature gets associated with constituents, and those known from analyse s o f a "classical," well-establishe d syntacti c featur e suc h a s struc tural Cas e (i n particular, the uncontroversially not lexically-relate d Nominativ e Case). As emphasized at the outset, the claim here involves parallelism onl y with respect to purely formal properties and associated parameters ; the role, interpre tation and range of occurrence of our syntactic [+FOCUS ] featur e vs. structura l Case feature s are radically different . Thi s i s precisely th e typ e o f situatio n on e can expec t withi n a modula r theor y o f grammar , wher e phenomen a aris e fro m the interactio n o f a complex arra y o f autonomou s sets o f principles . The evidenc e presented in thi s study show s that th e reaso n fo r th e existenc e of S-structur e "designated" Focus positions in language s is purely syntactic, in the narro w sens e o f th e term . That is, th e nee d fo r over t movemen t int o a n S -
Structural Focus, Structural Case, and The Notion of Feature-Assignment 5
9
structure A-bar position specifi c to FOCUS i s not du e to "early" satisfaction o f some fundamentally LF-induced requirement suc h as scope assignment, contrar y to claims t o this effec t i n previous studies . Thi s conclusio n i s implie d firstl y b y discrepancies foun d i n "designated" Focu s language s between th e position whic h the (moved ) Focu s phras e mus t occup y a t S-structure an d the LF-positio n fro m where i t takes scop e (se e sectio n 3.1) . (Thi s type of fact provides motivatio n fo r LF movemen t fro m A-ba r t o A-ba r position. ) Furthermore , th e non-LF-relate d nature of the phenomenon o f designated S-structure Focu s position s i s confirmed by th e significant range of variation in terms of location i n hierarchical structur e exhibited b y suc h position s across , an d eve n within , languages , a s w e demon strated in the foregoing sections. Notice that if one maintained that such S-struc ture Focu s position s reflec t th e locatio n o f Focu s i n L F representations , n o uniform LF-representation coul d be postulated for identificational FOCUS i n UG. This i n turn woul d go agains t the well-motivate d an d widel y accepte d assump tion tha t th e L F componen t is ; not subjec t t o language-particula r variation , du e to th e unavailabilit y of appropriat e evidenc e fo r acquisitio n i n th e primar y lin guistic data. Regarding th e case a t hand, note th e problem tha t a child acquirin g an i n sit u Focu s languag e (lik e English) woul d fac e whe n tryin g to "discover " the LF-representatio n fo r identificationa l Focu s i n his/her languag e withi n thi s non-uniform LF framework. Hence we have good reasons to maintain that S-struc ture designated Focus position s ar e not (necessarily ) reflections o f LF-represen tations, but have independent origins/motivation i n requirements o f overt syntax , involving grammatica l formatives, suc h as the syntactic featur e [+FOCUS] . It is als o wort h pointin g ou t i n thi s contex t tha t no t onl y scop e assignmen t fail s to provide the motivation for overt movement into an S-structure Focus position ; neither ca n suc h movement s b e attribute d to needs o f "clausal typing, " o r satis faction o f selectiona l requirements , a s ha s bee n propose d fo r th e superficiall y similar case of Wh-interrogative movement (see e.g . Chen g (1991)). Clauses con taining identificational Focus vs . those withou t such Focu s d o not play distinc t roles i n selectio n fo r clause-typ e o r i n an y othe r respec t involvin g th e whol e clause, i n contrast t o interrogative vs . declarativ e clauses , whic h do . The present study leaves open a number of issues related to the syntax of Focus, such a s for instance the mechanis m by whic h constituents containe d withi n th e [+FOCUS]-marked phrase occupying the "designated" Focu s positio n may receiv e Focus interpretation . Obviously som e proces s o f percolatio n woul d hav e t o b e invoked, but it s nature—in light of properties motivate d fo r other featur e perco lation processes (e.g . percolation in pied-piped wh-phrases)—an d its consequence s for both the PF and LF components await further investigation . Similarly , we have not discusse d her e th e potentia l existenc e o f multipl e Focu s construction s i n designated Focu s languages , involving all but one Focus constituent i n situ. What the framework of assumptions outlined in this stud y seems t o predict i s that des ignated Focu s language s wil l tur n ou t no t t o hav e tru e identificationa l Focu s interpretation—like a moved Focu s constituent does—for i n situ phrase s i n thei r alleged multipl e Focus constructions. A furthe r issu e regardin g ou r conceptio n o f [+FOCUS ] assignmen t may b e raised i n ligh t o f recen t proposal s involvin g th e reanalysi s of al l instance s of structural Cas e "assignment " a s case s o f Cas e checkin g takin g plac e withi n
60
Discourse Configuratianal Languages
SPEC-head configurations . Give n the formal parallels i n distribution and range of variation between the two syntactic features established i n this study, it is not inconceivable tha t further , deepe r investigatio n o f th e hierarchical arrangemen t and interaction among the full rang e of functional categories an d their domain s of "checking" in the clausal projectio n o f designated Focus languages wil l make such a reductio n empiricall y feasibl e wit h respect t o [+FOCUS ] a s well . This would mean the elimination o f feature-assignment unde r government (Le. , value (i) of our parameter 4 "configuration o f assignment" i n (28) above) , as well a s the elimination of the option o f feature-transfer subject t o adjacency (i.e., value (i) o f ou r "mod e o f assignment " paramete r 5 in (28)) , Yet before an y such re analysis ca n b e considere d seriously , on e mus t recognize th e existenc e o f a n important differenc e between various feature-acquisition/licensin g processes , pointed out in section 4.3 . Som e instances of alleged feature-checkin g processe s within SPEC-head configurations seem to involve a requirement of strict adja cency—as e.g. [+•FOCUS ] in Hungarian, as well as Nominative Case in Vata (se e 4.3), wherea s othe r feature-checkin g processe s d o permi t som e element s (adverbials and/or parentheticals) intervening between the feature-checking head and th e SPEC positio n t o b e checked—as e.g. [+WH] checking in SPE C o f CP in English (as in our example (7)) , as well a s Nominative Case in English (a s in examples (27 ) discusse d i n 4.3). I n the absenc e o f a n independent explanation , this variation with respect t o the requirement of strict adjacency seem s to undermine the plausibility of a full reductio n of all "feature-licensing" processe s to a single (SPEC-head ) checking mechanism. 19 Finally, it is worth noting that in contrast to most account s adopting the ide a of [+FOCUS ] assignment/checking (includin g those i n the present volume) , ou r proposal identifies the source/assigner of the [+FOCUS] feature in UG not as one single X°(= F ) category ; rathe r i t assume s tha t an y functional hea d withi n th e clausal projectio n has the potential t o be affiliated wit h the featur e [+FOCUS]. Which one(s) can actually bear the feature in any particular languag e is a matter of parametric variation (see our parameter 2 in (26)). A t the same time, this "parasitic" affiliation o f the [+FOCUS] feature with a variety of distinct head s raise s some interesting conceptual questions regarding th e role of differen t type s of syntactic features in relation t o the notion "category" which are beyond the scope of the present study. On the other hand, the primary empirical motivation for this assumption, as argued in the study, is quite clear-cut : th e specific cross-linguis tic variation found wit h respect to the hierarchical location of S-structure Focus positions. Thoug h one may attempt to develop som e alternativ e wa y of captur ing this variation so as to avoid the assumption of multiple sources/assigners fo r the [+FOCUSJ feature, one of the "designated" Focus language s studied, namely Hungarian, turns out to provide som e prima facie direct evidence suggestin g the correctness o f the multiple source hypothesis. Consider th e following sets of data: (29) (Csak) JANOS nem AZ (JJSAOOT dobta el. only JOHN-NO M no t TH E NEWSPAPER-AC C thre w awa y ('the (only ) perso n suc h that it's not THE NEWSPAPER that he threw awa y is JOHN" )
Structural Focus, Structural Case, and The Notion of Feature-Assignment 6
1
(30) a . *(Csak) JANO S A Z UJSAGO T dobt a el . only JOHN-NO M TH E NEWSPAPER-AC C thre w awa y ('the (only ) person suc h that it' s TH E NEWSPAPER that h e threw awa y is JOHN') b. *Ne m JANOS (csak)A Z tJJSAGO T dobt a el . not JOHN-NO M (only) THE NEWSPAPER-ACC thre w away The unacceptability of (30a ) confirms the prediction o f our basic account: th e designated Focu s position , whic h we claime d t o b e th e SPE C o f I P positio n i n Hungarian, ca n accommodat e onl y a singl e phrase , an d n o furthe r (adjunction ) Focus movemen t i s possibl e No w i n ligh t o f this , th e ful l acceptabilit y o f th e "multiple Focus movement" sentence in (29) may seem unexpected. Yet the con trast betwee n (29 ) an d (30a ) fall s i n plac e naturall y within ou r conceptio n o f [+FOCUS] assignment . Notice th e presenc e an d locatio n o f th e negativ e mor pheme new, in (29) . Arguably , this elemen t i s th e hea d o f Neg P i n Hungarian (also observe the ungrammaticality of (30b), whic h indicates that nem is not som e particle o f phrasa l negatio n tha t induce s Focu s interpretation) . Give n this , th e obvious conclusion is that not onl y I but als o Neg is a source/assigner o f [+FO CUS] (fo r it s SPEC ) i n Hungarian . In contrast , theorie s o f Focu s postulatin g a unique categor y F fo r [+FOCUS ] woul d see m t o hav e n o accoun t fo r th e para digm (29 ) vs . (30) .
Notes 1. The typ e of Focus w e are concerned with here is "identificational" i n the sens e tha t it implie s exhaustiv e listing . Instance s o f suc h Focu s ar e manifeste d e.g. b y th e clef t construction o f English, Wh-questions, and their "natural" answer s (see Horvath (1981) , (1986)), a s wel l a s by "ore/>>"-phrases . 2. Lasnik an d Stowell (1991) sugges t tha t what makes Focu s phrase s behav e lik e tru e quantifiers i s tha t "the y contai n a cover t operato r 'only.'" Evidenc e fro m Hungarian "on/;y"-phrases indicates that the presence o f this morpheme itsel f i s well-forme d only if the phras e receive s th e [+FOCUS ] feature we ar e postulating. 3. Brody's (1990 ) discussio n of his condition (b) fails to make clear whethe r he really means FP (as stated), or perhaps [SPEC,FP ] instead , on the analogy of the Wh-Criterion . Which o f thes e tw o possibilitie s i s adopte d makes no differenc e for ou r discussion ; th e choice i s importan t though with respec t to the LF representation assigned t o i n sit u Fo cus phrase s in multiple-Focu s sentences . 4. A relevant issue not addressed by Brody is how phrases such as a moved (interrogative) Wh-phras e bein g i n SPEC ! of C P would receiv e th e [+FOCUS ] featur e in English . He state s that free assignment of this feature is possible onl y i n the "prepositiona l part " of th e sentence, whic h for him i s VP in Hungarian and IP in English. This woul d i n turn necessitate the existence of some special [+FOCUS]-assigne r fo r SPEC of CP in English, which i s a problemati c consequenc e i n hi s framework , especiall y fo r th e cas e o f wh phrases in embedded questions . 5. What we argu e here i s no t tha t there coul d b e n o functiona l hea d F (projectin g an FP), o r that there could be no Focus-Criterion (though it does make such proposals much less motivated) , but rathe r that the rang e and type s of variatio n we observ e canno t and
62
Discourse Configurational Languages
should not be attributed to these, because there is a more adequate account available, based on the independently needed featur e [+FOCUS ] i n conjunction with independently motivated parameter s o f feature-assignment. 6. Evidenc e fo r thi s V-preposin g i s provide d b y th e post- V positio n o f th e otherwis e preverbal incorporate d particles, suc h as e.g. el= 'away ' i n examples (4b ) vs. (4d) . 7. For a discussion of possible alternativ e accounts o f cases of apparent non-adjacency as i n (8) , se e sectio n 4.3 . 8. In Brody' s framework, phrases i n A-positions receiv e th e [+FOCUS ] featur e freel y in Hungarian , as the y d o i n th e English-typ e languages . 9. An are a tha t appear s interestin g an d challengin g fro m th e poin t o f vie w o f th e al leged rol e o f V is Focu s withi n infinitival clauses. W e will no t tak e u p thi s topi c in th e present stud y howeve r sinc e i t woul d lea d u s to o fa r int o a discussio n o f th e synta x of Hungarian. 10. The issu e o f whic h functiona l head(s ) may assig n th e [+FOCUS ] featur e wil l b e discussed i n sectio n 4 . A proposal simila r to ou r "lexicalization" hypothesis, namely , Vraising int o I t o make i t "verbal " i n order to enabl e the latte r t o assig n FOCUS , i s pre sented als o i n Tuller (1992 ) fo r post-V Focu s i n som e Chadi c languages . 11. The reason why we specif y "e x situ " here is that even designate d Focu s language s have in sit u phrases tha t come t o be interpreted a s Focus: these occur i n multiple Focu s sentences, whic h appea r t o b e analogou s t o multipl e Wh-question s i n language s lik e English. W e will no t discus s such constructions i n thi s study . 12. Following Walter s (1979) , P 2 in the glosses mean s a particular Tense marker , an d SM stand s for "subjec t marker. " 13. In the glosses SP= subjec t prefix , PP=pronominal prefix , CP= nomina l clas s prefix , and T= tense/aspec t affix . 14. Neither coul d ne b e claime d t o b e a n F tha t ha s undergon e raisin g int o C (thu s accounting fo r it s pre-Focu s S-structur e position); Clements (1984 ) provide s example s within which ne occurs i n an embedded claus e following an independent, lexically fille d C position . 15. Notice also tha t trying to claim that examples lik e (17 ) involv e the Focusing o f the whole IP would be problematic as well. I f the IP were Focused, ne would be expected t o occur t o th e lef t o f th e whol e IP, i.e., t o the lef t o f th e subject-NP . 16. A similar proposal—involving V-raising to I—is made with respect to Nominativ e Case assignmen t in Vat a as wel l a s i n Dutc h by Koopma n (1983). 17. Confirming evidence tha t th e Focu s positio n in Hungaria n is indee d a n A-bar po sition a t S-structure comes fro m th e fac t tha t phrases in the Focus positio n o f Hungarian license parasiti c gaps , a s demonstrated in Horvath (1987 , sectio n 2.3) . 18. O f course , du e t o th e fac t tha t Italia n ha s nul l expletives , example s lik e (27b , c ) are no t testabl e i n this language . So there stil l i s som e chanc e tha t Belletti's analysi s is tenable fo r Italian. 19. The present paper was completed before Chomsky's (1992) " A Minimalist Progra m for Linguisti c Theory" appeared . The analysi s presented her e i s incompatibl e wit h thi s latter mor e restrictiv e framework in obviou s ways. It is fairl y clea r what the genera l di rection o f a n attemp t t o recas t th e accoun t o f Focu s propose d her e i n term s o f thi s minimalist framework would be; it may involve the postulation o f a "strong" vs. "weak" distinction fo r ou r [+FOCUS ] feature , and motivat e movemen t b y th e nee d fo r feature checking, o n the analogy of Chomsky' s proposa l for Case an d wh-movement (see a suggestion to this effect i n Chomsky (1992, p. 45)). Yet given that the "minimalist program" seems to imply the same account for the syntax of Focus a s for several other A-bar movement constructions, such as wh-interrogatives and topicalization, one would need t o ex -
Structural Focus, Structural Case, and The Notion of Feature-Assignment 6
3
plore how the rather striking difference s observable in the properties an d cross-linguisti c variation of these constructions ca n be derived before such a mechanical extension of the wh-movement accoun t is adopted. This latte r task is undertaken in research i n progress .
References Belletti, A. (1990 ) Generalized Verb Movement, Rosenber g & Sellier, Turin . Borer, H . (1986) "I-Subjects, " Linguistic Inquiry 17 , 375-416. Brody, M. (1990) "Remark s o n the Order o f Elements i n the Hungarian Focus Field, " in I. Kenesei , ed. , Approaches to Hungarian vol . 3. , JATE, Szeged . Burzio, L. (1981 ) Intransitive Verbs and Italian Auxiliaries, Doctoral dissertation , MIT , Cambridge, Massachusetts . Cheng, L . (1991 ) On the Typology of Wh-Questions, Doctora l dissertation , MIT , Cambridge. Chomsky, N. (1976) "Condition s on Rules of Grammar," Linguistic Analysis 2 , 303-351. Chomsky, N. (1981 ) Lectures on Government and Binding, Foris , Dordrecht . Chomsky, N . (1986 ) Barriers, MIT Press, Cambridge , Massachusetts. Chomsky, N. (1992 ) " A Minimalist Program fo r Linguistic Theory," i n MIT Occasional Papers in Linguistics 1 , Cambridge, Massachusetts . Clements, G . N. (1984) "Bindin g Domain s in Kikuyu," Studies in the Linguistic Sciences 14.2. Farkas, D . (1986) "O n the Syntactic Position o f Focus i n Hungarian," Natural Language and Linguistic Theory 4 , 77-96. Fassi Fehri , A . (1989 ) "Generalise d I P Structure, Case , an d VS Word Order," i n I. Lak a and A. Mahajan, eds. , Functional Heads and Clause Structure, MI T Working Pa pers i n Linguistics vol . 10 , Cambridge, Massachusetts . Grimshaw, J. (1991 ) "Extende d Projection, " ms. , Brandeis University, Waltham, Massachusetts. Horvath, J. (1981 ) Aspects of Hungarian Syntax and the Theory of Grammar, Doctora l dissertation, UCLA, Los Angeles . Horvath, J. (1986) FOCUS in the Theory of Grammar and the Syntax of Hungarian, Foris , Dordrecht. Horvath, J . (1987 ) "O n Model s wit h a VP-less Phras e Structur e an d Two (A)symmetry Phenomena," i n I. Kenesei, ed. , Approaches to Hungarian vol . 2. , JATE, Szeged . Kayne, R. (1989 ) "Note s o n English Agreement," ms. , CUNY , New York. Koopman, H. (1983 ) The Syntax of Verbs: From Verb Movement Rules in the Km Languages to Universal Grammar, Doctoral dissertation, McGill University , Montreal. Koopman, H. , an d D . Sportich e (1990 ) "Th e Positio n o f Subjects, " ms. , UCLA , Lo s Angeles. Kuroda, Y.-S. (1988 ) "Whethe r W e Agree or Not: A Comparative Syntax of English an d Japanese," Lingvisticae Investigationes 12 . Lasnik, H., and T. Stowell (1991 ) "Weakes t Crossover, " Linguistic Inquiry 22 , 687-720. May, R. (1985 ) Logical Form, MI T Press, Cambridge , Massachusetts . Ortiz de Urbina, J. (1986) Some Parameters in the Grammar of Basque, Doctoral disser tation, Universit y of Illinois , Urbana-Champaign. Pollock, J.-Y . (1989 ) "Ver b Movement , UG an d th e Structur e of IP," Linguistic Inquiry 20, 365-424 . Rizzi, L. (1991) "Residual Verb Second and the Wh-Criterion," ms., University of Geneva, Geneva.
64
Discourse Configurational Languages
Saito, M . (1985 ) Some Asymmetries in Japanese and Their Theoretical Implications, Doctoral dissertation , MIT, Cambridge , Massachusetts . Travis, L. (1984 ) Parameters and Effects of Word Order Variation, Doctoral dissertation , MIT, Cambridge , Massachusetts . Tuller, L . (1988 ) "Th e Synta x o f Postverba l Focu s Constructions, " ms. , Universit y o f Leiden, Leiden . Tuller, L . (1992 ) "Th e Synta x of Postverba l Focu s Construction s in Chadic, " Natural Language and Linguistic Theory 10 , 303-334. Walters, J. (1979 ) "Focu s in Aghem," in L. Hyman, ed., Aghem Grammatical Structure, Southern Californi a Occasiona l Papers in Linguistics #7, Lo s Angeles.
3 Aspects of Discourse Configurationality in Somali MARCO SVOLACCHIA LUNELLA MEREU ANNARITA PUGLIELL I Dipart. di Linguistica — Terza Universita degli Studi di Roma
1. Introduction1 Focus an d t o som e degre e topi c ar e central t o th e synta x of Somal i a s ther e i s obligatory focu s o f on e element i n every main , declarative clause. Unlike th e focused ver b i n Somal i th e focuse d N P i s syntactically marked i n tha t i t mus t occur i n a specific position . Th e focused elemen t i s also lexically marked by an identifying particle , referre d t o a s focus (indicator) particle i n th e literature . Owing to its obligator y syntactic an d lexical indicatio n o f focus, Somali ca n be called & focus prominent language. I t can als o b e considered a topic prominent language (L i an d Thompson, 1976 ) i n tha t topic i n th e for m o f N P dislocatio n and resumptive pronoun s i s a primary featur e of Somali. The ai m of this pape r is to give an overview of the main aspects o f discours e related phenomena in Somali, investigating some basic relations between their semantics and syntax . The specifi c questions w e will address concern : (1) th e syntacti c categor y an d the structura l position s whic h focu s an d topic constituent s belong to; (2) th e relation betwee n syntacti c focalizatio n an d its semantics ; (3) th e role an d syntactic : properties of different kind s of focus constructions: nominal focu s structure s (containin g baa/ayaa), th e so-calle d verbal focalizatio n (th e type with waa), a s well a s cataphoric focal ization (th e waxaa sentences); (4) th e relation between topic and focus structures and the occurrence of subject resumptive pronouns. 65
66
Discourse Configurational Languages
It wil l b e show n tha t bot h topi c an d focu s nomina l structure s occup y [Spec , CP], thoug h fo r topic s thi s i s a base-generated position . Th e discours e rol e o f topic an d focu s NP s i s associate d wit h th e presenc e o r absenc e o f th e focu s marker i n Comp. Thoug h th e Somal i dat a ar e compatible bot h wit h a n interpre tation o f focu s a s a syntacti c featur e associate d wit h som e categor y (Horvath , 1986, an d this volume) o r as a function featur e within a focus projection (Brody , 1990), w e wil l adop t th e forme r view , locatin g th e focu s particl e i n th e Inf l node fro m whic h i t undergoes movemen t t o Com p i n the case of nominal focal ization (Lecarme , 1991) . We propos e tha t post-verba l waxaa focu s structure s ar e generate d a s righ t adjoined t o IP. Further, we will show that subject resumptive pronouns in Somali occu r when ever a subject is either base generated in A'position, eithe r a s an initial, interna l or post-verbal topic , or moved acros s a CP node. The y ar e ungrammatical when the subjec t is eithe r i n [Spec , IP] , o r moved t o th e C P focus position . An interestin g resul t achieve d b y th e analysi s w e hav e develope d i s tha t i n Somali ther e i s n o nee d t o stipulat e a specifi c orde r betwee n topi c an d focu s constituents. A consequence i s tha t C P in Somal i mus t b e recursiv e sinc e bot h topic an d focu s NP s ar e represented there . This articl e i s organize d a s follows. Section 2 presents basi c fact s abou t So mali syntax . Sectio n 3 , i n whic h som e pas t proposal s ar e briefl y reviewe d an d new one s ar e advanced , deals wit h topic an d focus, an d presents th e theoretica l framework an d the hypotheses leading to ou r analysi s of discours e relate d fact s in Somali . Finally , Sectio n 4 summarize s an d comment s o n th e mai n result s o f the discussio n at thi s stage o f our research .
2. Basics In thi s sectio n w e ar e going to review som e aspect s o f Somal i whic h constitut e the minima l backgroun d fo r the forthcomin g discussion . As mentione d above , Somal i require s tha t i n ever y mai n sentenc e a focu s particle be present. One can differentiate betwee n "nominal " an d "verbal" focalization, dependin g on the for m o f the focu s marker : baa/ayaa an d waxaa2 indi cate nominal focus, whil e waa indicates verbal focus, a s seen in (la, c ) and (Ib): (1) a . naa g *(baa ) libaa x aragta y woman F M lio n sa w 'A WOMAN ha s see n a lion' b. Cal l moo s *(waa)-u u cunay C. banan a FM-h e at e 'Cali HAS EATEN a banana' c. Cal i *(waxaa)-u u cuna y moo s C. FM-h e at e banan a 'Cali ha s eate n a banana' 3
Aspects of Discourse Configurationality in Somali 6
7
Each o f the above sentence s i s ungrammatical without its FM. Baa i s right-adjacent t o th e N P i t marks , waa i s left-adjacen t t o th e Verba l Comple x (se e be low), while th e elemen t focuse d by waxaa i s located afte r th e verb . No F M ca n occur i n embedde d o r untensed clauses (imperative , jussive, potential, etc.) , an d only one focus is license d i n an y one sentence: 4 (2) a . Cal i ba a sheega y i n Muus e (*baa/*waa ) yimi d C. F M sai d tha t Muus e FM s cam e 'CALI said tha t Muuse has come' b. run-t a (*baa/*waa ) sheeg ! truth-the F M F M tel l 'Tell the truth' +WH element s i n Somal i mus t be focused , eithe r b y baa, o r waa o r waxaa (see below , Section 3) ; this entail s tha t multiple WH-questions canno t occu r i n Somali. Clitic resumptive pronouns ar e a majo r featur e i n Somal i syntax . Objec t resumptive pronouns corresponding to argument s must alway s occu r (wit h 3r d pers. pronoun=0) irregardles s of the presence an d position o f the full coindexe d object NPs . Consider th e following examples: (3) a . (aniga ) shala y ba a la- y dila y me yesterda y F M IMP-m e bea t 'Someone beat m e YESTERDAY' b. shala y ba a (aniga ) la- y dila y yesterday F M rn e IMP-m e bea t 'Someone beat me YESTERDAY' c. shala y ba a la- y dila y (aniga ) yesterday F M IMP-m e bea t m e 'Someone beat me YESTERDAY' d. anig a ba a shala y la- y dila y me F M yesterda y IMP-m e bea t 'Someone beat M E yesterday' In (3d) the object N P is focused, whil e in (3a-c) it is not, occupying all possibl e positions i n th e sentence . I n al l cases , th e objec t resumptiv e pronoun s mus t occur. On th e contrary , the condition s tha t rul e th e occurrenc e an d positio n o f th e subject resumptive pronoun ar e fairly complex , depending on the presence, posi tion an d natur e (i.e. lexica l vs . pronominal NP) o f their antecedent : (a) th e subject resumptive pronoun cannot occur when the subject NP either i s focused o r i s la, a n impersonal clitic:
68
Discourse Configurational Languages
(4) a . wiilal-ki i baa/*ba y moo s cunay d boys-the F M FM-the y banan a eatin g 'THE BOY S ar e eating a banana' b. moo s baa/*bu u l a cunaya a banana F M FM-h e IMP eatin g 'Someone i s eatin g A BANANA' Note that the subject resumptive pronoun cliticizes o n the FM, when present, and a contractio n take s plac e (e.g . baa + uu 'he ' yield s buu; baa + ay 'they , she ' becomes bay). (b) th e subjec t resumptiv e pronoun i s optional whe n a nominal subjec t come s after th e F M an d before th e Verba l Complex : (5) moo s baa/bu u wiilki i cunaya a banana F M FM-he boy-th e eatin g 'The boy i s eatin g A BANANA' (c) th e subject resumptive pronoun is, finally, obligatory i n all other cases (not e 6c, i n whic h th e subjec t i s a n independent pronoun) : (6) a . wiilk i moo s *baa/bu u cunaya a boy-the banan a F M FM-h e eatin g The boy , he i s eating A BANANA' b. moo s *baa/bu u cunaya a wiil-k i banana F M FM-h e eatin g boy-th e 'He is eatin g A BANANA, th e boy' c. moo s * baa/ baan anig u cuna y banana F M FM- I I at e 'I ate A BANANA' The distribution of the subject resumptive pronouns will be derived i n §3.4-5. Two main structura l cases ar e recognizable i n Somali: nominative , the case of the subject , an d accusativ e (o r rathe r non nominative), whic h i s th e unmarke d case (bot h morphologically an d syntactically). A third case, genitive, ha s no rel evance t o this paper. Subjec t marking is fairly complex , involvin g both segmen tal and prosodic elements. 5 A striking feature of Somali i s that the subject marke r appears a t the en d o f th e maximal subjec t phrase , e.g. : (7) a . nin-k w man-the(S) 'The man' b. nin-k a iy o gabar-t w man-the an d woman-the(S ) 'The ma n an d the woman '
Aspects of Discourse Configurationality in Somali 6
9
c. [nin-k a imanayaa ] wa a walaal-ka y man-the coming(S ) F M brother-m y "The man who is coming i s m y brother'6 Subject focusin g entail s a numbe r of consequences : First, a focuse d subjec t NP loses it s subjec t mar k (8b) . Secondly , it s resumptiv e pronou n canno t occu r (8c). Thirdly , th e agreeing ver b has a special type of conjugation, th e restricted paradigm (ReP) , characterize d essentiall y by the reductio n o f som e agreemen t features o f the normal ("extended") paradig m (8d). 7 These facts ar e exemplifie d below: (8) a . niman-kaa s ba a hili b cunay d men-those F M mea t eating(ReP ) 'THOSE MEN are eating meat * b. * niman-kaasu ba a hili b cunaya * men-those(S) F M mea t eating(ReP ) c. * niman-kaas ba y hili b cunaya " men-those F M mea t eating(ReP ) d. * niman-kaas ba a hili b cunayaa n men-those F M mea t eatin g The sam e propertie s whic h characteriz e subjec t focusin g als o characteriz e sentences i n whic h a subjec t W H element, either as a n interrogative o r a relative, occurs : (9) a . max-a a dhacay ? what-FM happened(ReP ) 'WHAT'S happened?' b. *mux-H « dhacSy ? what-FM-he happene d c. nin-ka j [0( 0 n a t. dhegeysanaya l baana n aqoo n man-the W H thai : u s listening(ReP ) FM-not- l kno w 'I do not know THE MEN WHO ARE LISTENING T O US' d. * ninka; [0 ; 0 a y ts na dhegeysanaya-a ] baana n aqoo n man-the W H tha t they us listening(ReP)-(S ) FM-not- i kno w e. [Cal i iy o Maxamed j [0i o o jaamacadd a k a shaqeeya]-a] wa a C. an d M. W H tha t university i n work(ReP)-(S ) F M walaalo brothers 'Cali and Maxamed, wh o work a t the university , ar e brothers ' f. *[Cal i iy o Maxamed j [0{ o o a y jaamacadd a k a shaqeeyaan] ] C. an d M . W H tha t the y universit y i n work(3p)(S ) waa walaal o FM brother s
70
Discourse Configurational Languages
As the examples abov e show , the subject resumptive pronoun cannot occu r (9b) ; the subjec t marke r i s droppe d (9d) ; th e ver b ha s a restricted paradigm endin g (9f). Note that in relative clauses (bot h appositive, displaying the complementize r oo (9e) , an d restrictive, withou t an overt on e (9c) ) th e WH operato r (subjec t i n the above examples) i s alway s null . There i s genera l acceptanc e tha t Somal i i s a SOV language. Basi c wor d or der, however, canno t be directly deduce d from th e order o f full NPs , since focalization an d topicalization largel y determin e thei r position an d thus th e apparen t free wor d order. So , a simple two-argumen t sentence meaning roughly "Cali at e a banana " coul d b e translate d in Somali with , at least , th e followin g sentences , all of which are grammatical (in the translation the focused elemen t i s written in capitals, whil e the topic is italicized) : (10) a . Cal i moo s bu u cuna y S C. banan a FM-h e at e 'Cali at e a BANANA' b. Moo s bu u Cal i cuna y O banana FM-h e C . at e 'Cali at e a BANANA' c. Moo s bu u cuna y Cal i O banana FM-h e at e C . 'Cali at e a BANANA' d. Moo s Cal i ba a cuna y O banana C . F M at e 'CALI ate a banana' e. Cal i ba a moo s cuna y S C. F M banan a at e 'CALI ate a banana' f. Cal i ba a cuna y moo s S C. F M at e banan a 'CALI at e a banana'
O (FM-SRP) V (FM-SRP ) S V (FM-SRP)V S S (FM) V (FM ) O V (FM)V O
As one can see, th e element focused by baa is alway s i n preverbal position . One reason fo r saying that Somali is SOV is that the impersonal subjec t cliti c is the leftmost , hence th e highest , element in the Verba l Complex . Th e othe r i s that, a s w e have see n above , th e subjec t resumptiv e pronoun , deal t wit h belo w in more detail, is not incorporated into the Verbal Complex but belongs to a higher node, while object resumptive pronouns are. Both facts, along wit h others we shall discuss later , strongl y sugges t th e subject-object asymmetr y well know n in mor e familiar languages. 8 On the contrary, th e relative order of elements i s strictly determine d insid e the Verbal Complex, a templati c fiel d maximall y constituted fro m righ t t o left , b y (1) the verb (wit h an optional incorporated bar e noun , generally formin g a lexi cal uni t wit h it) ; (2 ) som e adverbial s of plac e (e.g . hor 'i n fron t of, ' ag 'near, ' etc.) and 'manner' (kala 'alone, * wada 'together') ; (3) German hinlher like deicti c particles, indicating, in rough terms, movement toward or away from th e speake r
Aspects of Discourse Configurationality in Somali 7
1
(soo 'her' ; sii 'bin') ; (4 ) a possessive (usuall y 2nd pers , kaa, kiiri), whos e us e will be explained shortly below; (5) some relational elements, traditionally calle d "prepositions" (e.g . ku 'in , by means of, ' la 'with, ' etc.); 9 (6 ) one object clitic ; finally, la, impersonal subjec t clitic . Consider th e following examples : (11) a . wu u [ i far a saaray ] FM-he m e finger s put-o n 'He has beaten me' b. Cal i ba a buuggiis a [i- i ka a siiyay ] C. F M book-hi s me-t o your s gav e 'CALI gave me his book fo r you' c. wa y [i s ho r joogaan ] FM-they sel f fron t stan d 'They stan d on e opposite th e other' d. annag a ba a [ la-yno-o so o ordayay ] us (EMPH ) F M IMP-us-t o DEIC T ra n 'Someone was running towards US ' e. Cal i ba a [ 0 u geeyay ] C, F M (it ) t o brough t 'CALI brought it(7him/her/them ) t o him(/her/them/it)' Note tha t 3r d pers. objec t clitics are not phoneticall y realized, a s see n i n (lie). The templatic nature of the Verbal Complex is clearly shown by (1 Ib), in which kaa, normall y a 2nd pers. possessive, here spell s ou t 'you ' (O) . Th e reaso n fo r this is that only on e overt objec t clitic ca n occur i n the Verbal Complex (n o restrictions hold instead for null clitics). Thus, if two overt pronominals are require d (no matter if selected b y one or more heads) one is realized as a possessive, whic h is assigne d it s ow n position i n the Verbal Complex . I n (lib), fo r example, th e verb siin, 't o give * ha s tw o objects, buuggiisa, 'book-his ' an d i, 'me, ' whil e u 'to, for' (i-i focal) an d left-detache d (t o have it s scale-denot ing sid e receive a link interpretation). The S-structur e positio n o f focus-prepose d phrases ca n b e argue d t o be a reflection no t o f their partia l statu s a s foci bu t o f their partial statu s as links. Then , assumin g the IS fo r (50b ) i s identica l t o its S structure, th e lin k force o f Fido i s correctl y reflecte d at IS . It s foca l side , how ever, i s not . Fortunately , there i s a transformationa l device tha t wil l allo w th e value-denoting sid e o f Fido t o b e i n sit u an d th e scale-denotin g sid e t o b e left detached a t the sam e time : reconstructio n (Chomsk y (1977)) . Reconstructio n i s available t o map S-structure s like (51a ) t o LF structure s like (51b) : (51) a . [ whose motherj j di d you see tr b. [ whosej ] did you se e tj mothe r An LF like (51a ) woul d not yield the correct interpretatio n for this question ('fo r which x did you see x's mother') . By reconstructing a part of the wh-phrase back into th e claus e the righ t interpretive resul t is achieved.
