UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The postposing construction in Japanese Rosen, Eric Robert 1995

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

Item Metadata

Download

Media
831-ubc_1996-0595.pdf [ 6.45MB ]
Metadata
JSON: 831-1.0087263.json
JSON-LD: 831-1.0087263-ld.json
RDF/XML (Pretty): 831-1.0087263-rdf.xml
RDF/JSON: 831-1.0087263-rdf.json
Turtle: 831-1.0087263-turtle.txt
N-Triples: 831-1.0087263-rdf-ntriples.txt
Original Record: 831-1.0087263-source.json
Full Text
831-1.0087263-fulltext.txt
Citation
831-1.0087263.ris

Full Text

T H E POSTPOSING CONSTRUCTION IN JAPANESE by ERIC ROBERT ROSEN B.G.S. Simon Fraser University, 1975 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIREMENTS FOR T H E DEGREE OF MASTER OF ARTS in T H E F A C U L T Y OF G R A D U A T E STUDIES Department of Linguistics We accept this thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH COLUMBIA October 1996 © Eric Robert Rosen, 1996 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 LjY\Ql {/) $>j\C 4 The University of British Columbia Vancouver, Canada -6 (2/88) 11 A b s t r a c t T h i s work exam ines t h e s y n t a c t i c n a t u r e o f t h e p o s t p o s i n g c o n s t r u c t i o n i n J a p a n e s e , w h i c h i s a c o n s t r u c t i o n commonly h e a r d i n c a s u a l c o n v e r s a t i o n , i n w h i c h a c o n s t i t u e n t o f t h e s e n t e n c e o c c u r s t o t h e r i g h t o f t h e m a t r i x v e r b , o u t s i d e o f i t s c a n o n i c a l p o s i t i o n . T h i s c o n s t r u c t i o n a p p e a r s t o v i o l a t e a s t r i c t • c o n d i t i o n i n J a p a n e s e on t h e l i n e a r o r d e r i n g o f c o n s t i t u e n t s w i t h i n a p h r a s e : n a m e l y , t h a t a l l c o n s t i t u e n t s o f a m a x i m a l p r o j e c t i o n must p r e c e d e t h e h e a d o f t h e p r o j e c t i o n . The q u e s t i o n s I s e e k t o answer abou t t h e p o s t p o s i n g c o n s t r u c t i o n a r e as f o l l o w s : 1. Does a p o s t p o s e d p h r a s e have a s y n t a c t i c r e l a t i o n s h i p w i t h t h e r e s t o f t h e s e n t e n c e o r i s i t a mere " a f t e r t h o u g h t " t o t h e s e n t e n c e . 2. Does t h e p o s t p o s i n g c o n s t r u c t i o n obey s y n t a c t i c c o n s t r a i n t s ? 3. I f p o s t p o s e d p h r a s e s a r e a s y n t a c t i c phenomenon, a r e t h e y d e r i v e d by movement o r by b a s e - g e n e r a t i o n ? 4 . Why does t h e p o s t p o s i n g c o n s t r u c t i o n a p p e a r t o v i o l a t e t h e " l e f t - a t t a c h i n g " n a t u r e o f J a p a n e s e l i n e a r p r e c e d e n c e ? The f ramework o f my a n a l y s i s i s " c l a s s i c a l " g o v e r n m e n t - b i n d i n g t h e o r y o f t h e 1980 's and e a r l y 1990 ' s ; I a l s o draw on some o f t h e a p p r o a c h e s o f t h e M i n i m a l i s t programme o f Chomsky (1995). The d a t a I u s e a r e n a t i v e - s p e a k e r j udgemen ts o f p o s t p o s e d J a p a n e s e s e n t e n c e s . The m a i n c o n c l u s i o n s I come t o abou t t h e p o s t p o s i n g c o n s t r u c t i o n a r e as f o l l o w s : 1. T h e r e a r e s y n t a c t i c c o n s t r a i n t s on p o s t p o s e d p h r a s e s t o t h e e x t e n t t h a t we c a n o n l y e x p l a i n p o s t p o s i n g as a s y n t a c t i c phenomenon r a t h e r t h a n as an a f t e r t h o u g h t : f o r e x a m p l e : Ill (a) Postposed phrases show evidence that they are subject to subjacency-l i k e e f f e c t s . (b) Traces of postposed phrases show evidence of a head-government l i c e n s i n g requirement. 2. Postposed phrases show Condition C reconstruction e f f e c t s which are most e a s i l y explained by movement of the postposed phrase. 3. The unique nature of postposed phrases as sole exception to the s t r i c t l e f t -attaching nature of Japanese can be explained by the uniqueness of the p o s i t i o n that they occur i n : I post that they are right-adjoined to the root node of the sentence. Under the minimalist framework of Chomsky(1995), a p o s i t i o n adjoined to the root node i s the only p o s i t i o n that i s not i n the minimal domain of a head. iv TABLE OF CONTENTS Abstract i i Table of Contents i v Acknowledgement v i i 0. Introduction 1 0.1 Assumptions and framework 3 0.2 Japanese as a s t r i c t l y " l e f t - a t t a c h i n g " language 7 0.3 Postposing as a syntactic phenomenon 11 0.4 A comparison of postposing, scrambling, and t o p i c a l i z a t i o n . 16 1. Evidence for movement of postposed phrases. 24 1.1 Postposing from matrix clauses 35 1.2 Postposing from embedded clauses 50 1.3 Summary 61 2. L o c a l i t y constraints on postposed phrases I subject/object asymmetries 63 2.1 Postposing a r e l a t i v e clause out of a non-matrix CP 64 2.2 Postposing a r e l a t i v e clause out of a wh-island 68 2.3 Postposing a genitive out of a DP insi d e an adverbial clause: 72 2.4 Postposing a r e l a t i v e clause out of an adverbial clause 74 2.5 Postposing a dative out of a r e l a t i v e clause 75 2.6 Postposing a r e l a t i v e clause out of a r e l a t i v e clause 76 2.7 Postposing out of koto clauses that are subjects of t r a n s i t i v e verbs 80 V 2.8 Postposing a r e l a t i v e clause out of a matrix subject 2.9 Summary 82 83 3. L o c a l i t y constraints on postposed phrases II --i s l a n d e f f e c t s that are r e l a t i v i z e d with respect to a [+N] feature 88 3.1 Postposing out of a r e l a t i v e clause: worse deviance when a DP i s postposed than when another category i s postposed 91 3.2 Postposing out of other c l a u s a l structure with no [+N] head: no [+N] domain e f f e c t s when DP's are postposed 105 3.3 Conclusions 107 3.4 A r e l a t i v e clause as a weaker i s l a n d than other kinds of adjuncts 108 4. Postposing out of wa-marked phrases 115 4.1 Are topics islands for extraction? 119 4.2 Discriminating between d i f f e r e n t types of adjuncts 125 4.3 Postposing out of VP topics 135 4.4 Postposing out of non-case-marked CP topics 137 4.5 wa-marked purpose clauses 140 4.6 Do CP topics pattern with DP topics with respect to Saito's diagnostics? 146 4.7 L e x i c a l government of subjects i n Japanese 147 4.8 Summary and r e l a t e d issues 148 vi 5. Postposing as a root clause phenomenon: an explanation i n terms of Minimal Domains 160 5.1 Introduction 160 5.2 A member of the domain of no category 161 5.3 The head parameter i n a l e f t - a t t a c h i n g language 170 5.4 Proposed conditions on l i n e a r ordering 174 5.5 Summary 183 6. Non-syntactic factors i n postposing 184 6.1 Postposed phrases l i k e to be "heavy" 185 6.2 Analysis of postposed subject.with object i n canonical p o s i t i o n 186 7. Summary 191 Appendix A: Early Analyses of the Postposing Construction 193 Appendix B: No-marked phrases 195 V l l Acknowledgement I wish to thank a number of people whose support made t h i s research p o s s i b l e . My s u p e r v i s o r , Michael Rochemont, was instrumental i n l e a d i n g me to and g u i d i n g me through t h i s p a r t i c u l a r research t o p i c . When, four years ago, I took on the r a t h e r dubious e n t e r p r i s e of t r y i n g to re-enter the academic world a f t e r a h i a t u s of s e v e r a l decades, i t was l a r g e l y through h i s encouragement and support that I managed to progress t h i s f a r . The other two members of my committee, Rose-Marie Dechaine and Hisatsugu K i t a h a r a were a l s o extremely generous w i t h t h e i r help and encouragement. Rose-Marie provided a wealth of suggestions about the e x p o s i t i o n and argumentation of the manuscript as w e l l as making me aware of analyses of s y n t a c t i c phenomena that r e l a t e d to my research that I had not been aware of. Hisa's i n s i g h t i n t o Japanese syntax and f a m i l i a r i t y w i t h the most current models of syntax such as the M i n i m a l i s t programme were extremely v a l u a b l e . Tomio Hirose and A k i h i k o Uechi spent countless hours i n a s s i s t i n g me w i t h judgements of sentences. Without t h e i r help, t h i s research would have been impossible. E a r l i e r stages of t h i s research b e n e f i t t e d g r e a t l y from conversations w i t h Henry Davis, J i l a Ghomeshi and W i l l Thompson. I a l s o wish to thank the S o c i a l Sciences and Humanities Research Council of Canada, grant #410-92-1379 "The Syntax and Phonology of Focus" f o r t h e i r support during the e a r l y stages of t h i s research. F i n a l l y , I wish to thank a l l the members of the Department of L i n g u i s t i c s at the U n i v e r s i t y of B r i t i s h Columbia -- both f a c u l t y and graduate students --f o r p r o v i d i n g a very f r i e n d l y and supportive working atmosphere. I n t r o d u c t i o n 1 The postposing c o n s t r u c t i o n i n Japanese has been discussed by a number of authors (see below); most r e c e n t l y by Endo (1989) and Whitman (1991b). The co n s t r u c t i o n i s an apparent v i o l a t i o n of an otherwise s t r i c t c o n d i t i o n that a l l phrases i n the .language are h e a d - f i n a l . Endo (1989) describes postposing as "the s t r u c t u r e i n which one or more c o n s t i t u e n t s appear f o l l o w i n g the matrix p r e d i c a t e , " and comments that i t occurs f r e q u e n t l y i n both c o n v e r s a t i o n a l and inf o r m a l speech and i n c e r t a i n forms of w r i t i n g such as poems, cartoons and advertisements. Endo's data show that a wide v a r i e t y of elements can be postposed to the r i g h t of the matrix verb i n Japanese sentences: f o r example, nominative, a c c u s a t i v e , d a t i v e , and g e n i t i v e DP's; PP's and ob l i q u e s ; a l l types of embedded clauses ( i n c l u d i n g r e l a t i v e c l a u s e s ) ; complex DP's, adjunct phrases of a l l types, t o p i c phrases, AP's when they modify a noun (but not AP p r e d i c a t e s ) . She discusses previous analyses of the c o n s t r u c t i o n , namely "non-movement" analyses i n Inoue (1978), Kuno (1978a&b), and S a i t o (1985) and "movement" analyses i n Haraguchi (1973) and Kuroda (1980). In her a n a l y s i s , she argues that postposing i s a r e s u l t of s y n t a c t i c rightward movement because i t shows subjacency e f f e c t s . A study of t h i s c o n s t r u c t i o n i s p a r t i c u l a r l y u s e f u l f o r understanding the p r o p e r t i e s of s y n t a c t i c movement i n Japanese because, apart from the w e l l - s t u d i e d phenomenon of scrambling 1, Japanese does not show as much evidence of s y n t a c t i c movement as languages l i k e E n g l i s h , because of i t s w h - i n - s i t u and base-generated t o p i c s 2 . Scrambling appears to be r e s t r i c t e d to a l i m i t e d range of phrase-types e.g. CP's and accu s a t i v e and d a t i v e DP's, and does not permit movement of 'See Abe (1993), Nemoto (1993), Saito (1985(, (1986), (1989), (1992), (1994a), Yoshimura (1992) for analyses of scrambling in Japanese. 2See Saito (1986). 2 genitives, r e l a t i v e clauses, or subjects 3, a l l of which can be postposed. Because the postposing operation i s possible for a wider v a r i e t y of phrase types than scrambling, i t has the p o t e n t i a l to shed some l i g h t on c e r t a i n syntactic phenomena of Japanese that are d i f f i c u l t to examine i n the context of scrambling. Although i t seems l i k e l y that discourse and other non-syntactic factors are relevant i n explaining the phenomenon of postposing, i n t h i s paper I s h a l l l i m i t a t tention to the synta c t i c properties of the postposing construction using c l a s s i c a l generative p r i n c i p l e s . In a number of places i n my analysis I draw f r e e l y from minimalist p r i n c i p l e s of Chomsky (1995) -- i n p a r t i c u l a r , the concept of a minimal domain -- but I have not adopted the e n t i r e Minimalist programme as the framework i n which I am examining postposing. The following i s a summary of the conclusions I come to about the postposing construction: 3333 1. The postposing construction shows clear evidence that i t i s a syntactic phenomenon rather than an "afterthought" construction. Postposed phrases exhibit a set of l o c a l i t y constraint patterns which include (i) subject/object asymmetries ( i i ) r e l a t i v i z a t i o n with respect to a [+N] feature, and ( i i i ) conditions on traces of postposed phrases that require government by a l e x i c a l [+V] head. 4 2. Postposed phrases show evidence that they are derived by movement rather than base-generation. The evidence for movement i s based mainly on Condition C 3See Saito (1985) sec. 3.2 for evidence that long-distance scrambling of a subject is not possible. Saito also argues that short-distance scrambling of subjects is not possible, and explains the subject/object asymmetry with respect to scrambling in Japanese by a difference in case-assignment. He proposes that subjects, unlike objects, are inherently case-marked in Japanese. 4(See Cinque (1990) and Chung (1994) for similar proposals.) 3 reconstruction and anti-reconstruction e f f e c t s that are observed when a name i s embedded i n a postposed phrase whose canonical p o s i t i o n i s c-commanded by an element coreferent with the name. A movement hypothesis i s also consistent with subject-object asymmetries and r e l a t i v i z e d - m i n i m a l i t y - l i k e phenomena, which are more e a s i l y explained through a movement ana l y s i s . 3. Postposing shows a heaviness e f f e c t that i s reminiscent of rightward Heavy NP S h i f t and r e l a t i v e clause extraposition i n English. 4. The concept of minimal domains i n Chomsky (1995) explains the facts th^t ( a) r i g h t adjunction can only occur to the root node i n Japanese and (b) l e f t adjunction appears to be possible to any p r o j e c t i o n except the root node. Before discussing these points I f i r s t give some examples that show that postposing i s syntactic and not merely an "afterthought" tacked onto the sentence. Then, as background, I give a summary of the s i m i l a r i t i e s and differences between the three main types of d i s l o c a t i o n i n Japanese: scrambling, postposing and t o p i c a l i z a t i o n . 0.1 Assumptions and framework In t h i s analysis I look at postposing from the point of view of " c l a s s i c a l " generative theory. In p a r t i c u l a r , I assume the following: - X' theory (e.g. Jackendoff (1977)) - binding theory (Chomsky (1981)) - p r i n c i p l e s of head-government (Chomsky (1981)) - the ECP based on a head government and antecedent government requirement (Lasnik & Saito (1984)) - extended projection, a f t e r Grimshaw (1991): s p e c i f i c a l l y that IP and CP are i n 4 the extended projection of V and NP, DP and PP are in the extended projection of N - specifier positions only occur on the l e f t , regardless of the order of head-complement or head-adjunct I also draw some principles from the more recent minimalist programme of Chomsky (1993 - 1995) - - spec i f i ca l ly , the concept of the domain of a head - - but my account is not s t r i c t l y minimalist. 0.1.1 Reconstruction For convenience I shall follow the analysis of reconstruction of Chomsky (1995), using the "copy theory" of movement. In the Condition C reconstruction examples that I shall present in chapter 1, I shall assume that the reconstruction of a postposed phrase is explained by the fact that a copy of the phrase is le f t behind in si tu and survives u n t i l LF, even though i t is deleted somewhere on the path of the derivation that leads to the phonetic component of the grammar. In the same chapter, I shall assume that the "anti-reconstruction" effects that I shall present are explained by the fact that certain types of adjuncts such as relative clauses behave as i f they adjoin to the DP they modify after that DP has undergone movement.5 In a copy theory model, such an assumption w i l l explain why elements within such a relative clause do not appear to suffer Condition C reconstruction effects . I shall use the term "anti-reconstruction" to describe this lack of Condition C reconstruction effects typical ly seen in names that are embedded in a relat ive clause that modifies a phrase that undergoes reconstruction. I use the term "certain types of adjuncts" above in referring to relative 5In a D-structure model, the idea that phrases such as relative clauses attach to the projection they modify after that phrase has undergone movement will mean that such a relative clause will not exist at D-structure. This creates a problem in that under such an analysis, with a D-structure model, the Projection Principle cannot hold at D-structure. 5 c l a u s e s because g e n i t i v e m o d i f i e r s o f a DP, w h i c h , I am assuming i n my a n a l y s i s t o be a d j u n c t s , do show C o n d i t i o n C r e c o n s t r u c t i o n e f f e c t s . I s h a l l n ot at t e m p t t o e x p l a i n t h i s d i v e r g i n g b e h a v i o u r o f r e l a t i v e c l a u s e s and g e n i t i v e s w i t h r e s p e c t t o r e c o n s t r u c t i o n . I n o t h e r r e s p e c t s , i n t h e examples I p r e s e n t i n t h i s p a p e r , g e n i t i v e s and r e l a t i v e c l a u s e s p a t t e r n s i m i l a r l y . F o r example, t h e y show s i m i l a r b e h a v i o u r w i t h r e s p e c t t o t h e i r p o s t p o s a b i l i t y i n v a r i o u s e n v i r o n m e n t s , w h i c h makes i t n a t u r a l t o assume t h a t g e n i t i v e s a r e a d j u n c t s , l i k e r e l a t i v e c l a u s e s . The o n l y e x c e p t i o n t o t h i s a s s u m p t i o n i s t h e c a s e of g e n i t i v e s t h a t a r e arguments o f nouns such as " s t u d y " i n p h r a s e s t h a t a r e a n a l o g o u s t o "the s t u d y o f E n g l i s h " i n E n g l i s h . We s h a l l a l s o see t h a t g e n i t i v e s , u n l i k e s u b j e c t s , do not show SC e f f e c t s , w h i c h s u g g e s t s t h a t g e n i t i v e s a r e i n an a d j o i n e d r a t h e r t h a n s p e c i f i e r p o s i t i o n . I f p o s t p o s e d p h r a s e s (and a l s o A ' - s c r a m b l e d p h r a s e s , w h i c h a r e r e l e v a n t i n s e c . 1.1.2.2) a r e d e r i v e d by t h e copy t h e o r y of movement o f Chomsky (1995:202-210) , t h e n t h e r e must be an o p e r a t o r - v a r i a b l e r e l a t i o n a t LF between some element i n t h e p o s i t i o n o f t h e p o s t p o s e d p h r a s e and t h e c a n o n i c a l p o s i t i o n o f t h e p o s t p o s e d p h r a s e . I n o r d e r f o r t h i s t o o c c u r , t h e r e must be "complementary d e l e t i o n " a t LF of e v e r y t h i n g e x c e p t an o p e r a t o r p h r a s e i n t h e o p e r a t o r p o s i t i o n ; i n t h e t r a c e p o s i t i o n , "the copy of what rema i n s i n t h e o p e r a t o r p o s i t i o n d e l e t e s . " I n t h e examples d i s c u s s e d i n Chomsky (1995:202-210), t h e r e i s an o v e r t w h - o p e r a t o r . There i s no such o v e r t o p e r a t o r i n a p o s t p o s e d o r s c r a m b l e d p h r a s e . I n t h e examples I s h a l l p r e s e n t , i t w i l l be n e c e s s a r y t o assume t h a t a t l e a s t some o f t h e m a t e r i a l o f t h e p o s t p o s e d p h r a s e i n t h e v a r i a b l e p o s i t i o n i s not d e l e t e d a t LF. One way i n w h i c h s u c h an LF c o n f i g u r a t i o n might o c c u r i s t h a t t h e r e i s an empty o p e r a t o r a t LF i n t h e p o s i t i o n o f t h e p o s t p o s e d p h r a s e and a v a r i a b l e a s s o c i a t e d w i t h t h e p o s t p o s e d p h r a s e t h a t o c c u r s i n t h e in s i t u p o s i t i o n a t LF. I s h a l l assume t h a t some k i n d o f c o n f i g u r a t i o n o f t h i s t y p e o c c u r s a t LF w i t h p o s t p o s e d p h r a s e s but I s h a l l not t r y t o d e t e r m i n e t h e e x a c t n a t u r e o r s t r u c t u r e o f t h e empty o p e r a t o r o r t h e v a r i a b l e . The f o l l o w i n g d i a g r a m i l l u s t r a t e s r o u g h l y t h e p r o p o s e d c o n f i g u r a t i o n , w i t h u n d e r l i n e d e l e m e n t s b e i n g d e l e t e d at or by LF . XP represents the postposed phrase, Op an empty operator , and x a v a r i a b l e . (1) CP / \ CP Op[XP] / \ / \ x [XP] 0.1.2 C-command In order to argue that there i s a c-command r e l a t i o n between a postposed phrase and the empty category i n i t s canonical p o s i t i o n I s h a l l assume that c o n d i t i o n s f o r c-command can be vacuously s a t i s f i e d when the c-commanding p o s i t i o n i s one that i s adjoined to the root node. Consider the f o l l o w i n g s t r u c t u r e : (2) XP / \ XP YP / \ ZP We want to be able to say that YP c-commands ZP, where no category dominates XP. The c-command c o n d i t i o n says that YP c-commands ZP i f f : (a) YP does not dominate ZP (b) every p r o j e c t i o n that dominates YP dominates ZP 7 C o n d i t i o n (a) above i s c l e a r l y s a t i s f i e d . C o n d i t i o n (b) above h o l d s v a c u o u s l y : no p r o j e c t i o n d o m i n a t e s YP s i n c e XP o n l y c o n t a i n s Y P . I s h a l l a l s o make t h e f o l l o w i n g a s s u m p t i o n s abou t J a p a n e s e . Some o f t h e s e a s s u m p t i o n s a r e no t c r u c i a l t o my a n a l y s i s b u t a r e made f o r e a s e o f e x p o s i t i o n . 1. T o p i c s a r e i n [ S p e c , C P ] . 2. G e n i t i v e s and r e l a t i v e c l a u s e s a r e a d j o i n e d t o a p r o j e c t i o n o f D. 3. V P - i n t e r n a l s u b j e c t 4. no h e a d movement o f V t o I 5. C l a u s e l e v e l a d j u n c t s s u c h as c o n c e s s i v e c l a u s e s a r e a d j o i n e d t o a p r o j e c t i o n o f I. 6. koto i s a c o m p l e m e n t i z e r t h a t i n t r o d u c e s c a s e - m a r k e d c l a u s e s 7 . n o m i n a t i v e a rgumen ts o f u n a c c u s a t i v e v e r b s s u c h as nigeta " t o r u n away" a r e u n d e r l y i n g l y i n o b j e c t p o s i t i o n 0.2 J a p a n e s e as a s t r i c t l y " l e f t - a t t a c h i n g " l a n g u a g e The f o l l o w i n g d a t a show t h a t i n J a p a n e s e , f o r a l l t y p e s o f c a t e g o r i e s , b o t h comp lements and a d j u n c t s o b l i g a t o r i l y p r e c e d e t h e h e a d o f t h e p h r a s e t h e y o c c u r i n . 0.2.1 Complements must p r e c e d e heads N P : (3a) [ e i g o - n o [ b e n k y o o j ] E n g l i s h - g e n s t u d y " t h e s t u d y o f E n g l i s h " 8 (3b) * [ [benkyooN] e igo-no] N P NP study Engl ish-gen | 'the study of E n g l i s h " N' / \ N XP PP: (4a) [Nihon-[kara P ] ] Pp k i t a . Japan-from came "He/she/I came from Japan." * PP (4b) *[[kara P ] Nihon] P P k i t a | from Japan came P' "He/she/I came from Japan." / \ P XP VP: (5a) [e-o [kaki v] ]VP-wa s i n a i . [p ic ture-acc draw]-top do-neg •k "As for drawing a p i c t u r e , I don ' t do i t " . VP !5b) * [ [kaki v ] e-o]V P-wa s i n a i V [draw p i c t u r e - a c c ] - t o p do-neg / \ "As for drawing a p i c t u r e , I don ' t do i t " . V XP 9 0.2.2 Adjuncts must precede heads NP: (6a) [Taroo-ga/no k a i t a [eN]]N P-ga s u k i . Taro-nom/gen drew picture-nom l i k e a b l e "I l i k e the p i c t u r e that Taro drew." * NP (6b) *[[eN] Taroo-ga/no k a i t a ] - g a s u k i . / \ [picture -nom/gen drew]-nom l i k e a b l e NP XP "I l i k e the p i c t u r e that Taro drew." VP: (7a) [zyoozu-ni e-o [kaki v] ]VP-wa s i n a i . [ s k i l l f u l l y p i c t u r e - a c c draw]-top do-neg "As f o r drawing a p i c t u r e s k i l l f u l l y , I don ' t do i t " . * VP (7b) *[e-o [kaki v] zyoozu-ni] V P -wa s i n a i / \ [p ic ture-acc draw s k i l l f u l l y ] - t o p do-neg VP XP "As for drawing a p i c t u r e s k i l l f u l l y , I don ' t do i t " . 10 IP: (8a) Taroo-wa [[[kuruma-ga kowareta k a r a ] P P korarenakat-[taj I P] t o ] C P i t t a . Taroo-top car-nom broke because couldn't-come C said "Taro said that he couldn't come because h i s car broke down." (8b) *Taroo-wa [[korarenakat-[ta z] [kuruma-ga kowareta k a r a ] P P I P] t o ] C P i t t a . Taroo-top car-nom broke because couldn't-come C said "Taro said that he couldn't come because h i s car broke down." * IP / \ IP XP 0.2.3 Postposing as a sole exception to left-attachment Postposing i s a s t r i k i n g exception to the l e f t - a t t a c h i n g nature of Japanese. When a phrase i s postposed, i t s occurrence to the r i g h t of the root node seems to v i o l a t e the s t r i c t head-final constraint on l i n e a r order of phrases. (9a)(canonical sentence) (From [ l o c . subj. V]I] koko-ni neko-ga i r u here-loc cat-nom exist "There i s a cat here." Endo(1989:85)) IP / \ VP I / \ PP V / \ DP V 11 (9b) (postposed sentence) IP [ l o c . V I] s u b j . / \ koko-ni i r u , neko-ga. IP DPj h e r e - l o c e x i s t cat-nom / \ ( I t ) ' s here - - the ca t . VP I / \ PP V / \ e i V 0.3. Evidence that postposing i s a s y n t a c t i c phenomenon Because postposing appears to be such an anomaly i n the language with respect to the issue of l i n e a r order of c o n s t i t u e n t s , there has been a tendency to view i t as a n o n - s y n t a c t i c phenomenon, expla inable as an "afterthought" c o n s t r u c t i o n . In t h i s s e c t i o n I s h a l l present evidence that postposing i s subject to s y n t a c t i c c o n s t r a i n t s . I s h a l l show that sentences with a postposed phrase that contains a name show C o n d i t i o n C r e c o n s t r u c t i o n e f f e c t s when there i s a pronoun c o r e f e r e n t i a l with the name i n the main sentence that c-commands the canonica l p o s i t i o n of the postposed phrase. This evidence for r e c o n s t r u c t i o n of the postposed phrase suggests that there i s a s y n t a c t i c r e l a t i o n between the postposed phrase and the res t of the sentence. Consider (10) and (11). (10) i s the non-postposing v e r s i o n and (11) i s the same sentence with an element postposed. (10) [ [ [kare. i / : i -ga [TaroOi-no tomodati -o] k seme-ta I P] -koto/no-ga C P ] akiraka da I P] . him -nom [T . -gen f r i e n d s - a c c ] blamed- COMP-nom evident copula It i s evident that hei blamed [ T a r o * i / : j ' s f r iends] k . 12 (11) [ [ [kare, i / : i-ga e k seme-ta I P]-koto/no-ga C P] a k i r a k a da I P] , [ T a r o O i - n o tomodati-o ] k . hiirii-nom blamed- COMP-nom evident copula T^-gen f r i e n d s - a c c I t i s evident that het blamed themk -- [TarOi's f r i e n d s ] k . In (11), postposing can occur only i f the pronoun subject of the embedded clause i s not coindexed w i t h the name contained i n the postposed phrase. This f a c t suggests that there i s some s y n t a c t i c r e l a t i o n between the postposed phrase and the main sentence and that some k i n d of b i n d i n g v i o l a t i o n occurs when there i s coreference. I f postposing were a mere "afterthought" c o n s t r u c t i o n , there should be no s y n t a c t i c r e l a t i o n between the postposed phrase and the main sentence and no b i n d i n g v i o l a t i o n should be o c c u r r i n g here. In chapter 1 I s h a l l make a d e t a i l e d a n a l y s i s of these kinds of sentences and s h a l l propose an e x p l a n a t i o n f o r the i m p o s s i b i l i t y of coreference. 0.3.1 Other analyses of postposing as a s y n t a c t i c phenomenon Whitman (1991b) describes postposing as "adjunction to the r i g h t of the main verb" and c i t e s examples (W(20,21)) from Kuno (1978) that show that postposing i s subject to subjacency. These examples a l s o support the hypothesis that postposing i s a s y n t a c t i c phenomenon r a t h e r than an afterthought to the sentence. Endo (1989) proposes that the postposed phrase undergoes rightward s y n t a c t i c movement to a p o s i t i o n adjoined to the matrix S' node (CP i n my a n a l y s i s . ) She supports t h i s a n a l y s i s w i t h examples that suggest that postposing shows subjacency e f f e c t s such as the adjunct c o n d i t i o n : f o r example she shows that a postposed phrase cannot be e x t r a c t e d out of an adjunct such as a r e l a t i v e 13 clause 6 (12), or adverbial clause (13). Examples (12)-(8) below and t h e i r judgements are from Endo (1989) . (12) *[ N P[ s,mari ga et katta] mise]-ni i t t a no, d r e s s - O i . [E77b] Mari-NOM bought store LOC went COMP dress-ACC I went to the store where Mari bought -- a dress. (Object of r e l a t i v e clause cannot be postposed.) (13) * [ P P ej mita ato-de] piano-o rensyuu-suru yo, t e r e b i - O i -[E130a] saw a f t e r piano-ACC practise-do EMPH TV-ACC I ' l l p r a c t i s e piano a f t e r I have watched -- TV. (Object of temporal adjunct clause cannot be postposed.) (14) [ P P terebi-o mita ato-de] e± rensyuu-suru yo, piano-o [E130b] TV-ACC saw a f t e r practise-do EMPH piano-ACC I ' l l p r a c t i s e a f t e r I watch TV -- piano. (Object of matrix clause can be postposed.) (15) koko-ni [ N P e± wine-ga aru yo, [s. kinoo katta] £ [E13a] here-LOC wine-NOM exi s t EMPH yesterday bought Here i s the wine -- that I bought yesterday. 6Some of the examples that I sh a l l present w i l l show that postposing out of an adjunct island such as a re l a t i v e clause can be grammatical. In the following example, which i s discussed in sec. 1.2.3.1, postposing of the genitive violates any version of Subjacency. (i) [Hanako-ni [ei omotya-o] watasita kodomo-ga] nigeta, TaroOj-no. H.-dat toy-acc handed-over child-nom f l e d T.-gen The c h i l d who handed over the toy to Hanako ran away -- Taro's (toy) I s h a l l argue that postposing shows subjacency-like effects but l o c a l i t y constraints on postposing, although they w i l l be shown to exist, do not exactly follow any constraint model that I know of. 14 ( R e l a t i v e c l a u s e m o d i f y i n g m a t r i x s u b j e c t c a n b e p o s t p o s e d 7 . ) ( 1 6 ) * [ N P et w i n e ] - w a k o k o - n i a r u y o , [ s , k i n o o k a t t a ] i [ E l 3 b ] w i n e - T O P h e r e - L O C e x i s t E M P H y e s t e r d a y b o u g h t * A s f o r t h e w i n e , h e r e i t i s - - t h a t I b o u g h t y e s t e r d a y . ( R e l a t i v e c l a u s e m o d i f y i n g t o p i c c a n n o t b e p o s t p o s e d . ( B u t s e e f o o t n o t e 5 ) ) H o w e v e r , h e r a n a l y s i s , w h i c h r i g h t - a d j o i n s t h e p o s t p o s e d p h r a s e t o C P , l e a v e s a n u m b e r o f f a c t s u n e x p l a i n e d . O n e i s t h e f a c t t h a t i f p o s t p o s i t i o n i n J a p a n e s e i s r i g h t a d j u n c t i o n , t h e n w h y i s n o o t h e r t y p e o f r i g h t w a r d a d j u n c t i o n p e r m i t t e d i n t h e l a n g u a g e ? F o r e x a m p l e , a n e l e m e n t c a n , u n d e r c e r t a i n c o n d i t i o n s , b e p o s t p o s e d f r o m a n e m b e d d e d c l a u s e , b u t i t c a n o n l y o c c u r t o t h e r i g h t o f t h e m a t r i x c l a u s e : i t c a n n o t a d j o i n t o t h e r i g h t o f a n e m b e d d e d c l a u s e . 8 ( 1 7 ) i s t a k e n a g a i n f r o m E n d o ( 1 9 8 9 ) . ( 1 7 ) [ e i b u s n o t t e i k u n o ] C P m e n d o - k u s a i d e s y o o ? a t t i - k o t t i i . [ E 6 a ] b u s r i d e - I N F L g o - i n f l C O M P l o t - o f - t r o u b l e - i n f 1 c o p u l a - A S P t h e r e - h e r e T o g o r i d i n g a b u s i s a l o t o f t r o u b l e i s n ' t i t ? - - h e r e a n d t h e r e . I n ( 1 8 ) b e l o w , w h e n t h e s a m e p h r a s e i s r i g h t - a d j o i n e d t o a n o n - r o o t n o d e , s e v e r e d e v i a n c e r e s u l t s . ( 1 8 ) ( a ) * [ [ [ e ± b u s n o t t e i k u ] I P a t t i - k o t t i j I P n o ] m e n d o - k u s a i d e s y o o ? b u s r i d e - I N F L g o - i n f l t h e r e - h e r e C O M P t r o u b l e - A S P c o p u l a - A S P T o g o r i d i n g a b u s - - h e r e a n d t h e r e - - i s a l o t o f t r o u b l e , i s n ' t i t ? T h e g a - m a r k e d D P i n t h i s e x a m p l e m a y b e a n u n a c c u s a t i v e . E v i d e n c e i n c h a p t e r 2 s u g g e s t s t h a t i t i s d i f f i c u l t t o p o s t p o s e o u t o f a n u n e r g a t i v e s u b j e c t . 8 M u l t i p l e r i g h t a d j u n c t i o n s t o t h e m a t r i x c l a u s e d o a p p e a r t o b e p o s s i b l e . S e e E n d o ( 1 9 8 9 ) f o r e x a m p l e s o f m u l t i p l e r i g h t a d j u n c t i o n s . I d i s c u s s t h i s p h e n o m e n o n i n c h a p t e r 5 . 15 ( 1 8 )(b)* [ [ e ibus notte iku n o ] C P a t t i - k o t t i j C P mendo-kusai desyoo? bus ride-INFL g o - i n f l COMP there-here trouble-ASP copula-ASP To go r i d i n g a bus -- here and there -- i s a l o t of trouble, i s n ' t i t ? (18)(c) structure of (18) (a) * CP1 I \ I \ I \ CP2 C / \ \ CP2 [post'd phr. ] i IP / \ / \ IP C /. \ / -no [lot-of-trouble-copula] I' / \ [e; to go r i d i n g a bus] 16 (18) (d) structure of (18) (b) / \ / \ / \ C P 2 C / \ C I P / \ / \ I P c / \ / -no [lot-of-trouble-copula] I' / \ I' [post'd phr.]i / \ [ej to go r i d i n g a bus] I propose a so l u t i o n to t h i s problem i n chapter 5. In summary, the postposing construction shows good evidence of being a synta c t i c phenomenon. It also exhibits properties that are d i f f e r e n t from two other types of d i s l o c a t i o n i n the language -- scrambling and t o p i c a l i z a t i o n . 0.4. A comparison of three types of d i s l o c a t i o n i n Japanese The following i s a summary of the s i m i l a r i t i e s and differences between the three phenomena of scrambling, t o p i c a l i z a t i o n , and postposing. I use the term " d i s l o c a t i o n " below as an umbrella term for a l l types of discontinuous dependencies including both movement and base-generation. Scrambling 17 - l e f t d i s l o c a t i o n from an argument p o s i t i o n -analysed as movement -case-marking occurs as for phrases i n s i t u -case-marker-drop not possible for scrambled objects -only argument DP's, PP's and CP's may be scrambled -clause-bound scrambling can be A-movement (See Saito (1994a:74)) T o p i c a l i z a t i o n - l e f t d i s l o c a t i o n from canonical p o s i t i o n of phrase -can be derived by e i t h e r movement or base-generation (see Saito (1985)) except for PP topics which cannot be base-generated - o b l i g a t o r i l y the leftmost element of the sentence 9 -marked by p a r t i c l e wa rather than by case-marker - t o p i c a l i z e d PP's and phrases headed by p a r t i c l e ni are t o p i c a l i z e d i n the form XP-ni-wa or XP-P-wa -DP's other than genitives (see (19) below), adverbs, obliques, and CP's can be t o p i c a l i z e d . V P - t o p i c a l i z a t i o n i s possible with l i g h t verb suru (See (13) below.) The following example shows that a VP can be t o p i c a l i z e d i f a l i g h t verb occurs a f t e r the t o p i c a l i z e d VP i n order to support the i n f l e c t i o n : (19)[hon-o yomi V P]-wa s i - n a - i read-top do-neg-infl As for reading books, I don't/he-she doesn't. "Contrastive" wa-marked phrases, which are not obligatorily the leftmost element of the sentence, are to be distinguished from true topics. Postposing 18 -always the rightmost element i n sentence -marked with case-marker or p a r t i c l e that occurs canonically - a l l DP's including genitives, adverbs, CP's, r e l a t i v e clauses can be postposed -A' movement (see below) The following examples suggest that postposing cannot be A-movement.10 F i r s t , consider the following examples from Saito (1994b:74) that show that a clause-internally-scrambled pronoun can A-bind a r e c i p r o c a l : (20) a. ?*otagai i-no sensei-ga karerai-o hihansita (koto) each-other-gen teacher-nom they-acc c r i t i c i z e d (fact) Each o t h e r ' S i teachers c r i t i c i z e d their^. b. Pkareraj-o otagaii~no sensei-ga tx hihansita (koto) they-acc each-other-gen teacher-nom c r i t i c i z e d (fact) Each o t h e r ' S i teachers c r i t i c i z e d theirs. c. ?*Masao-ga otagaii~no sensei-ni karerai-o syookai-sita (koto) -nom each-other-gen teacher-nom them-acc introduced (fact) Masao introduced them to each other's teachers. d. karerai-o Masao-ga otagaii-no sensei-ni syookai-sita (koto) them-acc -nom each-other-gen teacher-nom introduced (fact) Masao introduced them to each other's teachers. In (a) above, the r e c i p r o c a l i s not bound, which accounts for the deviance. °These examples were pointed out by Hisatsugu Kitahara (p.c). 19 In (b) the improved grammaticality can be explained i f we posi t that the scrambled pronoun "them", that c-commands the r e c i p r o c a l , can A-bind i t . S i m i l a r l y , i n (c) the r e c i p r o c a l i s not bound but i n (d), the scrambled pronoun appears to A-bind the r e c i p r o c a l which i t c-commands from i t s scrambled p o s i t i o n . The following sentences are i d e n t i c a l to the four above except that rather than scrambling the pronoun, we have postposed i t . We f i n d that no improvement r e s u l t s when the pronoun i s postposed. If we assume that precedence does not figur e i n binding, we can explain t h i s fact i f the postposed pronoun i n (b) and (d) above i s not i n an A-position, unlike the scrambled phrase i n the previous set of examples. (21) a. ?