COPULAR SENTENCES IN RUSSIAN
Studies in Natural Language and Linguistic Theory VOLUME 70 Managing Editors Marcel den Dikken, City University of New York Liliane Haegeman, University of Lille Joan Maling, Brandeis University Editorial Board
Guglielmo Cinque, University of Venice Carol Georgopoulos, University of Utah Jane Grimshaw, Rutgers University Michael Kenstowicz, Massachusetts Institute of Technology Hilda Koopman, University of California, Los Angeles Howard Lasnik, University of Maryland Alec Marantz, Massachusetts Institute of Technology John J. McCarthy, University of Massachusetts, Amherst Ian Roberts, University of Cambridge
The titles published in this series are listed at the end of this volume.
COPULAR SENTENCES IN RUSSIAN A Theory of Intra-Clausal Relations
by
ASYA PERELTSVAIG Cornell University, USA
A C.I.P. Catalogue record for this book is available from the Library of Congress.
ISBN 978-1-4020-5792-2 (HB) ISBN 978-1-4020-5793-9 (e-book)
Published by Springer, P.O. Box 17, 3300 AA Dordrecht, The Netherlands. www.springer.com
Printed on acid-free paper
All Rights Reserved © 2007 Springer No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.
ABSTRACT
This book investigates intra-clausal relations, namely, the relations that obtain between the elements in a clause; in particular, three types of intra-clausal relations are investigated here: phrase-structural relations, thematic relations, and case relations. This investigation is based on a detailed study of copular sentences in Russian. With respect to phrase-structural relations, it is argued that not all syntactic structures are asymmetrical. Rather, it is proposed that under certain conditions – namely, when the two input elements have the exact same features – Merge will result in a symmetrical structure. This requirement for identical features leads to a more parsimonious analysis of certain copular sentences where the interpretation derives directly from the syntactic structure, without postulating a special “identity copula”. As for thematic relations, it is claimed that there is no one-to-one correspondence between thematic positions and structural positions (contra the strong version of UTAH, Baker 1988). Instead, a more flexible theory of thematic relations is proposed. It is also proposed that theta-assignment is not a necessary condition for the DP interpretation. Rather, a DP can be interpreted if it establishes a coreference relationship with another theta-marked DP. This analysis extends to Left Dislocation and Pronoun Doubling constructions. Finally, case relations are said to be loosely tied to thematic relations; it is maintained that non-argument DPs – namely, those that are merged as neither complements nor specifiers of a lexical head – need not be case-marked in syntax at all and appear with the morphological default marking (which is nominative in Russian). Several conceptual and empirical problems for the alternative “agreement in case” analysis of Russian copular sentences with the nom-nom pattern are identified and discussed and this alternative is rejected. The analysis developed in this book accounts for a number of peculiar properties of Russian copular sentences, including their interpretation, case marking patterns, and such syntactic properties as extraction, inversion, binding possibilities and unaccusativity diagnostics.
v
TABLE OF CONTENTS
List of Tables
ix
Abbreviations Used in the Examples
xi
1. Copular Sentences in Russian: an overview
1
1.1. 1.2. 1.3. 1.4. 1.5. 1.6.
Introduction ......................................................................................... The Scope of the Study ...................................................................... Approaches to Copular Sentences ...................................................... Theoretical Framework ....................................................................... The Proposal ....................................................................................... Organization of the Book ...................................................................
2. Nouns and Their Extended Projections 2.1. 2.2. 2.3. 2.4. 2.5. 2.6. 2.7.
Lexical vs. Functional Categories ...................................................... The DP Hypothesis ............................................................................. The DP Hypothesis in Russian........................................................... The Locus of Referentiality................................................................ Post-Copular DPs and NPs in Russian............................................... Cross-Linguistic Evidence .................................................................. Summary .............................................................................................
3. Phrase Structural Relations And Merge: Symmetry Or Antisymmetry? 3.1. 3.2. 3.3. 3.4. 3.5. 3.6.
The Antisymmetry Hypothesis ........................................................... Challenging the Antisymmetry Hypothesis........................................ Merging Symmetrical Structures ........................................................ Symmetrical Structures ....................................................................... Agreement in Symmetrical Structures................................................ Summary ............................................................................................. vii
1 3 6 8 10 14 17 17 19 21 23 25 31 36
39 40 42 47 50 54 57
viii
CONTENTS
4. Thematic Relations
59
4.1. Theta Theory: an Overview................................................................ 60 4.2. A Richer Thematic Theory: Higginbotham (1985) and Modifications ...................................................................................... 62 4.3. Interpretation of Copular Sentences with the nom-nom Pattern....... 66 4.4. Interpretation of Copular Sentences with the nom-instr Pattern ..... 74 4.4.1. Adjectives and nouns as deficient -markers........................ 75 4.4.2. Theta-role deconstruction and non-local theta-marking ....... 85 4.5. Meaning Differences: a Review ......................................................... 94 4.6. Summary ............................................................................................. 101 5. Case Relations 5.1. 5.2. 5.3. 5.4.
Case Theory ........................................................................................ The Status of Nominative Case .......................................................... Case Marking in Copular Sentences with the nom-nom Pattern ...... Case Marking in Copular Sentences with the nom-instr Pattern .... 5.4.1. Case marking on internal arguments of verbs ...................... 5.4.2. Inherent case marking in Russian.......................................... 5.5. Case-Marking with Copula-Like Verbs in Russian ........................... 5.6. Case Marking on Depictives in Russian ............................................ 5.7. Summary .............................................................................................
103 103 106 112 116 117 120 125 130 136
6. Conclusion
137
Acknowledgements
139
Notes
141
References
151
LIST OF TABLES
Table 1. Table 2. Table 3.