144
Discourse Configurational Languages
Reconstruction ca n b e employe d t o reflec t th e dua l informationa l natur e o f focus-preposed phrase s in much the same way . Let us assume th e IS representio n in (52 ) fo r (50b) : (52) [ t[+scale], [[ IP (they ) t 2 (it) FIDO ! ] named 2 ]] The value-denotin g sid e o f the preposed phras e i s reconstructed bac k int o I P t o receive a focal interpretation , but the trace lef t in the left-adjoined slo t preserve s the scale-denotin g sid e o f Fido, an d dul y receive s a lin k interpretation . Give n Ward's arguments for the dual informational force of focus-preposed phrases , th e existence o f thi s constructio n cease s t o b e a vali d stimulu s fo r positin g a n ab stract rule of focus-raising. The leftwar d detachmen t of the focus-preposed phras e is no t a function o f it s partia l statu s as a focus, but rathe r o f it s partia l functio n as a link. What signals its partial status as focus at the surface is, a s expected, it s H* L phrasal tune, not its syntacti c position. Summarizing thi s section , i t appear s tha t th e motivation use d t o marshal l th e existence o f a n abstrac t rule o f focus-raisin g i n Englis h i s no t a s soli d a s i t ap peared. The fact s o n weak crossove r wit h focus ar e not clear cu t an d the rul e o f focus-raising itsel f i s problemati c i n a numbe r o f ways . Th e frontin g o f foca l phrases i n focus-preposin g canno t b e use d a s a stimulu s fo r abstrac t focus-rais ing either . The abstrac t I S configuration propose d i n Sectio n 5 does no t encom pass a rule of focus-raising . Given the issues raise d i n thi s section, this is no t a s undesirable a s it originall y may hav e seemed . 8. The Place of IS in the Model In th e T-model of grammar (Chomsky and Lasnik (1977) , Chomsk y (1981) ) on e single leve l o f representation, LF , is availabl e a s a n interfac e wit h th e interpre tive component s i t presumably feeds . I f this is so, the followin g questio n arises : what is the relationship between LF and IS (or Culicover & Rochemont's F-struc ture) and , mor e generally , wha t is th e plac e o f I S i n the mode l o f grammar ? There have been severa l proposal s i n the literature t o determine wha t th e ex act locu s o f informationa l representation is , especiall y fo r focus . Som e o f thes e proposals assum e informatio n packaging is represente d a t LF (Chomsk y (1976 ; 1981), Huan g (1982) , Horvat h (1986) , an d Rochemon t (1986)) . Other s assum e it is at an abstract level o f representation derived fro m L F called LF ' (cf . Huang (1984)). A third group of works leaves the question open (Koopma n and Sportiche (1982) an d Culicove r an d Rochemon t (1983)) . If focu s an d groun d ar e take n t o be logico-semantic element s o f some sor t it makes sense , conceptually , t o assum e that thei r correc t locu s o f representatio n i s LF . Taking th e approac h adopte d i n this study , however, ther e i s littl e motivatio n t o assum e tha t IS (o r F-structure ) has t o b e identifie d wit h LF. In fact , th e spiri t behin d th e Gricea n divisio n of labo r i s that interpretation i s not a monolithi c process. Instead , ther e ar e a se t o f system s wit h mor e specifi c interpretive tasks. Logical semantic s is one such system; information packaging,
Structural Properties of Information Packaging in Catalan 14
5
implicature computation , and i.llocutio n ar e others . Th e su m o f al l thes e inter pretive processe s result s i n globa l interpretation . If interpretatio n i s no t mono lithic, th e syntacti c interfac e leve l tha t serve s a s inpu t t o interpretatio n canno t be monolithi c either . A singl e leve l i s no t sufficien t t o interfac e th e differen t interpretive system s successfully : th e pur e structura l representation o f logico semantic relation s an d the pure structura l representation o f information packag ing mus t b e carrie d ou t a t differen t level s o f abstraction . There is syntactic evidenc e tha t add s to the conceptual argument s fo r the non identity o f I S an d LF . Consider exampl e (53a) : (53) a . [ Ja va n fe r e l descompt e a algun s CLIENT S ] . already di d th e discoun t t o som e client s '(They) alread y gav e som e client s th e discount. ' b. [ IP a algun s clienlSj [ IP ja va n fe r el descompte t j ] ] At LF, following the traditional analysis of quantifier-raising, the quantifier phrase a alguns clients 't o som e clients ' mus t raise an d left-adjoi n t o IP , as i n (53b) . From this position th e quantifier binds a variable i n A-position an d c-command s its scope. The logical meanin g of the quantified expression i s thus appropriatel y represented a t LF and fed into the logico-semantic componen t fo r interpretation . But sentenc e (53a ) i s a n all-focus instruction. As such , give n the I S configu ration propose d i n Sectio n 5 , it s IS representatio n mus t hav e al l materia l unde r the cor e I P node , i.e. , nothin g shoul d b e left-adjoine d t o IP . This sentenc e re quires a n L F structur e like (53b ) an d a n I S structur e like (54): (54) [ Ip ja va n fe r e l descompt e a algun s clients ] The configuration s of (53b ) an d (54) ar e not an d cannot be the same. Therefore , LF an d IS canno t b e th e same . I f (54 ) wer e bot h th e I S an d the L F representa tions o f (53a) , th e quantificationa l forc e o f alguns clients woul d b e lef t unrepresented a t the relevan t interfac e level . If , alternatively , (53b ) i s propose d for bot h representations, alguns clients, whic h is part o f the focus, i s erroneousl y interpreted as a link because o f its structural position at IS. If alguns clients wer e a link , i t woul d be encode d a n suc h a t S-structure : (55) [ Ip a algun s clients. j [ Ip ja els n fer e l DESCOMPT E t t ] ] 1 va to som e client s alread y object di d th e discount The conclusion is that quantifier-raising and link left-detachment canno t take place at th e sam e leve l o f abstraction . Unles s on e o f th e tw o rule s i s abandoned , I S and L F mus t b e maintaine d a s separat e level s o f representatio n (cf . (44 ) abov e for simila r evidence) . Once th e non-identity of LF and IS is determined, th e possibility indicate d b y Koopman an d Sportich e (1982 ) an d Culicover an d Rochemont (1983 ) tha t IS i s derived fro m L F mus t b e examined . O n thi s hypothesi s S-structur e i s mappe d onto L F an d LF , i n turn , i s mappe d ont o LF ' (= IS) . Th e workin g hypothesi s
146
Discourse Configurational Languages
underlying thi s paper , however , i s tha t I S i s directl y derive d fro m S-structure . This i s evident fro m th e mapping function yieldin g IS , a s was describe d i n Sec tions 5 an d 6 , which i s designed , fo r both Catala n an d English, t o tak e S-struc ture representations as input. It is very hard to gather empirical evidence t o argue that on e hypothesi s i n (56 ) i s better than the other ,
but ther e ar e conceptua l arguments, based o n th e economy-of-derivation condi tion i n Chomsky (1991) , tha t suggest that a direct derivation of IS fro m S-structure i s superio r t o a derivatio n from LF . The spiri t o f Chomsky' s conditio n is t o minimiz e derivation mappings . B y a "least effort " principle, if a give n result ca n b e achieve d b y a simpl e derivatio n and a complex one , onl y th e forme r shoul d b e legitimate . Fo r instance , a give n movement operatio n should not be posited i f it is t o be undone in the nex t deri vation. Chomsky's exampl e is the lowering of inflection to V at S-structure. This is a "costly " operation , sinc e i t i s undon e a t L F whe n V+inf l i s raise d t o th e original INF L position. Th e I S situatio n is analogou s t o th e cas e describe d b y Chomsky. Conside r (53 ) again , repeated a s (57a) , wit h it s L F representatio n i n (57b) an d it s I S representatio n i n (57c) : (57) a . [
ja va n fe r el descompt e a algun s CLIENT S ] already di d th e discoun t t o som e client s '(They) alread y gave som e client s th e discount.' b. [ IP a alguns clientSj [ Ip ja va n fer el descompt e t } ]] c. [ IP ja va n fer e l descompte a alguns clients ] Ip
At S-structur e the quantifie r appears i n situ . At L F i t must raise an d left-adjoin to IP, but a t IS i t mus t appear in sit u again . If LF mediate s betwee n S-structur e and IS , (57a ) mus t b e transforme d ont o (57b ) befor e i t i s mappe d ont o (57c) , despite th e fact tha t (57a) an d (57c) ar e identical. I t is a two-step derivatio n with the secon d ste p reversin g th e outpu t of th e firs t step . This i s precisel y th e typ e of derivation that Chomsky (1991) judges not permissible, al l things being equal . The derivatio n tha t yield s I S directl y fro m S-structure , i n contrast , i s straight forward an d maximall y economical (i n th e cas e o f (57 ) th e derivatio n applie s vacuously). Assuming economy , then, a direc t mappin g betwee n I S an d L F i s a bette r alternative. A s fo r th e derivatio n of L F fro m S-structure , i t proceed s i n th e standard way. The resulting picture is a parallel mapping between S-structure and
Structural Properties of Information Packaging in Catalan 14
7
two autonomou s abstrac t level s o f representation: L F and IS. The inclusio n o f a separate leve l o f IS, derive d directl y fro m S-structure, severel y affect s the stan dard view of what grammar musit consist of . The T-model of grammar, represente d in (58a) , ha s n o roo m fo r suc h a level. A model o f gramma r tha t doe s incorpo rate i t ca n be modele d a s in (58b ) (th e dotte d lin e mean s tha t furthe r strata ar e probably neede d t o represent othe r relations) :
Chomsky (1991) views S-structure as a contact level betwee n several "fundamen tal" level s o f representation. In other words, S-structure is not a fundamental level but just wher e al l fundamenta l level s meet . Th e role o f S-structur e a s a contac t level i s enhanced and becomes eve n mor e plausibl e i n a model lik e (58b) . Eac h level o f pure representation is related to S-structur e by a different spoke , an d S structure is literally the hub, from where or to where all the information is passed . At S-structur e al l informatio n fro m th e pur e level s o f representatio n mus t b e recoverable t o b e passe d o n t o phonolog y an d th e physica l realit y o f th e utter ance, o r vice versa . This sectio n ha s presente d syntacti c evidenc e tha t support s th e conceptua l arguments for the non-identit y of IS and LF an d the direct derivatio n o f IS fro m S-structure. Thi s proposal , o f course , entail s a mode l o f gramma r tha t depart s somewhat fro m the traditiona l T-model .
9. Conclusion The analysi s of th e structura l representation of logico-semantic notion s evolve d hand in hand with the study of their interpretive import in a comprehensive theor y of logica l semantics . I n contrast , th e stud y o f th e structura l representatio n o f informational relation s proceeded, for th e mos t part , without a parallel compre hensive theor y o f informatio n packaging . Once suc h a theor y i s take n int o ac count, expectation s abou t wha t th e syntacti c representation o f informatio n packaging i s ar e likel y t o change. The pape r ha s take n a particula r theory o f informatio n packaging a s a poin t of departur e and looke d int o the structura l realization o f informational relation s as define d i n thi s theory . Immediately , i t become s clea r that , i n Catalan , infor mational relation s are a major determinant of surface syntactic structure . I n En glish-type languages, however, the overt structural representation of information packaging is mostly the responsibility of prosody. One single informational meaning i s realize d by mean s of differen t structura l configurations in differen t Ian -
148
Discourse Configurational Languages
guages. Despit e thi s contrast , thi s pape r suggest s tha t ther e i s a n abstract leve l of pur e informationa l representation , IS , wher e difference s betwee n language s in th e over t realizatio n o f informatio n packagin g ar e neutralized . I S mediate s between surfac e synta x an d th e informationa l component , jus t a s L F mediate s between surfac e synta x an d logica l semantics . Thus , th e interpretiv e rule s tha t bleed IS are the sam e fo r all languages, independen t o f the overt structura l real ization o f information packaging . This work , o f course, focuse s o n only tw o language-types . I n the futur e othe r language-types, severa l o f whic h ar e represente d i n th e paper s i n thi s volume , must be taken int o account so as to check th e validity of the proposal. Consider ation o f othe r language-type s i s likel y t o lea d t o change s i n th e actua l I S con figuration propose d her e o r t o th e establishmen t o f som e kin d o f parametri c variation i n relation t o IS. What is less likely t o change is the conviction tha t th e representation o f informatio n packagin g i s a s importan t a par t o f synta x a s th e representation o f theta structure , case structure , or logico-semantic structure , an d that the syntax/information-packaging interface is analogous t o the syntax/logico semantics interface . This , o f course , ma y entai l a chang e i n ou r standar d con ception o f grammar, but it may just be the price to pay if information packaging , which ha s lon g remaine d evasiv e t o forma l linguistics , i s t o be integrate d int o the genera l theor y o f language.
Notes * A much earlie r version o f thi s wor k wa s presente d a t th e 16t h GLO W Colloquiu m a t Cambridge University in April, 1990 , an d a firs t writte n version o f tha t talk appeared a s part of Chapters 5 and 6 in my dissertation (Vallduvi (1992a)). I am indebted to R. Frank, A. Kroch, C . Heycock, J. Hoeksema, M . Moser, E . Prince, B . Santorini, R. Zanuttini, and audiences a t presentation s o f thi s work fo r a showe r o f comment s an d suggestions , an d to K . E. Kis s fo r he r interes t i n thi s wor k an d he r patienc e a s edito r o f thi s volume . O f course, an y error s an d omission s remai n m y ow n responsibility . Thi s pape r wa s written while I was at the Universitat Autonoma de Barcelona o n a postdoctoral researc h fellow ship fro m th e Research an d Technology Divisio n o f the Spanish Ministry o f Science an d Education. 1. Informatio n a s define d here i s differen t fro m th e mor e comprehensiv e notio n of in formation i n Situation Theory, wher e the information conveyed b y an event, linguistic o r nonlinguistic, i s wha t we lear n about a situatio n fro m tha t event . 2. Heim's notio n o f fil e i s actually closer t o a discourse model , see n a s a participant's evolving mode l o f the current dicourse, than to ( a subset of) the knowledge-store, under stood a s a general cognitive construct. Here, onl y the basic file-card metaphor is adopted and extrapolate d t o th e knowledge-stor e wit h th e necessar y changes . N o referenc e wil l be mad e t o othe r aspect s o f Heim' s theor y (metho d o f truth-valu e computation , etc.) . 3. As inferrabl e fro m th e contents of thi s section , informatio n packaging is held t o b e autonomous wit h respect t o logica l semantics . Thi s vie w i s a t odds wit h anothe r family of approache s tha t take precisely th e opposit e tack . In Vallduvi (1992a) som e o f th e ar guments tha t woul d apparentl y favo r th e latte r approaches are discusse d and countered . As for th e notion of "identificational" focu s (Kis s (this volume), Horvath (this volume)), it largel y correspond s to th e tailfu l instruction s i n th e instruction-base d system, bu t i t i s
Structural Properties of Information Packaging in Catalan 14
9
stripped o f it s logico-semanti c exhaustivenes s impor t (cf . Hor n (1991 ) fo r th e vie w that exhaustiveness i n focu s i s a conversational implicature) . 4. Author s wh o believ e wh-word s i n wh-question s ar e narro w foc i (e.g . Rochemon t (1986), Horvath (thi s volume) ) d o not agree wit h the claim tha t prominence i s necessar ily associate d wit h focus , give n tha t in wh-questio n wh-word s ar e no t prominent . Thi s belief, however , i s not unproblematic (cf . Erteschik-Shir (1986)) . 5. Sentences (8b ) and (8c) ar e marginally possible wit h an extreme metalinguisti c flavor a s a correction o f pronunciation or other aspec t o f the utterance. I n Hungarian, which has a fixed foca l prominence a s well, the sam e phenomenon exist s (cf . Horvat h (1986)). 6. A different analysi s of left-detachmen t i s foun d i n Cinqu e (1990) , wh o argue s fo r the base-generated statu s of left-detached phrases. As for the analysis of clitics, th e structural description adopte d here entail s a view o f clitics as base-generated preverball y following Stroze r (1976 ) an d Sune r (1988) , inte r alia . 7. Alternatively , th e objec t ma y b e weakl y pronominalized . Weakl y pro-nominalize d elements (Catala n clitics, English unstressed pronouns) remain within the core IP , but are not "visible " qu a informationa l elements . Thei r presenc e i n th e claus e i s du e t o othe r structural requirements . Wea k pronoun s appear withi n th e focus-delimitin g brackets i n some example s below , but must not be considered par t of the focus . Strong pronouns, in contrast, ma y ente r a n informational instructio n as full-fledge d element s o f th e focu s or the ground . 8. Positing a detachment of the [+TNS ] ver b to a clause-peripheral adjunctio n position may see m an unusual proposal. However , if VO adjoins to IP, the trac e in 1 0 is properl y governed b y the detached [+TNS]-carryin g verb, a necessary conditio n t o licens e verba l empty categories , accordin g to Koopma n (1984). Also , thi s proposa l runs counter t o th e Head Movement Constraint (HMC), but see Torrego (1984 ) an d Kayne (1990) for analyses wher e X ° is adjoine d t o X P a s well . 9. This analysi s of focus-preposing i s able t o generate some string s that are underivable in th e traditiona l analysis , but unfortunatel y i t als o overgenerates . Conside r (i-iii ) «i)=(21)): (i) E
l MERCEDE S e l Jord i n o sa p qu i s ' h a comprat . the Mercedes th e Jord i n o know s wh o sel f ha s bough t 'The MERCEDES Jord i doesn't kno w who bought.' (ii) a . E L MERCEDE S e l Jordi qui s'h a compra t n o sap . b. E L MERCEDE S n o sa p el Jordi qui s'h a comprat . c. E L MERCEDE S r. o sap qu i s'h a compra t e l Jordi . d. E L MERCEDE S qu i s'h a compra t no sa p el Jordi . e. E L MERCEDE S qu i s'h a compra t el Jord i no sap . (iii) a . *E L MERCEDES n o sa p el Jordi s'h a compra t qui . b. *E L MERCEDES qu i no sa p s'ha compra t el Jordi .
(i) an d th e additiona l strings i n (ii) , som e o f whic h th e traditiona l analysi s canno t ac count for , ar e correctl y predicte d b y th e right-detachmen t analysis . Bu t thos e string s i n (iii) ar e also predicted t o be correct, even thoug h they are utterly ungrammatical. Apparently, th e verba l strin g and th e complementize r hav e t o b e right-detache d i n on e singl e group fo r reason s tha t wil l hav e t o remai n unexplaine d fo r th e tim e being . Th e smal l residue i n (iii ) canno t be accounte d for , but, in an y event , mor e case s ar e covere d her e than i n th e traditiona l analysis. Within grammaticality , som e o f th e sentence s i n (ii ) ar e
150
Discourse Configurational Languages
less natura l o r frequen t tha n others. An accoun t of suc h gradation will no t b e attempte d here. 10. Root h (1985) an d Horvat h (1986) dismis s earlie r counterexample s t o focu s wea k crossover (i n e.g . Sola n (1984) ) becaus e the y involv e a "fre e variable " reading , wher e focus i s coreferential with a pronoun to the left b y chance. (47b) , however , is not subjec t to thi s criticis m becaus e i t reflect s a tru e boun d reading . I n (47b ) his parents i s a con trastive link and is therefore realized with an L+H* pitc h accent (cf. Section 4) . This does not affec t th e crossove r effec t i n an y way.
References Adams, M. (1987) "Fro m Old French t o the Theory o f Pro-drop," Natural Language and Linguistic Theory 5, 1-32 . Antinucci, F. and G. Cinque (1977) "Sull'ordin e delle parole in italiano: 1'emar-ginazione," Studi di Grammatica Italiana 6 , 121-146 . Baltin, M . (1982 ) " A Landing Site Theor y o f Movemen t Rules," Linguistic Inquiry 13 , 1-38. Bonet, E. (1990 ) "Subject s i n Catalan, " The MIT Working Papers in Linguistics (Papers on Wh-movement) 13 , 1-26 . Bonet, S. , an d J . Sol a (1986 ) Sintaxi generativa catalana, Enciclopedi a Catalana , Barcelona. Chafe, W.L. (1976) "Givenness , Contrastiveness, Definiteness, Subjects, Topics, and Point of View, " i n C . Li , ed. , Subject and Topic, Academi c Press , Ne w York . Chomsky, N. (1976 ) "Condition s on Rules of Grammar," Linguistic Analysis 1, 75-109. Chomsky, N . (1977 ) "O n Wh-movement, " i n P . Culicover e t al. , eds. , Formal Syntax, Academic Press , Ne w York. Chomsky, N . (1981 ) Lectures on Government and Binding, Foris , Dordrecht . Chomsky, N. (1991 ) "Som e Note s on Economy of Derivation and Representation," i n R . Friedin, ed. , Principles and Parameters in Comparative Grammar, MI T Press , Cambridge, Massachusetts. Chomsky, N., and H. Lasnik (1977) "Filter s and Control," Linguistic Inquiry 8 , 425-504. Cinque, G . (1990 ) Types of A'-dependencies, MI T Press , Cambridge , Massachusetts . Contreras, H . (1991 ) "O n th e Positio n o f Subjects," i n S.D . Rothstein , ed., Perspectives on Phrase Structure: Heads and Licensing (Syntax and Semantics 25) , Academi c Press, New York. Culicover, P. , and M . Rochemon t (1983 ) "Stres s an d Focu s i n English, " Language 59 , 123-165. Erteschik-Shir, N. (1986) "Wh-question s and Focus," Linguistics and Philosophy 9 , 117 149. Farkas, D . (1986) "O n th e Syntactic Position of Focus i n Hungarian," Natural Language and Linguistic Theory 4 , 77-96. Fernandez-Soriano, O. (1989) "Stron g Pronouns in Null-Subject Languages and the Avoid Pronoun Principle," The MIT Working Papers in Linguistics (SOIL I) 11 , 228-239. Gazdar, G. (1979) Pragmatics: Implicature, Presupposition, and Logical Form, Academi c Press, Ne w York. Halliday, M.A.K . (1985 ) An Introduction to Functional Grammar, Arnold, London . Heim, I . (1983) "Fil e Change Semantics an d the Familiarity Theory o f Definiteness," i n R. Bauerle et al. , eds., Meaning, Use and Interpretation of Language, d e Gruyter, Berlin an d Ne w York.
Structural Properties of Information Packaging in Catalan 15
1
Horn, L. (1981 ) "Exhaustivenes s and the Semantic s o f Clefts, " NELS 11 , 125-142. Horn, L. (1989 ) A Natural History of Negation, Universit y o f Chicago Press , Chicago . Horvath, J. (1986) FOCUS in the Theory of Grammar and the Syntax of Hungarian, Foris, Dordrecht. Huang, C.TJ. (1982) "Move-w h in a Language Without Wh-movement," The Linguistic Review 1 , 369-416. Huang, C.T.J. (1984) "On the Distribution an d Reference o f Empty Pronouns," Linguistic Inquiry 15 , 531-574. Jackendoff, R . (1972) Semantic Interpretation in Generative Grammar, MIT Press, Cambridge, Massachusetts . Jacobs, J. (1986 ) "The Syntax of Focus an d Adverbials in German," i n W. Abraham and S. de Meij, eds., Topic, Focus, and Configurationality, Joh n Benjamins, Amsterdam and Philadelphia. Kayne, R. (1990) "Romanc e Clitic s an d PRO," NELS 20.2 , 255-302. IS. Kiss, K . (1981 ) "Structura l Relation s in Hungarian, a 'Free * Word Order Language, " Linguistic Inquiry 12 , 185-213 . Koopman, H. (1984 ) The Syntax of Verbs, Foris , Dordrecht . Koopman, H. and D. Sportiche (1982 ) "Variable s an d the Bijection Principle," The Linguistic Review 2, 139-160. Kroch, A., B. Santorini and C. Heycock (1988) "Bare Infinitives and External Arguments," NELS 18 , 271-285. Pierrehumbert, J . an d J . Hirschber g (1991 ) "Th e Meaning o f Intonationa l Contour s i n Discourse," i n PR. Cohe n e t al., eds. , Intentions in Communication, MI T Press , Cambridge, Massachusetts. Prince, E.F . (1986) "O n the Syntactic Marking of Presupposed Ope n Propositions," CLS (Parasession) 22 , 208-222. Prince, E.F . (1992) "Th e ZPG Letter: Subjects , Definiteness , an d Information-Status, " in S.A. Thompson and WC, Mann, eds., Discourse Description, Joh n Benjamins, Amsterdam and Philadelphia. Reinhart, T . (1982 ) "Pragmatic s an d Linguistics : A n Analysi s o f Sentenc e Topics, " Philosophica 27 , 53-94. Rochemont, M . (1986) Focus in Generative Grammar, John Benjamins, Amsterdam and Philadelphia. Rochemont, M . (1989 ) "Topi c Island s and the Subjacenc y Parameter, " Canadian Journal of Linguistics 34 , 143-170 . Rooth, M. (1985) Association with Focus, Doctoral dissertation, University of Massachusetts, Amherst. Rossell6, J. (1986 ) Gramatica, configuracions i referenda. Per una teoria alternativa del PRO-drop romanic, E'octoral dissertation, Universitat de Barcelona, Barcelona . Sola, J . (1992 ) Agreement and Subjects, Doctora l dissertation , Universita t Autdnoma de Barcelona. Solan, L. (1984 ) "Focu s and Levels o f Representation," Linguistic Inquiry 15 , 174-178. Steedman, M. (1991) "Structur e an d Intonation," Language 67, 260-296. Strozer, J. (1976 ) Clitics in Spanish, Doctora l dissertation , UCLA , Los Angeles . Suner, M . (1988 ) "Th e Role o f Agreemen t i n Clitic-double d Constructions, " Natural Language and Linguistic Theory 6 , 391-434. Thwing, R. , and J. Walter s (1937 ) "Focu s in Vute, " Journal of African Languages and Linguistics 9, 95-121. Torrego, E. (1984) "O n Inversion in Spanish and Some of its Effects," Linguistic Inquiry 15, 103-129.