*otagaii-no sensei each-other-gen teacher Each other'Si teachers ga kareraj-o hihansita (koto) nom they-acc c r i t i c i z e d (fact) c r i t i c i z e d then^. b. ?*otagaii-no sensei each-other-gen teacher Each other'Si teachers ga t i hihansita -- karerai-o. nom c r i t i c i z e d -- they-acc . c r i t i c i z e d -- theirs. c. ?*Masao-ga otagaii-no sensei-ni karerai-o syookai-sita (koto) -nom each-other-gen teacher-nom them-acc introduced (fact) Masao introduced them to each other's teachers. d. ?*Masao-ga otagaii-no sensei-ni syookai-sita -- karerai-o. -nom each-other-gen teacher-nom introduced -- them-acc . Masao introduced to each other's teachers -- them. These data also suggest that i t i s not possible to scramble a phrase and then postpose i t . If i t were, we ought to be able to f i r s t scramble the pronoun 20 i n (c) and (d) above to an A-position from which i t can bind the r e c i p r o c a l , and then postpose i t to achieve the surface order. Comparison A s i g n i f i c a n t d i f f e r e n c e between postposing and the other two common types of d i s l o c a t i o n i s that a wider range of phrase types can be postposed than scrambled or t o p i c a l i z e d . I s h a l l not give a f u l l inventory of a l l the phrase types that can or cannot be t o p i c a l i z e d or scrambled, but I s h a l l i l l u s t r a t e below that genitives and r e l a t i v e clauses, both of which are phrase types that are important i n my analysis, can be postposed but not scrambled or t o p i c a l i z e d : (22) (a) t o p i c a l i z e d genitive i s ungrammatical •TaroOi-no-wa, Hanako-ga [ei hon-o N P] yonda. T.-gen-top H.-nom book-acc read As for T a r o's, Hanako read (his) book. (b) scrambled genitive i s ungrammatical *Taroo -nOj Hanako-ga [ej hon-o N P] yonda. T.-gen H.-nom book-acc read *Taro's, Hanako read (his) book. 1 1 This sentence can only receive the interpretation where the genitive modifies the subject, in which case it has not been scrambled. 21 (c) postposed genitive is grammatical Hanako-ga Oi hon-oNP] yonda, Taroo-nOi. H.-nom book-acc read T.-gen Hanako read (his) book -- Taro's. If we drop the genitive case-marker in (12), the sentence is s t i l l ungrammatical (23) *Tarooi-wa, Hanako-ga [ei hon-oNP] yonda. T.-top H.-nom book-acc read As for Taro's, Hanako read (his) book. (24) (a) topicalized relative clause is ungrammatical: * [CPTaroo-no-katta] i-wa, Hanako-ga [ei hon-oNP] yonda. T.-gen-bought-top H.-nom book-acc read *As for Taro bought , Hanako read (such a) book. (b) scrambled relative clause is ungrammatical * [CPTaroo-no-katta] i Hanako-ga [ei hon-oNP] yonda.12 T.-gen-bought H.-nom book-acc read *Taro bought , Hanako read (such a) book. l2This sentence can only have the interpretation where the relative clause modifies the subject: "Hanako, whom Taro bought, read a book." 22 (c) postposed r e l a t i v e clause i s grammatical: Hanako-ga [e £ hon-o N P] yonda, [ C PTaroo-no-katta] L H.-nom book-acc read T.-gen-bought Hanako read a book -- that Taro bought. 0.4.1 A comparison between postposing and other types of movement or d i s l o c a t i o n (a) Like t o p i c a l i z a t i o n , postposing seems to: 1. occur only i n an A' p o s i t i o n 2. have only one s i t e a v a i l a b l e (b) Whitman (1991b) argues that a postposed phrase may not contain new information but doesn't specify whether he means discourse-new or hearer-new information. This claim seems to be generally true but there may be q u a l i f i c a t i o n s . 1 3 (c) Postposing d i f f e r s from scrambling and t o p i c a l i z a t i o n i n that r e l a t i v e clauses and genitives can be postposed but not scrambled or t o p i c a l i z e d . (d) Like scrambled phrases, postposed phrases must be case marked as they would be i n t h e i r canonical p o s i t i o n . (e) Wa-marked topics including t o p i c a l i z e d VP's (See (25) below.) can be postposed. For example, ga-marked phrases can be legitimately postposed ((47) i n sec. 1.1.2.2 i s grammatical.) If a ga-marked phrase i s focused, then such sentences may be counterexamples to Whitman's claim that exhaustive l i s t i n g focused phrases cannot be postposed. I s h a l l not try to deal here with the question of whether a l l ga-marked phrases are focused. 23 25)si-nai, [hon-o yomi V P]-wa do-neg read-top He/she doesn't -- as far as reading books i s concerned 24 1. E v i d e n c e f o r a movement a n a l y s i s o f p o s t p o s e d p h r a s e s I n t h i s c h a p t e r I show t h a t p o s t p o s e d p h r a s e s show C o n d i t i o n C r e c o n s t r u c t i o n and a n t i - r e c o n s t r u c t i o n e f f e c t s . These phenomena a r e e a s i l y e x p l a i n e d i f p o s t p o s e d p h r a s e s undergo s y n t a c t i c movement. The d a t a I examine i n t h i s c h a p t e r a r e d i v i d e d i n t o two main s e c t i o n s : s e c t i o n 1.1 i n w h i c h I examine p o s t p o s i n g when t h e gapped p o s i t i o n i s c o n t a i n e d i n a m a t r i x c l a u s e , and s e c . 1.2, i n w h i c h t h e gapped p o s i t i o n i s c o n t a i n e d i n an embedded c l a u s e . I n each o f th e two s e c t i o n s I w i l l f o l l o w t h e same s t r a t e g y : t o l o o k a t p o s t p o s i n g from a v a r i e t y o f p o s i t i o n s : complement p o s i t i o n , s u b j e c t p o s i t i o n , t o p i c p o s i t i o n ( o n l y r e l e v a n t f o r m a t r i x c l a u s e s ) , and from d a t i v e s and o b l i q u e p o s i t i o n s . The f o l l o w i n g d i a g r a m shows t h e p o s t p o s i n g e n v i r o n m e n t s I s h a l l be l o o k i n g a t . (26) CP / \ TOPIC / \ I P C \ / \ VP I / \ SUBJECT \ / \ DATIVE/OBLIQUE \ / \ COMPLEMENT V 1.0.1 A l e f t w a r d movement a c c o u n t o f p o s t p o s i n g I f p o s t p o s i n g i s s y n t a c t i c movement, t h e n i t i s alw a y s movement t h a t 25 r e s u l t s i n rightward p o s i t i o n i n g of the postposed phrase. A number of d i f f e r e n t possible derivations, could lead to such a r e s u l t : rightward movement i s not the only possible option. For example, the postposed phrase could be leftward moved to a p r o j e c t i o n high i n the structure of the sentence and then the rest of the sentence could be moved l e f t to an even higher p o s i t i o n above the postposed phrase. The following diagram i l l u s t r a t e s how such a d e r i v a t i o n might work with leftward movement to s p e c i f i e r p o s i t i o n s : (27) (a) FP n F i ' FP, 26 IP / \ / \ XP (b) I F i ' I FP 2 / I IP / \ / \ 27 (c) / I / \ FP2 t i / | XPi F 2 ' I The phrase XP ends up to the right of IP, the rest of the sentence, as a result of two steps of leftward movement to the specifiers of FP2 and F P i . Although such a derivation should not, in principle , be ruled out, I shall assume the simpler derivation of rightward movement, after giving evidence in favour of syntactic movement of postposed phrases. 1.0.2 The nature of the gap When a phrase is postposed, i t i s construed with a gapped position in the main sentence. In the following examples, I shall try to determine the nature of this gap. I shall show in the examples below that this gap cannot be a pro. If i t were, we would not be able to explain the condition C reconstruction effects that are observed, since i t is structure inside the postposed phrase - -spec i f i ca l ly a name embedded in the phrase - - that experiences the observed binding violat ions . If the gap is a pro, we have no way of representing the structure inside the element that that pro represents. The relevant structure is as follows: 28 (28) IP / \ IP YP(postposed phrase)j / \ VP Y' / \ / \ pronoun; V name; Y' / \ / \ e j V The gap associated with the postposed phrase i s e^. Neither of the pronoun nor the name; can c-command the other. If ej i s pro, then there i s no way we can e s t a b l i s h a s t r u c t u r a l r e l a t i o n between pronoun ; and naraej to explain why they cannot be coindexed. I f . on the other hand, ej i s a trace of movement, a copy of YP survives t i l l LF i n the p o s i t i o n of e s under a reconstruction account. Pronoun; does c-command name; within t h i s copy, which explains the binding v i o l a t i o n . The foregoing i s the essence of the kind of reconstruction account I s h a l l present for the examples that w i l l follow. Those examples w i l l lead me to conclude that the gap i s a trace of movement, not a pro. 1.0.3 The nature of the postposed p o s i t i o n If we assume that postposed phrases are not derived by successive leftward movement, and that s p e c i f i e r p o s i t i o n s cannot occur on the r i g h t , then the postposed phrase must be i n either an adjoined or a complement p o s i t i o n . If i t i s i n a complement p o s i t i o n then the main sentence, which I s h a l l assume i s a CP, must be i n the s p e c i f i e r p o s i t i o n of a high functional p r o j e c t i o n that hosts the postposed phrase as i t s complement, as we see i n the structure below. 29 (29) XP / \ / X' / / \ CP X YP(postposed phrase) / _ \ main sentence This w i l l mean that there can be no c-command r e l a t i o n between the postposed phrase and the gap with which i t i s construed i n the main clause: only m-command. Although I do not see any way that the above configuration should be ruled out, I s h a l l instead adopt the hypothesis that the postposed phrase i s i n a p o s i t i o n right-adjoined to the highest p r o j e c t i o n of the sentence, as i l l u s t r a t e d below. This hypothesis w i l l make i t po s s i b l e for there to be a c-command r e l a t i o n between the postposed phrase and the gap with which i t i s construed. In addition, because i t i s the gap and not the postposed phrase that has a l o c a l semantic r e l a t i o n with some head, and because I know of' no evidence i n other languages of a d i s l o c a t e d element being found i n a complement p o s i t i o n , i t seems more p l a u s i b l e that the postposed phrase i s not i n a complement p o s i t i o n . (30) XP / \ XP postposed phrase / \ main sentence The postposed phrase always attaches to the r i g h t of the rightmost element i n the main sentence. Accordingly, I s h a l l assume that i t attaches to the highest 30 p r o j e c t i o n for which we have evidence of an overt head: ei t h e r IP or CP, depending on the sentence. 1.0.4 A reconstruction analysis of binding v i o l a t i o n s with postposing I s h a l l f i r s t present one example of an apparent binding theory v i o l a t i o n that occurs with a postposed phrase and show how i t can be explained as a Condition C reconstruction e f f e c t i f we assume movement of the postposed phrase. Then I s h a l l show how the same approach can be used with other examples of postposing to argue for a movement analysis. Consider again, (11), repeated here as (31): (31) (a) postposed object contains a name that i s coindexed w i t h a pronoun i n the main sentence: [ [ [kare, i / : j-ga e k semeta I P] -koto-ga C P] akiraka da I P] , [Tarooj-no tomodati-o] k. him.i/d-nom blamed- COMP-nom evident copula Ti.-gen friends-acc [[ PRONOUN-NOMi e k V I C]CP-nom A I I P] [NAME-GEN DP] k It i s evident that he A blamed them -- TarOi's f r i e n d s . (b) ( s t r u c t u r e o f O l a ) 31 IP / \ IP DP K ( p o s t p o s e d p h r a s e ) / \ \ CP I' D' I / \ I \ C AP I GENi D' / \ cop. IP C VP / \ PRONOUN-NOMi V' / \ e t V I n t h i s example, we need t o e x p l a i n why d e v i a n c e r e s u l t s when t h e pronoun s u b j e c t o f t h e embedded c l a u s e i s c o i n d e x e d w i t h t h e name embedded i n t h e p o s t p o s e d p h r a s e . C o n s i d e r t h e f o l l o w i n g f a c t s : 1. The g e n i t i v e m o d i f i e r GENi o f t h e p o s t p o s e d p h r a s e cannot c-command PRONOUN-NOMi o u t o f DPk. The f o l l o w i n g example (32) shows t h a t i n Japanese, a g e n i t i v e cannot b i n d a p h r a s e o u t s i d e o f t h e DP i t o c c u r s i n . I n (32), t h e s u b j e c t c-commands t h e name, but t h e g e n i t i v e embedded i n t h e s u b j e c t does not c-command t h e name out o f i t s p h r a s e , as we can see by t h e l a c k o f C o n d i t i o n C v i o l a t i o n 32 that occurs when the g e n i t i v e and name are coindexed. (32) k a r e ^ n o tomodati-ga Tarooj-o semeta hin^-gen fr iend(s) -nom r a c c blamed HiSi f r i e n d ( s ) blamed TarOi. 2. PRONOUN-NOMi cannot c-command the postposed phrase DPk or anything i n s i d e i t s ince i t i s embedded i n s i d e the CP matrix subjec t . I s h a l l assume that PRONOUN-NOM cannot have scrambled outside of the embedded c lause to a p o s i t i o n where i t might c-command the postposed phrase, f o l l o w i n g evidence i n Sai to (1985) that l o n g - d i s t a n c e scrambling of a subject i n Japanese i s r u l e d out . 3. If ek i s a pronominal ra ther than a t race , no b i n d i n g theory v i o l a t i o n should occur as a r e s u l t of a r e l a t i o n between PRONOUN-NOMi and ek s i n c e : a) the two elements are not coindexed b) a pronominal cannot represent s t ruc ture that occurs " i n s i d e i t " . s u c h as GENi Thus we cannot f i n d any r e l a t i o n between the p o s i t i o n s of the two coindexed phrases : the g e n i t i v e w i t h i n the postposed phrase and the embedded pronoun subject , that could e x p l a i n the apparent b i n d i n g v i o l a t i o n that occurs . We need, then, to consider the p o s s i b i l i t y that the postposed phrase has moved and that a copy of t h i s phrase i s l e f t behind at LF that undergoes a b i n d i n g v i o l a t i o n . The f o l l o w i n g diagram shows how, i f we assume, f o l l o w i n g Chomsky (1995), that a copy of the moved phrase i s l e f t behind i n the p o s i t i o n where i t o r i g i n a t e d , and that the copy survives i n the LF component, we can e x p l a i n the ungrammaticality of (31) when coindexing occurs as a case of C o n d i t i o n C r e c o n s t r u c t i o n . 33 (33) IP / \ IP DPk / \ \ CP I' GEN; D' / \ C AP I / \ copula IP C I' VP / \ PRONOUN-NOM; V / \ DPk V | \ GEN; D' DPk, the postposed phrase, i s shown above i n i t s c a n o n i c a l p o s i t i o n , i n s i d e embedded VP. I f , when i t i s postposed, a copy i s l e f t behind (as i n Chomsky(1995)'s copy theory of r e c o n s t r u c t i o n ) , PRONOUN-NOM; w i l l c-command the coindexed name GEN;, causing a v i o l a t i o n of Con d i t i o n C of the c l a s s i c a l b i n d i n g theory. (A name must be free everywhere.) As f u r t h e r evidence, consider the a n t i - r e c o n s t r u c t i o n e f f e c t observed below: . ' 34 (34)(a) [ [ [kare ? i / : i -ga semeta I P ]-koto-ga C P ] akiraka da I P] , [Taroo ; -o yonda tomodati-o] j . him, 1 / : j-nom blamed- COMP-nom evident copula T j . - a c c c a l l e d f r i e n d s - a c c It i s evident that he; blamed them - - the f r i e n d s that c a l l e d Taro ; . (b) (s t ructure of (34)(a) IP / \ IP \ / \ \ CPi I ' \ I / \ \ C AP I \ | copula \ IP \ I DPk I ' \ I D ' VP / \ / \ CP2 D' PRONOUN-NOMi V / \ / _ \ / \ NAME; DPk V In t h i s example, i f CP 2 , the r e l a t i v e clause modifying DPk i s adjoined to the s t r u c t u r e of DPk a f t e r i t has been postposed, CP2 w i l l not be part of the canonica l DPk and no C o n d i t i o n C v i o l a t i o n w i l l occur . This t y p i c a l " a n t i -r e c o n s t r u c t i o n " e f f e c t that occurs with r e l a t i v e clause embedding, w i l l e x p l a i n the fact that (34) i s only marginal ly deviant when the name i n s i d e the postposed phrase i s coindexed with the pronoun subject of the embedded c l a u s e . 35 We s h a l l see i n section (1.1.1.1) below that when the postposed phrase contains a complement clause rather than a r e l a t i v e clause that a name embedded i n such a clause i s subject to Condition C reconstruction, as we would pr e d i c t . 1.1 Postposing from matrix clauses I s h a l l now examine s i m i l a r cases of apparent binding theory v i o l a t i o n s that occur when a name inside a postposed phrase i s coindexed with a phrase i n the non-postposed part of the sentence that c-commands the canonical p o s i t i o n of the postposed phrase. 1.1.1 Postposing matrix DP complements When an object i s postposed from a matrix clause, a name contained i n i t cannot corefer with the matrix subject: (35) kare. i / ; j-ga semeta, Taroo ;-no tomodati-o. he.i/;j-nom blamed Ti-gen friends-acc He.i/:j blamed (them) -- T a r O i ' s friends. The deviance that occurs when there i s coreference i n t h i s sentence can be explained i n a s i m i l a r manner to our account of (31) above. We can ru l e out a l l p o s s i b i l i t i e s of d i r e c t binding between coindexed elements except for the p o s s i b i l i t y , here, that the postposed phrase occurs i n a p o s i t i o n low enough to be c-commanded by the matrix subject. However, other evidence makes reconstruction a more l i k e l y p o s s i b i l i t y . When we embed the name i n (35) i n a r e l a t i v e clause, the sentence su f f e r s only marginal deviance when there i s coreference, as we see i n (36) below: 36 (36) ?karei-ga semeta, [TaroOi-o yonda g a k u s e i ] - o . hei-nom blamed [ T . ; - a c c c a l l e d s tudents] -acc He; blamed (them)^-- [the students who c a l l e d T a r o ^ . The marginal grammaticali ty i n (36) versus the severe ungrammaticality when there i s coreference i n (31) can be explained as an a n t i - r e c o n s t r u c t i o n e f f e c t i n exac t ly the manner we d i d f o r (34) above. If d i r e c t b i n d i n g of the postposed phrase occurs , then we have no explanat ion for the improved grammatical i ty of (36). 1 .1 .1 .1 Postposing matrix CP complements In (37) below, a name coreferent with a pronoun i n the main sentence i s embedded i n a complement ra ther than a r e l a t i v e c l a u s e . (37) *karei-ga h i t e i - s i t a , [TaroOi-ga okureta k o t o ] - o . hei-nom denied T.—nom was-late C-acc he; denied (it.,) - - [the fact that Taro; was late] . , . (37) was judged to be worse than (36). This i s what we might expect i f r e c o n s t r u c t i o n i s occurr ing since embedding i n a r e l a t i v e clause w i l l t y p i c a l l y produce a n t i - r e c o n s t r u c t i o n e f f e c t s whereas embedding i n a complement clause w i l l not . T h i s contrast i s expla inable by the a n a l y s i s of r e c o n s t r u c t i o n of Chomsky (1995) . A complement c lause , not being an adjunct , cannot be added to the s t r u c t u r e a f t e r a phrase i t has occurred i n has moved. In the next example, a CP headed by complementizer to i s postposed. We s h a l l see again that apparent b inding theory v i o l a t i o n s occur that can be 37 explained by C o n d i t i o n C r e c o n s t r u c t i o n . 1 (38) *Sanae wa, k a r e ^ n i e, i t t a , [Yootyani ga r i p p a da t o ] j . Sanae TOP him^ DAT s a i d Yochani-nom handsome i s COMP Sanae s a i d to hirr^ - - that Yochani i s handsome - - . If the clause i s postposed by movement, ' i t w i l l be reconstructed to a p o s i t i o n where i t i s c-commanded by the coindexed pronoun, which w i l l cause a C o n d i t i o n C v i o l a t i o n . On the other hand, i f the postposed phrase i s base-generated, we can only e x p l a i n the ungrammaticality of (38) i f the postposed phrase i s attached lower than the i n d i r e c t object " h i m - d a t i v e " . Such a c o n f i g u r a t i o n might be p o s s i b l e under the f o l l o w i n g circumstances . If the verb "say" occurs i n a s h e l l s t r u c t u r e 1 5 as shown below, with i t s CP complement o r i g i n a t i n g as s i s t e r to the verb, the i n d i r e c t object i n s p e c i f i e r of that VP, and an empty verb taking that VP as i t s complement, then i f head movement of the verb "say" i s delayed u n t i l LF, the postposed phrase could a d j o i n to V of the lower VP and be c-commanded by the i n d i r e c t o b j e c t . The complementizer to introduces clausal complements of certain verbs in Japanese -- for example "to say", "to think" or "to hear". Clauses headed by to differ from complement clauses headed by koto or no in several ways: for example, long-distance scrambling can freely occur out of to clauses but marginally out of koto or no clauses. (See Saito (1994) for examples of scrambling out of koto clauses.) (ii) to clauses are never case-marked but koto and no clauses are always case-marked. The difference between the two types of clauses will be significant in chapters 2, 3 and 5r, where we shall see that they exhibit slightly different locality constraints with respect to postposing. 'See Larson (1988), Hale and Keyser (1993), Chomsky (1993). 38 (39) VP V / \ VP V / I e i . o . V I \ V postposed phrase / \ CP V say We could f o r c e the postposed phrase to a d j o i n i n a higher p o s i t i o n by embedding the matrix clause i n a no-da c o n s t r u c t i o n , which would appear to have the s t r u c t u r e i n (40). In t h i s c o n s t r u c t i o n , the matrix clause i s followed by complementizer no, which i s , i n tur n , followed by the copula. (See sec. 1.2.1.2 f o r f u r t h e r examples of t h i s c o n s t r u c t i o n . ) (40) V I \ CP V I da(copula) C / \ IP C /_\ no matrix c l . I f we t e s t (38) embedded t h i s way, we s t i l l get an apparent C o n d i t i o n C r e c o n s t r u c t i o n e f f e c t . Here, the e f f e c t cannot be due to d i r e c t b i n d i n g of the 39 postposed phrase, since the coindexed dative pronoun cannot c-command the postposed phrase. (41) *Sanae wa, k a r e ^ n i e., itta-n-da, [Yootyani-ga rippa da to]-. Sanae TOP hin^-dat said-C-V Yochan;-nom handsome i s comp The fact i s that Sanae said to hirr^ that he± i s handsome Yocharii. (42) *Sanae wa, karej-ni kood i t t a , [Yootyani ga rippa da to]6. Sanae TOP hin^ DAT this-wayj said Yochani-nom handsome i s COMP Sanae sa i d to hiir^ -- that Yochanj i s handsome --In (42) a resumptive pro-form for the postposed CP i s ins e r t e d i n place of the gap where the postposed clause should be. (42) i s grammatical when there i s no coreference. The fact that a Condition C reconstruction e f f e c t seems to occur i n (42) suggests that the resumptive pronoun "t h i s way" acts l i k e a v a r i a b l e and i s i n an A' r e l a t i o n with the postposed phrase, with the postposed phrase being reconstructed into the place of the v a r i a b l e at LF. This example i s important in the l i g h t of the general question about the nature of the gap i n a sentence from which a phrase i s postposed. We have needed to consider the p o s s i b i l i t y that t h i s gap i s pronominal element rather than a trace of movement. The example above shows that when we make the gap an overt pronominal element, i t shows an A'-rather than an A - r e l a t i o n with the postposed phrase. In t h i s section I s h a l l examine the e f f e c t s that occur on a name embedded i n a postposed phrase when the postposing occurs i n two d i f f e r e n t environments: 1.1.2 Postposing matrix subjects 1. when the canonical p o s i t i o n of the postposed phrase i s c-commanded by a topic phrase 40 2. when the canonical p o s i t i o n of the postposed phrase i s c-commanded by a scrambled object 1.1.2.1 Topic as c-commander Since v/a-marked topics o b l i g a t o r i l y occur as the leftmost element of the sentence they must c-command the rest of the sentence. Consequently, a topic coreferent with a name embedded i n a postposed subject can be used to test for reconstruction of the subject: (43) *karei-wa, suki na no, T a r o O i - n o tomodati-ga. hej-top l i k e a b l e cop comp -gen friend(s)-nom *As for him i ( he l i k e s (them) -- Taro;'s friend(s) In (43) , the presence of the complementizer no i n the matrix clause should force the postposed phrase to adjoin to a p r o j e c t i o n at least as high as CP. If the topic i s base-generated i n [ s p e c , CP], the question of whether i t w i l l c-command an element adjoined to CP w i l l depend on one's d e f i n i t i o n of c-command. (44) CP / \ CP postposed phrase / \ topic C / \ C \ no If, for X to c-command Y, i t i s required that every p r o j e c t i o n that dominates X dominates Y, then i n the above case, the topic does not c-command the postposed 41 phrase, since CP dominates the topic but only one segment of CP, not the whole projection, dominates the postposed phrase. Under such an assumption, the topic i n (43) w i l l not bind the postposed phrase d i r e c t l y and the Condition C v i o l a t i o n must be explained by reconstruction of the postposed phrase. 1 6 Once again, embedding the name i n a r e l a t i v e clause improves the grammaticality of the sentence, suggesting the occurrence of anti-reconstruction e f f e c t s . (45) ?*karei-wa, suki na no, T a r o O i - o yonda tomodati-ga. he^-top l i k e a b l e cop comp -acc c a l l e d friend(s) *As for hirr^, he l i k e s (them) -- the friends who c a l l e d T a r O i . 1.1.2.2 Scrambled object as c-commander Another way of t e s t i n g the binding of a reconstructed subject might be to use a scrambled object as a c-commander for the reconstructed subject, with a fl o a t e d q u a n t i f i e r marking the s i t e of the postposed subject. 1 7 However, there i s a problem with t h i s test i f we assume, following Saito (1992) that clause-i n t e r n a l scrambling can be A' movement. ieThere is still the possibility that the topic has its own functional projection that is above CP: TopP / \ topic CP / \ CP postposed phrase If this is the case, then the topic will directly bind the postposed phrase. However, the anti-reconstruction effects we see in the next example cannot be explained by the above structure. See Saito (1985:212) for evidence that quantifiers such as "three people" cannot be moved themselves from their canonical position in the sentence but can mark the position of the trace of a moved DP with which they are associated. 42 F i r s t , notice that i n a canonical sentence, a q u a n t i f i e r construed with a subject can be stranded when the subject i s postposed: (46) [e k san-nin] kare i / : J-o semeta, [TaroOi-no tomodati-ga] k. three-people he i / : i-acc blamed i-gen friends-nom The three (of themk) blamed himi/;j -- [TarO;'s friends] k We also f i n d that the subject can s t i l l be postposed when the object i s scrambled to the l e f t of .the sub j e c t - q u a n t i f i e r . (47) karei-o e., san-nin semeta, [Hanako-no tomodati-ga]j hei-acc three-people blamed -gen friends-nom Three (of themj) blamed hin^ -- Hanako's (three) friends If the scrambled object i s i n an A-position, then we should expect to f i n d a Condition C reconstruction v i o l a t i o n when there i s coreference between the pronoun and the name. We find, however, that for some speakers, the following sentence i s not bad. (48) ??karei-o san-nin t i semeta, [TaroOi-no tomodati-ga] d hei-acc three-people blamed -gen friends-nom Three (of their^) blamed himi -- TarOi's (three) friends This fact could be explained i f we adopt Saito (1992)'s hypothesis that the scrambled pronoun can be i n an A' p o s i t i o n . The following diagram shows the presumed LF structure of (48) when copies of the postposed phrase and of the scrambled pronoun are l e f t behind. 43 (49) CP / \ CP Opj[ DP] / IP / \ OPi I' / \ VP I / \ DP \ / \ \ [Taro ' S i friends x d] V If the scrambled pronoun "him" i s i n an A' p o s i t i o n , no Condition C v i o l a t i o n w i l l r e s u l t , since a name must be A-free by Condition C. The s l i g h t deviance that t h i s structure creates could be explained as a weak crossover v i o l a t i o n . The pronoun "him" A'-binds two posit i o n s that can be A' bound: the name Taro ; and the v a r i a b l e x i # neither of which c-commands the other. We get a s i m i l a r e f f e c t when the postposed phrase i s i n s i t u , as we see in (50) below. (50)??karei-o Taroo-no tomodati-ga semeta. hin^-acc i _gen f riend (s)-nom blamed Him 1 ( Taro's friends blamed. / \ [him Xi] V 44 A weak crossover explanation of the s l i g h t deviance of (48) i s not completely straightforward since a number of assumptions about weak crossover would have to be made to explain the r e s u l t s . Is weak crossover s o l e l y an LF phenomenon? Does i t occur d e r i v a t i o n a l l y or representationally? Answers to these questions w i l l bear on a weak crossover analysis of examples l i k e (48) . In addition, the fact that examples l i k e these are only s l i g h t l y deviant for some speakers makes i t possible that some other phenomenon i s causing the deviance. If we substitute for the q u a n t i f i e r san-nin a case-marked q u a n t i f i e r such as san-nin-de, the sentence becomes grammatical. This would be explained by the fact that a qua n t i f i e d DP/PP such as " three-people-inst" i n (51) below acts d i f f e r e n t l y than a bare q u a n t i f i e r l i k e "three-people" and cannot undergo q u a n t i f i e r f l o a t , being a separate DP or PP. Since i t cannot undergo Q-float l i k e "three-people", i t cannot "anchor" the p o s i t i o n of an associated DP. Thus (51) i s grammatical because the subject could have o r i g i n a t e d to the l e f t of the object. (51) kare^o san-nin-de semeta, [TaroOi-no tomodati-ga] j he^acc three-people-inst blamed -gen friends-nom Three (of t h e r O j ) blamed hin^ -- TarOi's (three) friends 1.1.3 Postposing matrix topics Data i n Endo (1989) show that topics are commonly postposed i n Japanese -- i n fact they are one o f the most common types o f elements to be postposed. It i s d i f f i c u l t to test f o r movement of a postposed topic by binding of a reconstructed phrase since a wa-marked topic when i t i s i n canonical p o s i t i o n i s o b l i g a t o r i l y the leftmost (and therefore highest) element i n the sentence: nothing i n the main sentence can c-command the s i t e of a postposed t o p i c . 45 Suppose that we were to postpose a topic phrase. Any other phrase that remains i n the non-postposed part of the sentence w i l l occur to the right of the empty category that w i l l occur i n the canonical p o s i t i o n of the to p i c : (52) eL XP V I p o s t p o s e d - t o p i C i The only way XP can c-command ei i s i f XP occurs to the r i g h t of a pr o j e c t i o n that dominates ei: (53) YP • / \ CP XP / \ ei C / \ / \ But the only time t h i s configuration occurs i n Japanese i s when XP i s i t s e l f the postposed phrase. Thus the only other possible configuration for an XP i n the sentence that i s to the r i g h t of the canonical p o s i t i o n of a postposed topic i s as follows: 46 (54) CP / \ CP postposed topiCi / \ C / \ / \ XP In t h i s configuration XP c l e a r l y does not c-command ei so we are unable to use our reconstruction diagnostic. These elements canonically occur to the ri g h t of the subject. When they are postposed, they can be tested for movement i n the same way that we d i d with postposed objects i n sec. 1.1.1 above. When a name i s embedded i n a postposed dative or oblique we f i n d that the name cannot be coindexed with the subject, suggesting a Condition C reconstruction e f f e c t . The (a) sentences below give the canonical order of elements i n the same sentence with no postposing. (55) (a) karei-ga [Taroo„i/;j-no tomodati-ni] k puresento-o ageta. 1.1.4 Postposing matrix datives and obliques he-nom -gen friend(s)-dat present-acc gave Hei gave a present to TarOi' s f riend (s). (55) (b) karei-ga e k puresento-o ageta, [Taroo. i / 3-no tomodati-ni] k he-nom present-acc gave -gen friend(s)-dat hei gave a present -- to TarOi's f r i e n d ( s ) . 47 (56) (a) kare^ga zutto [Taroo„i/;i-no ie-made] a r u i t e - k i t a . he-nom all-the-way -gen house-as-far-as walk-went he A walked a l l the way as far as TarOi's house. (56)(b)kare^-ga zutto a r u i t e - k i t a , Taroo.i/:i-no ie-made. he-nom all-the-way walk-went -gen house-as-far-as hei walked a l l the way -- as far as Taro^s house. 1.1.5 Postposing matrix genitives In t h i s section I w i l l test postposing of a genitive no-marked phrase from subject and object environments, as shown i n the following structure: (57) CP . . . [ p o s t p o s e d phrase] I IP I VP / \ (subject)DP \ | \ D' \ / ' \ V GENITIVE D' / \ (object)DP V I D' / \ GENITIVE D' 48 In (58) a genitive associated with an object i s postposed. The sentence i s good without coreference and bad with coreference. (58) *karei-ga tomodati-o semeta, TaroOj-no hej-nom friends-acc blamed T. rgen He blamed h i s friends -- Taro's (friends) Once again we must determine whether the sentence i s bad because of d i r e c t binding of the postposed phrase, i n t h i s case, by the subject. Compare (58) to a sentence i n which the name i s more deeply embedded insi d e the g e n i t i v e : i . e . insi d e a r e l a t i v e clause. (59) kare^ga tomodati-o semeta, TaroOi-o yonda gakusei-no hei-nom friends-acc blamed T. (-acc c a l l e d students-gen hei blamed (their.,) friends -- [the students who c a l l e d T a r o J / s (friends) Example (59) was judged to be much better than (58), i n fact, almost f u l l y grammatical. The contrast between the two examples again suggests a n t i -reconstruction, supporting the hypothesis that there i s movement of the genitive phrase. In (60) below, the postposed genitive i s associated with the subject. (60) tomodati-ga karei-o semeta, Taroo ;-no friends-nom hei-acc blamed T.—gen His friends blamed him -- Taro's (friends) (60) was judged grammatical for speakers who associate a postposed genitive 49 h i e r a r c h i c a l l y 1 8 , that i s , with the highest possible phrase i n the sentence of which the ge n i t i v e could be a modifier. It suggests that the postposed phrase i s not attached low enough to be c-commanded by the object. 1.1.6 Postposing matrix adjuncts Because the canonical p o s i t i o n of an adjunct that has c l a u s a l material so that i t can contain a name can be d i f f i c u l t to determine i n a matrix sentence, i t i s d i f f i c u l t to test for Condition C reconstruction e f f e c t s with postposed matrix adjuncts. Accordingly, i n t h i s chapter, I s h a l l only look at postposing of embedded adjuncts. If an adjunct i s embedded i n a complement CP, i t s p o s i t i o n w i l l be c-commanded by a pronoun subject of the matrix clause and possible Condition C reconstruction e f f e c t s can be tested i n the same manner as we d i d i n e a r l i e r subsections. See sec. 1.2.4 for a discussion of postposing embedded adjuncts. 1.1.7 Summary of r e s u l t s of postposing i n matrix environments In environments where we are able to test postposed phrases. i n matrix environments for Condition C reconstruction e f f e c t s , we f i n d that these e f f e c t s occur. In some cases i t i s d i f f i c u l t to t e l l whether these e f f e c t s are due to d i r e c t binding of a name embedded i n the postposed phrase by a coindexed element i n the main part of the sentence. In the next section, i n which I look at postposing i n embedded environments, i t w i l l be easier to make sure that no such d i r e c t binding can occur, since the p o t e n t i a l antecedent of the name i n the postposed phrase can be embedded i n a p o s i t i o n from which i t cannot c-command the postposed phrase. For speakers who associate a postposed genitive hierarchically, only the highest available DP in the sentence can host the trace of the postposed phrase; other speakers seem to associate a genitive linearly: for them, only the rightmost DP in the sentence can be associated with the genitive. For these speakers, this example is bad. 50 1.2 Postposing from embedded clauses 0 s. When we examined phrases postposed out of matrix clauses i n the previous section, we faced the problem of determining whether apparent Condition C v i o l a t i o n s were a r e s u l t of d i r e c t binding of a low-attached postposed phrase or by reconstruction of the postposed phrase. When the postposed phrase originates i n an embedded clause, d i r e c t binding of the postposed phrase by a coindexed antecedent i s no longer possible since we can place the antecedent i n the embedded clause where i t w i l l c-command the s i t e the postposed phrase i s construed with but not the postposed phrase (see diagram below). We s h a l l see i n t h i s subsection that Condition C e f f e c t s also occur i n embedded structures, suggesting movement. If movement occurs i n these sentences then movement should also be possible i n matrix sentences. (61) CP / \ CP VP I V / \ CP V postposed phrasei / \ / __\ DP(name)i pronoun-j / \ \ / \ / \ -no c-command r e l a t i o n between DPj and pronoun.. t< 51 In t h i s subsection I look at the e f f e c t s of postposing d i f f e r e n t kinds of elements out of d i f f e r e n t kinds of islands. For each type of postposed element being considered, I have chosen types of embedding that most e a s i l y allow the occurrence of a pronoun that (a) can be coindexed with a name that i s embedded in the postposed phrase; (b) c-commands the canonical p o s i t i o n of the postposed phrase; and (c) occurs low enough i n the structure that i t could not d i r e c t l y bind the postposed phrase. In order that these conditions can be met, not every possible type of i s l a n d has been considered for each type of postposed element. For a more thorough elaboration and t e s t i n g of d i f f e r e n t types of islands, see chapter 3, " L o c a l i t y E f f e c t s . " 1.2.1 Postposed embedded DP complements 1.2.1.1 Postposing out of koto or no clauses With the exception of CP's headed by non-case-marked complementizer to (see subsection 1.1.5) complement clauses i n Japanese are usually headed by case-marked complementizers koto or no. I follow Whitman (1991a) i n considering these elements to be C s rather than N's. The following examples i n which a phrase i s postposed out of a koto or no clause, are repeated from sec. 0.1: (62) [ [ [kare. i / :j -ga e k seme-ta I P]-koto/no-ga C P] akiraka da I P] , [TaroOi-no tomodati-o] k. hirrii-nom blamed- COMP-nom evident copula Ti.