Differences between existential and copula byt’ “be” ..................... Synonymous stative predicates in Russian ....................................... Unaccusativity diagnostics in Russian for three types of predicates ........................................................................................... Table 4. Realization of arguments....................................................................
ix
4 77 83 84
ABBREVIATIONS USED IN THE EXAMPLES
acc caus cop dat def distr emph f fut gen imper imperf inf instr m n nom nomin pass perf pl prep pres pst sf sg smlf subj
accusative case causative morpheme copula dative case definite distributive emphatic particle feminine future tense genitive case imperative imperfective infinitive instrumental case masculine neuter nominative case nominalizing morpheme passive perfective plural prepositional case present past short form adjective singular semelfactive subjunctive
xi
CHAPTER 1
COPULAR SENTENCES IN RUSSIAN: AN OVERVIEW
1.1.
INTRODUCTION
It is dangerous to theorize without data, my friend Watson. –Arthur Conan Doyle, The Adventures of Sherlock Holmes
This book is concerned with copular sentences, namely sentences containing the copula be (Russian byt’) as the main verb, as well as some related structures. The main focus here is on copular sentences in Russian, which exhibit various “quirks” not present in English. The main contrast that is accounted for in this book involves the form of the nominal (or adjectival) phrase following the copula (usually such phrases are called “predicates” but in order to remain analysis-neutral I call them “post-copular phrases”). These post-copular phrases can appear in either nominative or instrumental case: (1)
ˇ a. Cexov byl pisatel’. Chekhov was writer.nom “Chekhov was a writer.”
ˇ b. Cexov byl pisatelem. Chekhov was writer.instr “Chekhov was a writer.”
For the ease of exposition, I refer to sentences like (1a) as the nom-nom (i.e., nominative-nominative) pattern, and to sentences like (1b) as the nom-instr (i.e., nominative-instrumental) pattern. The contrast in (1) immediately raises a number of questions: (2)
a. Why do post-copular phrases need to be case marked at all? b. Why are there two possible case markings? c. Is there a difference in syntactic structure between these two patterns? If so, does it involve the syntax of the copula, the syntax of the post-copular phrase, or both? d. Is there a difference in meaning between these two patterns? If so, what exactly is it?
These are the questions that are discussed in great detail in the course of this book; here, I merely outline the direction of the analysis. The answer to the first question appears to be morphological rather than syntactic: nominal roots in Russian are bound, that is they must be “completed” by morphemes that indicate a number of morphological features, among them case. Thus, even nouns that appear outside of syntactic context necessarily appear in a case form, specifically as nominatives. 1
2
CHAPTER 1
Given this, one might expect that post-copular phrases in Russian would appear in the nominative. So what gives rise to the instrumental alternative? The answer to this question is syntactic: the two patterns correspond to two different syntactic structures, and the differences between these syntactic structures involve both the syntax of the copula and the syntax of the post-copular phrase. In fact, I argue in the course of this book that the differences involving the copula and those involving the post-copular phrases are interrelated in a systematic way. In particular, the nom-instr pattern is argued to differ from the nom-nom pattern in that the former but not the latter involves a richer type of a small clause core, involving a head absent in the nom-nom pattern, whereas the nom-nom pattern is argued to involve a “bare”, head-less small clause. In addition, I argue that the post-copular phrase in the nom-instr pattern is a bare NP, whereas the post-copular phrase in the nom-nom pattern is a DP. Furthermore, I show that the differences in syntactic structure correspond to differences in meaning between these two patterns: even though the two sentences in (1) are translated into English in the same way, speakers of Russian share an intuition that the two sentences do not mean the same thing (i.e., cannot be used felicitously in exactly the same situations). This intuition has been explored in many prior works; however, a clear and encompassing analysis has been missing so far. In this book, I show that the nom-nom pattern give rise to an identity interpretation imposed by the syntactic structure, whereas the nom-instr pattern is interpreted as asserting a property ascription due to the head of the small clause core. Another important point about the contrast in (1) is that the copula byt’ “be” is the only verb (in standard Russian) that allows this contrast. Non-copular verbs virtually never allow the nom-nom pattern (exceptions from colloquial Russian are discussed in Section 5.5 below):1 (3)
a.
*Mama {pocelovala / uvidela / požalela / razbudila} devoˇcka. mother.nom kissed / saw / pitied / woke-up girl.nom intended: “The mother {kissed / saw / pitied / woke up} the girl.” b. *Mama {proˇcitala / porvala / poljubila / vybrosila} èta kniga. mother.nom read / tore / liked / threw-out [this book].nom intended: “The mother {read / tore / liked / threw out} this book.”
The ungrammaticality of the nom-nom pattern applies not only to obviously lexical verbs, such as above, but also to copula-like verbs, semantically similar to the copula byt’ “be” (in Russian, svjazocˇ nye glagoly lit. “linking verbs”):2 (4)
ˇ Cexov {javljalsja / kazalsja / poˇcitalsja } veliˇcajšim pisatelem. Chekhov was / seemed / was-revered [greatest writer].instr “Chekhov {was / seemed / was revered as} the greatest writer.” ˇ b. *Cexov { javljalsja / kazalsja / poˇcitalsja } veliˇcajšij pisatel’. Chekhov was / seemed / was-revered [greatest writer].nom intended: “Chekhov {was / seemed / was revered as} the greatest writer.” a.