152
Discourse Configurational Languages
Valimaa-Blum, R . (1988 ) Finnish Existential Clauses: Their Syntax, Pragmatics, and Intonation, Doctora l dissertation, Ohi o State University , Columbus. Vallduvf, E . (1991) "The Rol e of Plasticity in the Association of Focus an d Prominence, " ESCOL 7 , 295-306. Vallduvf, E . (1992a ) The Informational Component, Garland , Ne w York. Vallduvf, E . (1992b ) "Focu s Construction s in Catalan, " in C . Laeufer and T.A. Morgan, eds., Theoretical Analyses in Romance Linguistics, Joh n Benjamins, Amsterda m and Philadelphia. Vallduvf, E. (1993) "Catalan as VOS: Evidence from Informatio n Packaging, " in W. Ashby et al. , eds. , Linguistic Perspectives in the Romance Languages, Joh n Benjamins , Amsterdam an d Phildelphia. Ward, G.L . (1988 ) The Semantics and Pragmatics of Preposing, Garland , New York.
6 An F Position i n Western Romanc e JUAN URIAGEREKA University of Maryland, College Park"
1. Some Background It is has been argue d tha t functional categories other than C, I, and D exist. The Infl skeleton has been "split" into categories tha t include several Agr nodes, Tense, and Aspect ; likewise, the De t skeleto n ha s bee n argue d to presen t variou s Agr nodes, Number, o r a Possessio n node . Splittin g th e Com p skeleto n ha s a lon g tradition, Reinhart's (1979) analysis was based on the premise o f two such node s in certai n languages , th e origi n o f wha t i s no w commonl y referre d t o a s "C P recursion". And again several nodes have been argued to correspond to the Com p skeleton, a matter I want to discus s here . But firs t I want to lay out my ground-rules. One , thi s i s a research program . Surely ther e ar e alternative s i n terms o f features i n single categories , instead o f positing multipl e categorie s eac h correspondin g t o a feature. I wil l no t explor e this alternativ e vie w here . Two, bot h a formal and a substantive issu e aris e th e minute on e posits a new category. The formal on e is proving the existence of the category i n terms of som e empirical evidence. The substantive one is determining what that category is, how it differ s fro m others , etc . Th e secon d issu e i s too ambitious a t this stage , an d I will onl y have som e conjecture s t o offe r abou t the natur e o f th e categor y I a m positing. The first issue itsel f i s non-trivial. It has to be determined whethe r th e hypothesized categor y i s a single category o r an array o f categories, is a head o r a maximal projection, is higher or lower than other categories within the Phras e Marker. The mai n purpose o f thi s pape r i s t o deal wit h thes e sorts of question s for a given category I hypothesize . Three, evidenc e adduced for a given proposal is relative to a framework, and it ma y count only within that framework. I assum e th e essentials o f Chomsky' s (1993) Minimalist program , This means , t o star t with , tha t I canno t assum e 153
154
Discourse Configurational Languages
levels o f representation othe r tha n PF an d LF , operations tha t ar e no t economi cal, parameter s that ar e not trivial , etc. The thir d of m y ground-rule s has consequence s tha t shoul d b e clarified , par ticularly give n th e contex t wher e thi s articl e i s appearing . A t issu e i s whethe r there shoul d be a level o f Information Structure (IS), a s posited by Valduvi (this volume). I wil l no t assum e suc h a level, an d my reason s ar e o f two sorts . A separate level o f IS is supposed t o encode representation s tha t have no con sequence fo r th e trut h condition s o f a sentence . The assumptio n behin d thi s i s that L F i s a leve l whic h is truth-theoretica l i n nature . I disagre e wit h bot h th e proposal an d the assumption . In th e minimalis t framework a s wel l a s on th e work s leadin g t o it s definitio n of LF, it i s not th e cas e that LF ha s anythin g to do with truth, or for tha t matte r notions tha t ente r int o th e calculation s o f truth , suc h a s arguments , predicates , operators, variables, and the like. The primitives of LF are syntactic, and include chains of different sorts , the conditions of their formation, operations o f ellipsis , and littl e else . L F interface s with semanti c level s (cal l thos e Intentional / Con ceptual structure , ICS) wher e perhaps a truth-functional apparatus has it s place , as wel l a s othe r non-syntacti c devices. I n othe r words , i t is perfectl y legitimat e to map LFs to structures encoding matters of focus, background, and so on, without a separate leve l o f IS . In fact , w e could no t ma p suc h discourse matter s otherwis e i n th e minimalis t framework. Th e syste m i s designe d aroun d level s o f competenc e which , i n Chomsky's words , follow from "virtual conceptual necessity". It is far from clea r that conceptual necessity forces the postulation of a level whic h is not even necessarily a competenc e level , bu t seem s mor e lik e a n interfac e wit h th e perfor mance systems that Chomsky places outside of the syntax proper (i.e. , th e optimal mapping o f PF an d LF) . Within th e minimalis t framework , the notio n o f discours e configurationality is coheren t onl y as a metaphor, since discours e is i n essence a matter of perfor mance. O f course , performance-dependen t doe s no t mea n vague , idiosyncratic , or senseless. Fo r instance, for the minimalist program semantic s i s performative , and fe w domain s hav e bee n show n t o b e s o precise , rule-governed , an d ful l o f sense a s semantics . However , thi s systematicit y doe s no t hav e t o b e expresse d configurationally, an d very possibl y i s not expressed thi s way . This i s al l t o sa y that i f an y effect s o f discours e in fact pla y a role i n th e configurationalit y of a sentence, these must be expressed by wa y of a category which is recognizable a t LF, and i s bor n i n th e synta x proper. Apart fro m no t bein g abl e t o accep t a separat e level o f IS , I als o hav e diffi culties wit h th e assumptio n tha t i t expresse s matter s whic h ar e no t truth-func tional. As i s wel l know n from th e literatur e o n focus , a sub-clas s o f sentence s exhibiting this phenomenon have truth-theoretic consequences (se e th e Introduction to this volume) . In turn, it might seem a s if standard topicalizations hav e n o truth-theoretic consequence , bu t thi s may no t b e true—a t leas t fo r non-contrastive topics that create their own categorical judgement (see Raposo and Uriagereka (1993)). This i s al l t o sa y tha t core phenomen a within what we ma y cal l Information Theor y ar e a s truth-functiona l as others .
An F Position in Western Romance 15
5
The poin t just mad e doe s rto t o n itsel f mea n tha t informatio n theoretic phe nomena must be represented at LF, anymore than any other phenomenon that has effects o n th e trut h condition s o f th e sentenc e must . Bu t eve n i f th e minimalist view o f L F wer e give n u p i n favo r o f a leve l o f logica l for m i n th e traditional philosophical sense , matter s o f Informatio n Theory woul d stil l hav e t o b e ex pressed i n that would-be level. That is, information theoretic matter s do not carry enough conceptua l weigh t fo r us to posi t a separate leve l of representation , an d either L F or logica l for m see m perfectl y fi t t o d o the job o f representin g them . However, I d o accep t th e empirica l fac t tha t discours e affect s configurations . Hence, a s a consequence o f the ground-rule s I se t up fo r myself , I a m virtually forced t o posit a syntacti c categor y tha t encode s information-theoreti c issues . I promised a conjecture i n this respect , an d with it comes a disclaimer. I have no t found an y conclusiv e evidenc e tha t ther e ar e separate functiona l categorie s t o express matter s of topic , focus , emphasis , contrast , etc. Al l of thes e hav e a n as pect i n common : the y encode th e point of view o f a speake r or som e othe r sub ject, i n a manner to be clarified immediately. I therefore assume that one category alone serve s a s an all-purpos e devic e t o encod e a point o f view . I will call th e category in question "F" . Uriagerek a (1988 ) borrows th e insight behind thi s categor y fro m Chomsk y (1977) . I explicitly too k F to merely stan d for "functional, " an d in other places I used F as a mnemonic fo r a "further" pro jection, o r "focus" wit h the vague import o f emphasis, contrast , information-encoding device , etc . All I mean i s this: F encodes poin t o f view. The clai m is that all information theoretic operation s nee d t o be mediated throug h a point of view. That is , whe n emphasi s appear s i n a sentence , someon e i s responsibl e fo r tha t emphasis. Ol d or new informatio n is old or new for someone . Eve n th e usag e of a referrin g expression presuppose s a speake r wh o assume s responsibilit y (mis takenly o r otherwise ) for callin g someon e Smith' s murdere r or Jones, an d simi lar issue s aris e fo r deixis , anaphora , etc . The genera l vie w is this: a syntactic theory formally expressing dependencie s goes a s fa r a s establishing vali d chains an d nothin g else. These chain s hav e in dexation mechanisms, bu t the indices carry no semantic value . Value assignment is a matter which i s dependent o n a discourse, whic h i s to sa y the poin t of view of a speake r o r som e othe r subject . Thi s i s wha t pragmatic s i s about . M y pro posal i s that th e synta x has on e designate d nod e whic h enter s int o th e determi nation o f what are vali d valu e assignmen t for forma l indice s (i.e . relations ) tha t the synta x creates. Ther e are syntactic rules determining possibl e point s o f view in a sentence, an d only a sub-class of the logically possible point s of view that a pragmatic theor y allow s satisfie s th e syntacti c derivation. Take th e sentenc e John's mother believes that Mary likes NOTHING WHATSOEVER. I t ha s a n interpretatio n wher e th e emphasi s o n th e objec t o f likes ha s a characteristi c illocutionar y force . Interestingly , the forc e i n questio n ma y b e related t o eithe r th e speake r o f th e sentenc e (say , I) , o r th e subject , John' s mother—but no t to John. That is , I may be reporting a belief tha t John' s mothe r has: "Mar y like s NOTHIN G WHATSOEVER! " Bu t John' s mothe r migh t als o have a n arra y o f beliefs : "Mar y doesn' t lik e peanuts ; sh e doesn' t lik e candy , either; in fact, Mar y doesn't sve n lik e cookies or ice-cream." The n I am entitled
156
Discourse Configurational Languages
to conclude tha t John's mother believes Mary likes NOTHING WHATSOEVER ! John's mothe r neve r had tha t belief , bu t ha d enoug h beliefs abou t Mar y fo r m e to conclud e what I did . Even i f w e recreat e th e scenari o fo r John instea d o f hi s mother, the sentence in point cannot express John's illocutionar y force—only th e speaker's an d th e mother's . This differenc e i n emphasis can be capture d in differen t ways . My proposal is to do it through a syntactic device: the emphasized element move s t o the spec of an F category, which is to be interpreted in the performative components a s ex pressing th e poin t o f vie w o f whicheve r subjec t i s relevan t (th e speake r o r a n embedded subject) . Covert o r over t rule s o f movement allo w certai n chain s in volving F, and not othe r chains . Ideally, thi s alone wil l predic t th e clas s o f deri vations whic h fee d ICS fo r a n appropriat e poin t o f vie w t o b e expressed . Perhaps i t is worth emphasizing that movement of an element to the spe c o f F does no t translate into emphasizing that element at LF. The movement itself i s a (covert o r overt ) operatio n to satisf y som e L F requirement of the sor t discusse d by Chomsk y (1993) . However , onl y element s that are appropriatel y place d i n F will b e visibl e fo r IC S t o b e interprete d as emphasize d b y whicheve r relevan t subject i s responsibl e fo r th e emphasis . T o illustrate this, conside r John thinks that NOTHING does Mary like an d NOTHING does John think that Mary likes. Crucially, in the firs t o f these two sentence s i t is John wh o i s responsible fo r th e emphasis, whil e in the second instance it is I who is responsible fo r it. A way t o express thi s is t o freeze th e displace d NOTHING i n the plac e wher e i t ha s bee n moved—say the specifier of F. While in the sentences seen before, at LF the object of likes coul d mov e t o th e lowe r [spec , FP ] o r th e highe r one, 1 in thi s instanc e the movement has taken place overtly, and minimalist considerations would make it sta y where we see it. This i s presumably mapped into ICS in a trivial manner, with th e element i n th e spe c o f FP expressin g the poin t of vie w of it s closes t c commanding subjec t (John i f NOTHING i s i n th e lowe r spec , myself i f NOTHING i s i n th e highe r spec , assumin g some versio n o f John Ross' s performative analysis.) Th e synta x ha s decide d th e possibl e IC S interpretations , eve n i f th e interpretation prope r happen s a t that level . Examples of this sort can be created wit h focali/ation, topicalization , the mod e of presentation of referential expressions (see Uriagereka (forthcomin g a), Raposo and Uriagereka (1993)). In the present context this means two things. First, when I argu e that an element ha s raise d t o F this a s suc h has n o interpretation; i t i s a formal mechanis m which will have various consequences, dependin g o n the na ture o f th e move d elemen t (whethe r i t i s a n operator , a n argument , a predicate , etc.; se e Uriagereka (forthcoming b)). Second, th e last few paragraphs are intended as a conjectura l motivation on th e substantiv e issue behin d F. Finally, the minimalist program imposes a view on parametric variation. Within the Romance languages explored here it is reasonable t o assume that while over t matters associate d t o F wer e quite productive in the Low Middle Ages an d stil l are in contemporary dialects which I will call "archaic," these matters are residual or non-existen t i n othe r dialects . More generally , example s presente d i n thi s volume sho w that variatio n exist s wit h respec t to discours e configurationality .
An F Position in Western Romance 15
7
The minimalis t framewor k captures intralinguisti c difference s through trivia l morphological specifications. Assumin g a hypothesized category F , we have three logical possibilities. First, F has "strong" feature s which must be checked by PF, or F does no t have them. If it does, som e hea d X must raise t o F to check thes e features, whic h given the strategy of Greed (an element move s only to satisf y it s own requirements) entail s that the head that raises to F must itself contain F features tha t are matched i n association to the F projection. It is then logicall y pos sible tha t the head X that raises to F may or may not bring strong feature s from the lexicon which ar e visible at PF. The second possibility is incompatible with a stron g F : a n X head withou t F features will not b e abl e t o raise t o F without violating Greed, and then the derivation wil l crash a t PF, for F remains visible at that level i n violation of the principle of Full Interpretation. However , a n X head without F feature s i s compatibl e wit h a wea k F . Finally, a n X hea d wit h F features ma y be compatible wit h a weak F. In this instance, associatio n of X to F is not in order to meet the requirements of both F and X, but onl y those o f X. (Whatever X's requirements are. ) The proposa l yield s thre e logicall y possibl e languages , which I wil l sugges t are paradigmatically represented by Galician-Portuguese, French , and Spanish— within Romance . Bu t quit e aside o f th e empirica l adequac y of thi s stat e o f af fairs, th e theory force s u s in this direction. Ther e i s little els e we can do within the system , whic h extend s i n principl e t o parametri c difference s pointe d ou t throughout thi s volume .
2. The Basic Patterns One o f th e fundamenta l difference s amon g Romanc e languages lie s i n wher e pronominal clitic s ar e place d i n eac h language . For instance , withi n Western Romance, th e patterns in (1) occur in tensed clauses : (1) Pattern s of Western Romanc e A: French, (spoken) Brazilian Portuguese(?) B: [+tense] Castillian Spajiish , Aragonese , Catalan , .. . C: ; [+tns] [+tns ] Portuguese, Galician , Leonese , most "archaic" dialect s I wil l no t explain thes e patter s (see Uriagerek a (forthcomin g a ) for a n analysis and variou s references). Fo r now , let me point out a few other , apparentl y unre lated properties associated to the third pattern: (2) Propertie s apparentl y associate d t o Patter n C : (i) Over t F(ocus ) elements (ii) Over t Focus movement
158
Discourse Configurational Languages
(iii) "Recomplementation " (iv) "Sandwiched " Dislocation s (v) Over t expletive s (vi) Persona l infinitive s withou t Aux-to-Comp (vii) Interpolatio n of element s betwee n clitics an d V Examples of (2i) can be see n in the er element appearin g i n medieval Galicia n texts, a s describe d b y Matto s e Silv a (1989) : (3) a . Quand o chegou.. . e vi o qu e no n podi a passar,.. . na when arrived.Il l an d saw.II I tha t no t could.Il l pas s in-th e segunda vegad a er mete u o mant o e n aquel a agu a second tim e E R inserted.Il l th e rob e i n tha t wate r 'When he arrived and saw that he could not pass, th e second tim e he did inser t the rob e i n tha t water' b. E est o pode s prova r qu e Deu s ouvio a t a ora£o n s e and thi s can.I I prov e tha t Go d heard.Il l th e you r praye r i f me achare s aqu i depoi s qu e aqu i er veere s me found.I I her e afte r tha t her e E R came.I I 'And you can prove that God heard your prayer if you found me her e after her e yo u di d come ' Traditional grammarian s noted tha t thi s for m "expresses contraposition, " "pro vides a pleonastic reinforcement, " o r "reinforce s th e sens e o f th e verb. " Thes e linguists equate er to even, only, an d similar adverbials , which obviously makes this elemen t loo k lik e a focusing marker. Simila r "extra" element s ca n b e see n in severa l contemporary dialects: 2 (4) X a e u sabi a qu e e l fa s cae r n a poza ! already I knew. I tha t i t were.I I fal l in-th e puddl e 'I knew tha t you wer e goin g t o fal l i n the puddle ' This Galicia n xa (lit : 'already') has nothing to do with completion of an event or anything associated t o a time past; all it does is add illocutionary force to a more neutral statement . Whether thi s sort o f elemen t i s adverbia l o r is integrate d int o th e skeleto n o f the sentenc e is a harde r question. Kiparsk y (1986) argue s tha t suc h element s a s Tense ar e grammaticalization s of adverbia l elements, an d th e sam e i s arguabl y true o f othe r functiona l categories . Bu t regardles s o f whethe r thi s i s generall y true, it does no t answer the real question : of the two possibilities tha t UG grants for focusin g elements—adverbial s o r functiona l heads—whic h on e i s a t issu e here? Propert y (iib ) suggests that indeed a hea d i s a t stake . A s i s wel l known, many Romanc e languages show the Focu s movemen t in (5) :
An F Position in Western Romance 15
(5) moita s cousa s li e e u dixera ! (Galician many thing s da t I said.had. I 'I had said MANY THINGS t o him/her'
9
)
The questio n is wher e the focuse d moitas cousas 'man y things ' i s moving . Of course, thi s coul d be a n adjunction—but i t coul d als o be a n instance o f move ment to a spec. If the latter is th e case, the obvious questio n i s what this spec is, and havin g a n F head woul d then clearly b e handy. Notice, incidentally , tha t if mere adjunctio n i s a t issue, we probably should expect the possibility o f several focus movement s (which does, not obtain) , and that the sor t of movemen t in (5 ) be fairl y genera l acros s languages—whic h i s not. Therefore, adjunctio n here i s highly suspect . Note als o th e position of the clitic i n (5) . This is entirely impossibl e i n mos t modern Romanc e dialects, bu t i t i s quit e systemati c in archai c forms—thi s i s property (2vii) . Bein g deliberatel y naiv e abou t th e clitic i n (5), w e can explai n its positio n i f ther e is a n extr a host fo r th e elemen t i n th e relevan t dialects . A question I return to in sectio n 4 is whethe r this could be F. Another archai c trai t tha t i s productiv e i n som e Romanc e language s i s recomplementation, which as Higgins (1988) notes i s associated t o dislocation s we may refer to a s "sandwiched": (6) a .
To p - S'; S ' >
Com p - S.
51. Topi c sentence s an d relativ e clause s ar e simila r i n tha t the y d o no t exhibi t Subjacency effect s an d emplo y resumptive pronouns (cf. Kuno (1973)). However, there
330
Discourse Conflgurational Languages
are stil l man y differences between them : A -ki phras e ma y appea r i n th e topi c positio n but canno t be the head o f a relative clause; th e aboutness relatio n betwee n th e topic an d the res t o f th e sentenc e i s no t observe d i n a relative clause , an d therefore , som e unac ceptable topic s ma y be heads o f relative clauses , a s show n in (i) . (i) a . * i iywu-nun i kos-i muli coh-ta . this reason-To p thi s place-sub water-su b good- M 'Speaking of thi s reason, th e wate r of this place i s goo d (fo r it).' (lit. ) b. i kos-i muli coh-u n k u iyw u this place-su b water-su b good- C th e reaso n 'the reaso n wh y wate r is good here ' 52. The topic marker is also morphologically the same as the complementizer o f a relative claus e (cf . (ib ) i n n . 51) . Here , I sugges t tha t -nun a s th e topi c marker i s a n affi x which is morphologically subcategorize d for [+N ] whereas the complementizer o f a rela tive claus e i s morphologicall y subcategorize d fo r [+V] . 53. Not e tha t th e followin g sentenc e wit h a relativ e claus e i s grammatica l wher e a varaiable is A-bound, as in (107). Thi s suggest s that a variable in (i ) is als o pronominal and subjec t to (105b ) o r (109 ) below . (i) He ; is th e mant whO j John me t tj . 54. Ki m (1989 ) has , in fact , suggeste d that wh-phrases ar e quantifiers in Korean , bu t given th e observation that wh-phrases move to certain designate d position s governe d b y a [+wh ] questio n marke r an d give n thei r semantic s tha t differ s fro m th e semantic s o f quantifiers, i t seems that they are basically differen t fro m quantifiers . Note also that given the discussion i n n. 10 , one ma y sugges t tha t in LF the orde r amon g th e topic, foc i an d quantifiers i s this : th e topic—foci—quantifiers . I f thi s orde r i s correct , quantifier s may not be topic s i n Korean just a s foci may not, whic h is tru e (cf. th e discussion o n (103)). 55. Fiengo e t al. (1988, 87 ) also observes that syntactic Pied Pipin g o f complex N P is restricted i n English. Fieng o e t al. (1988), Le e (1987) an d Choe (1988, 97-98) also not e some problem s wit h th e ide a tha t Pie d Pipin g explain s why Subjacenc y doe s no t hol d inLF.
References Aoun, J., N . Hornstein, and D. Sportich e (1982 ) "Som e Aspects o f Wide Scop e Quanti fication," Journal of Linguistic Research 1 , 69-95. Atlas, J.D . (1991 ) "Topic/comment , presupposition , logica l for m an d focu s stres s implicatures: th e case o f focal particles only an d also," Journal of Semantics 8-1 / 2, 127-147 . Bach, E . (1970) "Questions, " Linguistic Inquiry 2, 153-166 . Baker, C . L . (1970 ) "Note s o n th e Descriptio n o f Englis h Questions : Th e Rol e o f a n Abstract Questio n Morpheme, " Foundations of Language 6, 197-219 . Barwise, J. , an d R . Coope r (1981 ) "Generalize d Quantifier s an d Natura l Language, " Linguisitics and Philosophy 4 , 159-219 . Belletti, A. (1990 ) "O n th e morpholosyntacti c nature of th e sequenc e 'Aux+Pas t participle' i n Italian, " i n J . Mascar 6 an d M . Nespor , eds. , Grammar in Progress, 25-32.
Focus and Topic Movement in Korean and Licensing 33
1
Blok, P . I. (1991 ) "Focu s and presupposition," Journal of Semantics 8-1/2 , 149-165 . Bresnan, J . (1970 ) "O n Complementizers : Toward s a Syntacti c Theory o f Complemen t Types," Foundations of Language 6 , 297-321 . Brody, M. (1990) "Remark s o n tfc. e Orde r of Elements i n the Hungarian Focus Field, " in I. Kenesei, ed., Approches to Hungarian. Vol. 3: Structures and Arguments, JATE , Szeged, 95-121 . Browning, M. (1987) Null Operator Constructions, Doctoral dissertation, MIT, Cambridge , Massachusetts. Chang, S.-J. (1973) A Generative Study of Discourse: Pragmatic Aspects of Korean with Reference to English, Languag e Research 9.2. (Supplement), Seou l National University, Korea . Chang, S.-J. (1990) "Discourse contex t and grammar," in Ene (Language) 15 , Linguistic Society o f Korea , Seoul , Korea, 499-538. Choe, H.-B. (1928/35) Wuli Malpon (The Grammar of the Korean Language), Chengumsa, Seoul, Kore a (the eighth versio n (1980)). Choe, H.-S. (1985) "Remark s o n Configurationality Parameters," i n S. Kuno et al., eds., Harvard Studies in Korean Linguistics I, Hanshin Publishing Company , Korea, 14 — 29. Choe, H.-S. (1987) "Syntactic Adjunction, A-chain and the ECP—Multiple Identical Cas e Construction i n Korean," NELS 17, University of Massachusetts, Amherst . Choe, H.-S . (1988) Restructuring Parameters and Complex Predicates—A Transformational Approach, Doctoral dissertation , MIT, Cambridge, Massachusetts . Choe, H.-S. (199I/i n prep.) "Restructuring , Head-movement and IP structure," paper presented a t Seoul Internationa l Workshop on Generative Grammar , Seoul, Korea . Choe, J.-W . (1985 ) "Pitch-Accen t an d q/wh Words in Korean, " i n S . Kuno , et al., eds., Harvard Studies in Korean Linguistics I, Hanshi n Publishin g Company , Korea , 113-123. Choe, J.-W . (1987) "L F and Pied Piping, " Linguistic Inquiry 18.2. , 348-353. Choi, K . Y. (1989) "L F Movement : the Wh-island Constraint, " i n S . Kuno , e t al. , eds., Harvard Studies in Korean Linguistics III, Hanshi n Publishing Company, Korea , 213-234. Chomsky, N. (1964 ) Current Issues in Linguistic Theory, Mouton , The Hague . Chomsky, N. (1970) "Remark s o n Nominalization," in R. Jacobs an d P. Rosenbaum, eds., Readings in English Transformational Grammar, Waltham. Chomsky, N. (1971 ) "Dee p structure , surfac e structure, and semantic interpretation, " i n D. D. Steinberg an d L. A. Jakobovits, eds., Semantics, Cambridge University Press , Cambridge, 183-231 . Chomsky, N. (1973 ) "Conditions ; o n Transformation," i n S . Anderson, an d P . Kiparsky, eds., A Festschrift for Morris Halle, Holt, Rinehart and Winston, New York, 232286. Chomsky, N. (1976) "Condition s o n Rules of Grammar," Linguistic Analysis 2, 303-351. Chomsky, N. (1977 ) "O n WH Movement," in P . Culicover, T . Wasow, and A. Akmajian, eds., Formal Syntax, Academi c Press , Ne w York, 71-132. Chomsky, N. (1981) Lectures on Government and Binding, Foris , Dordrecht . Chomsky, N. (1982 ) Some Concepts and Consequences of the Theory of Government and Binding, MI T Press, Cambridge , Massachusetts. Chomsky, N. (1986a) Knowledge of Language: Its Nature, Origin and Use, Praeger, New York. Chomsky, N. (1986b) Barriers, Linguisti c Inquiry Monograph Thirteen, MIT Press, Cambridge, Massachusetts.
332
Discourse Configurational Languages
Chomsky, N. (1989 ) "Som e Notes on Economy of Derivation and Representation," i n I. Laka an d A. Mahajan, eds., Functional Heads and Clause Structure, MI T Working Paper s i n Linguistic s Vol. 10, MIT, Cambridge, Massachusetts. Culicover, P . W. (1991) "Topicalization, Inversion, an d Complementizer s in English," in D. Delfitto, et al., eds., Going Romance and Beyond, OT S Working Papers, University o f Utrecht , Utrecht. Diesing, M . (1988 ) "Bar e Plura l Subject s and the Stage/Individua l Contrast, " ms. , University o f Massachusetts at Amherst. Dik, S . C. (1980 ) Studies in Functional Grammar, Academic Press , Ne w York. Emonds, J. (1976) A Transformational Approach to English Syntax, Academic Press, New York. En$, M . (1991 ) "Th e Semantics of Specificity, " Linguistic Inquiry 22 , 1-25 . Fiengo, R., and J. Higginbotham (1981) "Opacity i n NP," Linguistic Analysis 7 , 395^22. Fiengo, R., C.-T.J. Huang, H. Lasnik, and T. Reinhart (1988) "The Syntax of Wh-in-Situ," in H . Borer , ed., Proceedings of the Seventh West Coast Conference on Formal Linguistics, 81-98 . Fodor, J. , an d I . Sa g (1982 ) "Referentia l an d Quantificationa l Indefinites," Linguistics and Philosophy 5, 355-398. Goldsmith, J . (1981 ) "The Structure of wh-Questions in Igbo, " Linguistic Analysis 1, 367-393. Grimshaw, J. (1979) "Complement Selection and the Lexicon," Linguistic Inquiry 10 , 279326. Haig, J . H . (1976 ) "Shado w Pronou n Deletion i n Japanese," Linguistic Inquiry 1, 363 371. Han, H.-S . (1987) The Configurational Structure of the Korean Language, Doctora l dissertation, The Universit y of Texas at Austin. Higginbotham, J . (1980 ) "Pronoun s an d Boun d variables," Linguistic Inquiry 11 , 679 708. Higginbotham, J., an d R. May (1981 ) "Questions , Quantifier s and Crossing," Linguistic Review 1 , 41-80. Holliday, M.A.K. (1967) "Note s on transitivity and theme in English: Part 2," Journal of Linguistics 3 , 177-274 . Hong, S.-S. (1985) A and A-bar Binding in Korean and English: Government and Binding Parameters, Doctora l dissertation , UConn . Hooper, J . B. , an d S . A. Thompso n (1973 ) "O n th e Applicability of Roo t Transformations," Linguistic Inquiry 4 , 465-497. Horvath, J. (1986) FOCUS in the Theory of Grammar and the Syntax of Hungarian, Foris , Dordrecht. Huang, C.-T. J. (1982 ) Logical Relations in Chinese and the Theory of Grammar, Doctoral dissertation , MIT, Cambridge, Massachusetts. Jackendoff, R . (1972) Semantic Interpretation in Generative Grammar, MIT Press, Cambridge, Massachusetts. Jo, M.-J . (1986) Fixed Word Order and the Theory of the Pre-verbal Focus Position in Korean, Doctora l dissertation, Universit y of Washington. Kayne.R. (1983 ) "Connectedness, " Linguistic Inquiry 14 , 223-249. Kim, A . H.-O. (1982) " A Universal of WH-Questions and its Parametric Variants, " in R. Schneider, K . Tuite , an d R . Chametzky , eds., Papers from the Parasession on Nondeclaratives, Chicag o Linguistic Society, 107-118 .
