UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

From sentence to discourse : integrated explanations for certain linguistic phenomena in Japanese Ono, Mieko 1990

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-UBC_1991_A1 O56.pdf [ 10.35MB ]
JSON: 831-1.0100456.json
JSON-LD: 831-1.0100456-ld.json
RDF/XML (Pretty): 831-1.0100456-rdf.xml
RDF/JSON: 831-1.0100456-rdf.json
Turtle: 831-1.0100456-turtle.txt
N-Triples: 831-1.0100456-rdf-ntriples.txt
Original Record: 831-1.0100456-source.json
Full Text

Full Text

FROM SENTENCE TO DISCOURSE: INTEGRATED EXPLANATIONS FOR CERTAIN LINGUISTIC PHENOMENA IN JAPANESE By MIEKO ONO B.A., Tokyo Women's Christian College, 1973 M.A., Tokyo Women's Christian University, 1975 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (DEPARTMENT OF ASIAN STUDIES) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August 1990 © Mieko Ono, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of A o a n S t u d i e s The University of British Columbia Vancouver, Canada Date N r w 1 Q I Q QD DE-6 (2/88) i i Abstract This thesis examines those aspects of language in which syntax and extra-syntact ic factors interface. There are three major approaches to d i scourse study (interpretation of a sentence in discourse): i) Discourse Study without Syntax: Any 1 inguist ic phenomena can be explained through discourse; syntax is dissolved into discourse study. i i) Discourse Study Interacting wi th Syntax: Syntact ic rules and discourse functions interact or intermingle wi th each other. i i i ) Modular Approach to Discourse Study: Syntax is auto-nomous, but can feed information into other ext ra-syntact ic components to obtain the f inal interpretation of a sentence in context. Approach ( i i i ) is adopted here, where a "Government and Binding" (Chomsky (1981)) type of generative grammar is assumed as the syntact ic framework. Four l inguist ic phenomena in Japanese are chosen for case studies of the mode of interaction between syntax, semantics, and pragmatics. i i i Chapter 2 examines the Japanese re f lex ive zibun. The c o -reference problem is solved through syntact ic rules for the anaphoric use and discourse rules for the referential use. Chapter 3 examines demonstrat ives. Since they are or iginal ly used as de ix is , the problem is mainly d i s c u s s e d in semant ic and discourse arenas. The comparison between pronoun and demonstrative is also discussed. Both Chapter 2 and Chapter 3 deal with the problem of c o r e f e r e n c e , w h i c h is one d i s a m b i g u a t i o n m e c h a n i s m in the comprehension of discourse. Chapter 4 examines quant i f ier interpretat ion. Th is problem involves another type of d isambiguat ion mechanism. The scope interpretation being represented in LF, why a certain reading is obtained in the actual discourse environment is explained from the viewpoint of the human attention system (conscious and unconscious). Chapter 5 examines the part ic le wa, which is most commonly considered a topic marker or an old information marker. Wa marks a certain semantic structure in syntact ic representation and such a wa-sentence has an important funct ion in d iscourse organizat ion. The nature of the cont ras t iveness assoc ia ted wi th a wa sentence is explained in this light. In this modular type of approach, the phenomena which were formerly explained by fa i r ly complex sets of rules have become more transparent, and some seemingly conf l ic t ing analyses done in the past iv a r e n o w c o n s i d e r e d a s a n a l y s e s of d i f f e r e n t a s p e c t s o f a s i n g l e phenomenon . ) V T A B L E OF C O N T E N T S Abstract i i Table of Contents v Acknowledgements ix Chapter 1: Introduction 1 Notes(Chapter 1) 7 Chapter 2: The "Reflexive" Pronoun in Japanese 2.0 Introduction 8 2.1 Three Types of Pronoun Usage 8 2.2 The Referential Use of Zibun 12 2.3 The Anaphoric Use of Zibun 18 2.4 Anaphor, Pronominal, and Variables 2.4.1 The Binding Theory 24 2.4.2 Governing Domain 25 2.4.3 Topic-Bound vs. Subject-Bound Topic-Antecedent 30 Topic vs. Subject 37 Zibun as a variable 40 vi 2.4.4 Zero-topic and Referential Use 42 2.5 Zibun and Kuno and Kuroda's Extra-Syntactic Notions 47 2.6 Examination of Discourse Data 52 2.7 Conclusion 61 Notes (Chapter 2) 63 Chapter 3: Japanese Demonstratives: Ko-So-A 3.0 Introduction 68 3.1 The Use of the Demonstratives 3.1.0 Introduction 70 3.1.1 Ternary vs. Binary 73 3.1.2 Territory Formation: Physical Distance vs. Psychological Distance 79 3.1.3 Speaker's Territory and Egocentricity 85 3.2 The Use of Ko/So in Discourse 3.2.1 Introduction 101 3.2.2 The Ko/So - Antecedent Relation 106 3.2.3 Problems in the Use of So 122 3.3 Conclusion 132 Notes (Chapter 3) 134 v i i Chapter 4: Quantifier Reading 4.0 Introduction 137 4.1 Quantifier Interpretation 4.1.1 Quantifier Reading in Japanese 137 4.1.2 Linear Order Hypothesis and Subjecthood 1 40 4.1.3 Wide/Narrow vs. Same/Different 1 46 4.2 "Attention" Systems in Discourse 153 4.3 Focus of Interest, Quantifier, and Word Order 1 58 4.4 Conclusion 166 4.5 Addendum 4.5.1 Grosz's 'Focus' 1 67 4.5.2 Kuno's 'Focus' 1 71 4.5.3 Rochemont's'Focus' 177 Notes (Chapter 4) 180 Chapter 5: The Function of the Particles Wa and Ga 5.0 Introduction 184 5.1 The Particle Wa 184 5.1.1 A Characteristic of the Use of Wa 185 5.1.2 Wa and Human Judgement 197 5.1.3 Judgement Marker vs. Topic Marker 200 5.1.4 Wa in Contrastive Reading "Contrastiveness" and the Categorical Judgement....203 v i i i "Contrastiveness" and Wa (I): Particle Deletion 206 "Contrastiveness" and Wa (II): Contrastive Construction 21 1 "Contrastiveness" and Wa (III): Indefinite Nouns 217 "Contrastiveness" and Wa (IV): Subject NPs and Definiteness Effect 222 The Nature of "Contrastiveness" 230 5.2 Ga and Wa 5.2.1 Ga vs. Wa 240 5.2.2 Wa, Ga, and Discourse 243 5.3 Conclusion 251 Notes (Chapter 5) 253 Chapter 6: Concluding Remarks 259 Bibliography 262 Appendix 285 ix A c k n o w l e d g e m e n t s T h i s t h e s i s h a s been c o m p l e t e d due to t he h e l p of m a n y p e o p l e . F i r s t , I w o u l d l i k e t o t h a n k m y r e s e a r c h s u p e r v i s o r , P r o f e s s o r M a t s u o S o g a , f o r h i s e n c o u r a g e m e n t and p a t i e n c e a s w e l l a s f o r i n n u m e r a b l e p i e c e s of a d v i c e and s u g g e s t i o n s a l l t h r o u g h m y g r a d u a t e s t u d i e s a t t h e U n i v e r s i t y of B r i t i s h C o l u m b i a . W i t h o u t h i s h e l p , I c o u l d no t h a v e m a n a g e d to s o l v e the p r o b l e m s w h i c h a r o s e in the c o u r s e of c o m p l e t i n g m y w o r k . I w o u l d a l s o l i k e t o e x p r e s s s p e c i a l t h a n k s t o P r o f e s s o r M i c h a e l R o c h e m o n t . My d i s c u s s i o n s w i t h h i m h a v e g i v e n m e g r e a t i n t e l l e c t u a l s t i m u l a t i o n , and h i s i n s i g h t f u l s u g g e s t i o n s h a v e a l w a y s been " e y e - o p e n e r s " to m e . I l e a r n e d a l o t f r o m h i m , no t o n l y a b o u t l i n g u i s t i c s , bu t a l s o f r o m h i s i n d e f a t i g a b l e s p i r i t in s o l v i n g p r o b l e m s . My s p e c i a l t h a n k s go t o P r o f e s s o r s H i r o k o T e r a k u r a a n d K e n - i c h i T a k a s h i m a a s w e l l . T h e i r c o n s t a n t e n c o u r a g e m e n t and s u g g e s t i o n s n e v e r f a i l e d to r e m i n d me of t h e b r i g h t s i d e of l i f e w h e n I w a s a l m o s t o v e r w h e l m e d by t he d i f f i c u l t i e s I f a c e d , and t h e y h e l p e d me r e g a i n the c o u r a g e to c o m p l e t e my w o r k . I am a l s o d e e p l y i n d e b t e d to P r o f e s s o r s A s h o k A k l u j k a r and P a t r i c i a S h a w f o r t h e i r c o m m e n t s , e n c o u r a g e m e n t , and s u p p o r t . P r o f e s s o r s N a o m i M c G l o i n and Guy C a r d e n c o n t r i b u t e d t o t h e r e v i s i o n o f my t h e s i s w i t h t h e i r i n v a l u a b l e c o m m e n t s . Many t h a n k s t o t h e m , e s p e c i a l l y t o P r o f e s s o r C a r d e n f o r h i s g e n e r o u s o f f e r of t i m e f o r d i s c u s s i o n . I w o u l d f u r t h e r l i k e to e x p r e s s m y a p p r e c i a t i o n to P r o f e s s o r s Bob L e v i n e and S u s a n n e C a r r o l l , v i s i t i n g r e s e a r c h e r s i n t h e L i n g u i s t i c s d e p a r t m e n t , w h o gave me v a l u a b l e i npu t to a p a r t of m y e a r l i e r d r a f t . My f e l l o w s t u d e n t s a l s o p a r t i c i p a t e d in m y t h e s i s in d i r e c t and i n d i r e c t w a y s : s o m e of t h e m r e a d a l l o r p a r t s of m y m a n u s c r i p t s i n t h e i r v a r i o u s s t a g e s ; s o m e h e l p e d t o c l a r i f y p r o b l e m s in d i s c u s s i o n s ; s o m e o f f e r e d m e t e c h n i c a l a s s i s t a n c e ; and a l l o f t h e m m a d e m y g r a d u a t e l i f e m e m o r a b l e . T h e s e p e o p l e a r e : H e l e n L i s t , K a t h y H u n t , B r u c e B a g e m i h l , H e n r y D a v i s , M a s a h i k o N a k a t a , G a r y A r b u c k l e , and R o s e m a r y H a d d o n . I e s p e c i a l l y a p p r e c i a t e M a s a h i k o N a k a t a ' s c o n s t a n t f r i e n d s h i p a l l t h r o u g h m y g r a d u a t e l i f e a t U. B. C. A n d v e r y s p e c i a l t h a n k s go to G a r y A r b u c k l e f o r h i s a s s i s t a n c e in t y p i n g out and p r o o f - r e a d i n g my m a n u s c r i p t s . W i t h o u t h i s h e l p in s e t t i n g up the m a t e r i a l , m y t h e s i s w o u l d not have been c o m p l e t e d . I f e e l t h a t I am v e r y f o r t u n a t e to have had s o h e l p f u l a G r a d u a t e A d v i s o r a s P r o f e s s o r K i n y a T s u r u t a a n d s o p a t i e n t and r e s o u r c e f u l a d e p a r t m e n t s e c r e t a r y f o r g r a d u a t e s t u d i e s a s R a c h e l R o u s s e a u . M a n y t h a n k s f o r t h e i r s u p p o r t . L a s t , I w i s h to s a y "Domo a r i g a t o o g o z a i m a s i t a " to my m o t h e r Y o s h i k o Ono, b r o t h e r K o y a , and m y d e a r g r a n d m o t h e r Y o s h i . 1 Chapter 1 Introduction There are many l inguist ic phenomena where syntax and discourse interact (Kuroda 1973, Inoue 1982). As for the mode of this interaction, Kuno's series of works (1978, 1979, and elsewhere) for example, called Functional Grammar, integrate syntactic and discourse rules into rule sets based on functional perspectives. In his framework, the phenomena by which syntact ical ly well-formed sentences turn out to be i l l - formed sentences in discourse are given interesting explanations: e.g., i f the v io la t ion of the discourse rules is the resul t of the inevitable application of syntactic rules, the penalty is low (the sentence is not felt to be too bad) (1978, Chapter 3). In the spectrum of interaction, considering Kuno's system to occupy the middle ground, one end w i l l be occupied by Chomsky's generative grammar, which claims an autonomous system of grammar; and the other by discourse-oriented grammar (Givon 1979). This thes is assumes a "Government and Binding" type of generative grammar, and several l inguis t ic phenomena are analyzed. Many important works of functional grammar are also taken into consideration in the course of research, as they deal wi th the area in 2 which syntact ic and discourse phenomena overlap. The reason for adopting the generative grammar approach is that the theory of the autonomy of grammar and the modular account of human language provides an explanation for a complex phenomenon by means of a combination of much s impler rule sets and pr inciples , instead of employing rules which are directly applied to the phenomena and which attempt to explain everything at once; and furthermore makes it possible to explain independent phenomena by the use of the same principles (integrated explanation of l inguistic phenomena). In this framework, formal grammar 1 and other human cognitive functions are viewed as autonomous, each being a modular component of language. What actual variety of modules language consists of is s t i l l under investigation. Newmeyer (1 983) presents one potentially accurate but preliminary sketch, shown below: • (1) "Language" [Newmeyer: p. 3] 3 As for the formal grammars, Extended Standard Theory and Government Binding Theory assume the following model: (2) D-structure Syntactic Move-alpha S-structure LF Move-alpha Phonetic Form (PF) Logical Form (LF) [Saito (1985): p. 1 1 ] LF and PF are syntactic representations for semantics and phonology, respectively. LF is followed by LF' as assumed in Chomsky (1977). Semantics by i tself , which may possibly form an independent module, may be distinguished from pragmatics, 2 which can be broken down into several modules (which probably includes those conversat ional p r inc ip les proposed by Grice (1975) as one of the modules, as Newmeyer's model suggests). The interact ion between syntact ic , semantic, and pragmatic phenomena w i l l be viewed in a s i m i l a r direction as in Dretske (1972): (3) ...the semantical differences between C (U) and C (u") are to be represented as having their source in the apparently non-semantical differences between U and U' - is by PF Move-4 cal l ing the difference while acknowledging that certain pragmat ic d i f fe rences , jus t as ce r t a in s y n t a c t i c differences, figure importantly in the semantic analysis of some expressions in which they are embedded. (C(U) is a larger l inguis t ic expression in which an expression U is embedded.3 [M. 0.]) [p. 427] In this l ight , passive, for example, would be viewed as having a syntactic representation as a sentence, but the distr ibution of passive constructions is conditioned by discourse (Will iams 1977, Oosten 1984). The si tuations are the same when a certain pragmatic phenomenon is closely connected to a semantic phenomenon, a semantic phenomenon to a syn tac t ic phenomenon, a syntac t ic phenomenon to a pragmatic phenomenon (cf. Kuno's functional grammar), and so forth. For example, Rochernont (1985) is a work carried out wi th in the framework of the autonomy of grammar wi th modular components, which accounts for English focus as in (4): (4) Focus, it is concluded, is a syn tac t i ca l ly represented notion wi th systematic though varying phonological and semantic representations, [pp. 177-178] In this thesis , I have selected four l ingu is t i c phenomena in Japanese, some of which have been studied largely in syntax, some of which have been studied for the most part in pragmatics (context-dependent aspect), and some of which have been studied from both angles, but not often as the result of the interaction between syntax and pragmatics; and I invest igate how the modular account of these 5 phenomena provides us wi th a much simpler but integrated explanation for them. The f i rs t two chapters deal wi th the problem of coreference. In Chapter 2, the reflexive pronoun zibun is examined. This topic has been examined mainly in syntax. However, I w i l l argue that syntactic rules and pragmatic (discourse) accounts of the coreference problem of zibun w i l l provide a fuller explanation of the phenomenon as a whole. In Chapter 3, demonstrat ive pronouns are examined. The coreference problem of demonstrative pronouns is considered to be mainly one of a pragmatic nature, since their original function is deixis. The central concerns of the investigation are the semantic nature of the demonstrative, and the demonstrative in discourse anaphora. The former w i l l occupy a good deal of space, since it is related to the manner of "demonstration"; in the latter, demonstratives are compared wi th pro-nouns in terms of the usage in discourse anaphora. In Chapter 4, I w i l l examine interpretation of quantifier nouns in a sentence. Quantifier interpretation has been considered in syntactic terms, but I w i l l show that the choice between the readings is determined by extra-syntactic factors. In Chapter 5, the property and function of the particle wa, which is best known as a "topic" marker, is discussed. The examination w i l l be carried out from the standpoints of semantics, syntax, and pragmatics. 6 In the course of this thesis, some of the apparently confl ict ing analyses done in the past may become, not alternative solutions to the same problem, but merely studies of the different aspects (modular components) of phenomena. Independent l inguis t ic phenomena, on the other hand, may be found to be dominated by the same principle of cognitive functioning. Even certain phenomena which are seemingly particular to Japanese can be recognized as manifestations of functions of human cognition, ones which may happen to be strongly dominating factors in Japanese, but scarcely specific to it alone. 7 Notes(Chapter 1) 1 Formal grammar i t se l f is modular (cf. Chomsky (1981), Nakajima (1985-86), and LGB-type generative grammar. 2 Gazdar (1979) puts i t this way: Pragmatics = Meaning-Truth Conditional [p. 2], Wilson (1975), on the other hand, proposes non-truth conditional semantics. In this thesis, I assume that semantics is at least truth-conditional (with the possible addition of non-truth conditional and conventional meaning (?); cf. Wilson), and that pragmatics is the context-dependent aspect of meaning, although the line between the two has not been clearly defined so far. 3 U and IT are, for example, sentences having contrastive features (stress), which Dretske regards as pragmatic features, adapting the definit ion of pragmatics in C. Morris and R. Carnap {international Encyclopedia of United Science, Chicago, 1955). Although his view of semantics and pragmatics differs from the one I have adopted in this thesis, the point made by Dretske offers an idea of how to look at the mode of interface between syntax, semantics, and discourse. 8 Chapter 2 The "Reflexive" Pronoun in Japanese §2.0 Introduction The aim of this chapter is to demonstrate how zibun, which is generally considered to be the Japanese ref lexive , establ ishes a coreferential relation to another noun. My concern is not the generation of zibun but i t s disambiguating mechanism in terms of the inter-pretation. Although zibun is often regarded as a counterpart of English reflexives such as myself, yourself, or himself, zibun also exhibits a fair number of s imi la r i t i e s to pronouns in English. This characterist ic of zibun w i l l be taken into serious consideration in this chapter. §2.1 Three Types of Pronoun Usage According to Evans (1980), there are three categories ' in the usage of pronouns. Let us cal l them the referential use, the anaphoric use, and the bound variable use, in consideration of the relationship between a pronoun, i ts referent, and the expression where the referent is pinned down as a l inguistic form, if any. 9 Referential Use (1) Look at him! (e.g., pointing at a man walking over there) (2) I heard that the company which John was working for went bankrupt. I wonder how he is making his l iving now. Anaphoric Use (3) John said that he w i l l go to Mexico next month. (John = he) (4) John loves his mother. (John = his) Bound Variable Use (5) Everyone loves his mother. In (1) and (2), the object as a referent must actually be pointed out or described as sal ient in some manner. In determining i ts reference, pragmatic factors are involved. In (3) and (4), a coreferential relation is established between the NP John and the pronoun he since John and he refer to the same person. In (5), although everyone by i t se l f cannot make a reference to any person, each member of a set of everyone can be i ts referent just like the bound variables in logic. In examples (3) and (4), the reading that he is not John belongs to the referential use. 2 Lasnik (1979) proposes a non-referent ia l r u l e ; 3 the rule syn tac t ica l ly determines when two NPs are "not coreferential" and leaves the rest of the cases without discriminat ing referential cases (1) and (2) and anaphoric cases (3) and (4). As a consequence, the Lasn ik- type approach resu l t s in the coreference (but not non-coreference) relation is eventually determined in the pragmatic domain. 10 However, Evans considers the existence of case (5) as very crucial , and places emphasis on the connection (though not immediate) 4 between (3)-(4) and (5), drawing a dis t inct ion between "intended coreference" and " r e f e r e n t i a l l y dependent r e l a t i on (g rammat ica l coreference)." Rochemont (1983) makes the further assumption that "a pronoun is re ferent ia l ly dependent on some NP only if the two phrases are coindexed." Now, in the case of the Japanese reflexive zibun, although i ts anaphoric aspect tends to get primary attention, I must point out that despite the name "reflexive," which in English denotes behavior which is syntact ical ly different from the pronoun (cf. Chomsky (198 0 ) , zibun in fact exhibits all three usages of pronouns, as shown in (6) - (9): Referential Use (6) Ara, ara, darenimo yarazu ni zibun bakari ga anybody (to) not giving self only NOM tabete iru wa. eat -ing 'Oh my, without giving it to anybody, only self is eating.' (7) Zibun wa hi tori de saki ni kaetta rasi i yo. self TOP alone earlier went home seem 'It seems that self alone left earlier.' [Since zibun is usually identified wi th English self, the tentative translations are given using that word, although the best and most idiomatic rendering of these examples would be with the pronoun he. ] Anaphoric Use (8) Taroo wa [zibun no oya] o taisetu ni suru. Taroo TOP self's parent(s) ACC be attentive to 'Taroo is attentive to self's parent(s).' Bound Variable Use 5 (9) Daremo ga [zibun no oya] o taisetu ni suru. Everyone NOM 'Everyone is attentive to self's parents.' Taking the Evans-type approach (also cf. Reinhart (1983)), I w i l l divide (6) - (9) into two categories: (8) and (9) w i l l be called anaphoric, and (6) and (7) referential in this chapter. In the following sections, I w i l l f i rs t examine the referential use of zibun, which is pragmat ical ly "bound" 6 in discourse, then the anaphoric use, which is syntac t ica l ly bound. I w i l l also show how coreference under syntactic res t r ic t ion and coreference conditioned by discourse interact on a single instance of zibun in a sentence. A 12 sentence is taken to be both a minimum unit of discourse and a syntactic product. §2.2 The Referential Use of Zibun Firs t , observe the fol lowing examples (10) - (12). Imaginable contexts for these sentences would be internal monologues including interjectional expressions. (10) Zibun ga ikeba yokatta. self NOM go (provisional) be preferable-PAST '(I wish) self (= I) should have gone.' (11) Konna toki koso zibun ga sikkari sinakereba. in this very situation self NOM should become courageous 'In this very situation self (= I) should become courageous.' (12) Kore wa zibun mo ukauka site irarenai zo. self (NOM) too be off one's guard can-not-be-ing 'Well, self (= I) too can not be off 'self's' guard.' In an internal monologue, the speaker is the only one in the "world" created by his consciousness (internal world); in other words, he occupies an absolute position and does not have to identify his position in relat ion to other people. Zibun is uniquely chosen in an internal monologue, instead of the f i r s t person pronouns such as ore, boku, watasi, and the like, the choice of which is socio-cultural ly determined; 13 i.e., consideration of the outside world is involved. However, zibun is not merely the substitute for the f i r s t person pronoun /; it is rather "self" in opposition to "other." (13) - (15) show clear instances of the self/other opposition: (13) Hito wa hito, zibun wa zibun da others (contrastive) self (contrastive) COP 'Others are others, self is self.' (14) Anotoki wa [NP [NP zibun ga nigeru] no] ga that time TOP [ self SUB escape] NOM seiippai de, tanin no koto made doing everything in one's power COP other's matter as well kangaete irarenakatta. think not be-ing-PAST 'At that moment, all self could do was to escape and could not think about others as well . ' (15) Hito wa ninotugi de, mazu zibun ga kawaii. others second f i rs t self dear 'Others are second; f i rs t , self is dear.' "Self," which is opposed to "other" is not necessari ly fixed on the speaker. Zibun may also be the hearer, as in (16) - (18), or a third party, as in (19) - (21): 14 (16) Mazu zibun ga saki ni tabete miro yo. f i rs t self NOM before me eat try (imperative) '(You) try/eat first!' (17) Zibun no koto na-n-da kara zibun de denwa sinasai. 'self's' matter it is since self (in person) telephone (imperative) 'Since it is self's own matter, phone them yourself.' (18) [Basu ni nori-okureta] no wa zibun no sei bus to miss COMP TOP self 's (own) fault desyoo isn't it 'The reason you missed the bus is self's own fault, isn't i t? (19) Zibun bakari tabete iru wa. only eat -ing 'Only self is eating.' (20) Zibun ga yareba i i noni naa. self NOM do (provisional) it is good though l -wish '(I wish) self would do it.' 15 (21) [ N P [ S Zibun ga kaita] tegami] ni zibun ga henzi o kaite self NOM wrote letter to reply ACC iru wa. '(Look at him.) self is wri t ing a reply to the letter which self wrote.' The three zibun s in (22) form a "minimal trio" each of which refers to a different person: (22) a. Zibun ga siyoo. (1st person) w i l l do 'Self w i l l do i t : b. Zibun ga sinasai. (2nd person) do (imperative) '(Self) do it!' c. Zibun ga site iruyo. (3rd person) do -ing 'Self is doing it, isn't he?' (22a) is a vol i t ional expression which requires a 1st person subject; (22b) is an imperative which requires a 2nd person subject; and in (22c) -te iru yo implies that the speaker is an observer. With the help of these l ingu is t i c devices and/or a shared environment between the speaker and the hearer, the referr ing objects w i l l be c l a r i f i ed . Interestingly enough, this kind of deictic use seems to be controlled by a s i m i l a r pr inciple which dominates ko-so-a demonstratives wi th 16 respect to the "speech arena" formation. (Cf. Chapter 3 for Japanese Demonstratives.) To summarize, zibun in the referential use has the fol lowing properties: 1) Zibun, unlike the typical personal pronoun, is free from considerations of gender and honorific. 2) Zibun contrasts with "other" in the self-other opposition; 3) the "self" which is opposed to "other" can be shifted according to context; that is, zibun can be 1st person, 2nd person, or 3rd person. Now, zibun in a discourse bound case is assumed to show the restricted mode of the same use (the deictic use). In wri t ten discourse, zibun can be either a narrator who is the wr i te r himself or a narrator who is one of the characters whose internal feelings are described in the current discourse. In this sense, each sentence can be understood through i ts embedding in a higher abstract sentence such as (23): 7 (23) <NP wa> [ ] to omotta (thought) kanzita (felt) handansita (judged) mita (observed) etc. [NP = performer (wri ter /a character from whose point of view the story is told)] 17 However, these outer frames of sentences in (23) seldom, at least in a consistent manner, exist as l inguistic forms in actual discourse; they do not have real forms, except for the occasional realizations such as (24) - (27), when the discourse environment permits them. (24) Watasi wa omowazu [zibun ga ikeba yokatta] to I TOP involuntarily COMP sakenda. cried out 'I involuntarily cried out that self (I) should have gone in person. (25) Hanako wa [zibun ga sikkari-si-na-kere-ba] to ketui sita. TOP COMP decided 'Hanako decided that self (she) should brace (herself) up.' (26) Taroo wa [zibun mo ukauka site irarenai zo] to omotta. TOP COMP decided 'Taroo thought that self (he) should not be off selfs' guard.' (27) Sono otoko wa [zibun no ban ga kita] koto the man TOP [self 's turn NOM came] COMPUhe fact) o si t ta. ACC realized 'The man realized that self's (his) turn came.' 1 8 However, once sentences have the forms of (24) - (27) instead of (24') -(27') below, they possibly fal l into the domain of anaphoric use 8 just like the (28) - (28') pair does. (24') Zibun ga ikeba yokatta! 'I should have gone in person.' (25') Zibun ga sikkari-si-nakere-ba.. . 'I should brace myself up." (26') Zibun mo ukauka-site-irarenai zo. 'I should not be off my guard.' (27') Zibun no ban ga kita.' 'It's my turn.' (24) - (27) vs (24') - (27') are analogous to (28) vs (28'): (28) He is running! (referential) (28') John said that he was running (when he saw Jane), (anaphoric) In this section, zibun in the referential use has been observed in typical settings. The next section w i l l concern anaphoric use. §2.3 The Anaphoric Use of Zibun Zibun d iffers from English ref lexives in the fo l lowing two respects: i) long distance coreference (i.e., zibun does not observe clause-mate condition); ii) i ts antecedent is the subject of the sentence. 19 In this section, the coreference rule for anaphoric use of zibun w i l l be examined. As for the coreference in syntax, the notion "c-command" (29) is gaining significance in recent literature: (29) C-Command: Node A c (constituent)-commands node B if the branching node most immediately dominating A also dominates B. [Reinhart: 1976] S NP2 V NP i c-commands N P 2 N P 2 cannot c-command NP] Now, assuming that Japanese is a configurational language, 9 zibun also seems to observe the general coreference rule for pronouns (cf. Saito (1985)) which is given in (30): (30) Zibun cannot c-command its antecedent. Rule (30) defines the case where zibun is disjoint in reference as shown in (3 1). 20 NP2 PP V NP P 3 (i) NP3 is not eligible as the antecedent of NP 1 and NP2. (ii) NP2 is not eligible as the antecedent of NPi . Thus in (32) NP2 is zibun; then, NP3 cannot be i ts antecedent: (32) [s, Tarooj wa zibunj ni [s 2 Zirooj koso Hanakok NP, TOP NP 2 to N P 3 (emphatic) N P 4 ga suki nan n da] to tubuyaita]. NOM like i t - i s - tha t COMP muttered Taroo muttered to self that it is Ziroo, not me, who likes Hanako.' 21 In (33b) and (34b), zibun is not coreferential wi th Taroo since it does not observe Rule (30) (zibun does c-command Taroo ): (33)a. Tarooj ga [NP zibunj no kodomo] o sikatta. self "s child ACC scolded Taroo scolded self's child.' S NP VP Taroo-ga NP V Zibun no kodomo-o sikatta b. *Zibuni ga [NP Tarooi no kodomo] o sikatta. 'Self scolded Taroo's child.' S NP VP Zibun-ga NP VP Taroo no kodomo-o sikatta 22 (34) a. Tarooi ga [NP zibuni no ie o katta hito ] [ self 's house ACC bought person] o sagasita. ACC looked for 'Taroo looked for the person who bought self's house.' b. *Zibunj ga [NP Tarooi no ie o katta hito] o sagasita. 'Self looked for the person who bought Taroo's house.' Next, as is wel l -known, the antecedent of zibun is subject (subject-antecedent condition), zibun also observes Rule (35): (35) The antecedent must c-command zibun . In (36), Taroo does not c-command zibun; although zibun cannot c-command Taroo; thus, Taroo cannot be i ts antecedent (unless Taroo is mentioned somewhere in discourse and referred to by zibun in referential use): (36) * [ N P [ S Zibuni ga katte iru] inu] ga [NpTarooi no self SUB was keeping dog NOM Taroo 's neko] ni kamituita. cat to bit 'The dog which self was keeping bit Taroo's cat.' 23 Also observe (37a) - (37c): (37)a. *[NpTaroo no sigotoba] wa [Npzibun no kazoku] Taroo's workshop TOP self's family (NOM) sae hairenai. even cannot enter 'As for Taroo's workshop, even self 's family cannot enter.' b.* [NpTarooj no kazoku] wa [Npzibunt no sigotoba] ni hairenai. 'As for Taroo's family, they cannot enter self 's workshop.' c * [NpZibun, no sigotoba] wa [NpTarooj no kazoku] sae hairenai. 'As for self's workshop, even Taroo's family cannot enter it." In all three examples, Taroo does not c-command zibun because of NP nodes and all get Note that if one takes Taroo no kazoku instead of Taroo in (37b) as the antecedent of zibun, then it is grammatical wi th no violat ion of rules. Now in a configuration like (31), if NP3 is zibun, the potential antecedents are NP) and NP2 (Rule 30); and both NP 1 and NP2 c-command zibun in NP3 posit ion (Rule (35)). However, the subject-antecedent condition for Japanese ref lexives a l lows only NP1 to be a possible antecedent. Consequently, zibun in subject position as in (38a) is considered as a referential case while (38b) is anaphoric: 24 (38) a. Zibun ga ikoo. NOM shall-go 'Self (I) shall go.' b. Taroo wa [zibun ga ikoo] to itta. TOP said Taroo said that self (he) shall go.' Thus, Rule (30), (35) and the S-A (subject-antecedent) condition explain basic cases of zibun coreference in anaphoric use. Cases which do not fall in the domain of these rules and conditions are considered as referential. In the next section, the nature of these rules is examined in more detail wi thin the framework of universal grammar, to see how the Japanese zibun behaves therein. §2.4 Anaphor, Pronominal, and Variables §2.4.1 The Binding Theory The binding theory (Chomsky 1981) c l a s s i f i e s nominal expressions into three categories: anaphor, pronominal, and R-expression, and states the conditions for each category as in (39): (39) Binding Condition [p. 1 88] (A) An anaphor is bound in i ts governing category. (B) A pronominal is free in i ts governing category. (C) An R-expresssion (e.g., name and variable) is free. 25 The governing category is defined as in (40): (40) Governing Category [p. 1 88] a is the governing category for fi if and only if a is minimal category containing fi and a governor of fi, where a = NP or S. The notion "binding" is defined as in (41): (41) a is X (= A or A')-bound by fi if and only if a and fi are coindexed, fi c-commands a, and fi is in an X-position. [p. 184] Under the definit ion of the governing domain in (40) and the binding conditions in (39), zibun is considered neither as a anaphor (cf. the long-distance coreference) nor as a pronominal (cf., the subject-antecedent property). This conflict w i l l be pursued in the next section. §2.4.2 Governing Domain Zibun is distributionally different from English reflexive: zibun can occur in the subject position and possessive case, whereas English reflexives cannot. 1 0 In both of these cases, zibun cannot be an anaphor, since they are free in governing categories if a possessive bearing NP as wel l as S is taken as the governing category (one possible assumption about such NPs). 26 Because of a number of d i f f i cu l t ies involved in i ts formulation, Huang (1982) proposes a modif icat ion of the governing category (40), which is given below in (42): (42)a. Governing Category [Huang, p. 337] a is a governing category for fi if and only if a is the minimal category containing fi, and a S U B J E C T which, if fi is an anaphor, is accessible to fi. b. SUBJECT is AGR (agreement) or a subject of NP and S. c. "Access ib i l i t y " is defined as fol lowing (Chomsky 1981, p.21 1 ff.): (i) a is accessible to fi is if and only if in the c -command domain of a and assignment to fi of the index to a would not violate (ii). (i i) *[y..5..], where y and 8 bear the same index (/ wi thin /' condition). If a S U B J E C T is accessible to fi (= an anaphor or a pronominal), then an anaphor and a pronominal share the same governing category; if a S U B J E C T is not accessible to fi, then an anaphor and a pronominal have different governing categories since the access ib i l i ty is irrelevant to a pronoun. Thus, wi th fi in the subject (= S U B J E C T ) posit ion as in (43), and governed inside a , 1 1 if fi is an anaphor then a is not i ts governing category since an anaphor cannot be i ts own subject ( / w i t h i n /' condit ion), whereas if fi is a pronominal then a is i ts governing category. 27 (43) [a fit Ii Given a sentence like (44), if zibun is an anaphor, S is the governing category instead of NP and it is bound in the S; if zibun is a pronominal, NP is i ts governing category and it is free in the NP. (44) [s Tarooi ga Hanako o [NP zibuni no ie] de sekkyo-sita.] NOM ACC self 's house at lectured 'Taroo lectured Hanako at self's house.' In either case, Taroo = zibun reading is obtained. The identity of zibun is obscure, but the coreference is s t i l l accounted for. However, this type of analysis is not free from problems both for the anaphor hypothesis and the pronominal hypothesis. Suppose zibun is an anaphor, the analysis cannot explain the zibunj reading in (45) since NPj is outside of its GC (governing category): (45) a. [s 3 Tarooj ga [s 2 HanakOj ga [s] zibunj/j ga tadasii] to NOM NOM NOM right omotte-iru] to omotta]. is-thinking thought 'Taroo thought that Hanako was thinking that self was right.' 28 b. [s 2 Yamadai ga [s, Matidaj ga [ N P ziburii/j no kuruma] NOM NOM 's car de ziko-o-okasita] to.itta]. wi th had-an-accident said 'Mr. Yamada said that Mr. Matida had an accident by self's car.' In (45a), since zibun in subject position is not an accessible subject, S2 instead of S i is GC and Taroo is outside of G C . 1 2 In (45b) Sj is GC and Yamada is outside of G C . 1 3 If zibun is a pronominal, on the other hand, the problem of GC is not a problem any more since a pronominal is free in GC but may (or may not) be coindexed with NP outside of GC. Then, the fact that zibun is always coindexed with the subject becomes a very peculiar character is t ic for a pronominal controlled by condition B in (39). (Sportiche's (1986) explanation for this is that the long distance coreference property of zibun is that of a bound pronoun (i.e., zibun is either an anaphor or a bound pronoun). I w i l l pursue this point further in § Now, the possible resolution for the problems of the governing category w i l l be found in Manzini and Wexler (1987) and Wexler and Manzini (1987). In Manzini and Wexler and Wexler and Manzini, five values of parameter for GC (and two values of parameter for the proper antecedent) are set up.in universal grammar. 29 (46) GC parameter [M & W: 1987a, p. 53] 7 is a governing category for a iff y i s the minimal category which contains a and a. has a subject, or b. has an INFL, or c. has a TNS, or d. has an indicative TNS, or e. has a root TNS. (47) Binding [ibid, p. 64] a. An anaphor is bound in its governing category by a proper antecedent. b. A pronominal is free in i ts governing category from proper antecedents. (48) Antecedent parameter [ibid, p. 64] A proper antecedent for a is a. a subject fi, or b. an element fi whatsoever. In Japanese, the value (e) w i l l be selected in (46) for GC and (45) are nicely accounted for as anaphor cases. In the next section, this anaphor solution is further investigated in connection with topic-antecedent cases. 30 §2.4.3 Topic-bound vs. Subject-bound § Topic-Antecedent The topic sensit ive property of zibun has been pointed out by Kuno (1 973), Matsuda (1 975), and Kitagawa (1 981), among others. First of a l l , i t is important to distinguish two types of topics, i.e., Topic 1 and Topic 2. For Topic 1 the topic NP is associate wi th the gap in S. (49) TaroOj wa [s ei Hanako o syookai sita] TOP ACC introduced 'As for Taroo, (he) introduced Hanako.' (50) Hanakoi wa [s Taroo ga e, syookai sita] TOP NOM (ACC) introduced 'As for Hanako, Taroo introduced (her).' In (49), the topic corresponds to the subject; in (50), the topic corresponds to the object. (The "e" is either a pro or a trace in Saito (1985); the "e" is a trace in Kuroda ((1988) and elsewhere).) Note that wi th Topic 1, topicalizat ion does not affect the coreference relation of the subject-antecedent condition, as can be seen in (51): (51) a. Taroot ga Hanako o zibun, no nooto de butta. NOM ACC self "s notebook INS hit 'Taroo hit Hanako with self's notebook.' 31 b. Taroo-j wa [s ei Hanako o zibuni no nooto de butta.] TOP (NOM) ACC 'As for Taroo, (he) hit Hanako with self's notebook.' c. HanakOj wa [s Tarooi ga ej zibunj/*j no nooto TOP NOM (ACC) de butta.] 