COPULAR SENTENCES IN RUSSIAN
3
The only two exceptions are stat’ “become” in colloquial Russian, which patterns with the true copula byt’ “be”, and nazyvat’sja “be called” with a particular word order (I return to these exceptions in Section 5.5 below). Given this limitation of the nom-nom pattern to the copula byt’ “be”, whatever analysis is proposed to explain its peculiar behavior should not overgeneralize to other verbs, including those that are semantically similar to byt’ “be” – a fault common to many previous analyses of copular sentences in Russian but remedied in this book. Finally, the analysis developed in this book is further supported by the examination of copular sentences in other languages, in particular in Italian and Norwegian. Yet, the goals of this book are not limited to explaining the empirical contrasts mentioned above. Copular sentences are used here to probe into the nature of three important intra-clausal relations: phrase-structural relations, thematic relations and case relations. Phrase-structural relations are the backbone of syntax; they define what kinds of structures the syntactic theory allows. For instance, in recent years it has been a matter of a considerable debate among syntacticians whether syntactic structure is necessarily asymmetrical (Kayne 1994) or whether some degree of symmetry is allowed (Moro 2000). Thematic relations are those between predicates and referential expressions; they include -assignment and predication. In this book, I consider the question of how many different thematic relations there are (i.e., how thematic positions are saturated), and how each one of them is restricted. Furthermore, I show that thematic relations are intimately connected to case relations. As far as the latter type of relation is concerned, I draw a distinction between syntactic case specifications and morphological case realizations, and argue that case marks whether a given noun phrase is a specifier or a complement of a lexical head. Before I embark on the exploration of these issues, a few preliminaries are in order. They will be dealt with in this first chapter. Section 1.2 outlines the scope of this study; in particular, I justify the reasons for excluding existential byt’ “be” from consideration, as well as my reasons for excluding copular sentences with certain types of post-copular phrases. In Section 1.3, I review some previous approaches to copular sentences and to the differences between predicational and equative/identity copular sentences. Section 1.4 is dedicated to a brief overview of the theoretical framework adopted in this book, namely the Minimalist framework. In Section 1.5, I present the structures proposed for copular sentences in Russian and the issues raised by these structures. Finally, in Section 1.6, I outline the course of the discussion in the rest of the book. 1.2.
THE SCOPE OF THE STUDY
As is the case in many other languages, Russian has a number of uses for byt’ “be”: existential, locative, possessive (alienable and inalienable, of which the former is illustrated below), auxiliary for future tense of imperfective verbs, predicative, and equative:3
4 (5)
CHAPTER 1
a.
b.
c.
d.
e.
f.
Do revoljucii v Moskve byli konki. Existential before revolution in Moscow were horse-trams.nom “Before the Revolution there were horse-trams in Moscow.” Den’gi byli na stole. Locative money.pl.nom were on table “The money was on the table.” U Tani byli den’gi. Possessive at Tanya were money.pl.nom “Tanya had money.” Deti budut igrat’. Auxiliary children be.fut play.inf “The children will play.” Gnomy byli sušˇcestvami rabotjašˇcimi. Predicative gnomes.nom were [creatures laborious].instr “(The) gnomes were laborious creatures.” Lenin byl Vladimir Uljanov. Equative Lenin.nom was [Vladimir Ulyanov].nom “Lenin was Vladimir Ulyanov.”
However, it turns out that not all of these instances of byt’ “be” are the same. In particular, I will assume, following Chvany (1975: 5), that “Russian sentences with byt’ fall into two syntactically distinct types”: one type involving an existential verb byt’ (which she calls ∃ and the other – the copula byt’. The differences between the two kinds of byt’ are summarized in the Table 1 below (adapted from Chvany 1975: 53). According to Chvany, the existential byt’ appears in existential, locative and possessive constructions, whereas the non-existential/copula byt’ appears in Table 1. Differences between existential and copula byt’ “be” Tests
Existential byt’ “be”
Copula byt’ “be”
How sentential negation is expressed?
net (Ego net doma. ‘He.gen is not at home’)
ne + byt’ (On ne byl pisatelem. “He.nom was not a writer.”)
Case of the subject under negation?
genitive
Nominative
Allows raising-to-object?
no
yes (Ja sˇcitaju Vanju geniem. “I consider Vanya a genius.”)
Allows raising-to-subject?
no
yes (Vanja kažetsja nerjaxoj. “Vanya seems sloppy.”)
Has lexical derivatives?
yes (e.g., bytie “being”)
no
Has syntactically similar synonyms?
yes (e.g., sušˇcestvovat’ “be, exist”)
no
COPULAR SENTENCES IN RUSSIAN
5
predicative constructions (and, presumably, in equative constructions as well, though she does not discuss them at all). Furthermore, Chvany shows that, perhaps surprisingly, the future auxiliary byt’ is a “contextual variant” of the existential byt’. A detailed discussion of the application of the tests above to different uses of byt’ goes far beyond the scope of this book; the interested reader is referred to Chvany’s work for details. In this study, I focus on the non-existential copula byt’ “be” in Russian and consider the the relation between predicative and equative uses of byt’ in Russian. In particular, I argue in the course of this book that the predicative and the equative byt’ are related through semantic bleaching.4 An alternative approach would be to postulate two distinct lexical items for the predicative and the equative byt’ “be”; most proponents of this approach assign the equative byt’ “be” a distinct meaning, encoded as part of the lexical entry. In this book, I will argue against such an approach; in particular, I will show that there is no need for a distinct equative (i.e., identity) copula. Rather, the meaning of equative sentences derives from their syntactic structure. Thus, even though I do not argue that all uses of byt’ can be reduced to one lexical item, I do argue that predicative and equative uses can be so reduced. Another restriction on the scope of this study will concern the possible post-copular phrases. Any non-verbal category can appear in the post-copular position, including noun phrases, as in (1), phrases headed by long- or short-form adjectives (in Russian, polnye and kratkie prilagatel’nye; see below), PPs and a lexically restricted set of adverbs (which are typically fossilized prepositional phrases). In this study, I do not consider post-copular PPs and fossilized adverbs, such as the following: (6)
a.