Focus and Topic Movement in Korean and Licensing 33
3
Kim, A. H.-O . (1985 ) The Grammar of Focus in Korean Syntax and Its Typological Implications, Doctoral dissertation , University o f Souther n California . Kim, S.-W. (1989 ) "Th e Q P Status of Wh-Phrases in Korean and Japanese," in E. J. Fe e and K . Hunt , eds., Proceedings of the Eighth West Coast Conference on Formal Linguistics, 358-372 . E. Kiss , K . (1987) Configurationality in Hungarian, Reidel , Dordrecht . Kratzer, A. (1989) "Stage-leve l and Individual-level Predicates, " ms., University of Massachusetts a t Amherst. Koopman, H . (1984 ) The Syntax of Verbs: From Verb Movement Rules in the Kru languages to Universal Grammar, Foris, Dordrecht . Kuno, S . (1973) The Structure of the Japanese Language, MI T Press, Cambridge , Mas sachusetts. Kuno, S . (1978) "Japanese : A Characteristic OV Language," in W. P. Lehmann, ed., Syntactic Typology, 57-138 . Kuroda, S.-Y . (1965) Generative Grammatical studies in the Japanese Language, Doc toral dissertation , MIT , Cambridge , Massachusetts . Laka, I. (1989 ) "Constraint s on Sentence Negation," MIT Working Papers in Linguistics 10, 199-216 . Lasnik, H. , an d M . Sait o (1984 ) "O n th e Natur e o f Proper Government, " Linguistic Inquiry 15 , 235-289. Lee, C.-M . (1989 ) "(In)definites , Cas e Markers , Classifiers an d Quantifiers i n Korean, " in S . Kun o et al. , eds. , Harvard Studies in Korean Linguistics HI, Hanshi n Pub lishing Company , Korea, 469-488 . Lee, E.-J . (1987) "EC P and Subjacency in LF and Hierarchical Orde r o f WH- Adjuncts," in S . Kun o et al. , eds. , Harvard Studies in Korean Linguistics II, Hanshi n Pub lishing Company , Korea, 93-103. Lee, H.-J . (1990 ) Logical Relations in the Child's Grammar: Relative Scope, Bound Variables, and Long Distance Binding in Korean, Doctoral dissertaion , U C Irvine . Mardcz, L . K . (1989 ) Asymmetries in Hungarian, Doctora l dissertation , Universit y o f Groningen. May, R . (1985 ) Logical Form: Its Structure and Derivation, MIT press , Cambridge , Massachusetts. Mitchell, E . (1991 ) "Evidenc e fro m Finnis h fo r Pollock's Theor y o f IP, " Linguistic Inquiry 22 , 373-379. Moon, G.-S . (1989 ) "O n Wh-Movemen t an d the Distribution o f Null Arguments," i n S . Kuno e t al. , eds. , Harvard Studies in Korean Linguistics III, Hanshi n Publishin g Company, Korea , 309-318 . Pollock, J.-Y . (1989 ) "Ver b Movement , Universa l Grammar , an d th e Structur e o f IP, " Linguistic Inquiry 20 , 365-424. Reuland, E.J., an d A.G.B. ter Meulen, eds., (1987) The Representation of (In)definiteness, MIT Press, Cambridge , Massachusetts. Rizzi, L. (1990 ) Relativized Minimality, MI T Press, Cambridge , Massachusetts . Rochemont, M. S., and P. W. Culicover (1990) English focus constructions and the theory of grammar, Cambridg e University Press, Cambridge . Ross, J . R. (1967) Constraints on Variables in Syntax, Doctora l dissertation . MIT , Cam bridge, Massachusetts . Saito, M . (1985 ) Some Asymmetries in Japanese and their Theoretical Implications, Doctoral dissertation , MIT, Cambridge , Massachusetts.
334
Discourse Configurational Languages
Suh, C.-M. (1989) "Wh-construction s in Korean," in S. Kuno et al., eds., Harvard Studies in Korean Linguistics /// » Hanshin Publishing Company , Korea, 517-526 . Suh, C.-S . (1989) "Interrogative s an d Indefinit e Words i n Korean : Wit h Referenc e t o Japanese," i n S . Kun o e t al. , eds. , Harvard Studies in Korean Linguistics HI, Hanshin Publishin g Company, Korea, 431-456. Whitman, J. (1989) "Topic , Modality, and IP structure," in S . Kuno et al., eds., Harvard Studies in Korean Linguistics III, Hanshin Publishing Company, Korea, 341-356 . Whitman, J. (1991 ) "String vacuou s V to COMP," ms., Cornell University . Williams, E. (1978) "Across-The-Boar d Rule Application," Linguistic Inquiry 9 , 31-43.
11 The Theory of Syntactic Focalization Based on a Subcategorization Feature of Verbs MI-JEUNG JO Pusan Susan University, Korea
1. Introduction A phenomenon suc h as scrambling is a diagnostic o f nonconfigurationality. It i s implied tha t scramblin g i s re:sponsible for som e degre e o f 'flat ' structur e (Hale , 1982). Language s wit h scramblin g suc h a s Korea n an d Hungaria n als o exhibi t movement t o configurationall y arranged A'-position s suc h a s Topi c an d Focu s positions. Sinc e syntactic Focusing uniformly requires immediate adjacency to a verb, the former precedes an d is hierarchically higher than the latter (cf . Horvath (1981), E. Kis s (1981 ) an d Jo (1986)) . This configurationa l phenomenon i n the so-called 'nonconfigurational ' language s has invite d various accounts . Two competing theories o n syntactic Focus movemen t have been propose d in Hungarian. Horvath (1981, 1986 ) argues that a verb assigns the feature [+focus] to a constituen t moved t o th e pre- V position ; assignment of Focu s unde r government an d adjacenc y is treated lik e assignmen t of abstrac t Cas e b y a verb . I t is adopted fo r Basque in Azkarate et al. (1982) an d Ortiz de Urbina (1983) , and for Korea n i n J o (1986) . E . Kis s (1987a ) propose s a n alternativ e analysi s of Hungarian Focus; the Focus position i s immediately dominated by S', and a constituent of S can be Focused by upwar d movement, whil e Topic is immediatel y dominated by S". 1 Followin g Ki m A.H-O (1985), 2 we will refer to this analysis as th e Multi-Leve l Hypothesi s an d t o Horvath' s proposa l a s th e Verb-Inheren t Hypothesis. Orti z d e Urbin a (1989) , alterin g his previou s position , propose s a similar analysi s in Basque; a Focused elemen t (an d an interrogative Wh-phrase) is moved to the specifie r po s ition of CP, which is a typical position for a n operator. In these analyses, the pre-verbal (henceforth, pre-V) Focus position has nothing t o d o wit h th e ver b itself ; th e adjacenc y to a ver b i s a resul t o f othe r 335
336
Discourse Configurational Languages
phenomena—i.e. the verb-initial basi c wor d order i n Hungarian and verb-move ment in Basque (i.e., the V-second phenomenon) triggered by an operator whic h is move d t o the specifier positio n o f CP. Th e Verb-Inheren t Hypothesi s ha s several advantage s tha t the Multi-Level Hypothesi s canno t offer . First , the con figurational relatio n betwee n Topi c an d Focu s position s i s naturall y accounte d for unde r the Verb-Inheren t Hypothesis; a Focused elemen t ha s t o ge t assigne d the Focus featur e from th e pre-V position, while a Topic constituent is interprete d by virtu e o f bein g in a n A'-position outside a n S . However, i n th e analysi s tha t both Topic and Focus are in sentence-external operato r positions, th e precedenc e and structura l superiority o f Topic ove r Focu s positio n i s virtuall y stipulated . Secondly, th e Verb-Inheren t Hypothesis ca n provid e a n explanatio n wit h ty pological relevance , s o that not only pre-verbal Focusin g i n language s like Ko rean, Hungarian , Basque, Armenia n an d Turkish but als o post-verba l Focusin g in Aghem can be regarded as two variants of one "verb-adjacent, verb-governed " Focus-assignment rul e (Horvat h 1981) . I n contrast, th e Multi-Leve l Hypothesi s cannnot sugges t a principle d accoun t fo r language s whos e syntacti c Focusin g commonly exhibit s th e immediat e adjacenc y t o a verb . Unde r thi s theory , eac h language is subjec t t o a language-specific account. Fo r example , w e cannot ex pect tha t a proposal made for the pre-verbal syntacti c Focusing i n Hungarian o r Basque sheds light on that in Korean, which has neither V-initial basic orde r no r V-movement triggere d b y a n operator i n the specifie r position . Given thes e advantages , w e wil l carefull y examin e a majo r proble m o f th e Verb-Inherent Hypothesi s pointe d ou t b y proponent s o f th e alternativ e hypoth esis—the questio n o f whethe r th e inheren t feature o f verbs , [+focus] , i s base d on thei r subcategorizatio n property . W e will attemp t t o renovate th e hypothesi s by proposin g a solutio n fo r the questio n an d other problem s o f th e Verb-Inher ent Hypothesis relate d t o it. In contrast t o languages like English , w e will sho w that languages with syntactic Focus suc h as Korean have an A'-position governe d by and adjacent to V. Armenian (Comrie 1984 ) an d Turkish (Underhill 1976) dat a (cf. note s 1 3 and 14 ) also suppor t this assumption; a Focused constituen t (o r an interrogative Wh-word ) in th e pre- V positio n i s i n complementar y distributio n with a nonargument such as a predicate constituent . Therefore, thoug h its existence has no t been widel y accepte d i n th e literatur e th e A'-position governe d b y and adjacen t t o V is th e essential characteristi c o f languages wit h syntactic Fo cus. The notio n o f VP-internal A'-position justified throug h Korea n Focu s con structions wil l be empiricall y teste d i n its Hungarian counterparts .
2. Focus Movement and the Two Configurational A'-Positions in Korean The ai m of this section i s to sho w how syntacti c Focus constructions i n Korea n are derive d b y movement s t o th e tw o A'-positions, Topi c an d th e pre- V Focu s positions. Thi s topi c require s understandin g tha t constituent s i n a Korea n sen tence are clearly divided into "scramblable" versus "nonscramblable" ones. Sec tion 2. 1 concern s th e former ; Cho e H.-S . (1988 , 1989)' s characterization s o f
The Theory of Syntactic Focaliiation 33
7
scrambling in Korean based o n its configurationality will be introduced. Sectio n 2.2 discusses nonscramblabl e constructions i n Korean. Finally, sectio n 2. 3 introduces derivatio n o f the Focu s construction s i n Korean . In Hale (1978)'s early characterization of nonconfigurational V-final languages such a s Japanese an d Korean, onl y the position o f a verb is fixed. 3 Furthermore , the category V in Korean does not undergo an y syntactic movemen t rule, excep t perhaps for movement like the string vacuous movement of V to COMP proposed by Whitma n (1991). Sinc e th e positio n o f a ver b i s fixed , w e will sho w tha t a verb an d a predicat e complemen t immediatel y precedin g i t als o constitut e a n unscramblable syntactic unit. We will furthe r sho w that Focus formation in Ko rean relie s o n th e configurationalit y sustained betwee n th e tw o constituent s u p to the PF component. This relation will be contrasted wit h the configurationality existing betwee n a verb an d argument s which ma y be dissolve d b y scramblin g in a later grammatica l level.
2.1. Scrambling of Arguments and the Configurational VP in Korean First, w e introduce Cho e H.-5J . (1988 , 1989)' s argument s fo r th e existenc e o f a configurational V P i n Korean . Despit e th e fre e wor d orde r phenomenon , sh e provides various syntactic motivations for postulating a VP node in Korean. First, Korean ca n hav e a VP topic whic h preposes th e constituen t mad e u p o f a n ob ject an d a verb, leavin g th e tense marker s (i.e. , the present tens e -0 o r -n ) with the pro-verb ha 'do' : (1) [ton-u l pel-ki]-nu n apeci-k a ha-si-0-k o [ton-u l money-Ace earn-i n g-Top father-Nor n do-Hon-Pres-an d money-Ac e ssu-ki]-nun emeni-k a ha-si-n-t a spend-ing-Top mother-Nor n do-Hon-Pres-De c 'As for making money, (my ) father does , an d a s for spendin g money , (my) mother does. ' Second, V P idioms ar e eas y t o fin d i n Korean , whic h consist o f a bar e objec t noun an d a verb. 4 This als o indicate s VP constituency. (2) Suil- i miyekkuk-u l mek-ess-t a Nom seawee d soup-Ac e eat-Past-De c 'Suil faile d i n an exam.' In addition to the straightforward evidence fo r a syntactic VP,5 Korean exhib its a subject-objec t asymmetr y wit h respec t t o bindin g conditio n C . Th e coreference possibilit y o f th e objec t pronou n i n (3a ) suggest s th e structura l su periority o f the subjec t NP : (3) a . Suilj-u y emem-k a ku a ;-lul salangha-n-t Gen mother-Nor n he-Ace love-Pres-De c 'Suil's mothe r love s him.'
338
Discourse Configurational Languages
b. * kurka Suilj - u y emeni-lu l salangha-n-t a he-Nom Ge n mother-Ac e love-Pres-De c 'He loves Suil' s mother. ' Korean als o show s cross-over effects . A sentence lik e (4b ) i s ungrammatical , given tha t a Korea n Wh-phras e in-sit u i s move d a t L F (cf . Huan g (1982) ) an d leaves a variable whic h canno t be th e anteceden t o f a pronoun t o it s left : (4) a . nookoo-k a pro/cak i (uy ) emeni-lu l salangha-0-pnikk a who-Nom h e Ge n mother-Ac e love-Pres- Q 'Who loves hi s mother?' b. * pro/caki (uy ) emeni-k a nookoo-lu l salangha-0-pnikk he Ge n mother-Ac e who-No m love-Pres- Q 'Who doe s hi s mother love? '
a
Choe furthe r show s tha t scramblin g i n Korea n i s clause-bounde d an d tha t i t does not affec t bindin g and variable binding; the scramble d sentence s i n (5 ) and (6) maintain th e sam e grammaticalit y a s the correspondin g non-scramble d ver sions i n (3 ) and (4) . (5) a . ku-lu l Suil-u y emeni-k a salangha-n-t a he-Ace Ge n mother-Nor n love-Pres-De c b. *Suil-u y emeni-lu l ku-k a salangha-n-t a Gen mother-Ac e he-No m love-Pres-De c (6) a . pro/caki j emeni-lu l nookoo ;-ka salangha-0-pnikk a he mother-Ac e who-Nom love-Pres- Q b. * nookoo-lul pro/caki j emeni-k a salangha-0-pnikk who-Acc h e mother-Nor n love-Pres- Q
a
Based o n the abov e test , Cho e conclude s tha t scramblin g i n Korean , i n contras t to tha t i n Japanese, 6 is a non-syntactic phenomenon. Since th e structur e of a sentence i n Korea n needs a configurational VP, Choe H.-S. assume s tha t scrambling is a nonconfigurational phenomenon occurrin g at the leve l o f the PF component wher e the Projection Principl e n o longer nee d b e observed. Accepting Hale (1982)' s positio n that languages with scrambling sho w some degre e o f fla t structure , she argue s that scramblin g i n Korea n i s triggere d by th e applicatio n o f a Restructurin g Rul e whic h categoricall y an d morpho logically amalgamate s th e thre e categorie s V , INFL and COM P (i.e. , sentenc e endings)7 an d 'flattens ' th e structur e of a sentence; fo r her , C and INF L i n Ko rean hav e th e feature s [ + CD] (i.e. , categoria l dependency ) an d [ + MD ] (i.e. , morphological dependency ) i n th e lexicon , an d the y trigge r over t restructuring illustrated i n th e followin g phrase-structur e markers (Choe H.S . 1989:282) .
The Theory of Syntactic Focalization 33
9
After th e amalgamatio n o f INF L an d C wit h a verb , subjec t an d objec t NP' s become sister s a s in (7b) , an d scrambling is possible i n th e altere d structur e i n PR 2.2. Non-Scramblable Pre-Verbal Constituents in Korean Despite th e amalgamatio n of V-INFL-COMP in al l finit e sentence s i n Korean , pre-V constituents strictly subcategorized by verbs are not subject to scrambling. They include nonarguments such as predicate NPs, PPs and APs; they are strictly subcategorized, but they have no theta-role. The italicized constituent s i n the following sentence s canno t b e altere d b y scrambling , an d n o constituen t ca n b e moved i n between a verb an d a predicate complemen t (th e parenthese s i n eac h sentence mar k the possible positio n fo r a n adverb): (8) a . Suil - i ( ) hoecang-i ( * mancangilchilo) toe-ess-t a Nom chairman-Part unanimousl y become-Past-De c 'Suil became chairma n unanimously.' b. Suil - i ( ) sinsa-lo ( * encena) poi-n-t a Nom gentleman-a s alway s seem-Pres-De c 'Suil alway s seems t o be a gentleman.' c. Suil - i ( ) phikonha-e ( * onul) poi-n-t a Nom tired-In f toda y seem-Pres-De c 'Suil seem s t o be tire d today. ' d. Sunae-k a Suil- ul ( ) chinku-lo (* hangsang) yeki-n-t a Nom Ac e friend-a s alway s consider-Pres-De 'Sunae alway s considers Sui l a friend. '
c
In contrast , an adver b can freel y interven e between a subjec t N P an d th e ver b (with a predicate complemen t immediately preceding it ) (8a)-(8c ) an d between
340
Discourse Configurational Languages
an objec t N P an d the ver b (8d) , a s the possible positions for th e adverb s i n th e above indicate. The data in (8) show that scramblin g in Korea n i s limited t o arguments; a verb and an immediately preceding predicate complement ar e immune from nonconfigurationa l phenomena. In the Government-Bindin g (GB) framework, deletion i s another typ e of rul e occurring between S-structur e and the PF component. Both arguments and a verb selecting a n argumen t i n Korea n ca n underg o deletio n unde r identity . Forwar d deletion i s applie d t o the identica l objec t N P in th e sentence s i n (9) , whil e th e backward gappin g i s applied t o the verb -ilk 'read ' i n the sentence s i n (10) : (9) a . Suil - i k u cip-u l sa-ss-k o Sunae - ka k u cip-u l Nom th e house-Ac e buy-past-an d N M th e house-O M cangsikha-ess-ta decorate-past-dec 'Suil bought the house, an d Sunae decorated th e house.' b. Suil- i k u cip-ul sa-ss-k o Sunae-ka 0 cangsikha-ess-t a 'Suil bought, and Sunae decorated th e house.' (10) a . Suil- i sinmun-u l ilk-0-k o Sunae-k a chaek-u l Nom newspaper-Ac e read-Pres-an d No m book-Ac e ilk-nun-ta read-Pres-Dec 'Suil reads a newspaper, an d Sunae reads a book.' b. Suil- i sinmun-u l 0 Sunae-k a chaek-ul ilk-nun-t a 'Suil read s a newspaper, an d Sunae , a book.' However, a predicate NP selected by an intransitive verb is not subject to forward deletio n under identity, a s shown in (11) . No r ca n verbs selectin g a predi cate N P underg o backwar d gapping , an d s o th e ver b shoul d b e repeated , a s in (12): (11) a . Sunae-u i apeci-nu n uisa-i-ess-k o Sunae-nu n machimna e Gen father-To p doctor-is-past-an d To p finall y uisa-ka toe-ess-t a doctor-Norn become-past-dec 'Sunae's fathe r wa s a doctor, an d Suna e finally ha s become a doctor.' b. *Sunae-u i apeci-nu n uisa-i-ess-ko Sunae-nu n machimnae 0 toe-ess-ta (12) a . Suil- i hoecangi toe-ess-k o Sunae-k a Nom chairman-Par t become-Past-an d No m pu-hoecang-i toe-ess-t a vice-chairman-Part become-past-Dec 'Suil became chairman , and Sunae became vice-chairman. ' b. *Suil- i hoecang-i 0 Sunae-ka pu-hoecang- i toe-ess-t a
The Theory of Syntactic Focalization 34
1
A predicate NP and a verb selecting such a complement behave differently fro m an argumen t and a ver b selectin g i t wit h respec t t o deletion , a s th e contras t between (9 ) an d (11 ) an d that between (10 ) and (12) demonstrate . Both the pre-V predicate complement and a verb selecting i t are excluded fro m the two PF phenomena, scramblin g and deletion. Following Horvat h (1981, 1985) , Jo (1986) attributes the resistance of the pre-V predicate complement and the verb to scramblin g an d deletion t o their constituenthood ; as in Hungarian, this combination constitutes a n intermediate projection, V (i.e. , "a small VP"). Therefore, we assume that a Korean sentence like (8d) has the structure (13) in D-structure, which ha s th e tw o projection s o f V (i.e., th e intermediat e V an d th e maxima l VP):
As far a s the two constituents are concerned, Korean seems to be qualified a s in a configurational language, as defined i n terms of the Projection Principle by Hale (1983:26): (14) Th e Configuratioriality Paramete r a. I n configurationa l languages, the Projection Principl e hold s o f th e pair ( LS (lexica l structure), PS (phonologica l structure)) . b. I n non-configurational languages , the Projectio n Principle holds of L S alone .
342
Discourse Configurational Languages
2.3. Movements To Topic and Focus Positions No predicate complement s i n the pre-V position i n (8) can be scrambled. Amon g them, however, the predicate NP in the sentence (8a) can be proposed to the Topic node, which is an A'-position locate d in the leftmost position o f a sentence, a s in (15a). Th e Topi c positio n i n Korea n i s a landin g sit e fo r a n N P o r a PP , bu t a predicate PP in a sentence lik e (8b ) cannot be topicalized becaus e of its proper ties simila r t o a n AP (cf. Jo (1986 ; Ch . 4)), a s show n in (15b) . (15) a. [
hoecang]-un [c Suil-i (mancangilchilo ) t toe-ess-ta ] chairman-Top No m unanimously become-past-De 'As fo r chairman, Sui l becam e one. ' b. *[ ppsinsa-lo]-nun [ c, Suil-i (enceyna ) t poi-n-ta ] gentleman-as-Top No m alway s seem-Pres-De c Np
c
Sentence (15a) shows that movement to the Topic nod e is qualitatively differen t from scramblin g i n Korean. 8 Topicalizatio n o f a nonargumen t suc h a s a predi cate complement is predicted by the Theta Criterion; an A'-position ma y provide a landin g sit e fo r eithe r a n argument or a nonargument. The Topi c positio n i n Korea n ha s bee n know n a s a base-generate d A'-posi tion, sinc e a Topic whic h ha s a wider semanti c scop e tha n the subjec t NP , as in the followin g sentence , ha s n o original position withi n the clause : (16) kkoc-u n [ c cangmi-ka yeppu-0-ta ] flower-Top rose-Nor n pretty-Pres-De c 'As fo r flowers, rose s ar e pretty.' The exac t structura l location o f th e Topi c nod e depend s o n ho w w e defin e a finite clause i n Korean. In Choe H.-S. (1988) , root clauses i n Korean are defined as a n endocentric categor y headed b y COMP , as illustrate d i n th e phrase-struc ture marke r in (la). Sh e argues tha t post-tense verba l suffixe s whic h determin e the typ e o f sentence s ar e COMP ; the y ar e obligator y i n roo t clauses , an d the y are i n complementar y distributio n wit h a n embedde d claus e nominalizer complementizer suc h as -ki. Although Wh-Q phrase s i n Korea n d o not underg o Wh-movement, the y have to be in the scope of the interrogative sentenc e marke r -ni I -upnikka; otherwise , the y are interpreted a s quantified phrases, a s shown i n (17). The interrogative Wh-phras e in (17a) can be freely scramble d withi n a clause as in (18a), but it cannot be preposed t o the position o f Topic, a s in the sentenc e (18b). (17) a . Suil- i muess-u l sa-ss-n i (SO V order ) Nom what-Ac c buy-Past- Q 'What di d Suil buy? ' b. Suil- i muess-u l sa-ss-upnit a Nom what-Ac c buy-Past-De c 'Suil bough t something.'
The Theory of Syntactic Focalization 34
3
(18) a . muess-u l Suil - i sa-ss-n i (OS V order ) what-Acc Nor n buy-Past- Q 'What did Suil buy?' b. *[ cp muessi-un [ c, Suil-i t ; sa-ss-ni ] what-Top No m buy-Past- Q The impossibilit y o f structur e (18b ) i s attribute d by Cho e H.-S . (1988 ) t o th e structural position of Topic whic h i s outside the scope of the interrogativ e sen tence marke r (i.e., COMP); the Wh-phrase in (18b) , muess 'what ' is not within the scope o f the interrogative ending -ni. Therefore, the logical position fo r Topic constituents mus t be higher tha n COMP—the Spec positio n immediatel y domi nated by CP (cf. (7a)); it is an A'-position optionall y license d b y COMP, the head of a finite clause. 9 In conjunctio n wit h preposin g a pre- V constituen t (i.e. , a predicat e NP ) t o Topic, a subjec t N P i n (15a) , Suil ca n b e furthe r move d t o th e empt y pre- V position t o ge t Focused. Wit h th e subjec t move d t o th e pre- V position , phono logical primac y falls o n it, an d th e Topic interpretatio n fo r th e prepose d predi cate NP in (15a ) disappears , even if th e Topic nod e is filled : (19) [
hoecang-u n [c, [!p t; ( ) [ vp [v, SUILri ( * mancangilchilo) chairman-Top No m unanimousl y toe]]- ess]-ta]] become-Past-Dec 'It is SUI L who became chairma n (unanimously). ' CP
In surfac e structur e a Topi c constituen t may b e linearl y adjacen t t o a Focuse d constituent, as suggested in the Focus construction (19) . However , no evidence establishes tha t the pre-verba l Focu s positio n i s locate d outsid e th e clause . O n the contrary , th e impossibl e positio n o f adverb s marke d i n (19 ) provide s com pelling evidence that the pre-verbal Focus position i n Korean coincides wit h that of th e pre- V predicat e constituent s given i n (8)—th e positio n governe d b y an d immediately adjacen t to a verb. 10 Focus constructio n formation in Korean, as shown in (15a) an d (19), i s a substitution whic h involve s movement s to th e two A'-positions, th e Topic nod e li censed b y COMP and the pre-V Focus positio n license d by a verb which select s a predicat e N P complement. I t i s extremely restricted , becaus e dislocatio n o f a predicate complemen t t o the Topic positio n i s limited t o a predicate N P (cf. the Topic sentence s i n (15)) . A constituent which can get Focused i n Korean seem s to be reduce d t o the subjec t NP in a sentence lik e (8a ) (cf . it s Focuse d versio n (19)). But ther e i s a way out fro m thi s limitatio n in Korean; a n object NP , a PP and a subject N P i n a sentence whos e mai n predicat e doe s no t subcategoriz e a pre-V constituen t can ge t Focused i n clef t sentence s a s in (20). Th e structur e of the cleft Focu s sentenc e is analogous to the Focus construction (19) :
344
Discourse Configurational Languages
(20) a . [Suil- i san kes]-u n [ c, tj k u CHAEK ri-0-ta] Norn buy-Pas t thing-To p th e book-is-pres-De c 'It is the BOOK tha t Sui l bought.' b. [Suil- i kan kos]-u n [c t; DUBROVNIK r i-0-ta ] Nom go-Pas t place-To p is-Pres-De 'It i s DUBROVNIK wher e Suil went.' c. [ ku chaek-u l sa- n salam]-u n [c t; SUIL-i-0-ta] the book-Ac e buy-Pas t person-To p is-Pres-De 'It is SUI L wh o bought th e book.'
c c
The un-Focused versio n of the sentences abov e are assumed to be similar t o (8a) . For example, (20a ) is derived fro m a sentence lik e the following, where the pre V position i s occupie d b y th e NP which is appositiv e to the subjec t NP: (21) [ c. [jp ku chaek- i [ v, [NP Suil-i san kes ] -i]-0]-ta ] the book-Nor n No m buy-Past thin g -is-Pres-De c 'The book i s what Sui l bought. ' The copul a -i 'is ' i s on e o f the verb s whic h select s a predicate N P withou t th e Case particl e -ka (cf.(8a)) , an d thus the Focused constituent s of the sentence s i n (20) ar e unambiguously in the pre- V position . I n other words , th e derivatio n of the cleft Focu s sentence s appear s t o be subject to the identical limitatio n a s that of (19) ; onl y th e N P i n th e subjec t positio n ca n b e move d t o th e pre- V Focu s position b y substitution . To summarize, Focus formation in Korean, though extremely restricted, clearl y exhibits th e tw o propertie s significan t for th e theor y o f syntacti c Focusing : (1 ) the pre-verba l Focu s positio n coincide s wit h th e positio n o f predicat e comple ments, whic h i s governe d b y an d immediatel y precede s a verb ; (2 ) th e pre- V position is a n A'-position i n which subcategorized nonarguments occur. At leas t in th e case of Korean , the tw o propertie s ar e inseparably related ; th e latte r fol lows fro m th e former . Th e firs t propert y lead s u s t o adop t th e Verb-Inheren t Hypothesis, rather tha n the Multi-Level Hypothesis; namely, a Focused elemen t is move d t o th e pre- V positio n fo r th e assignmen t o f th e featur e [+focus ] b y a verb. Thi s positio n furthe r ha s t o the n admi t th e existenc e o f a VP-internal A' position.