'As for Hanako, Taroo hit (her) wi th self's notebook.' In (51), zibun is coreferential w i th subject NP, which is in turn coindexed w i t h topic NP; as a result , Taroo and zibun obtain a coreference relation. This does not hold for (51c) since topic NP is coindexed wi th the object, which cannot be coreferential wi th zibun. Also observe (52): (52) Tarooi wa [zibun*j ga ei itiban yoku sit te-iru], TOP NOM very wel l know 'As for Taroo, self knows him very well . ' Zibun in (52) is only interpreted as referential. This subject-bound property is observed in a relative clause case like (53a). In an example like (53b), the ambiguity of zibun is nicely accounted for: 32 (53)a. [NpCsTaroo-j ga ej ziburij/*j no ie de fssyo ni benkyoosita] home at together studied HanakOj] 'Hanako whom Taroo studied wi th at self's home' b. Tarooi wa [ 5 ej [NP ts ej zibun ,/j no kuruma 0 aratte-iru] In (53b), ej is coindexed wi th the head noun of the relative clause and e, is coindexed wi th topic, and unlike ej in (53a), each e in the subject positions is coreferential wi th zibun. Therefore, in the case of Topic 1, the subject-antecedent condition and not topic-antecedent, s t i l l holds. The situation is different in the case of Topic 2, where there is no corresponding argument position of the verb, as in (54) and (55): (54)a. Tarooi wa [zibunj ga tyokusetu soko ni itta.] TOP car ACC is-washing otokoj] 0 nagamete-iru]. man ACC is-watching 'Taroo is watching a man who is washing self's car.' TOP NOM in person there went 'As for Taroo, self went there personally.' b. Tarooj wa [zibuni no otooto ga itta.] TOP 's younger brother NOM went 'As for Taroo, self's younger brother went.' 3 3 c. Taroo, wa [[ziburij ga katte-ita] inu] ga byooki ni natta. TOP self NOM kept dog NOM sick became 'As for Taroo, the dog which self was keeping became sick.' ( 55 ) a. Tarooj wa [itiban si ta no musume]j ga zibunj/*j to TOP 2 youngest daughter NOM self wi th dookyosite iru. semantically anomalous) live -ing (zibun = Taroo ) 'As for Taroo, his youngest daughter lives wi th self.' b. Tarooj wa hahaoyaj ga [zibun,/j no zaisan] o TOP 2 mother NOM self's property ACC kanrisi te-iru. manage -ing (zibun = Taroo/hahaoya ) 'As for Taroo, his mother manages self's property.' There are cases in which the establishment of coreference is apparently d i f f i c u l t . 1 4 The examples in (56) are such cases: (56) a. ?Taroo wa Hanako ga zibun to kekkon sita. TOP 2 NOM with married 'As for Taroo, Hanako married self.' 34 b. ?Taroo wa Hanako ga zibun o syookai sita. ACC introduced 'As for Taroo, Hanako introduced self.' These, as a matter of fact, are good i l lustrat ions of the nature of Topic 2. With Topic 2, whether the connection between the topic and the rest of the sentence is appropriate or not is determined pragmatically. In isolation, these sentences have low acceptability because of the lack of pragmatic informat ion 1 5 . The coreferential relationship between zibun and Taroo in these cases is not impossible, simply open. Therefore, the statement that the zibun coreference is sensitive to topic means that it is sensit ive to Topic 2. The question then, is what is the nature of topic-antecedent as opposed to subject -antecedent. In the fo l lowing , I w i l l examine Topic 2 and i t s coreferential relationship to zibun. According to Kuno (1973, p62ff) Topic 2 is created through a process of "Subject ivizat ion" from either NP-no (possessive) or NP-n/ (locative) as shown in (57) and (58). (57)a. John no otoosan ga sinda. (=Kuno's (23)) 's father NOM died 'John's father died.' b. John ga otoosan ga sinda. (Subjectivization) NOM NOM c. John wa otoosan ga sinda. (Thematization) TOP 35 (58) Bunmeikoku no dansei ga heikin-zyumyoo ga c iv i l i zed countries 's male 's average l i fe span NOM mizikai . (=Kuno's (27)) is-short 'It is the average l i fe span of men of c iv i l i z ed countries that is short.' Kuno claims that sentences such as (59) (=Kuno's (18) (1 973, p25 1)), and (56) above, which do not have a corresponding NP-no or NP-m sentence, are base generated. (59) a. [sSakana wa [stai ga ii.] fish TOP red snapper NOM good 'As for fish, red snapper is good.' b. [sHana wa [s sakura ga ii.] blossoms TOP cherry-blossom NOM good 'As for flowers, cherry blossoms are the best.' On the other hand, Kuroda (1986 and elsewhere) labels Topic 2 as the "major s u b j e c t " 1 6 and argues that unlike Topic 1 sentences (for which Kuroda has maintained the movement analysis since 1965), these sentences are base-generated and get the nominative case marker ga in pos i t ion , then are moved under S' pos i t ion jus t l ike Topic 1. Consequently, Kuroda considers examples l ike (60) (given in Mikami (I960)) as a separate category of wa sentences. 36 (60) a. [NpSinbun o yomitai] hito wa/*ga koko ni newspaper ACC read-want person TOP here in arimasu. exist 'As for those who want to read newspapers, they are here.' b. Basyo wa/*ga okunai-setu ga attooteko datta. place TOP indoor-theory NOM predominant was 'As for the place (of the murder), the "indoor" theory was predominant." Now back to the topic of zibun coreference. If the cases where zibun is coreferential wi th Topic 2 are considered as pragmatically determined, then the ungrammaticality of the Topic I case in (51c) (repeated here as (61)) w i l l be left without explanation, since the functional difference between Topic 1 and Topic 2, i f any, is far from clear in terms of zibun coreference. (61) . HanakOj wa [sTarooj ga ej zibunj/*j no nooto de butta.] TOP NOM (ACC) 'As for Hanako, Taroo hit (her) with self's notebook.' On the other hand, if this coreference is considered a syntactic phenomenon, it is puzzling why both subject and topic are chosen as the antecedents. I shal l pursue the la t te r case further under the investigation of the nature of topics in syntax. 37 § Topic vs. Subject F i rs t , let us look at the previously proposed topic structures. (62) gives a summary of Kuno and Kuroda's analyses introduced in the previous section. (62) a. Kuno's Analysis (1 973)...base-generation Ex. [sSakana wa [stai ga ii.]] (cf (59)) b. Kuroda's Analysis (1986)...movement Ex. [s'Sakanaj wa [stj [stai ga ii.]] On the other hand, Saito (1985) suggests two derivations for topic sentences as in (63). (63) a. Tarooj wa [sHanako ga ei suki da.] 'As for Taroo, Hanako likes him.' b. [sTarooi wa [sHanako ga tj suki da.]] In (63a), the relationship between N P - w a and S is "licensed" by the "aboutness relation" (Saito 1985, p.287) Hoji (1985) further assumes that the base-generated NP-wa is Kuno's "thematic" topic and NP-wa created by movement is Kuno's "contrastive" topic as shown in (64) (=Hoji's (73)). (64) a. "Thematic" Ex. [ S " N P r w a [ s-[ sNP-gf3 [ v p ei V]]]] b. "Contrastive" Ex. [ s N P r w a [ s N P - g f 3 [ V P ti V]]] 38 Hoji ident i f ies the "thematic" topic as the E(xpression) of Banfield (1973). Now, evidence in favor of the s tructure (64a) rather than (62a)/(62b) for Topic 2 comes from examples such as (65). (65)a. Taroo wa kodomo-tati wa mina dokuritu sita. T0P2 children TOPI all be-independent 'As for Taroo, as for his children, (they) all get up on their own.' b. Taroo wa otootoi wa e\ kodomo o T0P2 younger brother TOPI child ACC daigaku ni yaranakatta. university to did-not-send 'As for Taroo, as for his younger brother, (he) did not send his child to a university.' c. Taroo wa kodomoj wa Hanako ga ei sodatete-iru T0P2 child TOP1 is- ra is ing (e.g. Hanako is his ex-wife) 'As for Taroo, as for his child, Hanako is raising (him).' Since Topic 2 precedes Topic 1 (S/S 1) in these examples, the following structure (66) can be suggested. 39 (66) Topic 2 S'/S S Now, using the recent version of X'-theory described by Chomsky (1986) which fac i l i ta tes the comparison of structures despite their apparent diversity, (66) may be represented as (67). (68) It is [cpJohn [cpwho left.]] Assuming (67) is the correct structure for Topic 1 and Topic 2, the posit ion occupied by Topic 2 is that of a smal l -c lause s u b j e c t 1 7 as possibly assumed in English cleft constructions such as (68); while Topic 1 under CP(=S') instead of IP(=S), as suggested by Kuroda (1988), is in the specifier 's position and is not a subject in the same sense of Topic 2. This explains why NP-wa , which is under S', by Kuroda's movement analysis, and binds the object, cannot be the antecedent in (61). Furthermore, this gives a consistent view for the syntact ic structure of zibun coreference wi th the subject-antecedent condition. (67) CP 40 Although I have to leave the details of this analysis for future research, I assume this as one possible syntactic explanation. In the next sect ion, another p o s s i b i l i t y for the var iable interpretation of zibun is examined for syntactic topic coreference. § Zibun as a Variable The claim that zibun (and null pronouns, but not overt pronouns) can be a variable has been made several times in the l i terature (Hoji (1982), Saito and Hoji (1983) among others). Saito and Hoji (1983) claim that the difference between (69a) and (69b), which have the configuration shown in (70), is due to the fact that zibun exhibits a weak crossover effect while kare does not. (69) a. [s Jiroj-o [s [NP Hanako-ga kare|-o kirat teiru koto]-ga ACC NOM him ACC disl ike fact NOM [ V P t i yuuutuni siteiru]]]. depressed make ( = S & H ( 1 5 b ) ) '(Lit.) J iroj , the fact that Hanako disl ikes himi has depressed.' b* [s J i r o r o [s [NP Hanako-ga zibunj-o kirat teiru koto]-ga [ V P tj yuuutuni siteiru]]]. (=(15c)) (70) a. QOj [ kare, tj ] b* QOj [ zibunj tj ] (weak crossover) (QO: quasi-operator = scrambled NP) 41 If zibun is a variable and kare is not (which is assumed to be true), (70b) is accounted for as a result of violat ing the Biject ion Principle (Koopman and Spontiche (1982)), which prevents an operator from binding more than one variable. This assumption leads to the stipulation given in (7 1): (7 1) When zibun is locally A'-bound, it must be construed as a variable. (Saito & Hoji, p. 255) However, Saito & Hoji eventually claims that instead of the Biject ion principle, which requires many st ipulations to make it work in (70), (70b), which exhibits a weak crossover effect, is better explained by the general principle given in (72): (72) A variable cannot be antecedent of a pronoun or an anaphor that it does not c-command, [p. 256] A stipulation such as (71), in fact, has problems. Observe (51b), repeated here as (73): (73) Tarooi wa [s ei Hanako o zibuni no nooto de butta.] TOP (NOM) DO 'As for Taroo, he hit Hanako with self's notebook.' Suppose (73) is derived by movement as Kuroda and Saito suggest. Then since zibun is locally A'-bound by Taroo which, in turn, binds the trace, the sentence violates the Biject ion Principle. However, in real i ty the sentence is grammatical. If zibun is an anaphor, i t has to be A-bound; i f zibun is a variable, it has to be A'-bound. Since Topic 2 is in the A'-posi t ion, if (7 1) is a correct generalization, zibun has to be a variable. But this is 42 not the case. This conclusion indirectly supports the possibi l i ty of the alternative "subject" analysis for Topic 2 presented in the previous section. In the next sect ion, I w i l l examine how discourse affects coreference. §2.4.4 Zero-Topic and Referential Use Huang (1982 and 1984) claims that the empty topic binds the object EC (empty category) in Chinese. (74) [ topSi ] , [Zhangsan shuo [Lisi bu renshi ei] say not know '[Hirrij], Zhangsan said that Lis i didn't know ei.' [Huang (1984), p. 542] He also claims that this type of zero-topic binding is only allowed in a "topic-prominent" language like Chinese. In fact, the topic NP deletion observed for Chinese in Tsao (1977) is s t r ik ingly s imi la r to Japanese N P - w a deletion (cf. Mikami (1960), Kuno (1978) and Terakura (1984)). Huang's zero-topic is motivated by the "topic-chain" phenomenon in discourse as shown in (75): (75) [[TopiCj] ] [ [ ei ] ] [ [ e, ] ] [ [ ei ] ] Huang assumes that this coindexing in (75) can be done in LF' module fol lowed after L F 1 8 (cf. (2) in Chapter 1). That is , in Huang's inter-pretive framework, the coindexing in syntax provides the coreferential 43 relationships in a sentence and the coindexing in the type above provides information for a sentence to be interpreted in discourse. Now in Japanese, (13) in §2.6 also provides an example of zero analysis for Topic 1. In examples like (76), let us also assume that they have a zero topic which is Topic 2 as in (77); then zibun is, in fact, not referential but anaphoric and coindexed (in syntax) wi th Topic 2. (76) a. zibun ga zikani soko ni i t ta NOM in person there to went 'Self went there in person.' b. Taroo ga zibun no kodomo o syuzyutu-sita NOM self 's child ACC operated Taro operated self's child.' (77) a. [ej] [zibunj ga zikani soko ni itta.] TOP 2 b. [ej] [Tarooj ga zibunj/j no kodomo o syuzyuti-sita.] TOP 2 The problem is that unlike Topic 1, since Topic 2 is an extra-NP, it is not clear whether a sentence has a zero Topic 2 or not. The discourse concept of a 'global topic' could offer one possible solution for this. Assuming that "global topic" (discourse topic) is what a discourse is about, while "local topic" (Topic 1 and Topic 2) is what a sentence is about, i t is observed that the "global topic" and Topic 2 have s imi la r functions; i t has often been pointed out that Topic 2 has a discourse function s imi l a r to such expressions as Taroo to ieba, "speaking of 4 4 Taroo," Taroo no koto da keredomo, Taroo no ken ni tuite da ga ne, "talking about things concerning Taroo." Then, is i t the case that all referential cases are, in fact, anaphoric ones and bound by Topic 2? To see this point, f i rs t compare examples (78) and (79b). (78) and (79b) are identical, except that i t is apparent that the subject-antecedent condition is violated in (79b), while it is not in (78). (78) Taroo ga Hanako o [zibun no tomodati no mae] de self 's friend in front of LOC home-hazimeta. start to praise 'Taroo started to praise Hanako in front of self's friend.' [Zibun = Taroo] (79) a. Sorekara Hanako ni nani ga okotta no. after that EXP what NOM happened Q 'Then, what happened to Hanako?' b. Taroo ga Hanako o zibun no tomodati no mae de home-hazimeta no sa. 'I w i l l te l l you, Taroo started to praise Hanako in front of self's friend.' [Zibun = (Taroo)ZHanako } Suppose we account for this antecedent shift phenomenon 1 9 in a fashion parallel to the zero topic analysis. That is, Hanako is the global topic in the mini -d iscourse (79) and is coindexed wi th the Topic 2 in 45 discourse interpretation as shown in (80) and zibun then is Hanako. Notice here that the subject-antecedent condition is not violated, since Hanako (= zibun ) is the Topic 2 but not the direct object (which is also Hanako). (80) Hanako -[Discourse Information] Hanako-o zibun no home-tomodati hazimeta no mae-de However, (79b) is not exactly a Topic 2 sentence. Example (81), which is in fact (79b) wi th the NP overt, is, in fact, ungrammatical. (81) * [s-Hanako wa [sTaroo ga Hanako o zibun no tomodati mae de homehazimeta.]] Now let us replace (79b) with (82a). (82) a. Taroo ga kanozyo/e o zibun no tomodati she no mae de homehazimeta no sa. b. [ei [Tarooj ga kanazyoj/ei o zibunj/j no tomodati TOP 1 no mae de homehazimeta no sa.]] 46 If (82a) has the structure of (82b) with/without the resumptive pronoun kanozyo, then zibun is coreferential wi th the zero topic which is, in turn, coindexed wi th the discourse topic Hanako. However, as Kuno (1978, p104) reports, it is rather rare, though not impossible, that a non-subject NP in the preceding discourse occurs as a zero topic. Also, (82b) is a counter example to cases such as (61) which does not allow coindexing between Topic 1 and zibun20. Therefore, although Topic 2 has a function s i m i l a r to that of the discourse topic, (79b) does not necessarily have Topic 2 as the zero form. The same thing can be said of pairs (24-27) vs. (24'-27') in relation with the performative analysis in terms of the abstract "localization" of discourse function. Let us return our attention to the data (4) in §2.6 (the relevant portion repeated here as (83)). (83) Johnj wa [Np[sono ie ni zibun*j o kakumatte-TOP 1 that house in self ACC sheltered lureta] Bi l l ] o uragitta-nodaroo. ACC betrayed-would-have 'Perhaps it was the case that John betrayed B i l l , who sheltered self in that house.' Zibun in (83) is not John but Mary, which has occurred earl ier in the previous discourse. Mary is the protagonistic character and from her point of view the story is narrated. Thus, the referential use of zibun, which is identified wi th the narrator and/or the protagonistic character can override syntact ic 47 coreference as in cases such as that shown in (83) (John* zibun), where the syn tac t ic antecedent does not coincide w i t h the narrator / protagonist. There are also discourses which do not have part icular narrators (cf. Kuroda's reportative and non-reportative dis t inct ion in Kurado (1973b)). In connection wi th this topic, I w i l l examine two key notions which, although they are not syntactic concepts, are relevant to zibun coreference. §2.5 Zibun and Kuno and Kuroda's Extra-Syntactic Notions Kuno (1973) claims that the notion of "awareness" plays the key role in determining the choice between zibun and kare 'he', as he states in (84): (84) Zibun in a constituent clause (A) is p r e f e r en t i a l wi th a noun phrase (B) of the matrix sentence only if A represents an action or state that the referent of B is aware of at the time it takes place or has come to be aware of at some later time. In other words, zibun appears in subordinate clauses only when the clauses represent the internal feeling of the referent of zibun and the f i r s t person pronoun appears in i ts place in the direct representation of the internal feeling, [p. 322] In Kuno's analysis , apparent free var iat ions of "reflexive" zibun / "pronominal" kare in cases like (85) - (87) are not free variations as far as an ex t r a - syn tac t i c level is concerned. In other words, the dis t r ibut ion of those two is dominated by an extra-syntact ic factor, which is "awareness." 48 (85) Taroo wa [Ziroo ga zibun/kare o nantonaku TOP NOM ' se l f /h im ACC somehow ^ sakete-iru to] omotta. be-avoiding COMP thought 'Taroo thought that Ziroo was somehow avoiding him.' (86) Taroo wa [kyoodai-tati ga zibun/kare o TOP siblings NOM self /him ACC tayor i -n i - s i t e - i ru koto] ga wazurawasikatta depend-on COMP NOM was-annoying 'Taroo felt that it was annoying that his siblings depended on him.' (87) Taroo wa [Nptteki no supai ga sirokuji-chu zibun/kare TOP enemy 's spy NOM all the time ' s e l f /h im o mihatte-ita] toki] mo fudan to kawari-nakatta. ACC was-watching when too usual was-the-same 'Taroo was his usual self even when the enemy's spies were watching him.' As a native speaker myself, I certainly detect that the choice of the use of these two controls, say, the nuance of the utterance. The problem wi th a notion like "awareness" is the diff iculty in defining what exactly it means or even in proving that it exis ts , although in tu i t ive ly the concept does not seem to be far off what native speakers sense in examples like (85) - (87). 49 Kuroda (1973 a and b) deals wi th the same problem. Kuroda (1973a) argues that, contrary to Kuno's claim of "awareness" factor in zibun coreference, the following example (88) (= (5) in Kuroda, p. 139) does not involve "awareness": (88) Oedipus wa Jacasta ga zibun o unda ie de ima wa bore house now kodomotati to koohuku-soo-ni kurasite-imasu. children wi th happily l iv ing- i s 'Oedipus now lives happily wi th his children in the house where Jacasta bore him.' At this point, Oedipus did not have the sl ightest idea of the fact that Jacasta had borne him. Thus, Kuroda claims the difference between reportive style and non-reportive style determines the choice between zibun/kare in cases like (89): (89) Johnj wa Bi l l j ga zibunj/j/karej o hometa toki Mary no soba ni ita. 'John was by Mary when B i l l praised himself.' According to Kuroda, if the story is told from the narrator's point of view (omnipresent but not omniscient) kare is used (reportative); on the other hand, if the story is told from the subject's (= John in this case) point of view, zibun is used (nonreportative). (In either case/- reading is intact.) Thus, zibun is used in (89) not because John is aware of the event but because John's viewpoint is employed. 50 Kuno's "awareness" and Kuroda's "viewpoint" do seem to be related to the "self-other" opposition scheme of zibun in §2.2. That is, in (85)-(87), the events are viewed from Taroo as the center of consciousness and the self 's awareness of the events naturally follows from the "self-centricity". To extend this discussion any further would take us too far from the main point of this chapter. However, the "self-other" opposition schema, which I assume to be the essential characterist ic of the use of zibun, is not a sudden invention of any sort. The Japanese language has often been called an "epistemological language" (Kuroda 1973; McCawley 1978) which l ingu i s t i ca l ly different iates what is related to "I" and what is not. The concrete examples given in the literature quoted above are in the following form: (90) a. (Watasi wa) soo omou / kangaeru / s inziru / omotte-iru / kangaete-iru I so believe 'I believe so.' b. (Watasi no yuuzin wa) soo *omou / *kangaeru / *s inzi ru / omotte-iru / kangaete-iru my friend believe 'My friend believes so.' According to Akatsuka, the distinction between "my" belief and someone else's belief (which "I" can never know for sure) creates this difference. The usage of some adjectives is another case, which is mentioned in Kuroda. An example of this is: 51 (9 Da. Watasi/*Taroo wa atui. hot 'I am hot.' b. *Watasi/Taroo wa atu-gatte-iru. hot-feel-(being in a certain state) 'Taroo is hot.' Such epistemological characteris t ics are a strongly dominating factor in Japanese, which can be extended to broader aspects of the language than normally thought, possibly including the in-group/out-group dist inct ion in the honorific system and the principles governing the use of the koso-a-do demonstratives (cf. Chapter 3). What is noteworthy in both Kuno's and Kuroda's hypotheses is that the subject NP which is independently defined in grammar is also the si te of manifestation of such conceptions as "awareness" and "view-point," which have extra-syntactic natures.. Now, to review the posi t ion, there are two uses of zibun: referential use and anaphoric use; the former is pragmatically bound and the latter is syntact ical ly bound. Every zibun is potentially ambiguous. Then the s tructural information makes a certain prediction, namely possible and impossible coreference relationships. The contextual information provides who is ident i f ied w i t h zero topic, who is construable as the performer of abstract performative frame, who is the narrator, and the like. Although the final decision on the choice of the antecedent involves complex pragmatic factors, there is now a series of 52 candidates for the most l i k e l y antecedent, e i ther s t ruc tura l or contextual, for zibun. In the fol lowing section, I w i l l analyze some discourse data drawn from texts to test the foregoing predictions. §2.6 Examination of Discourse Data [Data] (1) . . ^ l i : i t M A m t i : i t ^ A ^ * . J : a H o ^ A \ iftlHE ~>tzmoxn. mi>ifi&ftoit^ss^oit. a & MK). S # o £ v S £ f I 9 . U ^ D o S J f c , i * c o ^ * f c o v ^ T S o f c . ( R ® SO • 129) (2) . . . I ^ M ^ ^ W ^ A ^ t f c K ^ H ^ t u ^ . #£-MlTfc£ KM*. &mz%^mzLfri>. m*zt>Ltz. {Fimmnbiph • 123) (3) * « 5 £ < A ^ & l * & i f c j S £ a < fcW£&#*£. ^ ^ J i t ( r a ^ M ^ ^ • 141) 53 (4) N a r / m - w a henzi-o mat-te tyuuibukaku zibunm-r\\ sosoi-de i - ru 6///|<-no me-kara kao-o so ras i - t a . B/'//|<-no g iwaku-ga zibunm-r\i muke-rare- te i - r u koto, z/fcunm-ga otooto-no John\-r\\ yar -ase- ru dake no r iyuu-ga ar-u koto, kotoni otooto-ga keisatu-ni ima tukamatte-i-ru koto - korera-no kan-gae-ga subayai hirameki-de Marym-no atama-no naka-o tuukasi-ta. Osoraku (dareka-ni tanom-are-te) John\-wa sono ie-ni zibunm-/kare\-o kakumat-te kure-ta Billy-o uragit- ta no dar-oo. Isi no yowai John\-no koto dakara. Sikas i zibunm-wa mukankei da. 'Marym turned away from the stare of Billy,, who was watching 5ELFm attentively and wait ing for an answer. That Billys suspicion was turned on 5ELFm [i.e., that Billy suspected SELFm], that 5ELFm had a good reason to have [herm] younger brother John\ do that, especially that [herm] brother^ was in the custody of the police - these ideas rushed through [herm] mind. Perhaps i t was the case that (being asked to) John\ betrayed Billy, who sheltered SELFm/him\ in that house. Because Johns was infirm of purpose. But 5ELFm was not a party to that.' (Oshima's original.) (5) E S T * ' ^ X ^ t z & ^ m ^ k 9 fcS ^mtz^tcb. —&i,zi&\,\x&i>fstff^fr'f(izmtztiLV. s a i g a s X%< g L t z ) & ± S # W;^ftfrfc - ^ - i £ < H t M f &LX*t:frt>mt>'J?l,&r>X hit. t>K ISI t % < X\Z S b h b . o O ' S r V ^ f f i t t S r S f c U f c . < f t » - 1 4 1 ) tfit:. *ixi**&^tz&&ifimL&#<nm^*x&zmzm^m&tz fc-S^##&;L'?:££*5&L,T'b*. ^ fctfpfc "Cft & . (ftIff • 245 ) 246 ) (8) 4*tiJK#<aE#0#5<a#£a»&a»jEo/£. (10) r j / ^ t ^ j fcS^lifflfrtsifc. r^Lfr-tZtmfrZ? i bSftttTK-ytz. g ^  ® a a ^  (2) <7) BR O 3 t e ^ & J I * » f l i * £ . ^ s & S & A l ^ f c t e l i l l S f t f c & c T ) ^ . * $ 8 £ 5 £ t i ftkr&M^fc^'^T'te&v^ ^ j g j ; 0 fc&*Lji.&ib*>Ji&fcf^SE ^ - y a ^ ^ W M -96) (12) A f B f c V ^ t f . - # T < O j & ^ \ %*->X^r> I, *(iZm mbxvt^tzhL^Xo (t'lt^) (13) u ^ u a f r i B i m ; : . T^m^xma. &M&Z.il&V&&i:mt& coi>cr>t LXxKfeL. Z o Ltifo%.l,zmt>ti. % a L tz&fe coil cr> JiX ± # T v ^ < 0 ^ f f c ^ 3 ifc*SJBi-6. fo&MVkXte. Ziiififrtitoft » £ 3 g * f L T V ^ f c * ) * ! . i# i^A£fc&£frt>fc 'Otf>B W f c & ^ S i ^ T v^S*»^0. fc £ L X a t t ) l T \ i 9 o o & o l * ^ f l i ( : 5 : ^ 3 t *R1Mtf)ff i i$®vM fc IX LfrMmtihtzZ a . **ifc44J: o t i n f c = f i r s « a ^ « ^ < r fc^'-c^ -so^fc . - c ^ s . $:5t3:fc*i&:K3fifc:* a S o t v ^ ^ o ^ f c Ltufo. R i S f t e i f c . £_ it*>J3B<*>Atf fc a n ^ ^ T ^ S 3 i # f c |S] t**31fc:& & Atf fc*H*rtfv*-0*S v ^ v ^ & f t A J l - C f c * S^tiS A^conf fc a $ S * . T V * ( ^ £ 3 5 : A H • 83-4) [TEXTS] Shiga, Naoya: "Wakai," in Gendai Ninon Bungaku Taikei 34 , Tikuma Shoboo, Tokyo. Shimizu, Tooru: Shiisyuposu no Shinwa, by A. Camus, Sintyoo Bunko, Tokyo. Tsu j i , Kunio: "Enkei-Gekizyoo kara , " in Kita no Misaki, Sintyoo Bunko, Tokyo. 56 [The Skeletons of the Relevant Sentence Structures in the Data] 1. [Watasi ga atta kagiri de wa daremo ga NOM met as far as TOP everyone NOM zibun no koto sika hanasanakatta. self's matter (ACC) but did not speak. 'As far as I know, everyone spoke of nothing but self's own matter.' 2. [[rel-cl] otoko ya [rel-cl] onna-tati] ga [zibun no mae] man and women NOM s e l f s front o mitumete ita. ACC were gazing at 'The man who...and the woman who...were gazing in front of selves.' 3. Watasi ga...sore o [zibun no siya] ni I NOM it ACC self s' field of vision LOC tutumi-konda. put into (put things in perspective) '...and I put things into perspective.' 4 [cf. Data] 5. Ti t i j wa ei zibun ga wakaranakatta. father TOP ' s e l f NOM (object) could not recognize 'My father could not recognize self.' 57 Ti t i j wa ei [sobo ga [zibun no uti] e kuru] TOP [grandmother NOM self s' home GO come] koto o iyagatta. COMP ACC hated. 'My father was unwill ing for my grandmother to come to self's home.' Tumai wa ei [zibun no atama] o monda. wife TOP self 's head ACC massaged 'My wife massaged self's head.' [ ] to zibun ga itta. COMP self SUB said 'Self said that...' Zibun^ wa ei [Np[zibun2 no zutuu] ga naotte iru] no] self i TOP [ 'self2 ' s ' headache NOM has gone] ni kizui ta to noticed 'Self noticed that self's headache was gone.' Zibun1 1 wa ei [Np[zibun2 no me] ga kagayaku] no] self 1 TOP [ s e l f 2 's eyes NOM light up] o kanjita. ACC felt 'Self felt that self 's eyes lighted up.' 58 11. Eien yori mo karej wa ei [zibun no yuuki to ronsyoo eternity-than he TOP [self's courage and proof no hoo] o aisu. (rather)] ACC love 'He loves self's courage and proof more than eternity.' 12. Taroo to ieba, otooto ga [zibun no sinyuu] speaking of Taroo younger brother NOM self 's friend to iede site simatta rashii yo. with has run away from home seems you-know 'Speaking of Taroo, his younger brother seems to have run away with self's friend.' 13. [ej[ei[Np[Np[s[Np[Zibuni no syuui] no hito-bito] ga idaite iru] [[self's surrounding people] NOM have] [Npsamazama na sinnen]], [ ]] demotte [NP...[Npkatei]] o [various kinds of beliefs,] wi th supposition ACC sasaete iru]]. support -ing '0 supports this supposition [that...] wi th various kinds of beliefs which those around self usually have and...' In (1), zibun is daremo. Daremo is the c-commanding subject. Thus, i t sa t i s f i es Rule (35) and the subject-antecedent condition. Furthermore, s ince daremo is a quantified NP, this is a typical syntact ical ly bound (bound variable) case. Notice that watasi is in the 59 adverbial clause, thus it cannot c-command zibun. The referential use of zibun is also possible. However, the context excludes this possibil i ty. In (2), there is the same case as in (1). Here a plural NP acts in the same way as a quantified NP in Japanese (cf. Hoji (1982)). In (3), the c-commanding subject is watasi. Thus, watasi is coreferential wi th zibun. Also zibun can be referential, referring to the narrator. However, since the subject is the same as the narrator, the two coincide and the reference remains the same. In (4), second sentence from the bottom, John is the subject-topic; thus it is a possible antecedent. Mary is a performer in an invis ible performative sentence, such as Mary wa...to omotta "Mary thought...". Oshima (1979) claims that if one replaces zibun by kare, then kare is John; otherwise zibun only has a zibun-Mary reading. However, the fact is that zibun is ambiguous between John and Mary or the subject and the performer. The reason why only the zibun-Mary reading is possible is related to a type of discourse strategy: that is, zibun is used consistently to refer to Mary in the previous discourse. Examples (5), (6), and (7) are all taken from Naoya Siga's "Wakai" (Reconciliation). The story is narrated by the hero who appears as zibun, and zibun is consistently used to refer to the narrator. In (5), the local topic titi binds the subject. (Notice that the object zibun is marked by ga.) Thus, titi is a possible antecedent as far as the structure is concerned. However, the context (especially the 60 style of narration) c lar i f ies that zibun is referential and refers to the narrator. In (6), the subject sobo and the subject titi are possible antecedents, but again the context specifies that zibun refers to the narrator. In (7), tuma is the subject. But the head mentioned is not the wife's head but rather her husband's head. He is the narrator. Examples (8), (9) and (10) are also taken from Shiga's Wakai; however, the uses of zibun i l lus t ra ted therein are a l i t t l e different from those in the previous examples. In (8), the matrix subject zibun is referential by position and refers to the narrator. Note that this sentence sounds like an objective observation: the narrator describes his own action. This is because zibun is coreferential wi th a performer in an invis ible performative sentence such as X wa...to mita, 'X observed...'; thus, the narrator is the observed object (= zibun ) and the observer (= X) simultaneously. (9) and (10) provide another interesting case. In (9), zibun \ is referential by position. Zibun 2 is either referential or anaphoric. If anaphoric, zibun 2 is coreferential w i th the subject zibun \ . If referent ia l , each zibun can have a different reference as a logical possibil i ty. However, zibun \ and zibun 2 are coreferential contextually. (1 0) exhibits the same case. 61 In (1 1), zibun can be either referential or anaphoric by position; however, i t is anaphoric by context and bound to the c-commanding subject kare, which in turn is referential, referring to Geete (Goethe). The final cases are related to Topic 2. Zibun in (12) is ambiguous between the Topic 2, Taroo, and the subject otooto. In (13), zibun is coreferential with the subject which is null but bound by Topic 1. Topic 1, in turn, is also null but identif ied wi th huzyoori na ningen (absurd man) through the topic chain. §2.7 Conclusion Syntax cannot give the complete answer to the coreference poss ibi l i t ies of Japanese reflexives. However, it is tooextreme a move to leave the whole issue to a theory of pragmatics, since zibun s t i l l observes cer ta in syntact ic res t r ic t ions . The d i s t inc t ion between referential and anaphoric uses, which captures the nature of zibun, is more than just assuming zibun as the English reflexive equivalent . 2 1 That is, every zibun is potentially ambiguous between referential and anaphoric. Anaphoric cases are the ones whose coreference relations to other NPs are determined by sentence grammar which says, detai ls aside, that zibun has to be bound to the subjects (including Topic 2 ) . 2 2 If a sentence has a zero topic (which may get a coindex through the subject in case Topic 1), i ts referent is determined in discourse. And such a process is assumed to be in the formal representation of discourse structure, i.e., in the LF' module. 62 Thus, the basic schema is: the sentence grammar feeds the information such as potential candidates for the coreference (i.e., general "subjects" including Topic 2) to the non-syntactic modules where the final disambiguation in the coreference interpretation takes place. The LF' module of grammar identifies what is the entity of the topic. If it is null in syntax, Topic 1 is easily identifiable through topic chain in surface structure of discourse; while the Topic 2 case is more abstract and i ts nature is not fully clear yet, but it is related to the "ground" on which the story is told (cf. Chapter 5, mu l t i -wa sentences) or topic in a global sense. Now, subject which is syntact ical ly ident i -fiable candidate for the zibun coreference is also identified, in turn, as the si te of "awareness" or "viewpoint ." 2 3 If these are key notions to determine the coreference in discourse, it explains why the subject-bound topic, as opposed to the object-bound topic, is coreferential wi th zibun. The cases where zibun is coreferential w i th the narrator (referential case) seem to fal l in the same pragmatic domain controlled by a notion such as "viewpoint'V'awareness." 63 Notes (Chapter 2) 1 There is a fourth category, "E-type pronouns." However, these three are sufficient for the purpose in this chapter. 2 Actual ly (3) and (4) are s t i l l ambiguous between the coreference given by sentence grammar and the coreference given by discourse rule (referential use) since the pronouns are preceded by their ante-cedents. However, in (i), such a poss ib i l i t y is excluded. The coreference reading is solely determined by sentence grammar. This was pointed out to me by M. Rochemont (personal communication). (i) His mother loves John, (he = John) 3 If NP 1 precedes and kommands N P 2 , and N P 2 is not a pronoun, then NPi and N P 2 are noncoreferential. Definition: A kommands B if the minimal cycl ic node dominating A also dominates B. [Lasnik, 1976, p. 15] 4 The following pair does not exhibit a parallelism, unlike (4) and (5): (i) His mother loves everyone, (he * everyone) (ii) His mother loves John, (he = John) 5 In Japanese, zibun ' self , but not kare 'he', has this type of use. 6 The term "bind" is used here in an intui t ively understandable sense, not a syntactically defined one for anaphoric use. 7 Kuno's direct discourse analysis (e.g. 1973) is developed on this aspect of zibun. 8 'Anaphora' or 'anaphoric' expressions are different from 'anaphor': the lat ter is defined in the binding theory; the former means the expressions which have coreferential NPs in the same sentence. 9 The conf igura t iona l i ty does not necessar i ly dic ta te the tree s t ructure as in (29) (cf. Hoji [1982] for a binary branching 64 hierarchical structure). However, the choice of the configurational representations does not affect our analysis. 0 Zibun in object position and clause-bound cases like (i) in general are sl ightly awkward: (i) a. Hanakoi ga zibuni o hometa. 'Hanako praised 'self." b. Tarooi ga zibuni o syookai-sita. 'Taroo introduced 'self." (For an unknown reason, if zibun-de (by him/her self) is added, these sentences sound more natural.) There are peculiar cases with zibun-o. Once sentences in (ii) are embedded, unlike (i) where zibun can have long-distance coreference, they s t r i c t ly observe the clause-mate condition, as in (i i i ) : (ii) a. Taroo wa zibun o korosita. 'Taroo suppressed 'self." b. Taroo wa zibun o migaita. 'Taroo cultivated 'self." ( i i i ) a. HanakOj wa [Tarooi ga zibuni/*j o korosita] to omotta. 'Hanako thought that Taroo suppressed 'self'.' b. HanakOj wa [Tarooj ga zibunj/*j o migaita] to omotta. 'Hanako thought that Taroo cultivated 'self'.' The nature of zibun wi th these subjective state verbs might be that of an ordinary noun which has a meaning l ike "ego" or "mind." An expression such as kare wa zibun to iu mono o motte iru, 'He is his own man' contains such a zibun. This assumption is compatible wi th the following claim of Kuno's: 65 Zero pronominalization in the possessive case: the more predictable the reference of the possessor is, the more possible i t is that the possessive phrase becomes a zero pronoun, (translated by M. 0. [Kuno 1983]) Then, what is coreferential wi th Taroo is not zibun but the zero pronoun in the possessive position of zibun. Even so, however, it is s t i l l necessary to explain why the zero pronoun refers to Taroo and not Hanako. The explanation w i l l be given in exactly the same manner as it is given here, for the zero pronoun for this kind of zibun is used as the inalienable possession. 1 1 This condition is needed because of small clauses and believe type contexts. 1 2 In the fol lowing sentence, i t is a l i t t l e d i f f icu l t to make a com-parison between / and j and zibun^ but this is due simply to memory capacity: (i) [s 3 Tarook ga [s 2 HanakOj ga [s, Zirooj ga [ N P zibunj/j/?k NOM NOM NOM no ie] de yopparatte-iru] to itta] to itta]. 's house in drunk said said 'Taroo said that Hanako said that Ziroo was drunk in 'self's' house.' 1 3 Possessive is a problematic case for GC. GC for (i) often receives irregular treatment and assumes S instead of NP. (cf. Anderson (1979)): (i) [s John washed [ N P his car.]] 1 4 M. Soga's comments on Topic 2 called my attention to these kinds of examples, reflecting the nature of Topic 2. I also owe example (59) to Soga. 1 5 Terms which imply soc io-cul tura l bonds supply good contextual information for zibun as coreferential wi th a topic. The use of 66 kinship terms like musume (daughter), hahaoya (mother) in (55) is such a case. 1 6 Shibatani (1976-77 and the reference cited there) argues that the f i rs t NP in (i) can not be coreferential wi th zibun ; thus can not be a subject. (i) Yamada-sensij ga musukoj ga zibun*i/j ni prof. NOM son NOM to unzarisi te-iru. is-disgusted 'It is Prof. Yamada whose son is disgusted wi th self.' However, it is possible for zibun to be coreferential wi th the f i rs t NP as shown in (ii). (ii) Yamada-sensii ga musukoj ga zibuni/j no ie de house at netuite-iru. be-i 11-in-bed 'It is Prof. Yamada whose son is i l l in bed in self 's house.' The reason why zibun^ gets "*" may be that "zibun ni unzarisite-iru" is a clause-bound expressions s imi lar to the cases (ii) in note 10. 1 7 M. Rochemont pointed this out to me. 1 8 M. Rochemont pointed out to me that Discourse Representation Theory in Heim (1982) might implement this phenomenon in formalization. 1 9 This phenomenon was f i r s t observed, to my knowledge, by Heizo Nakajima and reported in Kitagawa (1981). 2 0 Some native speakers have informed me that (61c) is s l ightly better than the scrambling case, i.e., wi th Hanako-o instead of Hanako-wa. This may suggest the (61c) could be improved in a certain discourse 67 environment. However, the discourse context l ike (78) is, at least, not the one for (61c) to improve. 