Ol’ga byla v xorošem nastroenii. Olga was in good mood “Olga was in a good mood.” b. Ol’ga byla {navesele / nastorože / zamužem}. Olga was tipsy / on-guard / married “Olga was {tipsy / on guard / married}.”
prepositional phrase
fossilized adverbs
The reason for leaving these aside is that they do not exhibit external case marking and hence are irrelevant to the discussion of the nom-nom and nom-instr patterns, although the complement of the preposition has or has originally had some nonnominative case marking.5 For the same reason (i.e., lack of case marking) I do not considered postcopular phrases headed by short-form adjectives in any great detail. Many Russian adjectives exhibit an alternation between two morphological forms: the so-called short- and long-form adjectives (cf. Timberlake 2004: 288–292). Although a detailed consideration of this alternation goes beyond the scope of this book, a few brief remarks are in order. In terms of inflectional morphology, both types of adjective forms decline for gender and number. However, only long-form adjectives have case paradigms as well. Short-form adjectives, on the other hand, exhibit no case alternation; historically, they correspond to nominative forms in full-fledged Old Russian paradigms and some idiomatic expressions in modern Russian still preserve
6
CHAPTER 1
short-form adjectives in modifier positions with their original case forms (examples from Kozyreva and Khmelevskaja 1972: 19), but this is no longer productive: (7)
a.
na bosu nogu on bare.ACC foot.acc “barefoot” b. ot mala do velika from small.GEN to big.gen “everyone; people of all ages” c. sred’ bela dnja among white.GEN day.gen “in full daylight”
Another distinguishing property of short-form adjectives is that (in modern Russian) they cannot appear in prenominal modifier positions (except in a few fossilized idiomatic expressions, such as the ones above). This distribution is illustrated below: (8)
a.
Dom byl novyj. long-form / post-copular house.nom was new.nom “The house was new.” b. Novyj dom stojal na gore. long-form / modifier new.nom house.nom stood on hill “The new house stood on {a/the} hill.” c. Dom byl nov. short-form / post-copular house.nom was new.sf “The house was new.” d. *Nov dom stojal na gore. short-form / modifier new.sf house.nom stood on hill intended: “The new house stood on {a/the} hill.”
Some researchers relate the two deficiencies of short-form adjectives and explain their inability to serve as modifiers by their lack of case. Whether this is a correct explanation or not, what is important here is that short-form adjectives do not exhibit the nominative-instrumental alternation in the post-copular position; therefore, they are not directly relevant for the present investigation.6 Thus, in this book I focus on the distribution and structure of copular sentences with post-copular phrases headed by nouns and long-form adjectives. 1.3.
APPROACHES TO COPULAR SENTENCES
Where others say they are finished, we begin. – Yiddish saying
Copular sentences have been in the center of research among linguists, philosophers and logicians at least since Plato and Aristotle. Although a detailed review of this
COPULAR SENTENCES IN RUSSIAN
7
enormous body of literature goes beyond the scope of this book, a brief review will be beneficial in highlighting the contribution of this study.7 As mentioned in the previous section, I leave aside the uses of the copula byt’ “be” that Chvany (1975) unifies under the label “existential” (including not only existential uses as such, but also locative, possessive and future tense auxiliary uses). This still leaves us with two senses or uses of the copula: the predicative and the equative (or identity) byt’ “be”. These are illustrated again below: (9)
a.
Gnomy byli sušˇcestvami rabotjašˇcimi. gnomes.nom were [creatures laborious].instr “(The) gnomes were laborious creatures.” b. Lenin byl Vladimir Uljanov. Lenin.nom was [Vladimir Ulyanov].nom “Lenin was Vladimir Ulyanov.”
Predicative
Equative/Identity
As Bertrand Russell put it: The proposition Socrates is a man is no doubt “equivalent” to Socrates is human, but it is not the very same proposition. The is of Socrates is human expresses the relation of subject and predicate; the is of Socrates is a man expresses identity. It is a disgrace to the human race that it has chosen the same word is for those two such entirely different ideas as predication and identity – a disgrace which a symbolic logical language of course remedies. [Russell 1919: 172, cited in Moro 1997: 254] So where does the difference between predication and equation come from? Usually, one of the following two approaches is taken: either this difference derives from there being two (typically homophonous) copulas or it comes from post-copular phrases. The former approach has been widely accepted, both among traditional grammarians (e.g., Benveniste 1966, Halliday 1967, Quirk and Greenbaum 1973, Kahn 1973, Salvi 1988a, among others) and in more modern schools of grammar. For instance, Halliday (1967: 66, cited in Moro 1997: 298) makes a distinction between three lexical verbs be belonging to three different classes: whereas his class one copula means “exists, happens, is found or located” and is thus irrelevant for the present study, his class zero copula means “can be characterized as, has the attribute of being” (i.e., copula of predication) and his class two copula means “identifies or is identifiable as, can be equated with” (i.e., copula of equality or identity).8 Likewise, Quirk and Greenbaum (1973: 353, cited in Moro 1997: 298) make the following generalization about the uses of the copula: “… be is commonly used to introduce a characterization or attribute of the subject… but with complement noun phrases it also commonly introduces an identification of the subject”. In Montague grammar (Montague 1973, Dowty et al. 1981: 229, Partee 1976, 1999), a distinction is made between two kinds of be: the be of predication and the distinct be of identity, distinguished by the types of their arguments. The be
8
CHAPTER 1
of predication, as in John is tall, takes two arguments of types e and <e, t>. The semantics of this version of be is: Px[P(x)]. In other words, it simply applies the predicate to the subject. On the other hand, the be of identity, as in Clark Kent is Superman, takes two arguments of type e. The semantics of this version of be is: xy[x=y]. In other words, it asserts the identity of its two arguments. This approach is also taken by Stowell (1989: 255), who distinguishes between a predicative be, which is a raising verb, and an identity be, which is “a two-place predicate conveying a relation of identity holding between two referential NPs”. The alternative approach is to treat the copula itself as the same across different uses and place the difference with the post-copular phrases. Under this approach, the copula takes two arguments of types e and <e, t>– and the phrases around the copula are type-shifted to accommodate its selection; this approach has been adopted by E. Williams (1983) and Partee (1986, 1987). There are two type-shifting functions used for this purpose: BE and ident. (10)
BE combines with generalized quantifiers (<<e, t>t>) and yields predicative expressions (<e, t>); ident combines with expressions of the type e and yields predicative expressions of the type (<e, t>)
In this book, I propose an analysis that combines parts of both of these approaches: the difference between the two types of copular sentences (which in the case of Russian are identifiable through case marking on the post-copular phrase) lies in the syntactic differences involving both the copula itself and in the post-copular phrase. In fact, the copula and the post-copular phrase are said to interact in a systematic manner giving rise to the syntactic and semantic differences between the two types of copular sentences. The proposal is detailed in the Section 1.5. But before I proceed with the proposal, a brief outline of the theoretical framework adopted in this book is necessary. 1.4.