3. The Critiqu e o n the Verb-Inherent Hypothesis Reexamine d 3.1. Subcategorization and Assignment of Focus The first questio n that the Verb-Inherent Hypothesis encounters i s whether ther e is any ground for assuming the feature [+focus] as an inherent property of verbs , analogous t o Cas e o r theta-role features . Ortiz d e Urbin a (1989 ) take s u p thi s questio n i n hi s criticis m o f Horvath' s (1981, 1985 ) analysis of Focus (an d WH-Q phrases) in Hungarian. He points out
The Theory of Syntactic Focalization 34
5
that he r analysis , whic h considers [+focus ] t o b e a featur e assigne d b y V o n a par wit h othe r feature s assigne d b y V, is essentially a stipulation, sinc e ther e is no selectiona l restrictio n betwee n verb s an d foci ; althoug h a syntacti c featur e [+focus] woul d be simila r t o a Cas e featur e i n th e structura l restrictio n o n it s assignment (i.e., governmen t and adjacency), this does not explain why it shoul d be the verbs tha t assig n the focu s feature. His criticism explicitly indicate s tha t the stipulativ e natur e o f Focus-assignmen t b y V wil l b e remove d i f i t ca n b e related t o a subcategorizatio n propert y characterizin g th e categor y V. Curiously enough , Horvat h herself , wh o originall y propose d th e Verb-Inheren t Hypothesis, actuall y agree s wit h Ortiz de Urbina's statemen t o f the problem b y making th e followin g remark : (22) Cas e i s intimatel y tie d t o subcategorization , wherea s th e notio n "FOCUS" has nothin g to do with it. (Horvat h 1985:128 ) The above statement exactly echoes Ortiz de Urbina's criticis m o f her own analysis. Althoug h sh e treat s bot h Focu s an d Cas e a s syntacti c feature s assigne d b y the categor y V , she admit s tha t there i s a fundamental difference between the m with respect to subcategorization . Horvath's characterization o f the pre-V position (i.e . th e landin g site for FO CUS/WH-Q), however , counter s the intuitio n stated i n (22) ; sh e establishe s th e pre-V node as a base-generated position in which a significant class of verbs take s a subcategorized complement (Horvath 1985: Ch . 1) . She therefore concludes that this positon i s an A-position (i.e., a potential theta-position) an d thus, the rule of FOCUS/WH-Q movement in Hungarian is movement into an A-position, although a trace lef t by such a movement is not an anaphor, but a variable. I n other words , there is a discrepancy in her description o f the nature of the feature [+focus] and that o f th e pre- V Focu s position ; whil e th e forme r supposedl y ha s n o relatio n with the subcategorizatio n propert y of a verb, the latte r is licensed by it. The conflicting properties o f the syntactic feature [+focus] an d the pre-V Fo cus position i n Horvath's analysi s contribute to undermining the gist of the VerbInherent Hypothesis she advocates, o n the one hand, and it invites further criticis m of th e hypothesis , o n the othe r hand : (1) if th e syntacti c featur e has nothin g t o do with subcategorization, why does a Focused constituen t i n Hungarian hav e t o be moved into the pre-V A-position where a subcategorized complement occur s (i.e., i n the case of substitution)? (2) how can a Focused constituen t wit h a thetarole b e move d int o th e pre- V A-positio n withou t violatin g th e Theta-Criterio n (E. Kis s 1987a , Ki m A.H-O 1985 , J o 1986) ? Contrary to the identical observations mad e by the proponents o f the two com peting theorie s o f syntacti c Focu s (i.e. , th e Verb-Inheren t vs . th e Multi-Leve l Hypotheses), we will argu e that assignmen t o f the feature [+focus ] i s intimatel y related t o a subcategorizatio n propert y o f V. Unlike a Case feature , however, i t is not directl y recognizable from the verb itsel f (i.e. , its lexical feature). There fore, th e featur e itsel f is not a subcategorizational feature of V, but thi s does no t necessarily mean that it has nothing to do with subcategorization. The version of the Verb-Inheren t Hypothesis we propos e her e wil l eliminat e th e discrepanc y
346
Discourse Configurational Languages
between th e featur e [+focus ] an d th e pre- V positio n wit h respec t t o subcategorization an d thu s wil l sho w tha t Orti z d e Urbina's criticis m i s no t a n authentic proble m fo r the hypothesis . In sectio n 3.1.1 , w e wil l conceptualiz e a potentia l A'-positio n license d b y a subcategorizational featur e characterizin g th e categor y V (i.e. , a n A'-positio n within a VP). Th e syntacti c Focu s position , whethe r i t be pre-verbal o r a postverbal, i s such a n A'-position whic h is governed by and adjacent to a verb. Con sequently, unlik e Horvath , w e analyz e Focu s movemen t a s movemen t t o a n A'-position. This characterizatio n o f the pre-V (o r post-V) A'-position i s crucia l for approachin g othe r relate d problem s o f the Verb-Inheren t Hypothesi s fro m a new perspective . I t dispenses wit h the problems concernin g th e Theta-Criterio n and Cas e conflict . In addition, we will also solv e the inconsistency i n predictin g the directionalit y o f governmen t fo r Focu s an d Cas e assignmen t i n Hungarian, from whic h Horvath' s versio n o f th e Verb-Inheren t Hypothesis suffers . Sectio n 3.1.2 discusse s a conflict between the Projection Principl e an d th e Theta Crite rion occurrin g i n the syntacti c Focus construction s i n Korean . 3.1.1. Predicate Complements and a Potential A'-Position in VP It is well known that the categor y V has th e richest complemen t syste m amon g lexical categories . I n the Government-Binding theory, each position satisfyin g the subcategorization feature s o f the lexica l hea d o f a construction i s considered a s a theta-positio n excep t fo r idiom chunk s (Chomsky 1981:35) . Therefore, a con cept suc h a s ' a subcategorize d nonargument ' doe s no t exis t i n thi s theory . I n contrast t o thi s assumption , we wil l sho w tha t V doe s subcategoriz e a comple ment to which it assigns neither Case nor a theta-role. Predicate nominals , predi cate adjective s an d predicate PP's ar e such constituents. Since th e selectio n o f predicate complement s i s a subcategorization propert y characterizing V, Emonds (1985:43) attribute s the asymmetry in the noun and verb complement system , no t onl y to a verb' s abilit y t o Case-mar k (Chomsk y 1981 , Stowell 1981) , bu t t o it s abilit y t o selec t predicat e complements . Fo r instance , we find tha t verbs which take predicate nominals either do not have correspond ing derive d nominals , o r their derive d for m cannot tak e a predicate nou n with out a n introductory as (Emond s 1985:43) : (23) * I was disappointed b y John's unexpecte d remaining a cook . The becomin g {of/as } a n adult entails responsibility. Her appearanc e {as / * 0 } the unwelcom e guest wa s embarrassing . Since th e so-calle d 'linkin g verbs ' suc h a s remain, become an d appear i n (23 ) do no t assig n Case , th e asymmetri c Case-markin g abilit y o f N an d V canno t account for why derived nominal s never tolerat e predicat e nouns . The example s in (23 ) indicat e tha t th e abilit y t o selec t predicat e complement s i s a more gen eral propert y o f V than it s abilit y to assig n Case . We also find tha t AP sisters to V are tolerated, whil e those to N are not, eve n though predicat e adjective s ar e no t Case-marke d b y V . This als o suggest s tha t
The Theory of Syntactic Focalization 34
7
the ability to select predicate complement s i s a source o f the asymmetry between N an d V (Emonds 1985:43) : (24) Joh n appeare d reluctan t to leave . *We were surprise d b y John's appearanc e reluctan t t o leave . That desser t taste d sweete r tha n candy. That dessert's taste sweete r tha n candy overwhelmed us. Some verb s als o tak e predicat e nominal s an d adjective s attribute d t o a n ob ject NP, as in (25). Theses constructions have been known as small clauses sinc e Stowell (1981) . The corresponding derive d nominal s also canno t select a predicate complement : (25) a . Joh n consider s Mary ( a fool/rich} . b. *John' s consideratio n o f Mary { a fool/rich} . c. Bil l prefer s hi s stea k rare . d. *Bill' s preference o f his stea k rar e e. W e elected John secretar y f. *0u r electio n o f John secretar y The predicate nominal s and adjectives in (23), (24) and (25) neither receive Cas e nor a theta-role fro m a verb , thoug h the y ar e subcategorize d b y V ; the y rathe r assign themselve s a theta-role t o a subjec t o r a n object NP. 11 The verb' s abilit y to selec t predicat e complement s i s a universal propert y o f V. The base-generated Pre- V constituent s i n Korea n presented i n sectio n 2. 2 at test to this property of verbs. It suggests that V in fact ca n license an A'-position which i s immun e fro m bein g assigne d Cas e an d theta-role . Althoug h thi s position is normally occupied by predicate complements, we expect i t to be a candidate for a potential A'-position whic h an operator suc h a s a Focused constituen t and a WH-Q phrase can move into without violating the Theta-Criterion o r caus ing Cas e conflict . I t resemble;; an A'-position outside a n S, although i t is located within a VP of an S. Therefore, w e can say that V has a n ability t o license a n A'position b y virtu e o f a subcategorizatio n propert y whic h select s predicat e complements. Obviously, th e positio n o f predicat e complement s doe s no t automaticall y qualify a s a potential landin g site o f Focus i n all languages. Sinc e the structura l conditions fo r Focus assignment , adjacency an d government, must be observe d as in the cas e o f abstrac t Case-marking, languages whose positio n fo r predicat e complements canno t satisf y the m ar e rule d out . Fo r instance , th e positio n fo r predicate nominal s and adjective s i n Englis h is not alway s adjacent to V, as th e sentences i n (25 ) show , and syntactic Focusing (i.e., the pre - o r post-verbal Fo cus by movement ) is not possible in thi s language. Similary, th e abstrac t Case assignment rule canno t work in a language such a s Korea n wher e the position s of th e subjec t o r th e objec t ar e no t alway s adjacen t t o th e Case-assigners . The syntactic Focu s mechanis m onl y work s i n language s wher e th e potentia l A' -
348
Discourse Configurational Languages
position (i.e . th e predicate complemen t position) i s adjacen t to V. Since Korea n has the predicate complement s i n the pre-V position , a s shown in sectio n 2.2 , i t has a potential A'-position adjacent t o V, where Focus assignment can take place.12 We assume tha t othe r languages with syntacti c Focu s als o hav e suc h a A'-position governed by and adjacent t o V. For instance, Armenian (Comrie 1984) 13 and Turkish (Underhil l 1976) 14 dat a sugges t tha t th e pre- V Focu s positio n i n thes e languages correspond s t o th e positio n o f predicat e complements . W e wil l tes t whether thi s prediction is borne ou t i n Hungaria n pre-V constituent s in sectio n 3.2.1. Given the analysis of the A'-position license d b y V as a potential landin g site for Focus , w e need n o stipulation to predict a language-specific directionality of Focus assignment . I t accord s wit h th e positio n wher e a subcategorize d non argument occur s i n a language . I f suc h a constituen t occur s i n th e pre- V posi tion, Focus assignmen t is leftward (a s in Korean), whereas i t would be rightward (presumably in Aghem), i f it occupies the post- V position. Thus , in ou r versio n of th e Verb-Inheren t Hypothesis , th e directionalit y o f Focu s assignmen t i s no t subordinated to the directionality of government for Case-marking and theta-marking. Th e direction of Focus assignment ca n be opposite to the Case-marking, as we se e i n Hungarian. We hav e mentione d tha t Horvath' s Verb-Inheren t Hypothesi s ha s th e majo r advantage ove r th e Multi-Leve l Hypothesi s i n tha t i t ca n provid e a typologica l explanation fo r bot h pre-verba l an d post-verba l syntacti c Focu s i n variou s lan guages, as two variants of one "verb-adjacent, verb-governed " Focu s rule . How ever, a s pointe d ou t b y Ki m A.H- O (1985:333-34) , he r predictio n o f th e directionality o f Focu s assignmen t suffers fro m a theory-interna l inconsistenc y in the analysis of Hungarian, on which her theory of Focus ha s been established . She says that the particular Focus positio n in a given languag e is determine d by a "directionality parameter," a s a property of the notion of government in Universal Grammar. In her theory, the directionality parameter is language-specific , and it should comply with a universal principle referred to as "Head Peripheralit y Condition" (26) : (26) Hea d Peripheralit y Conditio n (Horvat h 1985:60) : Given th e general P S rule schem a o f th e for m H» - (^...(9 - H-1 -(C j+1)...(Ck), in an y given instantiation of the rul e schema , either (C t ) ...(C p or (C j+1)...(Ck) mus t be 0. According t o th e Hea d Peripheralit y Condition , (a ) complements mus t b e gov erned b y thei r heads , an d (b ) th e governmen t relatio n i s unidirectional—i.e. , complements to any particular head-of phrase occur on one side of the head, rather than o n both sides . Thus , sh e predicts tha t Focus assignmen t i n Hungaria n i s in unidirectional with Case assignment. Recall that the pre-V position is an A-position i n he r analysis , and thu s bot h Focus an d Cas e assignmen t are predicte d t o be leftward . Contrar y to her prediction , in the followin g structure of Hungarian
The Theory of Syntactic Focalization 34
9
sentences give n by her the object NP occurs on the right side of V (i.e. , a pre-V constituent plu s a verb) :
Therefore, Hungaria n is the clea r example tha t directionalit y o f government fo r Focus assignmen t i s independen t from tha t fo r Case assignment . To su m up , thi s sectio n show s tha t th e syntacti c Focu s positio n adjacen t t o verbs i s grounded in a subcatagorization property o f verbs. Unlike other catego ries, verb s ca n selec t predicat e complement s to which neither Cas e no r a thetarole is assigned, and thus they can in additio n license a VP-internal A'-position . Although thi s attribut e of verb s is universal, onl y language s wit h syntacti c Focus are assumed to have the A'-position immediately adjacent to V, so as to meet the adjacency conditio n fo r Focus assignment. Then, movement t o the pre-V (o r post-V) Focus positio n shoul d be movement to an A'-position, rather than movement to an A-position a s proposed b y Horvath. This analysi s o f the pre-V Focus position ca n dispens e wit h th e problem s tha t he r proposa l ha s concernin g th e Theta-Criterion and Case conflict. Our prediction about the directionality of Focus assignment depend s o n the position wher e nonarguments are selected by a verb in a language . 3.1,2. Focus Assignment: A Conflict between the Projection Principle and the Theta-Criterion Both the subcategorization feature [+pre d comp ] and the syntactic feature [+focus] are inherent properties of the category V involving the constituent governe d by an d adjacent to it . Therefore, correlatin g th e tw o feature s inevitably poses a problem o f violating the Projection Principl e i n (28), more specificall y violatin g the subcategorization feature [+pre d comp ] , though such a violation is allowed by the Theta Criterion in (29) (i.e., an argument can be moved into the preverbal nonargument position) .
350
Discourse Configurational Languages
(28) Th e Projectio n Principl e (Chomsk y 1981:29 ) Representation a t each syntacti c level (i.e., LF, and D - and S-structure) ar e projected fro m th e lexicon , i n tha t they observ e th e subcategorization propertie s of lexica l items . (29) Th e Theta-Criterio n (Chomsk y 1981:36 ) Each argumen t bears on e an d onl y one theta-role, an d eac h theta-rol e is assigne d to on e and only on e argument. The analysi s o f Focu s assignmen t in th e presen t pape r bring s ou t a theoretica l tension existin g between the two principles o f UG, which has not been reveale d before b y other phenomena. We have shown in section 2. 2 that nonarguments in Korean observ e th e Projectio n Principl e i n a manner differen t fro m arguments , and w e will furthe r revis e th e principle t o accommodate th e violatio n occurrin g in th e preverbal Focu s assignmen t of Korea n (cf . (19)) . Although w e defined th e pre-V Focus position as an A'-position licensed by a verb in the preceding section, the gap between the subcategorization property of V and the featur e [+focus ] ha s not been bridged yet. The featur e [+focus ] is different fro m th e othe r syntactic features of V such a s Case an d theta-role, which are als o based on it s subcategorization properties, i n that the two latte r features are directly relate d t o the lexical properties of a verb. Therefore, the y are represented a t D-structure , thoug h Cas e assignmen t i s formall y checke d a t S-struc ture. I n contrast, the featur e [+focus ] is not detectable fro m th e ver b itself . I t i s the featur e of a ver b availabl e onl y a t S-structur e (i.e. , afte r Focu s movement ) as the inpu t to th e representations o f PF an d LF. The ga p betwee n th e subcategorizatio n propert y characterizin g th e categor y V and the feature [+focus] requires a conceptual reconciliatio n unde r our analysis. Given the fact tha t the featur e [+focus] is a feature of V available onl y a t Sstructure, w e sugges t tha t th e subcategorizatio n featur e [+pred.com p ] obligatorily "turns into it," whe n a moved argument occupies this position originally allowe d fo r predicat e complements . Unde r thi s vie w o f th e syntacti c fo cusing mechanism , bein g i n a nonargumen t positio n make s a n argumen t syntactically prominent, and the Focused constituen t finally acquire s phonologi cal an d semanti c prominenc e a t th e level s o f P F an d LF , respectively. I n othe r words, th e featur e [+focus ] i s th e othe r sid e o f th e subcategorizatio n featur e [+pred.comp ] , whic h i s assigne d t o a non-base-generate d pre- V constituen t under adjacenc y and government . In ou r analysi s o f Focu s assignment , w e no w se e ho w th e relatio n betwee n the subcategorizatio n propert y o f V an d th e featur e [+focus ] create s a tensio n between the two principles of UG, the Projection Principle an d the Theta-Criterion. If a verb licenses a VP-internal A'-position by virtue of a subcategorizatio n property, thi s means that it ca n b e a landing site for a n argument , provided tha t it is empty. In other words, the subcategorization feature [+pred comp ] (or [+ pred comp]) can be 'properly ' violate d in a certain context such as Focus assignment. Eve n thoug h it is VP-internal, an A'-position, by definition , is a potentia l landing site of an argument and thus, the feature [+pred. comp ] is opened t o a
The Theory of Syntactic Focalization 35
1
licit violation in the grammatical component where 'move alpha' is applied. Sinc e the Thet a Criterio n allow s th e violation , Focu s assignmen t present s a cas e i n which it is directly in conflict with the Projection Principle. I n fact, an y versio n of th e Verb-Inheren t Hypothesi s for Focu s assignmen t encounter s a dilemm a between th e tw o principle s of UG (e.g . Horvath' s analysis). 15 In the Government-Binding (GB) framework, however, the two principles hav e been assume d to be completely in accord. The Projection Principl e (28 ) mus t be observed o n ever y leve l o f th e gramma r except fo r th e P F component . Idiom s aside, eac h argumen t is assigned a theta-role by virtue of the theta-position tha t it o r its trac e occupie s i n LF a s to satisf y th e Theta-Criterion (29) . Sinc e ever y syntactic representatio n shoul d b e a projectio n o f themati c structur e (i.e . al l subcategorized position s mus t b e theta-marked ) i n G B theory , an y subcatego rization entail s theta-marking . I n other words , a VP-internal nonargument posi tion an d movemen t t o thi s positio n ar e no t theoreticall y allowed ; th e theor y predicts tha t a n A'-position must be a t leas t outsid e a n S . Therefore, a conflic t between the two principles has never arise n in this framework. The Multi-Leve l Hypothesis o n Focus-assignmen t (e.g . £ . Kis s 1981 , Orti z d e Urbin a 1989 ) i s compatible with the implications of the Projection Principle (28) proposed in GB theory, and thus it does not have the problem that th e Verb-Inherent Hypothesi s has to deal with . Now tha t a subcategorized. A'-position an d Focus assignmen t t o a constituent moved t o thi s positio n i n language s lik e Korea n empiricall y demonstrate s th e conflict betwee n the Theta Criterio n and the Projection Principle,16 a revision of the latter i s required; it i s reformulated a s holding only for arguments : (30) Th e Projectio n Principl e (Revised ) Representation a t each syntactic level (i.e. , LF , and D - an d S-structure) ar e projected fro m th e lexicon , i n that they observ e th e argument structure s of lexica l items . Idiomatic constructions will provide empirical support to our claim that a verb selecting a nonargumen t in Korea n license s th e pre-verba l A'-positio n i n orde r to assign th e feature [+focus ] to an argument moved to it. I t is well know n that an objec t N P i n a n idio m ha s n o themati c role , an d i t i s a n exceptio n t o a requirement o f th e Projection Principl e (28 ) (i.e. , non-revise d version) , namel y 'subcategorization entail s theta-marking ' (Chomsk y 1981:37) . In this sense , the object N P in an idiom is similar to a predicate complement subcategorized b y a verb. Conside r the idio m presented i n (2) , repeate d i n (31a) : (31) a . Suil- i miyekkuk-u l mek-ess-t a Nom seawee d soup-Ac e eat-past-de c 'Suil failed in the exam.' b. miyekkuk-u l Suil-i mek-ess-ta 'Suil at e seawee d soup. '
352
Discourse Configurational Languages
c. Suil- i miyekkuk-u l elma-cen-e y mek-ess-t a Nom seawee d soup-Ac e sometime-ago-i n eat-past-de c 'Suil ate seaweed sou p sometim e ago. ' The object N P in (3la) miyekkuk a s a nonargument cannot hav e its literal mean ing 'seawee d soup' ; nor does th e verb hav e its lexical meanin g 'eat ' i n the idi omatic expression ; the y constitute a complex predicat e phrase . However , a s soo n as th e objec t N P i s scramble d a s i n (31b) , i t i s immediatel y interprete d a s a n argument (i.e. , 'seawee d soup' ) a s the glos s shows . I t i s als o interprete d a s a n argument whe n the adverbial phrase in (31c), elma-cen-e 'sometim e ago,' inter venes betwee n th e ver b an d th e objec t NP . The dat a abov e indicat e tha t a nonargument objec t N P in an idiom i s like a predicate N P adjacent t o a verb in the sentenc e (8a) ; i t is interpreted a s a nonargument only if it is adjacent to and governed b y a verb selectin g it . The nonargumen t i n a n idio m exhibit s a furthe r similarit y t o a predicat e complement i n Topicalization an d syntactic Focusing, a s well. Whe n th e objec t NP without a thematic role in (3la) is proposed to the Topic position , th e idiom atic reading ca n be maintained, as in (32a), while scrambling i n (31b) forces th e same NP to be interpreted as an argument. More significantly, the verb mek- 'eat ' as a verb selectin g a nonargument object N P in the idio m can licens e th e pre- V Focus positio n and assigns the feature [+focus] to the subject N P moved int o it, as in (32b) : (32) a . miyek-kuk-u n [ c. Suil-i [ v, t mek]-ess-ta ] seaweed-soup-Top No m eat-past-de c 'As fo r failin g in a n exam, Sui l did.' b. miyek-kuk-u n [ c, t; [ v, SUILri mek]-ess-ta ] seaweed-soup-Top No m eat-past-de c 'It i s SUIL wh o failed i n a n exam.' Based on the behavior of the object NPs in an idiom a s in (31 ) and (32) , w e can treat the m a s a subclas s o f nonargument s occurring i n th e pre- V positio n (i.e. , predicate complement s selecte d b y a verb). 17 I f w e admi t th e existenc e o f a subcategorized nonargument, the object NP in idiomatic expression s i s no longe r an isolated exception to the Projection Principle. The verb mek- in the idiom (3 la), therefore, has the subcategorization feature [+pre d comp ] just like the verb toe 'become' in (8a) , despit e th e differen t Cas e realize d o n the surface . To summa rize, b y examinin g th e ver b in an idiom selectin g a nonargument object N P we have empirically supporte d ou r claim that the inherent feature of a verb [+focus ] is th e othe r sid e o f th e subcategorizatio n featur e [+pre d com p ] . Th e assign ment o f th e forme r i s triggere d b y th e violatio n o f th e latte r i n S-structur e a s predicted b y th e Theta Criterio n (29) and the revise d Projectio n Principl e (30) .
The Theory of Syntactic Focalization 35
3
3.2. Empirical Justifications for the Preverbal Nonargument Position in Hungarian A considerable numbe r o f languages have pre-V (o r post-V) syntacti c Focusing . The two competin g theories o n syntactic Focusin g hav e entirel y differen t posi tions on this quite prevalent phenomenon. As we have emphasized, th e most sig nificant advantag e of the Verb-Inherent Hypothesi s i s tha t it ha s th e possibilit y to generalize ove r the pre-verbal and post-verbal Focu s beyond language-specifi c differences, whil e th e Multi-Leve l Hypothesi s ha s t o rel y o n language-specifi c aspects o f each languag e to account for the pre-verbal syntactic Focusing i n languages like Korean, Hungarian, Armenian, Turkish and Basque. In this sense, th e Verb-Inherent Hypothesi s intend s t o b e universa l i n it s essentia l claim . There fore, i t i s crucia l t o tes t whethe r th e specifi c versio n o f th e Verb-Inheren t Hy pothesis w e have described i s borne out in other languages with a syntactic Focu s mechanism. In th e following , w e wil l empiricall y tes t th e versio n o f th e Verb-Inheren t Hypothesis we advocate in Hungarian, which has Focus movement as well as WhQ movemen t t o th e pre- V position . Mor e specifically , w e wil l inquire whethe r the notion of A'-position licensed by V and the process of assigning the featur e [+focus] based on subcategorization is also valid for the Hungarian counterparts. Section 3.3. 1 examine s non-Focuse d pre- V constituent s i n Hungarian . Though they are more heterogenous than the pre-V constituents in Korean, we will show that al l o f the m involve a predicate complemen t o r a par t o f a comple x predi cate. The pre-V node in Hungarian also functions as an aspect marke r of a verb. We wil l attemp t t o integrat e thi s language-specifi c propert y int o th e cross-lin guistic characterization o f the pre-V node as the A'-position governed b y a verb . 3.2.1. Non-Focused Pre-Verbal Constituents in Hungarian Since th e propert y o f th e bass-generated pre- V constituent s i n Korea n (cf . sec tion 2.2 ) i s the basi s fo r our theor y of Focus assignment , i t ha s primar y impor tance whether o r not i t also holds in a language like Hungarian. The two point s in questio n concernin g non-focuse d pre- V constituent s i n Hungaria n are : (1 ) whether they ar e constituents which are not directly theta-marked by a verb (a s opposed to Horvath); (2) whether they are base-generated constituents in the preV positio n selecte d b y a verb (as opposed t o E. Kiss). 18 As ha s alread y bee n discussed , Horvat h (1981, 1986 ) argue s tha t th e pre- V node i n Hungaria n is a n A-position, wher e a subcategorize d argumen t occurs . However, w e fin d predicat e complement s appearin g in this positio n amon g th e data give n by her. A predicate adjectiv e may appea r i n th e pre-V position , a s in the followin g example fro m Horvat h (1986:54): (33) a . A gyereke k nagyo n biiszk6 k volta k a dijr a the childre n ver y proud-pi , wer e th e prize-ont o 'The childre n were ver y prou d of the prize.' b. * A gyereke k volta k nagyo n btiszke k a dijr a the childre n wer e ver y proud-pi , th e prize-ont o
354
Discourse Configurational Languages
The ungrammaticalit y of th e sentenc e i n (33b ) show s tha t th e positio n o f thi s predicate constituent cannot be altered i n a non-Focused sentence . Sinc e th e preV constituent above (i.e., a predicate adjective) , though selected by the verb, ha s nothing t o d o wit h theta-marking, it doe s no t suppor t Horvath' s clai m tha t th e pre-V nod e i s a n A-position. A subcategorize d P P by a ver b lik e tesz 'put ' i s claimed t o be one of the non-Focuse d pre- V constituents , whic h constitutes th e most important basis for her characterization of the pre-V position. However , th e data involvin g such a pre-V constituen t (Horvath 1985:54 ) appea r t o be contro versial.19 Within the analysis of E. Kiss (1987a:44), al l constituents appearing in the preV position are moved there from a n S, including the Focus an d interrogative wh phrases, as the diagram in (34) show s (cf. note 1) . Therefore, non-Focuse d pre- V constituents ar e not base-generated .
Despite the different analysi s for defining the structure of the pre-V position, he r observations about these constituents are strikingly reminiscent o f the propertie s of th e base-generate d pre- V constituent s in Korean , as i n th e followin g remar k (E. Kis s 1987a:76) : (35) "Positio n F of the Hungarian sentence, i.e., the peripheral position mini mally c-commandin g S , immediatel y dominate d b y S 1, has turne d ou t to b e a multi-purpos e operato r position , ope n t o variou s type s o f operators taking scope ove r S : a focus, a focus also functioning as an interrogative operator, th e nominal/adverbia l par t o f a complex predi cate, o r th e adverbia l par t o f a comple x predicat e als o functionin g as an aspectua l operator. " Notice that except for a Focus operato r (an d a Focus functionin g as an interroga tive operator), al l pre-V constituents are nonarguments involving a complex predi cate. The complementar y distribution relation betwee n th e variou s non-Focuse d constituents an d a Focu s operato r i n th e pre- V position i s th e weakes t poin t of the Multi-Leve l analysis of the Hungaria n Focus . These non-Focu s constituent s
The Theory of Syntactic Focallzation 35
5
are strictl y subcategorize d b y a verb. However , thei r positio n withi n a sentenc e is no t fixed , a s X" * in th e phras e structur e (34 ) indicates . Th e complementar y distribution relation between the Focus an d a non-Focused constituen t in the preV nod e raise s tw o questions : (1 ) wh y i s a par t o f a comple x predicat e no t di rectly inserted i n the pre-V position in D-structure?; (2) what motivates movemen t of suc h constituent s to the pre- V nod e withou t getting Focused ? Bearin g thes e two questions in mind, we will examine the sentences involving non-Focused pre V constituent s give n by E. Kiss . In line with her observation in (35), E. Kiss provides ample examples involv ing pre-verba l predicate complements or a part of a complex predicat e (E . Kis s 1987a:62):20 (36) a . [ s,, Mno s [ s. 'orvos [ s lesz ]] ] ('= [ 1 stress] ) John docto r become s 'John becomes a doctor.' b. [ s ,, Jano s [ s , 'iigye s [ s volt]]] John skillfu l wa s 'John wa s skillful. ' c. [ s,, Jano s [ s. 'levelet [ s i r ]]] John lett;r-Ac c write s 'John i s letter-writing. ' d. [ s,, Ja"no s [ s. 'moziba [ s ment]] ] John cinema-t o wen t 'John wen t to the cinema.' e. [ s. Janos [ s,' fel [ smen t a le"pcson]] ] John u p wen t th e stair s 'John wen t upstairs.' The non-Focuse d pre- V constituent s in (36a ) an d (36b ) ar e prototypica l predi cate complements , namel y a predicat e N P and AP, while thos e i n (36c) , (36d ) and (36e ) ca n b e terme d 'incorporate d constituents ' o f a comple x predicate , though E. Kiss treats both types equally as the nominal/adverbial parts of a complex predicate, a s stated i n (35) . In addition to the predicate complements in (36a) and (36b), th e sentence (36c ) also provide s interestin g evidenc e for definin g th e pre-V nod e a s a VP-internal A'-position. The determinerless noun in the pre-V position, level-et 'letter-Ace ' is marked with the accusative ; Case, but accordin g to E. Kiss's observation , it is interpreted a s an nonargument—i.e., a nominal part of the complex predicat e 't o letter-write.' I t is very simila r to the object N P without a theta-role in a n idio m in Korean, as given in (31a); thoug h phrasal, levelet-ir 'letter-write' is interprete d as a singl e verb. 21 O n th e othe r hand , whe n a full-fledge d objec t N P (e.g . a n argument N P wit h a determine r suc h a s a level-et 'th e letter-Ace.' ) appear s i n the pre- V position , i t mus t bi s Focused; suc h a n argument normall y occur s i n a post-verbal position in a non-Focused sentence.22 E. Kiss' s movemen t analysis, however, counters her own observation ; if th e phrasa l construction trul y consti tutes a complex predicate, it simpl y shoul d be inserte d in th e pre- V positio n a t
356
Discourse Configurational Languages
D-structure, an d the pre-verbal bar e objec t N shoul d b e a VP-internal constitu ent.23 The nonargument object NPs in (3la) an d (36c) demonstrate tha t th e preV nod e bot h i n Korea n an d Hungaria n i s no t onl y reserve d fo r predicat e complements, bu t als o serve s a s a n 'extended ' nod e fo r th e categor y V (i.e., a part o f a complex verb) . Hungarian has a set of verb-particles, know n as 'verbal prefixes' i n traditional grh function a s directional adverbials. 24 Th e particle occurrin g i n the pre-V positio n of (36c) i s an instance of this construction. Like the predicate complements, thes e particl *be-van 't o be-in'), among others. Horvat h (1978b, 1985:55 ) emphasizes tha t a characteristic of Hungarian phrase structure is that particular verbs must be strictly subcategorized t o tak e (o r no t t o take ) on e o f thei r complemen t phrase s i n th e pre-V node, even though she claims that it is a regular argument position. E . Kiss's proposal (i.e. , movement to the pre-verbal A'-position outside an S) is an implausible solutio n because i t doe s no t try t o capture th e relation betwee n a non-Fo cused pre-V constituen t and a verb, whethe r the former is a phrase o r a part of a complex predicate ; i f strictl y subcategorized , a constituent shoul d b e base-gen erated an d governed b y a verb. Ou r proposal tha t th e pre- V nod e is a n A'-posi tion license d b y a verb , however , i s compatibl e wit h th e occurrenc e o f language-specific subcategorize d nonargument s in the pre-V position, such as the verbal particle s i n Hungarian. The pre-V nod e in Hungarian also plays a role in aspectual interpretation , a s observed by Horvath (1981, 1985) , Kiefer (1982 ) an d E. Kiss in (35). Th e gram mar of Hungarian has no systematic an d productive morpholog y fo r marking th e distinction between progressive an d perfective aspect, a s languages lik e Englis h do. Typically Hungarian sentences ar e vague with respect t o progressive vs . per fective aspect, and the actual interpretation depends on context. Though the views of thes e linguist s var y abou t th e rol e o f non-Focuse d pre- V constituents, 25 th e general agreemen t i s that the y ar e involve d in som e wa y o r other i n expressin g aspect. For E. Kis s (1987a:69-76) , Focus an d aspect markin g are in complementar y distribution i n th e A'-position dominate d b y S' (whic h is adjacent to V). Fo r in stance, sh e observe s tha t th e adverbia l par t o f a comple x predicat e suc h a s fel 'up', o r le 'down ' appearin g in thi s positio n ca n onl y functio n a s a n aspectua l marker whe n a Focuse d constituen t doe s no t exist . A sentenc e containin g a n adverbial fel 'up ' i n th e pre- V positio n (b y movemen t t o th e Focu s slo t i n he r analysis) wil l b e interprete d a s perfective , a s i n (37a) . I f i t i s lef t behin d i n S , and th e Focus positio n i s empty, the aspec t o f the sentenc e i s either progressiv e (37b), o r 'existential, ' expressin g tha t the action ha s at least onc e taken place i n the pas t o r wil l tak e plac e a t leas t onc e i n th e futur e (37c ) (E . Kis s 1987a:69 70): (37) a . [ s,, Janos [ s, ' fel [ s men t a lepcstfn]] John u p wen t th e stair s 'John went/ha s gon e upstairs.'