-gen friends-acc It i s evident that h e i blamed them -- T a r O i' s f r i e n d s . 52 (63) [ [ [kare ? i / j-ga e-j seme-ta I P] -koto/no-ga C P] akiraka da I P] , [Taroo £-o yonda tomodati-o] j . hirrii-nom blamed- COMP-nom evident copula Ti.-acc c a l l e d friends-acc It i s evident that he t blamed them -- the friends that c a l l e d TarOi. Recall the discussion of these examples i n sec. 0.1, which showed that the i m p o s s i b i l i t y of coreference i n (62) and the marginality of coreference i n (63) are explainable as Condition C reconstruction and ant i - r e c o n s t r u c t i o n e f f e c t s r e s p e c t i v e l y . 1.2.1.2 postposing out of no-da clauses In t h i s construction, which i s very common, a matrix clause i s embedded in a CP headed by complementizer no, and the CP i n turn i s embedded i n an IP with i n f l e c t i o n provided by the copula. (64) *[[[karei-ga seme-taIP]-nc,,]-daIP] , [ T a r o O i-no tomodati-o] j . hirt^-acc. blamed- COMP -copula Tt.-gen friends-nom The fact i s that they blamed him; -- Taro's frie n d s . (65) T f t t k a r e — g a e^  semeta I P]-n C P]-da I P] , [Taroo ;-o yonda gakusei ] -o^. hei-nom e s blamed- COMP -copula [T.i-acc c a l l e d students] - a c C j The fact i s that het blamed (them)^-- [the students who c a l l e d TarOjJj. Since the postposed phrase i n (64) and (65) i s adjoined to the r i g h t of the comp-copula construction i n which the clause i n which i t canonically occurs i s embedded, i t cannot be adjoined lower than the subject kare unless kare has for some reason been r a i s e d to subject p o s i t i o n of the highest clause. (See diagram below.) Again, coreference i s impossible when the name i s shallowly embedded and marginal when i t i s deeply embedded i n the postposed phrase. 53 CP / \ CP DPi (postposed) I / \ IP / \ | DP VP Taro s V / \ CP C C no V da (copula) -pronoun kare cannot c-command the name Taro unless the pronoun has raised, contrary to evidence i n Saito (1985) against r a i s i n g of subjects \ V / \ V 54 1.2.2 Postposing embedded subjects (67) *Sanae wa, karei n i [e^ rippa da to] i t t a , Yootyani ga. Sanae TOP hin^ DAT handsome i s COMP said, Yochani NOM Sanae said to hin^ that hej i s handsome -- Yochani. In (67), the pronoun kare w i l l only d i r e c t l y bind the postposed phrase i f the l a t t e r i s adjoined low i n the sentence -- for example to matrix V , as shown in (68) below 1 9: In such a case, there i s a p o s s i b i l i t y of d i r e c t binding of the pronoun by the postposed phrase. CP / \ .. DP IP Sanae-wa | (topic) VP V / \ VP DP / \ Yootyan-gai (postposed phrase) DP V kare-nii / \ (dat.) CP V / \ / \ The VP shell in the sructure below follows Larson (1988), Hale and Keyser (1993) and Chomsky (1993). 55 To control for any s t r u c t u r a l ambiguity i n (68) above, i n (69) the sentence i s embedded i n a no-da construction. Here, the pronoun cannot c-command the postposed phrase unless i t ra i s e s out of the embedded clause, which would be contrary to evidence of scrambling of subjects i n Saito (1985) . 2 0 (69) *Sanae wa, karej n i [ei rippa da to] itta-n-da, Yootyani ga. Sanae TOP hirr^ DAT handsome i s COMP said-comp-copula, Yochani NOM Sanae said to himi that hei i s handsome -- Yochani-20There is still the possibility that subjects cannot raise overtly (in accord with Saito (1985)) but can raise at LF; however, there is no evidence of anything that would motivate covert raising of a subject out of an embedded clause. 56 structure of (69) CP / \ CP DPi (postposed) | Yootyan-gai IP I VP I V / \ CP V I da (copula) C / \ IP c I VP I V / \ kare-nij V / \ CP V / \ / \ t i In t h i s example we get a Condition C reconstruction e f f e c t , suggesting that the postposed phrase has moved from p o s i t i o n t± above. 57 1.2.3 Postposing embedded genitives 1.2.3.1 Genitive postposed out of r e l a t i v e clause or a koto clause Example (70) shows that a genitive can be postposed out of a r e l a t i v e clause: (70) [Hanako-ni [ e i omotya-o] watasita kodomo-ga] nigeta, Tarooj-no. H.-dat toy-acc handed-over child-nom f l e d T.-gen The c h i l d who handed over the toy to Hanako ran away -- Taro's (toy) In (71) a genitive i s postposed out of a r e l a t i v e clause and i t s canonical p o s i t i o n inside the clause i s c-commanded by a pronoun inside the clause -- i n t h i s case a dative. The i m p o s s i b i l i t y of coreference between the name i n the postposed g e n i t i v e and the pronoun "him (dative)" suggests a Condition C e f f e c t . Note that the pronoun, because i t i s embedded i n the r e l a t i v e clause, cannot d i r e c t l y bind the postposed phrase. In (72) when the name i n the postposed phrase i s embedded more "deeply" i n a r e l a t i v e clause, we get t y p i c a l a n t i -reconstruction e f f e c t s . (71) *[karei~ni [ej omotya-o] watasita onna-no-ko-ga] nigeta, TaroOi-no hirtii-dat toy-acc handed-over girl-nom f l e d T.-gen The g i r l who handed him the toy ran away -- Taro's (toy) 58 (72) [karei-ni [ed omotya-o] watasita onna-no-ko-ga] nigeta, [TaroOi-ga t u r e t e - k i t a inu-no].j hin^-dat toy-acc handed-over girl-nom f l e d T.i-nora brought dog-gen The g i r l who handed hiir^ the toy ran away -- the [dog that TarOi brought]'s (toy). Examples (73) and (74) are s i m i l a r to (70) - (72) except that the postposed phrase canonically occurs inside a koto clause instead of a r e l a t i v e clause. Again, we see the same contrast: ungrammaticality when the name i s coindexed with a pronoun that c-commands the s i t e of the postposed phrase and grammaticality when the name i s embedded more deeply within the postposed phrase. (73) *Hanako-wa [kare^ga [ej omotya-o] kowasita koto-o] s i t t e - i r u , Taroo-no ;. H.-top he-nom toy-acc broke comp-acc knows T.-gen Hanako knows that hei broke the toy -- TaroOi's. (74) Hanako-wa [karei-ga [e^ omotya-o] kowasita koto-o] s i t t e - i r u , [TaroOi-o karakatta kodomo-no]d. H.-top he-nom toy-acc broke comp-acc knows T.-acc teased child(ren)-gen Hanako knows that heA broke the toy -- the child(ren) who teased TaroOi' s. These contrasts suggest that a condition C v i o l a t i o n occurs i n (71) and (73) by reconstruction of the postposed genitive back to the r e l a t i v e clause. The improvement i n grammaticality i n (72), i n which the name i s embedded within another r e l a t i v e clause again i s evidence for the reconstruction hypothesis. Notice also that we cannot explain the ungrammaticality of (71) by d i r e c t binding of the postposed genitive by the coindexed pronoun, since the pronoun A:are i s embedded inside the r e l a t i v e clause from which i t cannot c-command anything 59 outside the clause. The fact that genitives can be postposed at a l l i s somewhat surpri s i n g , given the fact that they cannot be moved leftwards under scrambling and cannot be t o p i c a l i z e d (See examples i n chapter 0.) Genitive phrases are discussed further i n Appendix B: The nature of no-marked phrases. 1.2.4 Postposing embedded adjuncts Endo (1989) has many examples of postposing adjuncts out of matrix clauses that were taken from taped conversations and were judged to be grammatical. In the following examples, the extraction s i t e of a postposed adjunct i s embedded. We see that i t i s possible to postpose an adjunct out of an object CP (75), a subject CP 2 1 (76), an adverbial clause (77), l e x i c a l DP (78), v o l i t i o n a l clause (79), and a concessive clause (80). In (80) the sentence i s marginal but not completely bad. (75) [[sono hoo-ga e £ t i k a i ] I P koto-o] C P wasureta, [ k o t t i kara iku y o r i ] i that (comparative)-nom near fact think here from go than I forgot that i t ' s closer -- than going from here. (76) [[sono hoo-ga i k i - y a s u i ] I P koto-ga] C P aru, k o t t i kara iku y o r i . that (compar.)-nom go-easy fact-nom be here from go than There are times when i t ' s easier to go -- than going from here In this sentence the verb of existence of which nominative case-marked complementizer koto is the subject, is likely unaccusative. See sec. 2.7. "Postposing out of koto clauses that are the subject of a transitive verb" for examples of island effects that do appear when complementizer koto is subject of a transitive verb. 60 (77) [[sono hoo-ga ei t i k a i ] I P no-ni] C P, ikanakatta, [ k o t t i kara iku y o r i l j that (comparative)-nom near comp went-neg. here from go than *Even though i t ' s closer, I didn't go than going from here. (78) [[[[sono hoo-ga e± t i k a i ] I P t o ] C P yuu] I P hanasi-o] N P, s i t e - i t a [ k o t t i kara iku y o r i h that (compar. ) -nom near comp say story-acc d i d here from go than He said ( t o l d a story) that i t ' s closer, -- than going from here. (79) [ei mondai-o tokoo t o ] C P omotta, [Taroo-no yarikata-de] ± problem-acc s o l v e - v o l i t . C thought -gen do-way-inst. I t r i e d to solve the problem -- i n Taro's way. (80) ??karei-wa mondai-o tokoo to s i t a no-ni dame datta, TaroOj-no yarikata-de he-top problem-acc s o l v e - v o l i t . C did i n - s p i t e - o f no-good was, -gen do-way-inst. In s p i t e of the fact that hei t r i e d to solve the problem e.,, i t didn't work -- [in Taroj's way]j. In the following sentences, a manner adjunct containing a name i s postposed out of a v o l i t i o n a l clause (81) and a concessive clause (82). We f i n d that the name cannot be coindexed with a pronoun i n the matrix clause that c-commands the canonical p o s i t i o n of the adjunct phrase. This again suggests reconstruction and therefore movement of the postposed adjunct. 61 (81) *karei-wa/ga [ej mondai-o tokoo t o ] C P omotta, [TaroOi-no yarikata-de]j he-top/nom problem-acc s o l v e - v o l i t . C thought -gen do-way-inst. Hej t r i e d to solve the problem -- i n Taro^s way. (82) * karei-wa/ga mondai-o tokoo to s i t a no-ni dame datta, Tarooj-no yarikata-de he-top/nom problem-acc s o l v e - v o l i t . C d i d i n - s p i t e - o f no-good was, -gen do-way-inst. In s p i t e of the fact that he £ t r i e d to solve the problem e,, i t didn't work --[in TarOj's way]j. 1.2.5 Summary of r e s u l t s of postposing i n embedded environments In summary, to the extent that we are able to test postposed phrases for Condition C reconstruction e f f e c t s , we see evidence for movement of a l l types of postposed phrases from a l l types of environments. 1.3 Summary Although the evidence i n favour of movement of postposed phrases i n Japanese i s not conclusive, the fact that reconstruction and anti - r e c o n s t r u c t i o n e f f e c t s are observed whenever i t i s possible to test for them makes movement the most natural hypothesis for explaining these e f f e c t s . A movement hypothesis i s further strengthened by the evidence for subject/object asymmetries (subject islands are worse b a r r i e r s for postposing than object islands) that we s h a l l see i n the next chapter. These e f f e c t s are also most e a s i l y explained by a movement account. There i s also the p o s s i b i l i t y that postposing may be derived by eit h e r 62 movement or base-generation, depending on the type and category of the postposed element. Although such a hypothesis i s not ruled out by the data I present here, I s h a l l adopt, for now, the simpler hypothesis of movement i n a l l cases, i n the absence of evidence that requires a base-generation hypothesis. 63 2. L o c a l i t y Constraint Patterns I -- Subject-object asymmetries In t h i s chapter I examine the f i r s t of two l o c a l i t y constraint phenomena that are evident i n data on postposing: subject-object asymmetries, and r e l a t i v i z a t i o n with respect to a [+N] feature. The examples that are presented below show that when a phrase i s postposed out of a subject, greater deviance r e s u l t s than i f i t i s postposed out of an object -- i n fact a subject appears to be a more severe i s l a n d for extraction than wh-islands, adverbial clauses, and r e l a t i v e clauses. We s h a l l see that i t i s postposing out of a subject that induces a strong v i o l a t i o n ; postposing of. a subject does not induce a s u b s t a n t i a l l y stronger v i o l a t i o n than postposing of an object. These r e s u l t s are s i g n i f i c a n t i n l i g h t of the fact that Lasnik and Saito (1992) f i n d that Japanese does not show evidence of the Subject Condition. In the following examples, I w i l l examine how postposing DP adjuncts such as r e l a t i v e clauses or genitives displays subject/object asymmetries. When a phrase i s postposed out of eit h e r a subject or an object i n eit h e r case some deviance r e s u l t s ; however, the degree of v i o l a t i o n i s c o n s i s t e n t l y worse for a subject i s l a n d . In each section below, a phrase i s postposed out of a subject or object that i s embedded in a weak i s l a n d such as a Jcoto clause, r e l a t i v e clause, concessive clause, or wh-island. We s h a l l see that only mild deviance r e s u l t s when a phrase i s postposed from these kinds of is l a n d s . A subject, on the other hand w i l l be seen to be a more severe i s l a n d for postposing. The following i s a summary of the kinds of environments I s h a l l be looking at: 2.1 Postposing a r e l a t i v e clause out of a non-matrix CP 64 2.2 Postposing a r e l a t i v e clause out of a wh-island: 2.3 Postposing a genitive out of a DP inside an adverbial clause: 2.4 Postposing a r e l a t i v e clause out of an adverbial clause: 2.5 Postposing a dative out of a r e l a t i v e clause: 2.6 Postposing a r e l a t i v e clause out of a r e l a t i v e clause: 2.7 Postposing out of koto clauses that are subjects of t r a n s i t i v e verbs 2.8 Postposing a r e l a t i v e clause out of a matrix subject In each of the above environments, a r e l a t i v e clause or genitive w i l l be postposed, stranding the nominal head that i t modifies. There i s a consistent contrast i n grammaticality that depends on whether the stranded nominal head i s a subject or an object. In a l l cases we w i l l f i n d greater deviance when the stranded nominal head i s a subject. 2.1 Postposing a r e l a t i v e clause out of a non-matrix CP In the following examples, a r e l a t i v e clause i s postposed, stranding i t s nominal head, which i s i t s e l f embedded in a koto clause. We s h a l l see that when we postpose a r e l a t i v e clause out of a DP, postposing i s s e n s i t i v e to the p o s i t i o n of the DP i n i t s VP: whether i t i s a subject ( s p e c i f i e r position) or object (complement p o s i t i o n ) . In sec. 2.1.1 I s h a l l examine the postposing of object-headed r e l a t i v e s ; i n sec. 2.1.2 the postposing of subject-headed r e l a t i v e s . In some of the examples i n t h i s section, the koto clause i s nominative; however, i n a l l cases, the verb of which the clause i s the subject i s i n t r a n s i t i v e and i s arguably an unaccusative construction. In section 2.7 I w i l l give evidence that nominative koto clauses show SC e f f e c t s when they occur as a subject of a - t r a n s i t i v e verb. Similar examples also occur i n sec. 5.8.2. In order to i s o l a t e the e f f e c t s of the subject- vs. object-headedness of the r e l a t i v e clause that the postposed phrase i s being extracted from, I do not introduce i n 65 t h i s subsection examples with koto clauses that are t r a n s i t i v e subjects. The following diagram i l l u s t r a t e s the relevant structure. (83) 6 6 CPi / \ CPi CP2 ( r e l a t i v e clause) I IP I VP V / \ (koto clause)CP 3 V I C I \ IP c I koto VP / \ (subject) DP \ I \ D' \ / \ \ (trace of r e l a t i v e ) t 2 D' V / _ \ , / \ (object) DP I D' / \ (trace of r e l a t i v e ) t 2 D' / _ \ 67 2.1.1 object-headed r e l a t i v e postposed 2.1.1.1 nominative koto clause as a b a r r i e r : ( p o s s i b l e unaccusative construction) (84) Masao-ga, okane-o nusunda koto-ga akiraka da, kinoo boku-ga moratta. -nom money-acc s t o l e comp-nom cl e a r copula yesterday I-nom received I t ' s c l e a r that Masao st o l e the money -- that I received yesterday. 2.1.1.2 nominative koto clause with koto-qa aru construction as a b a r r i e r , (possible unaccusative construction) (85) Masao-ga, okane-o nusunda koto-ga aru, boku-no kakusita. -nom money-acc s t o l e comp-nom exist I-gen h i d Masao has (on occasion) stolen money -- that I hid. 2.1.1.3 accusative koto clause as a b a r r i e r (86) Masao-ga, okane-o nusunda koto-o s i t t e - i r u , kinoo boku-ga moratta. -nom money-acc st o l e comp-acc know yesterday I-nom received I know that Masao st o l e the money -- that I received yesterday, (or) Masao knows that (they) s t o l e the money -- that I/he received yesterday. 2.1.2 subject-headed r e l a t i v e postposed: 2.1.2.1 nominative koto clause as a b a r r i e r : 68 (87) *akatyan-ga n a i t e - i r u koto-ga akiraka da, t o n a r i - n i sunde-iru. baby-nom c r y i n g - i s comp-nom cl e a r copula next-door-loc l i v e s It i s c l e a r that the baby i s crying -- (the one) that l i v e s next door. 2.1.2.2 nominative koto-qa am construction as a b a r r i e r : (88) *akatyan-ga naita koto-ga yoku aru, t o n a r i - n i sunde-iru. baby-nom c r i e d comp-nom much exist next-door-loc l i v e s There have often been times when the baby c r i e d -- (the one) that l i v e s next door. 2.1.2.3 accusative koto clause as a b a r r i e r : (89) *akatyan-ga n a i t e - i r u koto-o s i t t e - i r u , t o n a r i - n i sunde-iru. baby-nom c r y i n g - i s comp-acc know next-door-loc l i v e s I know that the baby i s crying -- (the one) that l i v e s next door. In the above examples we see that (a) the koto clause does not appear to be a b a r r i e r for extraction of the postposed phrase; (b) a subject ((87)-(89) i s a b a r r i e r for extraction of the postposed phrase whereas an object ((84)-(86)) i s not. 2.2 Postposing a r e l a t i v e clause out of a wh-island: In the following examples, a r e l a t i v e clause i s again postposed, stranding i t s nominal head, which t h i s time i s embedded i n a wh-island. Once again we s h a l l see that greater deviance r e s u l t s when the DP that the r e l a t i v e clause modifies i s i n a subject rather than an object p o s i t i o n . 69 CP! / \ CPj CP2 ( r e l a t i v e clause) I IP I VP I V / \ (+wh)CP3 V / \ / \ DP I D' / \ t 2 D' 70 2.2.1 object-headed r e l a t i v e postposed: 2.2.1.1 sing l e wh-island b a r r i e r : (91) ?*Masao-ga, sono okane-o nusunda ka-doo-ka wakaranai, kinoo boku-ga moratta. -nom that money-acc s t o l e comp not-clear yesterday I-nom received I t ' s not c l e a r i f Masao sto l e the money -- that I/he received yesterday. 2.2.1.2 double wh-island b a r r i e r : (92) ?*Masao-ga, sono okane-o nusunda ka-doo-ka wakaru no, boku-no kakusita? -nom that money-acc st o l e comp i s - c l e a r comp I-gen hid Do you know i f Masao sto l e the money -- that I hid? 2.2.2 subject-headed r e l a t i v e postposed: 2.2.2.1 single wh-island b a r r i e r : (93) **sono akatyan-ga n a i t e - i r u ka (doo ka) wakaranai, t o n a r i - n i sunde-iru. that baby-nom i s - c r y i n g comp not-clear next-door-loc l i v e I t ' s not cl e a r whether the baby i s crying -- (the one) that l i v e s next door. 71 2.2.2.2 double wh-island b a r r i e r : (94) **sono akatyan-ga n a i t e - i r u ka wakaru no, t o n a r i - n i sunde-iru? that baby-nom i s - c r y i n g comp clear comp next-door-loc l i v e Do you know i f the baby i s crying -- (the one) that l i v e s next door? The following sentence suggests that the demonstrative adjective sono i n (91) - (94) i s not necessar i l y acting as a resumptive for the r e l a t i v e clause --that i s , i t does not ru l e out movement of the r e l a t i v e clause. When sono co-occurs with the r e l a t i v e clause, we s t i l l get what appears to be a Condition C reconstruction e f f e c t when the name Masao i s c-commanded by the pronoun "he".22 (95) kare . . i / ? . . j-ga sono kuruma-o utta ka-doo-ka wakaranai, kinoo Masaoj-ga nusunda he-nom that car-acc sold comp not-clear yesterday -nom sto l e I don't know whether he sold that car -- that Masao sto l e yesterday. With no coreference between the pronoun and the name, (95) i s judged to have s i m i l a r grammaticality to (92); with coreference i t ' s worse; thus a reconstruction e f f e c t seems to occur here. In the foregoing set of examples the wh-island(s) appears to act as a weak b a r r i e r for extraction of the r e l a t i v e clause, inasmuch as a l l the examples are not completely grammatical. What i s i n t e r e s t i n g , though, i s that the degree of deviance i s more sharply affected by whether the head of the r e l a t i v e clause out of which the r e l a t i v e clause i s extracted i s a subject or an object than by the number of wh-islands i n which the clause i s embedded. 2 2 N o t i c e t h a t t h e d e m o n s t r a t i v e a d j e c t i v e sono c a n c o e x i s t w i t h a r e l a t i v e c l a u s e i n a c a n o n i c a l s e n t e n c e : ( i ) k a r e - g a , [ s o n o [ k i n o o M a s a o - g a , n u s u n d a ] ,,, k u r u m a - o ] u t t a k a - d o o - k a w a k a r a n a i , h e - n o m t h a t y e s t e r d a y - n o m s t o l e c a r - a c c s o l d comp n o t - c l e a r I d o n ' t know w h e t h e r h e ! s o l d t h a t c a r t h a t M a s a o 2 s t o l e y e s t e r d a y . 72 2.3 Postposing a genitive out of a DP inside an adverbial clause: Diagram (96) i l l u s t r a t e s the relevant structure. The adverbial clause appears to constitute a weak b a r r i e r for extraction of the gen i t i v e i n (98). But i n (97) the subject "car" out of which the genitive i s postposed i s a worse b a r r i e r for extraction than the object "money" i n (98) or the concessive clause i n both. (96) CP, I \ CPj DP2 (genitive) I IP I I' / \ (adverbial)PP I' I \ CP4 VP / \ /_\ / \ DP3 D' / \ t 2 D' / \ 73 2.3.1 subject DP b a r r i e r : (97) * kuruma-ga boku-o h i i t a no-ni, kega-wa nakatta, Taroo-no. car-nom me-acc ran-over although injury-top(contrast) not-exist -gen In s p i t e of the fact that a car ran over me, I'm not injured-- Taro's (car). 2.3.2 object DP (98) ??Masao-wa, okane-o nusunda no-ni, sikararenakatta, Taroo-no. -top money-acc st o l e although scold-pass.-neg-past -gen In s p i t e of the fact that Masao st o l e money, he wasn't scolded-- Taro's (money) 74 2.4 P o s t p o s i n g a r e l a t i v e c l a u s e out o f an a d v e r b i a l c l a u s e : I n t h e f o l l o w i n g examples, a r e l a t i v e c l a u s e i s p o s t p o s e d , s t r a n d i n g i t s n o m i n a l head, w h i c h i s i t s e l f embedded i n an a d v e r b i a l c l a u s e . (99) CP! / \ CP1 CP 2 ( r e l a t i v e c l a u s e ) C I IP I I ' / \ ( a d v e r b i a l ) P P I CP 4 / \ / \ DP 3 D' / \ t 2 D' / _ \ I ' \ VP /_\ 75 2.4.1 object-headed r e l a t i v e postposed: (100) (better than (98) ?Masao-wa, okane-o nusunda no-ni, sikararenakatta, boku-no kinoo moratta. -top money-acc sto l e although scold-pass.-neg.-past I-gen(subj) yesterday received In s p i t e of the fact that Masao sto l e money, he wasn't scolded -- (money) that I received yesterday. 2.4.2 subject-headed r e l a t i v e postposed (101) *akatyan-ga n a i t e - i r u no-ni, miyoo-ni sizuka datta, t o n a r i - n i sunde-iru. baby-nom i s - c r y i n g although strangely quiet was next-door l i v e In s p i t e of the fact that the baby i s crying, i t ' s strangely quiet -- the (baby) that l i v e s next door. In the above examples we see the same subject-object asymmetry as we d i d i n section 2.3. 2.5 Postposing a dative out of a r e l a t i v e clause: In the following examples, a dative argument of a r e l a t i v e clause i s postposed. Again, a subject (102) 2 3 i s a more severe i s l a n d than an object (103) . 23The matrix verb in this sentence may be unaccusative; nevertheless, we see an apparent SC effect, unlike what we saw for nominative koto clauses that were subjects of unaccusative verbs in sec. 2.1.1.1.1 cannot explain why a SC effect seems to occur here but not in sec. 2.1.1.1. 76 2.5.1 subject-headed r e l a t i v e as a b a r r i e r : (102)**puresento-o ageta hito-ga koko-ni i r u , Hanako-ni. present-acc gave person-nom here-loc e x i s t -dat The person who gave a present i s here -- to Hanako. 2.5.2 object-headed r e l a t i v e as a b a r r i e r : (103) ?*puresento-o ageta hito-o yonda, Hanako-ni. present-acc gave person-acc c a l l e d -dat I c a l l e d the person who gave a present -- to Hanako. bad but better than (102) 2.6 Postposing a r e l a t i v e clause out of a r e l a t i v e clause: In the following examples, a r e l a t i v e clause modifies an argument of a higher r e l a t i v e clause. The most deeply embedded r e l a t i v e clause i s postposed, stranding i t s nominal head. 77 (104) CPa / \ CPi CP2 ( r e l a t i v e clause) I IP I VP I V / \ I D' / \ ( r e l a t i v e clause)CP 3 D' / \ /_\ / \ DP5 I D' / \ E / _ \ 78 2.6.1 object-headed r e l a t i v e postposed: higher clause i s subject-headed (but with possible unaccusative verb i n matrix clause) (105) ?purin-o tabeta otoko-no-ko-ga nigeta, boku-no tukutta. pudding ate boy-nom ran-away I-gen(subj.) made The boy who ate the pudding ran away -- (the pudding) that I made. (106) ?kodomo-o h i i t a kuruma-ga kieta, asoko-de asonde-ita. child-acc ran-over car-nom disappeared over-there-loc was-playing The car that ran over the c h i l d disappeared -- (the child) who was playing over there. 2.6.2 subject-headed r e l a t i v e postposed; higher clause i s subject-headed (but with possible unaccusative matrix verb) (107) *kodomo-ga kaita e-ga kieta, t o n a r i - n i sunde-iru. child-nom drew picture-nom disappeared next-door-loc l i v e The p i c t u r e that the c h i l d drew disappeared -- (the child) who l i v e s next door. (108) *kodomo-ga kaita e-ga kieta, gakkoo-ni t u r e t e - k i t a . child-nom drew picture-nom disappeared school-loc took (animate obj.) The p i c t u r e that t h e . c h i l d drew disappeared -- (the child) who I took to school. (105), (106) are better than (107), (108) 79 2.6.3 object-headed r e l a t i v e postposed; higher clause i s object-headed: (109) ?purin-o tabeta otoko-no-ko-o s i k a t t a , boku-no tukutta. pudding-acc ate boy-acc scolded I-gen made I scolded the boy who ate the pudding -- (the pudding) that I made. (110) ?kodomo-o h i i t a kuruma-o tubusita, asoko-de asonde-ita. child-acc ran-over car-acc smashed over-there-loc was-playing I smashed the car that ran over the c h i l d -- (the one) that was playing over there. (109), (110) same as (105), (106) 2.6.4 subject-headed postposed; higher clause i s object headed: (111) *kodomo-ga kaita e-o suteta, t o n a r i - n i sunde-iru. child-nom drew picture-acc threw-away next-door-loc l i v e I threw away the p i c t u r e the c h i l d drew -- (the child) who l i v e s next door. (109), (110) better than (111) When we replace the matrix verb i n (105) with one that i s unergative, we get worse grammaticality: (112) ?*purin-o tabeta otoko-no-ko-ga n a i t e - i r u , boku-no tukutta. pudding ate boy-nom i s - c r y i n g I-gen(subj.) made The boy who ate the pudding i s crying -- (the pudding) that I made. This fact suggests that the nominal head of the outer r e l a t i v e clause i s showing 80 Subject Condition e f f e c t s as well. To summarize the r e s u l t s i n t h i s subsection, when a r e l a t i v e clause i s postposed, stranding i t s nominal head, which i s i t s e l f embedded i n a higher r e l a t i v e clause, the presence of a subject as head of eith e r the inner or outer r e l a t i v e clause w i l l r e s u l t i n greater deviance than i f both heads of the r e l a t i v e clauses are object-headed. 2.7 Postposing out of koto clauses that are subjects of t r a n s i t i v e verbs When a clause headed by complementizer koto i s used i n a sentence, i t usually i s not an argument of a t r a n s i t i v e verb, as was the case for the examples given above with koto clauses. In the following examples, a koto clause does occur as subject of a t r a n s i t i v e verb, thus assuring that i t w i l l be an underlying subject. (113) 81 CPi / \ CPi DP2 IP I' VP / \ {koto clause)CP 3 V / \ / \ / \ DP V DP D' / \ I t 2 D' / \ (114)?*Taroo-wa, [ei nomisugita koto-ga] kare-o korosita, u i s u k i i - O i . -top drank-too-much comp-nom him-acc k i l l e d whisky-acc The fact that, Taro drank ( i t j too much k i l l e d him --, whisky;. (115)?Taroo-wa, [et nomisugita koto-o] h i t e i - s i t a , u i s u k i i - O i . -top drank-too-much C-acc denied whisky-acc Taro denied that he drank ( i t j too much -- whisky A. 82 When the koto clause i s nominative and occurs as subject of t r a n s i t i v e verb " k i l l e d " i n (114), postposing out of i t i s much worse than i n (115), where the same clause i s accusative and i s object of the verb. We can assume, then, that the koto clause i s underlyingly a subject i n (114). In the examples we looked at in section 2.1, the fact that we di d not see SC e f f e c t s i n postposing out of nominative koto clauses may have been due to the fact that these clauses were surface subjects i n what are possibly r a i s i n g constructions (e.g. " i t i s clear that" ) or unaccusative constructions (e.g. verb of existence.) 2.8 Relative clause postposed out of a matrix subject In the following examples, we see that postposing a r e l a t i v e clause out of a subject i s worse than postposing out of an object. 2 4 (116) ?*akatyan-ga n a i t e - i r u , t o n a r i - n i sunde-iru. baby-nom cry next-door-loc l i v e s The baby i s crying -- (the one) who l i v e s next door (117) ??akatyan-o mita, t o n a r i - n i sunde-iru. baby-acc saw next-door-loc l i v e s I saw the baby -- (the one) who l i v e s next door. Endo (1989:113,#80c) gives the following example of a relative clause postposed out of a matrix subject which she judges as good: (89) kaki-ga oisikatta yo, Hirosima-de tabeta. oysters-nom delicious-past emph -loc ate The oysters were delicious — (the ones) I ate in Hiroshima. The grammaticality of this sentence, however, might be explained by the possibility that the inflected adjective "delicious", is a psychological predicate for which "oysters" is not underlyingly in subject position. 2.9 Summary 83 The following pattern emerges from the examples we have looked at i n t h i s chapter. A subject DP or a nominative-case-marked Jcoto-clause that i s subject of a t r a n s i t i v e verb appears to be a worse b a r r i e r for e x t r a c t i o n than a wh-island, an adverbial clause, or a r e l a t i v e clause. 2 5 In conclusion, the postposing construction i n Japanese i s evidence that subjects i n Japanese are strong islands for extraction -- stronger than wh-islands or at least some types of adjunct phrases. These r e s u l t s are s i g n i f i c a n t , given the fact that Japanese has been considered not to show subject-object asymmetries. C l a s s i c a l - t y p e ECP analyses of l e x i c a l government i n Japanese have concluded that subjects i n Japanese are l e x i c a l l y governed 2 6, for example as a r e s u l t of the subject remaining i n VP or by r a i s i n g of V to I, eit h e r of which would allow the subject to be l e x i c a l l y governed by V. In postposing i n Japanese we do not see subject/object asymmetries with respect to the p o s t p o s a b i l i t y of subjects and objects. For example, base-generated topics marked by wa appear to be islands for extraction of a postposed phrase i n cases where the trace of the postposed phrase i s not i n the domain of a l e x i c a l [+V] head. 2 7 Yet we f i n d that subjects pattern with objects i n that they can be postposed out of a topic . In the f i r s t sentence below an object i s postposed out of a to p i c . In the second, a subject i s postposed instead. The s l i g h t decrease in grammaticality we see in the second sentence i s nowhere near "One type of structure that I have not yet considered is a matrix subject modified by a genitive. For some speakers, postposing a genitive out of a matrix subject is bad, but not so for others. In example (60) we saw evidence of a sentence that was judged grammatical in which a genitive was postposed out of a subject, but not all speakers judge such sentences as grammatical. I leave the problem of the islandhood of subjects with respect to genitives unsolved here. 26See, for example, Huang (1982), Koopman and Sportiche (1985), Whitman (1991). 27 This idea is discussed in detail in chapter 4: "Postposing out of topics." The examples below are repeated later in that section. 84 the degree we expect to see with an ECP e f f e c t that occurs i f subjects are not l e x i c a l l y governed. (See chapter 4 for further discussion.) (119) C P[dareka-ga ei tabetesimatta koto]-wa, t a s i k a da , [ano wagasi-o] ±. someone-nom ate-up comp-top c e r t a i n copula that sweet-acc It' s c e r t a i n that someone ate (iti) up that sweety (120) ? Cp[ei Wagasi-o tabetesimatta koto]-wa, tasika-da, Taroo-gai sweet-acc ate-up comp-top certain-cop. -nom I t ' s c e r t a i n that (he) ate up the sweet-- Taro. A possible explanation for the fact that subject/object asymmetries sometimes appear i n Japanese (as i n the case of the SC e f f e c t s on postposing we saw above) and other times do not (as we see for postposing out of topics) i s that when subjects, are "heavy" by v i r t u e of being modified by a CP, something about t h e i r information structure requires them to r a i s e to an A' p o s i t i o n i n order to take scope over the rest of the sentence. 2 8 Suppose that subjects i n Japanese are normally L-marked, ei t h e r by remaining i n VP or by r a i s i n g of V to I. This would explain the l i c i t status of traces of subjects postposed out of weak islands such as we see above in (120). But i f a subject i s made heavy by the addition of c l a u s a l material, as was the case in a l l the examples we looked at i n postposing out of subject clauses -- for example out of a subject-headed r e l a t i v e clause -- i f i t i s forced to r a i s e i n such a case to an A' p o s i t i o n , i t would become an i s l a n d for e x t r a c t i o n . 2 9 Could a phrase be postposed out of such a subject before i t raises? If i t d i d and i f the postposed phrase adjoined higher than the r a i s e d subject then we This idea was suggested by Michael Rochemont ( p . c ) . 'This raising could even occur at LF if the adjunct condition was representational rather than derivational. 85 would v i o l a t e s t r i c t c y c l i c i t y , since we would need to go back to a lower cycle to r a i s e the subject, (see (121) below) If the r a i s e d subject attached higher than the postposed phrase then the trace of the postposed phrase would not be c-commanded by the postposed phrase, v i o l a t i n g the Proper Binding Condition. (See (122) below.) Thus r a i s i n g of the subject would rule out postposing out of i t . (121) CP / \ CP postposed phrasej IP / \ landing s i t e I' of subject \ VP / \ DP V / \ 86 (122) XP / \ landing s i t e CP of subject / \ CP postposed phrase; I IP \ I ' \ VP / \ DP V / \ t i Under our hypothesis, could a heavy subject be postposed i n i t s entirety? It could i f movement to a postposed p o s i t i o n i s i n i t s e l f s u f f i c i e n t to s a t i s f y whatever requirement there might be for the heavy subject to take scope. In fact, we s h a l l see i n sec. 6.1 that postposed phrases " l i k e to be heavy". This does not imply the converse -- that heavy phrases l i k e to be postposed. But i f postposed phrases are frequently heavy, i t suggests that i n order to be r e a d i l y postposed, heavy phrases need not f i r s t move leftward to some high A' p o s i t i o n to s a t i s f y some requirement for heavy phrases and then move to a postposed p o s i t i o n . If t h i s hypothesis about r a i s i n g of subjects with c l a u s a l material i s correct, then what we have been seeing i n the examples i n t h i s chapter i s not r e a l l y a Subject Condition phenomenon but rather a kind of A' p o s i t i o n condition: i . e . that c e r t a i n A' posit i o n s are strong islands for extraction. It i s necessary 87 to say "certain" A' positions since other phrases that are arguably i n an A' p o s i t i o n such as topics and adverbial phrases do not show as strong i s l a n d e f f e c t s as heavy subjects, as we s h a l l see i n the next two chapters. 