THEORETICAL FRAMEWORK
This study is set in the Chomskian framework of modular generative grammar, in particular in the Minimalist framework (Chomsky 1995 and later work). However, it will be shown that it is necessary to make further amendments to the Minimalist theory. In this section, I briefly outline the Minimalist model of the grammar, with particular reference to four modules that are especially relevant for the study of copular sentences: (i) phrase structure, (ii) thematic and subject-predicate relations, (iii) case, and (iv) movement. The main goal of the Minimalist Program is to reduce the grammar to “virtual conceptual necessity” (Chomsky 1995: 169), or as Epstein and Hornstein (1999: xi) put it: “all things being equal … more is worse; fewer is better”. As the first step in this direction, Chomsky (1995: 169) reduces the number of levels of representation to only two “(conceptually necessary) interface levels”: A-P (articulatory-perceptual) and C-I (conceptual-intentional) interfaces (see also Chomsky 2000: 113). Unlike the
COPULAR SENTENCES IN RUSSIAN
9
previous notion of S-structure, spell-out is not a level at which principles of grammar can apply but rather an operation that strips the representation of the features that are relevant only to PF, leaving the residue , which is mapped onto LF by the same kinds of operations that are used to form .9 Moreover, the level of D-structure has been eliminated as well in favor of the model where the two basic structure-building operations Merge and Move apply in successive turns, so that there is not necessarily a stage in the derivation at which the structure contains all the elements in their basegenerated (merged) positions. Thus, unlike the previous Government and Binding model, Minimalism is radically derivational. There are no output conditions other than those required by the interfaces themselves, and structure is built derivationally by alternating applications of the two basic operations – Merge and Move. As a further simplification of the grammar, Minimalism eliminates standard X-Bar Theory: “minimal and maximal projections are not identified by any special marking, so they must be determined from the structure in which they appear” (Chomsky 1995: 242). Intermediate projections do not exist at all because they are not required by bare output conditions.10 Structure is created by multiple applications of Merge, the “operation that forms larger units out of those already constructed” (Chomsky 1995: 243). According to Chomsky (1995: 246), “the operation Merge(, is asymmetric, projecting either or , the head of the object that projects becoming the label of the complex formed.” However, this latter point will be challenged in this book, in particular in Chapter 3. Theta Theory, which has played an important role in the Government and Binding model of the grammar, loses much of its importance in the Minimalist model. Thus, in Minimalism thematic relations are merely expressions of certain configurations: “a -role is assigned in a certain structural configuration; assigns that -role only in the sense that it is the head of that configuration” (Chomsky 1995: 313, see also Chomsky 2000: 103, 127). The effect of the Function Saturation Principle of Heycock (1994), which made the projection of subjects obligatory, is produced by the new formulation of the EPP, which is “divorced from Case” (Chomsky 1995: 282) and further extended from T to C and v (Chomsky 2000: 102). In contrast to thematic relations, which take place in lexical domains, case and agreement are morphological features, which are checked in functional domains. In contrast to Theta Theory, movement receives considerable attention in Minimalist research. In the Minimalist Program, movement is driven by morphological feature checking – some feature must be checked in order for movement to take place. Chomsky (1995: 280) calls this restriction on Move the Principle of Last Resort, and derives it from the Principle of Full Interpretation, which requires that interface representations contain no elements inaccessible to that interface. This means that features that cannot be interpreted by either of the interface levels, the so-called [–Interpretable] features, must be checked (and erased) prior to the interfaces. Features are checked by entering into the checking domain of a head that contains a corresponding feature. The checking domain is created by adjunction to an XP, or a head, or substitution. Case in the Minimalist Program is considered as one such [–Interpretable] feature which must be checked prior to the interfaces (see Chomsky
10
CHAPTER 1
2000: 102, 119). Therefore, it is nothing more than motivation for movement. It should also be noted here that in much of the Minimalist research nominative case is treated on a par with accusative, both being called structural cases. In this book, I will adopt a different view of case in general and of the nominative case in particular. These issues are discussed further in Chapter 5 of this book. In addition to the principle of Last Resort, Move is restricted by Economy conditions, in particular the Minimal Link Condition. Both the Last Resort and the Minimal Link Condition are incorporated into the definition of Attract F, cited below from Chomsky (1995: 297). (11)
K attracts F if F is the closest feature that can enter into a checking relation with a sublabel of K.