] ('= [ 1 stress])
The Theory of Syntactic Focalization 35
7
b. [, . Hno s [,.[,'ment'fe l a '16pcs6n]] ] John we:n t u p th e stair s 'John wa s going upstairs. ' c. [ s,. Jdno s [ s ,[ s ' ment fe l a 16pcs6n]] ] John we;n t u p th e stair s 'John has alread y gone u p the stairs. " In othe r words , i f a Hungaria n sentenc e i s syntacticall y marke d fo r aspect , i t cannot contain a Focus; i f it contains a Focus, it cannot be syntactically marke d for aspect . I f two elements ar e mutually exclusive, i t usually indicate s tha t the y share some common grammatical property. However, it is difficult t o detect suc h a propert y (o r properties) relatin g a Focused constituen t an d aspect markin g i n E. Kiss' s analysis . The complementar y distributio n betwee n Focu s an d th e as pect markin g in Hungarian, therefore, is purely accidenta l i n her analysis . While fo r E . Kis s only a n adverbia l par t i n a comple x predicat e suc h a s fel 'up' i n (37) involves aspect markin g in Hungarian , as explicitly state d i n (35), the surfac e position o f an y base-generated pre- V constituen t determine s aspec t marking i n Horvath's (1985:77-79) analysis; depending o n whether i t is locally postposed o r not, aspect o f a ver b i s eithe r progressiv e o r perfective . Sh e thu s provides thre e interpretiv e template s fo r aspec t base d o n this observation: 26 (38) Interpretive Template for ASPECT (Horvat h 1985:78 ) a. [ v. Xmax V ] > PERFECTIV E b. [ v. V Xmax ] > PROGRESSIVE c. Elsewher e > VAGUE She assumes, however, the pre-V node i s a regular A-position. Fo r suc h analy sis, i t i s a myster y why thi s particula r A-position (i.e. , th e pre- V node) , rathe r than a post-verbal A-position, is employed to express aspect. Futhermore , aspec t marking i s no t generall y directl y relate d t o a n argumen t o f a verb , bu t i t i s a specification o f a ver b itself . I n brief, Horvath fail s to full y integrat e th e func tion o f the pre- V nod e i n aspec t markin g into her characterizatio n o f th e pre- V node, an d thus it remains a n accidental phenomenon. 27 Although the role o f the pre-V node in determining Hungarian aspect i s indi rect (i.e. , th e nod e filled wit h a base-generated pre- V constituen t vs. th e empt y node), w e assum e that thi s functio n i s nontrivial and tha t it mus t be a relevan t structural propert y o f th e pre- V position . Neither doe s E . Kis s no r Horvat h at tempt to relate suc h a function t o each's structura l characterization o f the pre- V node (i.e., an A'-position immediately dominated by S' for the former, and an Aposition governed by V for the latter). We will attempt to integrate this function of the pre-V node into our characterization o f it, namely its being a n A'-positio n licensed b y a subcategorization propert y o f a verb. In orde r to defin e a proper structura l locus fo r a n aspectua l device , i t seem s to be a prerequisite to examine the category INFL (or AUX), since this categor y is closel y relate d t o aspect-marking . The categor y INF L has bee n mor e o r less
358
Discourse Configurational Languages
established i n Universa l Gramma r eve r sinc e Chomsky' s (1957 ) proposa l fo r English. It is distinct in its syntactic behavior fro m othe r syntactic categories such as a categor y V—i t label s a constituen t that include s element s expressin g th e notational categorie s o f Tense, Modalit y an d Agreement. Althoug h certai n fea tures of INFL may vary from languag e to language, possible variation s ar e sup posedly restrictivel y parameterize d (i.e. , th e presenc e o r nonpresenc e o f agreement). What i s significan t fo r ou r discussio n i s tha t in recen t studie s o f INFL , fea tures involvin g the aspec t o f a ver b suc h a s Progressive an d Perfectiv e ar e no t included amon g th e possibl e categorie s postulate d unde r th e INF L nod e a t D structure. This suggest s tha t elements expressin g aspec t ar e generated unde r the VP node, i f a language has a n overt aspec t system . English has an explicit and productive aspec t system , and its interaction wit h INFL (AUX ) has been wel l studie d fro m th e earliest stage s o f generative gram mar. Chomsk y (1957 ) propose s tha t th e auxiliar y verbs expressin g aspec t i n English, have an d be, are generated by the phrase structure rule fo r AUX, as i n (39): (39) AU X >
Tense (Modal ) (have+en) (be+ing)
Although the rule (39 ) adequatel y captures the ordering constraint s among auxiliaries i n English , variou s subsequen t work s o n Englis h AU X (o r INFL ) (Jackendoff 1972 , Emond s 1976 , 1985 , Culicove r 1976 , Akmajian , Steel an d Wasow 1979 ) sho w that the two auxiliary verbs fo r aspect, have and be, behav e like verbs through their regular inflection with respect to Tense and VP-deletion.28 The structure of predicates in a sentence like Bill could have been studying Spanish i s give n as follows (Akmajian, Stee l an d Wasow 1979) :
The Theory of Syntactic Focalization 35
9
In either Culicover (1976)' s or Akmajian, Stee l an d Wasow (1979)'s representa tion o f aspect, the two important structural features of aspectual verbs i n English for ou r discussio n are : (1 ) the y ar e generate d b y th e bas e rul e fo r VP ; (2 ) th e position o f the aspectual auxiliar y verbs i n VP is adjacent to INFL on the oppo site sid e o f branchin g fo r arguments , sinc e aspec t shoul d furthe r interac t wit h tense. W e assum e tha t thes e tw o propertie s concernin g aspec t markin g ar e universal, despite language-specifi c differences at work in the actua l expressio n of aspect . Returning to aspect marking in Hungarian, the indirect role of the pre-V node seems t o be cruciall y related t o its structura l position i n a sentence (i.e. , its po sition wit h respec t t o th e VP node an d INFL). Thoug h Hungaria n has n o over t aspectual auxiliar y generated unde r VP, whether the pre-V node in a VP is empty or not can express aspec t by virtue of its structural properties. Horvat h (1985:64) explictly locate s INFL preceding th e VP as shown in phrase marke r (27). 29 The pre-V nod e in (27) is compatible with the two structural properties extracte d fro m the analyse s o f th e aspectua l auxiliar y verbs i n English, thoug h th e nod e is di rectly governe d by th e hea d V; the pre-V nod e is dominated by VP and is adja cent to INFL on the side where no argument appears. 30 Therefore, it seems to be no acciden t tha t the pre- V nod e in Hungaria n functions a s a n aspec t marker , if we postulat e it a s an A'-position adjacen t to a verb. Summarizing, thoug h th e rol e o f th e pre- V nod e functionin g as a n aspec t marker i n Hungarian seems t o be a highly language-specifi c phenomenon , i t i s unlikely tha t thi s languag e resort s t o a totall y ad hoc mean s o f expressin g as pect, such as making use o f an operator positio n dominate d by S ' (E. Kiss) or of a regular A-position wher e an argument may occur (Horvath) . There is of cours e no compellin g a prior i argumen t tha t a n A'-position subcategorize d b y a ver b should functio n a s a n aspec t marke r a s well . However , ou r characterizatio n o f the pre- V nod e i n Hungaria n (i.e. , a n A'-position license d b y th e categor y V ) seems t o be a plausible account for it s indirec t role in expressin g aspect , give n that there is an independently postulated INFL node in Hungarian which precedes the VP , as i n th e phras e marke r propose d by Horvat h (27) , an d tha t th e pre- V node is no t o n th e "argumen t side" of the V. To reca p thi s section , th e examinatio n o f th e non-Focuse d pre-verba l con stituents in Hungarian has show n that they ar e selecte d nonargument s such a s a predicate complement , a detenninerless object NP or a verbal particle forming a part of a complex predicate . Th e various non-Focused pre-verba l constituent s in Hungarian confir m our characterizatio n o f th e pre- V nod e a s a n A'-position li censed b y a verb , whic h has been previousl y establishe d throug h analyzin g the Korean counterpart s whic h have les s variety . Our examinatio n o f non-Focuse d pre-verbal constituent s is limite d to onl y Korean and Hungarian . However, th e Armenian (Comri e 1984 ) an d Turkis h (Underbil l 1976 ) dat a (cf . note s 13 , 14 ) suggest that our theory is in the right direction. Further research o n the nature of constituents occurrin g in th e pre- V (o r Post-V ) positio n i s necessar y i n other languages wit h syntacti c focalization .
360
Discourse Configurational Languages
3.2.2. Extended Focus Assignment: The Abstract Pre-Verbal Node for Focused Constituents and Wh-Q Phrases in Hungarian This section examines whether the process o f assigning the feature [+focus] base d on subcategorization is valid in Hungarian. It will also address the different prop erties o f the pre-verbal Focus nod e in Korean and Hungarian; in the former onl y verbs which select a predicate complement license the pre-verbal Focu s position , while i n th e latte r an y verb , includin g of cours e verb s selectin g a pre-verba l constituent, ca n licens e th e syntacti c Focu s positio n int o whic h both Focuse d constituents an d interrogative Wh-phrases ar e moved . Our analysi s of syntacti c Focusin g i n Korea n vi a assignmen t o f a n inheren t feature by a verb has the following implications for other languages with similar phenomena. Th e notio n o f assignin g th e featur e [+focus ] o n th e basi s o f th e subcategorization featur e [+pre d com p ] severely restrict s th e possible deriva tions o f syntacti c Focu s constructions . Sinc e th e featur e [+focus ] i s dependen t on the subcategorization featur e [+pred com p ] , its assignment is possible only when a verb select s a pre-verbal nonargument . Accordingly, Focus movemen t i s essentially limited to a substitution process; a Focused constituent replaces a basegenerated pre-V constituent. It means that the potential Focus positio n (pre- V o r post-V position ) must b e fille d b y a lexica l ite m i n D-structur e an d tha t th e nonarguments base-generated in thi s position canno t get focused . Under Horvath's Verb-Inherent Hypothesis, Focus movement in Hungarian also exhibits substitution ; base-generated pre- V phrase s mus t b e postpose d i n orde r that a Focused phras e can be moved int o the pre-V position. 31 The effec t o f Focusing a n argumen t in a sentenc e wit h a base-generate d pre- V complemen t i s demonstrated i n the followin g constructions (Horvath , 1985:101): 32 (41) Substitutio n Proces s fo r Derivin g Hungaria n Focus Construction s a. Mar i asztalra tett e a z edenyeke t (Unmarke d fo r Focus ) table-onto pu t th e dishes-Ace . 'Mary pu t the dishe s o n a table.' b. Mar i AZEDENYEKET ; tett e asztalra l{ (th e Object N P Focused ) the dishes-Ace pu t table-ont o 'It i s THE DISHES tha t Mary put on a table.' The assignmen t o f th e featur e [+focus ] i s consisten t wit h a subcategorizatio n property o f the ver b in th e above constructio n an d thus the restrictio n o n deriv ing Focu s construction s i s satisfied. Focu s movemen t i n Hungarian, however, i s still allowed , eve n i f the ver b i n a sentence doe s not select a pre-V constituent . This indicates tha t the pre-verbal Focu s positio n i n this language is licensed no t only b y th e subcategorizatio n featur e o f a specifi c verb , bu t i s a n abstrac t A' position license d b y al l verbs ; th e potentia l feature [+focus ] i s a n inheren t fea ture o f th e categor y V . As a result , th e derivatio n o f Focu s construction s i n Hungarian i s muc h more productiv e than i n Korean. The abstrac t pre- V Focu s positio n i n Hungaria n now seem s t o counte r ou r analysis o f th e pre-verba l A'-positio n a s license d b y a ver b an d it s
The Theory of Syntactic Focalization 36
1
subcategorization property . This aspect of syntactic Focusing i n Hungarian, however, shoul d b e examine d in the ligh t of other related phenomen a whic h distin guish i t fro m Korean . Hungaria n and the othe r language s known with syntacti c Focusing suc h a s Basque, Armenian , Turkish an d Aghem, unlike Korean , hav e obligatory Wh-Q Movement to a landing site identical to that of a Focused con stituent (i.e. , the pre-V position o r the post-V position) . We assume tha t a n abstract pre-V node, licensed no t by a specific class of verb but by the category V, is someho w due t o this fact . In Hungarian, like syntacti c Focusing, a Wh-Question als o involve s substitution (cf . Horvath 1985 : sectio n 1.7) . Whe n the object NP in (41a) i s questioned, the base-generate d pre- V constituen t must b e postposed , s o tha t th e interroga tive Wh-phrase can be move d to its place, a s shown below: (42) Mit j tet t asztalra t ; Mar i ? what-acc. pu t table-ont o Mar y 'What di d Mary put o n a table?' The natural answer to the requested question (42) should be a Focused sentenc e like (41b) , wher e the requested information occupie s th e pre-V Focu s position ; an unfocuse d sentenc e like (4la ) canno t be a well-formed response to questio n (42).33 Furthermore, i n Hungaria n a Wh-Question formation occurs, whethe r a ver b in a sentenc e select s a pre-V constituen t or not ; a n interrogativ e Wh-phras e i s obligatorily move d int o th e abstrac t pre- V A'-position a s i n (43b) , eve n wher e the verb select s no pre-V constituent. The correct respons e t o this questio n i s a Focus sentence (43c ) (th e sentenc e (43a ) i s its non-Focused version) : (43) a . Mar i olvast a a z ujsdgo t Mary read-def . th e newspaper-Ace . 'Mary rea d the newspaper.' b. Mit ; olvasot t t ; Mari ? What-Acc. read-indef. Mar y 'What did Mary read?' c. Mar i A Z OlSAGOT j olvast atj Mary th e newspaper-Ac e read-def. 'It is THE NEWSPAPER that Mary read.' Due to the obligatory Wh- Q phrase movement t o the pre-V positio n i n an interrogative sentenc e and its matching response, Hungaria n has t o have a n abstract and general pre-V A'-position int o which both interrogative Wh-phrases an d foci are moved. There ha s bee n a long-standin g and informal observation o n similaritie s be tween Wh-question s an d Focu s construction s in th e linguisti c literature . Th e parallelism i s base d o n syntactic , distributiona l similarities betwee n interroga tive Wh-phrases and Focus constituents in various languages such as Hungarian
362
Discourse Configurational Languages
and Basque , a s wel l a s o n th e intuitio n tha t i n discourse , th e Wh-phras e o f a question is replaced b y a Focus constituen t in the response, while the presuppo sition of the question an d of its natural response remai n constant. To make thes e observations explicit , Horvat h (1986:118 ) has propose d th e followin g principl e holding i n UG : (44) Th e FOCUS Constrain t on the Wh-Q Operato r A non-echo questio n interpretatio n can b e derive d onl y i f th e Wh- Q operator bears th e featur e FOCUS a t LF. Given that Focus assignment in languages like Hungarian is limited to apply only to th e pre- V node , a s Horvath' s Verb-Inheren t Hypothesis proposes , i t follow s that a well-forme d non-ech o Wh-questio n interpretatio n i s obtaine d onl y i f th e Wh-phrase occupie s th e pre-V positio n a t the level o f S-structure . Notice, however, tha t the Focus constrain t on the Wh-Q operator (44) i s inapplicable t o Korean , whic h ha s in-sit u Wh-Q phrase s lik e Chinese . Korea n fur ther constitute s a counterexampl e t o a cross-linguisti c predictio n tha t Horvat h makes regarding the distribution of interrogative Wh-phrases in the surface structure (Horvath 1985:122) : (45) Th e syntactic position(s) in which non-echo interrogativ e Wh- phrase s can appear in a language L will be identical t o or be a proper subset of the positions i n whic h FOCUS-constituents can appea r i n languag e L . Korean furnishe s a n exampl e wher e th e distributio n o f Wh- Q phrase s doe s no t coincide wit h Focused constituents , though th e distributio n o f the latte r i s syn tactically limited to the pre-V position as in Hungarian. It constitutes the reversed counterpart o f English whic h has a n in-sit u Focu s constituent but a structurally restricted distributio n of Wh-Q phrases. I n these tw o languages, the identit y between Wh- Q an d Focu s operator s i s no t syntacticall y expressed i n S-structure , though the y will occup y th e sam e operator positio n i n LF. Finally, Focu s assignmen t in Korean , which i s restricte d t o substitution , ex cludes base-generate d pre-V constituent s from gettin g Focused . I n contrast, th e Hungarian counterpart is not subject to such a restriction. According t o Horvath , Focus assignment to them is optional, while the constituents move d into the pre V positio n ar e obligatoril y Focuse d (cf . not e 22) . Sh e account s fo r th e latte r phenomenon i n term s o f th e Bindin g theory concernin g variable s (i.e. , bindin g condition C ; an R-expression i s free) an d a constraint on well-formedness a t L F (i.e., a variable must be A'-bound by an appropriate operato r i n LF) proposed i n the Government-Bindin g theor y (Chomsk y 1981) . He r accoun t ca n captur e obligatoriness o f Focu s Assignmen t i n sentence s (41b) , (42) , (43b ) an d (43c ) including the Korea n Focus construction in (19) , where a variable is lef t b y Fo cus/Wh-Q movement. Nevertheless, this accoun t is independen t of th e question of why the base-generated pre-V constituent in Hungarian (i.e., in a construction
The Theory of Syntactic Focalization 36
3
without a variabl e lef t b y Focu s movement ) ca n b e optionall y Focused , i n constrast to the Korea n counterpart. The optiona l Focu s assignmen t fo r the m i s also related t o the fact that a n interrogative Wh-phras e an d a Focused constitu ent occup y a n identical syntacti c position i n Hungaria n (i.e. , Focu s assignmen t applies t o bot h constituents) . A s the base-generate d pre- V constituent s ar e op tionally questione d (no t b y obligator y Wh-movement , bu t b y a n in-sit u Wh phrase), they also can be Focused. Otherwise, it is impossible t o provide a natural response to a question i n which one of the base-generated pre- V constituent s i s questioned. Now, i t seem s tha t substitutio n in th e formatio n o f Korea n Focu s construc tions represent s a n archetyp e of the syntacti c Focus mechanism , wherea s Hun garian and Basque represent an extended version of syntactic Focusing, whic h is not limited to substitution due to the obligatory Wh-movement to the pre-V po sition. Wha t i s commo n t o al l thes e language s is tha t the pre- V Focu s positio n coincides structurall y with the position o f a subcategorized constituent whic h i s not directl y theta-marke d by a verb .
3.3. Downgrading Movement and Focus Raising in LF Allowing downward movement to the pre-V position and Focus raising (and WhQ raisin g in Hungaria n and Basque) for bindin g in L F i s anothe r seriou s prob lem fo r Th e Verb-Inherent Hypothesis pointed ou t by th e competin g theory (E . Kiss 1987a , Orti z de Urbin a 1989) . The pre- V positio n governe d by an d adja cent t o V i s a n innermos t position o f V P in a sentence , an d movemen t i n thi s position i s inevitabl y downward, as show n in (19 ) an d (41) . Sinc e th e variabl e left b y th e movemen t canno t b e boun d a t S-structure , Focus/Wh- Q operator s should eventuall y undergo LF raising in order to be in a higher position t o bind variables. Though the necessity of LF-raising has been accepted i n literature, thi s criticism suggest s that postulating downgradin g movement is undesirable, sinc e it requires raisin g the operator s i n LF rather than in S-structure . This section attempts to discuss the question of downgrading Focus movemen t and Focu s raisin g i n L F base d o n ou r characterizatio n o f th e pre- V positio n i n section 3.1.1. The following tw o points are crucial for our discussion i n this sec tion: (1 ) th e pre- V nod e i s a n A'-position; (2 ) i t i s th e positio n wher e a base generated predicat e complement or a part of a complex predicat e occurs . Downgrading movemen t leaves a variable which is no t properly boun d i n S structure. Horvath argues that th e binding theory i n th e GB-framewor k actuall y allows suc h a syntacti c movement; th e binding conditio n concernin g variable s simply require s tha t 'a n R-expressio n i s free. ' Sinc e w e hav e argue d a t lengt h that th e pre-V nod e i s a n A'-position, no w th e question is ; wh y must the Focu s operator tha t ha s alread y undergon e syntacti c A'-movemen t b e raise d i n LF ? Although raising of Focu s (an d of a n interrogative Wh-phrase i n Hungarian ) i n LF is necessary t o meet th e vyell-formednes s conditio n whic h prevent s a n ope n sentence containing an unbound variable, this does not provide an authentic explanation fo r th e obligator y movement in LF.
364
Discourse Configurational Languages
Under our analysis, the obligatory L F movement i s not motivated b y binding of th e variable, bu t by reconstruction o f a predicate o f an S. The pre-V positio n is an A'-position licensed b y verbs whic h select a nonargument, an d thus i t pro vides a landing site fo r a Focus or an interrogative Wh-phrase whic h i s an argu ment i n Korea n an d Hungarian . However , i n contras t t o a n A'-position outsid e an S , the pre-verbal A'-position canno t be maintaine d i n L F due to th e requirement tha t th e base-generate d predicat e constituent s i n th e VP-interna l A'-position shoul d b e reconstructed .
4. Conclusion We hav e argue d tha t ther e i s a correlatio n betwee n a verb' s abilit y t o selec t a predicate complemen t (i.e. , subcategorization ) an d syntacti c Focu s assignmen t at length. In our analysis, the feature [+focus ] appear s t o be much mor e simila r to Case than Horvath has previously observed (cf . the statement i n (22)). A consequence o f ou r theor y i s tha t Focu s assignmen t i s predicte d t o b e possibl e b y an infinitiv e verb a s accusativ e Case assignment . Indeed , formatio n of a Focus construction i n a n infinitiv e context i s possibl e i n Hungarian , presumabl y b y movement t o the abstract pre- V nod e a s in (46 ) (E . Kiss , 1987a , 231-32): 34 (46) a . [ s,, Jano s [ s .[ s ' szeretne[s., [ s. a 'lenyegro l [ beszelni vegre]]]]] ] John woul d lik e th e essence-about to-spea k 'John woul d like t o spea k ABOUT THE ESSENCE a t last.' b. [ s,,[s,[s Ne m tudtun k [ s,,[s,mirol [ s beszelni ]]]]]] not know-w e about-wha t to-spea k 'We don't kno w wha t to spea k about. ' Nevertheless, th e parallelism of a finite an d an infinite verb in assigning Focus 35 is not always maintained. In some contexts, a Focused constituen t must be moved into the node preceding a finite verb , rather than that preceding a n infinite verb: (47) a . akaro m olvasn i a jelent6seke t I want-Pres-Def . t o read th e reports-Ac e 'I want to read th e reports.' b. A JELENTESEKET; akarom olvasni t ; 'It is THE REPORTS tha t I want to read.' c. *akaro m A JELENTESEKET; olvasni t ; The example s abov e indicat e tha t ther e i s a significan t interaction betwee n th e feature [+focus ] and tense. I n this sense, a verb's abilit y to select a nonargument (i.e., Focus-marking ) differ s fro m it s abilit y t o selec t a n argumen t (i.e. , Case marking). Our theory of syntactic Focusing based on subcategorization, however, has not provided a n account of why onl y th e categor y V can selec t a predicate complement an d thu s ca n assig n Focus—mor e generally , wh y onl y a lexica l categor y
The Theory of Syntactic Focalization 36
5
which ca n directl y combin e wit h tens e suffixe s (e.g . both verb s an d adjectives in Korea n (cf. note 12) ) exhibits suc h an ability. We will leave this question for further research .
Notes *I wish t o expres s m y gratitud e to th e Korea n Researc h Foundatio n fo r givin g me a n opportunity t o work i n Budapest (1990-2), and the Research Fun d fo r Junior Scholars . I am grateful to Katalin E. Kiss for encouraging m e to present a n early versio n o f this pa per in the summer school workshop held i n Dubrovnik and inviting me to contribute it to this volume. I also appreciate a valuable comment from Jo e Emonds. I would like to thank to Pd l Nyiri , Yong-Gu k Ki m an d a n anonymou s Hungaria n compute r technicia n wh o helped m e ou t whe n I ha d bafflin g problem s wit h my ne w compute r syste m during my settling-down perio d i n Budapest . I a m also gratefu l t o Gret e Dalm i fo r discussion s o n Hungarian an d her friendship . O f course al l errors an d shortcomings o f this paper ar e on my shoulders . 1. In her recent paper (in this volume), E. Kiss assumes that the Focus i n Hungarian is in Spec VP (i.e., Spec of the VP which includes the base-generated subject ) and that Topic is immediatel y dominate d by S . This proposa l i s anothe r variant o f th e Multi-leve l Hy pothesis. 2. Kim Alan H.-O (1985 ) propose s a functional approach t o pre-verbal Focusing (i.e. , Immediate Preverba l Focu s i n hi s term ) an d ver b Focusin g (i.e. , Focusin g o f th e finit e verb itself); h e attributes thes e t o three majo r principles o n the discourse level , an information flo w principl e (i.e. , new information occurs immediatel y o n the right o f the sen tence), a verb-fina l constraint in th e wor d orde r i n Korean , an d a preemption conditio n of the sentence domai n (i.e., pragmatically salient material may not follow the finite verb). Although pre-verbal Focusing i s preponderant in verb-final languages such as Korean and Turkish, hi s approac h cannot accoun t for Hungarian, whose wor d order i s not subjec t t o the verb-fina l constrain t (th e V-initia l orde r i n E. Kiss' s analysi s an d the SV O orde r i n Horvath's analysis) . 3.The V-firs t orde r o f Hungarian proposed b y E . Kis s als o reflect s th e fixe d positio n of th e verb . Th e V-fina l orde r o f Korea n ca n b e violate d onl y i n a roo t sentenc e fo r a stylistic purpos e (i.e., a literary style) (cf. Jo 1985 : Ch.l) . However, it is not clear whether such orde r i s a resul t o f V-movement ; i t ca n b e postposin g o f othe r constituent s fo r Focusing. 4. The objec t nou n i n a n idiom lose s it s idiomati c sense , i f i t i s expande d (e.g . modi fied b y a prenominal adjective) , o r if it is separate d fro m a verb by scrambling . I t has n o thematic rol e a s in English idioms . W e will sho w tha t such a n objec t nou n behaves lik e a predicate NP in the pre-V nod e in Topicalization an d Focus formatio n in sectio n 3.1.2 . 5. Choe H.-S. claim s that the existence of syntactic passives i n Korean also support s a configurational VP . However, i t i s no t clea r whethe r they ar e derive d b y NP-movemen t as i n English . NP-movemen t i n a passive no t onl y presuppose s a syntacti c V P but als o Case assignmen t by adjacency . I n Korean , th e subjec t positio n i s b y n o mean s adjacen t to the Case-assigne r INFL , an d :hu s NP-movement in a passive woul d be unmotivated . 6. Saito and Hoji (1983) argu e that scrambling in Japanese does affec t bindin g as shown in (1 ) an d thu s that it i s a n instanc e of 'mov e alpha. '
366
Discourse Configurational Languages (i) a . *karei-g a mada [Mary-ga Johnpni okutt a tegami]- o yond e ina i (koto ) he-Nom yet Nor n -t o sen t letter-Ac e rea d have-no t fac t '*He; has not read th e letter Mar y sent t o John/ b. [Mary-g a Johnj-ni okkut a tegami]-o karej-g a mad a yonde ina i (koto )
Choe H.-S . (1985 ) shows , however, tha t the Korea n sentenc e (2 ) corresponding t o (Ib ) is ambiguous; the coreferential reading is possible wit h stres s on the pronoun ku 'he ' bu t not wit h a neutral intonation : (ii) */O.K . [Mary-ka John;-eke y ponae-0- n phyenci]-lu l kUj-k a ilk-c i Nom -t o send-Pres-Mo d letter-Ac e he-No m read-t o ani-ha-ess-ta not-do-Past-Dec Therefore, sh e concludes that scrambling in Korean is a nonsyntactic phenomenon which may no t chang e binding. 7. Korea n doe s requir e a n independent INFL node . Even thoug h INFL element s usu ally amalgamate in the verbal morphology, V can be separated from the m in various con structions, as shown by the VP Topicalization in sentence (1) . 8. In Japanese both Topicalization and scramblin g have been argue d t o be adjunction; the former , adjunction t o S ' and th e latter , adjunctio n t o S (Saito 1985 , Hoj i 1985) . 9. Th e complementizer-nominalize r fo r a n embedde d claus e -ki, which i s i n comple mentary distributio n with a sentenc e ending , therefore , ca n als o licens e the Topi c posi tion, a s in the following sentence wher e the Topic constructio n (12a ) occurs withou t the declarative sentenc e marke r -to : (i) ( na-nun) [ NP [CP hoecang-un [ c, Suil-i t toe-ess-ki]]]-lu l I-Top chairman-To p No m become-Past-Comp-Ac c pala-n-ta wish-Pres-Dec 'I wish tha t a s for chairman, Suil ha s become one. ' 10. For Choe H.-S. (i n this volume), there are two possible Focu s position s i n Korean: the pre-V position an d the sentence-initial position whic h is in the right side o f the Topic of th e clause. He r proposa l onl y deal s wit h Focus appearin g i n the latter . 11. Emonds (1985 : Ch. 1 ) captures th e asymmetr y existing between N an d V in term s of direc t an d indirect theta-rol e assignment . In his theory , onl y V and P can directl y as sign a theta-role t o their complements; indirect theta-role assignmen t involves theta-rol e assignment usin g P , and thi s mechanis m permit s noun s an d adjective s hav e a comple ment system . Direct theta-rol e assignmen t mark s a theta-rol e t o onl y on e sister . Whe n V i s subcategorized for tw o sisters , a s in the sentence s o f (25), on e i s th e externa l argumen t of th e othe r (i.e., the subjec t is a n external argument). That is , V assigns a theta-role t o one o f the two sisters , an d the theta-role o f the secon d i s assigne d b y th e first, a s illus trated belo w (Emonds 1985:36) :
The Theory of Syntactic Focalization 36 11) V assigns theta-rol e directly t o Y
7 K
(2) Y assigns theta-role t o th e externa l argumen t NP. (3) Y
K
= PP,NP,AP,VP
We d o no t agre e wit h Emonds in tha t a predicat e complemen t i s directl y theta-marke d by th e verb . Th e presenc e o f a predicate complement supplement s th e predicatio n o f a main verb , whic h i s expresse d b y (Hi ) above . Thi s hypothesi s i s strange , becaus e th e argument theta-marked by the verb (i.e., the predicate complement Y K) in turn theta-marks another argument. 12. In a language like Korean, however, adjectives can also appea r in the main predicate positio n o f a sentence , an d the y ma y selec t a predicat e complemen t i n th e pre- A position similar to the intransitve verb toe 'become' in sentence (8a) ; the pre-A constituent i s usuall y a part of a complex adjective . Consequently, such a n adjective has abilit y to assig n Focus t o a consituent moved to the pre- A position. A sentence wit h an adjective occurrin g wit h a pre- A consituen t and it s correspondin g Focu s sentenc e ar e give n below (parenthese s mark position for the adverbia l phrase): (i) a . Sunae-k a (elilttae-puthe ) meli-k a ( * ) coh-0-t a Nom childhood-fro m head-Par t good-Pres-De 'Sunae has been smar t from he r childhood.' b. meli-nu n [ CP t; (__) SUNAE r ka (*_) coh-0-ta ] Top No m good-Pres-De c 'It is SUNAE who is smart.'
c
Therefore, ho w to incorporate th e V's (an d A's) ability t o select a predicate complemen t into th e theory o f verbal inflection appears to be a major issue to pursue . 13. Comrie (1984:7-8) observes that the preverbal position of the copula (i.e., the verb "be" i n th e presen t an d imperfect indicative) in Armenian must alway s be fille d wit h a predicate adjectiv e or a lexical verb (i.e., a part of a complex predicate), a s in th e sen tences i n (1)) , unless a Focused N P moves int o here. The sentence-fina l positio n o f th e verb "be" i n a nonfocused sentence like those in (1) must be reversed in an interrogativ e question (2a) (presumably by the postposing of the verb in the pre-copula position), wher e a Wh-word occupie s th e pre-V position: (i) a . Yerkink'—kapuy t e sky-DEF blu e i s 'The sky is blue.' b. Kapuy t e 'It i s blue.' c. * E kapuyt (ii) a . Petros- n INC ' e utum Peter-DEF wha t be-PRES-3S 'What is Peter eating?' b. *Petros- n INC' utu m e?