2 1 After I completed this chapter, I learned that the English reflexive is not so straightforward, and is also controlled by a certain discourse factor, (cf. Zribi-Hertz (1989)) 2 2 Psych-verb cases (McCawley (1976)) are not direct ly discussed in this chapter. For the treatment of this type of sentence in a GB framework, cf. Bel le t t i and Rizzi (1988). 2 3 H. Terakura has pointed out to me that the term "subject of consciousness" used in Ann Banfield (1982) (Unspeakable Sentences, London) may possibly be added to this l is t . 68 Chapter 3 Japanese Demonstratives: Ko-So-A §3.0 Introduction In the preceding chapter, I examined the Japanese ref lexive pronoun zibun, which is used referent ia l ly /anaphor ica l ly . In this chapter, I w i l l be examining Japanese demonstratives, wi th special attention to those used anaphorically as antecedents in discourse, and compare them wi th pro-nominal elements in discourse. I w i l l f i r s t examine the main issues concerning the nature of demonstratives and provide a general perspect ive regarding exact ly how Japanese demonstratives work. I believe that any factual observation which is made about their use in discourse should be re-examined in the light of thei r de i c t i c function. Then, I w i l l examine the d i s t r ibu t iona l difference between the /co-series and the so-series in discourse. One might say that such choices are more or less pragmatic matters: see, for instance, the following example, where we can employ either ko or so : 69 (1) Tantyoosa ni tomonau aki wa yoku hiroo monotony follow boredom often weariness to taihi sareru. Kore/sore wa tugi-no-yoo na be-compared this/that following riyuu ni yoru. 1 (Kobayashi: Tantyoo na Shigoto) reason 'Boredom caused by the monotony (of tasks) is often com-pared with weariness. This is for the following reason.' This type of problem is often regarded as a matter of, say, the language user's subjective ( = ko ^ob jec t ive ( = so ) at t i tudes toward the referent. However, my concern here is not s t y l i s t i c s , but rather the behavior of the demonstrat ive i t se l f , which may eventually be correlated wi th various kinds of phenomena, some purely l inguist ic and some not. A close examination of the nature of the referents of each demonstrative is one of the main aims of this investigation. In the course of the chapter, the character is t ics of the use of the various demonstratives w i l l be compared with that of pronouns. 70 §3.1 The Use of the Demonstratives §3.1.0 Introduction While English has a twofold demonstrative system, i.e. this-that/ here-there, Japanese has a threefold system, the /co-series, the so-series, and the a-series demonstratives, which are distributed over the various syntactic categories as shown in (2): (2) ko so a do (interrogative) noun 1 (object) kore sore are dore noun 2 (object/ person) koitu soitu aitu doitu noun 3 (place) koko koko asoko doko noun 4 (direction) kotti kotira sow sot ira atti at ira dotti dot ira nominal-adjective konna sonna anna donna determiner kono sono ano dono adverb koo soo aa doo For purposes of translation, Japanese demonstratives and English ones are taken to correspond in the following manner: 71 (3) Japanese ko so | a English this/here that/there The a-series are sometimes translated as "that one over there/' since objects pointed out by the a-series can be found further away than those pointed out by the so-series. The traditional views about the threefold system in Japanese demonstratives are based on this observation, namely, that the objects for ko-so-a may be arrayed according to the distance from the speaker, as shown in (4): (4)a. Kono kyuuri to, soko no tomato to, eeto, asoko this cucumber and that tomato and er... that ni aru no nasu kasira. Dattara are mo tyoodai. exist eggplant isn't if so that too give-me (At a greengrocer's) 'I w i l l take these cucumbers and those tomatoes and, er, is that one (over there) an eggplant? If it is, give me some of that, too.' b. cf. SPEAKER : cucumber -> tomato -» eggplant [ko ] [so] [a] Proximal Medial Distal The modified version of (4b), which is presented in (5), incorporates the existence of the hearer in this schema and is often considered to be a better-supported alternative: 72 The schema given in (5) indicates that something near to the speaker is referred to by ko ; something near to the hearer rather than the speaker is referred to by so ; and something distant from both the speaker and 0 the hearer is referred to by a. Sakuma (1936/enlarged ed. 1951) is often quoted' as the representative of the nawabari ' terri tory' theory developed from the view presented in (5). Sakuma claims that something in the speaker's terr i tory is referred to by ko ; something in the hearer's terr i tory is referred to by so ; and something in neither the speaker's nor the hearer's territory is referred to by a, as shown in (6): 1 73 (6) Speaker Hearer [a ] [Sakuma 1951] There have been many modifications of this theory after i ts formulation by Sakuma, but the idea of territory formation has been accepted as the foundation of studies of the Japanese demonstratives. Now, in the next three sections I w i l l investigate three core notions in such theories of the demonstratives: (i) ternary vs. binary; (i i) distance; and ( i i i ) the existence of the hearer. §3.1.1 Ternary vs. Binary Sakuma's demonstrative theory assumes a t r ipar t i te structure in t e r r i to r ia l formation. On the other hand, Mikami (1955) claims that ter r i tor ia l formation takes place in two sets of binary relations: ko vs. so and ko vs. a, as shown below in (7): 74 (7) a. b. [a ] > Case 1 Case 2 According to Mikami, there are two cases of psychological terr i tory formation wi th respect to koso-a use. In Case 1, ko and so share the oval field. The oval shape indicates that the f ie ld shared by the two has two centers: the speaker and the hearer. In Case 2, the terri tory of ko completely overlaps or absorbs the terr i tory of so, and ko in turn is opposed to a. The speaker is the center of ko in this case; however, a does not have a corresponding center. As the result of the combination of the binary sets indicated in Case 1 and Case 2, the apparent koso-a opposition is formed. In cases like that exemplified by (4), the shift from Case 1 to Case 2 takes place rapidly (Okamura 1972). This assumption may explain the lack of so-a pairs in the id io-matic expressions given in (8) below: 75 (8)a. koso p a i r s s o r e y a k o r e y a de t h a t and t h i s (and) w i t h s o k o k o k o ni t h e r e h e r e a t ' b e c a u s e of t h i s and t h a t ' ' o v e r t h e r e and o v e r h e r e ' soo koo s u r u u t i n i ' w h i l e d o i n g t h i s i n t h a t w a y i n t h i s w a y do w h i l e and d o i n g t h a t ' s o n n a k o n n a de ' b e c a u s e of c a s e s i n t h a t ( r e a s o n ) i n t h i s ( r e a s o n ) w i t h l i k e t h a t and c a s e s 1 i k e t h i s ' b. ko-a p a i r s a r e (ya) k o r e (ya) t h a t (and) t h i s (and) aa da koo da t h a t w a y t h i s w a y k a r e ( = a r e ) 2 k o r e t h a t t h i s k o k o k a s i k o ( = a s o k o ) h e r e t h e r e 'one t h i n g a n d / o r a n o t h e r ' ' t h i s and t h a t ' ' t h i s and t h a t / a b o u t ' ' o v e r h e r e and o v e r t h e r e ' c. * s o - a p a i r s -T h e p a r a d i g m d o e s n o t , of c o u r s e , p r o v i d e d i r e c t s u p p o r t t o t h e b i n a r y t h e o r y , b u t a t t h e l e a s t i t t e l l s us t h a t f o r s o m e r e a s o n so-a c a n n o t 76 form pairs in opposition. More interesting items of data on idiomatic expressions are given in (9) below: (9) dare sore who that (person) doko soko where there nanda ka( = a)nda itte (it is) what (it is) this saying 'so and so' 'such and such a place' 'on some pretext or other' The interrogative forms, dare 'who', doko 'where', and nani 'what' are combined wi th so or a but never wi th ko. This fact indicates a special property in ko, and I w i l l be returning to this topic in section 3.1.3. The next item of data to consider related to the binary theory is the possible range of choices among the demonstratives. Without any change in physical position, the speaker can point out the same referent, sometimes wi th ko, sometimes wi th so ; ko-so and ko-a are inter-changeable, respectively, under certain conditions. However, al l three members of the ko-so-a series are not interchangeable at the same time: R = a referent which is approximately the same distance from A and B, wi th the two facing one another. b. A: Kore/sore/*are wa nan-desu ka. this that what Q 'What is this / that? ' B: Kore/sore/*are ga rei no yatu sa. in question one 'This/that is the one I mean.' (1 Da. I R I A B 78 b. Hyottositara ano tubo, syuusyuuka ga nodo by some chance that vase collector kara te ga deru hodo hosigatte-iru to-ka-iu.... enthusiastically want I hear [(monologue) Kore o sagasu tame ni dorehodo this look for how much kuroo-sita-koto-ka.] have-a-hard-time 'Hey, is that the vase which I hear that many collectors are drooling over? [Finally I found it. What a hard time I had to track it down!]' Ko-so in (10) and ko-a in (1 1) are interchangeable, respectively; but it is not possible to use a in (10) and so in (1 1). The possible pairs in (10) and (11) correspond to the pairs given in the binary opposition. Again, this does not provide direct support for the binary theory, but the impl ica t ions are worth invest igating. In this chapter, I w i l l be assuming a modified binary system, of which the detai ls w i l l be discussed in §3.1.3. 79 §3.1.2. Territory Formation: Physical Distance vs. Psychological Distance Assuming that the primary function of demonstratives is deictic (Lyons: 1977), let us focus on the spatial use of demonstratives, where it is clear how S-H-R are arranged in physical space. The problem here is how we determine whether the referent is closer to S, closer to H, or close to neither S nor H. The fol lowing examples demonstrate that the physical distance involved is not the main factor in determining the "distance" (although it is not total ly irrelevant). (12) a. A: Soko wa moo sukosi mizikame ni tanomu yo. there a l i t t l e more shorter please (at a barber shop) 'Wil l you make it a l i t t l e shorter there?' Who can te l l which person is physically closer to the speaker's hair, the speaker whose hair it is , or the barber who is actually touching the speaker's hair? Also observe (1 3) and (1 4): (13) A: Kore ga koko no itamae no ziman-ryoori this here (this restaurant) chef special (dish) da yo. 'This is the chef's special.' 80 B: Yaa kore wa umasoo da. wel l this looks-good 'Well, t h i s / i t looks good.' (14) Kono usa-tyan wa nanto-iu onamae? this bunny what name 'What is this bunny's name?' In (13), the dish is on the table, in between the two speakers, or it is even possible in this instance for it to be physically closer to B; but A s t i l l uses ko. In (14), the utterance is made to a small chi ld who is playing wi th a stuffed animal: the addressee is touching the referent, the speaker is not, but nevertheless the speaker s t i l l uses ko. The fact is that the physical distance is very much affected by the psychological one. In (13), A's utterance implies that he is the host or at least patronizes the restaurant; and in this sense, the dish on the table is "closer" to A than to B. In (14), a friendly speaker is eager to show her closeness to the doll and thus to the g i r l : this is why she used ko instead of the alternative, so. The psychological parameters of distance perception may be i l lustrated by (15): 81 (I5)a. Each speaker (A, B, C, D) is almost the same distance from the referent, a scorpion. b. A: Saa, kore/*sore ga mekisiko de saisyuusita this Mexico at collected B: sin-syu no sasori da yo. new species scorpion 'Here it is. These are the scorpions of a new species which I collected in Mexico.' Dore dore! Kore/*sore wa sugoi ya. Mata korekusyon let me see this collection C: ga fueta ne. added 'Let me see! This is great. You've added to your collection again.' Konna/Sonna mono o tukamaeru tame ni this kind of/that kind of thing catch 82 harubaru makisiko made i t tekita-no-kai . all the way Mexico to went Aikawarazu kawarimono-da-ne. as usual eccentric 'You went all the way to Mexico to catch this/ that thing? You are as eccentric as ever! D: Kimoti-warui! *Konna/Sonna mono hayaku disgusting that thing quickly dokoka e yatte yo. ([in a whisper to C] somewhere send Anna/*konna/*sonna mono no doko ga i i -no-kasira , that thing where-good mattaku.) actually 'Disgusting! Take that thing away. (I don't understand what he sees in it, really!) In (15b), as an enthusiastic collector, A is "close" to his col lect ion; thus, he uses ko. B is on A's side, and shows his strong interest by also using ko. C is amazed by A's hobby: he may use so to express his disgust, or remain wi th ko to show his amazement, which counts as a form of interest. D is a person total ly out of sympathy wi th A's hobby. 83 She wants to create a psychological distance between herself and what she sees as A's bizarre collection, and so ko is avoided when speaking to the group as a whole. However, since D either considers C her ally or wants to win C over to her side, she uses a in the remark addressed privately to C (cf. (7b)). Thus, Sakuma's theory of terr i tory formation, i l lus t ra ted in (16) below, makes better sense when constructed on the basis of the participants' perception of psychological distance: (16) [ko } [so } speaker and hearer's territory R. Lakoff (1974) categorizes English demonstrative uses into three classes: Spatio-temporal deixis (pointing words); discourse deixis (anaphora); and emotional deixis. I do not agree wi th this c l ass i f i ca t ion , at least as far as Japanese is concerned, since even spatio-temporal deixis is not free from psychological assessments of 84 "distance," as I have demonstrated above. But what interests me most is the emotional de ixis which is used for "vividness" or creat ing 'closeness.' Lakoff's examples are duplicated here as (17) for this and (18)for that : (17)a. I see there's going to be peace in the Mideast. This Henry Kissinger really is something! ( = Lakoff's (8)) b. He kissed her wi th this/an/*the unbelievable passion. ( = (11)) (1 8)a. How is that throat? ( = (35)) b. That Henry Kissinger sure knows his way around Hollywood! ( = (45)) Lakoff's comment on the tftat-case is , "these are perhaps the most curious semantically, since the distance marker that seems to establish emotional closeness between speaker and addressee" (p. 351). However, this is no surprise at all if that is considered to be an undifferentiated equivalent of both so and a ; (1 8) would be a case of a in Japanese. The emotional closeness in (17) is between the speaker and the referent, while that in (18) is the speaker and the hearer, or the two who share the conversation. Lakoff's analysis of English this/that can easily be fi t ted into the schema for the Japanese ko and a in (16). The point on which I disagree wi th Lakoff is that, unlike her, I assume that spatio-temporal deixis is also affected by psychological factors. Lakoff's categorization of the demonstratives may be descriptively right, but I believe that three usages are generated from the fundamental schema of 85 the function of demonstratives, where the (psychological) distance factor is one of the parameters. The reason why I consider the term 'territory' should be preferred over 'distance' is that ' territory' implies a sense that the subject matter or the other parties 'belong' to the speaker; while 'distance' does not. Now, notice that in the case of a in (16), the speaker and the hearer form the territory whereas the referent does not belong to it; or in the case of so, the referent belongs to the hearer's terri tory but not to the speaker's. In the schema in (16), speaker, hearer, and referent participate in the formation of territory. However, is it the case that the hearer's existence is as significant a factor as that of the other two in terri tory formation? This is the problem that I w i l l be addressing in the next section. §3.1.3 Speaker's Territory and Egocentricity In the preceding section, I mentioned that the sense of distance is affected by the speaker's perceptions; however, it is also true that the physical distance constrains the perception to a certain extent. For example, in pitch darkness, the speaker can only use ko for a referent which he can touch or feel. A person cannot use so to refer to the food in his own mouth, and other people cannot refer to that same food with ko. Here, the assumption of a hierarchy of 'truth' in McCawley (1978) may lead to an explanation: 86 (19) A Hierarchy of 'Truth' A Ego's knowledge acquired through his sensory experiences. Ego's knowledge acquired through purely logical reasoning. Ego's knowledge acquired via someone else's knowledge. [McCawley (1978): p. 178] A thing which the speaker can perceive wi th certainty wi th in his ter r i tory is one whose existence the speaker can confirm wi th his senses. Such an object w i l l be the most certainly 'true' for him, in McCawley's terms, and w i l l thus be referred to by ko. Observe (20): (20) sense e x p e r i e n c e r n o n - e x p e r i e n c e r taste Kono/*sono/(ano) a z i nandaroo . *Kono/sono a z i t y o t t o hendaroo. (What i s t h i s / * t h a t t a s t e ? ) ( Isn ' t * t h i s / t h a t / ( t h e ) t a s t e a l i t t l e s t r a n g e ? ) s m e l l K o n o / * s o n o / ( a n o ) n i o i nanda roo . *Kono/sono n i o i doo omoo. (Wha t i s t h i s / * t h a t s m e l l ? ) (What do you t h i n k * t h i s / t h a t / [ the] s m e l l i s ? ) touch K o n o / * s o n o / ( a n o ) t e z a w a r i k a r a s u r u t o , a s a r a s i i . *Kono/sono t e z a w a r i a s a m i t a i d a r o o . ( J u d g i n g f r o m t h i s / ( t h e ) / * t h a t t e x t u r e , i t s e e m s to be l inen. ) ( * T h i s / t h a t t e x t u r e i s l i k e l i n e n , i s n ' t i t ? ) 87 f ee l i ng Konna/*sonna/(anna) s u g a s u g a -s h i i k i b u n w a h i s a s i b u r i da. *Konnani/sonnani k i b u n ga i i k a i ? (1 have not had t h i s / * t h a t r e -f r e s h i n g f e e l i n g f o r a long t ime . ) (You f e e l * t h 1 s / t h a t g o o d ? ) sound Kono/*sono/ano o to nandaroo. *Kono/sono o to i kaga . (Wha t i s t h i s / * t h a t / t h a t ( ove r t h e r e ) s o u n d ? ) (What do y o u t h i n k ^ t h i s / t h a t sound i s ? ; eg. the a d d r e s s e e i s u s i n g headphones . ) s igh t Kono/-*sono/ano e m i t a ko to ga a r u n a a . Kabe no * k o t t i / s o t t i g a w a w a donnahuu da i . (1 have seen t h i s / * t h a t p a i n t i n g b e f o r e . ) (Wha t i s ^ t h i s / t h a t ( the o t h e r ) s i d e of the w a l l l i k e ? ) In t h e c a s e s of ' t a s t e , ' ' s m e l l , ' and ' t o u c h ' i t i s of c o u r s e p o s s i b l e f o r t he s p e a k e r and the h e a r e r to e x p e r i e n c e the s a m e food , s m e l l , or t e x t u r e at the s a m e t i m e ; but s u c h i n s t a n c e s are exc luded . In cases concerning 'feeling,' 'taste,' 'smell , ' and 'touch,' a can be used only when the speaker is referring to past experiences; while a is perfectly all right for referring to present experiences observed through 'sight ' and 'sound.' However, in the la t te r case, ko and a are differentiated according to the physical distance from the source of sound or sight. When someone asks the experiencer about the experience, so is used. In other words, things (the experience in this case) which do not belong to the speaker are indicated by so. This fact casts the significance of the role of the hearer in the use of so into doubt. 88 In her categorization of emotional deixis, Lakoff claims that that creates a "closeness" or "togetherness" between speaker and hearer. According to Sakuma's schema, in the case of a, the hearer and the speaker belong to the same terri tory. In real i ty , however, it is not necessarily the case that the hearer is involved in the use of the a-series: typical examples are soliloquies like those in (20), but this may be involved in other situations as wel l . Examine ( 2 1 ) - (23): (21) Kimi niwa soozoo mo tukumai daroo ga anna sugasugasii you cannot-imagine probably but that refreshing kibun wa umarete hazimete no keiken sa. feeling for - the- f i r s t - t ime- in one's life experience 'You probably cannot imagine (this), but that was the f irs t time in my life I experienced that refreshing feeling.' (22) Ano soba wa umakatta naa. Kimi mo zehi itido that noodle was-good you by all means once tamesite goran yo. try 'Those noodles were darn good. You should try them sometime!' 89 (23) Ano tezamari kara suruto asa mitaidatta keredo, that texture judging-from linen seems-like but kinu mitai ni zyuryookan ga nakute ne. Zitubutu o si lk like weight do-not-have real thing siranai kimi ni setumei suru no wa tyotto don't know you explain a l i t t l e muzukasii kedo... diff icult 'Judging from that texture, it seemed like linen, but it is as light as silk. Since you have not seen the real thing, it is very diff icult to explain.' It would seem to be the case that the existence of the addressee is a somewhat secondary factor in the above cases: these are not monologues, and the addressees exist for the speakers. Thus, as an alternative, I propose the following hypothesis: the principle governing the d is t inc t ive use of the three demonstratives is established on an "egocentric" basis. The terr i tory is always formed wi th the speaker's ego as i ts center, and the way in which the object is referred to depends on whether it is within the speaker's ego domain or not. This egocentric hypothesis may seem a bit too speculative to be of use in explaining a l inguis t ic phenomenon. However, there are some other phenomena for which the egocentricity hypothesis gives a better interpretation than other existing analyses, or at least new openings for investigation: the 90 focus of interest (Zubin 1979, also cf. Chapter 4), for instance, and s tory-retel l ing (Bartlett 1932 reported by Zubin). Moreover, McCawley's hierarchy of 'truth' is based on the degree of distance from the speaker's ego. Now, returning to the problem wi th the demonstratives, the following diagram w i l l i l lustrate what seems to be going on: (24) [so ] [so ] The boundaries of the speaker's terri tory, which is formed on the basis of the speaker's ego, are elastic: something in the microworld may be referred to by ko, or the whole universe may be ko. The boundaries of a correspond exactly wi th ko. Thus, a is also elast ic , changing to conform with the speaker's territory. As the result of this territory formation (the flexible domain of ko ), something in the speech arena which does not belong to the speaker's terri tory w i l l be indicated by so, regardless of whether the hearer is inside or outside of this schema. In other words, whether the speaker's ego recognizes the other 91 existence in this schema or not is a secondary factor. For example, in an actual conversation, i t can be the case that the speaker stands in opposition to the other, and he may utter a sentence like (25): (25) Kore wa boku no omotya de, sore wa kimi no. this . I toy that you 'This is my toy and that is yours.' Here, the speaker tr ies to make a clear dist inct ion between something which is in his terri tory and something which is not, and by the nature of the conversation, the hearer exists in a domain which is opposed to the speaker's. Or observe (26) below: (26) a. A: Kono tubo doo omou. this pot how think 'What do you think of this pot?' b. B: (Dore dore) Aa, kore wa neuti mono da nee. this valuable thing '(Let me see) Well , this is a valuable one, you know.' In (26a), the speaker is A and the hearer is B, and A used ko ; in (26b), the speaker is B and the hearer is A, and B used ko. Here, the terr i tories of the two speakers overlap almost completely. Hearer's existence is not included in this schema, even though two people have taken turns in playing the roles of speaker and hearer in this conversation. The 92 speaker's position is placed in opposition to the rest of the "world," let us say, and this "world" is often represented by the hearer in conversational situations: ko (speaker) is opposed to so (non-speaker, possibly hearer). On the other hand, in a ko-a opposition, the speaker's ego does not contrast i t se l f w i th the rest of the world. Rather, a is wi th in the domain of the speaker's ego, or wi thin an extension or projection of the speaker's ego domain. Within the speaker's "world," ko, which indicates the "here" and "now" is opposed to a, which in turn indicates the "there" and "then." Seen in this light, it is easy to explain why in the a-series the togetherness of the speaker and the hearer is emphasized more than with the ^co-series. Below, I w i l l explain this point further. Firs t of a l l , if one is making a serious effort to play the hearer's role (it w i l l be necessary to make such an effort as long as one is interested in the conversation or wishes to be a hearer), one has to participate in the speaker's story in an attentive manner. If the speaker uses ko, the hearer w i l l make an effort to identify the /co-object. However, the hearer cannot refer to some object existing in someone's inner world using ko even if he succeeds in forming a clear image of the /co-object (cf. (20)). Thus, ko is ruled out in (27): 93 (27) A: Tyuugaku no dookyuu ni Yamada toiu junior high school classmate otoko ga ite ne. Koitu no henzin-buri wa man was this man eccentric manner do-hazurete i ta ne. extraordinary 'I have a junior high school classmate called Yamada. His eccentricity was extraordinary.' B: Sore de *kono/sono hito ima doosite-iru no. So this that person now is doing Q 'So, what is that/(the) friend doing now?' Notice that in the use of ko in (26), speaker and hearer are very close, and the two share the referent as the ko object. Yet the use of ko does not bring a s imi la r type of "togetherness" as is implied by the use of a. This is because in (26), the two te r r i to r i es overlap yet are s t i l l separate: there exist two egos as the center of each terr i tory in the same 'space.' On the other hand, in cases wi th a there is no ter r i tor ia l overlap as there is wi th the ko case given above. According to the Cooperative Principle of Conversation in Grice (1975), a conversation is formed on the basis of the mutual cooperation of the speaker and the hearer: the speaker may offer something which is more commonly shared between 94 the two. A l so , i f the hearer can relate to the speaker through identifying the a-referent, which exists in the speaker's mind as his "there" and "then," it means that the hearer can share, in a sense, the speaker's inner world. The more successfully the ident i f icat ion has been carried out, the closer the relationship established between the two: (28) A: Ano koro wa hontooni yokatta yo. Naa, Yoshida-kun that time really nice you know ya Kinosita-kun nimo ano toozi no seikatu o and too that time life taikensasete-yaritai yo. let-(them)-experience-for-them 'Those were the good old days. You know, I want to let Yoshida and Kinoshita experience what life was like at that time (for their sakes).' B: Ee, hontooni ano koro wa yoi zidai desita ne. yes really that time good time 'Yes, it was a really good time, wasn't it.' 95 C: Sono koro no koto o zehi hanasite-kudasai. that time thing by all means te l l - ( i t ) - for-me 'Please te l l me (for my sake) how things were at that time.' (29) A: Kimi to dookyuu no Tanaka-kun ne. you classmate 'You know your classmate Tanaka?' B: Aa, ano Tanaka ne. 'Oh, that Tanaka?' The curious thing here is why the referent is far from the speaker/ hearer terr i tory - for the speaker, further than the so-object in the actual si tuation - while at the same time, the speaker (and often the hearer) identifies the referent with "closeness." A possible explanation is as fo l lows: the speaker has the referent in his inner world , regardless of the hearer's acquaintanceship wi th i t , since things do exist in the speaker's ego world, and the ko-a opposition occurs wi th ko as his ego center and a as the extension of his ego domain: the former is d i f f i cu l t for the hearer to share; but the la t ter is more or less accessible and the speaker tends to choose the referent which is mutually f ami l i a r , since such a topic, shared between the two par t ic ipants in the conversation, makes it in teres t ing and thus practical . Certainly, though, the speaker can s t i l l use ko-a even if the 9 6 hearer cannot identi fy the referent, leaving the la t ter a contented listener. As for the so ser ies, according to the standard assumptions of terr i tory theories, the so-object is in the hearer's terr i tory. Observe (30): (30) A: Mukasi, onazi tyoonai ni taihenna wanpaku kozoo ga old days same town very naughty kid ite ne. Tasika tabakoya no hitori-musuko was i f - l - remember-r ight cigar store only-son datta to omou kedo, tonikaku koituno was (I) think though anyway this one rotten l i t t le brat was-no-comparison 'In the old days (when I was small) , there was a naughty kid in the same town I l ived in - if I re -member right, he was the only son of the [man who owned the] cigar store, I think - anyway, he (this one) was the most unbelievable rotten l i t t le brat.' warugaki-buri- t tara nakatta yo. 97 B: Aa, sono ko nara s i t t e - i ru wa. Ano ko no itazura oh that child (I) know that child mischief niwa zuibun nakasareta mono, very suffer greatly you know 'Oh, I know that kid. His mischief gave me a lot of trouble.' C: Hee.... Demo sono ko ga ima no hanasi to Yeah... but that child the-story-you-told dooiu kankei ga aru no. what relationship is 'Yeah. But what does that kid have to do with the story you told me?' Kitagawa (1978) characterizes the use of so in discourses like (30) as "to put the identity locus of the referent in the addressee's sphere, or, at least, away from the speaker's" (p. 236), and sono hito 'that person' means "the person whom you have mentioned" (p. 236), which sat isf ies one of Grice's conversational principles, "Be Relevant". However, in wri t ten paragraphs is the person whom the wr i te r has mentioned in the hearer's terr i tory? It is often the case in wri t ten discourse that the existence of the hearer ( = reader) cannot be specified. I see the so-phenomenon in this: the referent is simply outside of the speaker domain; however, in many cases, the hearer exists in such an "outside" 98 domain, and faces the speaker as an interlocutor. This can explain the use of so regardless of whether mutual part ic ipat ion between speaker and hearer is involved or not. If the demonstrat ive system is analyzed from the speaker -centered viewpoint used above, this may form a bridge between a type of focusing system and the d iscourse use of the demonstrat ives: the referent which is highlighted because of being in the speaker's ego domain may fa l l outside of that domain in the next instant, as the spotl ight of attention moves along wi th the f low of the discourse. I w i l l return to this topic later in this chapter. Now, in sect ion 3.1.1, I examined the binary theory as opposed to the ternary theory. Assuming the superior i ty of the binary framework on the basis of "egocentricity," I propose the fol lowing schema (31) for the koso-a demonstrative system in Japanese: (31) •so There exist two kinds of opposit ions in the system: i) speaker's "here and now" vs. "there and then" - [KO] vs. [A]; ii) speaker's cognitive world 99 vs. outside world - [KO] vs. [SO], The domain of ko is e last ic and so is that of a. The way ko and a are used in (32) indicates the nature of the ko-a opposition: (32) Hee, ano e ga kimi ga daifuntoo-site oh that painting you make-strenuous-efforts seriotosita to-iuu...[(monologue) Makasa made-a-successful-bid by no means konna tokoro de kono e to saikai-suru towa a-place-l ike-this this painting see-again omatte-inakatta naa.] did-not-think 'Oh, is that the painting you went all out to make a successful bid for? [I never imagined that I would see this painting again in a place like this.] The referent in the speaker's "there" is referred to by a. The Japanese sentence implies the existence of a hearer; thus, the purpose of the utterance is an objective identif ication of the referent. The physical distance is the dominant factor. Then, in the monologue, the speaker's strong concern for the referent is the controlling factor; the physical distance is almost irrelevant and the speaker recognizes the referent in his "here" domain. In wri t ten texts, apart from dialogues, a is seldom 100 used except In cases like (33), in which the wri ter believes he can share the same cognitive sphere: (33) Kono yoona syakaizyoosei no ugoki wa wareware l ike- th is the-state-of-society movement us ni ano 19xx-nen tooji no zyookyoo o that (year) then the-state-of-affairs omoidasaseru. make-(us)-recall. 'The way the state of society is developing reminds us of the way it was in (that) 19xx.' This usage is predicted by the nature of a : in wri t ten discourse it is not easy (though as the above shows, it is not impossible) to identify oneself or feel oneself related with the writer 's "there and then." As for the ko-so opposition, more elaboration is needed as regards the i m p l i -cations of being outside of the speaker's domain. The operation of the so series in discourse is, as a matter of fact, far from simple. In the rest of this chapter, I w i l l examine the use of so in discourse, and in the course of this the nature of ko and a w i l l also be discussed. 101 §3.2 The Use of Ko/So in Discourse §3.2.1 Introduction There are two major problems which w i l l be examined in the fol lowing sections. The f i r s t one concerns anaphoricity. The use of demonstratives is often classif ied into two categories: deixis (pointing out the reference) and anaphora (finding the referent in discourse). However, Kuno (1973 a, b) claims that ko is a "semi-anaphora" or "semi-demonstrative ( = deixis [M. 0.])" because of the "vividness" imparted by i ts use; while Mikami (1970) claims that a is always deixis. Compare (34) and (35): (34) Soo ieba Tanaka mo moo mago ga iru tosi speaking-of-that too already grandchild have age da naa. Ano otoko ga oziityan nante yobareru zu wa that man grandpa is -ca l led picture tyotto soozoo dekinai kedo.... a l i t t l e imagine cannot 'Speaking of that, Tanaka too is at the age where he could have grandchildren. I cannot picture that man (he) being called grandpa.' 102 (35) A: Mukasi onazi tyoonai ni tabakoya no hitori-musuko old days same town at tobacco store only-son de taihenna wanpaku ga ite nee. very naughty kid was 'In the old days, there was a very naughty kid who was the only son of the cigar-store (owner).' B: Sono/ano ko nara watasi mo s i t t e - i ru wa. that kid if I too know 'If you are talking about that kid, I know him.' If a as in (34) and (35) are deictic, then what about so in (35)? Kuno admits this type of use is "probably nonanaphoric but demonstrative" (1973a, p. 289). In fact, so in (35) is different from the so s in (36), but what exactly differentiates the usage in (36) from that in (35) is s t i l l far from clear: (36) Sore wa kahun-bunseki to iwareru hoohoo de kahun no that pollen-analysis is -cal led method pollen naka no keisi tu no bubun ga itai tosite nokori, sono syurui, in s i l i ca portion remain as leave that type ryoo kara kako no syokusei no hukugen o quantity from past plant-distribution reconstruct 103 s i , sore o toosite kikoo no hensen o siru toiuu that through climate change know mono de aru. (Sikata: Yoozo no kigen) 'That is a method which is called pollen analysis. The s i l i c a component of the pollen is left as remainder, and from that type and quantity we restore the plant d i s t r i -bution in the past, and through that we come to know the changes in the climate.' Thus, the question which must be addressed here is: what is the exact nature of the anaphoric ( = discourse) use of the demonstratives, ko, (a ), and so ? A second question which must be asked concerns the determining factor governing the choice between ko and so. Observe (37) and (38): (37) Naisin de motiron desi no soo ga zibun o in one's inmost heart of course disciple monk toki-husete kono hoo o kokoromi-saseru no o matte ita convince this method make-(him)-try wait no de aru. Desi no soo nimo Naigu no kono sakuryaku ga disciple monk too this strategy wakaranai hazu wa nai. Sikasi sore ni taisuru hankan must-know but that against i l l - fee l ing 104 yori wa sooiu sakuryaku o toru kokoro-moti no hoo ga than such strategy take psychology yori tuyoku kono desi no soo no doozyoo o ugokasita more strong this disciple priest sympathy appealed to no-de-aru. it w i l l be the case that (Akutagawa: Hana) 'In the depths of his heart, of course, he was waiting for his disciple monks to convince him and make him try this method. His disciples must have sensed this (his) strategy. But, more than the i l l - fee l ing towards that (such a) strategy, it would be the case that Naigu's psychology, which made him think in that way, appealed to the sympathies of this disciple.' (38) Naigu wa hazime, kore o zibun no kao-gawari ga sita at f i rs t this his face-metamorphosize sei da to kaisyaku-sita. Sikasi doomo kono because of interpret but somehow this kaisyaku dake de wa zyuubun ni setumei ga interpretation only adequately explain tsukanai yoo-de-aru. Motiron, Naka-doozi ya Ge-boosi ga of course and 105 warau genin wa 2Qk& ni am ni tigainai (ga).... laugh cause there was must (Akutagawa: Hana) 'At f i rs t Naigu interpreted this to mean that it was because his face had undergone a metamorphosis. It seemed that this interpretation as wel l could not explain things adequately. Of course, the reason why Nakadoozi and Geboosi laughed was there, (but)....' In (37), besides the original kono (sakuryaku)/sooiu (sakuryaku) combination, three other combinations would be theoretically possible: (39) a. [kono/sooiu) (original) b. [kono/*kooiu) c. [sono/sooiu] d. {sono/*kooiu} However, (39b) and (39d), which contain kooiu as the second component, are not acceptable. Why is this so? In (38), the combinations are as follows: (40) a. [kono/soko) (original) b. [kono/??koko] c. [sono/soko) d. [sono/??koko) Again, some pairs, like (40b) and (40d) are less preferred, if not entirely unacceptable, when compared wi th their so counterparts. What deter-mines the choice between ko and so here? 106 The two questions posed above are related, and involve many subsequent topics like pronoun vs. demonstrative, referential vs. non-referential, antecedents in discourse, and so on. In what fol lows, these topics w i l l also be examined in the course of the investigation of the two major questions. §3.2.2 The Ko/So - Antecedent Relation In this section, I w i l l discuss how ko and so are used in discourse. Firs t , let us look at the results of the analysis of actual texts in Ono (1975 and 1977) to see what differences there may be. Five sc ient i f ic a r t i c l e s 4 were chosen for this study of the demonstratives. The total number of occurrences of the demonstratives in question was 571: 253 cases of ko and 3 18 cases of so. The results which are relevant to this section's objectives are as follows: [1] i) The average distance between the antecedents and demonstratives (D = 1 means the antecedent is in the adjacent sentence). D /co-series 1.063 so-series 0.377 (t = 13.407 Pr << 0.001) 107 The antecedent of ko is found further distant than that of so in the previous discourse, that is , ko has a longer-distance power of referral . Within the data, cases in which the antecedent is more than two sentences away are only to be found with ko antecedents. i i) The location of ko/so antecedents Ko and its antecedent tend to be found in two separate sentences: i t is rare to find i ts antecedent wi th in the same clause as the ko, whi le the antecedent of so is usually found in the same sentence. sentence same different kore 5 21 53 74 sore 69 24 93 90 77 167 In extreme cases, the antecedent of so is found in the adjacent noun phrase in coordinate structures, e.g., "Aru Kapone to sono i t i m i , " "Al Capone and his gang." It is impossible to substitute ko for so here. 108 [2] 1) The type of the antecedent Ko has a sentential level antecedent as wel l as a word level one, while so prefers to have a word level antecedent. sentential (clause, sentence) word kore (wa/ga) 6 20 32 52 sore (wa/ga) 4 35 39 24 67 91 ii) The size of the antecedent Ko can refer to a fairly large portion of discourse, as shown below: number of sentences 2 3 4 5 6 12 20 26 kono/UP 0 0 2 1 0 0 1 0 4 kono yoona/ 6 3 1 1 1 0 0 0 12 koo iu koo/kono 1 1 1 1 2 1 0 1 8 yooni 7 4 4 3 3 1 1 1 24 On the other hand, it turns out that the antecedents of so are no more than single sentences. Besides [1] and [2] above, the following are additional interesting characterist ics of the data: 109 Postcedent: Ko can have a postcedent as wel l as an antecedent.7 Ex.: ...tugi'ni kore mo yooi ni dekiru pasu no hoohoo dearu. next this too easy can pass method Sore wa dai-hati-zu no yooni, hidari te o ue ni muke, it Figure 8 as left-hand upward put kahei o oyayubi to nakayubi to de sasaeru. Soko e... coin thumb and middle finger wi th hold there hoka no yubi wa kahei no ue o oou no dearu.... other finger coin ' on cover (Mazikku Shooziten) 'Next, this is also an easy method for a pass. It is, as Figure 8 shows, that you put your left palm upward and hold the coin with your thumb and middle finger. There,... cover the coin wi th the other fingers.' Reflexive Reference: Ko can refer to itself, i.e., the reflexive usage. Ex.1 Kono bun no kono gozyun wa bunpoo ni this sentence this word order grammar kanatte-imasu. agree 'This word order in this sentence is grammatical.' 