As seen from this definition, the notion of “closeness” plays an important role in Economy conditions. The definitions of “closeness” and “equidistance” from Chomsky (1995: 356) and the definition of minimal domain from Chomsky (1995: 299) are given in (12). (12)
Assume that is a feature or an X category, and CH is the chain (, t or (the trivial chain) . Furthermore, c-commands and is the target of raising. a. Equidistance: and are equidistant from if and are in the same minimal domain. b. Closeness: is closer to K than unless is in the same minimal domain as (a) or (b) , c. Minimal Domain: The minimal domain Min( (CH)) of CH is the smallest subset K of (CH) such that for any ∈ (CH), some ∈ K reflexively dominates . The domain (CH) of CH is the set of categories included in Max( that are distinct from and do not contain or . Max( is the smallest maximal projection including .
These notions will play an important role in the argumentation in throughout this book. However, I will also show that certain elements of the Minimalist theory need to be modified in order to account for Russian copular sentences. 1.5.
THE PROPOSAL
If morphological evidence does not fit in with our assumptions about the underlying structure, perhaps we should question those assumptions rather than immediately blame it all on the morphology. –Wierzbicka, The Case for Surface Case
In this book, I propose that the differences between the two types of copular sentences (which in Russian are identifiable through case marking on the
COPULAR SENTENCES IN RUSSIAN
11
post-copular phrase) are syntactic in nature. The first indication that this is indeed a syntactic phenomenon comes from the following contrast involving extraction, noted by Bailyn and Rubin (1991): while an instrumental-marked post-copular wh-phrase can be extracted, a nominative-marked wh-phrase cannot.11 (13)
a.
Kemi ty znaeš, cˇ to Saša byl ti ? who.instr you know that Sasha was “Who do you know that Sasha was?” ty znaeš, cˇ to Saša byl ti ? b. *Ktoi who.nom you know that Sasha was intended: “Who do you know that Sasha was?”
As shown throughout the book, this contrast is not the only piece of evidence for syntactic differences between the two types of copular sentences: they also differ in terms of inversion possibilities and these structural differences are responsible for both the differential case marking and the differences in meaning between them. So what are the syntactic structures associated with the two types of copular sentences? A heated debate has been raging in the literature as to the structure of small clauses and consequently of copular sentences (for a nice summary, see Cardinaletti and Guasti 1995); three types of approaches to the structure of small clauses have been proposed: a lexical projection approach, a functional projection approach and a “anomalous small clause” approach.12 In this book, I will adopt variants of the latter two of these approaches, for the two types of copular sentences. Before I outline my proposal in detail, I will briefly review the three approaches. • The lexical projection approach: a small clause is taken to be a lexical projection of the head of the predicate, which can be a noun, an adjective or a preposition. The subject of a small clause can be either in the specifier of the lexical projection of the predicative head (as in Stowell 1981, 1983) or adjoined to that projection (as in Manzini 1983, Heycock 1994, Rothstein 1995), if the distinction between modifier and adjuncts is drawn at all. These proposals are summarized in (14), where XP is the predicate category (either an NP, an AP or a PP): (14)
XP DP
X'/XP …X°…
• The functional projection approach: a small clause is analyzed as a projection of a special functional head which takes the lexical projection of the predicate
12
CHAPTER 1
(i.e., the post-copular phrase) as a complement. The exact label and nature of this functional head are being debated in the literature: for example, Sportiche (1995) and Guéron and Hoekstra (1995) label this functional head AGR , which allows them to capture the generalization that adjectival heads of small clauses exhibit agreement with their subjects in languages like French and Italian, even though small clauses are semantically tenseless; Contreras (1995) labels this functional head Asp , and Starke (1995) considers the same functional head to be a null V . According to an influential proposal by Bowers (1993), the functional head of small clauses has nothing to do with agreement or aspect and its sole purposes is encoding predication; he labeled this projection PrP and others subsequently used a more easily pronounceable (and possibly less confusing) label PredP (cf. Svenonius 1994); for ease of reference, I use the label PredP here.13 (15)
PredP DP
Pred' Pred°
XP
• The anomalous small clause approach: a small clause is taken to be an anomalous type of constituent in that it is not a projection of one head.14 In other words, a small clause under this approach is exocentric (cf. Moro 1997, 2000, who take this approach to small clauses in copular sentences).
(16)
SC = small clause DP
XP
In this book, I adopt a combination approach whereby small clauses with the nom-instr pattern include a small clause of the type similar to the one schematized in (15), while sentences with the nom-nom pattern include a small clause similar to (16); furthermore, I propose that the differences in structure are also associated with different types of post-copular phrases, namely, the endocentric small clause in (15) arises when the post-copular phrase is of a predicative type, such as an NP or an AP, whereas the exocentric small clause in (16) arises when the post-copular phrase is of a saturated expression, such as a DP.15 The two structures that I propose are schematized below (recall that in Russian, the two structures are most easily distinguishable by the overt case marking on the post-copular phrase: nominative in (17a) and instrumental in (17b)):16
13
COPULAR SENTENCES IN RUSSIAN
(17)
a. NOM-NOM pattern:
b. NOM-INSTR pattern
TP
TP T'
DPi
DPi FP
T° ti
T' FP
T° ti
F'
byt’ DP
νP
F°
DP
F°
F'
DP
ti OR: ti
byt’
ν'
ti
ν°
NP/AP
tbyt’
Before I proceed expounding on these structures, a quick note on verb raising is due. Although some researchers have argued that Russian is a verb-raising language similar to French (cf. King 1995: 96–103), the majority of Russian syntacticians (cf. Bailyn 1995, Sekerina 1997, inter alia) now agree that verbs in Russian raise not to T but only as high as an intermediate functional category, which for ease of reference I label throughout this book as F (the exact nature of this projection is beyond the scope of this book, but unlike much of the literature of Russian sentential syntax, e.g., Bailyn 1995: 50–52, I do not use the label Pred˚ for this head in order to avoid potential confusion with the category hosting the copula in sentences with the nom-instr pattern). In particular, the standard Pollockian adverb placement tests put Russian in the same category with English rather than French: adverbs typically precede rather than follow verbs in Russian, unlike in English. This is true both of lexical main verbs and the copula: (18)
a.