? G eat-PRES P
368
Discourse Configurational Languages
14. According t o Underbill (1976), Turkish has no need o f a linking verb 't o be' in the present tense , bu t i t i s needed i n othe r tense , fo r example , th e futur e tense ; thi s ver b i s olmak. At th e same tim e olmak als o mean s 'become * an d in this meanin g it may b e used an y tense. Th e use s o f th e ver b i n th e followin g sho w tha t th e pre- V positio n i n Turkish i s also occupie d b y a predicate constituen t (Underbil l 1976:149) : (i) a . Arkadasi m mekteb e gitti , dokto r old u friend-my school-t o go-pas t docto r become-Pas t 'My frien d wen t to school an d becam e a doctor.' b. Gocugu n eller i siya h oluyor child-poss han d blac k become-Pre s 'The child' s hand s are becoming black." Underbill (1976:72 ) describe s th e pre-verbal Focu s i n Turkish a s following: (ii) "Th e positio n of emphasis in Turkish is the position immediately before the verb; thus plac e i n tha t position an y wor d that you wis h t o emphasize . I n conversa tion, emphasis is also indicated by increased stress on the emphasized word (as in English). " An interrogativ e Wh-word or a phrase containing it i s considere d t o b e "emphasized, " and i t mus t be place d i n th e pre- V position in Turkish : (iii) a . B u film- i ki m gor-dii ? this film-Ac e wh o see-Pas t 'Who sa w this movie? ' b. Bun u hang i kati p yazdi ? this whic h cler k write-Pas t 'Which clerk wrote this? ' 15. I n Horvath' s version , th e requiremen t o f th e Projectio n Principl e i s accepted ; subcategorization entail s theta-marking . Th e pre- V constituent s in Hungaria n ar e subcategorized b y verbs , an d therefore , th e pre- V positio n i s considere d a s a n A-posi tion. Making use of new notions of th e Projection Principle an d the Theta Criterion , sh e (Horvath 1985 : Ch . 3) argues that a base-generated pre-V constituent in Hungarian which is locall y postpose d b y substitutio n observes th e tw o principle s (cf . not e 6) . However , the Focuse d constituent , which is move d fro m a n argumen t position t o the pre- V Focu s position, inevitably violates th e Theta Criterion (29) . Horvath's analysi s evidently shows that any version o f the Verb-Inherent Hypothesis canno t satisfy bot h the Projection Prin ciple and the Theta Criterion . Whil e her analysis meets th e requirement of the former, it violates th e latter . In contrast, ou r analysi s observes th e latter, but violate s th e former . 16. The possible violatio n of the Projection Principl e i n Focus assignment i s twofold: (1) one is a violation by the dislocated pre- V constituent ; (2) the other i s a violation b y a Focused constituen t moved to pre-V position, which is motivated by Focus assignment . As fo r th e former , neithe r does i t reall y violat e th e Projectio n Principl e no r th e Thet a Criterion. The dislocated predicate NP in the Korean Focus construction (19) violates the Projection Principl e because its trace in the Pre-V position is erased by the Focused con -
The Theory of Syntactic Focalization 36
9
stituent move d int o it . Althoug h G B theor y doe s no t predic t th e existenc e o f a subcategorized nonargument , it doe s no t preven t th e possibilit y tha t suc h a constituen t may leav e a trace i n an English sentenc e lik e what did John become t. The majo r moti vation fo r trace s i n G B theor y lie s i n th e transmissio n o f a theta-rol e (an d Case ) o f a subcategorized argumen t which h:is undergone movement, in order t o maintain the struc ture of arguments at every grammatical level. Horvath (1986: Ch. 3) argues that the ThetaCriterion, which is a general conditio n o n the well-formedness o f LF representation, ca n render th e stipulatio n o f th e Trace Erasur e Prohibitio n (Freidi n 1978:524 ) unnecessary . Since the trace o f the dislocated predicat e N P in the Focus constructio n (19 ) ha s no the matic role, it s erasur e causes n o violatio n of th e Theta Criterio n (J o 1986 : Ch . 4) . 17. Therefore, th e existenc e o f idiom s consistin g o f a bar e objec t N an d a ver b ap pears t o support for the constituen t V'(i.e., ' a smal l VP") , rather tha n VP in Korean (cf . Choe H.-S . 1988) . 18. We do not commit ourselves with respect to the different basi c word orders o f Hungarian propose d b y the tw o authors . Ou r argument s for th e pre-V nod e a s a VP-internal A'-position coul d b e compatibl e wit h th e bot h orders , a s fa r a s i t i s postulate d a s th e position governe d b y an d adjacen t to V. 19. Horvath (1985:54 ) analyze s th e preverba l constituent s i n th e sentence s belo w as PP's : (1) a . Mar i a z asztalr a tett e a z eddnyeke t the table-o n to pu t th e dishes-ac e 'Mary pu t th e dishes on the table. ' b. A boszorkdn y eg y csu f be'ka'va ' valtoztatt a a kirSlyfi t the witc h a n ugl y frog-translat . cas e change d th e prince-ace . 'The witch changed the prince int o a n ugly frog. ' Whether suffixe s -ra '-onto ' and -va (the translative Case ) ar e P or Case markers, som e native speakers—includin g E . Kis s (persona l communication)—d o no t agre e wit h he r judgement tha t the constituents i n the pre-V position o f these sentence s ar e non-Focused ; the NP wit h a determine r and the one wit h a modifie r are interprete d as Focuse d con stituents (i.e. , they ar e referential NP's wit h a theta-role). Onl y a determinerless N o r a bare N ca n be interprete d a s a non-Focused pre-V constituen t (i.e., a nonargument) , a s moziba (movie house-into) in E. Kiss's example (36d). For example, if a bare noun bek&vA (frog-translat. case ) occurs in the pre-V position of (Ib) it would be qualified a s a basegenerated pre- V constituen t (i.e., a par t o f a comple x predicate) . Therefore , neithe r d o these sentence s suppor t Horvath's characterization of th e pre-V nod e a s an A-position . Moreover, even i f th e pre- V constituents in the abov e sentence s ar e PP's, a s Horvath claims, th e pre- V positio n a s ar i A-positio n canno t stand . The N P complement s i n th e pre-V position of the above sentences apparently have a theta-role, namely Goal. Although some verbs strictl y subcategorizi; a PP it is not the verbs whic h directly assig n Case and a theta-rol e t o th e N P complement ; i t i s P whic h Case-mark s an d theta-mark s it s N P complement. Th e ver b doe s no t directl y involv e theta-rol e assignmen t o f the PP ; it s in volvement t o theta-markin g in a subcategorized P P i s indirect lik e the theta-markin g of the subject NP . Likewise, th e pre;-verbal NP in the sentences o f (1) are also Case-marke d and theta-marked , not by the verbs, but by the postpositional suffixes , -ra 'onto ' and -ba 'to.' Therefore, i t appear s tha t th e abov e sentences containing the pre-verba l P P d o no t pose a seriou s problem for th e generalizatio n o f th e pre-V Focus position as a n A'-posi tion license d b y V.
370
Discourse Configurational Languages
20. In addition to the examples give n i n (36), we find tha t the behavior o f the follow ing italicized pre- V predicate constituent s in the Hungarian is similar to the Korean coun terparts i n (8) ; the y mus t b e adjacen t to th e verbs , an d thei r positio n i s fixed , unles s a Focused constituen t occupie s th e pre-V position : (i) a . Ja"no s uriembernek ( * mindig) Idtszi k John gentleman-t o alway s seem s 'John (always ) seems t o be a gentleman.' b. Jdno s fdradtnak (*ma ) IStszi k John tired-t o toda y seem s 'John seem s t o be tire d (today). ' c. Edi t butdnak ( * mindig) tekin t enge m foolish-to alway s consider s m e 'Edit (always ) considers m e foolish. ' d. JSnos t beteggg (* sajnos) tett e so k munkaj a John-Ace sick-trans.Cas e unfortunatel y pu t muc h work-hi s '(Unfortunately,) to o much wor k mad e John sick. ' 21. In E. Kiss's analysis, the word order in post-verbal positions is not fixed (cf. phras e structure (34)). Therefore, ther e is no guarantee for a subcategorized constituent by a verb is adjacen t t o a ver b withi n a n S (i.e. , befor e it s movemen t t o th e pre-verba l Focu s position). 22. Horvat h (1985 : Ch . 3 ) account s for thi s fact a s follows: Since sh e postulate d th e base position of the object NP as post-verbal, the preposed objec t NP leaves a trace (i.e. , a variable) in its original position, which should be A'-bound in LF. If Focus Assignment fails t o apply to the object NP in the pre-V position , n o operator i s available t o bind th e variable lef t b y downgradin g movemen t (i.e. , movemen t t o th e positio n governe d b y and adjacen t t o th e verb ) i n th e L F representatio n o f th e sentenc e containin g suc h a n object NP. 23. A nonargument bar e N suc h a s tha t in (36c ) appearin g i n a post-verba l positio n can be independently accounted fo r i n terms o f aspect marking , which will be discusse d later i n thi s section . 24. Horvath analyzes these verba l particles as intransitive PP's , [p p [p ]]. Sinc e th e P without an NP complement i s not theta-marked by V, this analysis of the verba l particle s is compatible with the chracterization of the pre-V node as a subcategorized nonargumen t that w e have pu t forwar d here. 25. Horvath (1985:77), fo r example, points out that the only kinds of sentences i n Hungarian tha t are unambiguous with respect t o aspect ar e clauses wit h a non-Focused pre V constituent . However, E . Kis s (i n personal communication ) doe s no t agre e wit h this; progressive an d existentia l reading s requir e a n empt y Focu s slot , whethe r o r no t a n S contains a base-generated pre- V constituent. 26. The three interpretive templates in (38) invok e disagreements. The predictions tha t they mak e are no t identica l t o E. Kiss' s analysi s of Hungarian aspect (cf . not e 1 1 i n E . Kiss (1987a:74)) . For example, (38b ) doe s no t predict th e so-calle d existentia l interpre tation presente d i n E . Kiss' s exampl e (37c) . Furthermore , E . Kis s provide s counterexamples for each of them (in personal communication). The differences between the tw o analyse s on Hungaria n aspect , however, are no t ou r majo r concern . What is im portant fo r ou r analysi s is th e genera l agreement about th e rol e o f th e pre- V nod e i n expressing aspec t in Hungaria n (e.g . th e prefi x clearl y makes a sentenc e perfective).
The Theory of Syntactic Focalization 37
1
27. Horvat h (1985:58 ) notice s tha t th e pre-V constituent s exhibi t a significan t property indicating that they appear in a nonargument position; the y do not permit fre e right branching recursion like English adjective phrases i n prenominal position (i.e. , they cannot selec t posthea d complements;) , a s discusse d i n Emond s (1976) . Fo r instance , a P P selected b y a ver b lik e test 't o put ' i s a pre- V phrase , a s i n (la) ; a n N P i n th e pre- V phrase, however , canno t contai n a relativ e clause , a s th e ungrammatica l sentenc e (Ib ) shows: (i) a . Mar i a z asztalr a tett e a z ed£nyeke t Mary th e table-ont o pu t th e dishes-ace . 'Mary pu t the dishe: ; on the table.' b. * Mari a z asztalra , am i a z erkelye n dll t tett e a z Mary th e table-ont o whic h th e balcony-o n stoo d pu t th e ed6nyeket dishes-ace. 'Mary pu t th e dishe s o n the tabl e tha t stoo d o n the balcony.' 28. VP-deletio n i n Englis h can involv e each ver b phras e superscripte d i n the phrase marker (40) , an d th e rul e o f VP-deletion i s formalize d a s follow s (Akmajian , Steel an d Wasow 1979) : (i) V n Deletion Delete Vn , n 1 , optional 29. E. Kiss (1987a:45, 49 ) does not take up a definite positio n concerning the question of th e positio n an d statu s of INF L in Hungarian ; she say s tha t it i s no t pertinen t t o th e issues o f Hungarian syntax she deals with and that INFL in Hungarian is always realized on a main verb. However , a langaage in which INFL is realized on verbs such as Korean has syntactic motivations to postulate an independent INFL node (cf. Choe H.-S. (1988)) . 30. Horvath' s clai m tha t th e pre- V nod e i s a n A-position, however , i s no t onl y intu itively incompatibl e wit h its function i n aspect marking, it formally contradicts the Head Peripherality Conditio n (26) sh e has proposed a s a restriction o n the base rul e schema in UG. The restrictio n claim s tha t an y hea d o f phras e occupies a peripheral position a t it s own leve l o f projection . Whic h particula r sid e o f th e hea d i s non-nul l wil l constitut e a parameter o f UG , suc h a s 'V-in;.tial ' versu s 'V-final ' languages . Although the structur e of VP in Hungarian seems t o violate the Head Peripherality Condition in (26 ) b y allow ing certai n constituent s in th e pre- V position, a s th e phras e marker i n (27 ) shows , sh e argues that not onl y V but also V a s a lexical categor y may theta-mar k it s right sisters . Therefore, sh e says that Hungarian is basically a 'V-initial' language like English (i.e., it has right branching phrases). Sh e furthermore points out that an NP (with a postposition) appearing in pre-V position i s prohibited from fre e right-branching (cf. note 26). In othe r words, her argumen t can hold, onl y if the pre-V node is a nonargument position. Otherwise, Hungaria n would be a counterexample to th e Head Pheripherality Conditio n (26) , which allow s argument s in both side s o f a verb. 31. The process o f postposing the base-generated pre- V constituents to express the progressive aspec t is independentl y justified, as show n in the Interpretive Template fo r Aspect (38 ) tha t sh e has proposed. 32. In orde r to make the sentenc e in (41) noncontroversial, we have changed the nonFocused preverba l NP int o a bare N asztalra 'table-onto ' (cf . note 18) .
372
Discourse Configuration^ Languages
33. According to Comrie (1984:3) , Armenia n exhibits the preverbal Focus , and an interrogative pronoun in a question must be i n thi s position lik e Hungarian . However, h e reports that in the answer to a question, the constituent giving the requested informatio n (i.e., Focus) may not strictly observe th e preverbal rule ; the pre-V i s a possible position , but not a necessary position i n this language. Both (Ib) and (Ic) are to be understood a s answers t o the question o f (la) : (i) a . Petros- n IN C kerav ? ( = the marker fo r stress ) Peter-DEF wha t eat-AOR-3S G "What di d Peter eat? " b. Petros- s M I X9NJO R kera v DBF a n appl e eat-AOR-3S G "Peter at e an apple. " c. Petros- 3 kerav MI X9NJOR "Peter at e a n apple. " Therefore, th e preverba l rul e o f Focu s i n Armenian i s les s stric t tha n a languag e lik e Hungarian, a t leas t i n a n answe r t o a n interrogativ e question. 34. In Korean, in order to obtain syntactic Focus, a pre-V constituent is required t o be dislocated int o Topic position , as shown in (15a) an d (19); sinc e thi s position i s license d only by a finite claus e in Korean (cf. note 9) the derivation of a Focus constructio n i n an infinitive i s blocked, a s below : (i) a . Sunae-k a Suil- ul [ vp hoecang- i toe]-ke y ha-ess-t a Nom Ac e chairman-No m become-in f do-past-de c b. *Sunae-k a t j [ vp hoecang-u n SUIL ;-ul toe]-k e ha-ess-t a Nom chairman-To p Ac e become-in f do-past-de c Focus movement in Korean, nonetheless, is valid in an embedded nominalized finite clause as i n (2a ) and in a finite clausa l complement a s in (2b) , whic h both may hav e the Topi c node: (ii) a . Sunae-k a [ NP [IP hoecang-un t ; SUIL ri toe-ess]-um]-u l Nom chairman-To p No m become-past-ing-Ac c alli-ess-ta inform-past-dec 'Sunae informed that it i s SUIL wh o became chairman. ' b. Sunae-k a [ cp hoecang-un t ; SUILj-i toe-ss-ta-ko ] Nom chairman-To p No m become-past-dec-comp alli-ess-ta inform-past-dec 'Sunae informe d that it i s JOHN who became chairman. ' The impossibilit y o f derivin g a Focu s constructio n fro m a Korea n infinitiv e a s i n th e sentence (Ib) , therefore , doe s no t undermine the Verb-Inherent Hypothesis. 35. Furthermore , a Focused elemen t / a Wh-Q phrase in a n embedded claus e ca n ap pear i n th e pre- V Focu s positio n o f uppe r clauses ; lik e COMP-to-COM P movement , a Focus-to-Focus rul e (an instanc e of 'Mov e alpha' ) ma y appl y leaving a trace i n th e in -
The Theory of Syntactic Focalization 37
3
tervening pre-V Focus positio n (Horvat h 1986 , for Hungaria n and Azkarate et al . 1982 for Basque) . Of course, we cannot expect that the pre-V position of an infinitive verb is exactly lik e a full-fledge d A'-position fo r tha t of a finite verb into whic h an interrogative wh-phras e can b e successivel y move d (i.e . Focus-to-Focus movement ) i n Hungaria n (cf . Horvath 1981 an d E. Kis s 1987a) .
References Akmajian, K. , S . Steel , an d T. V/asow (1979 ) "Th e Category AUX i n Universa l Gram mar," Linguistic Inquiry 10 , 1-64 . Choe, H.-S. (1988) Restructuring Parameters and Complex Predicates, Doctora l disser tation, MIT, Cambridge, Massachusetts. Choe, H.-S . (1989) "Restructurin g i n Korea n an d Hungarian, " i n L . Marac z an d P . Muysken , eds., Configurationality, Foris , Dordrecht . Chomsky, N. (1957) Syntactic Structure, Th e Hague : Mouton. Chomsky, N. (1981 ) Lectures on Government and Binding, Foris , Dordrecht . Comrie, B . (1984 ) "Som e Forma l Propertie s o f Focu s i n Moder n Easter n Armenian, " Annual of Armenian Linguistics 5 , Clevelan d Stat e University , Cleveland, Ohio , 1-21. Emonds, J. (1976) A Transformational Approach to English Syntax, Academic Press, New York. Emonds, J.(1985 ) A Unified Theory of Syntactic Categories, Foris , Dordrecht. Fukui, N . (1986 ) A Theory of Category and Its Application, Doctora l dissertion , MIT, Cambridge, Massachusetts. Hale, K . (1982 ) "Preliminar y Remark s on Configurationality," NELS 12 , 86-96. Hale, K. (1983) "Warlpiri and ths Grammar of Non-configurational Languages, " Natural Language and Linguistic Theory 1, 5-47. Hoji, H . (1986) Logical Form Constraints and Configurational Structure, Doctora l dissertation, University of Washington, Seattle, Washington. Horvath, J. (1981 ) Aspects of Hungarian Syntax and the Theory of Grammar, Doctora l dissertation, UCLA , Los Angeles, California. Horvath, J. (1986) Focus in the Theory of Grammar and the Syntax of Hungarian, Foris , Dordrecht. Huang, C.-T.J . (1982 ) Logical Relations in Chinese and the Theory of Grammar, Doctoral dissertation , MIT , Cambridge, Massachusetts . Jo, M.-J . (1986) Fixed Word Order and the Theory of the Pre-Verbal Focus Position in Korean, Doctoral dissertation , Universit y of Washington, Seattle, Washington. Katada, F. (1991) "The LF Representation of Anaphors," Linguistic Inquiry 22, 287-313. Kim, A. H.-O. (1985) The Grammar of Focus in Korean Syntax and its Typological Implication, Doctoral dissertation , USC , Los Angeles, California . E. Kiss, K. (1981) "Structura l Relations in Hungarian, a "Free" Word Order Language, " Linguistic Inquiry 12 , 185-213 . E. Kiss , K . (1987a) Configurationality in Hungarian, Akademia i Kiado , Budapest . E. Kiss, K . (1987b) "I s VP Universal?" in I Kenesei, eds. , Approaches to Hungarian v.2, Szeged, Hungary. Ortiz d e Urbina, J. (1989 ) Parameters in the Grammar of Basque, Foris , Dordrecht . Saito, M . (1985 ) Subject-Object Asymmetries in Japanese and Their Theoretical Implications. Doctora l dissertation . MIT, Cambridge, Massachusetts.
374
Discourse Configurational Languages
Saito, M. , an d Hoji, H.(1983 ) "Wea k Crossove r an d Move Alpha in Japanese," Natural Language and Linguistic Theory 1 , 245-259. Stowell, T. (1981) Origins of Phrase Structure, Doctoral dissertation , MIT , Cambridge, Massachusetts. Underbill, R . (1976 ) Turkish Grammar, MI T Press, Cambridge , Massachusetts . Whitman, J. (1991 ) "Strin g Vacuous V to COMP, " ms.,the paper presented i n GLOW . Williams, E . (1980 ) "Predication, " Linguistic Inquiry 11 , 203-238. Willams, E . (1983 ) "Agains t Smal l Clauses," Linguistic Inquiry 14 , 287-308. Yoon, J.-Y. (1990 ) Korean Syntax and Generalized X-Bar Theory, Doctora l dissertation , University o f Texas, Austin, Texas.
12 Focus in Quechua PIETER MUYSKE N Universiteit van Amsterdam
1. Introduction This stud y is a firs t attemp t to analyz e what is know n about focu s in Quechu a within the perspective of recent views of clause structure as a layered se t of func tional projections. The primary data will be drawn from the varieties of Quechua spoken i n south-centra l Peru , suc h a s Ayacucho an d Cuzco , althoug h I refer t o studies o f other varietie s suc h a s Huanuco Quechu a (Webe r (1989) ) a s well. It i s necessar y t o relat e th e stud y of focu s i n Quechu a t o a numbe r of othe r aspects o f its grammar. First of all, the relation o f focus markers to various other particles involved in structuring the information flow i n the utterance. These ar e often referre d to as 'independen t suffixes ' and are discusse d in sectio n 2. Cru cial her e i s that focus markers double a s evidential particles . Second, the types of clauses; (in terms of tense features, syntactic independence, etc.) tha t focus can occur in, and their relation to tense, agreemen t markers , and complementizers. Thi s is the subjec t of the main analytical section o f this paper , 3. Particularly importan t her e i s the relatio n betwee n focu s an d what i s terme d 'Main Tense.' Third, th e relatio n betwee n focus, sententia l an d contrastiv e negation , an d question particles , tha t I dea l wit h in sectio n 4 . Finally , i n 5 a brie f summar y with som e points of further rssearch and som e unanswered questions .
2. Independent Suffixes Quechua suffixe s (whic h constitute the majo r pattern o f th e morpholog y o f th e language) are ofte n classifie d a s nominal, verbal, and independen t or class-free. Leaving the first tw o types for exemplification in the sentences cited below, here I will be concerne d with the las t group, which derives its label fro m bein g abl e 375
376
Discourse Configurational Languages
to attach to any constituent in the sentence. In addition t o describing th e genera l features of thi s group, I wil l briefly discus s thei r possibl e morphological statu s as clitics.
2.1. Preliminary Classification and Features The element s tha t wil l concer n u s primaril y here ar e thos e i n (1) , whic h mar k the evidentia l status of th e informatio n in th e utteranc e (i n additio n t o markin g focus, a s w e wil l se e in sectio n 2) : (1) evidential s -mi/-n direc t experienc e (glos s DE ) -si/-s hearsa y (glos s HS ) -cha/-ch dubitativ e (gloss Dub ) This clas s i s sometime s labele d validators , a ter m referrin g t o th e fac t tha t th e speaker indicates his willingness to validate what is being communicated. In this perspective -mi is labeled 'affirmative, ' sinc e the speaker can stand for information which is based on direct experience. I adopt Weber's (1989) argumentation, however, that the term evidential is more appropriate. The evidentials have a long and a shor t form , depending on whethe r the y follo w a consonan t o r a vowel , respectively. A second element o r pair of homophonous elements i s the question an d negation particle -chu. While i n northern Peruvian Quechua thi s se t is formall y differentiated, i n the varieties we are concerned wit h there is no immediate reaso n not to treat them as a single element, expressing something like 'questio n of truth suspended.' I t i s a comple x matter , t o whic h I tr y t o giv e onl y a preliminar y answer below in section 4, whether in fact i t is a single element. Fo r the sak e of clarity, I will gloss the m a s separate particles . A third class, listed in (2), can appropriately be labeled validators , expressing the commitmen t of the speake r t o the informatio n presented: (2) validator s -lla delimitative , 'just ' -puni definitive , 'certainly ' -ya emphatic , 'indeed ' In (3) I give a few examples of the elements presented s o far. In (3a) the validator -puni i s combined with the evidentia l -mi/-n, an d in (3b ) th e questio n particle chu i s combined wit h the evidentia l -cha/-ch. (3) a . wasi-puni- n house-Def-DE 'It definitel y i s a house.' b. nuqa-chu-ch ? I-Q-Dub 'Perhaps me? '
Focus in Quechua 37
7
A further, highly frequen t an d important set of independen t suffixe s doe s no t mark the factual status of the information presented, but rather its relation to other bits o f informatio n presented I n th e discourse . (4) discourse markers -qa topi c marke r -taq contras t -raq continative , 'still ' -na inceptive , 'already ' -pas 'an d also, ' indefinit e In (5 ) i t i s show n ho w thes e element s ca n b e used , ofte n i n combinatio n with evidentials an d validators : (5) a . tusu-ru-man-ku-pas-ch a dance- Asp-Pot-pl-Ind-Dub 'They coul d eve n danc e (emphati c conjecture). ' b. wallpa-q a wacha-n-raq-ch u chicken-To lay= egg-3-still-Ne g 'The chicken ha s not layed a n egg yet.' c. warmi-yki-na-ta q maqa-ru-su-nk i wife-2-already-Cont hit-Asp-3- 2 'And the n your wif e hi t you.' In (5a ) th e discours e marke r -pas i s combine d wit h the validato r -cha, i n (5b ) continuative -raq with the negation element -chu, and in (5c) inceptiv e -na with contrastive -taq, Readers familia r with studie s o n sententia l particle s i n othe r languag e fami lies will be aware that often th e precise interpretation of the elements I have been describing i s highly specific t o a discourse context. In (6 ) th e combinator y an d orderin g possibilities o f th e element s mentione d so far ar e schematically presented . I t is clear tha t the evidentials ar e alway s the final element s i n th e phonologica l word , an d tha t th e evidential s canno t b e combined wit h th e topi c marker . For th e rest , man y combinations ar e possible , though i t ca n be imagine d tha t mor e tha n thre e independen t suffixe s woul d b e exceptional. (6) Order (base d o n Parker, 1965; slightl y simplified)
378
Discourse Configurational Languages
Although other particles ca n contribute t o focus, the evidentials ar e most directl y involved i n markin g focus, an d I wil l limi t myself t o thi s se t in wha t follows .