1 10 Ex.2 Katute no kimi no tomo Kibi deatta kono zibun. former you friend this self (kono = katute no...zibun) (Nakazima: Sangetuki) 'I (self) who was once your friend Kibi. ' [5] Negation: So can refer to a statement embedded in the negation, but ko cannot. 8 Ex. Kore wa hannin dewa nai. Sonna/*konna bakana he offender is-not that this absurd koto ga aru hazu ga nai. thing never can be 'He is not the murderer. Such an absurd thing cannot be the case.' (cf. Kare wa hannin nanka-zyaa-nai. Koo tubuyaite-mite in this way mutter mo nanni naroo.) no use 'He is not the murderer. It is no use to mutter in this way.' Let us examine the implicat ions of these findings. Fi rs t , the demonstratives in discourse use are not straightforwardly related to the superficial ( = physical) distance in discourse; i.e., it is not the case that the antecedent is closer to the demonstrative ko than the demon-strative so in the temporal order in discourse. 9 Next, the deictic nature 111 of ko is strongly exhibited in the data. That is, the reflexive use in [4] is one thing, and the postcedent in [3] is another. In the latter, this type of use implies that what the speaker wants to say is already in his mind preceding i ts appearance in discourse: in other words, i t supports the assumption that he actually refers to something which exists in his mind in cases involving ko. In the former, Example 1 is a clear-cut case of the reflexive use of the deixis to the "meta-sentence" of the actual sentence. The use in Example 1 lends great support to that in Example 2, where ko refers not to the expression but to the actual existence in the f ic t ional wor ld created by the discourse. Let us now consider the negation, [5]. A negative expression has, as a matter of course, i t s affirmative counter-part. According to Ishigami (1983), the difference between affirmative sentences and negative sentences is based on the difference in how to choose (the direct) material for one's utterance. Ishigami i l lus t ra tes this point using examples like that reproduced in (41), and the figure duplicated here with my own interpretations as (42): (41 )a. Shingoo wa akai signal red 'The signal is red.' b. Shingoo wa aoku-nai. signal NEG 'The signal is not green. (42) [affirmative] da (it is) 1 12 (d i rect mater ia l ) object (observed fact) (s ignal i s red) (s ignal i s red) speaker-speech act [negative] nai (it is not) (d i rect mater ia l ) object (observed fact) (s ignal i s green) (s ignal is red) speaker • speech act 1 13 Ishigami argues that in (41a), the direct material is [shingoo ga akai (koto)] "(the fact that) the signal is red," which matches the "reality" which the speaker sees in front of his.eyes. On the other hand, in (41b), the direct material is [signal is green], which the speaker gets not from the observation of "reality" but his (subjective) choice, and this material contradicts what he actually sees; then the utterance [signal is not green] is obtained. This argument i t se l f is established on the basis of a metaphysical conception. The interesting point given my present concerns is: the negative sentence expresses the negation of the existence of what the speaker conceptualizes. In Chapter 5, a sentence which contains a certain use of the particle ga, which is called neutral description (Kuno, 1973) or thetic sentence (Kuroda, 1965 and else-where) w i l l be introduced, and Ishigami's theory of negation can explain why this type of sentence is very diff icul t , if not impossible, to negate (a fact pointed out by Kuroda and others). That is, a negative utterance is not a direct observation of the outside world. Another important point is that Ishigami's schema about negation claims that f i r s t you have to conceptualize something to exist as the direct material for i ts negation. In light of this, the fol lowing con-ditional case (43) could be given a s imi lar explanation: 1 14 (43) Moshi kyoo no gogo teiden ni nattara, sonna/*konna if today's afternoon power-failure this koto wa nai to omou ga, sugu kenkyuuzyo ni thing not think but immediately laboratory hatudenki o karini ikinasai. power-generator borrow go 'If you have a power failure this afternoon - I don't think that w i l l be the case - go immediately to the laboratory to borrow a generator.' Koo refers to the subjunctive idea [kyoo no gogo teiden ni naru ], not [Mosi kyoo no gogo teiden ni nattara ], which includes the [if ] frame. Keeping this point in mind, I w i l l examine the ko- antecedent relation in the following. Firs t of a l l , what is the implication of the tendency that ko has to refer to larger chunks of discourse, at the least sentential s tate-ments (cf. [2])? Compare (44b) and (44c) wi th (44a): 1 15 (44)a. Tantyoosa ni tomonau aki wa yoku hiroo monotony follow boredom often weariness to taihi sareru. ( = (1) in this chapter) be-compared 'Boredom caused by the monotony (of tasks) is often compared with weariness.' b. Kore wa tugi-no-yoo na riyuu ni yoru. this following reason 'This is for the following reason.' c. Sore wa tugi-no-yoo na riyuu ni yoru. this following reason 'This is for the following reason.' In (44c), sore can be replaced by [tantyoosa ni tomonau aki ga yoku hiroo to hikakusareru no/?koto 1. while in (44b) kore is replaced by [tantyoosa ni tomonau aki wa. yoku hiroo to hikakusareru to-iu-no/koto 1. No is the nominal izer in Japanese; thus, although so refers to the previous sentence, precisely speaking so refers to a nominal concept. On the other hand, ko refers to a sentential concept, i.e., to is a quotative marker and toiukoto is different from the simple koto . 1 0 (Notice that the particle wa is used in the latter, but i ts place is taken by ga in the former, since wa cannot occur in NP according to the general rules for 1 16 its use.) The levels of conceptualization seem to be different in ko and so. Next compare ko in (45): (45) Zyagaimo to ninzin o sainome ni kitte, tyotto potato carrot small cube cut a l i t t l e yawarakai-gurai ni yudete, kore o guriinpiisu to soft boil this green peas mazete-kudasai. mix 'First cut potato and carrot into small cubes and boil them until a l i t t l e tender, and mix this with green peas.' Although the referent cannot be pinned down in a l inguist ic form, i.e., in a noun form (cf. this is accusative in the fol lowing sentence), it is obvious that the referent is something which is prepared in the way described in the preceding sentence: the diced and boiled potato and carrot. This is a good example of how deixis works in discourse. The ko- referent is in the speaker's terr i tory or in the ' reflection' of his inner world. It must have some reali ty at least in his mind, though it does not necessarily have to exist for other people. On the other hand, in (written) discourse, the wri ter tries to relate his thought to the readers (including the wr i t e r himself). In (45), the previous discourse has established such an existence as the referent: the referent's existence is inferrable for the reader from the previous discourse. The reader has 1 17 received sufficient information from the previous discourse to figure out what is in the speaker's mind at the moment in question. It is often the case that the ko- reference is hard to pin down as a l inguist ic form, a word, a phrase, or a sentence in a discourse (Ono, 1975 and 1977); and as Kuno (1973) points out, the referent has a "vividness" in context. Although it is not impossible for it to refer to some concrete object mentioned in discourse, the ko- referent normally takes a sentence or sentences. A l l these facts indicate that ko refers to some imaginary existence evoked in the context. 1 1 Thus it takes a good deal of discourse (at least a sentential expression) to establish a plausible referent in context. Negative and conditional clauses are either to negate or to suppose a certain existence; the imaginary existence of something is the foundation upon which these expressions are based, and ko can refer to such kinds of existences as wel l . Thus, I support the contention that ko is always deixis, regardless of whether it is used in an actual pointing manner or used as the discourse anaphora. Now then, what is the difference between ko and so ? In (45), kore can be replaced by sore but it lacks the "vividness" which is attached to the ko- use. Roughly speaking, the anaphoric use of so can be described as "what l/you have mentioned in the previous discourse" (Kitagawa (1978) assumes that Grice's "be relevant" is operating here); while ko is described as a discourse deixis. It is, in fact, possible to use so to refer to a statement including a negative or an /'f-frame, as in (46) and (47), as wel l as excluding them (cf. Ex. in [5] and (43)): 1 18 (46) Kare wa hannin dewa nai. Sonna/*Konna uwasu wa he murderer is not that/ this rumor sinzite inai. Tada mondai wa... (I) believe not only problem 'He is not the murderer. I don't believe that (such a) rumor. But the problem is...' (47) Mosi kesa nesugosazu ni, ano hikooki ni if this morning without-oversleeping that airplane notte-itara.... soo/koo kangaeta dake de hiyaase ga took in this way think only cold sweat nizimideru. break out 'If I hadn't overslept this morning and had taken that airplane...the very thought makes me break out in a cold sweat.' The difference in ko/so in (47), for example, is that while soo kangaeta dake de can be easily replaced with [mosi...tara ] to kangaeta dake de, what koo refers to is a hypothetical situation, "I had not overslept this morning and had taken that airplane" but not the expression [mosi...tara ]. 1 19 Now, we can readily understand why there are many conjunctive words of so origin in Japanese. Observe (36), (37), and (38), which are dupli-cated here as (48), (49), and (50) respectively: (48) Sore wa kahun-bunseki to iwareru hoohoo de kahun no that pollen-analysis is -cal led method pollen naka no keisi tu no bubun ga itai tosite nokori, sono syurui, in s i l i c a portion remain as leave that type ryoo kara kako no syokusei no hukugen o quantity from past plant-distribution reconstruct s i , sore o toosite kikoo no hensen o siru toiuu that through climate change know mono de aru. (Sikata: Yoozo no kigen) 'That is a method which is called pollen analysis. The s i l i c a component of the pollen is left as remainder, and from that type and quantity we restore the plant d i s t r i -bution in the past, and through that we come to know the changes in the climate.' (49) Naisin de motiron desi no soo ga zibun o in one's inmost heart of course disciple monk toki-husete kono hoo o kokoromi-saseru no o matte ita convince this method make-(him)-try wait 120 no de aru. Desi no soo nimo Naigu no kono sakuryaku ga disciple monk too this strategy wakaranai hazu wa nai. Sikasi sore ni taisuru hankan must-know but that against i l l - fee l ing yori wa sooiu sakuryaku o toru kokoro-moti no hoo ga than such strategy take psychology yori tuyoku kono desi no soo no doozyoo o ugokasita more strong this disciple priest sympathy appealed to no-de-aru. it w i l l be the case that (Akutagawa: Hana) 'In the depths of his heart, of course, he was waiting for his disciple monks to convince him and make him try this method. His disciples must have sensed this (his) strategy. But, more than the i l l - fee l ing towards that (such a) strategy, it would be the case that Naigu's psychology, which made him think in that way, appealed to the sympathies of this disciple.' (50) Naigu wa hazime, kore o zibun no kao-gawari ga sita at f i rs t this his face-metamorphosize sei da to kaisyaku-sita. Sikasi doomo kono because of interpret but somehow this 121 kaisyaku dake de wa zyuubun ni setumei ga interpretation only adequately explain tsukanai yoo-de-aru. Motiron, Naka-doozi ya Ge-boosi ga of course and warau genin wa soko ni aru ni tigainai (ga).... laugh cause there was must (Akutagawa: Hana) 'At f i rs t Naigu interpreted this to mean that it was because his face had undergone a metamorphosis. It seemed that this interpretation as wel l could not explain things adequately. Of course, the reason why Nakadoozi and Geboosi laughed was there, (but)....' In (48), ko would also be possible (providing so was substituted for ko throughout), but the use of so imposes an objective tone on the style, referring only to the informative aspect of the information carried in the previous discourse. In (49), if the viewpoint of the utterance which contains kono is a l i t t l e bit inclined towards Naigu, who has the strategy in his mind; on the other hand, in the last utterance, which contains sooiu, it is a l i t t l e bit towards desi no soo. If sooiu were to be replaced by kooiu, sakuryaku would be given a more prominent central topic status in this discourse, which cannot be the case. In (50), kono and soko are used in a s imi lar way; also, if soko were to be replaced by koko, the sentence would lose i ts tone of objective judgement. I w i l l summarize the differences observed in the above examples as follows: 122 (i) So is related to a certain mechanism which operates to pass information from the previous discourse to the current discourse. A certain objective tone is imposed on this use, since the referent is not in the speaker's current area of concern ( = territory). (ii) Ko is related to a certain mechanism which indicates where "we" ( = speaker and hearer, hearer, or wr i t e r and reader) have shared an experience through an imaginary scene built on the basis of information given by the discourse, manufactured by the use of "here and now" deixis ko. More sample discourses showing this type of "spotlighting" ko use w i l l be found in the Appendix. With this taken care of, the next section w i l l be concerned with some residual problems connected wi th so, which has a more compl i -cated nature than ko. §3.2.3 Problems in the Use of So In this section, the d i s t inc t ion between demonstratives and pronouns w i l l be discussed in terms of the use of so. In §3.1.0, the correspondence between that in English, a language which has a twofold demonstrative system, and so and a in Japanese wi th i t s threefold demonstrative system, was touched upon. The Japanese so, in turn, can apparently be translated into not only that but also it. Compare (51) and its Japanese counterpart, (52): 123 (51 )a. What is that on Ms. Kimura's desk? b. That/It is a stuffed bird from South America, I heard. (52) a. Kimura-san no tukue no ue ni aru sore wa nanda? b. Sore/0 wa nanbei no tori no hakusei da soo desu. Japanese is poor in third person pronouns, in contrast wi th a very rich demonstrative system, makes the situation a l i t t l e complicated. Kare/ kanozyo/karera "he/she/they" are newcomers to the language, which occas ional ly replace zero pronouns, and English it is generally translated using sore or a zero pronoun. (52) is a case of either that/it = sore or that = sore/it = 0 in the translation. Then is sore used as a pronoun or a demonstrative? Or is it the case that the function of sore does not exactly f i t into the that vs. it d is t inct ion in E n g l i s h ? 1 2 In what fol lows, the nature of so w i l l be examined in comparison with pronouns. One of the characterist ics of pronouns is their bound variable use (Evans (1980), Partee (1982), among others). So, unlike ko and zero pronouns, seems to have this use: (53) Kono mati dewa, subete no tatemono ga sono/0/*kono this town in every building rekisi 0 monogatatte-iru. history tel l "In this town, every building te l ls i ts own history.' 124 However, also observe the sl ightly different cases in (54): (54)a. Kono naka kara dore-demo suki na burooti o erande, this among whatever like brooch choose sore/0/*kore o konya no paatee ni site iki-nasai . that tonight party put go 'Choose whatever brooch you like of those, and go to tonight's party wearing it.' b. Kyoo kore kara depaato ni itte nanika today now department store go something sagasite-mimasu. Mosi yoi mono ga mitukattara, try to look for if good one find sore/0/*kore o Haruko-san e no purezento ni simasu. that present make 'Now I am going to the department store to look for some-thing (for that purpose). If I can find something good, I w i l l make it a present for Haruko.' 125 c. Kondo wa Kyooto no yukigesiki o tori tai ne. Sore/ 0/*Kore next time snow-scene take a picture that o Afurika ni iru Kyooto-umare no tomodati ni okuttara Afr ica is born friend send-if yorokobu-daroo naa. be-glad-wi l l 'Next time, I want to take pictures of snow scenes in Kyoto. If I send them to my friend from Kyoto who lives in Afr ica right now, he w i l l be delighted.' In these examples, sore does not refer to specif ic objects. These examples remind me of the E-type pronoun in (55), which are (6) - (8) in Evans (1980): (55) a. Few M.P.s came to the party but they had a good time. b. Few congressmen admire Kennedy, and they are very junior. c. John owns some sheep and Henry vaccinates them in Spring. In these examples, the pronoun is not bound by the quantifier. According to Evans, in (55c), for example, the sentence entai ls that "Harry vaccinates all the sheep John owns" (p. 339 ) , unlike the quantifier binding case (56): (56) Some sheep are such that John owns them and Harry vaccinates them in Spring. 126 Evans treats these cases in (55) as referential, although the pronouns do not refer to specific entities. Now, let us assume that the demonstrat ives are a lways referential regardless of whether they are in the deic t ic use or the anaphoric use. Ko does not have much trouble in adapting to this assumption, since i ts referential nature has been made clear in the discussions in the previous sections. In the case of discourse anaphora, the speaker's territory is formed in the speaker's inner realm, which has i ts ref lec t ion in the discourse. Then, so is defined as something outside of the speaker's territory: the referent is not the speaker's "now and here" but "then and there." If so, there already ex is t s a good background for so as wel l to be considered as referential a l l the time. Maclaran (1982) suggests adapting the idea of discourse referent in Karttunen (1976) for the non-specific referent cases. First look at (57), which is Karttunen's (34): (57)a. Harvey courts a girl at every convention. b. Most boys in this town are in love with a go-go dancer. The sentences (57a) and (57b) are both ambiguous between the specific reading and the non-specific reading. However (58), which is Karttunen's (35), has only the specific reading: 127 (58) a. Harvey courts a girl at every convention. She is very pretty. b. Most boys in this town are in love wi th a go-go dancer. Mary doesn't like her at a l l . Here, indefinite NPs "establish" discourse referents for the pronoun to refer to. Thus, although there is no specific fish which exists in (59), it can refer to it, since the imaginary discourse referent does exist in the speaker's mind, and thus supposedly in the hearer's mind as wel l : (59) B i l l wants to catch a fish and eat it for supper. The discourse referent exists, however, in a modal bound fashion; the referent is a temporary one. McCawley (1979) argues for Karttunen's proposal. He suggests taking the universe of discourse as time-dependent and time branch-dependent in order to avoid the universal quantifier reading for (60) ( = McCawley's (18d)): (60) If bl isters develop on the patient's body, you should bandage them. (For every bl is ter X, if X develops on the patient's body, you should bandage X.) 128 Thus, "bl is ters develop on the. patient's body" in (60) is causing a "temporary addition of a corresponding discourse referent to the context domain" (p. 384). If Kartunen's discourse referent is a step in the right direction, the original de ic t ic nature of demonstratives can ful ly u t i l i z e i ts referential function in discourse, where successive sentences may be bound by certain modals, mood, tenses, aspects, and so on, to create a certain imaginary world. This can be supported by the claim made by Stechow (1982) and others that the demonstrative entails the existence of the referent. If so, the level where this type of referential mention is descr ibed 1 3 is not in LF but in LF' . Now, although Maclaran (1982) concludes that there are no genuine bound variable cases in English demonstratives, Japanese sono seems to differ in this respect. Unlike ko in kono, so in sono in general is in fact ambiguous between the following two cases in (61): (6 Da. Rei no tukue ni hikidasi ga tui te- i ru ne. Sono hikidasi in question desk drawer there-is that drawer no naka o mite-goran. inside look 'You know that/the desk which has a drawer. Look inside of the/that drawer.' 129 b. Monooki ni tukue ga simatte-aru ne. Sono hikidasi storage room desk store that drawer no naka o mite-goran. inside look There is a desk in the storage room. Look inside of i t s / (the) drawer.' (61a) is a typical so use in discourse, but in (61b), sono is being used in the sense of sore no "that 's," or perhaps closer to the "it 's" in English. Sono is potentially ambiguous between these two readings. However, the bound variable case (53) has only the use of sono found in (61b). In English there is no possessive case for demonstratives, although there are determiners. What is so in (53) then, in terms of i ts function, a demonstrative or a pronoun? In relation to this question, the next example (62) provides some interesting evidence: (62)a. Aru kapone to sono/0 i t imi . 'A l Capone and his gang.' b. Yosiko ga sono/0 hansei o okutta mura. hal f -a- l i fe spent vil lage 'The village where Yoshiko spent half of her life. ' 130 c. Sono/0 kodomo-tati ga sonkei-s i te- i ru t i t ioya children respect father 'The father whom his children respect.' In these cases, the so-that correspondence is ruled out. So, whose referents are non-human otherwise, apparently works like the personal pronouns here, or at least is used like "pro-forms." Notice in (62c), sono is cataphoric, which is normally not allowed for ko. Both the demonstrat ives and the pronouns can "stand for" expressions as a part of their functions; thus, the repetition of the same expressions may be avoided in this way. A typical example for this use of the demonstratives is given in (63), examples of which are not rare in certain styles of writ ing: (63) Hitori damatte tabi nideru, Taroo wa sonna otoko da. alone quietly leave for a trip such a man 'Quietly leaving for a trip alone, Taroo is such a man.' If i t is the case that the use of sono in (53) is the same as the usage exhibited in (62) - supposing some uses employ only the "stand for" function of demonstratives - sono is not referential there but rather to be identified wi th the bound variable use. So far, there is no conclusive evidence bearing on this dispute concerning the non-referential use in demonstratives (also cf. Stechow (1982)). What is clear so far is that 131 the way pronouns and demonstrat ives 1 4 refer is different, although both stand for the previous expression, as in (64): (64) He w i l l stay. I am sure of it. What a mistake. I am sure of that. (N. Parton, Vancouver Sun ) The difference in the ways a pronoun and a demonstrative refer is that the former "stands for" the NP in the simplest way; i.e., instead of using the same NP, the pronoun which has the same referent as the NP is used; the la t ter "stands for" the NP which is defined or identif iable in a discourse. In other words, in the case of demonstratives, what it stands for in (65) is not just onion, but onion which was stir-fried until its sweet flavor came out or onion which is in a certain state after the preceding cooking procedure. On the other hand, in the case of a pronoun, what it stands for is onion regardless of whatever state it is in now. (65) Tamonegi o zyuubun-ni amami ga deru made onion ACC well sweet flavor NOM draw until itamete-kudasai. Sorekara, sore o suupu no nabe ni s t i r - f ry then ACC soup 's pan to utusite-kudasai. transfer 'S t i r - f ry the onion until i ts sweet flavor comes out. Then, transfer that / i t into a soup pan. Sore in (65) can have both the demonstrative reading and the pronoun reading. Considering the fact that there are bound variable cases in (53) 132 and non-referential cases in (62), and the observation that it is usually translated wi th sore, it seems to be the case that the function of sore covers that of a demonstrative and a pronoun. 1 5 §3.3 Conclusion In this chapter, a good deal of space has been devoted to d i s -cussing the characteris t ics of Japanese demonstratives. The reason is that although various useful observations and assumptions have been put forward by many researchers, these mainly concern spatial deixis; there are quite a few noteworthy proposals about discourse use, but they are usually l imited to dialogue. The framework of the characterization has mainly been constructed on the basis of spatio (-temporal) deixis. Meanwhile, the observations presented by l inguis t s and often by rhetoricians about so-called anaphora demonstratives have succeeded in capturing certain aspects of demonstratives, but in the delimitat ion of the contextual environment. My main concern, apart from the nature of the demonstrative, is the use of demonstratives as one apparatus of discourse. It seems to me a reasonable procedure to take a f i rs t step towards the goal by getting a consistent view of the essential nature of demonstratives; otherwise, certain factual, character is t ic uses might be mistakenly considered as essential, when in reali ty such uses were of an a posteriori nature, manifested as a result of the influence of certain factors originating in the way demonstratives are employed. If the character is t ics of each use (spatial , temporal, and so on) can be referred back to some more basic nature (which is actually the case), a consistent view of the demonstratives in a l l spat io-temporal and 133 discourse uses w i l l have been obtained. It cannot be denied that the framework presented in §3.1 concerning such a basic nature contains some speculative assumptions. However, I believe that this is a viable attempt to deduce a s ignif icant relationship between 'facts' in the actual use of demonstratives; and that on the basis of such a framework, the "anaphora" use of the demonstrative can be reexamined. The difference between pronoun and demonstrative has also been discussed. In conclusion, anaphoric demonstratives are a l l referential by nature (except some uses of sore which should be handled in sentence grammar) and they can be bound in discourse (LF') but not in a sentence (LF). 134 Notes (Chapter 3) 1 Sore is used in the original text. 2 Ka is the old form of a. 3 * is used as a marker for "bad" or impossible formulations regardless of whether these are syntactic or not. 4 [Data corpus] (A) Hideo Suzuki (1974), "Hyogaki no Kankyo Fukugen," Kagaku vol. 44-7. (B) Daiichiro Sugimoto (1974), "Taiyo Nyutorino," Kagaku vol. 44-7. (C) Kazutaka Kobayashi (1974), "Tanch! na Shigoto no Dotai," Kagaku vol. 44-7. (D) Marikazu Shikata (1974), "Yojo no kigen to kakuritsu-katei," Seibutsu-kagaku vol. 26-2. (E) Shozo Osawa (1 965), "Ribosomu no Sei-gosei," Seibutsu-butsuri vol. 5-4. 5 Kore and kono (NP) behave differently: kono can refer farther than kore because of the (partially) identical NP of the antecedent which immediately fo l lows kono. Considering this extra factor, the comparison is made on the basis of the -re (noun) forms. 6 Kore/sore (noun form) which is used as the topic or the subject is chosen for the comparison. 7 Sentence (i) seems to be a case where, at the f i r s t glance, the postcedent is the whole fol lowing story. However, this style is rather new in Japanese and it is said to have come into use under the influence of the dummy subject construction in English: (i) Sore wa ame no s i to-s i to huru hen ni hito no that / i t rain quietly fall in a strange way one's 135 ki o meiraseru yoona ban desita. spir i ts depress evening 'It was an evening when the quietly fall ing rain depressed one's spi r i t s in a strange way' This claim may sound a l i t t l e too strong, since ko is used in s imi lar sentences apparently referring to non-negative portions of s tate-ments. However, the point is that konna cannot be used in the context of Ex. in [5], or, putting it in a different way, if konna is used, it refers to the whole event of his being accused as a murderer but not the previous statement by i tself . The fol lowing example permits a ko which refers to the statement including the negative: (i) Kare wa hannin nanka-zyaa-nai. Kore/sore wa hakkir i -s i te- i ru . 'He is not the murderer. This is for sure.' In the above (i); for example, once kore is changed to kono zizitu "this fact," sono zizitu is ruled out. The level of reference in the conceptualization of ideas seems to differ in ko and so. In German, the distance component of jener and dieser can be pinned down in discourse context according to the temporal order, l ike Japanese zensya 'the former' and koosya 'the latter'. As for the function of to/no/koto cf. McCawley (1978), Josephs (1 976), Watanabe (1 989), among others. I use the term 'discourse' in opposition to 'sentence,' and use 'context' as referring to the contextual aspect of discourse. It cannot be used deict ical ly (observation from M. Rochemont). M. Rochemont suggested to me that Discourse Representation Theory developed in Heim (1982) might be useful to describe this type of phenomenon for demonstratives. The demonstrative expressions like "this man" are compared with the definite expression "the man here" in Stechow (1982). According to Stechow, the former is directly referential while the latter is not, 136 and this difference is evident in the interpretation of the following example, (i) ( = Stechow's (4) and (5)): (i)a. Next year, I w i l l consult this fortune-teller. b. Next year, I w i l l consult the fortune-teller here. In (a), "this fortune-teller" means Mr. So-and-so whom the speaker can point out, whereas in (b), "the fortune-teller here" can mean the one who w i l l be found here next year, whoever he is, in addition to the interpretation as (a). 5 The alternative view is that there are two kinds of sore : one is a demonstrative, the other is a pronoun. 137 Chapter 4 Quantifier Reading §4.0 Introduction The third disambiguation mechanism in discourse, in addition to zibun coreference and demonstratives, is the quantifier reading in Japanese. This problem has been examined in syntax in connection with the formalism of the quantification in formal logic, but l i t t l e extra-syntactic work has been carried out. The purpose of this chapter is to determine factors which control the pragmatic aspects of quantifier reading, that is, why a certain reading is prominent and why one reading is chosen over another. §4.1 Quantifier Interpretation §4.1. i Quantifier Reading in Japanese A quantifier sentence like (1) is generally viewed as potentially ambiguous between (1 b) and (1 c): 138 (Da. Everyone admires someone. b. (Vx) (Ey) A D M I R E (x, y) c. Gy) (Vx) A D M I R E (x, y) May (1977) proposes two distinct LF representations for (1), which are created by two different operations on Quantifier Raising. However, May (1985) proposes a Scope Pr inciple which controls the interact ion between two operators. In either way, the ambiguity is captured in syntax and i ts disambiguation process is left for an extra-syntact ic component. On the other hand, examples l ike (2), which is a Japanese equivalent of (1), is considered as unambiguous1 by Kuroda (1970) while i ts scrambled version is ambiguous. This view is adopted by current GB literature like Hoji (1985 and 1986). (2)a. Daremo ga dareka o sonkeisiteiru (=( 1 b)) everyone NOM someone ACC admires 'Everyone admires someone.' b. Dareka o daremo ga sonkeisiteru. (=( 1 b)/( 1 c)) 'Someone, everyone admires.' Furthermore, Hoji (1986) gives the following paradigm (3). 139 (3) a. *QP-ga WH-o V b. WH-Oj QP-ga t i V (unambiguous: WH takes wide scope) c. QP-ga QP-o V (unambiguous: QP-ga takes wide scope) d. QP-Oj QP-ga t \ V (ambiguous) Assuming May's (1985) Scope P r inc ip l e , Hoji adds some generalizations (4) particular to Japanese to account for (3). (4) a. Scope interpretation is more restr icted in Japanese than in English in the sense that it reflects more closely the hierarchical relation at S-structure. b. In Japanese the wh-phrase always takes wide scope over a non-wh quantifier (but English does not have this restriction), [p. 9] Now, suppose syntax takes care of the scope interpretation in Japanese as in the way Hoji c la ims or something s imi la r to that, and feeds the information to semantics, the questions that w i l l arise are: i) why the scope reading is more or less close to the hierarchical relat ion at S-structure; i i ) why certain QPs tend to have wide scope; i i i ) how the ambiguity represented in (3d) is disambiguated. As for i i i ) , it is not the problem of syntax; for i) and i i ) , syntax offers syntact ic sides of i ts explanations but they are not the whole story. In what fol lows, the problems of quantifier interpretations w i l l be examined in order to answer the above questions. 140 §4.1.2 Linear Order Hypothesis and Subjecthood The phenomenon observed in (2) is interpreted in two ways: i) the word order affects the quantifier reading; i i ) subject NPs get wide scope reading. The f i rs t view is taken by Kuno (1973a, and the revised version in 1973b); the second one is taken by Keenan (1976) as one of his subject properties. F i rs t let us examine Kuno's claim concerning the linear order hypothesis (5) for Japanese. (5) Rule 1: In the basic word order structure, i f two quantifiers appear in the order of Qi - Q2, the obtained reading is the "same" vs. the "different." Rule 2: If the order of Q, - Q2 is changed into Q 2 -Qi as the result of leftward movement, the obtained reading is the "same" vs. the "same." Rule 3: If the order of Qi - Q2 is changed into Q2 -Qi as the result of rightward movement, the obtained reading is the "different" vs. the "same" - the interpretat ion of each Q is unchanged. [summary translation from Kuno 1973b; see also Kuno 1973a, footnote 33, p. 384,] Kuno's terms "same" vs. "different" roughly overlaps the terms "wide" vs. "narrow" but they are not exactly the same. They are in a sense hybrids of wide/narrow and group/distributive. This is especially true in Rule 2. Since Kuno's functional grammar does not differentiate syntactic 141 factors and nonsyntactic factors, the above rule set is better understood as more discourse oriented readings than just the wide/ narrow scope interpretation. Let us examine how Kuno's system operates. According to this rule set, the surface linear order rule which assigns the "same" vs. "different" readings for Qi - Q2 is only applicable to cases of basic word order. Rule 2 accounts for the "same" vs. "same" reading in scrambled sentences like sentences (b) in (6) and (7 ) , in oppo-si t ion to the corresponding basic word order sentences ( 6 a ) and (7a) . (6) a. Gonin no onna ga sannin no kodomo 0 sodateta. five women (nom) three child (acc) raised [same] [different] Five women raised three children.' b. Sannin no kodomo 0 gonin no onna ga sodateta. [same] [same] (7) a. Syoosuu no ningen ga takusan no kaisya 0 a few people (nom) many company (acc) [same] [different] gyuuzitte-iru. have under one's control 'A few people have many companies under their control. 142 b. Takusan no kaisya o syoosuu no ningen ga gyuuzitte-iru. [same] [same] Kuno claims that Japanese has the rightward movement for sentential adverbs, as wel l as the leftward movement in scrambling, as in (6) and (7). Example (8) (Kuno's (34), 1973b) represents such cases, and is accounted for by Rule 3: (8) a. Mainen oozei no kankoo-kyaku ga nihon ni kimasu. every year many tourists (nom) Japan to come ([same]) [different] (mainen = S-adverb) 'Every year many tourists come to Japan.' b. Oozei no kankoo-kyaku ga mainen nihon ni kimasu. many tourists (nom) every year Japan to come [different] vs. ([same]) (mainen = S-adverb) [same] vs. ([different]) (mainen = VP-adverb) 'Many tourists, every year, come to Japan.' Kuno argues that an example like (8b), in fact, represents either of the fol lowing two structures: one, the sentence is derived by moving the sentential adverb to the right. Rule 3 in (5) predicts that, unlike l e f tward movement, r igh tward movement does not affect the interpretation of the quantifier. This is why the "different" reading is obtained for many tourists. Two, sentence (8b) represents the basic 143 structure, i.e., every year is a VP-adverb. Therefore, the "same" reading is obtained for many tourists by Rule 1 in (5). Thus, Kuno contends that not only is the surface word order a relevant factor in determining the quantifier reading, but also the direction of movement. He also contends that the crossing is another factor. Let us examine this conclusion next. Japanese passives l ike (2) is treated as "a basic word order sentence as far as the quantifier reading is concerned" (Kuno 1973b, p. 273), which gives an ad hoc appearance in a Standard Theory framework such as that which Kuno adopted, especia l ly when his lef tward movement rule for the quantifier interpretation handles English passive cases in the way i l lustrated in (9) ( = Kuno's (14) in 1973b): (9)a. Every girl likes many bovs. Q1 ° 2 [same] [different] b. Many boys every girl likes. Q, Q ' 2 i [same] [different] 144 c. Many boys are liked by every gir l . Q2 Qi [same] [different] Here, the leftward movement creates the [same] vs. [different] reading, which contradicts Rule 2. This poses a minor problem to Kuno's contention. However, in current GB-type generative grammar, this is no problem at al l since passive sentences are generated as in (10) below, and Rule 1 in (5) accounts for the [same] vs. [different] reading in both Japanese and English. (10) A are liked many boys by every gir l . Furthermore, in (8), sentence (8b) is not necessarily considered as a case of the r ightward movement. Many tourists in fact is ambiguous, not only in (8b) but also in (8a). Observe example (11): (1 1 )a. Mainen oozei no kankookyaku ga nihon ni kimasu. every year many tourist (nom) Japan to come [same] vs. [same] (mainen = VP-adverb) b. Oozei no kankookyaku ga mainen nihon ni kimasu. many tourist (nom) every year Japan to come [same] [different] (mainen = VP-adverb) 'Many tourists come to Japan every year.' In the case of VP-adverbs, the [same] vs. [same] reading 2 in (1 la) (=(8a)) can be obtained by preposing mainen. Thus, whi le (8a) ([same] vs. 145 [different]) is a sentential adverb case, (11a) ([same] vs. [same]) is a preposed VP-adverb case. If this is so, then Rule 3 has to be dismissed from the rule sets in (5). In other words, the principle of quantifier interpretation can be stated wi th fa i r ly simple rules: Qi - Q 2 gets the [same] vs. [same] reading after scrambling by leftward movement; otherwise, sentences get the [same] vs. [different] reading. Next, let us consider Keenan's c la im about the wide scope property of the subject element in connection wi th the linear order hypothesis. Keenan's c laim is generally supported, except when a sentence contains a sentential adverb like (1 1 b) or Locative as in (12b): (12)a. Itutu no paatii de nanninka no kyaku ga five party at some guest (nom) [same] [different] dontyan-sawagi 0 sita. went-on-a-spree 'At five parties, some guests went on a spree.' b. Nanninka no kyaku ga itutu no paatii de dontyan-some guest five party [same] [same] sawagi 0 sita. 146 The readings in (13) represent Kuno's contention with the understanding that the basic word order of Japanese existential sentence is L + S + V instead of S + L + V, thus assuming that(12b) is the preposed case of (12a). If this is so, the conclusion to be drawn from this data is that the wide scope reading is not the property of the subject but the result of the posit ion, i.e., the 'sentence in i t i a l position, ' since in (12a) not subject but locative gets the "same" (i. e., "wide" in this case) reading; that is, the leftmost quantifier, which in most cases, is incidentally the subject, always gets the "same" reading in sentences wi th the basic word order. In fact, th is seems to be a general iza t ion of the phenomenon. However, there is a question to be asked here, which is: why, then, the sentence in i t ia l position gets the "same" reading. Before exploring the answer for this question, the problem associated wi th wide/narrow scope interpretation w i l l be discussed in the next section. §4.1.3 Wide/Narrow vs. Same/Different Fi rs t let us examine Kuno's terms "same" and "different" more closely in order to determine the relationship between his terms and the more general terms."wide/narrow." 