Ja dumaju, cˇ to Ivan cˇ asto celuet Mašu. I think that Ivan.nom often kisses Mary.acc “I know that Ivan often kisses Mary.” b. *Ja dumaju, cˇ to Ivan celuet cˇ asto Mašu. I think that Ivan.nom kisses often Mary.acc “I know that Ivan often kisses Mary.” [Bailyn 1995: 58, his (75)]
(19)
a.
Ja dumaju, cˇ to Ivan vsegda byl {durak / durakom}. I think that Ivan.nom always was fool.nom /fool.instr “I know that Ivan always was a fool.” b. *Ja dumaju, cˇ to Ivan byl vsegda {durak / durakom}. I think that Ivan.nom was always fool.nom / fool.instr “I know that Ivan always was a fool.”
14
CHAPTER 1
Thus, I assume that the copula (as well as lexical verbs) appears at spell-out in F ; furthermore, I assume that the subject appears at spell-out in Spec-TP, where it moves to satisfy the EPP (this is discussed in more detail in Section 2.5 below; cf. Lavine 2000, Lavine and Freidin 2002, Bailyn 2004b, inter alia). This is where the similarities between the two types of sentences end and the differences begin. First, even though the copula appears in the same position at spellout, it is not generated in the same position in the two types of copular sentences: in sentences with the nom-nom pattern, the copula is a functional category and as such it is merged in F , whereas in sentences with the nom-instr pattern, the copula is a lexical verbal element, merged in v . The reason I use the label v rather than Pred is to highlight its lexical rather than functional nature (the distinction between lexical and functional categories is discussed in Section 2.1).17 The reason I use v rather than V is because the copula is a light verb, as discussed in Chapter 4 below. Another major difference concerns the syntax of the post-copular phrase: in sentences with the nom-nom pattern, the post-copular phrase is a DP which merges together with the subject DP in a symmetrical structure, whereas in sentences with the nom-instr pattern, the post-copular phrase is a bare NP or AP which is selected by v as a complement. Several important issues arise in connection with these structures. The first issue is whether symmetrical structures can be generated in syntax; it is considered in detail in Chapter 3. The second issue is how the post-copular phrases in those two structures are interpreted semantically/thematically; this issue is the topic of Chapter 4. In that chapter, it is also shown how the two major differences between the two types of copular sentences are related to each other, namely how the difference in category of the post-copular phrase necessitates the difference in the lexical/functional status of the copula. The third issue is how the nominative and instrumental case marking arises from these structures; this is discussed in Chapter 5.
1.6.
ORGANIZATION OF THE BOOK
This book is organized as follows: the next chapter is concerned with the structure of noun phrases. In this chapter I define the distinction between lexical and functional categories and argue that referentiality in nominals is to be associated with the DP projection rather than with the lexical head N . Furthermore, I argue that the DP Hypothesis is to be extended to languages without overt articles, such as Russian. Specifically, I show that nominative and instrumental post-copular phrases are of different syntactic category: the former are DPs, whereas the latter are bare NPs (which pattern with APs rather than DPs). I further support this conclusion by cross-linguistic evidence that shows that in languages which distinguish DPs and bare NPs overtly, the latter pattern with APs and not with DPs in the post-copular position.
COPULAR SENTENCES IN RUSSIAN
15
In Chapter 3 I address the theoretical issues involved in merging symmetrical structures for one of the two types of copular sentences. I argue that the operation Merge conceived in the true Minimalist spirit can generate symmetrical structures, like the one proposed in this book for Russian copular sentences with the nom-nom pattern. The goal of the later chapters (Chapters 4 and 5) is to show that the structures proposed for Russian copular sentences are not only plausible, but also permit the capture of the range of properties that such sentences have. In particular, Chapter 4, which is concerned with thematic relations, shows that the structures I propose for copular sentences allow us to account for the interpretations associated with these sentences. In Chapter 5, I develop the Case Theory and provide an analysis of case marking in copular sentences in accordance with the structures I propose for these sentences. Chapter 6 summarizes the discussion throughout the book.
This page intentionally blank
CHAPTER 2
NOUNS AND THEIR EXTENDED PROJECTIONS
As mentioned in the previous chapter, the two types of copular sentences in Russian – those with the nom-nom pattern and those with the nom-instr pattern – differ in the category of the post-copular phrase (among other things): nominative post-copular phrases are DPs, while instrumental post-copular phrases are NPs. Note that this contrast is not self-evident at least for two reasons: first, it is often assumed that arguments are DPs, while predicates are bare NPs (cf. e.g., Longobardi 1994, inter alia); second, Russian lacks uncontroversial D-elements, that is articles, leading some to believe that Russian lacks the functional “superstructure” in nominals altogether. In this chapter, I consider these claims and argue that they are invalid. In particular, I maintain that Russian has the DP projection (although not in all nominals), as well as the intermediate functional projections, such as QP. Furthermore, I uphold the view that while argumental noun phrases need not be projected fully as DPs, post-copular phrases need not be smaller than DPs and in fact the post-copular phrases in sentences with the nom-nom pattern are full-fledged DPs. Before I proceed, a review of the concept of lexical and functional categories, which will be important throughout the book for the discussion of the NP/DP distinction, as well as the analysis of byt’ “be” as either a functional or a lexical head, is undertaken in the next section. 2.1.