88 3 . L o c a l i t y Constraints -- Part I I : Island e f f e c t s that are r e l a t i v i z e d with respect to a [+N] feature In t h i s chapter I examine a second l o c a l i t y constraint phenomenon that appears i n data on postposing. When a constituent i s postposed out of a DP pr o j e c t i o n such as a subject or object, the deviance i s worse i f the postposed element i s i t s e l f a DP such as a subject or object than i f i t i s an element with a [+N] feature such as a r e l a t i v e clause. Thus i n the structure below, i n which a DP modified by a r e l a t i v e clause i s embedded i n a higher r e l a t i v e clause, worse deviance r e s u l t s i f the embedded object DP1 i s postposed than i f i t s modifier, the lower r e l a t i v e clause CPX i s postposed. 89 (123) . I P / \ IP landing s i t e of postposed phrase I VP I V / \ DP 2 V I D ' / \ CP2 D ' I / _ \ IP I VP / \ DPj V I D ' / \ CPi D ' / _ \ These kinds of e f f e c t s are not explained by l o c a l i t y constraint models such as the Chomsky (1986) "Barriers" model or by R i z z i (1990)'s " R e l a t i v i z e d Minimality." In the structure above, DP 2 acts l i k e a b a r r i e r for extraction of DPj even though, by the Chomsky (1986) model, DP 2 i s l e x i c a l l y governed by matrix 90 V and therefore L-marked. CP2, being an adjunct, i s not L-marked, but the phenomenon that I s h a l l i l l u s t r a t e from the data below i s not islandhood of a r e l a t i v e clause: i n fact, we s h a l l see that a r e l a t i v e clause i n Japanese acts l i k e only a very weak i s l a n d for postposing. What we s h a l l see i s that DP2 blocks extraction (by postposing) of a category within i t only i n a r e l a t i v i z e d way: i t does not permit extraction of another DP that shares i t s [+N] feature, but i t does marginally permit extraction of an element that i s not [+N], such as a CP. These facts also d i f f e r from those that are accounted for by R i z z i ' s (1990) model, which does not deal with r e l a t i v i z a t i o n with respect to c a t e g o r i a l features such as [+N] . It also deals with the e f f e c t s of intervening heads rather than maximal projections. In the data on postposing, we do not see that the crossing of a [+N] head by another [+N] element i s what induces deviance. Consider, the following e a r l i e r example, repeated here as (124). 91 (124) (a) karej-ga semeta, [TaroOi-no tomodati-o] k. hej-nom blamed [T—gen f r i e n d s - a c c ] k He, blamed (themk) -- [TarOi's f r i e n d s ] k . (b) IP / \ IP DPk I VP / \ DP, V I / \ D' ./ \ DP; D' / \ NP D When object DPk, which has a [+N] feature, i s postposed i t crosses the head D and N of the extended proj e c t i o n DP, the matrix subject, which w i l l contain a [+N] head. Yet the head of DP, does not block the postposing of the object. Thus i n the data we s h a l l see below, i t i s not intervening heads, but rather a dominating XP that has a [+N] feature that w i l l appear to block the extraction of a YP with a [+N] feature, where XP contains YP. 3.1 Postposing out of a r e l a t i v e clause: worse deviance when a DP i s postposed than when another category i s postposed The following two examples i l l u s t r a t e the structure i n (123) above. We saw i n the preceding chapter that a r e l a t i v e clause can be marginally postposed out 92 of a r e l a t i v e clause when the r e l a t i v e clause i s object-headed. 3 0 We see below that when we postpose a subject or object out of the same r e l a t i v e clause, the r e s u l t i s worse than i f just the r e l a t i v e i s postposed, stranding i t s nominal head. (125) ?[[ ei purin-o jtabeta otoko-no-ko-ga] nigeta, [CPboku-no tukutta ] i . pudding ate boy-nom31 ran-away I-gen made The boy who ate the pudding ran away -- (the pudding) that I made. (126) *[ei tabeta] otoko-no-ko-ga] nigeta, [DPboku-no tukutta purin-o]i. ate boy-nom ran-away I-gen made pudding. The boy who ate (it) ran away -- the pudding that I made. The fact that postposed phrases prefer to be heavy (see chapter 6) means that we cannot explain the worse grammaticality of (126) by the fact that the postposed phrase i n t h i s sentence contains more material than i n (125) . In addition to r e l a t i v e clauses, genitives also induce weaker deviance when they are postposed out of an argument DP than when a subject or object i s postposed. In (127) below the postposed genitive must cross the DP "toy", the r e l a t i v e clause, and the matrix subject "baby" whereas i n (128), the postposed object of the r e l a t i v e clause must cross only the r e l a t i v e clause and the matrix subject; yet, (127) i s judged as better than (128). In order to explain why genitives group with r e l a t i v e clauses rather than subject or object DP's i n t h i s way, we need to assume that they do not have a [+N] feature. In Appendix B, I discuss genitives i n more d e t a i l . 3 0Endo (1989:114,#81c) claims that a r e l a t i v e clause cannot be postposed out of a r e l a t i v e clause; however, i n the example that she uses one of the clauses i s subject-headed. 31We do not see a SC e f f e c t here despite the nominative case of the "koto" clause, since t h i s i s l i k e l y an unaccusative c o n s t r u c t i o n . 93 • (127)(a) [ [e£ omotya-o] kowasita akatyan-ga] nigeta, [ano-ko-no] L. toy-acc broke baby-nom ran-away that-child-gen The baby who broke the toy ran away -- that c h i l d ' s (toy). 94 (127) (b) (structure of (127) (a) before postposing IP / \ IP landing s i t e of postposed phrase I VP I V / \ DP3 V | ran-away D' / \ • CP2 D' I / _ \ IP baby I VP / \ DP2 V | broke D' / \ DPX D' that c h i l d ' s / \ DPi i s postposed toy 9 5 (128) (a)?*[e 4 kowasita] akatyan-ga nigeta, [omotya-o]j. The baby who broke ( i t j ran away -- the toy;. (128)(b) structure of (128) a) before postposing IP / \ IP landing s i t e of postposed phrase V / \ DP3 V ran-away D' / \ CP2 D' I /_\ IP baby broke baby-nom ran-away toy-acc VP VP / \ DP2 V DP2 i s postposed broke D' / _ \ toy 96 Yet we f i n d that when we look at postposing out of other islands such as a wh-island or concessive clause, the reverse i s true. That i s , a r e l a t i v e or geni t i v e postposed out of a wh-island (see sec. 3.2.1) or adverbial clause (see sec. 3.2.2) i s worse than postposing a subject or object. The following data show the range of i s l a n d e f f e c t s that occur when d i f f e r e n t kinds of constituents are postposed out of a r e l a t i v e clause. 3.1.1 Survey of elements postposed out of a r e l a t i v e clause In t h i s section I s h a l l compare the degree of deviance that occurs when d i f f e r e n t categories of phrases are postposed from a r e l a t i v e clause. 3.1 .1 .1 Postposed ni-marked phrase In t h i s subsection, a phrase marked by p a r t i c l e ni, which I s h a l l assume to be a dative or lo c a t i v e DP, i s postposed. In a l l cases the r e s u l t i s f a i r l y ungrammatical. (129) ni 3 2-marked DP (dative interpretation) (out of object-headed clause): *?puresento-o ageta hito-o yonda, Hanako-ni. present-acc gave person-acc c a l l e d -dat I c a l l e d the person who gave a present -- to Hanako. 3 2 T h e p a r t i c l e ni m a r k s e i t h e r a d a t i v e o r l o c a t i v e D P . I n A p p e n d i x B I show t h a t t h e p a r t i c l e ni b e h a v e s m o r e l i k e a c a s e - m a r k e r t h a n a P . 97 (130) (Locative interpretation) (postposed out of subject-headed clause:) *buzitodo'ita hune-ga Kanada-kara da, minato-ni. s a f e l y - a r r i v e d ship -nom Canada-from copula p o r t - l o c The ship that a r r i v e d s a f e l y i s from Canada -- (arrived) i n port. (131) . (postposed out of object-headed clause:) *buzitodoita hune-o mita, minato-ni. s a f e l y - a r r i v e d ship-acc saw p o r t - l o c I saw the ship that s a f e l y a r r i v e d i n port. We see that "dative" ni-marked phrases are s l i g h t l y easier to postpose out of a r e l a t i v e clause than subjects or objects but the extraction i s s t i l l f a i r l y bad. Locative ni-marked phrases cannot be postposed out of e i t h e r object-headed or subject-headed r e l a t i v e clauses. If these elements are DP's, with a [+N] feature, then they i l l u s t r a t e the fact that the [+N] feature of head DP of the r e l a t i v e clause seems to block extraction of another [+N] element. 3.1.1.2 Postposed PP In t h i s subsection, a PP i s postposed out of a r e l a t i v e clause. Again, the examples are a l l ungrammatical. (132) (postposed out of subject-headed clause:) * b u z i t u i t a hune-ga minato-ni aru, [Kanada-kara] P P. s a f e l y - a r r i v e d ship-nom port-l o c exist Canada-from The ship that s a f e l y a r r i v e d i s i n port -- (arrived) from Canada. 98 (133) (postposed out of object-headed clause:) * b u z i t u i t a hune-o mita, [Kanada-kara] P P. s a f e l y - a r r i v e d ship-acc saw Canada-from I saw the ship that sa f e l y a r r i v e d -- from Canada. If we assume that PP's have a [+N] feature because they are i n the extended p r o j e c t i o n of N i n Grimshaw (1991) 's model, these data are consistent with our g e n e r a l i z a t i o n about the blocking of extraction of a [+N] category by a [+N] maximal p r o j e c t i o n that contains i t -- again, the head DP of the r e l a t i v e clause. 3.1.1.3 Adverb In t h i s example, an adverb i s postposed from a r e l a t i v e clause. The sentence i s marginal, but much better than the examples i n sections immediately above. This contrast further supports our g e n e r a l i z a t i o n : i f we assume that an adverb i s not [+N], we do not see the blocking e f f e c t of the [+N] head DP "boy". (134) ??purin-o tabeta otoko-no-ko-ga asoko da, [musyamusya]adv. pudding-acc ate boy-nom over-there copula ravenously The boy who ate the pudding i s over there -- ravenously. 3.1.1.6 c l a u s a l elements to clause: (non-case-marked CP) (135) ?Taroo-ni i t t a sensei-ga kaetta, [siken-ga enki-sareta t o ] C P . -dat said teacher-nom went-home exam-nom postponed comp The teacher who said ( i t j j to Taro went home -- [that the exam i s postponed] i. 99 koto clause: (case-marked CP) (136) s i t t e - i r u hito-ga sukunai, [siken-ga enki-sareta koto-o] C P. know person-nom few exam-nom postponed comp-acc The people who know are few -- that the exam has been postponed. s i - n i 3 3 purpose clause (case-marked bare VP): (137) iku hito-ga sukunai, eiga-o mi-ni. go person-nom few film-acc see-loc The people who are going are few -- to see a f i l m . A l l three of the above i l l u s t r a t e d categories are s u c c e s s f u l l y postposed from a r e l a t i v e clause. This r e s u l t i s consistent with our g e n e r a l i z a t i o n i f we assume that they a l l lack a [+N] feature. 3.1.1.6 Argument DP obj ect: (128) repeated here as (138): ?*kowasita akatyan-ga nigeta, omotya-o. broke baby-nom ran-away toy-acc The baby who broke (it;) ran away -- the toy;. subject: (139) *kowasita omotya-o boku-ga naosita, akatyan-ga. broke toy-acc I-nom fi x e d baby-nom I f i x e d the toy that e; broke -- the baby;. These two above examples contrast sharply with the examples of postposing In this construction the clause consists of the bare uninflecled form of the verb which is marked by particle 100 CP's adverbs, and VP's. The f a c t that a subject and object arguably have a [+N] feature supports our g e n e r a l i z a t i o n . 3.1.1.7 Hierarchy of p o s t p o s a b i l i t y c a t e g o r i e s Based on the evidence above, the f o l l o w i n g i s a h i e r a r c h i c a l l i s t of phrase types i n the order of the deviance that r e s u l t s when they are postposed out of a r e l a t i v e c l a u s e . 3 4 LEAST DEVIANT si-ni purpose clause koto clause to clause adverb, object-headed r e l a t i v e clause d a t i v e o b l i q u e obj ect subject MOST DEVIANT These r e s u l t s are s u r p r i s i n g i n that we would normally expect arguments, e s p e c i a l l y o b j e c t s , to be e a s i e r to e x t r a c t than adverbs or r e l a t i v e c lauses. From the h i e r a r c h y above, one can make the f o l l o w i n g g e n e r a l i z a t i o n : The l a s t four members of the hiera r c h y -- those that are the most d i f f i c u l t to postpose out of a r e l a t i v e clause -- can a l l be argued to be i n the extended p r o j e c t i o n ( i n the sense of Grimshaw (1991)) N - D - P: subjects and ob j e c t s are DP's and obliques d a t i v e s are e i t h e r DP's or PP's, depending on the sta t u s of " I have omitted genitives from this list because of the fact that genitives are somewhat problematic for this account. We need to assume that they do not have a [+N] feature, even though in some ways they appear to be DP's. I shall postpone the discussion of genitives until a later point in this subsection. 101 t h e i r p a r t i c l e . The f i r s t four members on the other hand could be argued not to be i n the N - D - P pro j e c t i o n : adverbs are usually not considered to have a [+N] feature, r e l a t i v e clauses i n Japanese have a surface form i d e n t i c a l to IP's, with no complementizer or case-marking, to clauses are not case-marked, suggesting no nominal feature, and purpose clauses are projections of V, taking the form such as : [XP-o V]-ni -acc -loc (see, for example, (137) above) The only elements among the f i r s t four whose status as lacking [+N] i s questionable are koto clauses. These elements resemble complex DP's i n that they are o b l i g a t o r i l y case-marked. Whitman (1991a), however, argues that A:oto clauses (as well as no clauses which have a s i m i l a r grammatical function) pattern d i f f e r e n t l y from complex DP's, for example by not allowing ga-no conversion 3 5. If we accept Whitman's argument that koto clauses are CP's, not DP's then koto clauses w i l l group with purpose clauses, to- clauses, and adverbs i n lacking a [+N] feature i n t h e i r extended p r o j e c t i o n . I can only give a d e s c r i p t i v e g e n e r a l i z a t i o n of the phenomena we are seeing here: I am unable to give an explanatorily adequate account of these data. The ge n e r a l i z a t i o n i s as follows: In the following structure, XPif+N] cannot bind a trace t A i f YP[+N] dominates, t^: 35Ga/no conversion refers to the phenomenon of converting the nominative case-particle ga to the genitive case-particle no to mark the subject of a relative clause. Subjects of relative clauses are more commonly marked with the genitive than the nominative case-particle. 102 XPJ+N] ... {YP[+N] ... t i ... }36 We would assume that a subject or object w i l l have t h i s [+N] feature on i t s head N, which w i l l manifest i t s e l f i n the whole DP of the phrase, for example by perc o l a t i o n of the [+N] feature to the top of i t s extended p r o j e c t i o n or by s e l e c t i o n of a DP with a [+N] D head by an NP. Thus when a DP such as a subject or object i s postposed out of a r e l a t i v e clause, i t must cross'the p r o j e c t i o n of the head N of the DP pr o j e c t i o n that the r e l a t i v e clause modifies, which also must have a [+N] feature. Such crossing of a p r o j e c t i o n with a [+N] feature by another p r o j e c t i o n with a [+N] feature appears to be what causes deviance i n the examples we have looked at above. On the other hand, when a r e l a t i v e clause i s postposed, we could argue that the CP r e l a t i v e clause i s an extended pro j e c t i o n of the V of the clause, and w i l l not have a [+N] feature. Thus no severe deviance occurs when the r e l a t i v e clause crosses the DP the higher r e l a t i v e clause modifies. We see t h i s contrast i n the following two examples: (128), (138) repeated again as (140): ?*{ei kowasita akatyan-ga[+N]} nigeta, omotya-Oi [ +N] . broke baby-nom ran-away toy-acc The baby who broke ( i t j ran away -- the toyi. " Although the structure here is expressed in left-to-right order of c-command, it is assumed that precedence does not figure in binding of a trace by a postposed phrase in Japanese. 103 (125) (repeated as (141)) ?{{€; purin-o [+N] } tabeta otoko-no-ko-ga [ +N] } nigeta, {IP[_N) {VP[_N]boku-no tuku}tta}i-pudding ate boy-nom ran-away I-gen made The boy who ate the pudding ran away -- (the pudding) that I made. In the case of PP's, we might argue that P, as an extended pro j e c t i o n of N-D also has a [+N] feature. P-like elements such as kara "from", made "as far as, u n t i l " , (h)e " i n the d i r e c t i o n of, towards" share with English prepositions the property that the phrases they head can serve as predicates, (See Appendix B on no-marked phrases for examples) and can modify nouns, whereas those headed by p a r t i c l e s such as ni (locative marker, dative) cannot. Datives seem to postpose more r e a d i l y than subjects and objects. I cannot o f f e r any explanation here of why t h i s might be so. As far as the remaining items i n the l i s t are concerned -- adverbs and various c l a u s a l elements, we might argue that they a l l lack the [+N] feature and thus w i l l not be blocked when they cross the [+N] head of the r e l a t i v e clause. We need to c r u c i a l l y assume that clauses headed by A:oto and no are CP's and not DP's and that "koto" and "no" are C s , not N's, as Whitman (1991a) suggests, even though they are case-marked l i k e nominal elements. The only other kind of element that remains to be discussed i s genitive phrases. We saw i n (127) that genitives can also be postposed out of r e l a t i v e clauses. If genitives are DP's l i k e subjects and objects, we would expect them to have a [+N] feature and to encounter the same kind of e f f e c t as we saw for nominative or accusative DP's when they are postposed out of a DP. But contrary to such expectations, we f i n d in the examples below that genitives can be postposed out of r e l a t i v e clauses (142), DP's (143) and other genitives (144). Thus genitives appear to be mysterious i n that they do not t r i g g e r a [+N] domain e f f e c t , even though we would expect them to be [+N] categories. 104 (127) repeated as (142) t [ei omotya-o] kowasita akatyan-ga] nigeta, [ano-ko-no] ±. toy-acc broke baby-nom ran-away that-child-gen The baby who broke the toy ran away -- that c h i l d ' s (toy). (143) inu-ga [ei neko-o] oikaketa, watasi-nOi dog-nom cat-acc chased me-gen The dog chased the cat --my (cat). (144) [eL otooto-no] purin-o tabeta, itoko-nOi brother-gen pudding ate cousin-gen I ate (his) brother's pudding -- my cousin's (brother's). In Appendix B I show that the p a r t i c l e no, which marks what we c a l l g e nitives i n Japanese also serves other functions: for example as a complementizer and as a form of the copula. It i s possible that t h i s p a r t i c l e serves as a kind of default functional head that serves a v a r i e t y of functions and i s as a r e s u l t unmarked for a [+N] or [-N] feature. Such a p o s s i b i l i t y would account for the fact that genitives group with r e l a t i v e clauses when they are postposed out of a [+N] XP. If no-marked phrases do not have a [+N] feature, then we should expect to f i n d only marginal deviance when we postpose a r e l a t i v e clause out of a no-marked phrase, just as we found when a no-marked phrase was postposed out of a r e l a t i v e clause i n (127) . Yet in the following sentences, greater deviance i s encountered than we might expect. ?*(145) [ei Taroo]-no hon-o nakusite-simatta, [ t o n a r i - n i sunde-iru] t. -gen book-acc lost-aux. next-door-loc l i v e I regret that I l o s t Taro's book -- (Taro) who l i v e s next door. 105 Are we seeing a SC e f f e c t here for the genitive, which may act l i k e a subject of the DP "book"? Such a hypothesis i s contradicted by the fact that i t i s p ossible to postpose a genitive out of a genitive, as we saw above i n (144). I leave t h i s problem unsolved. 3.2 Postposing out of other c l a u s a l structure with no T+N1 head: no [+N] domain e f f e c t s when DP's are postposed In the previous section we saw that severe deviance occurs when a category that we could po s i t to have a [+N] feature i s postposed out of another [+N] category. In the examples that follow i n t h i s section, I s h a l l look at what happens when we postpose various types of constituents out of a category that i s arguably not [ + N ]. As we might predict, we f i n d that the [+N] domain e f f e c t does not occur. In the absence of the e f f e c t we observed i n the previous section, we s h a l l f i n d that arguments postpose out of non- [ + N ]-islands better than adjuncts do: for example, when the i s l a n d i s a concessive clause or wh-island, objects seem to postpose better than r e l a t i v e clauses. This i s what we would predict to happen i n the absence of the blocking e f f e c t of [+N] categories on other [+N] categories, since arguments generally are more extractable than adjuncts. I s h a l l consider postposing from two types of isl a n d s : sec. 3.2.1 Postposing out of a wh-island sec. 3.2.2 Postposing out of a concessive clause 3.2.1 Postposing out of a wh-island In t h i s subsection we see that an object w i l l postpose better than a r e l a t i v e clause out of a wh-island: 106 3.2.1.1 r e l a t i v e clause postposed (146) ?*Masao-ga, ano okane-o nusunda ka-doo-ka wakaranai, boku-ga kinoo moratta. -nom that money-acc st o l e comp not-clear I-nom yesterday received I t ' s not c l e a r i f Masao sto l e that money -- that I received yesterday. 3.2.1.2 heavy object postposed (147) ??Masao-ga, nusunda ka-doo-ka wakaranai, ano boku-ga kinoo moratta okane-o -nom st o l e comp not-clear that I-nom yesterday received money-acc I t ' s not c l e a r i f Masao s t o l e ( i t ) -- that money that I received yesterday. 3.2.2 Postposing out of a concessive clause S i m i l a r l y when we construct a sentence i n which an element i s postposed out of a concessive clause, which as an adjunct has an i s l a n d e f f e c t , we f i n d again that an object w i l l postpose better than a r e l a t i v e clause: 3.2.2.1 r e l a t i v e clause postposed (148) ?*Masao-wa, okane-o nusunda no-ni, sikararenakatta, boku-ga kinoo moratta. -top money-acc st o l e although wasn't-scolded I-nom yesterday received In spite of the fact that Masao stole money, i t ' s a l l rig h t -- (money) that I received yesterday. 107 3.2.2.2 object postposed (149) ??Masao-wa, nusunda no-ni, sikararenakatta, boku-ga kinoo moratta okane-o. -top s t o l e although scold-pass-neg-past I-nom yesterday received money-acc In s p i t e of the fact that Masao st o l e ( i t ) , he wasn't scolded -- (money) that I received yesterday. 3.2.2.3 subject postposed It i s not c l e a r that subjects w i l l postpose out of islands as well as objects. Endo (1989:142) judges the following sentence to be bad. (150) * a s i t a aru no-ni, Ken-wa asonde-iru yo, siken-ga. tomorrow ex i s t although -top play emph exam-nom In s p i t e of the fact that i t ; i s tomorrow, Ken i s playing around -- the exanij. It i s possible that a contributing factor to the deviance of t h i s sentence i s the fact that the postposed phrase i s not "heavy" . Heavy phrases seem to postpose better than bare DP's with no c l a u s a l modification. (See chapter 5 for further examples.) 3.3' Conclusions The fact that i t i s easier to postpose an object than a r e l a t i v e clause out of an i s l a n d such as a concessive clause or a wh-island i s what we would expect since a r e l a t i v e clause has one further bounding node to pass through -- the DP that i t modifies. In the absence of a blocking of [+N] by [+N] e f f e c t i n these 108 examples we expect that an object should be easier to extract than a r e l a t i v e clause that modifies i t . Notice that the e f f e c t s we are seeing i n t h i s chapter are more l i k e subjacency e f f e c t s than ECP e f f e c t s . In the cases where i t i s easier to postpose a r e l a t i v e clause than an object, the contrast cannot have anything to do with whether the trace of the postposed element i s properly governed, since the trace of an object must be l e x i c a l l y governed by V. Furthermore, when a r e l a t i v e clause i s postposed and only mild deviance r e s u l t s , we must assume that i t s trace i s i n some way properly governed. In chapter 5 I w i l l show evidence that the trace of a postposed r e l a t i v e clause or genitive acts as i f i t i s governed by the V that also governs the DP that the genitive or r e l a t i v e clause modifies. Another notable fact about the data in t h i s chapter i s that islands such as r e l a t i v e clauses and adverbial clauses are only weak islands for postposing: i t i s s t i l l possible, under the righ t circumstances, to postpose elements out of them. The fact that we see gradations of marginal deviance i n many of these examples suggests subjacency rather than ECP e f f e c t s . In the next subsection I w i l l compare the i s l a n d strength of d i f f e r e n t types of c l a u s a l elements. 3.4 A r e l a t i v e clause as a weaker i s l a n d than other kinds of adjuncts When we factor out p a r t i c u l a r l o c a l i t y constraints such as the subject condition and blocking of extraction of [+N] by [+N], some elements w i l l postpose more e a s i l y out of a r e l a t i v e clause than out of other types of adjunct clauses. 3.4.1 adverbs postposed: In the following examples I w i l l compare the e f f e c t s of postposing an adverb out of e i t h e r a r e l a t i v e clause or a complement clause headed by koto. We 109 s h a l l see that the former case r e s u l t s in mild deviance and the l a t t e r i n a grammatical sentence. 3.4.1.1 adverb out of r e l a t i v e clause (151) purin-o tabeta otoko-no-ko-ga asoko da, musyamusya to. pudding-acc ate boy-nom over-there copula ravenously The boy who ate the pudding i s over there -- ravenously. 3.4.1.2 adverb out of koto clause (152) Taroo-ga kuru koto-ga akiraka da, moo sugu. -nom come comp-nom clear copula already soon It i s c l e a r that Taro w i l l come -- in a l i t t l e while, (judgement: not bad but worse than to clause or koto clause out of koto clause) In t h i s subsection I compare the e f f e c t s of postposing a non-case-marked CP out of a r e l a t i v e clause, a wh-island, a temporal adverb clause, and a complement case-marked CP. Once again we s h a l l see that the complement CP permits grammatical postposing and that the r e l a t i v e clause i s l a n d causes only mild deviance: s l i g h t l y less than a wh-island and d i s t i n c t l y less than a temporal adverb clause. 3.4.2 to-clause postposed 3.4.2.1 to clause out of r e l a t i v e clause (153) ?Taroo-ni i t t a sensei-ga kaetta, siken-ga enki-sareta to. -dat said teacher-nom went-home exam-nom postponed comp The teacher who said (it;) to Taro went home -- [that the exam i s postponed];. 110 3.4.2.2 to clause out of wh-island (154) ?Taroo-ga i t t a ka-doo-ka wakaranai, a s i t a konai to. -nom sa i d comp not-clear tomorrow not-come comp It' s not cl e a r i f Taro said (it) -- that he wouldn't come. (155) ??Taroo-ga i t t a ka-doo-ka wakaru no, a s i t a konai to? -nom said comp clear comp tomorrow not-come comp Do you know i f Taro s a i d (it) -- that he wouldn't come? 3.4.2.3 to clause out of temporal clause (156) *? Taroo-ga i t t e - k a r a , totuzen nigedasita, a s i t a konai to. -nom say-after suddenly ran-off tomorrow come-not comp Aft e r Taro said ( i t j , he suddenly ran off -- [that he wouldn't come tomorrow] i . 3.4.2.4 to clause out of koto clause (157) Taroo-ga i t t a koto-ga akiraka da, a s i t a konai to. -nom said comp-nom clear copula tomorrow comp What Taro said i s cl e a r -- that he w i l l come tomorrow. hierarchy of i s l a n d strength when a to clause i s postposed: The following chart summarizes the r e l a t i v e strength of i s l a n d e f f e c t s induces by each type of p o t e n t i a l i s l a n d we looked at i n the above examples. I l l GREATER DEVIANCE temporal clause > r e l a t i v e clause, wh-island > koto clause LESS DEVIANCE 3.4.3 Jcoto-clause postposed In t h i s subsection, I look at the e f f e c t s of postposing a case-marked CP out of a r e l a t i v e clause, reason clause, wh-island, and complement CP. Once again, the complement CP shows no i s l a n d e f f e c t s , but neither does the r e l a t i v e clause. The wh-island shows stronger i s l a n d e f f e c t s , and the reason clause shows s t i l l greater i s l a n d e f f e c t s . 3.4.3.1 koto clause out of r e l a t i v e clause (158) judgement:"almost perfect" s i t t e - i r u hito-ga sukunai, siken-ga enki-sareta koto-o. know person-nom few exam-nom postponed comp-acc The people who know are few -- that the exam has been postponed. 3.4.3.2 koto clause out of reason clause (159) ? * s i t t e - i r u kara, kyoo-wa asobanai, a s i t a siken-ga aru koto-o know because today play-neg tomorrow exam-nom e x i s t comp-acc Because I know ( i t ; ) , I'm not playing around today -- [that there's an exam tomorrow] t. (160) ?* himitu da-kara, sinpai s i n a i , Taroo-ga s i t t e - i r u koto-ga. secret copula-because worry do-not -nom know comp-nom Because i t j ' s a secret, I'm not worried -- [the fact Taro knows];. 112 3.4.3.3 koto clause out of wh-island (judgement: a b i t worse than extracting a to clause out of wh-island but s t i l l better than extracting a r e l a t i v e clause out of an object) (161)??Taroo-ga e; s i t t e - i r u koto-ga himitu ka-doo-ka wakaranai, [Masao-ga uso-o-tuita koto-o];. -nom know comp-nom secret comp not-clear -nom l i e - a c c - t o l d comp-acc It' s not cl e a r whether the fact that Taro knows i s a secret -- (that he knows that) Masao t o l d l i e s . (162)??Taroo-ga e; s i t t e - i r u koto-ga himitu ka-doo-ka wakaru no, [Masao-ga uso-o-tuita koto-o];?. -nom know comp-nom secret comp cl e a r comp -nom l i e - a c c - t o l d comp-acc Do you know whether the fact that Taro knows i s a secret -- (that he knows that) Masao t o l d l i e s ? 3.4.3.4 koto clause out of koto clause (163)Taroo-ga s i t t e - i r u koto-ga akiraka da, Masao-ga uso-o-tuita koto-o. -nom know comp-nom clear copula -nom l i e - a c c - t o l d comp-acc It i s c l e a r that Taro knows -- that Masao was t e l l i n g l i e s . hierarchy of i s l a n d strength when a koto clause i s postposed: GREATER DEVIANCE reason clause > r e l a t i v e clause > wh-island > koto clause LESS DEVIANCE 113 3.4.4 s i - n i purpose clause postposed In t h i s subsection, a case-marked bare VP that expresses purpose i s postposed out of the same types of p o t e n t i a l islands as we saw i n the immediately preceding subsection. Once again, the r e l a t i v e clause and the complement CP show no i s l a n d e f f e c t s ; the wh-island and the reason clause show very mild i s l a n d e f f e c t s i n t h i s case. 3.4.4.1 s i - n i purpose clause out of r e l a t i v e clause (164) iku hito-ga sukunai, eiga-o mi-ni. go person-nom few film-acc see-loc The people who are going are few -- to see a f i l m . 3.4.4.2 s i - n i purpose clause out of koto clause (165) Taroo-ga i t t a koto-o s i t t e - i r u , eiga-o mi-ni. -nom went comp-acc know film-acc see-loc I know that Taro went -- to see a f i l m . 3.4.4.3 s i - n i purpose clause out of reason clause (166) ?Hanako-ga i t t e - k i t e kara, bangohan-o ryoori-suru, yasai-o k a i - n i . -nom went because supper-acc cook vegetables-acc buy-loc Because Hanako went e i ( I ' l l cook dinner -- [to buy veg e t a b l e s ] ^ 114 3.4.4.4 si-ni purpose clause out of wh-island (167) ?Taroo-ga i t t a ka-doo-ka wakaranai, eiga-o mi-ni. -nom went comp not-clear film-acc see-loc I t ' s not cl e a r i f Taro went -- to see a f i l m . (168) Taroo-ga i t t a ka-doo-ka wakaru no, eiga-o mi-ni? -nom went comp clear comp film-acc see-loc Do you know i f Taro went -- to see a film? hierarchy of i s l a n d strength for postposing a si-ni purpose clause: reason clause, wh-island > r e l a t i v e clause, koto clause 3.4.5 Summary of comparison of i s l a n d e f f e c t s Although the patterns of r e l a t i v e grammaticality d i f f e r s l i g h t l y - f o r each type of element that i s being postposed, i n general a r e l a t i v e clause appears to be a weaker i s l a n d than a reason clause and in some cases i s no stronger an i s l a n d than a koto clause. This evidence requires us to consider the p o s s i b i l i t y that even i f r e l a t i v e clauses are adjuncts, we need to discriminate between d i f f e r e n t types of adjuncts. It would be useful to test other possible types of adjuncts as well: for example VP-level adjuncts versus sentence-level adjuncts. How might we explain the apparent weakness of r e l a t i v e clauses i n Japanese as a i s l a n d with respect to postposing? One p o s s i b i l i t y i s that r e l a t i v e clauses are adjoined to N' or D' rather than to DP or NP; Under the minimalist framework of Chomsky (1995), t h i s would mean that r e l a t i v e clauses are i n the minimal domain of the head to which they are related, whereas elements that are adjoined to an XP are not i n the minimal domain of head X. Another factor that may determine the r e l a t i v e i s l a n d strength of adjuncts i s whether the p r o j e c t i o n to which they are 115 adjoined i s l e x i c a l . If r e l a t i v e clauses are adjoined to N' rather than D', they are i n the minimal domain of a l e x i c a l head. The various po s s i b l e combinations of these types of p o s s i b i l i t i e s suggest a complex array of hierarchies of islandhood. For example, a complement of a l e x i c a l head i s usually considered not to be a b a r r i e r for extraction. But we might want to po s i t that a phrase i n the minimal domain of a l e x i c a l head i s a weaker b a r r i e r than one i n the minimal domain of a functional head, which might be the case for a sentence l e v e l adverb phrase. One piece of evidence that r e l a t i v e clauses are attached lower than a adjoined-to-DP p o s i t i o n i s the fact that they can follow determiner-like elements such as ano, kono, and sono. (169)ano Masao-ga kinoo yonda hon -nom yesterday read book that book that Masao read yesterday Another possible explanation for the weakness of r e l a t i v e clauses as islands for extraction i s that although they are adjuncts, they can be licensed i n a way s i m i l a r to L-marking by a l e x i c a l [+V] head i f they are i n i t s domain and i f no [+V] b a r r i e r intervenes. This idea i s proposed i n chapter 5. 4. Postposing out of wa-marked phrases In t h i s chapter I look at the e f f e c t s of postposing out of t o p i c a l i z e d phrases marked by the topic-marker wa and some r e l a t e d issues that a r i s e as a re s u l t of t h i s a n a lysis. In general, postposing out of iva-marked c l a u s a l constituents (e.g. CP's) i s l i c i t whereas postposing out of wa-marked non-clausal constituents (such as DP's) i s bad, although there are some exceptions to t h i s pattern. I attempt to explain these differences by p o s i t i n g that there i s a necessary syn t a c t i c d i s t i n c t i o n between d i f f e r e n t types of adjuncts -- those that 116 are i n the domain of a l e x i c a l head, and those that are not. I s h a l l propose that there i s the following l i c e n s i n g condition on traces of postposed phrases: 1. A trace t must be licensed i n Max(x) where x i s [+V]. 2. t i s licensed by x i f f x i s l e x i c a l In the f i r s t subsection I s h a l l d i s t i n g u i s h two sets of data i n which a phrase i s postposed out of a wa-marked topi c : a deviant set i n which the trace of the postposed phrase i s not i n the domain of a l e x i c a l head and a grammatical or less deviant set i n which the trace of the postposed phrase i s i n the domain of a l e x i c a l head. In Chapter 1 I argued that where i t i s possible to test for Condition C reconstruction e f f e c t s on a name embedded i n a postposed phrase, these e f f e c t s occur and are best explained by p o s i t i n g movement of postposed phrases. A movement hypothesis i s further strengthened by subject-condition e f f e c t s and r e l a t i v i z e d minimality e f f e c t s on postposed phrases that were seen i n chapters 2 and 3. In t h i s chapter I s h a l l continue to pursue a movement hypothesis for postposed phrases. I s h a l l also adopt Saito (1985)'s hypothesis that, with the exception of t o p i c a l i z e d l o c a t i v e s of the form XP-ni-wa, which he shows cannot be base-generated (discussed i n Appendix B) topics i n Japanese are e i t h e r base-generated i n a high functional p r o j e c t i o n or moved to that p o s i t i o n by A' movement. For the purposes of t h i s discussion I s h a l l assume that t h i s p o s i t i o n i s [Spec, CP]. This means that topics are not "L-marked" i n Chomsky (1986) terms: therefore, they should be p o t e n t i a l islands for extraction. Yet when we test for the islandhood of topics, we f i n d that in some cases topics behave as islands and i n other cases they do not, as the examples that follow w i l l 117 i l l u s t r a t e . See sec. 4.5. for a b r i e f discussion of Saito's a n a l y s i s . 