147 (13) Tasuu no otoko ga syoosuu no onna o homeru. many (pos) man (nom) a few (pos) woman (acc) praise 'Many men praise a few women.' (14) Syoosuu no onna ga tasuu no otoko ni homerareru. a few (pos) women (nom) many (pos) man by be-praised 'A few women are praised by many men.' The reading obtained for (13) is: each member of the "same" 'many' praises each member of the "different" sets of 'a few', as shown in (15a) and (15b). (15) a. a set of 'many' = (Bi l l , John,...Tom] a set of 'a few' = {'a few)', 'a few2 ' , . . . ' a few n '} a set of 'a f ewi ' = (Sarah, Jane,...Sue] a set of 'a few2* = (Julie, Mary,...Lucy] a set of 'a few n ' = (Nancy, Ann,...Karen] B i l l praises 'a f ewi ' John praises 'a f e w 2 ' a few women 149 On the other hand, in the reading in (14), each member of the same 'a few' (e.g., Karen, Julie,...Margaret) is praised by each member of the different sets of 'many,' as demonstrated 3 in (16a) and (16b). (1 6)a. a set of 'few' = {Karen, Julie,...Margaret] a set of 'many' = Cmanyi', 'many2'...'manyn'] a set of 'manyT = {Tom, Bill...Sam] a set of 'many2' = {John, Joe,...Adam] a set of 'many3' = {Jim, Ken,...Jeff] a set of 'manyn' = {Brian, Steve...Ben] Karen was praised by 'manyi' Jul ie was praised by 'many2' Margaret was praised by 'manyn' [the same 'few'] vs. [different 'many'] 150 Thus, what is clear is that Kuno's terms "same" and "different" are in fact referring to a set, and not an individual member. Now, the wide/narrow scope interpretat ion for the English equivalents of (13) and (14) is often described in a way somewhat like (1 7) and (1 8), respectively: 151 (1 7) Many men praise a few women. (Vx) (3y) PRAISE (x, y) (= For many men, there are a few women they praise.) (18) A few women are praised by many men. Oy) (Vx) B E - P R A I S E D (x, y) (= There are a few women such that they are praised by many men.) Furthermore, Kuno's "same vs. different" corresponds in a certain manner wi th loup's (1973) "col lect ive vs. individual." For example, the plural quanti f ier all, according to loup, gets the "individual" reading when it gets wide scope; i. e., "there can be as many garages as there are women" [p. 1 19]. On the other hand, when the singular quanti f ier gets wide scope, all gets the "col lect ive" reading, i.e., "women work together to build the garage." (19) A l l women built a garage. Using loup's terms, then (15b) can be explained in the fo l lowing way: each member of the "same" set of many men (= individual) praises "different" [A FEW WOMEN] (= set, not each member), i. e., there can be as many [A FEW WOMEN] as whatever numbers represented as many. A set is treated just l ike a singular quantifier. Now the problem is Kuno's "same" vs. "same" reading, 4 which is typical ly presented in subject-object scrambling cases l ike (20) (also cf. (6b) and (7b)): 152 (20) Syoosuu no onna o tasuu no otoko ga homeru. a few woman many man praise '(lit.) A few women, many men praise.' [same] vs. [same] If the "same" a few women, i . e., [A FEW WOMEN] (= set) is treated like a singular quantifier, then [MANY MEN] (= set) has to be "collective" as a set and obtains the "same" many men reading in (21): (21) r \ v J many men However, this type of "set" reading is not l ikely to be considered as a part of syntactic scope interpretation but it is rather taken as an extra-syntac t ic phenomenon. According to Contreras (1986), plural and 153 collective NPs have either "group" or "distributive" readings. 5 If so, it is the case that (21) gets the "distributive" reading, as wel l as the "group" reading shown in the i l lustrat ion, on the f i rs t quantifier. Likewise, the "group" reading, instead of the "distr ibutive" reading shown in the i l l u s t r a t i o n s , is also possible in (15b) and (16b). Given these assumptions, the wide/narrow interpretat ion in syntax, which is considered as ambiguous in this case, is s t i l l intact. So far, I have dealt with the question of how one interpretation of two quantifiers affects the meaning of a sentence which differs from the other interpretations involving (syntactic) scope interpretation and other non-syntactic factors. In the next section, I w i l l deal wi th the question asked earlier in this section, i.e., why does the sentence in i t ia l position get the "same" reading? I w i l l examine Zubin's 'focus' system assuming some sort of "centering" device ; is relevant to the interpre-tation of quantifiers. §4.2 "Attention" systems in discourse In Zubin's (1979) system, "focus" is controlled by the specific l imi ta t ions and biases of human perception. The speaker's "focus of interest," as Zubin ca l l s it, has at least two cognitive roots, Selective Attention and Egocentric Bias: 154 (22) SELECTIVE ATTENTION Every person has the following tendencies: 1. To focus interest on entities that are cognitively salient to him. 2. To focus interest on relatively few entities in the narrated scene, in comparison to the total range of entities available 3. To focus interest on one entity or on one set of homogeneous entities at a time. 4. To persevere in attention on one entity. 5. To reach a satiation l imi t where attention is shifted. [Zubin 1979: p.471] (23) EGOCENTRIC BIAS We process information about other human beings more readily than that about nonhumans, and we process infor-mation about ourselves most readily. [Zubin 1979: p.471] (24) EGOCENTRIC S C A L E speaker (ego) > hearer > other person > inanimate, concrete > abstract [Zubin 1979: p.478] Zubin attempts to draw a relationship between the cognitive bias of the semantic substance focus and the grammatical encoding. He concludes that nominative (grammatical subject) is used to indicate the 155 speaker's "focus of Interest" in German. Zubin claims that his focus-of-speaker's interest is different from the notion of 'topic' which tends f i rs t to be in an oblique case and then in the nominative; his 'focus' is always in the nominative from the beginning. And also, Zubin's 'focus' has nothing to do wi th voli t ional attention, in which "conscious effort" (to use Zubin's term) is drawn out from the hearer's side by the speaker. The distr ibutional difference among ' topic ' , 6 his 'focus', and 'conscious attention' is exhibited in (25) and (26): (25) Den Film habe ich schon dreimal gesehen. (top/acc) (focus/nom) 'The movie I have seen already three times.' (26) Mensch, ist der Josef scharf. Den mochte (top/attention/acc) ich mal besser kennenlernen. (focus/nom) (Attention is indicated wi th stress by Zubin.) 'Joseph is really something. Him, I would like to get to know better.' In real i ty , the dis t r ibut ion of Zubin's 'focus' largely coincides wi th that of 'topic' except in a marked structure l ike (25), since in languages which are not topic prominent, a nominative or (grammatical) 156 subject usually coincides wi th the sentential realization of a discourse topic. It is to be noted here that is referred to as "focus" in various theories, which is considered some kind of highlighting phenomenon, can be c lass i f ied into two categories: that of the preconscious attention system, like Zubin's, and that of the conscious attention system. As w i l l be shown in §4.5 Addendum, the conscious attention systems can be further divided into Grosz-type 'focus' which centers our attention on what the discourse is al l about, Kuno's "new" information-type 'focus' which draws our attention to what is new in the discourse, and Rochemont's focus, relevant to what is not "under discussion." According to Zubin, however, 'focus' is the grammatical encoding of " P R E C O N S C I O U S L Y disposed speaker's interest," which is a part of the cognitive property of the human attention mechanism. Although deictic use, word order, and stress are other devices in this mechanism by which a speaker's CONSCIOUS EFFORT at communication is conducted, but they are not related to Zubin's focus system. Now, I w i l l investigate the connection between subjecthood and Zubin's "focus of the interest." 7 First , the following two assumptions w i l l be taken as premises: (27) Assumption I: Human attention is restr icted to a specific entity. 157 (28) Assumption II: The nominative NP is the center of attention in the preconscious levels. Assumption I is derived from cognit ive propert ies of "select ive attention" in Zubin. Assumption II is Zubin's 'focus of the interest'. A hypothesis can be drawn from these assumptions, which is: (29) The subject NP receives the "same" reading by its inherent nature. The "sameness" is used in the sense of the one spec i f ic entry in opposition to diverse entries. The following examples wi th indefinite quantifiers i l lustrate this specific vs. non-specif ic 8 opposition: (30) a. Dareka ga minna o aisi teiru. someone (nom) everyone (acc) love 'Someone (specific) loves everyone.' [same] b. Minna ga dareka o aisi teiru. everyone (nom) someone (acc) love 'Everyone loves someone (unspecific).' [different] In this section, adopting Zubin's assumptions about the human attention system, an item or group of items represented as the subject is assumed to get specific attention and as a result it receives the "same" reading. In the fo l lowing section, this assumption w i l l be 158 investigated further in connection wi th cases where non-subjects get the "same" reading. §4.3 Focus of Interest, Quantifier, and Word Order In the preceding section, it is assumed that the subject gets the same reading. However, observe (31). The example sentence in (31) is ambiguous between the specif ic dareka reading and the non-specific one: (3Da. (Itumo) dareka ga (watasitati o) mihatteiru. (always) someone (nom) (we (acc)) be-watching 'Someone is always watching us.' b. Specific reading: the same 'dareka' a set of 'dareka' = {x], x = John, Bill,...Sue. c. Non-specific reading: different 'dareka' a set of 'dareka' = Cdarekai', 'dareka2 ',...'dareka n '} a set of 'darekai' = {xi), xi = John, Lucy,. . .Bill , a set of 'dareka2' = (X2J, x 2 = Mary, Jean,...Bob. a set of 'dareka n ' = (x n), x n = Lisa, Ron,...Tom 159 Why is it the case that (31a) is ambiguous in meaning between (31b) and (31c)? The key factor is a sentential adverb, itumo: without it , the preferred reading is (31b); with it, the preferred reading is (31c). Consider now Kuno's examples of the rightward movement in (8), discussed in §4.1.2, which is duplicated below as (32), and include some interesting cases of the mode of interaction between locative/temporal adverbs and subject quantifier phrases: (32)a. Mainen oozei no kankoo-kyaku ga nihon ni kimasu. every year many tourists (nom) Japan to come [different] 'Every year many tourists come to Japan.' b. Oozei no kankoo-kyaku ga mainen nihon ni kimasu. many tourists (nom) every year Japan to come [same/different] 'Many tourists, every year, come to Japan.' My interpretation of this phenomenon is as follows: every year belongs to the set of elements which cast the fol lowing part of the sentence into the frame of the l o c a t i v e / t e m p o r a l s i t u a t i o n a l se t t ing . Accordingly, every year binds the sentence in (32a) and the verb phrase in (32b), in terms of time. Thus, in (32a), every year has a wider scope than many tourists (subject); whi le in (32b), the ambiguity on the subject NP which is outside of the temporal frame surfaces because of 160 pragmat ic reasons. That is , in real i ty it is more l ike ly to have "different" many touris ts than the "same" many tourists . Note that besides the property of subject in our discussion, the quantifier reading is also affected by one's knowledge and beliefs about the world. The rule (29) is pragmatically sensitive by nature. Now, in addition to (29), cases like (32) a generalization such as (33): (33) Temporal/Locative phrases have the widest scope (and the "same" reading) because of their function as frame setters for a sentence. 9 Observe example (34): (34)a. Mainiti gonin no hito ga kuru. everyday five (pos) person (nom) come ([same]) [different] 'Everyday, five people come. b. Mainiti gonin no hito ga hatizyuttuu no everyday five (pos) person (nom) eighty (pos) [different]] ([same]) [[same]/[different] denwa o kakeru. telephone (acc) make-call Every day, five people make eighty calls. 161 In (34b), the subject has the wider scope than the non-subject items within the domain of the temporal frame of every day, which in turn has the widest scope. As a result, five people is ambiguous between the "same" and "different" reading in (34b), whi le i t tends to get the "different" reading in (34a); (33) overrides (34). Let us now consider word order as it is re la ted to the interpretation of quantifiers. (35) is a case of the reversed word order of (13), which is derived by Scrambling: 1 0 (35) Syoosuu no onna o tasuu no otoko ga homeru. a few (pos) woman (acc) many (pos) man (nom) praise 'A few women, many men praise.' The reading obtained is: the same 'a few' vs. the same 'many', not the same 'a few' vs. different 'many' as in (14). This is predicted by Kuno's Rule 2. The reading in (35) has two significant implications. Firs t , the subject NP receives the "same" reading not because it is the leftmost element (i.e., in sentence in i t i a l position) but because it is subject. Hence, the inherent a t tent ion-cal l ing nature of subjects is supported here. Second, the sentence- in i t ia l preposed item (hereafter, SIP) obtains the "same" reading. This can be explained if the val idi ty of the surface linear order principle is assumed. Thus, based on the above observation, one may advance the following (36): 162 (36) The SIP gains the "same" reading because of i ts position, (to be revised) However, this does not explain why the sentence in i t i a l position gets the "same" reading. Masunaga's (1982) notion of "bridge" seems to provide an explanation for that. According to her, the SIP constitutes a bridge between the previous discourse and the rest of the sentence; that is , the SIP has an anaphoric function and i ts antecedent is given in a l inguis t ic or non-l inguist ic form. I assume that it is because of this anaphoric nature that the SIP has to be specific. Thus, (36) w i l l be revised to (37): (37) The SIP gains the "same" reading because of i ts discourse anaphoric function. The "same" reading, originating in the anaphoric function as wi th the SIP, seems to hold the key to the "same" reading on topic NP's, too. Consider (38). (38a) is an instance of nominative topical izat ion; (38b) is one of accusative topicalization: (38) a. Sannin no gakusei wa hutatu no three (pos) student (top/nom) two (pos) purozyekuto o hikiuketa. project (acc) took-charge-of Three students took charge of two projects.' [the same 'three'] vs. [different/same(?) 'two'] 163 b. Hutatu no purozyekuto wa sannin no gakusei two (pos) project ( top/acc) three (pos) student ga hikiuketa. (nom) took-charge-of '(As for) two projects, three students took charge of them.' [the same 'two'] vs. [the same 'three'] N P - w a is ei ther contrast ive or thematic. But, all the top ica l i zed quant i f ier NPs are considered to be instances of contrast ive t o p i c s . n (For the detai ls of this argument see Chapter 5) Topic , in general, is anaphoric by nature, although the contrastive topic has its antecedent in the previous discourse, while this is not necessar i ly the case for the t h e m a t i c top ic (i.e., the antecedent could be in the un iversa l d i s c o u r s e ) . 1 2 Thus, the contrast ive topic is anaphoric in an even narrower sense than the thematic topic. If it is anaphoric, the referent should be determined uniquely; thus, the "dif ferent" re ferents are impossible in this case. Note, however, that the contrast iveness which is often regarded in the same light as exhaustiveness, does not af fect the reading. In the fo l lowing example, contrast ive reading is indicated by s t ress , which is often the c a s e : 1 3 164 (39) a. Sannin no onna no ko ga gonin no otoko no ko ni kisusita. three (pos) girl (nom) five (pos) boy to kissed 'Three gir ls kissed five boys.' b. Hutari no otoko ga vattu no pai o tairageta. two (pos) man (nom) eight (pos) pie (acc) ate-up 'Two men ate up eight pies.' [the same 'two'] vs. [different 'eight'] Here, in spite of the contrast, the "different" reading is s t i l l obtained for the second quantifier; that is, the stress gives the emphasis on the amount, but does not change the quantifier reading itself . Hence, the emphatic tone associated wi th the contrastive topic is not l ikely to be an influential factor in the quantifier reading in question, while the anaphoric nature is. Thus, we arrive at (40): (40) Topic gains the "same" reading because of i ts anaphoric function. The fol lowing (41) sums up the dominant readings and the generali-zations which account for them: (41 )a. Qi (subject) - Q 2 [same] [different] (by (29)) b. Qi (non-subject) - Q 2 (subject) [same] [same] (by (37) and (29)) 165 c. Qi (Top-subject) - Q2 [same] [different] (by (40) (and (29))) c' Q (Top) - Q 2 (subject) [same] [same] (by (40) and (29)) d. Qi (Loc/Temp) - Q2 (subject) [same] [same/different] (by (33) and (29)?) The surface linear order principle is descriptively correct, but it does not provide any explanation for the phenomenon. The set of hypotheses (29), (33), (37), and (40) discussed in this section are my attempt to resolve this problem. If (29) holds, the prominence of the s u b j e c t 1 4 in a sentential expression is supported from an ext ra-syntactic point of view. Zubin's assumption that something prominent to us, i.e. the center of attention, cannot be a diverse enti ty is empir ical ly supported in our perception of the outside world. As for (33), (37), and (40), they are related to the discourse function of those sentence i n i t i a l elements. Locat ive / tempora l adverbs provide a situational setting for the sentence; that is, they are used to cast the sentential discourse into a locat ive/ temporal frame. In this sense, topic also has this "framing" function. Utterances always occur within the framework of the topic. I assume that SIP in Scrambling and the contrastive topic both have vir tual ly the same function in terms of their anaphoricity. The difference between the two is: the lat ter has the particle wa, and the former does not. As is discussed in Chapter 5, the anaphoric aspect of NP-wa is rather secondary compared to i ts semantic 166 function, i.e., that of the categorical judgement marker. As the Grosz-type 'focus' c la ims, anaphoric use is closely related to our conscious attention system, as opposed to the unconscious attention system (i. e., Zubin's 'focus'), which is, in turn, t ied to subjecthood. Unlike the syntac t ic rules, the proposed general izat ions for ex t ra - syn tac t ic phenomena are sensitive to certain pragmatic factors related to our real world knowledge. §4.4 Conclusion This chapter has given an explanation of the disambiguation mechanism of quantifier reading in Japanese. The scope interpretation which is assumed to have LF representation in syntax provides a wide /nar row reading for the Qi - Q 2 sequence. However, actual readings obtained for such sentences are more or less influenced by some ext ra-syntac t ic factors. Contreras' "group" vs. "distr ibutive" reading is a relevant case for such influential factors. Kuno's "same" vs. "different," which offers a type of reading supported by our intuition, is also considered as another case of such meanings which are not purely syntactic, having a certain relationship to the wide/ narrow reading. "Same/different" overlaps wide/narrow for a sentence wi th the basic word order. On the other hand, in cases of scrambling, although syntax a l lows an ambiguous scope interpretat ion here, what is actual ly obtained in the discourse environment is more l ikely to be the "same" -167 "same" reading, which involves functions of the subject item and preposed non-subject item in discourse. §4.5 Addendum §4.5.1 Grosz's'Focus': Defined solely on a semant ic 1 5 basis in a processing model Grosz (1977, 1978, and elsewhere) considers focus to be an active process of attention-calling (she often uses the terms "focusing" or "highlighting"), and takes it to be formal ly representable on partitioned semantic networks. Her concept of 'focus' is formed on the basis of a task-oriented d ia logue , 1 6 to be incorporated into a computer processing model. The part icular nature of the task-oriented dialogue is that the discourse formation process is parallel to the actual task. Hence, this type of discourse provides an opportunity to check the descript ion-forming process wi th the actual chronology of the task without worrying about the degree of idiosyncrasy in the discourse style. Grosz's model (1977) of the focusing mechanism in task-oriented dialogues is as follows. A task as represented in discourse consists of hierarchially nested subtasks. When it is completed, the description of each subtask fades from focus; however, the parent task remains in focus. Therefore, when the descr ip t ion of a s ib l ing subtask is completed, discourse anaphoric expressions are related to the concepts 168 in the parent task in focus, not to the already completed sibling task(s). Thus, this is a two-fold focus system (Grosz, 1978); that is , the main task is "expl ic i t ly" focused while subtasks or objects involved are " i m p l i c i t l y " focused. The fo l lowing examples are given in Grosz (1977) ' 7 : (42) Discourse I (= Figure 11-7, p. 23) (E = expert; A = apprentice; [italics] = an object in focus, or a focused task) [M-Task] E: Good morning. I would like for you to reassemble the compressors. [S-Task i l E: [S-Task 2] E: A: I suggest you begin by attaching the pump to the platform ...(other Sub-Taski - i~ n ) Good. A l l that remains then is to attach the belt housing cover to the belt housing frame. A l l right. I assume the hole in the housing cover opens to the pump pulley rather than to the motor pulley. A: A l l right. The belt housing cover is on and tightened down. (60 utterances after the beginning.) [M-Task] E: Fine. Now, let's see if /'tj works. 169 (43) Discourse II (= Figure 11-14, p. 37) [S-Task,] A: How do I remove the flywheel? E: A: [S-Task 2] E: A: E: First loosen the two alien head setscrewsk holding it to the shaft, then pull it off. The two screws] are loose but I'm having trouble getting the wheel off. Use the wheel puller. Do you know how to use i t? No. Loosen the screwm in the center and place the jaw around the hub of the wheel, then tighten the screw^. Discourse I in (42) shows the nested hierarchical structure of tasks, i.e., (M-T (S-Ti ( S - T i ~ i , . . . 5 - T i ~ n ) , S - T 2 ) ) , while Discourse II in (43) consists of two sibling tasks, S-Ti and S -T2. The use of the discourse anaphoric expressions, including repetitions, indicates the task description units since they are l ike ly to have their antecedents, which are in focus, wi th in the same task descriptions: j = i in M-Task in (42); 1 = k in S-Taski and n = m in S - T a s k 2 in (43). Bul lwink le (1977) suggests using Grosz's type of 'focus' for anaphora disambiguation: 170 (44) a. Ira wants to have a party,- at his housej. It is going to be at 8 o'clock p.m. Invite {everyone at the lab; Mitch, Dave, Candy, Bruce, and Beth) to attend, (it = i) b. Ira wants to have a party,- at his housej. It is at 25 Wildwood Street. Invite (everyone at the lab; Mitch, Dave, Candy, Bruce, and Beth] to attend, (it = i) It refers to a party, not his house, unless his house rather than a party is in focus in this conversation. The same thing happens in a Japanese zero anaphora case: (45) Tanaka ga zibun no iei< de paateei o (nom) self's house at party (acc) yaru-rasi i yo. seems to throw, you know (mod) 0 Mitaka datte-iu keredo, iku kai. (0=1) (it is) I've heard although go (Q) 'I heard that (Mr.) Tanaka is going to throw a party at his own house. It's in Mitaka, I've heard. Are you going?' It seems to be natural to interpret that the locational information, Mitaka, is given for a party, not for his own house, in (45). Now, in much li terature, the term 'focus' is often used as just another name for what is traditionally called 'topic'. Grosz's 'focus' can be counted as one such example: 'topic' in her sense is the same entity which is referred to as 'global topic' in Chapter 2 (cf. also 'discourse 171 topic' in Chapter 5). As observed in Chapter 2, this type of topic frequently coincides wi th sentential topic in actual discourse. On the other hand, the topic-focus ar t icula t ion descended from the Prague School - which has had, and continues to have, a great influence on this area of study - presupposes the complementary distribution of topic and focus. Also "new information" in Praguian FSP (functional sentence perspective), "comment" in topic-comment (Hockectt, 1958), and "Rheme" in theme-rheme ( H a l 1 i d a y , 1967) have often been regarded as "focused" items, depending on the definition accorded the term 'focus.' In the fo l lowing two sections, I w i l l introduce Kuno's 'focus' and Rochemont's 'focus', which are controlled by principles largely contrary to that of Grosz's. §4.5.2 Kuno's'Focus': Information value represented in word order A series of important works by Susumu Kuno can be considered North American descendants of Praguian FSP. Kuno (1978) defines focus as new information, wi th word order arranged (in principle) in the sequence of old information to new information before a verb. That is, w i t h the exception of emphatical ly s t ressed cases, the posi t ion immediately preceding the verb is the most l ikely place for focus (cf. Kuno § 1 , especially §1.5. (46) = Kuno's (8) and ( lO) ) . ^ 172 (46) A: Ziroo wa Hanako to Bosuton ni i t ta? (top) wi th to went Did Ziroo go to Boston with Hanako? Ba: Un, Bosuton ni i t ta yo. yes to went (emphatic modal) 'Yes, they went to Boston.' Bb: *Un, Hanako to i t ta yo. 'Yes, he went to (Boston.) wi th Hanako.' (47) A: Ziroo wa Bosuton ni Hanako to i t ta? Ba: *Un, Bosuton ni i t ta yo. Bb: Un, Hanako to i t ta yo. Kuno regards new information as having higher informational value than iess new information. The test applied wi th the above examples is based on the non-deletabi l i ty of new information; the information immediately preceding the verb, Bosuton ni in (46) and Hanako to in (47), is new and cannot be deleted. However, later he changes the essential notion of newness to one of importance because of the phenomenon observed in the fol lowing model discourse given by Kuno (1982): 173 (48) A i You must have spent a fortune during your trip on hotels. Couldn't you stay with your friends or your friends' friends? Bi In some ci t ies , I did, but I had to stay in hotels in many cit ies. A 2 You f i rs t went to Paris, right? Did you stay in a hotel there? B 2 No, I didn't stay in a hotel 0 - I stayed wi th an old friend of mine who is studying music there. A 3 Did you stay in a hotel in London? I hear hotels are getting terribly expensive there. B 3 a. Yes, I stayed in a hotel 0 because I didn't have any friends there. b. Yes, unfortunately, in a hotel. c. *Yes, in London. In the above, 0 in B2 (= in Paris ) is older and less important; however, in B 3 - a , 0 (= in London ) is newer but less important, while in a hotel is older but more important. Thus, Kuno's deletion rule is revised from "deleting less new information rather than newer" to "deleting less important information rather than more important." Kuno's version of 'focus' predicts that items such as wh-words, which one would in tu i t ive ly consider to be "highlighted," w i l l be focused. The wh-question construction is the very thing which is 174 u t i l i zed to gain the information that the speaker desires; in other words, the wh-word is the most important piece of unfilled information. As a consequence, the piece of information in the answer which f i l l s the gap created by the wh-word is focused. However, Kuno's version does not predict cases involving the discourse anaphoric use of demonstratives. These deict ic signal words are also another device for "highlighting." If one accepts this sort of highl ight ing as a type of 'focus', then f i r s t of a l l , Kuno's new information strategy certainly does not work in these cases, since demonstratives have antecedents and are therefore not new in any sense. Second, even if one employs the notion of 'importance', the antecedent is not the important item in discourse in the same sense as the informational gap-f i l l ing word in an answer is. Therefore, the function of discourse anaphoric expressions cannot be accommodated in Kuno's focus system. At f i r s t glance, it might appear that Kuno's notion of 'focus' and i ts predictions of focused items are opposed to Grosz's. According to Kuno's informational value based 'focus', a topic or thematic item is less l ikely than other items to be focused. However, this is not always the case (cf. Kuno 1972, 1975): (49)a. Tikyuu wa marui. earth (top) round 'The earth is round. 175 b. Kinoo, Ikeda-san ga tazunete-kimasita. yesterday Mr. Ikeda (nom) come to v i s i t Ikeda-san wa mamonaku Amerika e iku to i t te-imashita. (top) soon America to go said that 'Yesterday, Mr. Ikeda came to v i s i t me. He said that he would go to America soon.' (50)a. Dare ga kore o kowasita no. who (nom) this (acc) broke (Q) 'Who broke this? ' b. Boku wa/*ga kowasite-nai yo. I (top/nom) s have not broken I'll te l l you (mod) 'I didn't break it.' ("I" contrasts wi th "someone" who did it.) c. (= Kuno's (2-2) in 1972) John to Mary to B i l l ga tazunete kimasita. and and (nom) vis i t ing come John wa/ga kudamono o kuremasita. (top/nom) fruit (acc) gave-me Mary wa/ga okasi o kuremasita. (top/nom) cookies (acc) gave-me 176 B i l l wa/ga hon o motte kite kuremasita. (top/nom) book (acc) bringing coming gave-me In contrast wi th the topics in (49), the topics in (50c), John and Nary (but not Bill ) a l l const i tute new information, since they are not predictable in the context (cf. Kuno 1972: 276). One might say that these topics (and also 6/7/ ) are a l l old information, since they are already mentioned in the previous discourse. However, it is misleading to assign an old-new distinction to one of noun quality (as in the above); the "old-new dist inct ion" is the dis t inct ion made on the information structure of discourse; thus, a sentence element is identified as new or old only in terms of how it is related to i ts embedded context . 1 9 On the other hand, according to the definition of importance, these topics are focused since they are non-deletable. In (50b), there is not much dispute over identifying boku (-wa) as new information and boku is also non-deletable, important information; thus the topic boku becomes focused. According to Halliday (1967, 1968, and elsewhere), his theme-rheme structure is different from old-new information, although the two can be overlapped: the former is the structure of a sentence (i.e., theme is the sentence in i t ia l element and rheme is the rest); whereas the latter is the information structure for discourse. According to the former view, the topic in topic-comment art iculat ions is more like a combination of the above two classes old and theme. 177 While Grosz's type of "focus" works to pivot the expanding discourse, Kuno's type of "focus" is relevant to the expansion i tself . Here there seems to exist two opposed dynamisms in discourse: one works to bring about convergence wi th the topic ( = Grosz's focus), the other works towards the expansion of discourse or the addition of new information to that already given. §4.5.3 Rochemont's 'Focus': a syntactically represented and contextually construable entity There are studies of 'focus' in generative grammar (Chomsky (1970), Jackendoff (1972) among others). A more recent work, Rochemont (1986) is an extensive study of 'focus' in English, which deals wi th phonological, syntactic, and discourse aspects of 'focus' from the viewpoint of generative grammar. Rochemont's 'focus', which is diagnosed by the well-formedness of question/answer pairs l ike (51) below, is treated as a syntactic notion, systematically associated wi th prominent stress and certain pragmatic environments. (51) a. Who did Tom cal l? b. Tom called Mary. focus It is not the purpose of this chapter to give an overview of Rochemont's complex theory of focus. What is relevant here is his notion c-construable in (52), using the pragmatic notion "under discussion," 178 which in turn provides a possible alternative to rather loosely defined concepts like old information. This c-construable is used to identify Rochemont's focus in discourse context. (52) a. If a is not c-construable, then a is focused, [p. 172] b. a is c-construable if (i) a is under discussion, or (ii) a is an indexical expression "Under discussion" means either that it is specif ical ly mentioned in the previous discourse, or mentioned using the expression wi th the same sense or the same reference. "Indexical expressions" are the personal pronouns /, you, we; locative and temporal adverbs such as here, there, today, last time; and verbs of appearance such as arrive. C-construable is further divided into two categories: "conventional" and "conver-sational," in the sense of Grice (1975). The rule for Contrastive focus in (53) makes an interesting claim: (53) If a /S is directly c-construable [ = the case of (i) in (52b), M.O.], where a /S is the result of extracting a from 5, and S is not c-construable, then a is a focus, [p. 175] This is to say that old information can be focused. At least, it is safe to say that a topic can be focused, considering the varied definitions of "old." Rochemont's rule also predicts that in Kuno's (50c), John, Mary, and 6/7/ are cont ras t ive ly focused (a /S = conversa t ional ly c -construable). Considering the fact that Kuno (1973) c lass i f ied topics into two categories, thematic topic and contrastive topic, Kuno's theory 179 also indicates s imi lar consequences, which naturally follow from seeing topic (Grosz's focus or Japanese wa- marked NP) and focus as two independent no t i ons , 2 0 neither two ends of a single spectrum nor in complementary distribution. They have two different functions, which I assume belong to different dynamisms of discourse formation. 180 Notes (Chapter 4) 1 There are some disagreements over the readings in (2a); however, even for those who disagree wi th the unambiguous reading, the preferred reading is s t i l l ( lb) , and (1c) is marginal. On the other hand, for those who disagree wi th the ambiguity treatment in ( la) , the unambiguous reading is also (1b). Whichever is the case (unambiguously obtained reading or preferred one), ( lb) has the prominence in reading. 2 There is no difference between the same 'every' and the different 'every.' 3 The entries of a few women can overlap. That is, in actuality, one may for example obtain the fol lowing cases, which are consistent wi th the assertion: (i) 'a few' : 'a few,' * 'a few 2 ' ; 'a few,' = 'a few2"; 'a few,' <t 'a f e w 2 ' ; 'a few," £ 'a few 2 ' ; 'a few,' ™ 'a f e w 2 ' . Thus, the "different" can include a case like 'a few, ' = 'a few 2" in (i) which is, in effect, equivalent to the "same" reading. However, we find these special cases only after checking into the situation in the real world. This is a consequence of the potential discrepancy between what is asserted in language and what actually takes place in reality. 4 Kuno (1971) c la ims that the subject-object scrambling case is ambiguous in terms of wide/narrow scope interpretation. However Kuno (1973 a & b) claims the unambiguous "same" - "same" reading for it. 5 Contreras specifies certain conditions for "distributive" reading. On the other hand, according to Wi l l i am (1986), any NP which has a "group" reading potentially has a "distributive" reading when it is in subject position. 6 It is wel l -known that Japanese, as a member of the so-cal led topic prominent languages, has the topic marker wa. An English sentence l ike (i) can be translated into Japanese either as ( i i ) or ( i i i ) , depending on the context: 181 (i) B i l l saw Mary at the park. (ii) Biru wa Merii o kooen de mita. (top) (acc) park at saw ( i i i ) Biru ga Merii o kooen de mita. (nom) On the other hand, English has topicalizat ion as in (iv); however, the s t ructural ly-marked topic sentence which is approximate to the Japanese wa sentence in discourse function is a lef t -d is located sentence like (v) (cf. Reinhart, 1982): (iv) Mary, B i l l likes. (Topicalization) (v) Mary, B i l l l ikes her. (Left Dislocation) It is not clear, however, whether the German topic sentence, for example (vi) given by Zubin, is functionally s imi la r to English (iv), (v), or neither of them: (vi) Den Film habe ich schon dreimal the movie (top/acc) have I (nom) already three times gesehen. seen. Further invest igat ion of the discourse function of this type of sentence is required. 7 Zubin uses nominat ive and (g rammat ica l ) subjec t in an interchangeable sense. Shibatani (1978) argues that in the m u l t i -nominative construction, which is one of the charac ter i s t ics of Japanese, only one of the several nominatives in a sentence is the subject. Problems to be examined here involve subjecthood, the status of the subject in syntactic theories, and the analysis of the mul t i -nominat ive construction. However, these w i l l be left for future research, as they do not affect the main lines of the argument. 