2.2. Morphological Properties In earlie r work , e.g . Muyske n (1981) , I have tried t o argue tha t th e se t of inde pendent suffixes i s really clitic-like , no t generated throug h word formation rules , but throug h phrase structur e rule s an d onl y attache d t o thei r host constituen t i n phonology. Three type s o f argument s were given . 2.2.1. Headedness Derivational affixe s potentiall y determin e th e category , an d hence th e distribu tion, o f th e constituen t the y ar e attache d to : nomina l affixe s creat e (o r 'head' ; cf. Williams, 1981 ) nouns , verbal affixe s creat e verbs , etc. The independent suffixes d o no t creat e element s o f a particular category : a noun wit h a n evidentia l marker remain s a noun, etc . 2.2.2. Selectivity Derivational affixe s ar e selectiv e i n tw o ways : the y obe y th e Majo r Categor y Restriction, cf . (7) , an d they ten d to obe y th e Unitary Base Hypothesis , cf . (8) . Both principle s ar e develope d i n the wor k o f Aronoff (1976) . In (7a) it is shown that the evidential -mi/-n ca n attach to the negation marke r mana, and in (7c) that the accusative case marker -ta cannot, even though i n (7b ) it i s show n t o b e abl e t o attac h t o adverb s suc h a s paqarin 'tomorrow. ' Th e latter form , bu t no t mana, is a majo r category , i t coul d b e argued ; henc e th e contrast. (7) a . mana- n hamu-n-chu . not-DE come-3-Ne g 'He/she does no t come.' b. paqarin-ta- n hamu-nqa . tomorrow-Ac-DE come-3F u 'He will come tomorrow. ' c. * mana-ta-n hamu-n-chu . not-Ac-DE come-3-Ne g A similar se t of examples ca n be taken to show that affixes, bu t not evidentia l markers, ar e constraine d b y the Unitary Bas e Hypothesis . I n (8a-c) it i s show n that cas e marker s (a s wel l a s evidentials ) ca n attac h t o nouns , adjectives , an d nominalized verbs , th e clas s o f [+N ] elements . Evidentials , bu t no t cas e mark ers, ca n attac h t o verbs, however , whic h are [-N] . (8) a . wasi-t a a. house-Ac house-3-D
' wasi-n-m i
E
Focus in Quechua 37
9
b. allin-t a b. ' allin-m i good-Ac good-D E 'the goo d one , well ' 'i t is good' c. hamu-sqa-n-t a c. ' hamu-sqa-n-m i come-Nom-3-Ac come-Nom-3-D E 'that he came Ac ' 'tha t he came DE ' d. *hamu-n-t a d. ' hamu-n-m i come-3-Ac come-3-D E Both criteri a distinguis h betwee n case marker s an d evidential s an d other inde pendent suffixes . 2.2.3. Vowel Deletion A third argumen t derives from th e phonological propertie s o f the evidentials. As can be seen in (9a,b), the evidentials have a long and a short form , depending on whether the base word ends irk a vowel or a consonant. As shown in (9a',b'), case markers suc h a s accusativ e ~ta and locativ e -pi ar e invariant . This contras t ca n be plausibl y relate d t o stres s patterns : cas e marker s ar e attache d befor e stres s assignment, evidential s after . (9) Evidential s Cas e markers a. wasi- n / wasi-n-m i a. ' * wasi-t / wasi-n-t house-DE house-3-D E house-A c house-3-A 'the/his house indeed ' 'hi s house-Ac' b. wasi- s / wasi-n-s i b. ' * wasi-p / wasi-n-p house-HS house-3-H S house-L o house-3-L 'the/his house, the y say ' 'i n his house'
a c i
o
Clearly, thi s contrast in phonological behavior accord s wel l wit h a difference in morphological status : we do not expect clitic s to influence word stress . 2.3. Problems Although the differences between the evidential suffixes an d classbound suffixe s such a s case marker s ar e well-established, i t i s less clear how to interpret the m in contemporary theoretical frameworks , fo r two reasons. First , th e introduction of th e theor y o f functiona l heads , particularly i n conjunction with the theor y o f incorporation propose d b y Baker (1988), makes the original distinctio n betwee n syntactic elements suc h as particles and morphological element s suc h a s derivational affixe s meaningless . Particle s may well be syntacti c heads, a s well a s affixes, befor e incorporation o f the base wor d takes place. Second, th e developmen t o f th e theor y o f level-ordere d Lexica l Phonolog y (Kiparsky, 1982 ) allow s fo r th e possibilit y o f different affi x classe s with differ ent properties . Thu s a global characterizatio n of affixes , wit h the possibilit y o f excluding element s no t conformin g to thi s characterization, ha s disappeared . I t
380
Discourse Configurational Languages
may thus be possible t o classify potential clitics simply as affixes undergoin g only post-lexical phonologica l rules . Therefore I wil l no t distinguis h between th e independen t suffixes , includin g the evidentials, an d elements such as the case markers fro m th e point of view of the syntax, in what follows. Their differen t morphological behavio r follows fro m the lexica l stratu m in whic h the y ar e attache d an d fro m th e typ e o f constituent they select. 1
3. The Syntax of Evidentials and Their Position in the Clause After thes e preliminar y remarks introducin g the evidential s an d explorin g thei r morphological statu s I now list some of the syntactic properties of the evidentials , before attemptin g a specifi c structura l analysis.
3.1. Constituent-external Evidentials are always constituent-external (where constituent is NP, AP, PP, but not necessaril y V P etc., a t least o n the surface) : (10) a . [ chay wasi-ta-n ] riku-nki-ch u that house-Ac-D E see-2- Q 'Do you see that house?' b. *[chay-m i wasi-ta ] riku-nki-ch u that-DE house-A c see-2- Q
3.2. Position in the Clause Evidentials occu r either o n the focused or rhematized, a s in (lla), or on the firs t constituent, a s in (lib): (11) a . cha y runa-q a Ayakuchu-ta- n ri- n that man-T o Ayacucho-Ac-D E go- 3 'That ma n is going to Ayacucho.' b. huk-s i ka-sq a hu k machucha-piwa n payach a once-HS be-S D on e ol d man-wit h ol d woman 'Once there were a n old man an d an ol d woman. ' (lla) show s tha t evidentials ma y also b e foun d o n anothe r constituen t tha n th e first one in a clause. These cases constitute marked positions for evidentials, which trigger focu s interpretatio n for the constituent to which the evidential is attached. Consider th e following examples, wher e the evidential fall s on a constituent different fro m th e firs t one : (12) a . Pidr u wasi-ta- n ruwa-n . Pedro house-Ac-D E make- 3 'It i s a house tha t Pedro builds. '
Focus in Quechua 38
1
b. Pidr u kunan-m i wasi-t a ruwa-sha-n . Pedro now-D E house-A c make-Pr- 3 'It is now that Pedr o is building th e house.' When th e evidentia l appear s o n the firs t constituent , th e sentenc e i s ambiguou s between focu s on the firs t constituent , (a) , an d no contrastive focus , (b) : (13) Pidru- n wasi-t a ruwa-n . Pedro-DE house-A c make- 3 (a) 'I t is Pedro tha t build s a house.' (b) 'Pedr o builds a house.'
3.3. Restriction to One per Clause Examples suc h a s (14) sho w tha t there ca n onl y b e one evidential pe r clause : (14) a . Mariya- s t'anta-t a ruwa-rqa- n Maria-HS bread-A c make-Pa- 3 'Maria, the y sa y mad e the bread.' b. Mariy a t'anta-ta- n ruwa-rqan Maria bread-Ac-D E make-Pa- 3 'Maria mad e th e BREAD.' c. *Mariya- s t'anta-ta- n ruwa-rqan Maria-HS bread-Ac-D E make-Pa- 3 'Maria, they say , made th e BREAD.' While bot h Mariya i n (14a ) an d t'anta-ta i n (14b ) ca n b e marke d wit h a n evi dential independently , th e tw o instance s o f markin g canno t occu r i n th e sam e clause, as in (14c) .
3.4. Restricted to Main Clauses Evidentials occu r onl y in main clauses, i.e. onl y in the scope of Main Tense (a s opposed bot h t o infinitival tense and to the 'dependent ' tense w e fin d i n adver bial subordinate d clauses an d nominalizations). Thu s 'mai n tense ' (15 ) i s fine : (15) a . Xwan-m i hamu-nqa . Juan-DE come-3F u 'Juan wil l come.' b. Hamu-nqa-chu ? come-3Fu-Q 'Will he come?' In adverbia l (cf. 16 ) and nominalize d (cf . 17 ) clauses, evidential s ca n occu r o n the final constituent , the verb, in the (a) examples, but arguabl y there they are in the domain of the main clause tensed verb. However, the (b) examples, in which
382
Discourse Configurational Languages
the evidential woul d b e on the firs t constituent , interna l t o the adverbial clause , are ungrammatical. (16) a . Qaynuncha w wiqchuku-spa-m i nana-chi-ku-ni . yesterday slip-Sub-D E hurt-Cau-Re1 'Since I slippe d yesterda y I have pain. ' b. * Qaynunchaw-mi wiqchuku-sp a nana-chi-ku-ni . yesterday-DE slip-Su b hurt-Cau-Re1 (17) a . Xwan-p a hamu-sqa-n-ta- n yacha-n i Juan-Ge come-Nom-3-Ac-D E know- 1 'I know that Juan has come.' b. *Xwan-pa- n hamu-sqa-n-t a yacha-n i Juan-Ge-DE come-Nom-3-A c kno w 1 The situation in non-matrix main tense clauses is rather complex. There is a class of formally coordinated clauses i n which the first claus e o f the pair is interprete d as restricting the referential domain of the subject of the second clause . Her e th e evidential i s possible o n both clauses, a s in (18) : (18) [Paqarin-m i chaya-mu-nqan chay-qa ] Xwan-mi . tomorrow-DE arrive-Cis-3Fu-D E that-T o Juan-D E 'The person tha t will come tomorro w i s Juan.' In thos e rar e cases , however , i n which th e main tens e claus e i s interpreted a s a complement (an d has a complementizer chay), th e evidential i s no t possible: (19) Mana-(*n ) ka-n-ch u chay-t a yacha-ni . not-DE be-3-Ne g that-A c know- 1 'I know that there is none.' The possibilit y o f -chu in (19) , an d the absenc e of -mi/-n, wil l be accounte d fo r in sectio n 4 below.
3.5. Not in Imperatives As show n in (20 ) an d (21), respectively , imperativ e an d exhortative expression s cannot contai n evidentials : (20) a . Mikhu-y . eat-Im 'Eat!' b. *Mikhu-y-mi . eat-Im-DE
Focus in Quechua 38
3
(21) a . Mikhu-sun . eat-lpl,Im 'Let's eat.' b. *Mikhu-sun-mi . eat-lpl,Im-DE 3.6. Gapping Evidentials canno t occur in a clause wher e the tensed ver b has been deleted , as shown i n (22) : (22) a . Mama-y-q a Qusqu-ta- n ri-ra-n ; tata-y-ta q chakra-ta . mother-1-To Cuzco-Ac-D E go-Pa- 3 father-1-Em p field-A c 'My mothe r wen t to Cuzco; m y father to the field. ' b. *...chakra-ta- n .. . field-Ac-DE This ma y be interprete d a s implying tha t evidentials nee d t o b e supporte d b y a tensed ver b also in S-structure. 3.7. Not on Post-verbal Constituents Evidentials d o not occur on elements t o the right of the main verb. Thus all fou r constituents in (23) ma y carry the evidential: (23) a . Mariy a Xwana-ma n libru-t a qu-n-m i Maria Juana-t o book-A c give-3-D E 'Maria gives the. book t o Juana. ' b. Mariya- n Xwana-ma n libru-t a qu- n Maria-DE Juan,a-t o book-A c give- 3 c. Mariy a Xwana-man-m i libru-t a qu- n Maria Juana-l:o-D E book-A c give- 3 d. Mariy a Xwana-ma n libru-ta- n qun Maria Juana-t o book-Ac-D E give- 3 However, the ungrammatically of (24b) shows that post-verbal constituents may not, even thoug h it is possible to have a post-verbal constituent , a s in (24a) : (24) a . Mariy a Xwana-ma n qu- n libru-t a Maria Juana-t o give3 book-A c b. * Mariya Xwana-ma n qu- n libru-ta- n Maria Juana-t o give3 book-Ac-D E We ca n se e i n (25 ) tha t i t i s th e mai n tens e ver b whic h form s th e limi t fo r evidentials, no t the lexica l content verb:
384
Discourse Configurational Languages
(25) a . llank'a-q-m i ka-rqa-n i work-Ag-DE be-Pa- 1 'I wa s a worker.' / * 'I used t o work.' b. llank'a- q ka-rqa-ni- n work-Ag be-Pa-1-D E * ' I was a worker.' / ' I used t o work.' In fact , i n pas t habitua l structure s suc h a s (25b) , i n whic h th e ver b ka- 'be ' i s used a s an auxiliary in a raising context, th e validator has t o occur o n the auxil iary to the right of the main verb. Otherwise the sentence has a predicate + copula interpretation. 3.8. Evidentiality and Case Evidentials cannot occur on certain elements that are unmarked for case. In Cuzco Quechua, three type s of adverbial expressions ca n be distinguished : (a ) manne r adverbs that need accusative case, suc h as allin-ta good-A c 'well' ; (b ) tempora l expressions tha t may occur wit h and without case marker s i n all positions, suc h aspaqarin 'tomorrow' ; (c) temporal nouns that only can occur withou t case in a clause-initial topic-lik e position , suc h a s dumingu 'Sunday ' i n (26a) . Conside r the followin g examples : (26) a . duming u feri a ka-nq a Sunday marke t be-3F u 'Sunday there wil l be a market.' b. dumingu-t a feri a ka-nq a Sunday-Ac marke t be-3F u c. dumingu-ta- n feri a ka-nq a Sunday-Ac-DE marke t be-3F u d. *dumingu- n feri a ka-nq a Sunday-DE marke t be-3F u The contrast between (26c ) an d (26d) suggest s that evidentials canno t appear on elements i n suc h a topic position . 3.9 Interaction between Evidentials and Topic Marking In th e majorit y o f Quechu a sentence s i n whic h both evidential s an d th e topi c marker appear , ther e i s a clear division o f the sentenc e betwee n th e topicalize d and th e rhematize d constituents , whic h ma y be expresse d i n a formul a adapte d from th e on e give n i n Weber (1989) :
Focus in Quechua 38
(27)
fv\ {X-qag} -EV
(XPJ
5
I .. . {Z-qa-* }
where only on e evidential ma y occur . Between nul l an d tw o -qa phrases ma y occu r a t th e beginnin g o f th e sentence , followed b y a constituent o r the ver b marked wit h a n evidential, eventuall y fol lowed b y between nul l an d three other -qa phrases . The example s i n (28) illustrat e typical Quechu a theme/rheme structures : (28) a . cha y runa-q a Ayakuchu-ta- n ri- n ( = 1 1 a) that man-T o Ayacucho-Ac-D E go- 3 'That ma n is going to Ayacucho.' b. Mama-y-q a Qusqu-ta- n ri-ran mother-1-To Cuzco-Ac-D E go-Pa- 3 'My mothe r wen t to Cuzco. ' Both Wolc k (1968 ) an d Leviasohn (1976 ) remark o n th e pre-verbal positio n a s the preferre d on e fo r th e constituen t with the evidential , a s i n thes e examples , but I hav e faile d t o discove r anythin g mor e tha n a frequenc y patter n i n thi s respect.
3.70. Intermediary Summing up and First Analysis Let me begin by summarizing; what we have learned so far about the evidentials . In 3.1. through 3.6 . i t transpires that evidentials tak e a constituent in their scop e in indicativ e main tens e contexts , but ar e a t th e sam e tim e a kin d o f clausa l operator: 1. Evidential s are always constituent-external (wher e constituent i s NP, AP, PP, but no t necessaril y V P etc., a t least o n the surface ) 2. Evidential s occu r either o n the focuse d or rhematized, or o n the firs t con stituent 3. Ther e ca n only b e one evidential pe r claus e 4. Evidential s occur only in main clauses, i.e. onl y in the scope o f Main Tense 5. Imperativ e an d exhortative expression s canno t contain evidential s 6. Evidential s cannot occur in a clause where the tensed verb has been deleted. On th e basis o f this w e can conclud e tha t evidential s have : (a) scop e a t S-structure either over the clause or over a n individual constituent (NP, VP ) (b) scop e a t LF over the clause for the evidential force, an d over a constituent for th e focu s
386
Discourse Configurational Languages
Their interpretatio n is roughly as in (29) : (29) I
have witnesse d / heard i t told / surmis e tha t there i s an X (where X is an event , a predicate o r an argument ) such that propositio n (.. . X ... )
Let us assume tha t there i s a functional projectio n Ev P with roughly the follow ing structure :
Now tak e a n ordinar y cas e i n whic h th e firs t constituen t carrie s th e evidentia l marker, a s in (13) , repeate d her e a s (31) : (31) Pidru- n wasi-t a ruwa-n . Pedro-DE house-A c make- 3 (a) 'I t i s Pedro tha t builds a house.' (b) 'Pedr o builds a house.' I will assume that the (a) reading, with focus, corresponds t o a structure in which Pedro ha s bee n raise d t o th e Spe c positio n o f EvP, whil e i n th e (b ) reading th e validator syntacticall y remain s external to IP (and has the whole IP in its scope), but phonologicall y encliticize s ont o th e firs t majo r constituen t o f IP , through something lik e (32) : (32) evidentia l X P = > 2 + 1 12
Notice tha t th e resultin g structure s ca n b e directl y interprete d i n term s o f (29 ) with th e differenc e tha t ther e i s eithe r a n even t o r a n individua l o f whic h th e evidential statu s is predicated . It is clear that evidentials wil l be constituent-external unde r the present analysi s and wh y the y wil l occu r eithe r o n a focused o r rhematized , o r on th e firs t con stituent. Sinc e th e evidentia l wil l hav e a n I P a s it s complement , ther e ca n onl y be one evidential pe r clause. We can explain that evidential s occur onl y in main clauses, tha t imperativ e expression s canno t contai n evidentials , an d tha t
Focus in Quechua 38
7
evidentials canno t occu r i n a claus e wher e th e tense d ver b ha s bee n delete d through th e assumptio n that Ev P selects a n IP headed b y a n elemen t wit h par ticular features, such as main tense (analyzed in Lefebvre and Muysken 198 7 a s containing a reference t o the momen t of speech) . Consider no w th e mor e comple x cases , allude d t o i n 3.7 . throug h 3.9. , an d exemplified i n a sentence suc h as (33) (= 28b) : (33) Mama-y-q a Qusqu-ta- n ri-ran mother-1-To Cuzco-Ac-D E go-PA- 3 'My mothe r wen t to Cuzco.' 7. Evidential s d o not occur on elements t o th e righ t o f V 8. Evidential s cannot occur on certain elements that are unmarked for Case. 9. Evidential s occur on constituents that are often precede d b y topicalize d constituents. I will assume that elements can be scramble d and adjoined left - o r right o f EvP, often wit h a topic or theme interpretation (marked by -qa or not). On the surfac e left-adjoined element s wil l occu r t o the lef t o f Spe c o f EvP. I have no clear analysi s of the precise natur e o f this left - o r right-adjunction , yet, and hence no explanation for the fact that the adjunction canno t be to IP itself (unless we assume tha t topicalized element s mus t be outside o f the scope of the evidential a s such) .
4. Interaction between Evidentials, Negation, and Questions In Quechu a there is a close link between the evidentials an d the non-factual particle -chu which mark s both question s and negation.2 The present study o f focu s in Quechua cannot pretend to provide a definitive analysis, but some remarks ar e called for . I will begin by briefl y analyzin g negation.
4.1. Negation Negation is generally marke d by an initial element mana and an enclitic particl e -chu, mentione d before . As see n in (34b) , we get ama in prohibitiv e (negativ e imperative) contexts . (34) a . Mana- n ri-n-chu . not-DE go-3-Ne g 'He does not go.' b. Am a ri-y~chu . not go-Im-Ne g 'Don't go. '
388
Discourse Configurational Languages
In (34a ) but not (34b ) th e evidential occurs , attache d t o the negation element . (35) t o (37) sho w that -chu is limited t o main tense contexts. In (35) w e have noun phrase-interna l negation . (35) cha y man a alli n run a that no t goo d ma n 'that bad man' In (36a ) negatio n -chu is marke d interna l t o a nominalized clause , an d i n (37a ) internal to a n adverbial clause . (36) a . *Man a ri-sqa-n-ta-ch u yacha-ni . not go-Nom-3-Ac-Ne g know- 1 'I know that (s)he does not go.' b. Man a ri-sqa-n-t a yacha-ni . not go-Nom-3-A c know- 1 (37) a . *Man a ri-sqti-n-ch u ri-saq-mi . not go-Sub-3-Ne g go-lFu-D E 'If (s)h e doe s not go I'll go. ' b. Man a ri-qti- n ri-saq-mi . not go-Sub- 3 go-lFu-D E Notice tha t the (b ) examples, withou t -chu, ar e fine . In contras t wit h th e evidential , -chu ca n occu r i n mai n tens e complemen t clauses suc h as (39). It s matrix equivalent, with the evidential , is given i n (40) : (38) Mana- n ka-n-chu . be-3-Neg not-D E 'There is none.' (39) [Mana-(*n ) ka-n-ch u chay-ta ] yacha-ni . not-DE be-3-Ne g that-A c know- 1 'I know that there i s none.' Like evidentials , negatio n ma y b e place d contrastively . Whil e i n (40a ) th e whole clause o r at least th e predicate i s negated , i n (40b ) i t i s an adverbial : (41) a . Mana- n Qusqu-ma n ri-ni-chu . not-DE Cuzco-t o go-l-Ne g 'I do not go to Cuzco. ' b. Mana- n Qusqu-man-ch u ri-ni . not-DE Cuzco-to-Ne g go1 'It i s not to Cuzco that I go (but somewhere else).' The analysi s I tentativel y propose assumes that three functiona l projection s ar e involved: th e Ev(idential)P , a Neg(ation)P , headed b y mana o r ama, an d a
Focus in Quechua 38
9
N(on)F(actual)P, headed by -chu.3 The NFP selects mai n tense IPs , just lik e th e EvP, bu t thes e nee d no t b e indicative , henc e th e possibilit y o f -chu wit h prohibitives. Th e NegP selects eithe r an NFP, yielding negation with -chu or any XP, yielding constituen t internal o r subordinat e clause-internal negation . I wil l illustrate these assumption s with the proposed D-structur e analysi s for (41b):
At S-structur e Qusqu-man 't o Cuzco ' i s move d int o the Spe c o f NFP , yielding Qusqu-man-chu, an d I wil l assum e withou t argumen t tha t evidentia l -mi/-n encliticizes ont o mana 'not ' i n the phonological representation .
4.2 Interaction between Evidential and the Question Marker Now tha t w e hav e a n analysi s of -chu a s headin g a non-factua l phrase (NFP) , the analysis of questions becomes straightforward . Yes-No questions ar e simply NFPs withou t a dominating negation phrase. A typical cas e woul d be (43) : (43) Hamu-nqa-chu . come-3Fu-Q 'Will he come?'
390
Discourse Configurational Languages
In m y analysis , th e tense d ver b hamu-nqa wil l mov e int o th e Spe c o f NFP , o r alternatively -chu wil l encliticize ont o th e verb, withou t there bein g movement . The fact tha t a question like (44) is ambiguous in the same wa y as was sketche d above fo r the evidential s supports thi s dua l analysis : (44) Qusqu-man-ch u ri-nk i Cuzco-to-Q go2 (a) 'I s i t to Cuzc o that yo u are going?' (b) 'Ar e you goin g to Cuzco? ' In th e (a ) cas e ther e ha s bee n movement , i n th e (b ) cas e encliticization . Whe n there i s onl y on e word , a s in (43) , yo u simpl y canno t tell. While normall y question s occu r withou t evidentials, the y ca n b e combined , as in (45 ) (= 3b) : (45) nuqa-chu-c h I-Q-Dub 'Perhaps me?' I have to assume tha t first nuqa is raised int o SpecAFP an d then the whole AFP is raised int o SpecEvP . Negative question s hav e the for m o f (46) : (46) mana-ch u hamu-nk i not-Neg come2 'Won't yo u come?' I will simply assume that -chu encliticizes ont o the head preceding it . In fact the limitation t o on e -chu i n negative questions i s on e o f th e argument s for assum ing tha t th e questio n an d the negation elemen t ar e identical. In Wh-questions , ther e i s n o -chu, a s ca n b e expected . (47 ) show s tha t th e question wor d i s moved int o th e SpecEv P position , an d is hence focused . (47) Pi- n chay-p i puri-n ? who-DE that-L o walk- 3 'Who walks there?' In earlier accounts, the fact that question words moved to the left althoug h Com p is o n the righ t i n Quechu a was alway s a bit o f a mystery .
Focus in Quechua 39
1
5. Conclusion I hope to have given an initial account of the complex interaction betwee n focu s and evidentiality, with some insight into their relation t o questions an d negation. A number of issues remain for further exploration , (a ) To what extent is the preverbal positio n privilege d a s th e positio n fo r th e constituen t marke d wit h th e evidential, and how does this affect my analysis? (b) Levinsohn (1975) notes that post-verbal position s i n Colombia n Quechu a ca n als o b e rhematized . Ho w gen eral i s this , an d wha t woul d b e th e implication s fo r m y analysis ? (c ) Quechu a varieties diffe r i n the extent to which -chu an d the evidentials ca n be combined ; how is this difference t o be represented? (d ) Are there arguments to force a choice between a movement t o SpecAFP/Ev P and a n encliticization i n P F analysis ?
Notes * I a m gratefu l t o al l researchers whos e earlier wor k on focu s an d evidential markin g I could rely on , and particularly Parker (1965) , Wolck (1968), an d Weber (1987). Many of the observations reported wit h respect to the relation between evidentials an d 'main tense' appear in Lefebvre an d Muysken (1988). I wish to thank the participants i n the HIL Spring 1992 semina r on Quechua syntax , and especiall y An a Alvarez an d Simon va n d e Kerke , for thei r comments . 1. In addition to these theoretical problems wit h the analysis in Muysken (1981), ther e is an empirical problem : th e status of genitive case . As shown in (i) and (ii), the genitiv e also has a short (i ) and a long (ii) form i n Quechua, under th e sam e phonologica l condi tions a s the evidentials . (i) wasi- q punkunhouse-Ge door- 3 'the doo r o f th e house* (ii) wasi-n-p a punkun -p house-3-Ge door- 3 'the door o f his house'
q/V
a/C
Originally th e short for m was -p, and the present -pa/-q alternatio n i s irregular. I n Cuzc o Quechua i t has bee n regularize d onc e agai n in that th e complete form tha t appears wit h the onl y monosyllabi c C V noun in the language, pi 'who' , is -qpa, as in (iii): (iii) pi-qp a ususin who-Ge daughter- 3 'whose daughter ' From this form both -q and -pa, in the appropriate contexts, ca n be derived throug h fairly regular phonologica l rules . It is clear fro m th e contras t in grammaticality betwee n (iv ) and (v), however, tha t th e genitive i s no t 'clas s free ' i n th e sam e wa y a s the evidentials .
392
Discourse Configurational Languages (iv) wasi-n-p a house-3-Ge (v) * hamu-n-pa come-3-Ge
One might object that it may be ungrammatical to case-mark a verb for independent rea sons, suc h a s the Case Resistanc e Principl e (Stowell , 1981) . Notice , however , tha t many verbs are case-marked i n Quechua, but only in a nominalized form. Thus, there is no strict correspondence betwee n phonologica l properties an d being class-free . That genitiv e cas e coul d b e analyze d as a syntacticall y independent elemen t i s clea r from it s interactio n wit h delimitative -lla. As can be see n i n (viii) , th e genitiv e pattern s with th e encliti c o r non-encliti c postposition s i n (vii ) i n bein g followe d b y -lla, rathe r than wit h th e cor e cas e marker s i n (vi), whic h are preceded b y -lla. (vi) *-man-ll a -lla-ma *-ta-lla -lla-t *-pi-lla -lla-p
n 'to a 'objective i 'locative
' ' '
(vii) -kama-ll a *-lla-kam a 'until ' -pura-lla *-lla-pur a 'among ' -hina-lla *-lla-hin a 'as , like' ladu-lla-pi *-lla-ladu-p i 't o the side o f (viii) -pa/-q-ll a *-lla-pa/-
q 'genitive
'
The patter n i n (vi-viii ) ca n b e explaine d b y assumin g that -lla attache s t o bar e nouns , and tha t postpositions , includin g th e genitiv e cas e particle , ar e i n fac t bar e nouns , subcategorized fo r a nou n phrase . Thi s woul d explai n wh y genitiv e i s th e onl y 'cas e marker' that may be followed by another case marker in Quechua, a long-standing puzzle , at least i n th e work reporte d o n in Lefebvre an d Muysken (1988) . 2. This sectio n owe s muc h to discussions wit h Ana Alvarez, who is pursuing researc h into th e statu s of functiona l categorie s i n Bolivia n Quechu a i n he r doctora l thesi s t o b e defended a t the Universida d Aut6noma d e Madrid . 3. Th e assumptio n that manalama hea d a phras e give s a natura l explanatio n fo r th e fact tha t the negation elemen t ha s moda l forc e itself .
References Aronoff, M . (1976 ) Word Formation in Generative Grammar, MI T Press , Cambridge , Massachsetts. Klavans, J. (1985 ) "Th e Independenc e o f Phonology an d Synta x in Cliticization," Language 61 , 95-120. Lefebvre, C. , an d P . Muysken (1987 ) Mixed Categories. Nominalizations in Ouechua, Kluwer Academic, Dordrecht . Levinsohn, S.H . (1975 ) "Functiona l Sentenc e Perspectiv e in Inga, " Journal of Linguistics 11 , 13-37. Muysken, P . (1981) "Quechu a Wor d Structure," i n F . Heny, ed., Binding and Filtering, Croom Helm , London , 279-327.
Focus in Quechua 39
3
Parker, G. (1965) Ayacucho Quechua Grammar, Mouton, The Hague. Weber, D . (1989 ) Hudnuco (Huallaga) Quechua Grammar. UCL A Publications i n Amerindian Languages, Lo s Angeles. Williams, E. (1981) "On the notions 'lexicall y related' and 'hea d of a word'," Linguistic Inquiry 12 , 245-274. Wolck, W. (1968 ) Foco y especificacion en quechua. CILA, Lima , Peru.
Abbreviations in the Glosses Ac accusativ e case Ag agentiv e nominalizer Asp (intensive ) aspect Cau causativ e Cis cislocativ e Cont contrastiv e DE direc t experience Def definitiv e Dub dubitativ e Emp emphati c Fu futur e tens e Ge genitiv e case: HS hearsa y Im imperativ e Ind indefinit e Lo locativ e cas e Neg negatio n Nom nominalize r Pa pas t tense pi plura l Pot potentia l mood Pr progressiv e aspec t Q questio n particle SD sudde n discovery tens e Sub adverbia l sutaordinato r To topi c marke r