3 7 3 8 I am assuming for the discussion i n t h i s chapter that genitives are adjoined to some proje c t i o n of D. For the following reasons, I have not adopted ei t h e r of the hypotheses that (a) genitives are complements of N or that (b) they are i n [Spec, DP]. Except i n cases where N i s a noun such as "study", that can c l e a r l y take an argument (see sec. 4.2.1 below), a genitive modifying a DP can exchange p o s i t i o n with a r e l a t i v e clause modifying the same DP, as we see i n examples (170) and (171) . (170) Taroo-no [Masao-ga kinoo yonda] hon -gen -nom yesterday read book Taro's book that Masao read yesterday (171) [Masao-ga kinoo yonda] Taroo-no hon -nom yesterday read -gen book Taro's book that Masao read yesterday. This fact would not be explainable i f genitives were arguments of N. The hypothesis that genitives are in a s p e c i f i e r p o s i t i o n i s not consistent with the fact that genitives do not show SC e f f e c t s when an element i s extracted from them, as we saw above i n example (144) i n sec. 3.1.1.7, repeated here as (172) : 37 S a i t o (1985) a r g u e s t h a t b a s e - g e n e r a t e d t o p i c s do n o t b e h a v e a s i f t h e y a r e d e r i v e d b y movement when t h e c a n o n i c a l p o s i t i o n w i t h w h i c h t h e t o p i c i s a s s o c i a t e d i s w i t h i n a n i s l a n d : i . e . t h a t i s l a n d d o e s n o t show i s l a n d e f f e c t s w i t h r e s p e c t t o t h e t o p i c . F o r e x a m p l e he a r g u e s t h a t a t o p i c c a n be a s s o c i a t e d w i t h a p o s i t i o n i n s i d e a s u b j e c t - h e a d e d r e l a t i v e c l a u s e . By c o n t r a s t , we saw i n c h a p t e r 2 t h a t a s u b j e c t - h e a d e d r e l a t i v e c l a u s e i s a s t r o n g b a r r i e r f o r p o s t p o s i n g . T h i s c o n t r a s t b e t w e e n t o p i c a l i z a t i o n a n d p o s t p o s i n g i n J a p a n e s e c o u l d be e x p l a i n e d i f t o p i c s a r e b a s e - g e n e r a t e d (as a r g u e d i n S a i t o (1985)) a n d p o s t p o s e d p h r a s e s a r e d e r i v e d by m o v e m e n t . 38 Some of the following examples are modelled after examples in Endo (1989). 118 (172) [e; otooto-no] purin-o tabeta, itoko-no; brother-gen pudding ate cousin-gen I ate (his) brother's pudding -- my cousin's (brother's). Accordingly, I s h a l l assume, unless otherwise noted, that the genitives i n the examples i n t h i s chapter have the following structure. (173) DP I D' / \ gen. D' / \ I am assuming adjunction to an intermediate p r o j e c t i o n of D rather than to DP since i f the configuration of the trace of a postposed genitive modifying a DP topic were as follows, postposing of the genitive w i l l not cross any maximal projections, since DP does not dominate the trace. In examples we s h a l l see below, postposing a genitive i n t h i s configuration i s ungrammatical. If the gen i t i v e were instead adjoined to D' as i n the structure above, DP w i l l be a non-L-marked b a r r i e r to extraction of the genitive under the model of Chomsky (1986) since (a) i t dominates the trace but not i t s antecedent and (b) i t i s not theta-marked by a l e x i c a l category. This i s the kind of explanation we need to maintain i n order to account for the ungrammaticality of sentences l i k e (176) below. 119 (174) CP / \ CP g e n i t i v e i / \ (topic)DP C / \ \ DP IP For convenience, i n the following discussion, I s h a l l r e f e r to topics that are construed with a subject p o s i t i o n as "subject topics", and those that are construed with an object p o s i t i o n as "object topics." 4.1 Are topics islands for extraction? As the following examples show, severe deviance r e s u l t s when we postpose an adjunct modifier of a DP wa-marked topi c . The postposed element i n the examples below i s e i t h e r a no-marked phrase or a r e l a t i v e clause that modifies the head noun of the topic DP. (The ( i i ) variant of each sentence i s the non-postposed version.) The following diagram shows the relevant structure. 120 (175) CP / \ CP XPi ( r e l a t i v e clause or genitive) / \ (topic)DP I D' / \ / _ \ t i D' / \ (176) (i) * [eL kuruma-wa] boku-o h i i t a , Taroo-no ; car-top me-acc ran-over -gen As for the car, i t ran over me, Taro's (car) ( i i ) [Taroo-no kuruma-wa] boku-o h i i t a . -gen car-top me-acc ran-over As for Taro's car, i t ran over me. (177) (i) *[ei purin-wa] tabeta, Taroo-no^ pudding-top ate -gen As for the pudding, I ate i t -- Taro's (pudding). ( i i ) [Taroo-no purin-wa] tabeta. -gen pudding-top ate As for Taro's pudding, I ate i t . (178) (i) *[ei kookyuusya-wa] taihen takai, C[,[Nihon-de tukutta] A. luxury-car-top very expensive Japan-loc made Luxury cars are very expensive -- (ones that are) made i n Japan. C \ IP 121 ( i i ) [ C P[Nihon-de tukutta] kookyuusya-wa] taihen takai yo. Japan-loc made luxury-car-top very expensive emph Luxury cars that are made i n Japan are very expensive. Yet when a constituent of a CP topic, such as a phrase headed by koto-wa or no-wa i s postposed, the sentence i s grammatical, or marginal at worst as i n the case of (181) . (179) C P [dareka-ga e; tabetesimatta koto]-wa, t a s i k a da , ano wagasi-o. someone-nom ate-up comp-top c e r t a i n copula that sweet-acc It ' s c e r t a i n that someone ate.(it;) up -- that sweet;. (180) C P[dareka-ga ei aketa to i u no]-wa, hontoo-rasii, [ano kinko-o];. someone-nom opened comp say comp-top true-modal safe-acc The fact that someone opened (it;) seems to be true -- the safe;. (181) ? c p[dareka-ga e; aketa koto]-wa keisatu-ni tutaeta, [ano kinko-o];. someone-nom opened comp-top police-dat t o l d that safe-acc As for the fact that someone opened ( i t ; ) , I t o l d the p o l i c e -- the saf e<. 122 (182)structure of (179)-(181) CPi / \ CPX XPi (postposed phrase) / \ (topic)CP 2 C / \ IP, IPi / \ VP V / \ t i V Examples (179) and (180) suggest that the topic p o s i t i o n i t s e l f i s not an absolute b a r r i e r for extraction of a postposed phrase. Notice that i n (178) the topic i s an object topic, as i s also the case i n (181) , which i s much less deviant than (178). Because, i n (178), postposing out of an object topic r e s u l t s i n severe deviance, the deviance of (176) and (177) i s not necessar i l y due to the fact that there the topic i s a subject to p i c . There are other differences between (176)-(178) and (179)-(181) that suggest a possible explanation for the contrast i n grammaticality between the two sets. One i s that the postposed phrases i n the deviant set are r e l a t i v e clauses or no-marked phrases -- l i k e l y adjuncts, whereas the postposed phrases i n the non-deviant set are arguments. Another diffe r e n c e i s that the .phrases that are t o p i c a l i z e d i n the deviant set are headed by l e x i c a l nouns whereas those i n the non-deviant set are headed by koto or no which are arguably non-lexical complementizers. The following examples, however, r u l e out both the question of 123 whether the t o p i c a l i z e d phrase i s l e x i c a l and the adjunct-argument status of the postposed phrase as possible factors that account for the contrast i n deviance between the two above sets of examples. F i r s t of a l l , the following three sentences show that postposing a r e l a t i v e clause out of a phrase headed by koto i s grammatical. (183) Masao-wa, okane-o nusunda koto-ga akiraka da, kinoo moratta. -top money-acc st o l e comp-nom cl e a r copula yesterday received It ' s c l e a r that Masao st o l e the money that I received yesterday. (184) Masao-wa, okane-o nusunda koto-o s i t t e - i r u , kinoo moratta. -top money-acc st o l e comp-acc know yesterday received I know that Masao st o l e the money -- that I received yesterday, (or) Masao knows that (they) s t o l e the money -- that I/he received yesterday. (185) Masao-wa, okane-o nusunda koto-ga aru, boku-no kakusita. -top money-acc stole comp-nom exist I-gen h i d Masao has (on occasion) stolen money -- that I hid. When we wa-mark the complementizer A:oto i n (183) we f i n d the sentence i s s t i l l marginal at worst: (186) ?Masao-ga, [et okane-o] D P nusunda koto-wa akiraka da, [kinoo moratta];. -nom money-acc st o l e comp-top clear cop. yesterday received I t ' s c l e a r that Masao s t o l e the money -- that I received yesterday. When we postpose a genitive instead of a r e l a t i v e clause out of a iva-marked koto clause the r e s u l t i s s t i l l grammatical: 124 (187) dareka-ga [ei kinko-o] D P aketa koto-wa, tasika da, [ano ginkoo-no];. someone-nom safe-acc opened comp-top c e r t a i n cop. that bank-gen It's c e r t a i n that someone opened the safe -- that bank's (safe). Thus we f i n d that i n (186) and (187) postposing a ge n i t i v e or r e l a t i v e clause out of a wa-marked phrase i s much better than i t i s i n (176) or (179). In (188) a clause s i m i l a r to the one. i n (187) i s headed by a l e x i c a l noun rather than koto or no, and the sentence i s s t i l l grammatical: (188) 0 P[dareka-ga e A aketa to yuu uwasa]-wa, hontoo datta, [ano kinko-o];. someone-nom opened C say rumour-top true copula-past that safe-acc The rumour that someone opened (it;) i s true -- the safe;, (good) 3 9 If we postpose a r e l a t i v e clause (189) or a no-marked phrase (190) instead of an argument DP out of a topic headed by a l e x i c a l noun, the sentence i s marginal, but s t i l l much better than any of (176)-(178). (189) ??Masao-ga, [e; okane-o] nusunda to yuu uwasa-wa hontoo da, [boku-ga kinoo moratta] L. -nom money-acc sto l e C say rumour-top true copula I-nom yesterday received The rumour that Masao stole the money i s true -- (the money) that I received yesterday. ' The fact that this sentence is judged as good seems to contradict data in Chapter 3 on relativized minimality effects. We should expect this sentence to suffer a relativized minimality violation as a result of the DP being postposed out of the lexical DP topic. I cannot explain the lack of relativized minimality effect in this sentence. 125 (190) ??dareka-ga [ei kinko-o] D P aketa to yuu uwasa-wa, hontoo da, [ano ginkoo-no] L. someone-nom safe-acc opened C say rumour-top true cop. that bank-gen It' s c e r t a i n that someone opened the safe -- that bank's (safe). Thus the deviance of (17 6) - (178) does not appear to be due to the fact that i n those examples, (a) the topic i s a l e x i c a l noun, and (b) the postposed phrase i s a genitive or r e l a t i v e clause. 4.2 Discriminating between d i f f e r e n t types of adjuncts In the examples that we looked at in the previous section we saw that in (176)-(178), when a genitive or r e l a t i v e clause was postposed out of a DP topic that they modified, the r e s u l t was severe deviance. On the other hand, in examples (179)-(190), when a constituent of a clause within a topic was postposed, no deviance or only marginal deviance resulted. I s h a l l now propose a difference between the two sets that accounts for t h e i r contrast i n deviance. Notice that the r e l a t i v e clause or no-marked phrase that i s postposed i n (176)-(178) i s arguably an adjunct of the wa-marked phrase whereas i n a l l the examples i n the second set i n which a r e l a t i v e clause or no-marked phrase i s postposed, the postposed element i s an adjunct of the object of the lower clause. The contrast i n grammaticality between (176)-(178) and (186), (187), (189), (190) suggests that adjuncts of arguments and adjuncts of non-arguments behave d i f f e r e n t l y with respect to t h e i r e x t r a c t a b i l i t y . I cannot give a precise d e f i n i t i o n of how we might characterize these kinds of d i f f e r e n c e s . A large proportion of syntactic research i n the generative t r a d i t i o n has been concerned with the behaviour of arguments rather than adjuncts: we do not have a well-developed theory of adjuncts. However, one 126 p o s s i b l e way of d i s t i n g u i s h i n g t h e s t r u c t u r e s o f t h e d e v i a n t s e t from t h e non-d e v i a n t s e t i s as f o l l o w s . I n t h e d e v i a n t examples ( 1 7 6 ) - ( 1 7 8 ) , t h e c a n o n i c a l p o s i t i o n o f XP, t h e p o s t p o s e d p h r a s e , i s not i n t h e domain o f a l e x i c a l head, as we see i n d i a g r a m (191) below. I use t h e term domain i n t h e sense o f Chomsky ( 1 9 9 5 ) . 4 0 I have been assuming t h a t g e n i t i v e s a r e a d j o i n e d t o D'. I n t h e diagrams below, I show t h a t as l o n g as t h e g e n i t i v e i s a t t a c h e d a t l e a s t as h i g h as an a d j o i n e d - t o - N P p o s i t i o n , i t w i l l be o u t s i d e t h e domain o f l e x i c a l N. The f o l l o w i n g two diagrams show t h e extremes o f t h e p o s s i b l e range o f att a c h m e n t s i t e o f t h e g e n i t i v e t h a t w i l l be i n a c c o r d w i t h t h e h y p o t h e s i s t h a t i t i s not i n t h e domain o f N. As d i s c u s s e d above, I am assuming i t i s not a d j o i n e d t o DP i n o r d e r t o e x p l a i n t h e f a c t t h a t t h e r e a p p ears t o be a b a r r i e r t o a n t e c e d e n t government o f t h e t r a c e i f we f o l l o w t h e model o f Chomsky (1986). (191)(a) CP ( l o w e s t p o s s i b l e / \ at t a c h m e n t ( t o p i c ) D P j C s i t e ) I / \ NP I P C / I I XP NP VP I / \ proi V o r t i See chapter 5 for an elaboration of the basic principles of Chomsky (I995)'s concept of domains. 127 (191)(b) CP ( h i g h e s t p o s s i b l e / \ att a c h m e n t ( t o p i c ) D P : C s i t e ) | / \ D' I P C / I I XP D' VP I / \ proL V o r t i As l o n g as XP, t h e g e n i t i v e o r r e l a t i v e c l a u s e t h a t m o d i f i e s t h e DP t o p i c i s a t t a c h e d no lo w e r i n th e D -P e x t e n d e d p r o j e c t i o n t h a n a p o s i t i o n a d j o i n e d t o NP, i t w i l l s t i l l be o u t s i d e t h e domain o f l e x i c a l N.41 It w i l l a l s o be o u t s i d e t h e domain o f m a t r i x V, i f we assume t h a t ( a ) , t o p i c s a r e a t t a c h e d t o a h i g h f u n c t i o n a l p r o j e c t i o n above IP and ( b ) , t h a t contra Whitman (1991a), V does not r a i s e t o C i n J a p a n e s e . 4 2 On t h e o t h e r hand, i n t h e g r a m m a t i c a l o r m a r g i n a l l y d e v i a n t examples (179 ) -(190), t h e t r a c e o f t h e p o s t p o s e d p h r a s e i s i n t h e domain o f a l e x i c a l head, namely V, as we see i n di a g r a m (192) below. I shall propose in sec. 5.2.1 immediately below that, based on the non-postposability of argument genitives out of topics, the trace of the postposed phrase must be in the domain of a head with [+V] features. If this is true, it will not matter if the genitive or relative clause is in the domain of N. 4 2See also Dechaine (1996) who argues that there is no head movement in Japanese. Yoon (1994) makes a similar argument for Korean. 128 (192) VP \ V / \ DP V D' / \ (trace of postposed phrase)- t D' In (192). we do not necessarily want to say that the s i t e of the postposed phrase i s head-governed. We saw in chapters 2 and 3 that r e l a t i v e clauses and genit i v e s act l i k e weak islands for postposing -- that i s , they behave l i k e adjuncts. We appear to need a more subtle apparatus for d i s t i n g u i s h i n g between the adjunct/argument status of d i f f e r e n t types of elements. The data we looked at above suggest that adjuncts i n the domain of a l e x i c a l head are less adjunct-l i k e than those that are not i n the domain of a l e x i c a l head. 4 3 The d i f f e r e n c e i n grammaticality between the two sets of data can be explained by the hypothesis that i n the deviant set, the postposed phrase i s not i n the domain of a l e x i c a l head and must cross a non-L-marked b a r r i e r ( i . e . the to p i c ) , whereas i n the non-deviant or mildly deviant set the trace of the postposed phrase i s also separated from i t s antecedent by a non-L-marked b a r r i e r but the trace _is i n the domain of a l e x i c a l head. Notice that i n the examples i n chapter 2 i n which an adjunct such as a r e l a t i v e clause or genitive i s postposed out of an i s l a n d such as an adverbial clause or r e l a t i v e clause, i n each case where only marginal deviance resulted, the postposed r e l a t i v e or genitive was in the domain of a l e x i c a l V. Consider the Grlmshaw (1990) introduces the concept of "a-adjuncts", which behave like adjuncts in some ways and like arguments in others. A-adjuncts resemble arguments in that they are licensed by argument structure. 129 following examples from chapter 2, repeated here as (193) ,(194), (195), and (196). The trace of the postposed phrase i s i n the domain of V "stole" i n (193) and (194), and of V "ate" i n (195) and (196). Thus the fact that these sentences are only mildly deviant, despite the fact that an adjunct has been postposed out of an island, i s explained by our hypothesis. (193) ??Masao-wa, okane-o nusunda no-ni, sikararenakatta, Taroo-no. -top money-acc sto l e although scold-pass.-neg-past -gen In s p i t e of the fact that Masao sto l e money, he wasn't scolded-- Taro's (money) (194) ?Masao-wa, okane-o nusunda no-ni, sikararenakatta, boku-no kinoo moratta. -top money-acc sto l e although scold-pass.-neg.-past I-gen(subj) yesterday received In s p i t e of the fact that Masao sto l e money, he wasn't scolded -- (money) that I received yesterday. (195) ?purin-o tabeta otoko-no-ko-ga nigeta, boku-no tukutta. pudding ate boy-nom ran-away I-gen(subj.) made The boy who ate the pudding ran away -- (the pudding) that I made. (196) ?purin-o tabeta otoko-no-ko-o s i k a t t a , boku-no tukutta. pudding-acc ate boy-acc scolded I-gen made I scolded the boy who ate the pudding -- (the pudding) that I made. 130 4.2.1 Can we postpose an element out of a topic i f i t i s i n the domain of a l e x i c a l N? Consider the case of genitives that behave as arguments, such as those we see i n English examples such as "the c i t y ' s destruction" (genitive as theme argument) or "England's defeat of the French army" (agent argument)). Such genitives are arguably i n the domain of the N head that assigns them a theta-r o l e . In (197), N "study" would appear to assign a theta-role to DP "English". (197) eigo-no benkyoo English-gen study the study of English Yet when we t r y to postpose an "argument genitive" l i k e the one above out of a topic we f i n d that there i s no contrast i n grammaticality with postposing a "possessive" g e n i t i v e " : (198) ?*benkyoo-wa hajimeta, eigo-no. study-top began English As for the study (of i t ) , I began (it) -- English. (199) ?*hon-wa yonda, Taroo-no book-top read -gen As for the book, I read i t -- Taro's. If we accept a formulation of the ECP that requires a phrase to be e i t h e r head-governed or antecedent governed, then we should expect an argument genitive to be head-governed by N and thus extractable, but in the example above we see that i t i s not. These data are explained i f we r e f i n e our hypothesis to say that to be extractable from an island, an adjunct must be i n the domain of a l e x i c a l [+V] head. 131 To summarize t h i s section, I s h a l l formally state the proposed condition and review how i t distinguishes between two contrasting sentences i n which a phrase i s postposed out of a topi c . Consider again the following two examples: (200) * [ e n purin-wa] tabeta, T a r o o - n O j . pudding-top ate -gen As for the pudding, I ate i t -- Taro's (pudding). (201) ??[dareka-ga [e, kinko-o] n p aketa to yuu uwasa-wa], hontoo da, [ano ginkoo-no] 1. someone-nom safe-acc opened C say rumour-top true cop. that bank-gen It's c e r t a i n that someone opened the safe -- that bank's (safe). Let us state the condition on traces of postposed phrases as follows: 1. A trace t must be licensed i n Max(x) where x i s [+V]. 2. t i s licensed by x i f f x i s l e x i c a l Suppose that t h i s l i c e n s i n g i s done i n two steps 4 4: (a) choose the smallest domain of a [+V] head that contains t (b) check that x i s l e x i c a l In (200) above, the trace of the postposed g e n i t i v e i s not i n the domain of any [+V] head so i t i s not licensed. I originally proposed this condition in a slightly different way. Formulating it in two steps more straightforwardly handles cases like those seen below in sec. 4.8.1. Thanks to Hisatsugu Kitahara (p.c.) for a suggesting this two-step formulation. 132 (202)(structure of (200)) CP / \ (topic)DP C I I D' IP / I I t D' VP /-\ I V In (201), by contrast, the trace of the postposed phrase w i l l be i n the domain of l e x i c a l embedded V "opened", which w i l l l i c e n s e the trace: 133 (203) (structure of (201)) XP / \ (topic)DP X ' I I D' IP I I NP VP I I N' V / \ CP N I rumour I VP I V / \ DP V | opened D' / I t D' NP /_\ safe 134 Notice that the smallest Max(x) that contains t, where x i s [+V], i s the embedded VP. Since i t s head i s l e x i c a l , t i s licensed, accounting for the lesser degree of deviance of (201) . (The marginality of the sentence could be due to a subjacency v i o l a t i o n . ) 4.2.2 Can a postposed genitive or r e l a t i v e clause o r i g i n a t e i n the domain of an argument pro? Another matter we need to consider i s the generally accepted base-generation account of wa-marked topics in Japanese. If a wa-marked topic i s coindexed with a pro in an argument p o s i t i o n , i s i t possible that a postposed ge n i t i v e or r e l a t i v e clause that modifies a topic originates not i n a p o s i t i o n adjoined to the topic but instead, adjoined to the pro that i s coindexed with the topic? If the trace of such a postposed genitive or r e l a t i v e clause were adjoined to an argument pro, then we should expect i t to be governed by the V of which the pro i s an argument. The following i s the hypothetical structure: (204) CP / \ CP XP2 / \ (topic)DPj C / \ IP C VP / \ p r O i V / \ t 2 proi 135 N o t i c e t h a t i n t h e above s t r u c t u r e , t h e r e i s no way t o r e c o v e r t h e p o s i t i o n o f t h e t r a c e . T h e r e f o r e , I s h a l l assume t h a t t h i s s t r u c t u r e i s n o t a p o s s i b l e d e r i v a t i o n o f t h e p o s t p o s e d p h r a s e i n t h e above examples. 4.3 P o s t p o s i n g out of VP t o p i c s So f a r we have l o o k e d a t examples i n w h i c h an element i s p o s t p o s e d out o f a t o p i c a l i z e d DP o r CP. A n o t h e r k i n d of element t h a t can be wa-marked i s t h e " c o n j u n c t i v e " b a r e form of the v e r b t h a t o c c u r s i n VP t o p i c a l i z a t i o n as i n (205) : (205) [ano k i n k o - o n i - d o - t o ake] V P-wa s i - n a i . t h a t s a f e - a c c t w o - t i m e s - P open-top do-not As f o r o p e n i n g t h a t s a f e a t a l l , I don't do i t . I t i s n o t c l e a r whether t h i s form i s t h e i n f i n i t i v a l form o f t h e v e r b o r w hether i t i s some k i n d of n o m i n a l i z e d form. I t o n l y o c c u r s i n t h e t y p e of VP t o p i c a l i z a t i o n we see i n (205) above o r i n p u r p o s e c l a u s e s ( d i s c u s s e d i n s e c . 4.5) i n w h i c h t h e c o n j u n c t i v e form i s marked by l o c a t i v e marker ni.) We s h a l l f i r s t c o n s i d e r t o p i c a l i z e d VP's. I n (206) we see t h a t i t i s p o s s i b l e t o p o s t p o s e an o b j e c t out o f a t o p i c a l i z e d VP. The g r a m m a t i c a l i t y o f (206) c o u l d be e x p l a i n e d by t h e f a c t t h a t t h e t r a c e of t h e p o s t p o s e d o b j e c t i s l e x i c a l l y g o v e r n e d by t h e b a r e V "open", i f t h e V has [+V] f e a t u r e s i n s p i t e o f i t s p r o j e c t i o n b e i n g t o p i c a l i z e d . (206) n i - d o - t o ake-wa s i - n a i , ano k i n k o - o . t w o - t i m e s - P open-top do-not t h a t s a f e - a c c As f o r o p e n i n g ( i t ; ) a t a l l , I don't do i t , -- t h a t s a f e ; . I n (207) below, a V P - l e v e l a d v e r b i s p o s t p o s e d out o f a t o p i c a l i z e d VP. 136 (207) ?*[ei garasu-o kowasi] V P-wa s i t a , konagona-nii. glass-acc break-top d i d into-pieces As for breaking the glass, I d i d i t -- into pieces. The sentence i s judged as being moderately ungrammatical. The contrast between (206) and (207) could be explained i f the trace of the postposed adjunct i n (207) i s outside the domain of V "break" -- for example i n a p o s i t i o n adjoined to VP. There are several ways i n which t h i s might be the case. One i s that the trace of the postposed adverb i s adjoined to VP, where i t w i l l be outside the domain of V. But r e c a l l that we wish to r e s t r i c t adjunction to XP. We saw i n sec. 4.2. that we need to rule out adjunction of a r e l a t i v e clause or ge n i t i v e to the DP i t modifies i n order to explain the islandhood of DP topics i n cases l i k e (177). S i m i l a r l y , i f the trace of the postposed adverb i n (207) i s adjoined to VP, there w i l l be no b a r r i e r between i t and i t s postposed, unless a t o p i c a l i z e d VP contains other phonologically n u l l material above the VP, for which we have no d i r e c t evidence. A second explanation for the deviance of (207) i s that the postposed adverb i s a r e s u l t a t i v e . We f i n d i n other languages as well that r e s u l t a t i v e s r e s i s t being d i s l o c a t e d from the VP with which they are associated. 4 5 45In English, for example, resultatives contrast with other types of adverbs such as instrumentals in their ability to be stranded by VP ellipsis: (i) *John broke the glass into pieces and Mary did so into pieces too. (ii) John broke the glass with a hammer and Mary did so with a hammer too. A better example for testing postposability of adverbs from a topicalized VP might be the following: (iii) (172) [e, garasu-o kowasi]VP-wa sita, kanazuti-de,. glass-acc break-top did hammer-inst As for breaking the glass, I did it - with a hammer (no judgement yet on this example) See Dechaine (1993) for a discussion of violations similar to (i) above in English. 137 To conclude t h i s section on postposing out of VP topics I s h a l l present some data that further support the hypothesis that the trace of postposed phrase inside an i s l a n d must be i n the domain of a l e x i c a l [+V] head. Consider the following p a i r : 4 6 (208) (repeated here from sec. 4.2.1 above) *[ei benkyoo]-wa hajimeta, [eigo-no ] i . study-top began English-gen As for the study (of i t ) , I began ( i t ) -- English. (209) [ej benkyoo-o hajime]-wa s i t a [eigo-no ] i . study-acc begin-top d i d English-gen As for beginning to study ( i t ) , I d i d -- English. Despite the close s i m i l a r i t y between the two sentences, there i s a d i s t i n c t grammaticality contrast. In the f i r s t example, the trace i s not i n the domain of matrix V "began", as explained above. In the second example, i t i s i n the domain of bare V "begin", whose VP has been t o p i c a l i z e d . The contrast i s i n accord with our hypothesis. 4.4 Postposing out of non-case-marked CP topics CP's headed by the non-case-marked complementizer to can also be marked as topics with wa: l6These examples were suggested by Hisatsugu Kitahara (p.c). 138 (210) a s i t a Taroo-ga kuru to wa47, omoenai. tomorrow -nom come comp top think-potent.-neg I don't think that Taro w i l l come tomorrow. In (211) and (212) we f i n d that postposing a subject or object out of such phrases i s grammatical. (211) a s i t a kuru to wa, omoenai, Taroo-ga. tomorrow come comp top think-potent.-neg -nom I don't think that (hej w i l l come tomorrow -- Taro;. (212) dareka-ga akeru to wa, omoenai, ano kinko-o. someone-nom open comp top think-not that safe-acc I don't think that someone opened ( i t j -- that s a f e j Can we test whether postposing an element outside the domain of V out of t h i s kind of construction w i l l be ungrammatical, as our hypothesis predicts? We cannot strand the complementizer and postpose the IP. One possible candidate i s a p r o p o s i t i o n a l l e v e l adverb. We need to be sure that the adverb i s construed with the embedded rather than the matrix clause. Consider the following examples. (213) *[ei dareka-ga ano kinko-o someone-nom that safe-acc I don't think that i t w i l l be -- surely. a k e r u to ] C F , wa, o m o e n a i , k i t t O ; open comp t o p t h i n k - n o t s u r e l y the c a s e that someone w i l l open t h a t s a f e 4 7 B e c a u s e p o s t p o s i n g n o r m a l l y o c c u r s i n c a s u a l s p e e c h , t h i s s e n t e n c e , a l t h o u g h g r a m m a t i c a l , w o u l d s o u n d m o r e n a t u r a l i f to-wa w e r e r e p l a c e d b y i t s more c o l l o q u i a l v a r i a n t nante. nance h a s a n e q u i v a l e n t m e a n i n g t o to-wa b u t a l m o s t a l w a y s o c c u r s i n n e g a t e d c o n s t r u c t i o n s , w h i c h s u g g e s t s t h a t i t i n d i c a t e s t h a t t h e c l a u s e i t h e a d s i s an N P I . 139 (214) * [e; s o o r i - d a i j in-ga j i n i n - s u r u daroo to] C [, wa kinoo k i i t a , k i t t O ; . Prime-minister-nom resign modal C top yesterday heard surely I heard that [e; the Prime minister w i l l resign] surelyj. In the two above sentences i t i s very d i f f i c u l t to i n t e r p r e t the adverb "surely" as modifying the embedded CP. Such a r e s u l t can again be explained by the fact that the trace of the p r o p o s i t i o n a l l e v e l adverb w i l l be outside the domain of a [+V] l e x i c a l head. Consider the structure below. (215) CPX / \ CPX postposed adverb 3 / \ / I A IP I I' / I B VP CPX i s the matrix clause, to which the postposed phrase i s right-adjoined. CP2 i s the t o p i c a l i z e d CP. The canonical p o s i t i o n of the postposed c l a u s e - l e v e l adverb w i l l be somewhere between p o s i t i o n A and p o s i t i o n B. The trace of the adverb w i l l not be i n the domain of embedded V as long as i t i s attached no lower than an adjoined-to-VP p o s i t i o n . We must also assume as we d i d for the case of other modifiers of topics, that the trace of the c l a u s e - l e v e l adverb i s not adjoined to the CP i t modifies. If i t were, there would be no b a r r i e r between i t and i t s antecedent. 140 4.5 wa-marked purpose clauses It i s also possible to wa-mark a purpose clause headed by the bare form of the verb marked by l o c a t i v e marker ni: (216) yasai-o k a i - n i wa, moo i t t e - k i t a . vegetables-acc buy-loc top already went As for to buy vegetables I already went. In (217) and (218) below, we f i n d that postposing an object out of t h i s kind of construction r e s u l t s i n a severely ungrammatical sentence, i n contrast to what we found for postposing out of other types of iva-raarked clauses. (217) * k a i - n i wa, moo i t t e - k i t a , yasai-o. buy-loc top already went vegetables-acc As for to buy (themj), I already went -- vegetables^ (218) *dareka-ga ake-ni wa, moo i t t e - k i t a , ano kinko-o. someone-nom open-loc top already went that safe-acc As for someone to open ( i t j , they already went -- that safei These examples are judged to be worse than what we would have i f the object remained i n i t s canonical p o s i t i o n and a modifier of the object were postposed, as we see i n (219) below. (219) ?*dareka-ga kinko-o ake-ni wa, moo i t t e - k i t a , ano ginkoo-no. someone-nom safe-acc open-loc top already went that bank-gen As for someone to open the safe, they already went -- that bank's (safe) 141 This s i t u a t i o n i s s i m i l a r to the [+N] domain e f f e c t s we saw for postposing out of r e l a t i v e clauses i n chapter 3. A s i m i l a r account i s possible for (217) and (218). Here, the t o p i c a l i z e d l o c a t i v e phrase i s a p o t e n t i a l i s l a n d , l i k e the r e l a t i v e clauses i n the examples of chapter 3. Suppose that the purpose clause i s headed by a nominal element i n order that i t can receive l o c a t i v e case. If t h i s i s so, (217) and (218) s u f f e r [+N] domain v i o l a t i o n s because the postposed DP crosses the [+N] head of the case-marked conjunctive form of the verb, i f we assume that t h i s verb has received nominal features i n order to be case-marked. In entertaining such an account, we must also consider the fact that when the purpose clause i s not iva-marked, we do not get the same kinds of i s l a n d e f f e c t s , as we see i n (220). (220) ake-ni i t t e - k i t a , ano kinko-o. open-loc go-particip.-came that safe-acc I went to open (i t ) -- that safe. What might explain the difference between (217) and (220)? Recall that i n a l l the examples of r e l a t i v i z e d minimality e f f e c t s i n chapter 3, a phrase was postposed out of an i s l a n d such as a r e l a t i v e clause. And i n many of the cases where we saw an apparent [+N] domain e f f e c t , the postposed element was an object, which ought to be head-governed. Thus the ungrammaticality of those examples may have been a severe subjacency e f f e c t rather than an ECP e f f e c t . Arguably, we are seeing the same kind of e f f e c t i n (217), given the improved status of (219). In (220), however, the trace of the postposed object i s not embedded inside an i s l a n d as i t i s i n (217), i f we assume that the purpose clause i n i t s canonical p o s i t i o n i s an argument 4 8. Thus, [+N] domain e f f e c t s only seem to occur when a non-L-marked b a r r i e r intervenes between the postposed element and i t s trace. 48Such a hypothesis is supported by the fact that only a limited number of verbs, typically verbs of motion that take locative arguments can occur with a purpose clause of the type being considered here. 142 We require a [+N] domain account to explain the fact that (219) i s s l i g h t l y better than (218). But there i s also the p o s s i b i l i t y that both (217) - (219) are deviant because the verb i n a purpose clause has [-V] features i n order that i t can be marked with l o c a t i v e case. If i t has these features, the postposed element w i l l not be i n the domain of a l e x i c a l [+V] head, which, under our hypothesis, would account for the ungrammaticality. As far as (220) i s concerned, there, even i f the trace of the postposed object i s not i n the domain of a [+V] head, there are no b a r r i e r s to antecedent government of the trace, as there are i n (217)-(219), which would explain the contrast. Another possible explanation for the ungrammaticality of (217) and (218) i s that when a phrase headed by ni i s t o p i c a l i z e d , i t i s i n a d i f f e r e n t p o s i t i o n than other kinds of t o p i c s . Such a hypothesis would be i n accord with Saito (1985) and Hoji (1985). Saito (1985) shows that PP t o p i c a l i z a t i o n d i f f e r s from DP t o p i c a l i z a t i o n i n that the former, unlike the l a t t e r , seems to obey i s l a n d constraints and w i l l not allow resumptive pronouns, whereas the l a t t e r w i l l marginally allow resumptive pronouns. Hoji (1985) argues that PP t o p i c a l i z a t i o n i s derived by movement rather than base-generation. 4 9 Examples (221) and (222) below show that wa-marked purpose clauses pattern with other wa-marked PP's i n not allowing resumptive pronouns and in not allowing construal with a p o s i t i o n inside a r e l a t i v e clause, unlike non-PP topics. (221 )* [yasai-o kai-nij—wa, sore^-ni i t t e - k i t a vegetables-acc buy-P-top that-P went As for [to buy vegetables] t -- I went to do thatj To determine the exact structure of a sentence with a PP topic or a non-PP topic would be difficult since both obligatorily occur as the leftmost element. Nevertheless, based on the evidence of Saito and Hoji, and the evidence I present below, I shall assume that there is some relevant structural difference and that purpose clauses pattern with PP topics. 143 (222) *[eiga-o mi-ni] i-wa [kinoo dekaketa kyakusan-ga] D P modoranakatta. film-acc see-P-top yesterday went-out guest-nom did-not-return As for [to see a movie] t, the guest who went out e; hasn't returned. To test the hypothesis that PP-topics are strong islands, unlike other topics such as DP topics, we need to test whether we can postpose out of other kinds of PP topics -- for example those of the form DP-ni-wa. However, i f we are to postpose a l e x i c a l l y governed element from such a phrase, we w i l l need to embed a clause i n t h i s DP. If the clause i s a r e l a t i v e clause, we have added another i s l a n d . Thus i t i s necessary to use a complement clause headed by a noun l i k e "rumour" that can take a CP complement. In (223) below, a r e l a t i v e clause i s postposed out of a PP topic, r e s u l t i n g i n a deviant sentence. In (224) a s i m i l a r r e l a t i v e clause i s postposed out of the same clause that i s embedded i n a DP topic and the r e s u l t i s marginal, but d i s t i n c t l y better than (223) . We get the same r e s u l t i n (225) when a genitive i s postposed out of a complement clause inside a DP t o p i c . (223) *[ei okane-o nusunda to i u uwasa-ni-wa], Taroo-ga maitta, [boku-no moratta] i . money-acc sto l e C say rumour-P-top -nom y i e l d e d I-gen received Taro was defeated by/yielded to the rumour c o p that he s t o l e money that I had received. 144 (224) ??Masao-ga, [e 4 okane-o] nusunda to i u uwasa-wa hontoo da, [boku-ga kinoo moratta] i . -nom money-acc s t o l e C say rumour-top true copula I-nom yesterday received The rumour that Masao sto l e the money i s true -- (the money) that I received yesterday. (225) ??dareka-ga [e^ kinko-o] D E, aketa to i u uwasa-wa, hontoo da, [ano ginkoo-no].^ someone-nom safe-acc opened C say rumour-top true cop. that bank-gen It' s c e r t a i n that someone opened the safe -- that bank's (safe). In (226) and (227) we see a s i m i l a r contrast. When an object i s postposed out of a complement clause inside a PP topic, the r e s u l t i s marginally deviant. When, i n (227), an object i s postposed out of a complement clause inside a DP topic, the r e s u l t i s much better. (226) ??