8 Although the 'same' is almost synonymous wi th 'specific, ' the scope relation and specif ici ty are closely related but not exactly the same; 182 if the speaker has a specific person in mind it takes wide scope but not vice versa (cf. Maclaran, 1982). Within Temporal/Locative, T seems wider than L: (i) Mainiti dokoka de nanninka ga umarete-iru. every day somewhere some numbers (of people) is-born ([same]) [different] 'Every day, somewhere, some number of babies are born.' (ii) Dokoka de mainiti nanninka ga umarete-iru. [same] ([different]) A movement analysis is assumed for scrambling (cf. Saito, 1985). However, whether the structure is derived by movement or not is not an essential problem in this analysis. Even if it is not derived by movement, the function of the SIP is intact. This c l a s s i f i ca t ion of topics is or iginal ly given by Kuno (cf. for example, 1973a). The following i l lustrates that a quantifier NP can occur in NP-wa position only when it has the contrastive reading: (i) Oozei no hito wa kita. many (pos) people (c-top/*t-top) came. Thematic topic: Kuzira wa honyurui da. whale (top) mammal (cop) 'A whale is a mammal.' Contrastive topic: Kaze wa huite-iru. wind is-blowing 'The wind is blowing (but it is not cold).' Other means to indicate contrastiveness are: the contrastive topic in question and the exhaustive-list ing ga (cf. Kuno 1973a), which is untestable because it is the nominative marker. It is not surprising then that the subject is responsive to zibun, the egocentric pronoun in Japanese, as the antecedent. 183 1 5 "Semantic" could be replaced here wi th "pragmatic"; i.e., it is used here in a wide, or non-truth conditional, sense. 1 6 Linde (1979) takes a s imi lar approach. 1 7 Supplementary explanations and c lar i f ica t ions have been added to these examples. 1 8 Kuno uses * to indicate sentences which are infel ic i tous in their contexts. Throughout th is thesis , * is used to indicate un-grammatical sentences as well as infelicitous sentences, in contexts where this does not cause any confusion. 1 9 Also cf. the argument of Reinhart (1985), which is briefly introduced in Chapter 5. 2 0 "Topic" has certainly created another "thicket" of terminology, even less transparent than the one associated wi th "focus." However, I assume Japanese NP-wa is a topic which happens to occur wi th a morphological marker. 1 84 Chapter 5 The Function of the Particles Wa and Ga §5.0 Introduction Many analyses and observations have been made concerning the part icle wa in terms of sentence structure, semantics, and discourse function, the last being the most popular topic. The main concern in this chapter is not to challenge a part icular analysis or observation, but rather to provide an integrated explanation of the nature of wa (and ga in that connection) which underlies the recognized character is t ics of the use of wa. In this way, some disagreements between the previous analyses may be resolved into mere te rminolog ica l clashes or d i f ferences in points of v iew. Af t e r th is is recognized, the reinterpretation of such disagreements w i l l shed new light on already existing analyses. §5.1 The Part icle wa The Japanese part icle wa is most commonly considered to be a topic marker or an old information marker. While these are intuit ively 185 appealing explanations, the vagueness of the concepts 'topic' and 'old information' causes diff iculty in defining the terms. In fact, it is quite possible to assume that the use of wa is not pr imari ly controlled by discourse factors such as these. §5.1.1 A Characteristic of the Use of Wa "I am a cat" is the famous opening sentence of Soseki Natsume's (1867-1916) 1905 novel Wagahai wa Neko dearu (/ am a cat ), which looks at human l i fe from the viewpoint of a house cat. Outside of f ict ion, though, this type of utterance is totally absurd. Japanese has a peculiar sentence type which is often called the "eel-sentence," named after a popular example given in (1): (1) Boku wa [unagi da], [ U N P Particle [eel ? ] V P A possible context for (1) is at a restaurant. Suppose a waiter is taking orders. Each customer says something l ike, "Boku wa sushi da (I am going to have sushi)"; "Boku wa soba da (I am going to have noodles)"; or "Boku wa unagi da (=( 1))." Or when fishing, one might say (1) in answer to a question such as, "What did you catch?" Last but not least, (1) can l i t e ra l ly mean "I am an eel" in a fictional context. Additional examples of this type are given in (2) below: 186 (2) a. T i t i wa kuruma da. my father car 'My father is a car/My father goes to work by car/My father's favorite pastime is repairing a ca r /e tc ' b. Otooto wa kyooryuu da. my brother dinosaur 'My brother is a dinosaur./My brother loves dinosaurs./My brother got a toy dinosaur./etc.' Suppose that someone asked you how your father goes to work. You could answer wi th (3a) or (3b), but a nonredundant answer like (2a) is not merely acceptable but somewhat preferred: (3) a. T i t i wa kuruma de sigoto ni ikimasu. my father car by work to go 'My father goes to work by car.' b. T i t i wa kuruma de ikimasu. 'My father goes by car.' In turn, (3b) is a well-formed answer to a question like, "What kind of stuffed animal did your brother get for Christmas?" This type of sentence has the form of [X wa Y da ], and is equated wi th the English copula sentence [X is Y], While English copula sentences 187 are translated into either [X wa Y da ] or [X ga Y da ], i t is often the case that "NP wa " is translated into English using "as for NP" as an approximation. (4) Translation equivalent: Japanese English X ga Y da • X is Y X wa Y da * "~~ " ^ As for X, it is Y The following translations ( D and (2') may describe the nature of such wa sentences: (1') Boku wa unagi da. 'As for me, I am an ee l / i t is (an) eel.' (2')a. T i t i wa kuruma da. 'As for my father, he is a car / i t is a car.' b. Otooto wa kyooryuu da. 'As for my brother, he is a dinosaur/it is a dinosaur.' The ambiguity between the above two readings has yielded a number of studies of these so-called "eel sentences." Now, the questions that should be asked here are: i) whether or not "da " in an "eel-sentence" is the Japanese equivalent of the English verb "to be" or the copula; and if so, i i ) whether or not the Japanese-188 spec i f i c par t i c le wa embodies the special function in the "eel-sentence" phenomenon (English copula sentences are used rarely, if at a l l , in the same way as (1) and (2) in Japanese). In what fol lows, I w i l l argue for the copula hypothesis for da, rather than the others, in investigating the function of wa. Let us f i r s t look at Okutsu (1978), which challenges the copula hypothesis. He proposes instead the propredication hypothesis, which maintains that da is the proform of (a part of) the predicate, such as -tabetai (to want to eat), -kudasai (please give), -tyumonsuru (to order), etc., as shown in (5): (5) Propredication Hypothesis: Okutsu (1978) Boku wa unagi ga tabetai 'I want to eat eel.' I eel Nom (obj) want to eat o tyuumonsuru 'I order an eel.' Acc. o Acc. 0 kure give (Imp) da 'Give me eel. Okutsu proposes the propredicat ion t ransformat ion 1 fo l lowed by (optional) par t ic le deletion. Okutsu's propredication hypothesis, however, is not we l l res t r ic ted, which is a crucial defect in theory construction. According to Okutsu, "da " can replace almost anything -1 89 verbs, adjectives, and nominal adjectives - in i ts derived forms wi th various suffixes, tenses, negative morphemes, etc., associating wi th various grammatical cases (!). This can be seen in (6): (6) a. Q: Kinoo kimitat i doko ni ita no. yesterday you (two) where LOC be-Past Q 'Where were you yesterday?' A , : Otooto wa gakkoo ni ITE brother school LOC be-(gerund) Boku wa uti ni ITA I home LOC be-Past A 2 : Otooto wa gakkoo DE, boku wa uti DA. ("da" in gerund) 'My brother was at school and I was at home.' b. Q: Kimi wa itu ko-rare-nai no. you when come-can-Neg Q 'When cannot you come?' A , : (Boku wa) suiyoobi ni K O - R A R E - N A I on come-can-Neg A 2 : (Boku wa) suiyoobi DA (ne). you know 190 c. Q: kimi wa nani o hanas-ase-rare-ta no. you what Acc speak-Causative-Passive-Past Q A | : boku wa zibun no kazoku ni tuite no hanasi o self's family about 's story K A K - A S E - R A R E - T A . A 2 : Boku wa zibun no kazoku ni tuite no hanasi DA (yo). Emp. In (6), "da " replaces the past form of the verb iru in (a), a negative form of the verb kuru in (b), and a causative-passive past form of the verb hanasu in (c). Furthermore, in Okutsu's analysis, wi th the exception of the equational case of (7c) which he admits is the copula case, even (7a) and (7b), which are usually not considered as "eel-sentences," 2 are also derivable from (8a) and (8b) respectively: (7) a. Boku wa Yamada da. I (name) 'As for me, I am Yamada.' b. Boku wa gakusei da. I student 'As for me, I am a student.' 191 c. Tokyo wa Nihon no syuto da. Japan 's capital city 'As for Tokyo, it is the capital city of Japan.' (8) a. Boku wa Yamada [to iu], as call (oneself) b. Boku wa gakusei [ni zokusuru], to belong 'I belong to the students.' To treat (7c) as an exception is poorly motivated. I cannot see why (8') below is not acceptable for (7c) as an underlying form, if (8a) and (8b) are acceptable for (7a) and (7b) respectively: (8') Tokyo wa Nihon no syuto to onazi-da. as same 'As for Tokyo, it is the same as the capital city of Japan.' It is true that Okutsu's main idea is not undermined by excluding (7c) from his proform analysis. But the point is that there is nothing to prevent including (7c) in "eel sentences" since da can replace anything in his theory. The revised vers ion of Okutsu's (1981) theory s tates the condition for the appl icabi l i ty of transformation: target elements should be "presupposed." What this "presupposition" means is not clearly 192 stated by Okutsu, but it seems to be s imi l a r to so-ca l led "old" or "known" information. (This s t i l l does not make the status of (7) any clearer, because of the nature of these widely used but loosely defined notions.) However, I must say that this revis ion does not put any essent ia l r e s t r i c t i on on his rule sets, but rather creates a less r e s t r i c t ed theory of grammar: sentence grammar has to u t i l i z e contextual information to make i t se l f fully functional. To progress toward the aim of a we l l - r e s t r i c t ed theory, the idea that "da " is the proform of various pre-dicative expressions w i l l have to be abandoned. For the purpose of deriving the "eel sentences," Kitahara 's pseudo-cleft hypothesis (1981) is a potential alternative to that of Okutsu. The "eel-sentence" is derived through i t s pseudo-cleft counterpart as shown in (9): (9) Pseudo-cleft Hypothesis: Kitahara (1 981) a. [Boku ga tabe-tai no] N P wa unagi da. I Nom want to eat eel 'What I want to eat is eel.' b. [Boku no no]Np wa unagi da. c. [Boku no]|s|p wa unagi da. d. Boku wa unagi da. 'As for me, it is eel.' 193 Kitahara cal ls the f i rs t "no" in (9b) the prenominal particle, like "no" in (1 O); and the second "no" the nominalizer particle as in (1 1). (10) Taroo no hon book 'Taroo's book.' (11) [Tenisu o suru nolNp wa tanosii tennis Acc play fun 'Playing tennis is fun.' According to Kitahara, "da " is a part of the pseudo-cleft construction, but different from the copulative "da " in (7). Kitahara's analysis does not require contextual information be taken into consideration; however, it shares the same fundamental basis; i.e., a l l information which is necessary for interpretation is already provided in the sentence as the underlying form. Thus Kitahara's analysis also places a large burden on sentence grammar, just as Okutsu's does. In conclusion, there does not seem to be any reason to assume two kinds of "da," and to treat the "eel-sentence" as a syntact ical ly special case. What is apparent here is that as far as syntax is concerned, the eel sentence is not different from those given in (7), and a l l of these [ X wa Y da ] sentences are c l a s s i f i e d as copulative 194 sentences. 3 However, the Japanese copula sentence tends to rely on context for interpretation more than the English one does. The question then w i l l be why Japanese copula sentences allow such apparently free combinations of X and Y. It seems that the particle wa has something to do wi th this phenomenon. To pursue this point further, let us move to another set of data which is not of the [X wa Y da ] type. First , let us consider example (12): (12) a. Taroo wa Tokyo e itta. to went 'As for Taroo, he went to Tokyo.' b. Taroo wa Hanako ga shookaisita. Nom introduced 'As for Taroo, Hanako introduced (him).' Both movement and non-movement analyses have been proposed for the structures of (12), which are represented in (13): 195 (13) a. 5/? NP Tarc-OjWa NP e i b. S/? NP Taroo, wa Hanako ga NP V l I e, syookaishita The wa-attached N P s , Taroo in (13a) and Taroo in (13b), are bound to the nominative in the former and the accusative in the latter. However, i t ta 196 there are also cases where a wa-attached NP is not bound to any arguments within S, as in (14): (14)a. Sakana wa [ s tai ga umai]. fish red snapper Nom good 'As for fish, red snapper is tasty.' b. Horidei wa [ shawai ga ii]. holiday Hawaii good 'As for holidays, Hawaii is good.' c. Taroo wa [sHanako ga sono kai ni deta]. Nom. the meeting to attended 'As for Taroo, Hanako attended the meeting.' In (14c), Taroo has something to do with Hanako's going to the meeting. For example, Hanako might be Taroo's assistant, and she went to a business-related meeting on his behalf. The examples in (14) remind us of the phenomenon seen in the copula case. Without context or real world knowledge, the relationship between the NP and the rest of the sentence looks rather arbitrary. Just as it does in some copulative sentences, the co-occurence of two nouns apparently creates a nonsensical sentence. What then is the role of wa in these sentences? It can be seen that wa marks the major break in a sentence. Looking at the matter 197 from another standpoint, wa apparently functions to join two otherwise unrelated elements into one sentence in (14). That raises a fundamental question: what makes a sentence a sentence? In the fol lowing, I w i l l look at Kuroda's (1965 & 1972) analyses of the sentence wi th wa and the sentence without wa, which holds the key to this question. §5.1.2 Wa and Human Judgment Adapting the framework of Brentano-Marty's theory of judgment for Japanese, Kuroda (1965) claims that the dist inction between the two types of human judgment explained below, thetic in (15) and categorical in (16), gains substantial support from Japanese, because of the particle wa: (15) Thetic (single) judgment: represents simply the recognition or rejection of material of judgment. [p. 154, Kuroda, 1972] (16) Categorical (double) judgment: consists of two separate acts, one, the act of recognition of that which is to be made the subject, and the other, the act of affirming or denying what is expected by the predicate about the subject. [p. 1 54, Kuroda, 1 972] In the categorical judgement (16), the particle wa marks the "subject" in Japanese. Note that Kuroda uses the terms 'subject' and 'predicate' not in the syntactic sense, but in the sense of tradit ional grammar or logic linked to Port-Royal logic . 4 Throughout this chapter, the term 'subject' 198 w i l l be used in Kuroda's sense. Although in tradit ional grammar and logic, the subject-predicate structure of a sentence corresponds to the s imi la r structure of judgment, in Brentano-Marty's theory, a sentence has the subject-predicate structure, but judgment may or may not have it. Kuroda argues that in Japanese, however, the subject-predicate structure in categorical judgment happens to be marked by the particle wa' as in (17). A sentence without the wa-marked phrase is 'subject-less' in judgment, and that is thetic. A thetic sentence like (18) is simply used to recognize an event. (17) a. Inu wa hasitte-iru. (= Kuroda's 7.2) dog (Nom) is-chasing 'The dog is running.' b. Inu wa neko o oikakete-iru. (= Kuroda's 8.2) dog (Nom) cat Acc is chasing 'The dog is chasing the cat.' (18) a. Inu ga hasitte-iru. (= Kuroda's 7.1) 'The/A dog is running.' b. Inu ga neko o oikakete-iru. (= Kuroda's 7.2) 'The/A dog is chasing a cat.' In the thetic judgment sentence in (18), inu 'dog' and neko 'cat' are simply participants in the event. On the other hand, in the categorical 199 judgement sentence (17), inu 'dog' unlike neko 'cat' is not just a participant but has a prominent status in the description of the event. The judgement underlying the example of the thetic case in (18b), for instance, is analyzed as (19) by Kuroda: (19) a. X is chasing of Y (kernel judgement) b. X is a dog (identification of participants) c. Y is a cat Categorical judgements are based on thetic judgements like (19) and, in addition, X is chosen among the participants and given prominent status, "subject," marked by wa. Such, in outline, is Kuroda's conceptual proposal on the nature of the wa- sentence. In this framework, once the event and the entity are recognized through judgment, the ref lect ion of this judgment act is real ized in sentence form as a wa- sentence. Regardless of whether an HP-wa has a direct connection to the rest of the sentence as a participant of the event described there, the wa- sentence is a syntactic real izat ion of the subject-predicate structure in judgment. Judgment underlies a well-formed sentential expression; then, syntax detects the structural well-formedness of the sentence. In other words, while syntax has to provide a cer ta in device to describe this phenomenon, the "ee l -sentences," even though their NPs lack thematic roles, are perfectly well-formed. As is indicated by wa, the sentences represent a cate-gorical judgement. 5 200 In the following sections, I w i l l investigate further consequences of assigning wa such a logico-semant ic funct ion , 6 i.e., marking a categorical judgement, in connection w i th the common functional approach to wa. §5.1.3 Judgement Marker vs. Topic Marker Wa has often been called the topic marker, but what the popular term "topic" refers to is a rather opaque concept. There have been many attempts to define "topic." While such definitions or c r i te r ia for "topic" are often vague, Japanese wa- marked NPs seem to coincide wi th the "theme" in the "theme-rheme" a r t i cu la t ion of the Prague school t rad i t ion . 7 Reinhart (1982) claims that the "theme," which is also her sentence topic, is determined relative to a discourse, in the way shown in (20) (= Reinhart (6)): (20) a. (Who did Max see yesterday?) Max saw Rosa yesterday, (topic = Max) b. (Has anybody seen Rosa yesterday?) Max saw Rosa yesterday, (topic = Rosa) Reinhart claims that topichood should be defined as "a R E L A T I O N between an argument and a proposition relat ive to a context" (p. 7; emphasis mine) instead of as a PROPERTY of the referents relative to a context, like "old information." 8 For the "relation" in question, Reinhart argues that 201 "aboutness" is the essential notion for T O P I C OF. In the semantic definitions of "about" by logicians, (21a) is simultaneously "about" the class of crows and the class of black things, and furthermore "about" the class of non-black things and the class of non-crows, since (21a) is equivalent to (21 b): (21) a. A l l crows are black. b. A l l non-black things are non-crow. What Reinhart cal ls a sentence topic is, on the other hand, ABOUT a single one of a l l the referring expressions in a given sentence; the context determines which one is the topic. Thus, she claims that "aboutness" is a pragmatic notion rather than a semantic one. Reinhart 's de f in i t ion of topichood w i t h "aboutness" as a pragmatic concept seems to be applicable to Japanese wa sentences too. The Japanese equivalents of (20) are given in (22), where it is apparent that the topics are morphologically marked by wa: (22) a. Max wa Rosa o mita. Acc saw b. Rosa wa Max ga mita. Nom saw Reinhart's PPA (Possible Pragmatic Assertion) is the formula to dist inguish a pragmatic assertion (= what a statement is about in a pragmatic sense) which is relative to a context, from all the possible 202 assertions for the proposition of a given sentence (= what a statement is about in a semantic or logician's sense). In Japanese, this distinction is not only assumed on an abstract level, but also the contextually chosen pragmatic assertion in PPA is morphologically marked by wa. The c r i t e r i a employed by Reinhart to identify the topic are Strawson's (1964) "subjecthood" c r i t e r i a . Strawson uses the term "subject" and "topic" interchangeably and what Strawson means by "subject" is the "subject" in tradit ional philosophical logic, as is the case w i th Kuroda. Now, according to Strawson's c r i t e r i a , what a statement is about (= topic) is: i) presumed knowledge, i.e., "assumed to be already in the audience's possession" (the pr inciple of the presumption of knowledge) in accordance w i t h the purpose of communication (the principle of relevance); ii) (tends to be) the center of the truth value assessment, i.e., a statement is verified according to whether it is a true/false statement about the topic. Now, whi le St rawson uses the term "topic" and "subject" interchangeably, Kuroda rejects the term "topic" and chooses "subject" because of the discourse-or iented nature of the former. Kuroda comments on this point as follows: (23) ...In fact, one might assume that the form of human judgement is uniformly something l ike the predicate form of modern logic, and any term, or set of terms, in such a form may be assigned the role of topic depending on discourse contexts. If such a claim is made, the concept of topic is kept val id as a val id concept, but the subject-predicate structure is not recognized as a (or the) form of judgement at a l l . [1972, p. 159] 203 However, "topic" in Reinhart's definit ion does not necessarily confl ic t wi th the "subject-predicate" structure in Kuroda's proposal. Strawson's (pragmatic) cr i ter ia are not the cr i ter ia for the R E L A T I O N but the c r i t e r i a for the properties of the chosen argument (= topic) in context. Reinhart's definition of topichood can therefore be divided into two parts: the relat ion between an argument and a proposition in a sentence, and the selection procedure of the topic in a discourse. I assume that the relationship underlying a topic sentence is context-free; that is, the subject-predicate relation and the "aboutness" relation reflects the internal structure of a sentence or the reinterpretation of that internal structure in discourse context. It is necessary to postulate something like the subject-predicate relation in semantics, or else one cannot explain why a sentence which contains an NP without any thematic role (cf. (1), (2), and (14)) can be realized as a sentence. §5.1.4 Wa in Contrastive Reading § "Contrastiveness" and the Categorical Judgement Kuroda (1965) assumes that the categorical judgement is a transformationally derived structure ( = syntactic structure) which is marked by wa, w h i l e wa i t s e l f has the inherent semantic feature, contrastiveness; that is, creating the contrastive effect and marking the categorical judgement are two different functions of the same wa, the former being primary and the latter secondary. On the other hand, Kuno 204 (1972) claims that there are two kinds of wa, the thematic wa and the contrastive wa. A natural consequence of his functional analysis is that Kuno postulates two types of wa. Hoji (1985) c la ims, however, that the functional 'd is t inct ion between "thematic" wa and "contrastive" wa have different syntactic representations. He proposes a base-generated structure (24a) for "thematic", and a movement structure (24b) for "contrastive": (24)a. Thematic wa S" NP V P 205 If wa marks an NP dominated by S", 9 then Taroo in (25) is non-contrastive; if wa marks an S dominating NP, then it is contrastive: (25) Taroo wa Tookyoo e itta. Tokyo to went 'Taroo went to Tokyo.' Hoji 's analysis is motivated by the poss ib i l i ty of variable/anaphora binding in "reconstruction" cases, which involves syntactic movement. If his analysis holds, here is a case where two different discourse-related functions, topic and contrastive in Kuno (1972), are represented by different syntactic structures. In general, there is no reason that a discourse function should be distinguished by i ts syntactic form, though an agreement might take place accidentally. In fact, Hoji 's syntactic d i s t inc t ion does not ful ly coincide w i th the d i s t inc t ion between "thematic" and "contrastive" in a functional sense, for the structure in (24b) is also shared by "scrambling" (Hoji 1985), which does not involve the part icle wa, and has a different discourse function (cf. Masunaga 1982). Furthermore, his contrastive wa examples are ones where wa has a heavy stress. While i t is observed in English that s t ress placement changes coreference relation (cf. Chomsky 1971 and 1976), which is handled in syntax, s t ress is not the primary means for conveying contrastiveness on topic in Japanese. Thus it is not clear whether Hoji is discussing a syntactic effect of stress placement , 1 0 where v i r tua l ly any element including wa can receive a stress, or an effect of (contrastive) wa. 206 In the next sect ions, I w i l l invest igate the nature of the contrastiveness in a wa-sentence. § "Contrastiveness" and Wa (I): Part icle Deletion In casual conversation, par t ic les are often dropped after the nominative and accusative NPs, as in (26): (26) a. Taroo 01 Hanako to Tookyoo e i t ta yo. with to went Emp. Taroo went to Tokyo with Hanako, you know.' b. Kimi 0i kore 02 si t te-(i)ru? you this know 'Do you know this?' According to Kuno (1973), the deletable particles are not ga and o, but wa and o; and the deleted part icle in a sentence l ike (26) is always wa instead of ga. Saito's (1982a) claim of an inherent case of the nominative NP is based on this observation of Kuno's. Evidence for wa-deletion comes from Japanese wh-words, which cannot be followed by wa, and in which part icle deletion is very di f f icul t , if not entirely impossible, as is the case in (27): 207 (27) a. Dare ga /?0 /*wa Hanako to kenkasita no. with quarreled Q 'Who had a quarrel wi th Hanako?' b. Dare ga /?0 /*wa Hanako ni denwa si ta no ka na. to phoned-l-wonder 'I wonder who phoned Hanako?' c. Dare ga /??0 /*wa yatta no. did Q 'Who did (this)?' Contrary to Kuno's claim, however, ga- deletion is quite possible. In the thetic sentences in (28), particle deletion s t i l l takes place: (28) a. (Ara,) ame 0 futte-(i)ru. (oh!) rain Nom is - fa l l ing '(Oh!) It's raining.' b. (A), dareka 0 kita. oh someone Nom came '(Oh!) Someone came (to the door).' 208 c. Mite. Ano hito 0 odotte-(i)ru. look! that man Nom is-dancing 'Look! That man is dancing.' The dif f icul ty of ga deletion after wh-words is due to another factor, possibly the focus nature of wh-words. (I w i l l come back to this topic in the section on ga. ) Likewise, no deletion of wa is reported in cases where i t is contrastive (Tsutsui, 1981 and 1984). It is reported that Emonds also points out the same fact in the data (in Hoji 1985). Some examples are provided in (29) to show that wa sentences in the contrastive construction and wa wi th stress cannot be deleted: 1 1 (29)a. Yamada wa/*0 motiron iku yo. Hayasi wa/*0 tabun of course go you-know probably ikanai ne. does-not-go l-would-say 'Yamada, of course, goes, you know. (But) Hayasi probably does not go, I would say.' b. Kudamono wa/*0 suki desu. • like 'I like fruit (but I don't like something else).' Accordingly, (30b) cannot be contrastive since a contrastive sentence cannot have 0 for wa, while (30a) is ambiguous depending on i ts context: 209 (30)a. Sono hon wa yomi-masita. that book read (polite) 'I read that book.' b. Sono hon 0 yomi-masita. This fact suggests that there is a strong correlation between wa and the contrastiveness which is not merely a contextual product. Terakura (1984) provides an interesting case in terms of the wa = 0 phenomenon. She claims that there exist cases where the absence of wa is obligatory. Such examples are given in (31): (31 )a. Yuube aru paatii e i t ta n da kedo ne. Soko de atta kimura to  yuu hito *wa/0 anata no kookoo no sempai da soo ne. (= Terakura (7b)) 'I went to a party last night. The person I called Kimura  that I met there. I understand (he) is a graduate from your high school.' b. Tyotto henna koto o ukagaimasu ga, Nihon e kaetta tomodati aa oite i t ta ueki *wa/0 watakusi no tokoro de wa basyo ga nai no de o-taku ni ikaga to omoimasite. Kanari ookii gomu no ki desu ga motte mairimasyoo ka. (= Terakura (7c)) 210 'I'm afraid I'm going to ask an odd question. A plant a  friend of mine left when he returned to Japan. I have no room for (it) at my place, so I'm wondering if you would like to have (it). (It) is a fairly big rubber plant. Would you like me to show (it) to you?' Terakura identifies the discourse function of these sentence in i t ia l NPs as being "not currently in addressee's consciousness" in Chafe (1976)'s sense, just like LD (left-dislocation) 2 in Prince (1983 and 1984). However, Terakura's examples l ike (31) can in fact have wa, contrary to what she has claimed, but then the discourse function of the NP {-wa ) have been changed 1 2 (the contrastive feature of wa is more or less apparent compared wi th 0 version) and they are no longer fel ici tous in those contexts. Thus, "obligatory (absence)" means "obligatory" at least on the level of discourse. However, (3 1) may also be syntactically different from so-cal led topic sentences. ' 3 It seems to me that (29) belongs to the same group of sentences as (32) below: (32) Hi tori damatte sake o nomu, Taroo wa sonna otoko da. alone quietly ACC drink such a man 'Drink(ing) sake alone and quietly, Taroo is such a person.' The example (32) does not f i t into the def in i t ion of categorical judgement, and there is no wa version for (32) since the sentence in i t ia l element seems like a preposed relative clause wi th sonna as the pro-form; i . e. [hitori damatte sake o nome( = sonna)] otoko "a man who 21 1 drinks sake alone and quietly." If (31) is a (32)-type sentence, (31) cannot be dealt wi th as a categorical judgement. Now, the non-deletabil i ty of wa in contrast ive environments shown in (29) and the contrastive wa version of (3 1) lend support to the assumption that wa has an inherent semantic feature. However, i f the particle wa is the one which contributes to the contrastiveness, why is it the case that (33), which is a typical non-contrastive ( = "thematic"), general statement has the contrastive reading given in (34)? (33) Kuzira wa honyuurui da. whale mammal 'A whale is a mammal.' (34) Kuzira wa honyuurui da ga same wa gyorui da. whale mammal but shark fish 'A whale is a mammal but a shark is a fish.' The next section w i l l be devoted to investigating this problem. § "Contrastiveness" and Wa (II): Contrastive Construction Contrastive reading is typical ly associated wi th wa- sentences like (35): 212 (35) Ani wa futuu no kaisyain da ga older brother ordinary company employee but otooto wa yuumei na komedian da. younger brother famous comedian 'Older brother is an ordinary company employee but younger brother is a famous comedian.' The use of wa here is called contrastive (Kuno, 1972). One might say, however, that the contrastive reading results from the contrastive context: each sentence in i so la t ion does not necessar i ly have a contrastive implication, as shown in (36): (36) a. Ani wa futuu no kaisyain da. 'Older brother is an ordinary company employee.' b. Otooto wa yuumei na comedian da. 'Younger brother is a famous comedian.' Nevertheless, there are reasons to assume that wa has something to do with contrastiveness. First , consider (37): (37) a. Ame wa hutte-imasu. ( = Kuno (18a)) rain falling-be 'Rain is falling.' 213 b. Oozei no hito wa paatee ni kimasita. ( = Kuno (19a)) many people party to come 'Many people came to the party.' In Kuno's themat ic-contras t ive d i s t inc t ion , although generic sentences l ike Kuzira wa honyurui da ("Whales are mammals") are typical examples which have the thematic wa, many sentences are ambiguous between thematic and contrastive (e.g. Biru wa gakusei da, " B i l l is a student"), and some only have the contrastive wa. Examples of the latter (unambiguously contrastive) case are in (37). The sentences in (37) cannot be thematic since they would violate the res t r ic t ion that a thematic NP must be generic or anaphoric, but they are perfectly al l right as contrastives, as shown in (37'): (37')a. Ame wa hutte-imasu. Demo kaze wa huite-imasen rain i s - f all ing but wind is-not-blowing kara daizyoobu desyoo. as alright w i l l be 'It is raining. But as the wind is not blowing, it w i l l be alright.' 214 b. Oozei no hito wa paatee ni kimasita ga paatee wa many people party to come but party tittomo moriagarimasendesita. at all was-not-a-success Many people came to the party but the party (itself) was not a success at a l l . It is l ikely the case that even if sentences like (37) were given without context, one would embed them in an imaginary context to fully interpret them. It is the sentences which demand such fe l ic i tous contrastive environments; they are ungrammatical otherwise. Sentences which are grammatical only in discourse are not allowed, on principle. Thus, the examples in (37) provide cases where the sentences in isolat ion already imply the contrastive reading; that is, wa seems to have a contrastive feature intr insic to it. However, the problem s t i l l remains as regards whether the sentences in (36) are different from the sentences in (35); are they two different wa, the contrastive wa for (35) and the thematic wa for (36), as Kuno c l a i m s ? 1 4 To answer this question, the role of the contextual environment w i l l be investigated next. First , consider (38) and (39): 215 (38) a. Ame ga hutte-kita. rain come-to-fall 'Rain has started to fall . ' b. Tuki ga dete-iru. moon is-there 'There is the moon.' (39) a* Ame ga hutte-kita ga tuki ga (mada) dete-iru. b. Ame ga hutte-kita ga tuki wa dete-iru. 'It has started to rain but the moon is ( s t i l l ) there.' c. Ame wa hutte-kita ga tuki ga dete-iru. d. Ame wa hutte-kita ga tuki wa dete-iru. Ame in (38a) and tuki in (38b) are non-generic and also non-anaphoric when one is describing the outside scene, which is a character is t ic of thetic sentences. Thus, it is necessary to use ga instead of wa in (38). However, once embedded in a [clause] ga [clause] construction, wa is neces sa ry ' 5 in both clauses, as in (39d), or at least in one of the two clauses, as in (39b) and (39c). If it were really the case that because of the [clause] ga [clause] construction, two events/ideas get contrasted, then (39a) should be grammatical without changing ga into wa, but it ac tual ly is n o t . ' 6 On the other hand, if the [clause] ga [clause] construction requires a wa- sentence(s) to play a key role in creating 216 the contrastive effect, one can explain why wa is needed in (39c). But the semantic function of the [clause] ga [clause] form i t se l f does not seem to have anything to do with the contrastiveness; it is very common to find non-contrastive sentences in this construction, as in (40) where ga can be translated as "and" rather than contrastive "but": (40)a. Boku no tomodati de Yamada to-iu no ga iru-n-desu I Cs) friend and called one Nom be ga kore ga taihenna kawarimono desite... and this person Nom quite a character be (Ger.) 'I have a friend called Yamada and this person is quite a character and...' b. Kono biiru wa nihon no desu ga nakanaka ikemasu this beer made-in-Japan be and pretty good yo. you know 'This beer is from Japan and it 's pretty good, you know.' .What the [clause] ga [clause] construction does here is to provide a felici tous environment ( = context) for the wa-sentence, just as in the case of "discourse conditioned" (Williams, 1977) sentences, for example a passive sentence which seems to require a specific environment in discourse, or a cleft sentence as a less controversial case . 1 7 Moreover, it is the wa sentence which is d i r ec t l y connected w i t h the 217 contrastiveness. The reason why the sentences in (36) in isolation are apparently neutral in terms of contrastiveness w i l l be explained by the nature of contrastiveness in § In the following two sections, I w i l l examine the nature of subject NP in wa-sentences in relation to the contrastiveness feature. § "Contrastiveness" and Wa (III): Indefinite Nouns Kuroda (1965, 1972) claims that his subject "cannot be indefinite specific" for the reason given below in (41): (41) Now assume that one has made a categorical judgement in which the subject is referred to by an indefinite noun in referential use. The speaker's mind must then have been directed not towards i ts i nd iv idua l i t y , but towards i t as a comple te ly unmarked representat ive of those individual en t i t i es that share the attribute represented by the noun used as the subject, say, dog. Then, if the speaker assigns a certain attribute expressed by the predicate to the subject, he would in fact be assigning this property to an arbitrary individual entity that might be named by the particular attribute used to refer to it, dog. In other words, he would, in effect, have made a generic statement concerning the class of objects, dogs, contradicting our in i t ia l assumption that he intended to make a statement about a specif ic object he is referr ing to. Thus, no subject in the logical sense can be indefinite specific. [Kuroda 1972, p. 167] Thus, he claims that (42a) can be translated wi th either wa or ga, but (42b) must be translated with ga. 218 (42) a. The dog is running. 'Inu wa/ga hasitte-iru. ' b. A dog is running. 'Inu wa*/ga hasitte-iru. ' The argument in (41) is explicated further in Kuroda (1965) as follows: a categorical judgement (also called "predication") consists of a logical premise and a conc lus ion , 1 8 as in (43a). Supposing (42b) were a wa-sentence, it would have to have the same form as (43b). (43) a. If X is A, then X is B. (is true) b. If it is a dog, then it is running. However, the i f -c lause in (43b) cannot be a logical premise of the predication. Thus, (42b) cannot be a wa- sentence. The reason why (44a) is ungrammatical w h i l e (44b) is grammatical is also explained by Kuroda in the same fashion: (44) a* A man is intelligent, b. A man is mortal. Man in (44b) has to be interpreted as generic, instead of a specific "a man"; while in (44a) a man cannot obtain such a generic (indefinite non-specific) interpretation; thus, the sentence is ungrammatical. 219 (45), on the other hand, is an apparent counter-example to this: if basu in (45) refers to one in sight of the speaker, that is an indefinite specific, "a bus." (45) Basu wa hasi t te-iru bus is-running '[A bus is running.]' (This translation is usually equivalent to the ga version of the sentence.) The sentence also sounds perfectly all right in contexts such as (46): (46) Are, basu wa hasitte-iru. Densya wa tomatte-iru why! electric train be-stopped noni naa. though (interjection) 'Why, [a bus is running], though the electr ic trains are stopped!' Nevertheless, (46) is not a genuine indefinite specif ic case, for the speaker is really giving a general statement on buses through discussion of a single instance. The speaker can consider the statement true in an area which is his concern at that moment. A generalization about the indefinite subject is: i f there is an indefinite noun in the subject position of the categorical judgement, it 220 must be non-specific. Thus, in (44a) "a man" has to receive the generic reading. Now, the cases which require serious attention here are those where an indefinite non-specific sentence semantically fa i ls to obtain the generic reading. The sentence is s t i l l grammatical if read with a certain contrastive implication. Consider (47) and (48): (47) Nanninka wa buzi nogareta some people safely escaped 'Some people escaped safely.' (48) a. Dareka wa byooki desu. (= Kuno (18c)) someone sick is 'Someone is sick.' b. Dareka wa (itumo) kesseki desu. someone (always) absent is 'Someone is always absent.' (47) and (48) are categorical judgement sentences (wa- sentences) wi th indefinite nouns as their subjects, "nanninka" and "dareka." They are grammatical only when read wi th a contrast ive impl ica t ion . For instance, in (47) "some who managed to escape" is contrasted wi th "the others who did not." The situation is exactly the s ame 1 9 in (48): the sentence may only have this reading wi th a contrast between "some" and "the others." More examples of this type w i l l be found in cases involving numerals: 221 (49)a. Sono paatee ni gozyuunin wa kita. that party to fifty people came 'Fifty people came to that party.' b. Yonin wa totyude dete-itta. four people in the middle went out 'Four people left in the middle.' Japanese does not have (in)definite particles. Gozyuunin in (49a) and yonin in (49b) can be definite, as "the f i f ty people" and "the four" respectively, and then they are no different from other categoricals. On the other hand, if gozyuunin and yonin are indefinite, then they must have a contrastive reading: the "four people" in contrast to, say, "the people who stayed there" in (49b). This contrastive phenomenon is also observed in Kuno (1 973). The questions to be asked now are: why is the contrast ive reading compulsory in these cases?; and again, what is the relationship between the generic reading ("thematic" in Kuno (1973)) and the contrastive reading ("contrastive") in indefinite nouns? These questions w i l l be answered in the next section. 222 § "Contrastiveness" and Wa (IV): Subject NPs and Definiteness Effect. There are some l ingu is t i c phenomena where the d i s t inc t ion between definite and indefinite plays a key role. This has often been called the "definiteness effect" or the "definiteness restr ict ion." A typical example of the DE is observed in the existential there- sentence in English, as in (50): (50)a. There is a table in this room, b^There is the table in this room. Only the indefinite NP a table is allowed in this construction. Unlike English, Japanese does not have (in)definite ar t icles. However, beyond the existence (or non-existence) of morphological marking devices, the notion of (in)definiteness comes into play in many languages, if not a l l , in the form of various "definiteness effects" (Gi l , 1987). Milsark (1977), however, claims that the definiteness res t r ic t ion on the there-sentence has been too narrowly defined. The NPs which are excluded from this environment are not only "the NP" but also a group of NPs which he called "strong" NPs. The "weak" versus "strong" dis t inct ion shown in (51) is intended to describe the actual restr ic t ions on there-sentences. 223 (51) WEAK a sm [of "sm" vs. "some" in Postal (1969): M.O.] number determiners 0 plural and mass determiner in nonuniversal reading STRONG "definites" the demonstratives pronouns possessive DET's universals all every each any when not polarity item of some 0 DET in universal reading [Milsark (1977), p. 8] Milsark identifies the dist inct ion of the "weak" versus the "strong" as that of "cardinality" versus "quantification." Here, a definite ar t ic le like the English the is viewed merely as a l inguist ic manifestation of the quantificational effect (cf. Chomsky (1975) for the as a universal 224 quantifier). After a l l , it is already clear from a language like Japanese that the existence of the ar t ic le is a rather different issue from the existence of the concept "definiteness." 2 0 Now, according to Milsark, words like some, many, few, and their analogous expressions are treated as ambiguous 2 1 between the non-quantified reading (weak) and the quantified reading (strong). For example, in his theory, the strong sense of "some" salesmen is nearly synonymous wi th "some" of the salesmen; but more precisely what the strong reading means is "some" of the l imited set of the group. That is, "some" people could be "some of the students" or "some of the people on the earth." What Milsark is not aware of, however, is that number determiners (numerals including a as a singular determiner) are also ambiguous between the quantified and non-quantified readings, or "strong" and "weak." The strong reading of a in a man is the generic reading, that is , the universally quantified man, whereas the weak reading is the numerical interpretation of "a." Returning to wa, a definiteness effect on the bare NPs construed as (in)definite, apparently more complicated than was the case wi th there, now appears more clearly when viewed in terms of the nature of the NP (-wa ) . 2 2 The key examples here are the indefinite noun subjects, as in (47) - (49), duplicated below as (52) - (54): 225 (52) Nanninka wa buzi nogareta some people safely escaped 'Some people escaped safely.' (53) a. Dareka wa byooki desu. (= Kuno (18c)*) someone sick is 'Someone is sick.' b. Dareka wa (itumo) kesseki desu. someone (always) absent is 'Someone is always absent.' (54) a. Sono paatee ni gozyuunin wa kita. that party to fifty people came 'Fifty people came to that party.' b. Yonin wa totyude dete-itta. four people in the middle went out 'Four people left in the middle.' Nanninka, dareka, gozyuunin, and yonin, in these examples, are respec-t ive ly member(s) of certain l imi ted sets: for instance, (52), some people of the sinking ship; (53), someone among my classmates; (54), fifty people on the invitation list, or four people at the meeting. The domain of the quantification is pragmatically defined: only context 226 determines what the "limited sets" in question are. Kuno's often-quoted claims in (55) are correct at some level of the description: (55) a. Wa is either thematic or contrastive b. Themes must be anaphoric or generic. Non-anaphoric nongeneric themes result in ungrammaticality. c. On the other hand, nonanaphoric nongeneric noun phrases can be contrasted. [Kuno 1973, p. 59-60] However, an integrated explanation of the nature of the NPs is to be found in a rather different manner. Fi rs t of a l l , I do not accept the binary approach to the function of wa, the division into thematic and contrastive, but rather consider these two the result of the interaction between the nature of the subject NP (semantics) and the context (pragmatics). Categorical judgement requires a quantificational nature for the subject NPs. "Fifty people" can be the uniquely identifiable entity in the context, "the f if ty people" (i.e., "definite" in the usual sense), or members of the uniquely identifiable set, "fifty people of the group." In the latter , these l imi ted sets are uniquely identif iable because the domain of quantification is pragmatically determined by context. Ladusaw (1982) also c l a ims that X of the pa r t i t i ve construct ion "xn of the X" (e.g., "three of the men") is "a group level ind iv idua l , denoting the set of a l l propert ies that the unique contextually relevant group of men have" (p. 238). This "group level individual" (as opposed to "individual entity") is an equivalent notion to Kuroda's subject quality of the categorical judgement (cf. (41)). Then, what the contrast ive reading in indefinite nouns means should be 227 considered in the light of the nature of the subject NP. To do so, let us look at indefinite noun cases in contrastive environments. A typical environment is given in (56): (56)a. Nanmin-tati wa hidoku tukarete-ita. Ooku ga/wa sono mati refugees very was-exhausted many the town ni sibaraku todomaru-koto-ni-sita. Daga nanninka wa ACC for a while decided-to-stay but some sarani saki ni susunda. further ahead went 'The refugees were very exhausted. Many of them decided to stay in the town. But some of them went further.' b. Takasi wa usagi o katte-iru. Nihiki wa massiro no (NAME) rabbit ACC be-raising two pure-white yatu, sanbiki wa makkuro, sosite nokori wa zenbu one three pure-black and the rest al l buti da. speckled 'Takasi is raising rabbits. Two of them are pure-white ones, three of them are pure-black, and the rest are speckled.' 228 In (56a), "many of them" contrasts wi th "some of them"; in (56b) "two of them," "three of them," and "the. rest" create the contrastive effect. Let us examine the subject NP closely in (57): (57)a. Takasi ga usagi o kai-hazimete itinen ni NOM start to raise (Gerund) one year to naru. Nanbikika wa moo sudeni kodomo o become some of them already young ACC unde-iru. have-borne 'One year has already passed since Takasi started to raise rabbits. Some of them have already borne young. b. Takasi ga usagi o kai-hazimete itinen ni naru. Sono uti no among them nanbika wa moo sudeni kodomo o unde-iru. 'One year has already passed since Takasi started to raise rabbits. Some of them have already borne young.' The difference between (57a) and (57b) is that (57b) has an "of the X" expression, sono uti no, which works to reinforce the connection between the whole number of rabbits and some (of them), while (57a) has no such phrase. An interesting phenomenon observed here is that the 229 contrastiveness seems to be a l i t t l e "faded" in (57b). Compare (58) with (59): (58) a. Nanninka wa buzi nogareta. some (people) safely escaped 'Some (of them) safely escaped.' b. Hansuu wa itumo atumari ni okureru half always meeting be-late 'Half (of them) are always late for the meeting.' (59) a. Sono fune no nanninka wa buzi nogareta. that ship 's some people safely escaped 'Some people in that ship escaped safely.' b. Watasi no guruupu no hansuu wa itumo atumari ni okureru my group 's half always meeting be-late 'Half of the people in my group are always late for a meeting.' Again, (58) has a stronger contrastive implication than (59) wi th "of the X." My assumption here is that the lack of an expression to l imi t the set in (58) indicates that the set is easily identifiable in the context. In other words, these sentences in (58) are strongly related to the previous discourse w i th i t s contextually unique set. On the other hand, the existence of an expression l imi t ing the set, in that ship or in my group 230 means that the set is less identifiable and needs explici t mentioning. In other words, without these expressions, the connection between the two sentences is weak. The relation between "strong relatedness" and "contrastiveness" w i l l be examined further in the next section, and an overall picture of the contrastive effect w i l l be provided there. § The Nature of "Contrastiveness" Through the observations and arguments in the previous sections, I have come to the conclusion that the contrast ive reading of wa sentences, which has often been used as a p r imi t ive notion, is the product of both the intr insic and the extrinsic nature of a wa sentence. There are two significant factors involved in inducing a contrastive quality in the sentence: one is related to the semantic property of subject NP of a categorical sentence, and the other is related to the contextual environment for the sentence in question. I w i l l discuss the former first . The fol lowing three sentences represent: a generic sentence, which is non-contrastive ( = (60a)); a non-generic sentence, which is ambiguous ( = (60b)); and a sentence wi th an indefinite noun as i ts subject, which is always contrastive ( = (60c)). 231 (60)a. Kuzira wa honyuurui da. 'A whale is a mammal.' b. Taroo wa gakusei da. 'Taro is a student.' c. Gonin wa kita. 'Five (of them) came.' Instead of postulating two kinds of wa, thematic and contrastive, their differences w i l l be explained in a uniform manner as follows: Suppose that "X is contrastive" means that X contrasts wi th Y. Then X and Y w i l l have something like the figure-ground distinction: the ground is needed for the figure to stand out. In (60c), gonin automatically receives the part i t ive reading because of the nature of subject NP of categorical judgement. The "of X" portion is the "ground" upon which the "figure" gonin stands out. A careful observation reveals, however, what actually contrasts is not just gonin and "of X" but five people's coming vs. the other people (= X - 5)'s not coming. Likewise, in (60b) where i t has the contrastive reading, Taro's being a student contrasts wi th the other's not being students or perhaps the other's not yet having a clear status in the set consisting of both kinds of people. In a generic statement l ike (60a), whale's being a mammal is set against the other (= whole set of living creatures minus whales)'s being in non-specified classes. Notice that generally speaking, when we 232 compare two, we always compare the two on the same ground. As in the case of figure-ground opposition, the vaguer and broader the ground is, the less salient the figure becomes. This is why (60a) is generally considered as non-contrastive. In §5.1.4 .5 , I used the term "strong relatedness" in connection with the overt/covert case of "of X." If X is obvious from context, it means that the contrasting party is clearly definable in discourse. In a generic sentence like (60a), where "of X" is conventionally definable by our knowledge of the world, the contrastive effect almost fades out. However, once (60a) is embedded in a context like that found in (61) ( = (34)), the situation becomes a l i t t l e different: (61) Kuzira wa honyuurui da ga same wa gyorui da. whale mammal but shark fish 'A whale is a mammal but a shark is a fish.' Now the context clar i f ies what contrasts wi th what; i.e., whale's being a mammal to shark's being a fish. The subject status of whale and shark makes the two nouns prominent; however, the charac ter is t ic of wa contrast is not item contrast (cf. the exhaustive ga in the fol lowing section), but rather statement contrast. I w i l l discuss the statement contrast in further detail next. 233 Firs t observe (62): (62)a. Taroo wa hon o katta ga book Acc bought but Ziroo wa (hon o) kawanakatta. did-not-buy 'Taro bought a book but Ziro did not buy one.' b. Taroo wa hon o katta ga Ziroo wa kuruma o katta. car bought 'Taro bought a book but Ziro bought a car.' c. (Sono okane de) Taroo wa ie o katta ga that money with house bought Ziroo wa bizinesu o hazimeta. business started '(With that money), Taro bought a house but Ziro started a business. In (62), one might say that Taroo and Ziroo are in contrast; however, when [X\-wa - Y i ] is contrasted with [ X 2 - w a - Y 2 ] , not only Xi and X 2 but also Y] and Y 2 contrast, as is evident from the resultant ungrammatical sentences 2 3 in (62'a): 234 (62')a* Taroo wa hon o katta ga Ziroo wa hon o katta. Taro bought a book but Ziro bought one.' b. Taroo wa hon o katta ga Ziroo wa kuruma o katta. 'Taro bought a book but Ziro bought a car.' However, in two contrastive statements, wa marked items serve as focal points, in the sense of Dretske (1972). Next, let us examine the case of mul t ip le-subjec t constructions in terms of the statement contrast. Observe (63): (63)a. [Supootu wa [(fuyu ni) sukii wa simasu]] ga... sports winter in ski do but 'As for sports, I ski in winter, but...' b. [Watasi wa [asa wa [koohi wa nomimasen.]]] I morning coffee do-not-drink 'I don't drink coffee in the morning.' In the multi-subject construction, it is observed that the leftmost NP-wa is thematic and the rest are contrastive (Kuno, 1973). This can be explained by assuming a nesting structure for those sentences. For example, the subject NP is sports in (63a) and among the possible 235 descriptions concerning sports, ski is chosen as the second subject. This relation in (63a) and (63b) is i l lustrated in (64a) and (64b) below: (64)a. In (64b), on the ground of the description about /, the description about morning is sal ient , which in turn serves as the ground for the description about coffee. Rather than the dichotomy between thematic and contrastive, the contrastive effect is gradual in this schema - the more restr icted the ground is, the more dis t inct ive the figure is - and this graduation, I believe, matches our intuition about the sentence. 236 Next observe (65): (65)a. Watasi wa asa wa koohii o nomimasen ga I morning coffee Acc do-not-drink but hiru (ni) wa (koohii o) nomimasu. daytime drink b. Watasi wa asa wa koohii wa nomimasen ga otya wa nomimasu. 'I don't drink coffee in the morning but drink tea.' c* Watasi wa asa wa koohii wa nomimasen ga hiru wa otya wa nomimasu. 'I don't drink coffee in the morning but I drink tea in the daytime.' d* Watasi wa asa wa koohii wa nomimasen ga hiru wa koohii wa nomimasu. 'I don't drink coffee in the morning but drink it in the daytime.' The paral lel construction provides a fe l ic i tous environment for the contrastive effect of wa sentences; in (65a) the description of morning and that of daytime contrast on the ground of the descr ipt ion concerning / ; in (65b), the description of coffee and that of tea contrast on the ground of the description concerning morning, for which the desc r ip t ion concerning / serves as the ground. There is a 237 res t r ic t ion against two nesting descriptions contrasting wi th another two nesting descriptions at the same time (cf. (65c)); in other words, once the descriptions contrast wi th each other, they cannot serve as the ground for another descr ip t ion (cf. (65d); notice that (65a) is grammatical). (65d) is probably a special case of (65c), or the other way around. Thus, (66), Kuno's counter-example for (65c) and (65d), which is reported in Enomoto (1983), is a case where two parallel descriptions concerning / contain more than one contrastive wa for each: (66) Boku wa syuumatu ni wa hon wa yomu ga zassi wa yomazu I weekend book read magazine do-not-read syuuzitu ni wa sinbun wa yomu ga hon wa yomanai weekday newspaper read book do-not-read 'On weekends, I read books but do not read magazines; however, on weekdays I read newspapers but do not read books.' However, this is not a genuine counter-example. There are two types of parallel here: one is my weekend's book-reading vs. my weekend's non-magazine-reading (also my weekday's newspaper-reading vs. my weekday's non-book-reading ); the other is my weekend's activity vs. my weekday's activity. That is, (66) has the following nesting structure of (66'a), which is also i l lustrated using a tree structure in (66'b): 238 (66')a [\ Boku wa [2 [3 syuumatu ni wa [4 [5 hon wa yomuls ga 5[zassi wa yomazuls U [3 syuuzitu ni wa [4 [5 sinbun wa yomub ga [hon wa yomanails Uh h h Boku-wa syuumatu-wa syuuzitu-wa hon-wa zassi-wa-(not) sinbun-wa hon-wa-(not) As shown in (66'b) the items in contrast have to be under s is ter nodes in the pa ra l l e l i sm. Thus, whi le (66'a) is a perfect ly we l l - fo rmed contras t ive sentence, (67b) below is not, since kuruma-wa and konpyuuta-wa do not have a s is ter relationship as i l lustrated in (67c): (67)a. Taroo wa kuruma 0 katta ga car bought Ziroo wa konpyuutaa 0 katta. computer bought 'Taroo bought a car but Ziroo bought a computer.' b* Taroo wa kuruma wa katta ga Ziroo wa konpuuta wa katta. 239 kuruma-wa konpyuutaa-wa Note that the contrast ive paral le l ism does not necessar i ly occur along with syntact ica l ly parallel structures like (65). Look at (68): (68) [Taroo wa katta] hon o Ziroo wa kawanakatta. bought book Acc did-not-buy 'Ziro did not buy the book Taro bought.' Here, the relat ive c l a u s e 2 4 and the main clause contrast; yet, Taroo-wa and Ziroo-wa are in a s is ter relationship in the contrastive paral le l ism scheme. Summing up, the in t r ins ic ground-f igure contrast of the wa-sentence on the one hand, and the ef fect of the ex t r ins ic contextual p a r a l l e l i s m 2 5 on the other, in teract to produce what we feel as contrastive in actual discourse. 240 §5.2 Ga and Wa §5.2 .1 Ga vs. Wa The copula sentences which we saw at the beginning of this chapter do in fact have ga counterparts, as shown in (69): (69) a. Boku wa susi da. '[lit.] As for me, it is sushi.' b. Boku ga susi da. '[lit.] i am sushi.' Min i -d iscourses for (69a) and (69b) are given in (70a) and (70b) respectively: (70) a. Q: Kimi wa nani (o taberu)? you what Acc eat 'What are you going to eat?' A: Boku wa susi da. b. Q: Dare ga susi (o tyuumonsita no)? who Nom ordered Q 'Who ordered sushi?' A: Boku ga susi da. Nani "what" in (70a) and dare "who" in (70b) are information-seeking words, and susi and boku "I" supply the information for them. Kuroda 241 cal ls this "focus in the sense of Chomsky (1970)." Kuno (1970) cal ls the use of ga here exhaustive l i s t i n g . 2 6 The interesting fact is that A in (70b) has its alternative (7 1): (71) Susi wa boku da. '[lit.] As for sushi, it is me.' Now then, let us assume tentatively that the choice of wa and ga marks the difference of the order of information as shown in (72): (72) a. Kimi wa nani? Boku wa susi da. ( = (69a)) Old New b. Dare ga susi? Boku ga susi da. ( = (69b)) New Old Susi wa boku da. ( = (71)) Old New Supposing that the old-new is marked by wa, and the new-old is marked by ga, then in (72b) the reversal in the information order seemingly determines the choice between the two particles. However, the problem is that although wa marks old-new order exclusively, ga may mark new-new order, which is the thetic case, in addition to new-old order. Thus, ga does not exclusively mark a reversed information order of the case of wa. Another problem is that the dichotomy of old-new or new-242 old divided by wa or ga in a sentence is not the structure found in many cases. For instance, look at (73): (73)a. Kimi wa nani o taberu? 'What w i l l you have?' Boku wa susi o tabevoo. I sushi wi l l - ea t old new old 'I w i l l have sushi' b. Kimi wa kono okane o doo suru no. you this money Acc how do Q 'What are you going to do with this money?' Boku wa kono okane o tvokinsimasu. I ( this money Acc save old old new 'I w i l l deposit this money (at the bank).' In (73a) wa occurs after old and the rest is new-old; in (73b) wa occurs after old and the rest is old-new. If the reversal shown in (72) is assumed to take place in predication instead of the informational structure, then both (69a) and (69b) are considered as categorical sentences. The difference between (69a) and (69b) is that in (69a) the predication is done in normal order, but in (69b) it is done in reversed order. Ogiwara (1986), in fact, develops this analysis of the reversed order of predication wi th in Montague Grammar. This line of approach 243 not only contributes to the establishment of an integrated theory of wa and ga, but also provides a good perspective to understand one of the mechanisms operating in discourse. I w i l l discuss this topic further in the next section. §5.2.2 Wa, Ga, and Discourse Unlike English, Japanese does not have clef t sentence con-structions equivalent in syntax. However, as far as discourse function is concerned, non-thetic ga sentences serve a s imi la r function as cleft sentences. 2 7 Let us take a further look at this point. Lacking cleft sentences, Japanese has pseudo-cleft sentences as in (74), which is in the construction of (75): (74)a. Kono tegami o kaita no wa Taroo da. this letter Acc wrote The person who wrote this letter is Taroo.' b. Taroo ga tyuumonsita no wa unagi da. nom order eel 'What Taroo ordered is eel.' 244 c. Taroo ga tegami o dasita no wa Pari (kara) da, letter sent Paris 'Where Taroo sent a letter is from Paris.' (75) [ N P [s ] no ] wa NP da. (The pre-copula NP may be followed by a case-marking particle.) Now, the status of no in the pseudo-cleft is controversial. Kitagawa and Ross (1 982) argues that this no is a pre-nominal modifier, the same one used in (76), under the assumption that no connects not only NP and NP but also S and NP (cf. Soga and Fujimura, 1978). (76) Watasi no hon I book [IIN P no [book]NP Thus the relat ive clause in pseudo-cleft is assumed to be headed by a null pronoun PRO instead of a real NP. On the other hand, Haig (1983) proposes that no is a nominal head of the relative clause, assuming two cases for it: re ferent ia l and non-referent ia l . An in teres t ing consequence of Haig's proposal is: if (74b), for example, is a case of referential no, then no works more like the pronoun "one" in English and there is no difference from the ordinary copula sentence; on the other hand, i f it is a case of non-referential no, then the sentence is 245 equivalent to the specificational pseudo-clef t 2 8 in Higgins (1972). In other words, in the referential case, the sentence is equivalent to (77): (77) Yamada wa watasi da. 'I am Yamada.' (77) is often used almost interchangably with (77'): (77') Watasi ga Yamada da. 'I am Yamada.' Ga in (77') has the focus/exhaustive l i s t ing reading. 2 9 On the other hand, the specif icational pseudo-cleft only allows wa, and interestingly enough, if the order is reversed (which, unlike predicational, is possible in English (cf. Wi l l i am, 1983)), ga (ex-l ist ing) is needed, as shown in (78) and (79): (78) a. Kono tegamo o kaita no wa Taroo da. 'The person who wrote this letter is Taroo.' b. Taroo ga kono tegami o kaita (no da). 'It is Taroo who wrote this letter.' (79) a. Taroo ga tyuumonsita no wa unagi da. 'What Taroo ordered is eel.' 2 4 6 b. Unagi ga Taroo ga tyuumonsita no da . 3 0 (78b), for example, has the thetic reading, but the focus reading is the one which pairs wi th (78a) in a functional sense. Now, although it is probably the case that the reversal of the order takes place either in semantics (Montague framework) or in syntax (Wil l iams) , which are almost identical in a functional sense, the alternation of wa/ga is determined by different discourse factors. (77) is used, for instance, in the following environment: (80) Yamada-san wa donata desu ka. who 'Who is Mr. Yamada?' Minasan Yamada to- iu otoko o osagasi no yoo desu ne. everybody called man look for seem Yamada wa (zitu wa) watasi desu. as a matter of fact 'It seems that people are looking for a man called Mr. Yamada. (As a matter of fact), Mr. Yamada is me.' In (80), topic continuity - once a topic is established, it is maintained for a while (cf. Givon (1983)) - is involved; Yamada is repeated in the third sentence. On the other hand, (81) is a typical environment for (77'): 247 (8 Da. Dare ga Yamada-san desu ka. who Mr. 'Who is Mr. Yamada?' Watasi ga Yamada desu. 'I am Yamada.' b. Anata no oture ga Yamada-san desu ka. your company 'Is your company Mr. Yamada?' lie, zituwa watasi ga Yamada (nan) desu. No as a matter of fact (emphatic) 'No, as a matter of fact, I am Yamada.' The exhaustive ga is used in a wel l - formed question-answer pair in (81a) or a parallel construction in (81b), just like focus in English (cf. Rochemont (1985)). Now, as has already been observed in the previous section, wa is associated wi th contrastiveness; and the exhaustive-l is t ing ga is also associated wi th a type of contrastiveness: watasi contrasts wi th anata no oture in (81b). In this sense, the contrastiveness in ga is not different from that found in wa in § That is, the NP is selected and contrasts wi th the rest of the members of a certain set, and the 248 contextual (including structural) paral le l ism induces the contrastive effect in discourse. (82) Watasi wa Tookyoo e ikimasu ga. I Tokyo to go Yasuda-san wa Nyuuyooku e ikimasu. New York to go 'I go to Tokyo but Mr. Yasuda goes to New York.' (83) Watasi (ga Tookyoo e iku no) de wa nakute I Tokyo to go is-not-the-case Suzuki-san ga Tookyoo e ikimasu. Tokyo to go 'It is not the case that I go to Tokyo, but it is the case that Mr. Suzuki goes to Tokyo.' The difference is , however, in the case of wa, i t is the statement contrast - the predicate should be different as wel l as the subject in (82) - but the subject becomes the focal point for the contrast; on the other hand, ga is not the statement contrast and the predicate may be identifiable as the same kind in interpretation. It seems to be the case that the exhaustiveness in ga is a secondary induced factor, part ial ly because of the nature of the question-answer environment - you have to give information no more or no less than what is needed (Grice's conversational maxim of Quantity) - and probably more fundamentally, 249 it would seem to be related to the non-"class" reading of the NP (-ga ), unlike the NP (-wa ) (cf. the nature of the subject in (41) in §, although the details are not susceptible of proof at this stage. Now, Japanese is often called a topic-prominent language because of the wa -marked NP, while in a language like English focus marking (constructionally/phonologically: cf. Rochemont (1985)) is far more prominent than topic. In Japanese, (sentence) topic is always marked by wa ; furthermore, when a discourse is carried out on the same topic, or the topic is obvious from context, NP (-wa ) is expressed by 0. The lat ter character is t ic , the deletabi l i ty (or, null topic) of NP-wa , may lead us to make a claim such as (84): (84) Any sentence can potentially have the NP-wa element at the sentence in i t ia l position in discourse, wi th the exception of thetic sentences. Zero topic analysis for zibun in Chapter 2 is based on the above assumption. Apparent thetic sentences may be categorical, if it is bound by a temporal/ locative element (with wa ), which is, in turn, 0, as shown in (85): (85) a. Hi ga tette-imasu. sun is-shining 'The sun is shining. 250 b. Soto wa/kyoo wa/0 hi ga tette-imasu. outside/today 'Outside/today the sun is shining.' The t ic sentences usual ly cannot be negated (Kuroda 1969 and elsewhere), as in (86a); but categorical can, as in (86b): (86)a* Hi ga tette-imasen. Neg b Soto/0 wa hi ga tette-imasen. Neg As is discussed under eel-sentences, the possibly non-thematic nature of NP-wa enables us to add a discourse topic, say "Mary," to a sentence during the course of interpretation, wi th the sense of "speaking.of Mary." When we have on the one hand a language which overtly marks what we are talking about (topic prominent) al l the time, and on the other a language which marks what is in focus (focus prominent), it may not be unreasonable to expect that these two languages w i l l use different strategies in discourse formation. I w i l l leave this problem, however, for more intensive studies of discourse structure. 251 §5.3 Conclusion The function of wa has been one of the most studied topics in Japanese t radi t ional grammar. It has been a big issue because it concerns not just problems of syntax or discourse but also problems involving various l inguistic components at the same time. In this sense, the problem of wa is one of the most interest ing cases for the applicat ion of the modular approach proposed in recent generative theories. I began my investigation of this topic wi th intensive studies of Kuroda's conceptual proposal of adopting the categorical and thetic dis t inct ion into Japanese sentences. According to Kuroda, categorical judgement happens to be morpho log ica l ly / syn tac t i ca l ly marked in Japanese because of the part ic le wa. As we l l , the wa- sentence as discussed in this chapter has a long history of discourse study, often relating to the commonly accepted views of wa as a topic marker or old information marker. The f i r s t several sections are devoted to just i fying the view of wa as the judgement marker, rather than the top ic /o ld information marker. Then, Kuno's c lass i f ica t ion of topic, contrastive vs. thematic, is examined to clar i fy the notion of contrastive as i t applies to wa-sentences. What we feel as contrastive in terms of wa actually has two sources: the inherent nature of N P ( - w a ) and contextual paral lel ism. The former is related to the semantic property of subject NP of categorical judgement. In the latter, the nature of wa contrast is 252 closely linked to i ts characteristic statement-contrast instead of i tem-contrast. The contrastive reading reported in the li terature, which is sometimes expressed in terms of graduation, is the result of the interaction between the property of wa and the contextual environment. By means of this analysis, the difference between contrastive wa and thematic wa is dissolved, and they are no longer dist inctive categories. Many of the topics discussed in each section above are not new, but rather c lass ica l ; but the purpose of my research here has been to propose an integrated framework for the nature of wa. Some of the seemingly conflicting analyses proposed in the past have, I believe, been reconciled by being reworked with different components in the over-all view of wa which is proposed here. 253 Notes (Chapter 5) 1 Base generation is a possible alternative. 2 (7a) and (7b) can have the "eel" reading; e.g., 'I am going to meet Yamada' for (7a) and 'I play the role of a student' for (7b). However, these are irrelevant for the present purpose. 3 We use the terms 'copula' and 'verb, to-be' interchangably. 4 "Subject in the logical sense" does not mean "logical subject" in generative grammar. 5 The truth conditional difference between the subject-predicate structure and the subjectless structure is argued in Kuroda (1969 & 1970). 6 "Judgement" is used as a term in logic, not in psychology, (cf. Kuroda, 1972, note 2.) 7 This term was or iginal ly used by members of the Prague School although Halliday of the London School contributed largely to the development of the concept. 'Theme' is the leftmost element of a sentence, which serves as "the s tar t ing point of the utterance," whereas 'rheme' is "what the speaker states about" the theme (cf. Brown and Yule (1983)). 8 The referents are constantly changing their roles in context: the same referent can be old or new, depending on context. Also look at (1) ( = Reinhart (37)): (i) A: Who did Felix praise? B: Felix praised HIMSELF. According to Reinhart, himself is new (focus expression), but it is also old, since it is the same person as Felix (topic expression.) Thus himself is old and new at the same t ime, which is a 254 cont radic t ion . However, the old-new d i s t i n c t i o n might not necessarily be made in terms of referents. That is, what is new is the information that [Felix praised X and X is someone] (the someone happening to be Felix himself in this case; also cf. §4.2.2 in Chapter 4). Hoji notes that the 5" is the same as the node E(xpression) of Banfield (1973). The proposal of the node E is also adopted in van Riemidjik and Zwartz (1974) for l e f t -d i s loca t ion as opposed to topicalization (reported in Reinhart, 1983). Hoji also points out that certain phrases l ike a P P - w a can be contrastive without stress. Rochemont (1 986) classif ies 'focus' into two categories: contrastive focus and presentational focus. While ga in a thetic sentence is related to the latter, non-deletable ga marks the contrastive focus only. In Terakura's analysis , these are not considered as deletion but simply non-wa sentences. Thus, once the sentences get wa, they are different sentences from Terakura's. Cf. note 10 in Terakura (1 984). In fact, (36) is ambiguous between the thematic and the contrastive, but i t is not a relevant issue. M. Soga has pointed out that (39a) is not so bad, especially when the order is reversed, l ike "tuki ga dete iru ga ame ga hutte kita." However, (i)c is very bad: (i)a. Taroo ga tatte-iru. Nom be-standing 'Taroo is standing' 255 b. Ziroo ga suwatte-iru. Nom be-sit t ing 'Ziroo is sitting. ' c * Taroo ga tat te- iru ga, Ziroo ga suwatte-iru. 'Taroo is standing but Ziroo is sitting. ' The reason why (39a) is felt s l ight ly better than (i)c probably has something to do wi th the fact that (39a) involves a cause and effect relation, i.e., rain fall ing is a counter-expectation of moon shining. The contrastive effect is produced pragmatically (by our knowledge of the world), as wel l . Look at the sentences in (i) and (ii): (i) Gonbe ga tane (o) makya ( = make ba), karasu ga/wa Nom seed Acc sow Conditional crow hojikuru. dig up 'When Gonbe sows seeds, crows dig them up.' (ii) Taroo ga/wa Tookyoo e itte, Ziroo ga/wa Oosaka itta. to go (Ger.) to go 'Taroo went to Tokyo and Ziroo went to Osaka.' Ba in (i) is a conditional marker, and the -te form ( = a gerund form) in (ii) is used to enumerate events. In these cases which can but do not necessarily have wa, two ideas contrast pragmatically. For the dis t r ibut ion of passive sentences in discourse, cf. Givon (1979) and Oosten (1984). There are also some arguments against such discourse constitutions in terms of passive. Haiman (1978) argues for the c lose r e l a t ionsh ip between conditionals and topics. 256 1 9 Dareka can be a specifiable someone, but this is not a relevant issue here. 2 0 Gil (1987) argues that the reason why some languages have the ( in)defini te a r t i c l e whi le some do not is cont ro l led by two independent parameters in the universal grammar: configurationality and the count-mass distinction. 2 1 M. Rochemont pointed out to me that sm and some are different l ex ica l i tems rather than being ambiguous. However, i t does undermine Milsark's contention over "weak" vs. "strong." 2 2 Kuroda (1969) also points out, without going into detai ls , that indefinite NPs can be used like "variables." 2 3 Contrary to wa, the particle mo is used in s imi la r environments to (62'), [X, - Y,] and [X 2 - Yi]: (i) Taroo ga hon o katte, Ziroo mo hon o katta. 'Taro bought a book and Ziro bought one, too.' 2 4 It has been observed that wa cannot appear in subordinate clauses. But an exception to this is the contrastive wa as shown in (i): friend is in contrast with someone else who did not buy a car: (i) [[ stomodati wa katta] kurumaW friend bought car 'The car which my friend bought' According to Kuroda (1972), subordinate structures do not represent judgement, but are "judgement materials without judgement forms." 2 5 The interaction between the negative and wa creates an interesting effect: 257 (i) Sanzyuunin wa konakatta. thirty people did-not-come 'Thirty people did not come.' The sentence can mean thirty people did not come although twenty-nine (twenty-eight, twenty-seven, e t c . . ) people did come. This reading involves the underlying expectation in the utterance. That is, the sentence implies that "thirty" is what the speaker had in his mind, but the actual number of people who came was less than that. An affirmative sentence like (ii) can also be interpreted in a s imi lar manner: (ii) Hyakunin wa kita. 'A hundred people came.' The sentence means that as many as a hundred people came. I consider that the type of contrastive implication involved in (i) is of a different nature than what I discuss in this section, although it originates in the function of wa. The part icle mo also exhibits a s i m i l a r (but differently manifested) type of problem: cf. Soga (1976). Two types of ga, thetic ga and non-thetic (exhaustive) ga can be viewed as focus markers for non-contrastive (presentational) focus and contrastive focus (Rochemont's (1986) terms) respectively. Kuno (1973) makes some in teres t ing observations about the s imi la r i ty between exhaustive-list ing ga in Japanese and stress and cleft constructions in English (cf. "Wa and Ga (Part I)" in Kuno (1973)). Also cf. Wi l l i ams (1983), where two types of pseudo-cleft are differentiated in his predication theory. If the predicate is stative, ga is the exhaustive-l ist ing except wi th numeral/quantifier NP like (i) (Kuno, 1973): 258 ( i ) T a k u s a n no h i t o ga b e z i t a r i a n da. m a n y p e o p l e v e g e t a r i a n 'Many p e o p l e a r e v e g e t a r i a n s . ' T h e f o l l o w i n g s e n t e n c e s a r e p r e f e r r e d t o ( 7 9 b ) p r o b a b l y b e c a u s e of t h e "-ga -ga " s e q u e n c e w h i c h i s not m u l t i - n o m i n a t i v e : (<) ( i ) a . U n a g i o T a r o o ga t y u u m o n s i t a . b. T a r o o ga u n a g i 6 t y u u m o n s i t a . p 259 Chapter 6 Concluding Remarks A sentence uttered in an actual speech situation is not free from context. Thus, the question is how formal grammar and ex t ra -grammatical factors interact for sentences to receive full compre-hension in actual discourse. With the purpose of giving answers to this question, four l inguist ic phenomena were examined in this thesis. The f i r s t two chapters concern the coreference problems of pronouns: reflexive pronouns in Chapter 2 and demonstrative pronouns in Chapter 3. The former has been studied mostly in syntax, often wi th analogies to the English reflexive pronoun; the latter has been studied from the viewpoint of pragmatics. In Chapter 4, quantifier reading was examined. A l l three topics are related to disambiguation mechanisms in discourse. The topic of Chapter 5 is of a sl ightly different nature, not a disambiguation system, but the so-called "topic" marker wa, one of the most interesting cases which has overt manifestations at each level of representation: semantics, syntax, and pragmatics. 