[ei nusunda to i u uwasa-ni-wa], Taroo-ga maitta, okane - O i -s t o l e C say rumour-P-top -nom y i e l d e d money-acc Taro was defeated by/yielded to the rumour c o p that he s t o l e -- money. (227) D P[dareka-ga e t aketa to yuu uwasa]-wa, hontoo datta, [ano kinko-o]^ someone-nom opened C say rumour-top true copula-past that safe-acc The rumour that someone opened ( i t j i s true -- the safe;, (good) Although I cannot give an explanation, the above r e s u l t s show that a PP-topic i s a stronger i s l a n d for postposing than a DP-topic. These r e s u l t s are 145 consistent with the evidence of Saito and Hoji that PP 5 0-topics behave d i f f e r e n t l y s y n t a c t i c a l l y than base-generated topics. If purpose clauses, which are of the form [ ... V-ni-wa] are of a s i m i l a r structure to PP topics, which are of the form [DP-ni-wa] , we can explain the fact that purpose clauses are stronger islands for postposing than base-generated topics by the fact that they are a d i f f e r e n t kind of construction, even i f we cannot give a precise explanation for the reasons why purpose clauses and PP-topics show these strong i s l a n d e f f e c t s . In section 4 .8 I propose a possible explanation for the d i f f e r e n t i s l a n d e f f e c t s shown by PP-topics versus base-generated topics with respect to postposing. I am referring to topics of the form XP-ni-wa as PP topics for the sake of giving them a label. In Appendix B I present evidence that particle ni is more likely a case-marker than a P. 146 4 .6 Do CP topics pattern with DP topics with respect to Saito's diagnostics? We observed i n sec. 4.1 that CP topics such as those headed by koto, no, or to do seem to permit extraction of a postposed phrase. If they pattern l i k e DP topics i n t h i s way, then we would expect them to also pattern l i k e DP topics with respect to the diagnostics Saito and Hoji use for comparing DP and PP topics with respect to movement versus base-generation. We also f i n d that. CP topics (marginally) allow resumptive pronouns. (228) was judged to be no worse than (229) . (228) ?[Masao-ga okane-o nusunda koto-wa] i ( Mary-ga s o r e - O j minna-ni hanasita. -nom money-acc s t o l e C-top -nom that-acc everybody-dat spoke As for [the fact that Masao st o l e money] i ( Mary t o l d i t 4 to everyone. (229) ?Tarooi-wa Hanako-ga kare^o hometa . -top -nom him-acc praised. As for Taro, Hanako praised him. Saito (1985) argues that coindexing a topic with a p o s i t i o n inside a r e l a t i v e clause i s better than scrambling a phrase out of a r e l a t i v e clause. Yet we do not f i n d that t h i s i s the case for CP topics: 147 (230) *[Masao-ga okane-o nusunda to i u koto-wa];, tadaima [kinoo e ; tutaeta h i t o - g a ] D P k i t e - i r u . -nom money-acc sto l e C say C-top right-now yesterday t o l d person-nom come-exist As for the fact that Masao sto l e money, the person who s a i d (so) yesterday i s here r i g h t now. (topic construed with p o s i t i o n i n s i d e r e l a t i v e clause) (231) *[Masao-ga okane-o nusunda to i u koto-o] i t tadaima [kinoo e ; tutaeta h i t o - g a ] D P k i t e - i r u . -nom money-acc sto l e C say C-acc right-now yesterday t o l d person-nom come-exist As for the fact that Masao sto l e money, the person who s a i d (so) yesterday i s here r i g h t now. (scrambled object construed with p o s i t i o n inside r e l a t i v e clause -- i . e . object scrambled out of r e l a t i v e clause) If CP topics behave the same as DP topics, Saito's analysis ought to predict that (230) i s better than (231), but t h i s i s not the case. I cannot o f f e r an explanation for the lack of contrast between (230) and (231) . It i s possible that CP topics are yet another type of construction that d i f f e r from both DP topics and PP topics. 4.7 L e x i c a l government of subjects i n Japanese The evidence we have seen so far suggests that topics i n Japanese are not L-marked and are islands for postposing i f a phrase postposed out of a v/a-marked topic i s not in the domain of a [+V] head -- i . e . a form of l e x i c a l government that extends to elements within the whole domain of a head, not just i t s complements. This raises the question of whether subjects i n Japanese are 148 l e x i c a l l y governed and whether they can be postposed out of t o p i c s . Recall f i r s t , (225), repeated here as (232), i n which an object ( l e x i c a l l y governed) i s postposed out of a top i c : (232) C P[dareka-ga e 4 tabetesimatta koto]-wa, tasika da , [ano wagasi-o];. someone-nom ate-up comp-top c e r t a i n copula that sweet-acc It' s c e r t a i n that someone ate (it;) up -- that sweet;. If we postpose a subject instead of an object i n (232), the r e s u l t i s only s l i g h t l y worse and nowhere near as bad as the deviant cases we looked at i n which the trace of the postposed element was not i n the domain of a l e x i c a l head. : (233) ? C P[e i Wagasi-o tabetesimatta koto]-wa, tasika-da, Taroo-ga; sweet-acc ate-up comp-top certain-cop. -nom It's c e r t a i n that (he) ate up the sweet-- Taro. This r e s u l t suggests that i f our hypothesis about l e x i c a l government i s correct, that subjects in Japanese are l e x i c a l l y governed -- under my assumptions here, as a r e s u l t of subjects being VP-internal. 4.8 Summary and r e l a t e d issues In summary, then, the grammaticality of postposing out of a wa-marked topic i n Japanese seems to depend on a number of f a c t o r s . One i s the necessity for the postposed phrase to be within the domain of a l e x i c a l [+V] head. Another i s the question of what kind of topic postposing i s occurring out of. It seems reasonable to assume, following Saito (1985) that there i s more than one type of wa-marked topic i n Japanese and that d i f f e r e n t types of topics occupy d i f f e r e n t 149 p o s i t i o n s . For example, as discussed above, the increased s e v e r i t y of postposing out of a PP topic as opposed to a base-generated topic could be due to a s t r u c t u r a l d i f f e r e n c e : that PP topics attach to a lower p o s i t i o n , as shown i n (234) below: (234) (i) XP | \ XP ZPi / I YP X' /_\ t i ( i i ) XP I \ XP ZP4 WP / I PP w /_\ t i (i) above represents the posited relevant structure of a sentence with a phrase ZP postposed out of a base-generated topic YP. In t h i s structure, XP w i l l not be a b a r r i e r to antecedent government of the trace t 4 by ZP since ZP i s contained i n but not included by XP. In ( i i ) on the other hand, the PP topic i s i n the s p e c i f i e r of a lower projection, WP. Here, WP constitutes an extra b a r r i e r to antecedent government 150 of t; by ZP. 5 1 There i s one possible problem with such an analysis, to which I s h a l l propose a s o l u t i o n . If PP topics are attached to a lower p r o j e c t i o n than base-generated topics, then we should expect that rightward extraction of a PP topic i n i t s e n t i r e t y ought not to be l i c i t . For example, PP i n the diagram above ought not to be L-marked by W, i f W i s a functional head. It also must move out of WP i n order to r i g h t - a d j o i n to the highest projection, i f we assume that a postposed phrase cannot attach lower than that. Thus WP ought to be a b a r r i e r for rightward extraction of PP, unlike the case of rightward extraction of topic base-generated i n the s p e c i f i e r of the highest projection. Yet we f i n d that PP topics can be postposed just l i k e other topics: (235) Taroo-ga sunde-iru, Tookyoo-ni-wa. -nom l i v e s Tokyo-loc-top Taroo l i v e s there -- in Tokyo. I now o f f e r the following explanation for the grammaticality of the example above. Saito (1985) gives good evidence that PP-topics are generated by movement. Let us suppose then that the l i c e n s i n g of base generated topics and PP-topics d i f f e r as follows. Base generated topics are base generated i n a high functional p r o j e c t i o n and are licensed there. PP-topics move to a high p o s i t i o n i n order to take scope but do not necess a r i l y need to be licensed by moving to "the checking domain", i n minimalist terms, of a high p r o j e c t i o n . If t h i s i s true, then movement from an p o s i t i o n that i s argument or complement of V d i r e c t l y to a postposed p o s i t i o n i s a l l that i s necessary for a PP topic to take scope. In that case, the trace of the postposed PP topic w i l l be L-marked by V, explaining the l i c i t n e s s of movement. Consider now the case of postposing out of a PP t o p i c . We 5 1 A n o t h e r f a c t o r t h a t we n e e d t o c o n t r o l f o r i s t h e q u e s t i o n o f w h e t h e r a w a - m a r k e d p h r a s e we a r e c o n s i d e r i n g i s a n i n s t a n c e o f i n - s i t u " c o n t r a s t i v e " w a - m a r k i n g r a t h e r t h a n t o p i c a l i z a t i o n . F o r e x a m p l e , i n e x a m p l e s l i k e (124) o r (128) i n w h i c h t h e w a - m a r k e d p h r a s e o c c u r s w i t h a n e g a t e d m a t r i x v e r b , i t i s p o s s i b l e t h a t we a r e s e e i n g h e r e a n i n s t a n c e o f " c o n t r a s t i v e " wa. 151 saw i n (217) and (218) that postposing out of a PP topic r e s u l t e d i n a deviant sentence. If a phrase i s postposed out of a PP topic before the topic i s raised, the topic i n i t s o r i g i n a l L-marked p o s i t i o n should not constitute a b a r r i e r ; however, such movement w i l l be ruled out by s t r i c t c y c l i c i t y : when the PP topic i s r a i s e d a f t e r i t s constituent has been postposed i t w i l l be necessary to return to a lower cycle, or, i n minimalist terms, the structure w i l l not be extended by r a i s i n g the PP topic, since a higher structure has already been created by postposing i t s constituent. In the foregoing analysis, I have proposed a modification to the c l a s s i c a l " d i s j u n c t i v e " version of the ECP that allows the l i c e n s i n g of the traces of adjuncts under c e r t a i n conditions: the trace of a postposed adjunct must be e i t h e r antecedent governed, or licensed by being i n the domain of a [+V] l e x i c a l head. This proposal explains the ungrammaticality of (219), repeated here as (236), since the trace of the postposed genitive w i l l not be l i c e n s e d by a [+V] head. (236)*purin-wa tabeta, Taroo-no. pudding-top ate -gen As for the pudding, I ate i t -- Taro's (pudding). 4.8.1 The status of sentence-level adjuncts If topics i n Japanese are b a r r i e r s to antecedent government -- for example, as a r e s u l t of not being l e x i c a l l y governed themselves -- the trace of the postposed genitive i n (236) w i l l be neither antecedent governed nor l i c e n s e d by a t+V] head. On the other hand, matrix-level concessive clauses or reason clauses, which appear to be sentence-level adjuncts that are not l i k e l y i n the domain of a l e x i c a l [+V] head, are r e a d i l y postposed. (see Endo (1989) for examples.) If they are postposed by movement, we need to assume that t h e i r traces 152 are antecedent-governed. If we assume that antecedent government requires c-command but not precedence 5 2 then the trace of a sentence-level adjunct can be antecedent governed i f i t occurs within the highest p r o j e c t i o n of the sentence. For example, i f a sentence-level adjunct such as a reason clause i s canonically l e f t - a d j o i n e d to a pr o j e c t i o n of matrix C and CP i s also the p r o j e c t i o n to which postposed phrases are right-adjoined, then there i s no b a r r i e r to antecedent government of the trace. The same would be true for (base-generated) topics i f they are generated i n [Spec, CP]. But when considering the matter of l i c e n s i n g of traces of postposed sentence-level adjuncts, we need to ask whether they are postposed by movement at a l l . We saw i n chapter 1 that arguments and low-level adjuncts such as r e l a t i v e clauses and genitives show evidence through Condition C reconstruction e f f e c t s of being postposed by movement. We also saw i n section 2.1 that postposing of these phrases shows subject-object asymmetries with respect to the grammaticality of postposing them out of a subject versus an object -- a phenomenon that we would usually associate with movement. But the same may not be true for other kinds of adjuncts. The following examples are designed to test whether a sentence-level adjunct i s postposed by movement. If we use the kind of Condition C reconstruction test that was used i n chapter 1, the only element that can c-command the s i t e of the postposed adjunct i s a t o p i c . But as was discussed i n chapter 1, i t i s possible that a topic i s i n a high enough p o s i t i o n to d i r e c t l y c-command a postposed phrase. In that case, we need to postpose a cl a u s e - l e v e l adjunct of an embedded clause. When we construct such examples we need to guard against a strong tendency for such adjuncts to be construed with the matrix rather than the embedded clause, as i s the case i n (237) below: 5 2 I f a n t e c e d e n t g o v e r n m e n t r e q u i r e d p r e c e d e n c e t h e n t h e t r a c e o f a n y p o s t p o s e d p h r a s e d e r i v e d b y movement c o u l d o n l y be l i c e n s e d by h e a d g o v e r n m e n t . I n s u c h a c a s e we w o u l d e x p e c t t h a t t h e p o s t p o s a b i l i t y o f a p a r t i c u l a r p h r a s e w o u l d o n l y d e p e n d on w h e t h e r i t was l e x i c a l l y g o v e r n e d a n d n o t o n w h a t k i n d o f i s l a n d i t i s e m b e d d e d i n . G i v e n t h e e v i d e n c e we saw i n e a r l i e r s e c t i o n s , i n w h i c h t h e g r a m m a t i c a l i t y o f p o s t p o s i n g t h e same t y p e o f e l e m e n t "would v a r y a c c o r d i n g t o t h e t y p e o f i s l a n d i t was e m b e d d e d i n , I s h a l l a s s u m e h e r e t h a t w h e r e p o s t p o s i n g i s d e r i v e d b y movement t h e t r a c e o f a p o s t p o s e d p h r a s e c a n be a n t e c e d e n t - g o v e r n e d f r o m t h e r i g h t . 153 (237) kare-ga, [e; omoikiri benkyoo-suru] to i t t a , [ a s i t a siken-ga aru kara] i. he-nom hard study C said tomorrow exam-nom exist because He said [e; he would study hard] --[because there i s an exam tomorrow]; The following sentence i s designed so that i t i s d i f f i c u l t pragmatically to construe the adjunct with the matrix clause: (238) Galileo-wa kinoo-no kaigi-de [[tikyuu-ga taira-soo da-kara] t a i r a de-aru t o ] C P i u koto-o h i t e i - s i t a . -top yesterday-gen meeting-loc earth-nom flat-modal cop.-because f l a t cop. C say C-acc denied G a l i l e o denied at yesterday's meeting the fact [ C Pthat the earth i s f l a t because i t looks f l a t ] . In (238) above i t i s possible to get a construal of the reason clause with embedded clause. Thus reason clauses can modify embedded clauses i n Japanese. However, when we t r y to postpose the reason clause i n (238) construal with the embedded clause becomes impossible: (239) *Galileo-wa kinoo-no kaigi-de [et tikyuu-ga t a i r a de-aru to i u koto-o] C P h i t e i - s i t a , [taira-soo da-kara]; -top yesterday-gen meeting-loc earth-nom f l a t cop. C say C-acc denied flat-modal cop.-because * G a l i l e o denied the fact [ C Pthat the earth i s f l a t e j at yesterday's meeting -- [because i t looks f l a t ] ; . This r e s u l t i s i n t e r e s t i n g because examples i n chapters 2 and 3 suggested that clauses embedded i n a A:oto or to i u Jcoto construction are at most very weak islands for postposing. For example, r e l a t i v e clauses and VP-level adverbs can 154 be postposed out of clauses headed by koto. (240) [Masao-ga, [e; okane-o] nusunda koto]-ga akiraka da, • [kinoo boku-ga moratta] i. -nom money-acc st o l e comp-nom cl e a r cop. yesterday I-nom received It ' s c l e a r that Masao sto l e the money -- that I received yesterday. (241) [Masao-ga, [et okane-o] nusunda koto]-ga aru, [boku-no kakusitaJi. -nom money-acc stole comp-nom exist I-gen h i d Masao has (on occasion) stolen money -- that I hi d . (242) [Masao-ga, [ei okane-o] nusunda koto]-o s i t t e - i r u , [kinoo boku-ga morattai. -nom money-acc st o l e comp-acc know yesterday I-nom received I know that Masao sto l e the money -- that I received yesterday, (or) Masao knows that (they) s t o l e the money -- that I/he received yesterday. (243)?[Taroo-ga ei kuru] koto-ga akiraka da, [moo sugu] ; -nom come comp-nom clear cop. already soon It' s c l e a r that Taroo w i l l come -- i n a l i t t l e while. In (240) - (242) a r e l a t i v e clause i s postposed out of a koto clause and in (243) a temporal adverb i s postposed marginally out of a koto clause. This suggests that i n (239) i t i s not the islandhood of the koto clause that i s responsible for the severe ungrammaticality that r e s u l t s when the reason clause i s postposed out of i t . Therefore, we might explain the contrast between (239) 155 and (240)-(243) by continuing to pursue our hypothesis that traces of postposed phrases must be eith e r antecedent governed or licensed by a [+V] head and by assuming that postposed adjuncts cannot be base-generated. 5 3 The r e l a t i v e clauses and VP-level adverbs i n (240)-(243) are possibly within the domain of V. In (239), on the other hand, i f the reason clause i s adjoined to a functional p r o j e c t i o n of IP of the embedded clause, i t w i l l not be i n the domain of embedded V, as we see i n the following structure: (239)(repeated again as (244)(a) *Galileo-wa kinoo-no" kaigi-de [e; tikyuu-ga t a i r a de-aru to i u koto-o] C P h i t e i - s i t a , [taira-soo da-kara]; -top yesterday-gen meeting-loc earth-nom f l a t cop. C say C-acc denied flat-modal cop.-because * G a l i l e o denied the fact [ C Pthat the earth i s f l a t e j at yesterday's meeting -- [because i t looks f l a t ] ; . I f i t w e r e p o s s i b l e t o b a s e - g e n e r a t e a p o s t p o s e d a d j u n c t , i t w o u l d b e d i f f i c u l t t o e x p l a i n why i t c o u l d n o t be c o n s t r u e d w i t h a p o s i t i o n i n s i d e a koto c l a u s e , g i v e n t h e e v i d e n c e i n S a i t o (1985) t h a t a b a s e - g e n e r a t e d t o p i c c a n be c o n s t r u e d w i t h a p o s i t i o n i n s i d e a r e l a t i v e c l a u s e , w h i c h i s t y p i c a l l y a w o r s e i s l a n d t h a n a koto c l a u s e . (I r e t u r n t o t h i s m a t t e r o f S a i t o (1985) i m m e d i a t e l y b e l o w . ) 156 (244)(b)(structure of (248a) CP / \ CP adverbi I IP I VP V / \ CP V(matrix) I C I IP(embedded clause) I I' / I t i I' I VP I V(embedded) Recall that we are assuming here that there i s no head movement of V to I i n Japanese. We must also explain why the trace i s not lice n s e d by matrix V. Recall that i n the grammatical examples where a r e l a t i v e clause or genitive modifying an argument DP was postposed, we needed to assume that the V that theta-marks the 157 argument that the r e l a t i v e clause or genitive modifies can somehow license the adjunct: (245) V / \ DP V I / I / I adjunct Yet we have a very s i m i l a r structure i n (244). It i s as i f the embedded CP pr o j e c t i o n i n (244) acts as a b a r r i e r to some governing r e l a t i o n of the adjunct by matrix V but that DP does not act as such a b a r r i e r i n (245) . Recall , now out proposed l i c e n s i n g condition on traces of postposed phrases: 1. A trace t must be licensed i n Max(x) where x i s [+V]. 2. t i s licensed by x i f f x i s l e x i c a l (a) choose the smallest domain of a [+V] head that contains t (b) check that x i s l e x i c a l I propose to explain the deviance of (244) by p o s i t i n g that the embedded CP, headed by complementizer koto has a [+V] feature. 5 4 If i t does, then embedded CP, not matrix VP, w i l l be the smallest [+V] category containing the trace i n (244). 5 4 W h i t m a n ( 1 9 9 1 ) p r o p o s e s t h a t koto c a n f u l f i l t h e r o l e o f a C . I f t h i s i s t r u e , a n d i f a C h a s a [+V] f e a t u r e a s a member o f t h e V - I - C e x t e n d e d p r o j e c t i o n ( i n t h e s e n s e o f G r i m s h a w ( 1 9 9 1 ) , t h e n koto may h a v e t h i s f e a t u r e i n t h e a b o v e e x a m p l e . 158 As a f i r s t step, we choose the CP as the smallest [+V] category containing the trace. As a second step, we determine whether i t i s l e x i c a l . In i t s r o l e as a complementizer, koto i s arguably not l e x i c a l . Thus the trace w i l l not be properly li c e n s e d and the deviance of (244) i s explained. I return now to the question of why phrases such as embedded cl a u s e - l e v e l adjuncts whose traces would not be l e x i c a l l y governed nor antecedent-governed, cannot be postposed by base-generation i n order to escape ECP e f f e c t s , e s p e c i a l l y i n l i g h t of the evidence i n Saito (1985) that base-generated topics can be construed with a p o s i t i o n inside a r e l a t i v e clause. If Saito's claim i s correct, why then could not a base-generated postposed reason-clause be construed with a po s i t i o n inside a koto clause in (244) given the evidence that koto clauses are much weaker islands than r e l a t i v e clauses? One p o s s i b l e explanation i s that when a base-generated topic i s construed with a p o s i t i o n elsewhere i n the sentence, that p o s i t i o n i s occupied by a pro. For example, i n the examples Saito gives for construal of a base-generated topic with a p o s i t i o n i n s i d e a r e l a t i v e clause, the p o s i t i o n i n question i s an argument DP that can be represented by a pro. In the case of a reason-clause on the other hand, i t would seem to be a reasonable assumption that the canonical p o s i t i o n of a postposed reason clause cannot be occupied by a pro -- i . e . that there i s no such pro-form i n the language --eit h e r p h o n e t i c a l l y empty or overt. 4.8.3 D i f f e r e n t types of adjuncts -- d i f f e r i n g degrees of islandhood The contrast in p o s t p o s a b i l i t y we have seen here between d i f f e r e n t types of adjuncts would lead us to predict that we' should also see them behave d i f f e r e n t l y from each other as b a r r i e r s to extraction. If sentence-level adjuncts are not i n the domain of a l e x i c a l [+V] head while VP-level adjuncts, and r e l a t i v e clauses and genitives that modify an argument DP are, we should expect to f i n d that the former are much more severe islands for extraction. This i s in fact what we f i n d . In chapter 3 we found that reason clauses were generally more 159 severe islands for extraction of postposed phrases than r e l a t i v e clauses. I have pursued an analysis here i n which traces of some adjuncts such as r e l a t i v e clauses that modify arguments can be licensed i n a way that other adjuncts, such as sentence-level adjuncts, cannot, because the former, but not the l a t t e r are i n the domain of a V. But regardless of whether we pursue such an analysis, i t would seem that a model of l o c a l i t y constraints that r e l i e s s o l e l y on whether a phase i s L-marked or not to determine i t s i s l a n d status cannot account for the gradation of grammaticality we see i n the data on postposing. We saw i n chapter 3 that d i f f e r e n t types of adjuncts a l l e x h i b i t d i f f e r e n t degrees of islandhood. For example, r e l a t i v e clauses appear to be weaker islands than other types of adjunct clauses. It might be possible to explain these gradations i n grammaticality by a subjacency account in which stronger islands have a greater number of b a r r i e r s -- for example that some types of clauses are CP's and others are CP's embedded i n PP's, which would constitute two b a r r i e r s instead of one, but t h i s kind of mechanism does not appear s e n s i t i v e enough to account for the fact, for example, that A:oto-clauses, r e l a t i v e clauses, temporal clauses, and wh-islands a l l show d i f f e r e n t i s l a n d e f f e c t s when d i f f e r e n t kinds of elements are postposed out of them, as we saw i n chapters 2 and 3. If we are seeing subjacency e f f e c t s i n those examples that depend on the number of b a r r i e r s crossed by the postposed phrase, these e f f e c t s may also be determined by whether the b a r r i e r i s L-marked or not, and whether i t i s i n an adjoined, s p e c i f i e r , or complement p o s i t i o n . There may also be other types of mechanisms that l i c e n s e phrases. For example, although topics do not appear to be L-marked, they are an obligatory element i n most types of Japanese sentence. This i s not a condition that we would associate with adjuncts. Are topics then licensed i n a d i f f e r e n t way than arguments, and i s i t t h i s l i c e n s i n g that permits some phrases to be postposed out of them? It would seem that we need a more well-developed theory of adjuncts and other non-arguments. 160 5. Postposing as a root clause phenomenon: an explanation i n terms of Minimal Domains 5.1 Introduction Endo (1989), and Whitman (1991b) have noted the fact that the postposing construction i n Japanese i s s t r i c t l y a root clause phenomenon. In t h i s chapter I show that under the framework of Chomsky (1995), in p a r t i c u l a r under the concept of a minimal domain, a p o s i t i o n adjoined to the root clause i s a unique p o s i t i o n that has no d i r e c t semantic or checking r e l a t i o n s h i p to the rest of the clause and uniquely serves as a s i t e for a postposed element. Although I draw from the ideas of the Minimalist programme i n t h i s chapter, the proposal that there i s rightward movement to a right-adjoined p o s i t i o n i s a c o n t r o v e r s i a l one, and i s not in accord with the Minimalist programme. Consider the following examples, which show that i n Japanese when a phrase i s postposed to the r i g h t , i t must be i n the rightmost p o s i t i o n . (246) [Endo 15;29] [ei basu notte-iku no] C P mendo-kusai desyoo? a t t i - k o t t i i . bus r i d e comp l o t - o f - t r o u b l e modal here-and-there To go here and there r i d i n g a bus i s a l o t of trouble, i s n ' t i t ? (247) *[ei basu notte-iku no] C P a t t i - k o t t i j mendo-kusai desyoo? . bus r i d e comp here-and-there l o t - o f - t r o u b l e modal To go here and there r i d i n g a bus i s a l o t of trouble, i s n ' t i t ? Example (247) above shows that a sentence-level adverb cannot attach to the r i g h t of an intermediate CP. Example (248) below shows that i t also cannot attach 161 to an intermediate IP. (248) *[ [ e i basu n o t t e - i k u ] I P a t t i - k o t t i i no] C P mendo-kusai desyoo? . bus rid e here-and-there comp l o t - o f - t r o u b l e modal To go here and there r i d i n g a bus i s a l o t of trouble, i s n ' t i t ? Whereas i n (246) the postposed phrase can be l i c i t l y adjoined to the right of the matrix verb, i t cannot appear to the ri g h t of the embedded IP or embedded CP. We also f i n d that i f a complementizer such as question p a r t i c l e ka appears to the r i g h t of the matrix verb, the postposed phrase must follow that element: (249) [e; basu notte-iku no] C P mendo-kusai desyoo ka? a t t i - k o t t i i -bus r i d e comp l o t - o f - t r o u b l e modal comp here-and-there To go here and there r i d i n g a bus i s a l o t of trouble, i s n ' t i t ? (250) *[e 4 basu notte-iku no] C P mendo-kusai desyoo a t t i - k o t t i i ka? bus rid e comp l o t - o f - t r o u b l e modal here-and-there comp To go here and there r i d i n g a bus i s a l o t of trouble, i s n ' t i t ? We can conclude from these data that the postposed element only attaches to the root clause and that i t as the rightmost element of the root clause. 5.2 A member of the domain of no category In t h i s section I show that a phrase that i s adjoined to the highest p r o j e c t i o n i s uniquely i n a p o s i t i o n that i s not i n the minimal domain of any head, under the d e f i n i t i o n of a minimal domain i n Chomsky (1995). The relevant minimalist d e f i n i t i o n s are as follows: 162 (251) Max(a) Max(a) = the smallest maximal p r o j e c t i o n i n c l u d i n g a . (252) Domain The domain of head a = the set of cat e g o r i e s i n c l u d e d 5 5 i n Max (a) that are d i s t i n c t from and do not con t a i n a . (253) Complement Domain The complement domain of a = the subset of the domain r e f l e x i v e l y dominated by the complement of the c o n s t r u c t i o n . (254) Residue The residue of a = the domain of a minus i t s complement domain. (255) Min(S) For a set S of c a t e g o r i e s , the minimal subset, Min(S) = the smallest subset K of S such that f o r any y e S, some P £ K r e f l e x i v e l y dominates y. (256) I n t e r n a l domain The i n t e r n a l domain of a = Min(complement domain of a) -- t y p i c a l l y i n t e r n a l arguments of a . 5 5 I a d o p t h e r e t h e d e f i n i t i o n o f m i n i m a l d o m a i n o f C h o m s k y ( 1 9 9 5 : 2 9 9 ) a s o p p o s e d t o t h e e a r l i e r v e r s i o n i n w h i c h t h e d o m a i n o f a was a s e t o f n o d e s contained i n Max(a) r a t h e r t h a n included i n Max(a). T h e d i f f e r e n c e i s c r u c i a l f o r o u r p u r p o s e s h e r e , s i n c e t h e l a t e r d e f i n i t i o n w i l l e x c l u d e a c a t e g o r y a d j o i n e d t o a m a x i m a l p r o j e c t i o n f r o m t h e d o m a i n o f t h a t p r o j e c t i o n . (257) Checking domain 163 The checking domain of a = Min(residue of a ) -- t y p i c a l l y involved i n checking i n f l e c t i o n a l features. These d e f i n i t i o n s w i l l be i l l u s t r a t e d i n the examples that follow. Consider now the following possible p o s i t i o n s for a d i s l o c a t e d phrase XP i n r e l a t i o n to a head Z. (At t h i s point I s h a l l not consider precedence, since i t has no s i g n i f i c a n c e i n the minimalist framework. (Chomsky (1995:334)) (258)i. adjoined to the complement of Z. Z' / \ YP Z / \ XP YP i i . s i s t e r to an intermediate p r o j e c t i o n of Z. ZP I Z' / I XP Z' I z 164 i i i . adjoined to the s i s t e r of an intermediate p r o j e c t i o n of Z. ZP I Z' / \ YP Z' / \ \ XP YP Z i v . s p e c i f i e r of Z. ZP / \ XP Z' \ Z v. adjoined to the s p e c i f i e r of Z. (Recall that we are not concerned here with l i n e a r order and that l e f t - a d j u n c t i o n and right-adjunction are equivalent for the purposes of t h i s discussion.) ZP / \ YP Z' / \ \ XP YP Z 165 v i . adjoined to ZP, the maximal proj e c t i o n of Z. ZP / \ XP ZP \ Z In case ( i ) , l e t Comp(Z) represent the complement domain of Z as defined above. Then XP, along with YP, i s i n the i n t e r n a l domain of Z. Comp(Z) i s everything r e f l e x i v e l y dominated by YP. XP i s part of the minimal complement domain of Z since the complement YP does not dominate i t . In case ( i i ) XP i s i n the checking domain of Z since i t i s d i s t i n c t from and does not contain Z and i t i s not dominated by any other member of the domain of Z. In case ( i i i ) XP i s i n the checking domain (Min(residue)) of Z, since YP, in the checking domain of Z, does not dominate i t . In case (iv) XP i s in the checking domain of Z. It i s i n the minimal domain of Z since i t i s i n the minimal set of categories that are contained by Max(Z) that are d i s t i n c t from and do not contain Z. It i s not a complement of Z; therefore i t i s i n the minimal residue of Z or i t s checking domain. In case (v) XP i s in the checking domain of Z. It i s i n the domain of Z since i t i s contained by Max(Z), and i t i s d i s t i n c t from and does not contain Z. It i s i n the residue of Z because i t i s not r e f l e x i v e l y dominated by the complement of Z. It i s i n the minimal residue ( i . e . checking domain) of Z since i t i s not dominated by YP, the s p e c i f i e r of Z. 166 Thus so far, with the exception of case ( v i ) , which we have not considered yet, XP i s i n every case in the minimal domain of Z: ei t h e r the checking domain or the i n t e r n a l domain. Consider now case ( v i ) . XP i s not i n the domain of Z since i t i s not included in Max(Z); i t i s only contained i n Max(Z). However, i f ZP i s dominated by a node of another projection, c a l l i t W, then ZP w i l l be i n the minimal domain of W for the same reasons that XP was i n cases (i) through (v). If ZP i s i n the minimal domain of W, then XP, adjoined to ZP, must also be i n the minimal domain of W since no other member of the domain of W dominates XP. The remaining case to be considered i s where XP i s adjoined to ZP and ZP i s dominated by no node of another p r o j e c t i o n : (259) ZP / \ ZP XP I Z In t h i s case, XP i s not i n the minimal domain of Z since i t i s not i n the domain of Z as explained immediately above. But i t cannot be i n the domain of any higher p r o j e c t i o n e i t h e r since there i s no p r o j e c t i o n above ZP. Thus we have a unique configuration i n which XP i s not in the domain of any p r o j e c t i o n . If a p o s i t i o n adjoined to the root node has the unique s t r u c t u r a l property of being i n the domain of no head then we ought to expect' i t to show unique sy n t a c t i c properties as well. Empirically, t h i s p o s i t i o n does show a s p e c i a l property i n Japanese as the only possible s i t e for r i g h t - d i s l o c a t e d phrases. In Japanese we can observe two other types of d i s l o c a t i o n of a phrase from i t s canonical p o s i t i o n : scrambling and t o p i c a l i z a t i o n . Suppose that there are (at least) three ways i n which a phrase can be d i s l o c a t e d from i t s canonical 167 p o s i t i o n : (a) base-generation i n the minimal domain of a functional p r o j e c t i o n that licenses i t . (e.g. base-generated t o p i c a l i z a t i o n ) (b) movement to a checking p o s i t i o n (e.g. PP t o p i c a l i z a t i o n or scrambling) (c) movement to a semantically- and checking-vacuous p o s i t i o n (e.g. postposing). I r e f e r to t h i s p o s i t i o n i n (c) as being both "checking vacuous" and "semantically vacuous." The p o s i t i o n can be considered "checking-vacuous" i n the sense that as I have shown above, i t i s the only p o s i t i o n that i s not i n the minimal domain of a head. This means that i t i s not in the checking domain of any category. If a constituent i n a checking p o s i t i o n i s interpreted by being checked by the head of the checking domain i n which i t occurs, then the postposing p o s i t i o n i s not e l i g i b l e for such i n t e r p r e t a t i o n , not being i n a checking domain. The p o s i t i o n can be considered "semantically vacuous" i n the sense that i t i s not i n the complement domain of any category. If complements receive a semantic i n t e r p r e t a t i o n by t h e i r r e l a t i o n to the head i n whose complement domain they occur, then a postposed phrase i s not e l i g i b l e for such an i n t e r p r e t a t i o n i n t h i s p o s i t i o n . To the extent that the postposed p o s i t i o n i s not i n the domain of a head, i t , i t cannot receive a semantic i n t e r p r e t a t i o n through i t s r e l a t i o n to any head i n whose domain i t occurs. On the other hand, I am not suggesting that postposing i s a semantically vacuous operation. Although I have chosen to r e s t r i c t my analysis to a purely syntactic one of postposing here, judgements of native speakers on postposing indicate that discourse factors are important i n 168 determining the legitimacy of a p a r t i c u l a r postposing operation. Suppose that there i s a higher node above the root node of the sentence that i s i n the discourse realm but not i n the syntactic realm. If t h i s were true, then the p o s i t i o n of postposed phrases would have some s t r u c t u r a l r e l a t i o n to the discourse component of the grammar, even though i t i s outside the domain of any head i n the syntactic component. If the postposed p o s i t i o n i s discourse-linked, i t w i l l have some semantic s i g n i f i c a n c e with respect to the o v e r a l l discourse even though i t may not with the root node of the sentence. Another p r e d i c t i o n that the hypothesis that the postposed p o s i t i o n i s "discourse-linked" makes i s that to have a r e l a t i o n with the discourse, we would predict that the postposed phrase should be attached high i n the sentence, which i s what we f i n d . If the postposed p o s i t i o n i s i n e l i g i b l e for a d i r e c t semantic or checking r e l a t i o n with a head, then we must ask how the postposed phrase receives i t s i n t e r p r e t a t i o n . To answer t h i s we must allow the p o s s i b i l i t y that movement and chain-formation can occur without a checking r e l a t i o n between the landing s i t e of movement and a l o c a l head. Although some types of movement (e.g. wh-movement) have been explained as movement to a p o s i t i o n i n which the moved phrase i s licensed by a functional head (e.g. a [+wh complementizer] i n the case of wh-movement) t h i s does not e n t a i l that a l l types of movement must undergo a feature-checking process. If postposing has a connection with the discourse component of the grammar, i t i s possible that the syntactic component and the discourse component, although separate, must interact i n some way. For example, postposing could be movement that takes place within the syn t a c t i c component of the grammar, as I have argued for by showing that postposing shows synta c t i c constraints, but that i t s landing s i t e i s licensed by the discourse component of the grammar --for example by a r e l a t i o n with a node that i s outside the syn t a c t i c domain but i n the discourse domain. 169 (260) DisP (in discourse component) / \ Dis CP (upper l i m i t of syntactic component) / \ CP XP(postposed phrase) lic e n s e d by a r e l a t i o n to DisP. The p o s s i b i l i t y that the landing s i t e of postposed phrases i s somehow licen s e d by the discourse component of the grammar i s consistent with the fact that postposing i s o b l i g a t o r i l y a s e n t e n t i a l phenomenon. We see i n the following examples that when a sentence fragment such as a DP i s uttered, for example, as an abbreviated answer to a question, such a phrase cannot l i c e n s e a postposed phrase on i t s r i g h t , even though such a postposed phrase would be adjoined to the root node of that sentence fragment 5 6: (261) [purin-o tabeta otoko-no-ko] D P pudding-acc ate boy the boy who ate the pudding (262) * [ [ t i tabeta] otoko-no-ko] D P, p u r i n - O ; D P ] ate boy pudding-acc the boy who ate (it) -- pudding The DP sentence fragment above i s arguably a bare DP with no IP or CP projections above i t . We might further p o s i t that there are no projections of a discourse component above that i n whose domain a postposed phrase could be li c e n s e d as i t would be i n (260) above. If we attempt to d i s l o c a t e a phrase to a right-adjoined p o s i t i o n other than to the root node, the phrase w i l l be i n the complement or checking domain of some 5 6These examples were suggested by Hisatsugu Kitahara ( p . c ) , who pointed out the s e n t e n t i a l nature of the phenomenon of postposing. 170 head and w i l l need to meet whatever l i c e n s i n g requirements there are for moving 'to the domain of that head. Ostensibly, t h i s i s what happens i n the case of scrambling and t o p i c a l i z a t i o n . 5 7 If movement or d i s l o c a t i o n to the domain of a head implies scrambling or t o p i c a l i z a t i o n , the question a r i s e s why we cannot scramble or t o p i c a l i z e to a right-adjoined p o s i t i o n lower than the root node. In the following section I s h a l l propose a formulation of the head parameter for Japanese that explains why right-adjunction i s only permitted to the root node in the language. My proposal w i l l also o f f e r an explanation of why postposing cannot occur to a p o s i t i o n l e f t - a d j o i n e d to the root node, and which ought to also be a p o s i t i o n i n the domain of no category. 5.3 The head parameter i n a l e f t - a t t a c h i n g language In t h i s section I examine two questions: a) Other than postposed phrases, no other type of r i g h t adjunction i s permitted i n Japanese. (Recall (246) - (250) above.) In a l l projections other than the root node, both complements and adjuncts o b l i g a t o r i l y precede heads: thus Japanese i s s t r i c t l y " l e f t - a t t a c h i n g . " Not only can a phrase not be ri g h t d i s l o c a t e d to adjoin to a non-root node but no type of adjunction to such a node i s possible -- for example an adjunct such as a r e l a t i v e clause cannot occur to the right of the NP or DP i t modifies. (263) [ [purin-o tabeta C P] otoko-no-ko-ga D P] nigeta pudding-acc ate boy-nom ran-away The boy who ate the pudding ran away. 5 7 W e m i g h t s p e c u l a t e , f o r e x a m p l e , t h a t t h e r e a s o n why s c r a m b l i n g a n d t o p i c a l i z a t i o n a l l o w o n l y c e r t a i n t y p e s o f p h r a s e s t o b e d i s l o c a t e d i s b e c a u s e some k i n d o f f e a t u r e - c h e c k i n g o f t h e m o v e d p h r a s e i s d o n e b y t h e f u n c t i o n a l h e a d t h a t l i c e n s e s s c r a m b l i n g o r t o p i c a l i z a t i o n . On t h e o t h e r h a n d , i f p o s t p o s e d p h r a s e s a r e n o t m o v e d t o t h e d o m a i n o f some h e a d , t h e r e o u g h t t o b e n o f e a t u r e c h e c k i n g . L a c k o f f e a t u r e c h e c k i n g w o u l d e x p l a i n why t h e r a n g e o f p h r a s e t y p e s t h a t c a n b e p o s t p o s e d i s much b r o a d e r t h a n t h a t o f p h r a s e s t h a t c a n b e s c r a m b l e d o r t o p i c a l i z e d . 171 (264)*[otoko-no-ko-ga [purin-o tabeta C P] D P] nigeta boy-nom pudding-acc ate ran-away The boy who ate the pudding ran away. In (263) the r e l a t i v e clause modifying "boy" precedes the head noun of the DP and the sentence i s good. In (264) the r e l a t i v e clause follows the noun i t modifies and the sentence i s bad. Why i s Japanese s t r i c t l y l e f t - a t t a c h i n g everywhere except at the root node? b) Many phrases that can be postposed cannot be l e f t - a d j o i n e d to the root node. The following examples are evidence that phrases such as genitives and adverbs, which can be postposed, cannot be l e f t - a d j o i n e d to the root node. (265) kinoo [ei ronbun-o] yon-da, Chomsky-nOi. yesterday paper-acc read-past -gen Yesterday I read (that) paper -- Chomsky's, (postposing of genitive) (266) *Chomsky-nOi kinoo [et ronbun-o] yon-da -gen yesterday paper-acc read-past *Chomsky's -- yesterday I read (his) paper, (fronting of genitive i l l i c i t ) If we explain (265) by adjunction of the genitive to the r i g h t of the highest p r o j e c t i o n (e.g. CP) , then l e f t adjoining i t to the same p r o j e c t i o n makes the sentence completely uninterpretable as far as the intended meaning i s 172 concerned. 5 8 The same argument would hold for (267) and (216) below. (267) kono syoosetu-wa e; omosiroi no yo, totemo;. t h i s novel-top i n t e r e s t i n g comp emph very-much This novel i s i n t e r e s t i n g -- very much so. (267) above i s taken from Endo (1989). If we t r y to front the postposed adverb, however, the r e s u l t i s ungrammatical. 5 9 (268) * t o t e m O i kono syoosetu-wa et omosiroi no yo. very-much t h i s novel-top i n t e r e s t i n g comp emph ? Very much so t h i s novel i s i n t e r e s t i n g . It i s as i f , when we get outside of the domain of a head, the head parameter has been reversed: righ t adjunction i s possible, but not l e f t adjunction. Although the process by which a phrase i s postposed cannot automatically apply leftwards to l e f t - d i s l o c a t e i t , there are two ways i n which a phrase can be l e f t - d i s l o c a t e d from i t s canonical p o s i t i o n iri Japanese: namely through scrambling or through t o p i c a l i z a t i o n . Scrambling i s l i m i t e d to a f a i r l y narrow range of phrase types. The fact that the scramblability of a phrase seems to depend on i t s Case suggests that some kind of feature checking i s involved i n scrambling. Under a minimalist 5 8 I a s s u m e h e r e t h a t f o r n o t f u l l y d e t e r m i n e d i n d e p e n d e n t r e a s o n s , a g e n i t i v e c a n n o t move t o a f e a t u r e - c h e c k i n g p o s i t i o n i n J a p a n e s e , s i n c e t h e r e i s n o e m p i r i c a l e v i d e n c e t h a t g e n i t i v e s c a n b e s c r a m b l e d . I f we a s s u m e t h e a n a l y s i s o f s c r a m b l i n g o f K i t a h a r a (1995) a s movement t o a f e a t u r e - c h e c k i n g p o s i t i o n t h e n i t a p p e a r s t h a t g e n i t i v e s e i t h e r l a c k p r o p e r f e a t u r e s f o r s c r a m b l i n g / f e a t u r e c h e c k i n g o r e l s e p o s s e s s f e a t u r e s ( p o s s i b l y t h e i r c a s e f e a t u r e s ) t h a t a r e i n c o m p a t i b l e w i t h t h e h e a d t h a t w o u l d c h e c k t h e m . 5 9 S o m e s p e a k e r s j u d g e t h i s s e n t e n c e a s b e i n g g r a m m a t i c a l . I t i s p o s s i b l e t h a t t h e a d v e r b c a n b e b a s e - g e n e a t e d i n s e n t e n c e - i n i t i a l p o s i t i o n h e r e , b u t i t i s n o t c l e a r w h e t h e r i t i s l e f t -a d j o i n e d t o t h e r o o t n o d e o r n o t . M o r e e x a m p l e s o f p h r a s e s d i s l o c a t e d t o t h e l e f t o f a t o p i c n e e d t o b e e x a m i n e d . 173 framework, we would expect that scrambling i s movement to a s p e c i f i e r p o s i t i o n rather than to an adjoined-to-XP p o s i t i o n since the l a t t e r i s not i n a checking domain. 6 0 T o p i c a l i z a t i o n requires topic marker wa to follow the t o p i c a l i z e d phrase, except i n c e r t a i n cases of c o l l o q u i a l speech where "wa-drop" can occur. It too i s more r e s t r i c t e d i n the phrase types that can be t o p i c a l i z e d than postposing. For example, as we saw i n sec. 0.4 genitives and r e l a t i v e clauses can be postposed but not t o p i c a l i z e d . These facts suggest that topics are licensed by some kind of functional head i n whose domain the topic occurs. Like scrambling, t o p i c a l i z a t i o n does not appear to be adjunction to the root node. If neither scrambling nor t o p i c a l i z a t i o n are l e f t - a d j u n c t i o n to the root node, we have a s i t u a t i o n i n Japanese i n which right-adjunction to the root node i s possible but not l e f t - a d j u n c t i o n . To r e c a p i t u l a t e then, the two questions I seek to answer here are: a) Why i s r i g h t adjunction only possible at the root node but not elsewhere? b) Why i s only r i g h t adjunction possible at the root node and not l e f t adjunction? 6 0 S e e K i t a h a r a (1997) f o r a n a n a l y s i s o f s c r a m b l i n g a s f e a t u r e c h e c k i n g b a s e d o n m i n i m a l i s t p r i n c i p l e s . 174 (269) root: CP *CP / \ / \ CP XP XP CP non-root: *CP CP / \ / \ CP XP XP CP In order to explain these facts I s h a l l propose c e r t a i n conditions on the l i n e a r ordering of constituents according to the way they show s t r u c t u r a l r e l a t i o n s h i p to each other in a sentence. These r e l a t i o n s h i p s are expressed i n terms of the r e l a t i o n between a head and an element that i s i n i t s minimal domain, i n the sense of the concept of minimal domain of Chomsky (1995) . I s h a l l state the conditions f i r s t , and then discuss what might motivate these conditions. 5.4 Proposed conditions on l i n e a r ordering For a terminal/head x l e t T(x) denote the largest set of terminals such that for every y £ T(x), Max(x) contains y. ( i . e . T(x) i s the set of terminals contained by Max(x).) Let T'(x) denote the largest set of terminals such that for every y £ T(x) and every y' £ T'(x), x precedes y. ( i . e . , T'(x) i s the set of terminals to the rig h t of those i n T(x)). Condition 1. Max(x) i s i n the minimal domain of head a i f and only i f there e x i s t s non-null T' (x) with oc £ T' (x) . Stated informally, Condition 1 means that i f a maximal p r o j e c t i o n Max(x) has been 175 b u i l t up and i f there are terminals to i t s r i g h t that are not contained i n Max(x), then one of those terminals must be a head with which Max(x) merges so that Max(x) w i l l be i n that head's minimal domain. Conversely, i f a maximal pr o j e c t i o n Max(x) i s i n the minimal domain of head a , then a must be to the r i g h t of Max(x). Condition 2. If Max(X) i s i n the minimal domain of . head Y, then x i s the rightmost member of T(x). ( i . e . Max(X) i s head-final with respect to complements and adjuncts) . Stated informally, Condition 2 means that heads may only s e l e c t l e f t - a t t a c h i n g maximal projections to be i n t h e i r minimal domain. Let us now look at how these proposed constraints bear on c e r t a i n structures i n Japanese. (270) Phrase l e f t - a d j o i n e d to root node: YP / \ XP YP This structure i s ruled out by condition 1. Since the " i f f " i m p l i c a t i o n works both ways, condition 1 requires that for head a , i f there e x i s t s a non-null T'(x) with a E T'(x), then Max(x) must be i n the minimal domain of a . In (10), the set of terminals dominated by the lower segment of YP constitute a non-null T'(x) with respect to head X, but there e x i s t s no head a i n t h i s set for which XP i s i n i t s minimal domain, since a phrase adjoined to YP i s not i n the minimal domain of Y. 176 (271) Phrase right-adjoined to root node YP / \ YP XP This structure meets condition 1. YP meets the condition since (a) YP i s in the minimal domain of no head and (b), T(Y) i s the whole phrase and T'(Y) i s n u l l . XP meets the condition for s i m i l a r reasons: XP i s i n the minimal domain of no head and no terminals exist to the right of T(X). The structure also meets condition 2 since neither XP nor YP i s i n the minimal domain of any head. (272) Phrase l e f t - a d j o i n e d to a YP, where YP i s i n the minimal domain of some head. ZP I Z' / \ YP Z / \ XP YP In the structure given above, YP i s the complement of Z. It would hold also i f YP were i n some other p o s i t i o n i n the minimal domain of Z. This example meets condition 1. T' (X) e x i s t s but Z i s a member of T' (X) and Max(X) i s i n the minimal domain of Z. It w i l l also met condition 2 i f we assume that XP i s l e f t - a t t a c h i n g . 177 (273) Phrase right-adjoined to YP, where YP i s i n the minimal domain of some head. ZP I Z' / \ YP Z / \ YP XP This example meets condition 1 since i t i s ZP whose minimal domain XP i s i n and Z e T'(X). However i t v i o l a t e s condition 2 since YP i s i n the minimal domain of a head Z and YP i s not head f i n a l : Y i s not the rightmost element of T(Y); X i s . These two conditions then have the r e s u l t that where root adjunction i s concerned, r i g h t adjunction i s possible but not l e f t adjunction, whereas where non-root adjunction i s concerned the reverse i s true. These are the facts that we observe for Japanese. These conditions w i l l also guarantee that outside of the case of r i g h t adjunction to the root, a l l projections w i l l be r i g h t headed, even i f they are not themselves in the minimal domain of a head. For example, the following structure w i l l be ruled out: (274) YP (root) / \ Y XP Although t h i s structure meets condition 2 (YP i s i n the minimal domain of 178 no head so i t doesn't have to be left-attaching) i t v i o l a t e s condition 1. XP i s i n the minimal domain of Y but there i s no non-null T'(X) such that Y E T"(X). These conditions w i l l also allow multiple r i g h t adjunction to the root node', where each successive right-adjoined phrase adjoins to the previously adjoined phrase: / \ YP XjP / \ XiP X 2 P / \ X 2 P For a l l i | i e {1, 2, ... n), XLP meets conditions 1 and 2. It meets condition 1 since T'(Xi) i s n u l l : MaxfXJ contains a l l terminals to the rig h t of X;. It meets condition 2 since XtP i s i n the minimal domain of no head, Multiple adjunction to the rig h t of the root node i s not possible i f any of X;P are adjoined to the l e f t of a subsequent r i g h t - d i s l o c a t e d phrase. For example, consider (276): (276) YP (275) YP XnP \ YP X,P / \ X 2 P XjP (276) v i o l a t e s condition 1 since T' (X2) i s non-null: T' (X2) i s the set of 179 terminals dominated by the lower segment of XXP. There i s no head a i n T' (X2P) such that X2P i s i n the minimal domain of a: X2P i s i n the minimal domain of no head. These predictions for multiple r i g h t adjunction are consistent with the empirical facts for Japanese postposing. In the data of Endo (1989) we do, in fact, f i n d examples of multiple r i g h t - d i s l o c a t i o n . The following examples are reproduced from Endo (1989). (277) e± e^  i i koto yo ne? [aa-yuu zyosei-ga i r u tte-yuu no-wa]i [zitu-ni]., good thing such woman-nom exi s t C V I C-top t r u l y I t ' s a good thing, i s n ' t i t -- that there i s a woman l i k e that -- t r u l y . (278) koko e£ ed e k kawaru no ne? yappari i ( [mainiti ne], naiyoo-ga k here change C as-expected every-day content-nom Here ( i t ) changes -- as expected -- every day -- the content. (279) et ima isogasii-n-da, [e, kyaku-ga kuru kara], asita^. now busy C V guest-nom come because tomorrow I'm busy now -- because a guest i s coming -- tomorrow. Example (279) i s p a r t i c u l a r l y i n t e r e s t i n g because an adverbial i s postposed out of an adverbial phrase which has i t s e l f been postposed. Endo notes that (279) i s ungrammatical i f only the temporal adverbial i s postposed and the adverbial clause out of which i t i s postposed remains i n s i t u . This mysterious fact can be explained by our analysis i f "tomorrow" i s right-adjoined to the adverbial "because" clause. 180 (280) CP / \ CP PP /_\ / \ I'm busy now PP adv. / _ \ tomorrow because a guest i s coming If the PP were i n s i t u , then i t would be a t y p i c a l i s l a n d for extraction of the temporal adverbial. But i n (280) the adverbial i s not outside the PP i s l a n d since i t i s adjoined to i t : i . e . i n Chomsky (1986) terms the non-L-marked PP cannot be a b a r r i e r for extraction because no b a r r i e r has been crossed by the adverb. At t h i s point I s h a l l speculate on what might motivate the above conditions. Suppose that order has some connection with processing -- for example with recovering a s y n t a c t i c representation from phonological representation by parsing a s t r i n g of terminal elements and converting them to a syntactic structure. The head parameter or i t s equivalent w i l l impose c e r t a i n constraints on the possible order of an XP and a head with which i t has a s y n t a c t i c r e l a t i o n . These constraints are necessary i n order to create an ordering of terminal elements that can be converted to syntactic structure by a parsing algorithm. Pursuing the idea that the concept of "minimal domain" of Chomsky (1995) i s fundamental to r e l a t i o n s between a head and a maximal projection, suppose that there i s a parsing algorithm that w i l l b u i l d up structure from a set of terminal elements i n a l e f t - t o - r i g h t d i r e c t i o n , creating maximal projections XP; and seeking to merge each maximal p r o j e c t i o n XP£ with a head Y or a p r o j e c t i o n of 5.4.1 Motivation for the conditions stated above 181 head Y so that XP; w i l l be i n the minimal domain of Y. Taking these assumptions, consider how a s i m p l i f i e d version of a parsing algorithm might work for a right-headed language. The s t r i n g of terminal elements i s parsed from l e f t to r i g h t . Each terminal element i s a head X;. The parser seeks to merge that head or a p r o j e c t i o n of i t with maximal projection(s) that have already been b u i l t up on i t s l e f t by an operation s i m i l a r to "Merge" i n Chomsky (1995). If i t finds such a maximal pr o j e c t i o n to i t s l e f t , and i f the head can s e l e c t that projection, the head X£ may merge with i t and the head projects. It continues to do t h i s with maximal projections on i t s l e f t u n t i l i t has considered them a l l as candidates for merging with, at which point a maximal pr o j e c t i o n X4P has been created. For each maximal pr o j e c t i o n XPt that i s created, i t looks for a head to i t s r i g h t that w i l l merge with projections in such a way that XP; w i l l be i n the minimal domain of a head. We could generalize t h i s requirement to say that each XP must be incorporated into the structure i n such a way that i t can be "interpreted" as having a r e l a t i o n to some head -- e i t h e r a semantic or checking r e l a t i o n , within the minimal domain of that head. This requirement must have one exception, however: an XP that has no remaining terminals on i t s r i g h t w i l l not be able to f i n d any head i n whose minimal domain i t can occur. This w i l l be true for the root node, and also for a postposed phrase. Thus we need to say that each XP that i s b u i l t up by the parsing algorithm must f i n d a head with which i t can have a "minimal domain r e l a t i o n " i f and only i f there i s a non-null set of terminals to the r i g h t of that XP. Suppose that the parsing algorithm creates a CP root clause and then finds that there are s t i l l terminals to i t s r i g h t that do not contain a head i n whose minimal domain the CP can occur. This i s what happens when these remaining terminals are postposed material. In such a case, the above condition can be met 182 i f the CP projects further to form another segment and adjoins the postposed material a f t e r i t has been parsed into an XP. We w i l l have the following structure: • (281) CP / \ CP XP Since T(C) i s the whole sentence, T(C) has no terminals to i t s r i g h t : i . e . T'(C) i s n u l l . Since CP has no terminals to i t s r i g h t i t does not need to look for a head to form a minimal domain r e l a t i o n with. The same i s true for XP. Notice that the condition that there be a non-null set of terminals to the ri g h t of XP i n order for i t to be required to f i n d a head i s necessary to handle the case of root clauses. It i s not created to allow postposed phrases but i t ends up allowing them. The conditions proposed above w i l l not allow phrases that are l e f t - a d j o i n e d to the root: (282) * CP / \ XP CP In such a structure, XP has terminals to i t s r i g h t but none of them i s a head with which XP has a minimal domain r e l a t i o n . I also proposed a second condition. In the parsing process, when each new terminal Y£ i s encountered, going from l e f t to ri g h t , i t "looks" for XP's to i t s l e f t to merge with head Yi. Suppose that although the parser builds up complex: structure, when i t looks back at the structure to the l e f t of the terminal i t i s 183 considering, i t only "sees",the head of each maximal p r o j e c t i o n that has been formed: [. . . X J [. - . X2] . . . Y Suppose further that i n order to "see" the boundary of each maximal projection, that maximal proj e c t i o n must be right-headed. Another way of conceiving t h i s idea i s that when considering whether a maximal p r o j e c t i o n to the l e f t i s a possible candidate for merging with the head i t i s parsing, the parser considers the rightmost element of each XP that has been formed and assumes i t to be the head of that XP, rather than examining the i n t e r n a l structure of the XP to determine where the head i s . If such a process i s to work, that XP must be ri g h t headed so that the head can be found t r i v i a l l y . Thus the second condition for operation of the parsing algorithm i s that each XP that merges with a head must be right-headed. 5.5 Summary In summary, then, the concept of a minimal domain of a head developed i n Chomsky (1995) predicts that there w i l l be a unique p o s i t i o n that i s adjoined to the highest p r o j e c t i o n which w i l l not be in the minimal domain of any head. If being i n the minimal domain of a head i s equivalent to having a d i r e c t r e l a t i o n to that head, then t h i s highest-adjoined p o s i t i o n w i l l escape from having to have a d i r e c t r e l a t i o n with any head and can serve as a semantically vacuous p o s i t i o n that can be used for, for example, a landing s i t e for c e r t a i n kinds of non-feature-checking-driven movement such as postposing. If we further adopt the formulation of the head parameter that I have proposed above, we can explain why adjunction to the root node can only occur to the r i g h t , and why r i g h t adjunction to a non-root node i s not possible. Notice that the conclusion here i s that at least i n l e f t - a t t a c h i n g 184 languages, a d j u n c t i o n of XP to YP i s severely r e s t r i c t e d , as Chomsky (1995) p r e d i c t s . The c o n s t r a i n t s I have proposed w i t h respect to d i r e c t i o n a l i t y i n Japanese are a f i r s t approximation f o r cap t u r i n g some noteworthy observations about postposing i n Japanese. The issue of d i r e c t i o n a l i t y i n s y n t a c t i c r e l a t i o n s appears to have r e a l e m p i r i c a l m a n i f e s t a t i o n s and consequently demands f u r t h e r research. 6. Non-syntactic f a c t o r s i n postposing As s t a t e d i n the i n t r o d u c t i o n , t h i s d i s c u s s i o n of postposing i s l i m i t e d to a p u r e l y s y n t a c t i c a n a l y s i s ; however, i t i s c l e a r than n o n - s y n t a c t i c f a c t o r s l i k e l y p l a y a r o l e i n the postposing c o n s t r u c t i o n as w e l l . In t h i s chapter, I b r i e f l y present data on some non-syntactic phenomena that bear on c e r t a i n questions about postposing. For example, I include examples that show that postposing i s subject to a heaviness e f f e c t : phrases that are "heavy" are e a s i e r to postpose than those that are not. This f a c t i s important i n some of my e a r l i e r examples, because i t shows that a p a r t i c u l a r ungrammatical sentence that has a heavy postposed element cannot be accounted f o r by the heaviness of the postposed element. I t i s a l s o important to my d i s c u s s i o n of the postposing of "heavy" subjects i n sec. 2.9. I a l s o give some examples of postposing e f f e c t s that I cannot e x p l a i n through a s y n t a c t i c a n a l y s i s , and which may r e q u i r e a c o n s i d e r a t i o n of dis c o u r s e f a c t o r s i n e x p l a i n i n g what makes a postposing o p e r a t i o n grammatical or not. 185 6.1 Postposed phrases l i k e to be "heavy" Although I cannot give a precise d e f i n i t i o n of what i t means for a phrase to be "heavy", the following example shows that postposing r e s u l t s i n a more grammatical sentence when the postposed element i s "heavy" -- for example by v i r t u e of containing c l a u s a l material, than when i t i s not. (283) : ??Masao-wa, okane-o nusunda no-ni, sikararenakatta, Taroo-no. -top money-acc st o l e although scold-pass.-neg-past -gen In s p i t e of the fact that Masao st o l e money, he wasn't scolded-- Taro's (money) (284) (postposed phrase made heavier) ?Masao-wa, okane-o nusunda no-ni, sikararenakatta, t o n a r i - n i sunde-iru Taroo-no. -top money-acc st o l e although scold-pass.-neg-past next-door l i v e s -gen In spite of the fact that Masao st o l e money, he wasn't scolded--Taro, who l i v e s next door's (money) In (284), the sentence shows a d e f i n i t e improvement when the postposed DP i s made heavier by adding a r e l a t i v e clause. This phenomenon i s reminiscent of Heavy NP S h i f t i n English as well as the r e l a t i v e heaviness that i s required for the extraposed clause i n r e l a t i v e clause extraposition i n English. For example, i n the following p a i r of English examples, we see that an NP that contains c l a u s a l material can be s h i f t e d to the r i g h t of i t s canonical p o s i t i o n whereas a non-heavy NP cannot. 186 (285) (a) I saw e; yesterday i n my backyard [that next-door neighbour of mine named Taro]. (b) *I saw e; yesterday i n my back yard Taro;. The (b) sentence i s highly unnatural unless the object "Taro" receives focus by being stressed. This fact shows that i n English as well, heaviness cannot be defined i n purely syntactic terms: discourse factors such as, i n t h i s case, focus, are important as well. 6.2 Analysis of postposed subject with object i n canonical p o s i t i o n For some speakers, when a ga-marked subject i s postposed, i t becomes d i f f i c u l t to coindex the matrix object with a name embedded i n the postposed subject (for example, embedded as a g e n i t i v e ) , even though such coindexing i s possible when the subject i s not postposed. I do not derive any firm conclusions from the following data, but have included i t because the way a speaker places these examples i n a discourse context seems to a f f e c t h i s or her judgements. Consider the following examples i n which a subject i s postposed: (286) TaroOi-no tomodati-ga kare—o semeta. (canonical sentence) -gen friend(s)-nom him-acc blamed Taro's friends blamed him. (287) (a) kare—o semeta, TaroOj-no tomodati ga. (underlining indicates stress) him—acc blamed T.j-gen f riend (s) -nom (They) blamed him; -- Taroj's f r i e n d s . 187 (287) (b)*kare i-o semeta, Tarooj-no tomodati ga. hin^-acc blamed T.i-gen friend(s)-nom (They) blamed himi -- TarOi' s friends. (288) ?kare 4-o semeta, [TaroOj-o yonda tomodati-ga] hiir^-acc blamed [T. ;-acc c a l l e d friends-nom (They-j) blamed him 4 -- [the friends who c a l l e d T a r o j j . For some speakers, (287) i s only possible when there i s stress on the verb "blamed" and no stress on the postposed phrase. In addition, kare must r e f e r to someone i n the speaker's in-group and "Taro's friends" to people i n the speaker's out-group. These judgements suggest that discourse factors strongly a f f e c t speaker-judgements of these examples. The "awkwardness" i n getting coreference unless c e r t a i n discourse and focus conditions are met may be due to a tendency for speakers to be s e n s i t i v e to context i n judging sentences. If we were to t r y to explain the d i f f i c u l t y of coreference for some speakers i n (287) i n purely syntactic terms then the following observations may be relevant: 1. Many speakers seem to judge (287) as being better than (288), even those who have less trouble getting coreference i n (287) . The fact that coreference becomes easier when the name i s embedded in a r e l a t i v e clause instead of a genitive i s highly suggestive of Condition C reconstruction. 2. If what we are seeing here i s i n fact a Condition C e f f e c t then we would have to assume that the subject i s being reconstructed from an A' p o s i t i o n to a p o s i t i o n where i t i s c-commanded by the object. This means that the object must have moved as we l l . It i s possible that postposing the subject induces some kind 188 of focusing e f f e c t on the object that requires that i t move to a higher p o s i t i o n . 6 1 If the object were o b l i g a t o r i l y r a i s e d (scrambled?) when the subject i s postposed, we might expect that t h i s i s a focusing e f f e c t , yet a speaker who had d i f f i c u l t y with coreference in (287) stated that kare cannot be focused by r e c e i v i n g stress i n t h i s example. 6.2.1 VP- and IP-level adverb test In t h i s subsection I s h a l l t r y to determine whether the deviance of (287) (b) i s due to forced r a i s i n g of the object when the subject i s postposed. Whitman (1991a) uses the p o s i t i o n of VP- and I P - l e v e l adverbs to determine whether a constituent i s i n VP or IP 6 2: i f the object appears to the r i g h t of a VP-level adverb such as "ravenously" i n (289) or (235) below, i t must be i n VP; i f i t appears to the l e f t of a sentence-level adverb, i t must have been outside of the VP: 6 1Michael Rochemont (p.c.) observes that i f postposing the subject requires r a i s i n g of the object, such movement i s reminiscent of l o c a t i v e i n v e r s i o n i n E n g l i s h , i n which one movement appears to hinge on another ( i . e . subject-aux. i n v e r s i o n on l o c a t i v e preposing.) If we were to claim that the object i s i n fact being focused i n sentences l i k e (287) we would need to t e s t t h i s c l a i m with question-answer p a i r s . Locative i n v e r s i o n i n E n g l i s h and postposing i n Japanese are not exactly p a r a l l e l , however, since the two movements involved i n l o c a t i v e i n v e r s i o n are leftward, whereas i n postposing, apparent r a i s i n g of the object i s leftward and I am assuming postposing to be rightward. 6 2Whitman (1991a) does not appear to have any d i r e c t independent evidence that the adjunction s i t e of what he r e f e r s to as "modal" (e.g. "surely") and "manner" adverbs (e.g. "hungrily", "unsteadily") d i f f e r s with respect to adjunction to IP or to VP. He shows that the p o s i t i o n of a subject with respect to these adverbs determines the e l i g i b i l i t y of a n e u t r a l d e s c r i p t i o n versus an exhaustive l i s t i n g reading of the sentence and c i t e s evidence i n works such as Diesing (1988) that (a) subjects of EL and ND sentences d i f f e r with respect to whether the subject i s VP-internal or external and (b) that "the ND reading i s g e n e r a l l y u n a v a i l a b l e for nominative subjects of i n d i v i d u a l l e v e l predicates i n the sense of Carlson (1978)." He a l s o notes that "[t]here are t h e o r e t i c a l and empirical d i f f i c u l t i e s with the hypothesis that the subject of i n d i v i d u a l l e v e l predicates o r i g i n a t e s external to VP." In my a n a l y s i s of postposing, I have been assuming that subjects are V P - i n t r e r n a l , i n order to explain, for example, why the trace of a subject postposed out of an i s l a n d does not seem to s u f f e r an ECP e f f e c t . It would be usefule to examine whether the traces of subjects of i n d i v i d u a l l e v e l predicates can be postposed out of an i s l a n d . 189 (289) ?(mori-de-wa) [VPmusyamusya banana-o t a b e ] - t e - i r u , gorira-ga. f o r e s t - l o c - t o p ravenously banana-acc i s - e a t i n g gorilla-nom In the forest ( i t ) i s eating bananas ravenously-- a g o r i l l a . (290) ?(mori-de-wa) banana-o [VPmusyamusya ta b e ] - t e - i r u , gorira-ga. f o r e s t - l o c - t o p banana-acc ravenously i s - e a t i n g gorilla-nom In the forest ( i t ) i s eating bananas ravenously -- a g o r i l l a . (291) (mori-de-wa) banana-o k i t t o [ V Ptabe]-te-iru, gorira-ga. f o r e s t - l o c - t o p banana-acc surely i s - e a t i n g gorilla-nom In the forest (it) i s surely eating bananas -- a g o r i l l a (291) was judged to be better than (289) and (290) . This r e s u l t could be explained i f the object undergoes r a i s i n g when the subject i s postposed. In (291) the object occurs to the l e f t of the sentence-level adverb "surely", which, according to Whitman's hypothesis about the p o s i t i o n of what he r e f e r s to as "modal" adverbs, adjoins to IP, not VP: If t h i s i s true, then i n (291) the object i s outside of VP. In (289) the presence of a VP-level adverb to the l e f t of the object suggests that here, the object has not been r a i s e d out of VP. The fact that (289) was judged as worse than (235) supports the hypothesis that the object must r a i s e when the subject i s postposed. What I cannot explain, however, i s why (290) i s marginal. If postposing the subject requires r a i s i n g of the object, i t i s possible for the object to have r a i s e d i n (290) to the same p o s i t i o n as i t does i n (291), since a l l we know in (290) i s that the object occurs to the l e f t of the VP-level adverb "hungrily." The structure of the three above examples i s shown below. 190 CP / \ PP c I IP / \ DP I' obj.(291) / \ adv. I' surely | VP / \ trace of postposed subj. V / \ adv. V' hungrily / \ DP obj.(289) 7. Summary 191 Considering the fact that Japanese i s a language well studied by l i n g u i s t s , the postposing construction has received r e l a t i v e l y l i t t l e a t tention i n the generative l i t e r a t u r e . Endo (1989) and Whitman (1991b) are the only recent studies I am aware of. Given the fact that scrambling and P P - t o p i c a l i z a t i o n are the only other forms of d i s l o c a t i o n i n the language that show good evidence of being derived by movement, a study of postposing holds the p o t e n t i a l for shedding new l i g h t on the nature of syntactic movement i n Japanese. Because the phenomenon of postposing has not yet been exhaustively studied, my analysis i s intended to be a " f i r s t pass" at explaining the phenomenon. The hypotheses I have proposed in t h i s paper w i l l need further refinement and modification and further t e s t i n g with new empirical data. In chapter 1 I argued that postposing i s derived by movement, not base-generation, because i t shows Condition C reconstruction e f f e c t s and a n t i -reconstruction e f f e c t s to the extent that we are able to test for them. In chapter 2 I showed that subject-object asymmetries are apparent i n postposing: s p e c i f i c a l l y , that a subject i s a more severe b a r r i e r for extraction than an object or even an adjunct. In chapter 3 I gave evidence that postposing shows r e l a t i v i z e d e f f e c t s with respect to a [+N] head: that i s , severe deviance r e s u l t s when a constituent with a [+N] feature i s postposed out of a higher phrase that also has a [+N] feature. This phenomenon does not apply to genitives, which, I suggest may be unmarked for a [+N] feature, possibly because the p a r t i c l e no that marks geni t i v e s i s a default functional head that i s unmarked for such features. In chapter 4 I examined the question of why postposing a r e l a t i v e clause 192 or a g e n i t i v e out of a topic i s l a n d i s sometimes grammatical and sometimes deviant. I explained the relevant data by proposing that d i f f e r e n t kinds of adjuncts behave d i f f e r e n t l y when they are extracted from an i s l a n d . We need to d i s t i n g u i s h between adjuncts that are i n the domain of a [+V] head from those that are not. The former set, but not the l a t t e r , show properties akin to being head-governed. Even though they are adjuncts, being i n the domain of a [+V] l e x i c a l head makes i t possible for them to be extracted, sometimes marginally, from a weak i s l a n d such as a top i c . In chapter 5 I showed that the p o s i t i o n I am assuming for postposed phrases -- right-adjoined to the root node, i s unique under the concept of minimal domains of Chomsky (1995) .1 also examined two questions concerning precedence in Japanese: (a) why i n a l e f t - a t t a c h i n g language l i k e Japanese r i g h t adjunction i s possible, but only to the root node and (b) why l e f t adjunction appears to be pos s i b l e except to the root node. I proposed an answer to these questions through a statement of the head parameter for Japanese. The conditions I proposed are motivated by the way sentences might be processed i n a l e f t - a t t a c h i n g language. 