260 while the LF' module of grammar w i l l identify the entity of the null element in the topic position. The final determination procedure among the el igible antecedents is complex and s t i l l not very clear; yet it is highly possible that the pragmatic factors like "awareness'V'viewpoint" control the final determination of the referent, where the structurally defined concept "subject" enters as one strong candidate for the site of the manifestation of such pragmatic conceptions. In Chapter 3, unlike zibun, demonstratives are shown to be always referential wi th the exception of the problematic cases created by those of the so- series, some of which exhibit a strong anaphoric nature or at least a possible disappearance of the dist inct ion between pronouns ( = proforms which stand for referents) and demonstratives ( = demonstration). The latter case may be handled in the formal representation of discourse structure (LF' module). In Chapter 4, quantifier reading was examined. The LF representation provides the wide/narrow reading for a quantif ier sentence. The final interpretation is determined by the interaction between the subjecthood, the sentence in i t ia l position which serves an important discourse function, and the "egocentricity." The egocentricity has turned out to have a wide influence on l inguis t ic phenomena in Japanese (which is often cal led an epistemological language), for example, in zibun in Chapter 2 and in the demonstrative ko-so-a dis t inc t ion in Chapter 4. In Chapter 5, adopting Kuroda's thetic vs. ca tegor ica l d i s t i nc t i on , i t is argued that a semantic s t ructure corresponds to a syntactic structure through wa as the marker of the ca tegor ica l judgement; and that syn tac t i c s t ruc ture , in turn, 261 corresponds to discourse structure, i.e., the function of the sentence topic marked by wa seems to have a significant influence on discourse organization. N P - w a has an anaphoric nature to a certain extent; however, i ts manner of reference is totally different from pronoun type anaphora, functioning rather as a "nexus" in the flow of discourse (cf. Grice's conversational maxim, "Be relevant."). The absence of NP-wa, as a consequence, indicates that the topic NP established in the previous sentence is s t i l l in effect, and this local discourse is about the same topic. The mode of interaction assumed from these analyses is that wi th a form as a medium, syntax, semantics, and pragmatics interface on it; a certain l inguist ic phenomenon has a function on each l inguist ic level (though i t might have a null function on some level(s) of representation). The autonomous system prevents two-d i rec t iona l interactions between two different levels of representation. In the case of zibun, for example, the ambiguity represented in syntax is eventually disambiguated outside of the syntact ic module. The case wi th the quantifier is apparently a l i t t l e different from the case of zibun, but not to a substantial degree; here the LF representation of the scope reading provides the ambiguity for certain cases l ike scrambling, and such readings are reinterpreted outside of syntax in a systematically related manner to the readings in LF. 262 B i b 1 i o g r a p h y Akatsuka, Noriko (1983) "Conditionals," Papers in Japanese Linguistics, Vol. 6, pp. 1-33. (1985) "Conditionals and the Epistemic Scale," Language, Vol. 61, No. 3, pp. 625-639. Akmajian, Adrian and Chisato Kitagawa (1976), "Deep-structure Binding of Pronouns and Anaphoric Bleeding," in Language, Vol. 52, No. 1, pp. 61-77. Aller ton, D. J. and A. Cruttenden (1979) "Three reasons for accenting a definite subject," Linguistics 15, pp. 49-53, Cambridge University Press, Great Britain. Al lwood , Jens (1981) "On the Dis t inct ions between Semantics and Pragmatics," in Wolfgang Kle in and W i l l i a m Levelf (eds.), Crossing the Boundaries in Linguistics, D. Reidel, Dordrecht. Anderson, M. (1979) Noun Phrase Structure, doctoral d isser ta t ion, University of Connecticut. Banfield, A. (1973) "Grammar of Quotation, Free Indirect Style, and Implications for a Theory of Narrative", Foundation of Language, vol. 10, pp. 1-39. Bedell, George (1972) "On No, " U. C. L. A. Papers in Syntax: Studies in East Asian Syntax, Vol. 3, pp. 1-1 9. B e l l e t t i , Adriana and Luigi R izz i (1988) "Psych-Verbs and e-theory", Natural Language and Linguistics Theory 6, pp. 291-352. 263 Breivik, Leiv Egil (1981) "On the Interpretation of Existential THERE," Language 57, pp. 1 -25. (1983) "On the Use and Non-use of Existential THERE, a review of Gary L. Milsark, Existential sentences in English, " Lingua, pp. 333-368. Brockway, Diane (1 981) "Semantic Constraints on Relevance, " in H. Parret, M. Sbisa, and J . Varschueren (eds.), Possibilities and Limitations of Pragmatics: Proceedings of the Conference on Pragmatics, Uribino, Italy; Benjamins, Amsterdam. Brown, Gi l l ian , and George Yule (1983) Discourse Analysis, Cambridge University Press, Cambridge. Bullwinkle, Candace (1977) "Levels of Complexity in Discourse for Anaphora Disambiguation and Speech Act Interpretation," Proceedings of the Fifth Joint Conference on Artificial Intelligence, pp. 43-49. Carlson, Greg N. (1977) "A Unified Analysis of the English Bare Plural," Linguistics and Philosophy 1, pp. 41 3-457. Chafe, Wallace L. (1976) "Givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View," in Charles N. Li (ed.), Subject and Topic, Academic Press, New York. Chang, Sandra and James McCloskey (1987) "Government, Barriers, and Small Clauses in Modern Irish", Linguistic Inquiry, vol. 18-2, pp. 173-237. Chomsky, Noam (1971) "Deep structure, surface structure and semantic interpretation," in D. D. Steinberg and L. A. Jakobvits (eds.), Semantics, University Press, Cambridge. (1974) "Questions of Form and Interpretation," Linguistic Analysis, Vol 1, No. 1, 1 975, pp. 75-1 09. 264 (1977) Essays on Form and Interpretation, Studies in Linguistic Analysis, North-Holland, Amsterdam. (1981) Lectures on Government and Binding, Foris Publications, Dordrecht. (1982) Some Concepts and Consequences of the Theory of Government and Binding, (Linguistic Inquiry Monographs 6), MIT Press, Cambridge, Mass. (1986) Barriers, MIT Press, Cambridge, Mass. Contreras, Heles (1986) "What does Everybody Want?" in M. Brame, H. Contreras and F. Newmeyer (eds.) A Festshrift for Sol Saprta, Noit Amrofer, pp. 55-64. Creider, Chet A. (1979) "On the explanation of Transformations," in Talmy Givon (ed.), Syntax and Semantics Vol 1 2, Discourse and Syntax, Academic Press, Inc., New York. Di jk, Teun A. (1 981) "Sentence Topic and Discourse Topic," in Studies in the Pragmatics of Discourse, pp. 177-193. Downing, Pamela (1 985) "A Discourse-Based Explanation for Quantifier Float Constraints in Japanese," ms., Linguistic Society of America, Seattle, December. Dretske, Fred I. (1972) "Contrastive Statements," The Philosophical Review, Vol. LXXXI, Cornell University, Ithaca, New York. Ehlich, Konrad (1982) "Anaphora and Deixis: Same, Similar , or Different?" in R. J. Jarvella and W. Klein (eds.), Speech, Place, and Action, pp. 73-99, John Wiley and Sons Ltd. 265 Enomoto, Yoshio (1 983) Relevance of Focus and Ground in Semantic Analysis, doctoral dissertation, State University of New York at Buffalo. Erteschik-shir , Nomi (1986) "WH-Questions and Focus," Linguistics and Philosophy, Vol. 9, No. 2, pp. 1 17-149, D. Reidel Publishing, Dordrecht. Evans, Gareth (1980) "Pronouns," Linguistic Inquiry 1 1.2. Facey, Kikuko (1983) "Japanese reflexivization and psychological semantics," Papers in Japanese Linguistics, Vol. 6, pp. 35-42. Fillmore, Charles J. (1971) "Santa Cruz Lectures on Deixis 1 971," Indiana University Linguistics Club (1975), Bloomington, Indiana. Firbas, Jan (1966) "Non-thematic Subjects in Contemporary English," Travaux Linguistiques de Prague 2, Academia, Editions de l'Academie Tchecoslovaque des Sciences (Prague), pp. 239-256. Fukuchi, Hajime (1985) Danwa no kozo, Taishukan, Tokyo. Gazdar, Gerald (1979) Pragmatics: Implication, Presupposition, and Logical Form, Academic Press, New York. Gi l , David (1987) "Definiteness, Noun Phrase Configurationality, and the Count-Mass Distinction," Chapter 10 in Eric J. Reuland and Al ice G. TerMeulen (eds.), The Representation of (In)definiteness, MIT Press, Cambridge. Givon, Talmy (1 979) On Understanding Grammar. Academic Press, New York. (ed.) (1983) Topic Continuity in Discourse: AQuantitive Cross-Language Study, Benjamins, Amsterdam. 266 Grice, H. Paul (1975) "Logic and Conversation," in P. Cole and J. L. Morgan (eds.), Syntax and Semantics J . Speech Acts, Academic Press, New York. Grosz, Barbara J. (1977) The Presentation and Use of Focus in Dialogue Understanding. Stanford Research Institute Technical Note 151, SRI Project 5844, A r t i f i c i a l Intelligence Center. (1978) "Focusing in Dialog," in TINLAP-2, University of Il l inois, pp. 96-103. (1978) "Focusing in Dialogue," The Technical Issues in Natural Language Processing, Vol. 2, University of Illinois. (1981) "Focusing and description in natural language dialogues," in Bennie L. Webber and Ivan A. Say (eds.), Elements of Discourse Understanding, Cambridge University Press, Cambridge. Gundel, Jeanette K. (1985) '"Shared Knowledge' and Topicality," Journal of Pragmatics, Vol. 9, pp. 83-1 07. Gunji,Takao (1983) "Generalized Phrase-Structure Grammar and Japanese Reflexivization," Linguistics and Philosophy, Vol. 6, No. 1, pp. 1 15-156. Haig, John H. (1979) "Real and Apparent Multiple Subject Sentences", Papers in Japanese Linguistics, vol. 6, pp. 87-125. (1982) "Prenominal Modification and the Distribution of NO in Japanese," ms., University of Hawaii at Manoa. (1983) "The Structure of Japanese Clefts," ms., Linguistic Society of America, Washington D. C , December. Hai'k, Isabelle (1984) "Indirect Binding", Linguistic Inquiry, vol. 15-2, pp. 185-39. 267 Haiman, John (1978) "Conditionals are Topics," Language, Vol. 54, pp. 564-589. Halliday, M. A. K. (1967) "Notes on t ransi t ivi ty and theme in English, Part 1," Journal of Linguistics, Vol. 3, No. 1, pp. 37-81. (1967) "Notes on t ransi t ivi ty and theme in English, Part 2," Journal of Linguistics, Vol. 3, No. 2, pp. 199-244. (1968) "Notes on t ransi t ivi ty and theme in English, Part 3," Journal of Linguistics, Vol. 4, pp. 153-308. Hasegawa, Nobuko (1980) "Three Reasons for Not Deriving Japanese Reflexives from Full NP's," Papers in Japanese Linguistics 7. (1981) "The VP Complement and'Control'Phenomena: Beyond Trace Theory," L inguistic Analysis 7.1. Hattori, Shiro (1 968) "KORE, SORE, ARE to this, that," in Eigo Kiso Goei no Kenkyu, Sanseido, Tokyo. Heim, Irene (1982) The Semantics of Definite and Indefinite Noun Phrase, doctoral dissertation, University of Massachusetts. Hendrix, Gary G. (1975) "Expanding the Ut i l i ty of Semantic Networks Through Partitioning," Advance Papers, International Joint Conference on Artificial Intelligence, Tbilisi, Georgian SSR, September. Higgins, Roger (1 973) On the Nature of Pseudoclefts, doctoral dissertation, MIT. Hirakoji , Kenji (1973) "Jibun Forms in Japanese," Papers in Japanese Linguistics 94, Vol. 2:2, winter, pp. 17-43. Hockett, Charles F. (1958) A Course in Modern Linguistics, New York: Macmillian. 268 Hoji, Hajime (1982) "The X-schema in Japanese and the * Parameter," ms, University of Washington/MIT. (1985) Logical Form Constraints and Configurational Structures in Japanese, doctoral dissertation, University of Washington, Seattle, Washington. (1986) "Scope Interpretation in Japanese and i ts Theoretical Implications", The Proceedings of WCCFL V. Hopper, P. J. (1 979) "Aspect and Foregrounding in Discourse," in Talmy Givon (ed.), Syntax and Semantics, Volume 1 2, Discourse and Syntax, Academic Press Inc., New York. Horn, Laurence R. (1 981) "Exhaustiveness and the Semantics of Clefts," in NELS 1 1, pp. 125-142. Horvath, Ju l ia and Michael Rochemont (1986) "Pronoun in Discourse and Sentence Grammar," Linguistic Inquiry, vol. 17-4. Huang, C.-T. James (1982) Logical Relations in Chinese and the Theory of Grammar, doctoral dissertation, MIT. (1984) "On the Distribution and Reference of Empty Pronouns", Linguistic Inquiry, vol. 15-4, pp. 531-44. Ide, Itaru (1958) "Daimeishi," in Bunpo Kakuronhen, Meijishoin, Tokyo. Inoue, Kazuko (1976) Henkeibunpo to Nihongo: Imikaishaku o chushin ni, Taishukan, Tokyo. (1976) "Reflexivization: An Interpretative Approach," in M. Shibatani, Syntax and Semantics 5; Academic Press, New York. (1982) "An Interface of Syntax, Semantics, and Discourse Structure," Lingua vol. 57, pp. 259-300. 269 (1 983) "'wa' to 'ga'," in K. Inoue (ed.), Nihongo no kihonkoza, pp. 45-56, Sanshodo, Tokyo. loup, Georgette (1 973) "Some Universals for Quantifier Scope," in Ralph W. Fasold and Roger I say (eds.), Analyzing Variation in Language. Ishigami, Teruo (1983), "Bun niokeru Hitei to Uso - Gengoshutai no Nodoteki-sokumen - , " Nihongogaku, vol. 2, Tokyo. Jackendoff, R. (1972) Semantic Interpretation in Generative Grammar, MIT Press, Cambridge. Josephs, Lewis S. (1975), "Complementation," Syntax and Semantics, Vol. 5, Academic Press, New York. Kamio, Akio (1979) "On the Notion Speaker's Territory of Information: a functional analysis of certain sentence-final forms in Japanese," in E. Bedell, E. Kobayashi, and M. Muraki (ed.), Explorations in Linguistics (Papers in Honor of Kazuko Inoue), Society of Linguistics and Literature, Tokyo. (1985) "Danwa niokeru Shiten," Nihongogaku, December, Vol. 14, pp. 10-21. Kamp, Hans (1982) "A Theory of Truth and Semantic Representation," in J. A. G. Groenendijk, T. M. V. Janssen, and M. B. J. Stoklof (eds.) Formal Methods in the Study of Language, Part I. Kaplan, R. (1966) "Cultural thought patterns in inter-cultural education," Language Learning, 16, pp. 1-20. Karttunen, L. (1976) "Discourse Referents," in J . D. McCawley (ed.), Syntax and Semantics, Volume 7, Notes from the Linguistic Underground, Academic Press Inc., New York. 270 Keenan, Edward L. (1974) "The Functional Principle: Generalizing the Notion of 'SUBJECT OF'," in Papers from the Tenth Regional Meeting, Chicago Linguistic Society. (1976) "Towards a Universal Definition of'Subject'," in Charles Li (ed.), Subject and Topic, Academic Press, New York. Keenan, Elinor Ochs (1976) "Foregrounding Referents: A Recon-sideration of Left Dislocation in Discourse," in Proceedings of the Second Annual Meeting, Berkeley Linguistic Society. Kempson, Ruth M. (1975) Presupposition and the Delimitation of Semantics, Cambridge University Press, Cambridge. (1977) Semantic Theory, Cambridge University Press, Cambridge. (1984) "Anaphora, the Compositionality Requirement, and the Semantics-Pragmatics Distinction," NELS, vol. 14, pp. 183-206. Kempson, Ruth M. and Annabel Cormack (1981) "Quantification and Pragmatics," Linguistics and Philosophy vol. 4, pp. 1 17-189. Kiss , Kataline (1981) "On the Japanese Double Subject Construction," The Linguistic Review I, Vol. 1 No. 2. Kitagawa, Chisato (1977) "NO -introduced Relative Clauses," Journal of the Association of Teachers of Japanese, Vol. XII, No. 2 + 3, pp. 229-247. (1978) "A Note on Sore and Are, " in E. Bedell, E. Kobayashi, and M. Muraki (ed.), Explorations in Linguistics (Papers in Honor of Kazuko Inoue), Society of Linguistics and Literature, Tokyo. (1981) "Anaphora in Japanese: Kare and Zibun," in Ann Kathleen Farmer and Chisato Kitagawa (eds), Proceedings of the Arizona Conference on Japanese Linguistics: The Formal 271 Grammar Sessions, Coyote Working Papers in Linguistics from A -> Z 2, University of Arizona, Tucson, Arizona, pp. 61-75. (1982) "Topic Constructions in Japanese," Lingua 57, pp. 175-214. Kitagawa, Chisato and Claudia Ross (1982) "Prenominal Modification in Chinese and Japanese", Linguistic Analysis, vol. 9-1. Kitahara, Yasuo (1981) Nihongo no Bunpo : Nihongo no Sekai 6, Susumu Ono and Saiichi Maruya (eds.), Tyuokoron-sha, Tokyo. Koopman, H. and D. Sportiche (1982) "Variables and the Bi jec t ion Principle", The Linguistic Review 2.2. Kuno, Susumu (1972) "Functional Sentence Perspective: A Case Study from Japanese and English," Linguistic Inquiry, Vol. 3, pp. 269-312. (1972) "The Position of Locatives in Existential Sentences", Linguistics Inquiry, vol. 2, pp. 333-378. (1973) Nihonbunpo kenkyu, Taishukan, Tokyo. (1 973a) The Structure of the Japanese Language. MIT Press, Cambridge. (1973b) Nihon-bunpo Kenkyu, Taishukan Press, Tokyo. (1975) "Three Perspectives in the Functional Approach to Syntax," in Papers from the Parasession on Functionalism, Chicago Linguistic Society. (1976) "Subject, Theme, and the Speaker's Empathy," in Charles Li (ed.), Subject and Topic, Academic Press, New York. and Etsuko Kaburaki (1977) "Empathy and Syntax," Linguistic Inquiry, Vol. 8, No. 4, pp. 627-672. 272 (1978) Danwa no Bunpd, TaishGkan Press, Tokyo. (1979) "On the Intersection between Syntactic Rules and Discourse Principles," in E. Bedell, E. Kobayashi, and M. Muraki (ed.), Explorations in Linguistics (Papers in Honor of Kazuko Inoue), Society of Linguistics and Literature, Tokyo. (1980) "The Scope of the Question and Negation in Some Verb-Final Languages," in Papers in Sixteenth Regional Meeting of Chicago Linguistics Society, edited by Jody Kreemer and Almerindo E. Ojeda, pp. 155-169. (1982) "Principles of Discourse Deletion," colloquium delivered in the University of Br i t i sh Columbia Linguistics Department. (1 982) "The Focus of the Question and the Focus of the Answer," in Papers from the Parasession on Non-declaratives, Chicago Linguistic Society, pp. 134-155. (1983) Sin nihon-bunpo Kenkyu Taisyukan, Tokyo. Kuroda, Shige-Yuki (1965) Generative Grammatical Studies in the Japanese Language, doctoral dissertation, M. I. T. (Published in 1979, Outstanding Dissertations in Linguistics, A Garland Series, Garland Publishing, Inc., New York and London). (1 969) "Remarks on the notion of SUBJECT with reference to words like ALSO, EVEN, orONLY, i l lustrat ing certain manners in which formal systems are employed as auxiliary devices in l inguist ic descriptions, Part 1," in the Annual Bulletin No. 3, the Research Institute of Logopedics and Phoniatrics, pp. 111-130. (1970) "Remarks on the Notion Subject wi th Reference to Words Like ALSO, EVEN, or ONLY, Part I", in Annual Bulletin, vol. 273 4, pp. 127-152, Logopedics and Phoniatrics Research Institute, Tokyo University. (1972) "The Categorical and the Thetic Judgement," Foundations of Language 9, pp. 153-185. (1 973a) "On Kuno's Direct Discourse Analysis of the Japanese Reflexive Zibun," Papers in Japanese Linguistics 2.1, pp. 1 36-147. (1973b) "Where Epistemology, Style, and Grammar Meet: A Case Study from Japanese," in Stephen R. Anderson and Paul Kiparsky (eds.), A Festschrift for Morris Halle, Holt, Reinhart, and Winston, New York, pp. 377-391. (1974) "Pivot-Independent Relativization in Japanese (I)," Papers in Japanese Linguistics, Vol. 3, pp. 59-93. (1 975-76), "Pivot-Independent Relativization in Japanese (I I)," Papers in Japanese Linguistics, Vol. 4, pp. 85-96. (1 976) "Headless Relative Clauses in Modern Japanese and the Relavancy Condition," Berkeley Linguistics Society, vol. 2. (1 976-77) "Pivot-Independent Relativization in Japanese (III), Interlude: Types of Japanese relatives," Papers in Japanese Linguistics, Vol. 5, (1 976-77), pp. 157-179. (1979) "(Ko) • So • A nitsuite," in Eigo to Nihongo, Kuroshio Shuppan, Tokyo. (1 979) The 'WHOLE of the Doughnut, Syntax and its Boundaries, E. Story-Scientia, Ghent. (1 985) "The Movement Theory of Wa Topics in Japanese", LSA meeting, Seattle. 274 (1986) "Movement of Noun Phrases in Japanese" in Takashi Imai and Mamoru Saito (eds.), Issues in Japanese Linguistics, Foris, pp. 229-272. (1988) "Whether We Agree or Not: A Comparative Syntax of English and Japanese", in Wil l iam Poster (ed.) Papers from the Second International Workshop on Japanese Syntax, CSLI, pp. 103-144. Ladusaw, Wil l iam A. (1982) "Semantic Constraints on the English Part i t ive Construction," in Daniel P. Flickinger, Marlys Macken, Nancy Wiegand (eds.), Proceedings of the First West Coast Conference on Formal Linguistics. Lakoff, Robin (1974) "Remark on T/i/'s and That," Papers from the Tenth Regional Meeting of the Chicago Linguistic Society. (1974) "Remarks on this and that, " Chicago Linguistic Society, Vol. 10. Lappin, Shalom (1984) "Pronominal Binding," ms., University of Ottawa, Ottawa, Canada. Lasnik, Howard (1976) "Remarks on Coreference," Linguistic Analysis ' 2.1. Levinson, Stephen C. (1983) Pragmatics, Cambridge University Press, Cambridge. L i , Charles and Sandra A. Thompson (1 976), "SUBJECT and TOPIC: A New Typology of Language," in Charles Li (ed.), Subject and Topic, Academic Press, New York. Linde, Charlotte (1979) "Focus of Attention and the Choice of Pronouns in Discourse," in Syntax and Semantics, Vol. 12, Discourse and Syntax, Academic Press, Inc., New York. 275 Linde, Charlotte and Will iam Labov (1975) "Spatial Networks as a Site for the Study of Language and Thought," Language, Vol. 51 No. 4, pp. 924-939. Lyons, John (1977) Semantics, Vol. 2, Cambridge University Press. Maclaran, Rose (1982) The Semantics and Pragmatics of the English Demonstratives, doctoral dissertation, Cornell University. Makino, Seiichi (1982) "Japanese Grammar and Functional Grammatical," Lingua 57, pp. 125-173. (1984) "Speaker/Listener Oreintation and Formality Marking in Japanese," Gengo Kenkyu, pp. 126-145. M. Rita Manzini and Kenneth Wexler (1987) "Parameters, Binding Theory, and Learnability", Linguistic Inquiry, vol. 18-3, pp. 413-444. Masunaga, Kiyoko (1982) "Bridging," in Proceedings of the Xlllth International Congress of Linguistics, Aug. 29th - Sep. 4th, 1982, Tokyo. Editors Shiro Hattori, Kazuko Inoue; The Linguistic Society of Japan, Taishukan, Tokyo. Materna, Pavel, Eva Hajicova, and Peter Sgall (1987), "Redundant Answers and Topic-Focus Articulation," Linguistics and Philosophy, Vol. 10-1, pp. 101-113. Matsuda, Hiroshi (1975) A Transformational Study of Japanese Reflexivization, M. A. thesis, The University of Br i t i sh Columbia. May, Robert (1977) The Grammar of Quantification, doctoral d i s -sertation, MIT. (1985) Logical Form: Its Structure and Derivation, MIT Press, Cambridge, Mass. 276 Maynard, Senko Kumiya (1981) "The Given/New Distinction and the Analysis of the Japanese Particles -WA and -GA," Aspects in Linguistics: International Journal of Human Communications, Vol. 14, pp. 109-132. McCawley, J. D. (1979) "Presupposition and Discourse Structure," in C -K. Oh and D. A. Dinneen (eds.), Syntax and Semantics, Volume 11, Presuppostion, Academic Press Inc., New York. McCawley, Noriko Akatsuka (1976) "Reflexivization: A Transformational Approach," in M. Shibatani, Syntax and Semantics 5, Academic Press, New York. (1978) "Another look at no, koto, and to : Epistemology and Complementizer Choice in Japanese," in J. Hinds and I. Howard (eds.), Problems in Japanese Syntax and Semantics, Kaitakusha, Tokyo. McGloin, Naomi H. (1976) "Negation," Syntax and Semantics 5, Academic Press, New York. Mikami, Akira (1 955) Gendai Goho Josetsu, Tokoshoin, Tokyo. (1960) Zoo wa Hana ga Nagai, Kuroshio-shuppan, Japan. (1972) Gendai goho shinsetsu, Kuroshio-shuppan, Tokyo. Milsark, Gary L. (1977) "Towards an Explanation of Certain Peculiari t ies of the Existential Construction in English," Linguistic Analysis, Vol. 3 No. 1, pp. 1-30. Miyaj i , Yutaka (1964) "Daimeishi," in M. Tokieda and Y. Endo (eds.), Koza Gendaigo, Volume 6, Kogobunpo no Mondaiten, Meijishoin, Tokyo. Morgan, J. L. (1975) "Some Interactions of Syntax and Pragmatics," in Syntax and Semantics 3, Academic Press, New York. 277 Murata, Yuzaburo (1982) Kind-eibunpd, Taishukan, Tokyo. Nakajima, Heizo (1 985) "Module Bunpo no Tenkai (Chd)," Gengo vol. 4, no. 12, pp. 88-96. (1 985) "Module Bunpo no Tenkai (Jo)," Gengo vol. 4, no. 11. (1986) "Module Bunpo no Tenkai (Ge)," Gengo vol. 5 no. 1. Nakamura, Masuru (1989) "Reflexives in Japanese", Gengo Kenkyu 95, Japan, pp. 206-230. Newmeyer, F. J. (1983) Grammatical Theory: Its Limits and Its Possibilities, The University of Chicago Press, Chicago and London. Nishio, Hiroko (1983) "Passives and Causatives in Japanese and Elsewhere, " Papers in East Asian Languages, vol. 1, Hawaii. Oehrle, Richard T. (1981) "Common Problems in the Theory of Anaphora and the Theory of Discourse," in H. Parret, M. Sbisa, and J. Varschueren (eds.), Possibilities and Limitations of Pragmatics: Proceedings of the Conference on Pragmatics, Uribino, Italy; Benjamins, Amsterdam, pp. 509-530. Ogihara, Toshiyuki (1986) "The Semantics of the'Obligatory Focus' Construction in Japanese," ms. (To appear in Texas Linguistic Forum.) Ohso, Mieko (1979) "A Constraint on the Antecedent of Zero Pronominalization in Japanese," in E. Bedell, E. Kobayashi, and M. Muraki (ed.), Explorations in Linguistics (Papers in Honor of Kazuko Inoue), Society of Linguistics and Literature, Tokyo. Okamura, K. (1972) "Daimeishi towa Nani-ka," in K. Suzuki and 0. Hayashi (eds.), Hinshibetsu Nihongo Bunpo Koza: Meishi, Daimeishi, Meijishoin, Tokyo. 278 Okutsu, Keiichiro (1 978), " 'Boku wa Unagi da' no Bunpd - Da to No - , Kuroshioshuppan, Tokyo. (1981) "Unagibun wa Koko kara kita ka," Kokugo to kokubun-gaku, vol. 58-5, Tokyo, pp. 76-88. Ono, Mieko (1975) "Shijigo no Kino: shij isaki Kettei no Keishikika o Mezashite - ," M. A. thesis, Tokyo Women's Christian University, Tokyo. (1975) "Bunsho ni Tsukawareta Shi j i go," Tokyo Joshi Daigaku Nihon Bungaku, Vol. 48, 1975; reprinted in Kokugogaku Ronsetsu-shiryo, Vol. 13, No. 14, 1977, pp. 11-31. (1984) "Ref lexive Pronouns in Japanese," in Japanese Studies in Canada, edited by Bernard Saint-Jacques and Matsuo Soga, Canadian Asian Studies Association. Onoe, Keisuke (1981) '"Wa no keizyoshi-sei to Hyogen-teki-kino," Kokugo to kokubun-gaku, vol. 58, Japan, Tokyo Daigaku Kokugo-kokubun Gakkai, Tokyo, pp. 102-1 18. Oosten van, Jeanne (1984) "The Nature of Subjects, Topics, and Agents: A Cognitive Explanation," doctoral dissertation, University of California, Berkeley. (Published by Indiana University Linguistics Club, 1 986.) Oshima, Shin (1979) "Conditions on Rules: Anaphora in Japanese," in E. Bedell, E. Kobayashi, and M. Muraki, Explorations in Linguistics, Kenkyusha, Tokyo. Oyakawa, Takatsugu (1973) "Japanese Reflexivization I," Papers in Japanese Linguistics 94, Vol. 2:1, summer, pp. 94-134. (1974) "Japanese Ref lexivizatiom 11," Papers in Japanese Linguistics Vol. 3, pp. 129-199. 279 Partee, Barbara Hall (1972) "Opacity, Coreference, and Pronouns," in Harman and Davidson (eds.), Semantics of Natural Language, D. Reidel, Dordrecht. (1982) "Compositionality," in F. Landman and F. Valtman (eds.), Varities of Formal Semantics, Foris Publications, Dordrecht. (1985) "Ambiguous Pseudoclefts with Un-ambiguous BE, " NELS 16. Postal, Paul M. (1969) "On so-called 'Pronouns' in English," in David A. Reibel and Sanford A. Schane (eds.), Readings in Trans-formational Grammar, Prentice-Hall , New Jersey. Prince, Ellen F. (1978) "A Comparison of Wh-Clefts and If-Clefts in Discourse," Language, Vol 54, IV, pp. 883-905. (1979) "On the Given/New Distinction," in Papers from the Regional Meeting, vol 15, 1 979, Chicago Linguistic Society, pp. 267-78. (1984) "Topicalization and Left-Dislocation: A Functional Analysis," in Sheila J. White and Virginia Teller (eds.), Discourses in Reading and Linguistics, Annals of the New York Academy of Sciences, Vol. 433, pp. 213-225. (1 985) "Fancy Syntax and 'Shared Knowledge'," Journal of Pragmatics, Vol. 9, No. 1, pp. 65-81. Putman, Hilary (1958) "Formalization of the Concept'ABOUT'," Philosophy and Science, Vol. 25, No. 2, pp. 125-130. Reinhart, Tanya (1980) "Conditions for Text Coherence," Poetics Today vol. 1, no. 4, The Porter Institute for Poetics and Semiotics, Tel Aviv University, pp. 161-180. 280 (1982) "Pragmatics and Linguistics: An Analysis of Sentence Topics," Indiana University Linguistics Club, Bloomington, Indiana. (1983) Anaphora and Semantic Interpretation, Croom Helm, London & Canberra. Rochemont, Michael S. (1 978) A Theory of Stylistic Rules in English, doctoral dissertation, University of Massachusetts, Amherst. (1983) "Coindexing and Coreference," ms., UBC. (1 985) "Focus in Generative Grammar, " The Linguistic Review 4, pp. 365-373. (1986) Focus in Generative Grammar, John Benjamins Publishing Co., Amsterdam/Philadelphia. Rooth, Mats (1984) "How to Get Even wi th Domain Selection," NELS (North Eastern Linguistic Society) vol. 1 4, pp. 377-401. Saito, Mamoru (1 982a) "Case and Government in Japanese," ms., M. I. T., Massachusetts. (1982b) "Case Marking in Japanese, A Preliminary Study," ms., MIT. (1984) "Three Notes on Syntactic Movement in Japanese" ms. MIT. (1 985) Some Asymmetries in Japanese and Their Theoretical Implications, doctoral dissertation, M. I. T., Massachusetts. (1986) "Scrambling as Semantically Vacuous A'-movement", ms., NYU Workshop on Phrase Structure. 281 Saito, Mamoru and Hajime Hoji (1983) "Weak Crossover and Move a i n Japanese", Natural Language and Linguistic Theory 1, pp. 245-59. Sakuma, Kanae (1936; enlarged ed. 1951) Gendai-nihongo no Hyogen to Goho, Koseisha-koseikaku, Tokyo. Sgall , Peter (1975) "Conditions of the use of sentences and a semantic representation of topic and focus," in Formal Semantics of Natural Language, edited by Edward L. Keenan. Shibatani, Masayoshi (1978) "Mikami Akira and the Notion of "Subject" in Japanese Grammar" in John Hinds and Irwin Howard (eds.) Problems in Japanese Syntax, Kaitakusha, pp. 52-67. (1976) Japanese Generative Grammar: syntax and Semantics 5, Academic Press. Shibatani , Masayoshi and Chiseko Cotton (1976-77) "Remarks on Double-Nominative Sentences", Papers in Japanese Linguistics, vol. 5, pp. 261-277. (1978) "Mikami Akira and the Notion o f 'Sub jec t ' in Japanese Grammar," in John Hinds and Irwin Howard (eds.), Problems in Japanese Syntax and Semantics, Kaitakusha Press, Tokyo. Shirai , Ken-ichi (1981) "Remarks on Predicates taking Ga for Object Marking," Descriptive Applied Linguistics, XV, International Christian University, Tokyo. Smith, Donald L. (1979) "Where THIS Refers to HERE in English and Japanese, "in Explorations in Linguistics (in Honor of Kazuko Inoue), Society of Linguistics and Literature, Tokyo. Soga, Matsuo (1976) "The Structure of Negation and Delimitational Particles in Japanese," Journal of the Association of Teachers of Japanese, Vol. XI, No. 1, pp. 65-83. 282 Soga, Matsuo and Tai j i Fujimura (1979) "A Note on Noun Modification in Japanese," Journal of the Association of Teachers of Japanese, Vol. XIII, No. 1. Sportiche, Dominique (1986) "Zibun", Linguistic Inquiry 17, pp. 369-374. Stechow, Arnim von (1981) "Topic, Focus, and Local Relevance," in W. Klein and W. Levelt (eds.), Crossing the Boundaries in Linguistics, pp. 95-130, D. Reidel Publishing Co., Dordrecht. (1982) "Three Local Deictics," in R. J. Jarvel la and W. Klein (eds.), Speech, Place, and Action, pp. 73-99, John Wiley and Sons Ltd. Strawson, P. (1964) "Identifying Reference and Truth-Values", Theoria 30, pp. 90-1 18. Szabolcsi, Anna "The Semantics of Topic-Focus Articulation," in J. A.G. Groenendijk, T. M. V. Janssen, and M. B. J. Stokhof (eds.), Formal Methods in the Study of Language, Mathematisch Centrum, Amsterdam. Takahashi, T. and M. Suzuki (1982) "Ko So : A no shi j i - ryoeki ni tsuite," in Kokuritsu Kokugo KenkyO hokoku 71, Kenkyu Hokoku, Vol. 3, Shuei Publishing Co, Tokyo. Terakura, Hiroko (1984) "Discourse Function of Sentence-Initial NP in Conversation," Descriptive Applied Linguistics XVIII, pp. 223-235, International Christian University, Tokyo. (1984) "The Deletion of the Theme of the Sentence and Discourse Structure," in Proceedings of the Nitobe-Ohira Memorial Conference on Japanese Studies, The University of Bri t i sh Columbia, Vancouver, Br i t i sh Columbia, Canada. 283 Tichey, Pavel (1978) "Questions, Answers, and Logic," American Philosophical Quarterly, Vol. 5 No. 4, pp. 275-284. Tsao, F. (1977) A Functional Study of Topic in Chinese: The First Step toward Discourse Analysis, doctoral dissertation, University of Southern California. Tsubomoto, Atsuro (1981) "It's A l l A/o: Unifying Function of No in Japanese," in Chicago Linguistic Society, 17, pp. 393-403. Tsutsui, Michio (1981a) "El l ipsis of wa in Japanese," ms., Middlebury Symposium of Discourse Analysis in Japanese, Middlebury. (1981b) "Topic Marker El l ips is in Japanese," Studies in the Linguistic Sciences, Vol. II, No. 1 Spring, pp. 163-179. (1983) "Ell ipsis of Ga, " Papers in Japanese Linguistics, Vol. 6, pp. 199-243. (1984) "'Wa' no Shoryaku," Gengo, Vol. 13 No. 5, pp. 112-121. Vennemann, Theo (1975) "Topics, sentence accent, e l l ipsis : a proposal for their formal treatment," in Edward L. Keenan (ed.), Formal Semantics of Natural Languages, University Press, Cambridge. Watanabe, Shinichiro (1979) "Nihongo no Bunretsubun nitsuite," in Eigo to Nihongo, Kuroshioshuppan, Tokyo. Weissenborn, Jurgen and Wolfgang Klein (eds.) (1982) HERE AND THERE, Cross-Linguist ic Studies on Deixis and Demonstration, John Benjamins Publishing Co., Amsterdam. Wexler, Kenneth and M. Rita Manzini (1987) "Parameters and Learnability in Binding Theory" in Thomas Roeper and Edwin Wil l iams (eds.), Parameter Setting, D. Reidel, pp.41-76. 284 Wil l iams, Edwin S. (1 977) "Discourse and Logical Form," Linguistic Inquiry, Vol. 8, No. 1, pp. 101-139. ; and Sperber, Dan (1979) "Ordered Entailments: An Alternative to Presuppositional Theories;" Syntax and Semantics 1 1, Academic Press, New York. (1980) "Predication," Linguistic Inquiry, Vol 1 1, No. 1, pp. 203-238. (1980-81) "Remarks on Stress and Anaphora," Journal of Linguistic Research, vol. 1 no. 3, pp. 1-15. (1983) "Semantic vs. Syntactic Categories," Linguistics and Philosophy 6, pp. 423-446. (1986) "A Reassignment of the Function of LF", Linguistic Inquiry, vol. 17-2, pp. 265-299. Wilson, Deirdre (1975) Presupposition and Non-truth-conditional Semantics. Academic Press, New York. Yamanashi, Masaaki (1986) Hatsuwakoi, Taishukan, Tokyo. Zribi-Hertz, Anne (1989) "Anaphor binding and Narrative Point of View", Language 65-4, pp. 695-727. Zubin, David A. (1979) "Discourse Function of Morphology: The Focus System in German," in Talmy Givon (ed.), Syntax and Semantics, Vol. 12, Academic Press, New York.' 285 A p p e n d i x [Data] -v^y £Ho#JC£> vnciF -y 7 u tw^x. *mizx-7'm%:<x. fc o'J?L±.W%bZ*>ZX>bfrtz&10 . A £ v^  £ £ XiXX £ n h a h &^c\b&XVp<c?). Hl^ra. 7 k £ £ L & # ^ - C » 9 > < fc . * >Vy (2^-fet3 < tt ^•ftfrg&fc. & f c 3 y 3 - T ' £ o $ 9 L * ; i ? f c £ o { 1 - £ - r . x-7" fcfcoUsOfcoT. ^ ^ y fc^ft£v^*£< 0/ittfc\ ^<ft^^y ( O f e v a ^ o ^ : } : , i fc'fco&Tfc <£ < fcfrofcfc. o £ 2" fcAtiT^ott^tf^fc^v-y i ^ ^ A ^ T . ± i l $ o # o 4 ? i * j f c t *> a CSDi a t o o 5 •ef i i fr fc ir^fc . t A & * > £ > h 3 < f l J o T * S £ L - £ f c £ . Tkfcfcgj; k ^ ^ t ^ t f i T t v ^ t S . fc t * j & * f c A # £ $ v ^ B f c : f l j o T . # f c : f c $ Ltz K) L%^X, Z<7)&&X. %.=¥-ttfrti<\Q>.X'X'b% < W It £ O . *»^>*>ttr fc ^ o / c ^ . *b->X. &tt<r>&1fifrhh <h ^tX^tztb. ±fvk £B L T . <7) <£> bco%s<.X3.te%'Z^tztbX. i ixfcffiLrfcv^T. fcfc^tffc;**-fcAixTv^*:*), y ^ ^ y v - x ^ f f o t . J L ^ A S o & c o m b ^ . t i & £ fc fc'tw. ^ L T , ftZXixX . v n ^ ^ f c ' r i < ^ F 7 ? > v^o>£ A * I i - fc fc r fcfcv^ Lfrr>tz*>. [2] *£-C3!5!«&&3S£ LfcV*i ^ * £ & £ o < * i B r « L*>fc^§tilf;£T'$&£^:k^o^ttfe*i£#> 0 H ¥ r r £ o <£dfc -?£>£ . HWfc v**tftf>bX,:lfc8i31*)»±5BJI"r. #tifc:L / , £ « o T V * f c t t * i f c * fc. v*&li*£#«;fo i i -o^ . fcfifcftfcliTo < 0 fcv^. ft<o*ni»<afcv*±»fc:fc&i'y$ 9 * r * i & < H i t , * f c l £ « B»»f Lfc»T *B3L. fc* i -e fcS . eOfl-ti*£«-*ifc*fc r ^ j i 0 fc r T . S S K - f SJSEfc :*< £ f c £ . *< fci: 9 » * > o f c f e i f c £ * T . «ajH* iii'*fa»fe=&ft-*«Sfc:i: 0 . * £ & - r . ± i i £ M £ a » < 0 M LXMm fcSBfcfc&ofci: A#fc£<0»3!&£A*U KSStfiS*.*: fe M » £ A * i S . ^ t t ( D i O ? i * * « l ^ a * ^ - o t t ? i « t t i r - C - C * S j S i T . fttf)I!itf>fMfc;£fifr^. MB fc&£>&£> L T & ^ T . & < * J I ^ £ £ £&v^feT'£>&. ISli-fc^i artftf*. ifiv^fc^lft^ao ££Bift«"B ft-fc^S J: a fcfiSifcifc. SWSrAiiS tmv^i:v^||i*^rfc<7)te*s . fc ofc=fc*Hi?frJ8*S v^(i^^LTfc# . jEflOfc^fc'MJKTfc ££^T^fc*£Sfcv* . li^;bfefr< . ^fc'fcfcfc&^feft&o 287 [33 r ^ E T ' ^ o T V ^ c / v f < . * ^ f c ^ * ^ B a ^ f f c 9 2r£r*B:fr L£ L «t o . ^ • f ! ^ ^ T ' - f A T ^ ^ ^ 7 A J H * 1 * 7 7 1 l f (1 180 cc) < L>^XcotfL&^-. ^ ' { f t ' i i S r ^ & ^ f c ( t t < \^coX\ £ £ t . < i < fc v > T A i i * - * . » ^ i f c « a 1 f * # * 9 £ > £ 9 A * i 9 S K fcv^fcfrttT . H n U f c S t < o t ^ t , L* fct<7 )± fc . V ^ j t / t i t ^ T < t l £ . E H ^ O i 5 f c d ^ ( 2 . p&fc 0 lilir L f c f c v > L W t f i f c \ #3fe{i# *S*>t>tf>T-*fc. ®£<7 ) *$ i i fc ' 5^ fc 5 5:£ fc . ^HTJS'JT'. SC^m^fc fc§§^T'L J: o . #fB<7 )Mtf ) fc d ^ f c f c ^ f c / - , . m & O f c 5 ^ f c { £ . L * L i -5 . 7 * y £ £ c # . ^ g o i i f c f c 3 ^ £ I £ < S L £ L * : £ 3 u = + # f t f c * * s < fc , *a**flj*iat. fc 3 £> £ 9 * T ' L j : 3 fc. W 9 3 r l * « F * S S * » ^ > 1 - f e ^ ^ i 9 ^ L J 5 £ > f c o a t . £ 1 2 . a fc^Tv^fcT^ott^T'St^ d 5 I B S f c f c v ^ - f . £ J L L * : 4 > . i> 3 - 0 ^ < 7 ) * f c o t f T . i ^ ^ t . « 8 < f c § . fcfc 3 0 > * f f t £ £ ! ^ T . S f c & f r & ^ l ^ j & H ^ i t ^ , ft«£T5tf>.Lfc:*;«£*IH£<flJoTs 3 3 f c r « U ± i i f c 1 t $ £ 3 ^ 3 ^ - H 2 . - A # f c * » = A f l " C t ± * fc d ^ S r f c - ^ D o t t 288 L # > U ©c\commii&fr£> 0 £ v * < fefc^.-C*>. fc&frfcifcti [4] - e ^ T ' , m ^ ^ ¥ t t 7 5 ? * £ $ v * 3 5 r £ (A10) T * f e * > T * a L 7 5 ^ x f c * f c % fci-e ^ y t ^ C ^ v ^ f , £ 1 \ mtb&Mizyyvx*x <mr>xmti*&b t T f c < i t\z. t = * 5 * » f e - C t . ftofcfe$££-f£% £ f c £ 3 f t » « # y £ # - y 7 " l I P ^ f c ^ - e . ± < X <®frLXXti t-t. £ fefcBb^Jfc I T . ^2# (^J35^r> A ) c / ^ n y n - v < fcn-fcfci^-c. ^ ^ f c y ^ ^ x ^ A f L . a \Zt#ycD\Q-frmteM:^mfrzfJ±^&ZlZbbX. tzz--?fr$£bLX . ftfc<7)204Hi. fc £ fc'££>£ £ * > l - T \ *ffi*»fe^* sSS±fe=SrV^ J: 3 t » * o { t 5 § f f i t t . < D i 3 U , 30#Lfcfe&9fc 'L. *-C*9> • t#*t . t i t f t t i . t f o < DtS < t>^fc£.<&tfX'&-tW* M@^»9>t<'k r a t . afcfe^T'^-t. mzt*v-mLxi»hmm*xtix. -v-ytm -fe^y^ ^ x f c ^ r 9 £ L f c . 1&<z>llwr*a»Wfci>cottzft^ z fc . 289 [53 *#»±v**SJKfc:.i.i-*&B8tt-6. o J t i a ^ i S l f r ^ H , m*b-i. fc # te. i » t 4 l : j a ^ H 5 ^ t , ^tiOJBcJgfrferE. E x3\<. ±£)?£*m^fc.frtr. ®fr£.m#>h. ® S ffl T J l i 11TJ 0) 38 tf> t& # * "T i l WT i V * <o t? £> £ . ^ 5 I P f c J £ L < ^ - ^ 9 . £ * > ± « fc # fc te . £ ^ ' M ^ £ t A § ' f c ! g S . <r<?)fc #H*»fe±<?)*»{ i . < i *S&v^J : a fcL. MJ£co*»*»fc fc [Observations] Sample (1): After the description of the rough procedure for each dish, the dish is referred to by ko. Sample (2): The main topic is (sake no) zoom'. Ko 1 — kono zooni 2 — kore = sanpei-ziru (sake no zooni) 3 — kono sakana -> sake 290 Sample (3): Ko 1 ..."miso " is the main concern 2 ...the topic is shifted to momen-doohu 3 ...dengaku is the topic of the whole paragraph 4 ...miso is now being focused; miso is the essential factor of dengaku Sample (4): Ko 1 — kono burausu (the focused object of the dyeing procedure). 2 — koko e 3 — koo-site These referents are in certain stages 4 — kore ga of the procedure 5 — koo-site Sample (5): The procedure is s t r ic t ly temporal-linear order in this case, and ko refers to a certain point of such a procedure to add the comment. In this kind of typical procedural discourse, the reader follows the procedure wi th some visualizat ion of what he is doing. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items