193 Appendix A Early Analyses of the Postposing Construction The following i s a b r i e f synopsis of pre-1989 analyses of the postposing construction as discussed i n Endo (1989) along with Endo's comments on these analyses. Inoue ( 19 7 8 ) 6 3 argues for a non-movement analysis, regarding postposed elements as a " p a r t i a l r e p e t i t i o n " of the main sentence. She groups together both sentences i n which the postposed element corresponds to a gap i n the main sentence and those i n which i t corresponds to a resumptive pronoun. Endo argues that the s o - c a l l e d resumptive pronoun examples, a l l of which use epithets such as a i t u , (a deprecatory form of ano hito, "that person") are i n fact a d i f f e r e n t construction from true postposed sentences that have a gap i n the main sentence. She shows that i n sentences involving epithets, the order of the postposed element and the epithet with which i t i s c o r e f e r e n t i a l can i n fact be reversed, with the epithet occurring as the postposed element. 6 4 She shows that i n the "resumptive" examples i t i s possible to have any phrase postposed, and coref e r e n t i a l with another phrase i n the main sentence as long as the l a t t e r e i t h e r (a) contains a demonstrative pronoun such as ano "that one" (or a form of i t as part of the morphology of a word of the phrase) or (b) gives more s p e c i f i c information about the former. Endo also shows that embedded clauses, adjective and adverbials can be postposed but that no resumptive pro-form i s a v a i l a b l e for them. Kuno (1978b) argues that postposed phrases are "afterthoughts." He gives an example of a l o c a t i v e phrase postposed out of an embedded wh-island which he said would be a serious v i o l a t i o n of Ross (1967)'s hypothesis that rightward movement i s upward bounded. Kuno proposes that what can be postposed i s constrained not s y n t a c t i c a l l y but semantically and pragmatically, saying that " ( i ) [p]ostverbal elements are either discourse-predictable [by the speakers assumptions] or supplementary; therefore, the sentences should have made sense without them65, ( i i ) Elements that would change the i n t e r p r e t a t i o n of the f i r s t part of the sentence cannot appear postverbally." Endo(1989:38) gives examples from taped conversations that show that elements that can change the i n t e r p r e t a t i o n of the f i r s t part of the sentence can appear postverbally -- for example negative sentences i n which a postposed negative p o l a r i t y adverbial such as anmari "(not) much" changes the meaning from "not at a l l " to "not much", or NPI sika "only, (no) more than" which o b l i g a t o r i l y occurs with negation changes the meaning from "not at a l l " to "only (three times)." Endo (1989:41) also shows that whereas small pro i n a main sentence can only be interpreted as d e f i n i t e , a postposed DP can be interpreted as being i n d e f i n i t e : that i s , i n the non-postposed part of the sentence, we have an empty category that receives an i n d e f i n i t e i n t e r p r e t a t i o n . If only the main part of the sentence i s interpreted, the gap l e f t by the postposed phrase w i l l be interpreted as a small pro, and can only receive a d e f i n i t e i n t e r p r e t a t i o n . But i f the whole sentence, including the postposed phrase i s interpreted, the gap can receive an i n d e f i n i t e i n t e r p r e t a t i o n . The following example from Endo (1989:41) i l l u s t r a t e s t h i s f a c t . This work i s w r i t t e n i n Japanese. 6 4 I n t e r e s t i n g l y , the example she uses i s i d e n t i c a l to one that Whitman (1991b) uses to discuss the use of resumptive pronouns i n postposed sentences. 6 5 T h i s a s s e r t i o n i s reminiscent of a s i m i l a r g e n e r a l i z a t i o n made i n Whitman (1991). 194 (293) (a) depaato-de e(pro) katta no? dept.-store-loc bought C Did you buy it/them at the department store? (only d e f i n i t e i n t e r p r e t a t i o n i s possible) (b) depaato-de ei katta no? nani-kai. dept.-store-loc bought C somethingj Did you buy something at the department store? ( i n d e f i n i t e i n t e r p r e t a t i o n i s possible) Kuno (1978a) follows Inoue (1978) i n regarding the postposed phrase as " p a r t i a l r e p e t i t i o n of an e l l i p t i c a l f i r s t c lause" 6 6 Kuno argues that the postposed element cannot represent "new information", thus disallowing ga-marked phrases i n most cases, with the exception "that [a] ga-marked element can appear sentence f i n a l l y i f the whole sentence expresses new information. 1 , 6 7 Kuno proposes that a postposing construction i s derived from two i d e n t i c a l clauses with complementary elements deleted i n each. He i s unable to explain why a phrase can be postposed out of an embedded question but not out of a r e l a t i v e clause. In discussing Kuno (1978a) Endo (1989:54) shows that i t i s i n fact possible to postpose, under c e r t a i n conditions, [discourse-]new information, such as phrases marked with contrastive wa or with "exhaustive l i s t i n g " ga and that constraints on whether these elements can be postposed are s y n t a c t i c . Saito (1985) notes three differences between what he c a l l s " r i g h t -d i s l o c a t i o n " and scrambling: (a) postverbal elements can appear only to the rig h t of a matrix verb (b) there i s no subject-object asymmetry for postposing but there i s for scrambling and (c) overt resumptive pronouns are possible i n postposed phrases (but not scrambled ones.) Endo (1989) argues against the claim that the pronoun-like elements that can be c o r e f e r e n t i a l with a postverbal phrase are i n fact resumptive pronouns, for reasons c i t e d above. Haraguchi (1973) argues for a movement analysis of postposing based on s e l e c t i o n a l r e s t r i c t i o n s that must be s a t i s f i e d between the postposed element and i t s head. Kuroda (1980) observes that s t r i c t head-final order i s always observed i n dependent clauses but that i t can be v i o l a t e d i n matrix clauses (i.e when postposing occurs) and he proposes that postposed sentences are derived by a transformation. In summary, a number of researchers have looked at postposing as a synt a c t i c phenomenon, but Endo's i s the only study I am aware of that (a) treats a large body of postposing data and (b) analyses the construction within a r e l a t i v e l y recent generative framework -- i . e . Chomsky (1981) or l a t e r . 6 6 E n d o ( 1 9 8 9 : 4 3 ) 6 7 q u o t e d f r o m K u n o ( 1 9 7 8 a : 7 6 ) i n E n d o ( 1 9 8 9 : 4 6 ) 195 Appendix B: No-marked phrases In chapter 3 I showed that no-marked phrases behave more l i k e r e l a t i v e clauses than l i k e DP's i n that they are not blocked by a maximal p r o j e c t i o n with a [+N] feature when they are postposed out of a DP. In order to explain t h i s behaviour of no-marked phrases I suggested that they are unmarked for the feature [+N] . In t h i s appendix I s h a l l explore the p o s s i b i l i t y that no i s a default functional head that serves many purposes i n the grammar and which i s unmarked for a [+N] feature. Types of no The p a r t i c l e no can serve a number of d i f f e r e n t functions, which are enumerated below: 1. combines with DP's to form phrases that modify other DP's. These phrases can have various kinds of semantic functions: a) possessor (294) Taroo-no hon book "Taro's book" b) a d j e c t i v a l (295) betu-no hon d i f f e r e n t book "a d i f f e r e n t book" 2. combines with PP's to allow them to modify a DP (296) Tookyoo-kara-no densya Tokyo-from t r a i n 'the t r a i n from Tokyo" 3. no i s the form of the copula that occurs i n r e l a t i v e clauses (297) haisya-no otooto dentist brother "my brother who i s a d e n t i s t " 4. complementizer that heads case-marked complement CP's (s i m i l a r to koto clauses) (298) Taroo-ga kaetta no-o s i t t e - i r u -nom went-home comp-acc know "I know that Taro went home." 5. complementizer found at the end of matrix clauses i n informal questions or "feminine speech" (299) ano hon-o motte-kita no? that book brought comp "Did you bring that book?" 6. marker of "genitive subject" of a r e l a t i v e clause (300) Masao-wa, okane-o nusunda koto-ga aru, boku-no kakusita. -top money-acc sto l e comp-nom exi s t I-gen h i d Masao has (on occasion) stolen money -- that I h i d . 196 (see Miyagawa (1991) for further examples and discussion of genitive subjects.) 7. "case-marked genitive" pro-DP 6 8 (301) Taroo-no-o -gen-acc "Taro's" can be a pro-form for a DP of the form: D P [Taroo-no DP] (302) Taroo-no hon-o -gen book-acc "Taro's book" T r a d i t i o n a l l y , these d i f f e r e n t forms of no have been considered to be separate l e x i c a l items -- for example #1 as the genitive case marker, #2, #3 as a copula V, #4, #7 as an N, etc. Here, I s h a l l investigate the p o s s i b i l i t y that there i s one no i n Japanese that serves as a default multi-purpose functional head. Problems with no as a genitive case assigner The p a r t i c l e no i s associated with what has been r e f e r r e d to as genitive case i n Japanese. There are two p o s s i b i l i t i e s for the r o l e of no: (!) that i t i s a case-marker that i s a phonetic r e a l i z a t i o n of the assignment of gen i t i v e case, which i s assigned i n some way -- for example inherently or (2) that i t i s a geni t i v e case-assigner: a functional head that assigns case to an element i n i t s domain. If we assume p o s s i b i l i t y (1) , that no i s a phonetic r e a l i z a t i o n of inherent g e n i t i v e case, then i t becomes d i f f i c u l t to explain a number of things. F i r s t , unlike English, Japanese allows multiple genitives for the same head noun. 6 9 If multiple genitives a l l receive case inherently, we must assume that they a l l have some theta r e l a t i o n with the head noun. Secondly, genitive phrases are r e a d i l y postposed. In English, movement i s not possible for elements that are assigned inherent case -- for example, genitives or datives cannot be t o p i c a l i z e d and cannot undergo heavy NP s h i f t 7 0 . F i n a l l y , genitives and r e l a t i v e s seem to be able to interchange t h e i r word order: see examples (3 0.3) and (304) below) If a r e l a t i v e clause can intervene between a genitive and the noun i t modifies, i t 6 8 F o r a d i s c u s s i o n o f t h i s k i n d o f c o n s t r u c t i o n s e e M u r a s u g i (1991) a n d S a i t o a n d M u r a s u g i ( 1 9 9 0 ) . 6 9 I n t h e f o l l o w i n g e x a m p l e , t h r e e d i f f e r e n t g e n i t i v e p h r a s e s m o d i f y t h e n o u n " p a n t s " . We a l s o s e e t h a t i t i s p o s s i b l e t o i n t e r c h a n g e t h e o r d e r o f t h e g e n i t i v e s . ( i ) T a r o o - n o n a t u y o o - n o r n i z u t a r n a - m o y o o - n o p a n t u - g e n s u m m e r - u s e p o l k a - d o t - p a t t e r n - g e n p a n t s " T a r o ' s p o l k a - d o t p a t t e r n e d summer p a n t s " ( i i ) n a t u - n o m i z u t a m a - m o y o o - n o T a r o o - n o p a n t u ( i i i ) m i z u t a m a - m o y o o - n o n a t u y o o - n o T a r o o - n o p a n t u 7 0 T h e f o l l o w i n g e x a m p l e s show t h a t E n g l i s h d a t i v e s a n d g e n i t i v e s c a n n o t u n d e r g o t o p i c a l i z a t i o n o r H N P S : ( i ) * J o h n , , M a r y g a v e t ( t h a t b o o k , ( d a t i v e c a n n o t be t o p i c a l i z e d ) ( i i ) * J o h n ' s , , M a r y g a v e [t , b o o k ] . ( g e n i t i v e c a n n o t be t o p i c a l i z e d ) ( i i i ) * M a r y g a v e t ; t h a t b o o k y e s t e r d a y [ t h e man who u s e d t o l i v e n e x t d o o r ] j . ( d a t i v e c a n n o t u n d e r g o HNPS) ( i v ) * M a r y g a v e J o h n [ t ( b o o k ] y e s t e r d a y ' [ t h e man who u s e d t o l i v e n e x t d o o r ] ' s ( . 197 would seem u n l i k e l y that genitives are assigned inherent case by t h e i r r e l a t i o n s h i p with the head noun. (303) Taroo-no [Masao-ga kinoo yonda] hon -gen -nom yesterday read book Taro's book that Masao read yesterday (304) [Masao-ga kinoo yonda] Taroo-no hon -nom yesterday read -gen book Taro's book that Masao read yesterday. For these reasons, case assignment by no as a case-assigning functional head seems a more p l a u s i b l e hypothesis. 7 1 Is no a P? One p o s s i b i l i t y that comes to mind i s that no i s a P, acting as a default case-assigner i n a manner p a r a l l e l to "of" i n English or "de" for French possessor DP's. However, i f no i s a P that assigns case, we need to be able to explain why a PP headed by no i s the only type of PP that can d i r e c t l y modify a noun. Consider the following examples: (305) *Tookyoo-kara densya Tokyo-from t r a i n the t r a i n from Tokyo (306) *Tookyoo-made densya Tokyo-as-far-as t r a i n the t r a i n to Tokyo (307) *mizu-ni oto water-in sound a sound i n the water (308) *Tookyoo-(h)e densya Tokyo-to t r a i n the t r a i n to Tokyo Phrases headed by P-like p a r t i c l e s kara "from", made "as far as", ni "to", " i n " , (h)e " i n the d i r e c t i o n of" can never d i r e c t l y modify nominal phrases. However, they can when t h e i r phrases have the p a r t i c l e no s u f f i x e d to them: 7 1 0 n t h e o t h e r h a n d t h e r e i s some e v i d e n c e t h a t g e n i t i v e c a s e i s a s s i g n e d d i f f e r e n t l y i n J a p a n e s e t h a n n o m i n a t i v e o r a c c u s a t i v e c a s e . N o m i n a t i v e a n d a c c u s a t i v e D P ' s c a n o c c u r w i t h b a r e q u a n t i f i e r s s u c h a s san-nin " t h r e e p e o p l e " , b u t g n e i t i v e s c a n n o t , a s shown b e l o w : ( i ) g a k u s e i - g a s a n - n i n s t u d e n t - n o m t h r e e - p e o p l e ( i i ) g a k u s e i - o s a n - n i n s t u d e n t - a c c t h r e e - p e o p l e ( i i i ) * g a k u s e i - n o s a n - n i n s t u d e n t - g e n t h r e e p e o p l e T h i s c o n t r a s t was p o i n t e d o u t b y H i s a t s u g u K i t a h a r a ( p . c ) . 198 (3 09)Tookyoo-kara-no densya Tokyo-from t r a i n the t r a i n from Tokyo (310) Tookyoo-made-no densya Tokyo-as-far-as t r a i n the t r a i n to Tokyo (311) mizu-no oto water-in sound a sound i n the water (312) Tookyoo-(h)e-no densya Tokyo-to t r a i n the t r a i n to Tokyo In the case of p a r t i c l e ni, i t must be dropped when no i s added. This suggests that n i may be a case-marker rather than a P. This matter i s discussed further below. There are several ways of explaining the above data. One i s that Japanese lacks the category P (except, perhaps, for no), and that the p a r t i c l e s that resemble P's (of which there are only about f i v e , excluding no) are simply case markers. If they are merely case-markers, then they cannot assign case and we can explain why some other element such as the p a r t i c l e no i s need to li c e n s e them when they modify a noun, which cannot assign case. But there are problems with t h i s proposal. F i r s t of a l l , these p a r t i c l e s seem to have f a i r l y s p e c i f i c semantic functions -- for example kara s i g n i f i e s o r i g i n , (h)e d i r e c t i o n , etc. If they are case markers, then we would expect that they are a r e a l i z a t i o n of case-assignment by some kind of theta-marking head, such as a V, and that each p a r t i c l e w i l l r e f l e c t a p a r t i c u l a r set of theta-case r e l a t i o n s . But i n examples (309) -(312) above, there i s no theta-marking head that takes nouns "Tokyo" or "water" as an argument. For example, i n (312), "Tokyo" i s not an argument of noun " t r a i n " . Nor does the p a r t i c l e no have any theta-assigning property here since i t has no semantic value of i t s own. Let us assume, then that the p a r t i c l e s i n the examples above are P's and that P's must be licensed by no when they modify a noun. The fact that phrases headed by no can modify nouns but those headed by p a r t i c l e s such as kara, "from", and made "as far as" cannot, suggests that no has d i f f e r e n t properties from these other p a r t i c l e s , which we have concluded above behave more l i k e P's. Not only does no behave much more l i k e a functional head, but when i t combines with PP-like phrases, there i s good reason to po s i t that the no we are seeing here i s the copular no that occurs i n r e l a t i v e clauses (See (297) above) and that examples l i k e (309)- (312) are a c t u a l l y r e l a t i v e clauses. In the following examples, we s h a l l see that the- set of P-like elements that can modify a noun by the i n s e r t i o n of no c l o s e l y matches the set that can form a predicate with i n f l e c t i o n provided by the copula. Consider the following examples: (313) Tookyoo-kara no densya -from t r a i n the t r a i n from Tokyo (314) Tookyoo-kara da. -from copula It i s from Tokyo (315) Tookyoo-made no densya -as-far-as t r a i n the t r a i n to Tokyo 199 (316) Tookyoo-made da. -as-far-as copula It i s (going) to Tokyo (317) Tookyoo-he no densya -to t r a i n the t r a i n to Tokyo (318) Tookyoo-he da. -to copula (319) eigo-de no benkyoo. E n g l i s h - i n s t r . study study(ing) using English (320) eigo-de da. E n g l i s h - i n s t . copula It i s by means of English. (321) *mizu-ni no oto - i n sound a sound i n the water (322) ?mizu-ni da. (See f t n . 72.) - i n copula It i s i n the water. The fact that p a r t i c l e s such as kara, made, and (h)e can both modify nominals by no-insertion and be followed by the copula and that ni cannot 7 2 i s most e a s i l y explained i f the former set of p a r t i c l e s are P's and the l a t t e r are case-markers. Notice that the nominative and accusative case-markers pattern with ni i n not allowing no i n s e r t i o n and not being able to be followed by the copula: (323) *Taroo-ga/o da. 7 3 -nom/acc copula It i s Taro (324) *Taroo-ga/o-no h i t o -nom/acc person the person who i s Taro Suppose that the f i r s t set of p a r t i c l e s (kara,"from" made,"as far as" 7 2 T h e s t a t u s o f ni i s a c t u a l l y u n c l e a r . A l t h o u g h p a r t i c l e n i c a n n o t o c c u r a f t e r p a r t i c l e no a s s e e n a b o v e , i n some c o n t e x t s i t c a n o c c u r w i t h t h e c o p u l a i n a s e n t e n c e w i t h a c o n t r a s t i v e i n t e r p r e t a t i o n , a s we s e e i n t h e f o l l o w i n g e x a m p l e , w h i c h was p o i n t e d o u t b y H i s a t s u g u K i t a h a r a ( p . c ) : ( i ) k a e r u - g a t o b i k o n d a - n o - w a (kawa d e n a k u ) i k e - n i d a . f r o g - n o m j u m p e d - i n - C - t o p r i v e r - c o p . - n e g . p o n d - i n t o c o p . What t h e f r o g j u m p e d i n t o was ( n o t t h e r i v e r b u t ) t h e p o n d . I t i s p o s s i b l e t h a t t h e e x a m p l e a b o v e , mizu-ni-da i s s i m p l y a w k w a r d , b e c a u s e t h e r e i s n o c o n t e x t g i v e n t o g e t a c o n t r a s t i v e i n t e r p r e t a t i o n . I l e a v e t h e q u e s t i o n o f t h e n a t u r e o f ni-m a r k e d p h r a s e s u n s o l v e d h e r e . 7 3 A DP t h a t i s c a s e - m a r k e d a s n o m i n a t i v e c a n o c c u r a s s u b j e c t o f a DP p r e d i c a t e t h a t i s l i c e n s e d by t h e c o p u l a , b u t i n s u c h a c a s e , t h e n o m i n a t i v e DP o c c u r s o u t s i d e o f t h i s p r e d i c a t e : T a r o o - g a g a k u s e i - d a . - n o m s t u d e n t - c o p u l a I t i s T a r o o who i s a s t u d e n t . 200 (h)e,"towards", are P's but that i n Japanese, PP's cannot occur i n t h e i r bare form as modifiers of a DP but must be licensed must be licen s e d by a functional head. When a PP occurs as a complement to a verb i t w i l l be licensed by the V or I head of the verbal p r o j e c t i o n . But when i t occurs as a modifier of a noun, i t must be licensed by functional head no. S i m i l a r l y , when a PP headed by one of the f i r s t set of p a r t i c l e s occurs with the copula, i t i s forming a predicate, and the copula serves simply to provide i n f l e c t i o n . If the second set of p a r t i c l e s ( i . e . ni "dative", ga "nominative", o "accusative") are s t r u c t u r a l case-markers, then they s i g n i f y a r e l a t i o n with a theta-assigning and case-assigning verbal head, since phrases they head occur as arguments of verbs or adjectives. Phrases that they head cannot occur as modifiers of a noun, since a noun cannot assign s t r u c t u r a l case. Nor can they be licens e d by no i n s e r t i o n since no does not assign a theta r o l e . If t h i s hypothesis i s correct, then no f i t s into neither group: i t does not r e f l e c t assignment of s t r u c t u r a l case by a verbal head nor does i t show the same properties as p a r t i c l e s that behave l i k e P's. In at least some of i t s roles, no behaves as a kind of default l i c e n s e r : i t licenses PP's that modify DP's as we saw above and i t provides i n f l e c t i o n i n i t s r o l e as the copula i n r e l a t i v e clauses. As far as the P-like p a r t i c l e s are concerned, we have seen that they must be lic e n s e d by no when they modify a noun. There are three environments i n which they can occur, as shown below. 1. as arguments 7 4 or adjuncts of a V (325) Tookyoo-kara Kyooto-made i t t a -from -as-far-as went I from Tokyo as far as Kyoto. 2. as modifiers of a noun, mediated by no-insertion (326) Tookyoo-kara-no densya Tokyo-from t r a i n the t r a i n from Tokyo 3. as predicates, mediated by the copula (small clause predicates are not possible i n Japanese) (327) Tookyoo-kara-da -from-copula It i s from Tokyo. If no i s a P, then i t exhibits i n e x p l i c a b l y d i f f e r e n t behaviour from a l l other P's i n the language i n that PP's headed by no can d i r e c t l y modify a noun but no other PP's can. No, in i t s r o l e as a li c e n s e r of bare PP modifiers of a noun patterns more l i k e verbal elements, since PP's can be licen s e d by (a) no, (b) i n f l e c t e d verbs, (c) the copula, and also by bare verbs as we see below. The following example suggests that a V head without i n f l e c t i o n i s s u f f i c i e n t to licen s e a PP: 7 4 I n t h i s e x a m p l e I a s s u m e t h a t Kyooto-made i s a n a r g u m e n t s i n c e t h e v e r b " t o g o " t a k e s a n o b l i g a t o r y g o a l a r g u e m n t . I t i s n o t c l e a r w h e t h e r t h e PP Tookyoo-kara i s a n a r g u m e n t o r a d j u n c t . 201 (328) Tookyoo-made i k i - w a s i - n a i . -P go-top do-neg As f o r going to Tokyo, I don't do i t . In t h i s case of VP t o p i c a l i z a t i o n , the PP i n s i d e the VP t o p i c i s not i n the domain of any overt i n f l e c t i v e head since the verb occurs i n i t s bare "conjunctive" form. I t i s noteworthy that i n Japanese, AP's p a t t e r n w i t h PP's i n that they need to be l i c e n s e d by i n f l e c t i o n or a copular element. A d j e c t i v e phrases i n Japanese must e i t h e r be i n f l e c t e d l i k e VP's i f the head a d j e c t i v e i s i n f l e c t i b l e , or i f not, (the case of "nominal" a d j e c t i v e s ) they must be supported by i n s e r t i o n of no or i t s v a r i a n t na. The l a t t e r p a r t i c l e i s another form of the copula. These requirements h o l d whether they are pr e d i c a t e s or whether they are m o d i f i e r s of a noun. Although I cannot e x p l a i n why PP's and AP's i n Japanese seem to re q u i r e t h i s l i c e n s i n g , we can at l e a s t make a d e s c r i p t i v e g e n e r a l i z a t i o n : i n Japanese, AP's and PP's cannot merge w i t h a [+N] category on t h e i r own but r e q u i r e some k i n d of mediation or l i c e n s i n g . I f no i s a f u n c t i o n a l head that l i c e n s e s these AP's and PP's i n the kinds of c o n s t r u c t i o n s we saw above, i t i s p o s s i b l e that i t i s doing the same t h i n g i n what we c a l l g e n i t i v e phrases. I f the no that occurs i n what we c a l l g e n i t i v e phrases i s the same no that l i c e n s e s AP's or PP's - - p o s s i b l y a f u n c t i o n a l head of some k i n d -- then we appear to have two ways i n which DP's i n Japanese are l i c e n s e d : (i) when they are case-marked i n a t h e t a - r e l a t i o n w i t h a l e x i c a l head (e.g. nominative, ac c u s a t i v e , and d a t i v e case-marked DP's) ( i i ) when they are not case-marked but are l i c e n s e d by a an element such as no or the copula L i c e n s i n g of what we c a l l g e n i t i v e s by no could be an example of the l a t t e r case. P r e d i c a t e nominals 7 5 would a l s o f a l l i n t o t h i s case since they are not case marked but appear to be l i c e n s e d by the copula, which can a l s o take the form no i n r e l a t i v e c lauses. The f o l l o w i n g example shows that p r e d i c a t e nominals cannot be case-marked: (329) Taroo-wa gakusei (*-ga) (*-o) da. -top student copula Taro i s a student. So f a r , we have looked at no i n i t s r o l e as: 1. a marker of g e n i t i v e s 2. the form of the copula that occurs i n r e l a t i v e clauses 3. a l i c e n s e r of PP's and AP's 7 5 T h e r e i s a q u e s t i o n o f w h e t h e r p r e d i c a t e n o m i n a l s a r e D P ' s o r N P ' s . I s h a l l a s s u m e t h a t t h e y a r e D P ' s s i n c e b o t h names a n d N P ' s w i t h d e t e r m i n e r - l i k e e l e m e n t s c a n o c c u r a s p r e d i c a t e n o m i n a l s : ( i ) a n o g a k u s e i - w a T a r o o - d a . t h a t s t u d e n t - t o p c o p u l a T h a t s t u d e n t i s T a r o . ( i i ) b o k u - n o h o n - o n u s u n d a n o - w a , ano. g a k u s e i - d a . my b o o k - a c c s t o l e o n e - t o p t h a t s t u d e n t - c o p u l a T h e o n e who s t o l e my b o o k i s t h a t s t u d e n t . 202 In 3. i f no i s the copula, i t may be a V or I head that provides supporting i n f l e c t i o n , with no semantic value. In order to unify our account of no, i t would be desirable to say that no i s a C rather than an I i n i t s copular r o l e ; however the following facts argue against t h i s p o s s i b i l i t y . Suppose that i n a copular construction, no i s a C, with a n u l l I: (330) haisya-no otooto dentist-cop brother [DP n u l l - I C ] C P DP "my brother who i s a den t i s t " Then we have no explanation for the fact that i n constructions where no i s a C, the i n f l e c t i o n that precedes i t cannot be dropped. (331) otooto-wa haisya-na-no brother-top dentist-copula-C My brother i s a d e n t i s t . (332) *otooto-wa haisya-no brother-top d e n t i s t - ( n u l l I)-C Consider now some other i n s t a n t i a t i o n s of no. 4. complementizer that heads case-marked complement CP's (sim i l a r to koto clauses) 5. complementizer found at the end of matrix clauses i n informal questions or "feminine speech" 6. marker of "genitive subject" of a r e l a t i v e clause 7. "case-marked genitive" pro-DP e.g. Taroo-no-o replaces Taroo-no hon-o In (4) and (5) no i s a C. In (6), no marks or assigns case to the "genitive subject" of a r e l a t i v e clause. Miyagawa (1991:35) argues that the I n f l of a r e l a t i v e clause i n Japanese does not contain AgrS to assign nominative case to the subject of the r e l a t i v e clause, which receives genitive case from the D head of the r e l a t i v e clause. He does not, however, explain why r e l a t i v e clauses with genitive subjects are defective i n not being able to assign nominative case. The i n f l e c t i o n a l morphology on the main predicate of a r e l a t i v e clause does not d i f f e r from that of a matrix clause except when i t s i n f l e c t i o n i s supplied by the copula, i n which case the copula takes the form no instead of da as discussed e a r l i e r . I have not adopted Miyagawa's proposal that a Japanese r e l a t i v e clause i s a DP. Suppose, however, we adopt part of Miyagawa's hypothesis: namely, that a r e l a t i v e clause with a 203 genitive subject i s for some yet unexplained reason 7 6, unable to assign nominative case to i t s subject. Then no, as a l i c e n s e r of the subject, i s serving a s i m i l a r function to the no that licenses a predicate nominal or PP i n i t s r o l e as the copula or that licenses a PP or un i n f l e c t e d AP that modifies a DP. In ( 7 ) , phrases headed by "case-marked no" function as pro-DP's. For example Taroo-no-o "Taro-gen-acc" i s a pro-form for [Taroo-no DP]-o. We might po s i t that t h i s construction has the structure [Taroo-no pro]-o and that pro here i s lic e n s e d by no as a functional head. In summary, i t seems possible to argue up to a c e r t a i n point that at least some types of no-marked phrases are headed by a default functional head. In some cases no has the ro l e of an I head and i n others the r o l e of a C. Throughout the data I have looked at on postposing, ge n i t i v e s at times exhibit anomalous or contradictory behaviour: for example, they behave l i k e adjuncts rather than s p e c i f i e r s i n that they do no show SC e f f e c t s , yet in the Condition C reconstruction examples we looked at i n chapter 1, they behave d i f f e r e n t l y from r e l a t i v e clauses i n that they do not show the a n t i -reconstruction e f f e c t s that are associated with adjunct embedding. And in postposing out of [+N] constituents, they pattern d i f f e r e n t l y from subjects, objects, datives, and obliques, as we saw in chapter 3. It seems that the nature of no-marked phrases i n Japanese requires further i n v e s t i g a t i o n . 7 6 I c a n o f f e r t h e f o l l o w i n g s p e c u l a t i o n ' a b o u t why r e l a t i v e c l a u s e s w i t h " g e n i t i v e " s u b j e c t s c a n n o t a s s i g n n o m i n a t i v e c a s e . A s a r g u e d a b o v e , x o t o c l a u s e s show s u b s t a n t i a l d i f f e r e n c e s f r o m r e l a t i v e a n d c o m p l e m e n t c l a u s e s t h a t make them l o o k m o r e l i k e C P ' s t h a n D P ' s . F o r e x a m p l e , t h e y a r e w e a k e r b a r r i e r s t o movement t h a n c l a u s e s h e a d e d b y l e x i c a l n o u n s a n d t h e y do n o t p e r m i t " g e n i t i v e s u b j e c t s . " I f we a c c e p t t h e h y p o t h e s i s t h a t Jcoto c l a u s e s a r e C P ' s , t h e n t h e f o l l o w i n g p a r a d i g m s u g g e s t s t h a t i n J a p a n e s e i t i s p o s s i b l e f o r a n o u n t o b e m o d i f i e d b y a n I P . . ( i ) [ T a r o o - g a N i h o n - n i k a e t t a t o i u ] X P u w a s a w a . . . - n o m J a p a n - l o c r e t u r n e d C s a y r u m o u r t o p T h e r u m o u r t h a t T a r o r e t u r n e d t o J a p a n . . . ( i i ) [ [ T a r o o - g a N i h o n - n i k a e t t a t o i u ] I P k o t o ] C P w a . . . - n o m J a p a n - l o c r e t u r n e d C s a y C t o p T h e f a c t t h a t T a r o r e t u r n e d t o J a p a n . . . I n ( i i ) , i f koco i s a C , t h e n t h e c l a u s e X P , h e a d e d by t h e i d i o m a t i c " c o m p - s a y " c o n s t r u c t i o n i s a r g u a b l y an I P , n o t a C P . B u t t h e same i d i o m a t i c c o n s t r u c t i o n o c c u r s i n ( i ) , m o d i f y i n g a l e x i c a l n o u n . I t w o u l d seem u n l i k e l y t h a t t h e same i d i o m a t i c c o n s t r u c t i o n h a s a d i f f e r e n t s t r u c t u r e i n ( i ) . t h a n i n ( i i ) . T h u s , i n ( i ) we seem t o h a v e a l e x i c a l n o u n m o d i f i e d b y a n I P . S u p p o s e , t h e n t h a t r e l a t i v e c l a u s e s w i t h " g e n i t i v e s u b j e c t s " a r e I P ' s a n d n o t C P ' s . T h i s s u p p o s i t i o n i s s u p p o r t e d b y t h e f a c t t h a t i n J a p a n e s e , c l a u s e s t h a t a r e c o m p l e m e n t s o f v e r b s o r a d j e c t i v e s h a v e o b l i g a t o r y c o m p l e m e n t i z e r s . R e l a t i v e c l a u s e s w i t h a n u l l C w o u l d a p p e a r t o b e a n a n o m a l y i n t h e l a n g u a g e . I f J a p a n e s e r e l a t i v e c l a u s e s a r e I P ' s , a n d i f t h e y i n v o l v e some k i n d o f o p e r a t o r - v a r i a b l e c o n s t r u c t i o n . ( p o s s i b l y w i t h a n empty o p e r a t o r ) , t h e n t h e r e i s no C P t o h o s t s u c h a n o p e r a t o r . S u p p o s e t h e n , t h a t w h a t e v e r p r o j e c t i o n a s s i g n s n o m i n a t i v e c a s e m u s t b e u s e d t o h o s t t h i s o p e r a t o r a n d t h a t i t c a n n o t b o t h a s s i g n c a s e a n d h o s t a n o p e r a t o r . I f t h i s h y p o t h e s i s i s c o r r e c t , t h e n we c a n e x p l a i n why r e l a t i v e c l a u s e s , w h i c h i n v o l v e o p e r a t o r - v a r i a b l e c o n s t r u c t i o n s w o u l d r e q u i r e n o - i n s e r t i o n t o a s s i g n c a s e t o t h e s u b j e c t a n d o t h e r c l a u s e s t h a t may be I P ' s s u c h a s c o m p l e m e n t c l a u s e s do n o t . What I c a n n o t e x p l a i n i s why g a - n o c o n v e r s i o n i n r e l a t i v e c l a u s e s i s o p t i o n a l a n d n o t o b l i g a t o r y . 204 Bibliography Abe, Jun (1993) Chomsky,N. (1981) Chomsky, N. (1986) Chomsky, N. (1995) Diesing (1988) Endo-(Simon), Mutsuko Grimshaw, Jane (1991) Hale, K and S.J. Keyser Haraguchi, S. (1973) Hoji (1985) Huang (1982) Inoue, K. (1978) Kayne, Richard Kitahara, Hisatsugu Koike, Satoshi (1994) Koopman & Sportiche (19 Kuno, S (1973) Kuno, S (1978) Kuno,S (1978) Binding Conditions and Scrambling without the A- A' d i s t i n c t i o n Univ. of Connecticut, D i s s e r t a t i o n Lectures on Government and Binding Dordrecht, Foris Barriers MIT Press Cambridge The Minimalist Programme MIT Press, Cambridge Bare plural subjects and the stage/individual contrast U.Mass ms. (1989) The Postposing Construction In Japanese, Ph.D. Dissertation, University of Michigan, Ann Arbor Extended Projection Ms. (1993) "On Argument Structure and the l e x i c a l expression of syntactic r e l a t i o n s , " i n The View from Building 20 MIT Press, Cambridge Remarks on Dislocation in Japanese, Ms., MIT D i s s e r t a t i o n • Logical Form Constraints and Configurational Structures in Japanese Univ. of Washington Logical Relations in Chinese and the theory of grammar Ph. D. Dissertation, MIT, Cambridge Nihongo no bunpoo-kisoku Taisyukan, Tokyo (1994) The Antisymmetry of Syntax Ms. Graduate Centre, CUNY (1997) (to appear) Elementary Operations and Optimal Derivations MIT Press Cambridge Leftward Movement in Japanese Relative Clauses T i l b u r g Conference on Rightward Movement 91) The position of subjects Lingua 85 The Structure of the Japanese Language MIT Press Cambridge "Theoretical Perspectives on Japanese L i n g u i s t i c s , " i n J. Hinds and I. Howard, eds. "Japanese, a c h a r a c t e r i s t i c 0V language," i n Lehmann, ed., Syntactic Typology, Univ. of Texas Press, Austin Larson (1988) "On the Double Object Construction" L i n g u i s t i c Inquiry 19:335-391 Lasnik, H. and M. Saito (1984) "On the Nature of Proper Government," L i n g u i s t i c Inquiry 15:235-289 Lasnik, H. and M. Saito (1992) Move-alpha MIT Press, Cambridge 205 Miyagawa (1991) R i z z i , L . (1990) Ross (1967) Saito, Mamoru (1985) Saito, M (1989) Saito, M (1992) Saito, M (1994) Saito, M (1994) Whitman, J (1991) Whitman, J (1991) Nonconfigurationality Within a Configurational Structure i n Japanese Syntax and Semantics Relativized Minimality MIT Press, C a m b r i d g e Constraints on Variables in Syntax Ph. D. Dissertation, MIT, Cambridge D i s s e r t a t i o n Some Asymmetries in Japanese and their theoretical implications MIT "Scrambling as semantically vacuous A' Movement" i n A l t e r n a t i v e Conceptions of Phrase Structure , eds. B a l t i n and Kroch, 182-200. Uni v e r s i t y of Chicago Press, Chicago Long-distance Scrambling in Japanese JEAL 1 Scrambling and the functional interpretation of wh-phrases Ms. Improper Adjunction (MITWPIL 1994) i n "Formal Approaches to Japanese L i n g u i s t i c s " String-vacuous V-to-COMP (GLOW Paper) Rightward Movement in Verb Final Languages Dokkyo International Symposium, Yoon, James (1994) Korean Verbal I n f l e c t i o n and Checking Theory, MITWPL 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0087263/manifest

Comment

Related Items