LEXICAL VS. FUNCTIONAL CATEGORIES
The distinction between lexical and functional categories is a very fundamental one in generative syntax. For example, Li’s (1990) generalization governing proper head movement is formulated in terms lexical vs. functional categories. (20) The Proper Head Movement Generalization (cited in Baker 2003: 53) A lexical head A cannot move to a functional head B and then back to a lexical head C. However, the nature of the distinction is not very clear. As Cardinaletti and Giusti (2000: 1) note, “it is sometimes not easy to decide whether an element belongs to either of the two classes [i.e., functional or lexical categories]”. So what are the criteria that make one decide whether a given element belongs to the class of functional or lexical categories? Based on Abney’s (1987) seminal work, Kerstens (1993) lists the following four criteria that distinguish functional categories from lexical ones:18 17
18 (21)
CHAPTER 2
a. Functional elements constitute closed lexical classes. b. Functional elements permit only one complement (which is a nonargument). c. Functional elements are usually inseparable from their complements. d. Functional elements “lack descriptive content”.
Yet, these criteria are not unproblematic and sometimes contradict one another with respect to a given item or set of items. Consider the “closed class” criterion first. The problem with this criterion is that “closeness” or “openness” of a given class is sometimes difficult to define. For instance, Baker (2003: 248) claims that Chichewa has only eight true adjectives but still takes Chichewa adjectives to be a lexical category. Similarly, Igbo is said to have a closed class of eight adjectives” (Bhat 1994: 41). On the other hand, languages routinely resist forming new prepositions, so P may be concluded to be a functional category (e.g., this is the conclusion drawn by Baker 2003: 303–325); however, there is a sense in which Ps (for instance, in Russian) can appear as either lexical or functional categories (cf. Franks and Yadroff 2002). The second criterion is also somewhat problematic. Since the introduction of the idea that theta-marking is strictly local and the analysis of ditransitive verbs in terms of verbal shells (see Larson 1988, and later work), lexical categories are believed to have also only one complement. Therefore, the number of complements a given head can have cannot be taken as a defining criterion for distinguishing lexical from functional categories. Below, I will come back to the claim that functional categories take non-argument complements. The third criterion listed by Kerstens has to do with the degree of separability of functional heads from their complements. The problematic nature of this criterion has been noticed by Abney (1987: 285, fn. 88) himself; he noted that determiners can appear without a lexical projection as a complement, as is indeed the case for pronouns, which he suggests should be analyzed as determiners (although there is evidence that pronouns in Russian are merged as Ns and move to D; cf. Franks and Pereltsvaig 2004). Finally, consider the claim that functional elements “lack descriptive content”. Unfortunately, the notion of “descriptive content” is vague. According to Abney (1987: 285), determiners must be analyzed as functional elements because they are not predicational. However, he notes that determiners constitute predicates over predicates (see also Barwise and Cooper 1981), thus undermining the claim that determiners are not predicational and also the claim that they, therefore, lack descriptive content. On the other hand, the presence of descriptive content does not always tip the scale in the direction of lexical categories. Thus, Baker (2003: 303–325) argues that prepositions are functional elements, even though a preposition like after can hardly be said to have less descriptive content than the verb follow (as mentioned in Section 1.2 above, I do not discuss prepositions and prepositional phrases in a great detail in this book). Instead of listing a number of properties that would help us decide whether a given category is lexical or functional, I propose to take the presence of a thematic grid as
NOUNS AND THEIR EXTENDED PROJECTIONS
19
the defining property of lexical categories. In other words, only lexical categories can discharge thematic positions because only lexical categories have such thematic positions to start with. The only way that a functional category can have a thematic grid is if it inherits it from a lexical category through percolation. The idea that functional categories do not have thematic grids has been long entertained in the literature (e.g., Pollock 1989), in particular with respect to auxiliary verbs (e.g., see Cinque 2004 on “restructuring” verbs, i.e., “functional” verbs as lacking thematic roles). Note also that this proposal ties nicely with the idea in Baker (2003) that (at least some) lexical categories can be distinguished by the theta-grids they bear (or the ways in which they can discharge thematic positions). According to Baker, verbs can theta-mark their specifiers, whereas adjectives cannot (this claim is discussed in great detail in Section 4.4.1 below). Note further that this definition of lexical category forces us to assume that nouns have theta-grids; otherwise, they can no longer be considered a lexical category. However, this goes against Baker’s specific claim that nouns cannot have a theta-grid (and have a referential index instead). I return to this point in much detail in Section 2.4 below. 2.2.
THE DP HYPOTHESIS
In this book, I adopt a version of the DP Hypothesis of Abney (1987), who proposed that a determiner is not a specifier of the nominal projection (as in Jackendoff 1977: 104–105), but rather the head of its own projection. Under this analysis, the D takes an NP as its complement: (22)
a.
b.
DP
Spec
D'
DP DP
D°
NP
the that every
king
Kim
D' D° 's
NP friend
According to Abney, the D node hosts articles (e.g., the, a), demonstratives (e.g., that, those), quantifiers (e.g., many, some, every).19 Moreover, in some languages the N head can move overtly into the D node; for example, it can move overtly in Hebrew, Romanian, Swedish, Norwegian, and in a more restricted way in Italian (see Dobrovie-Sorin 1987, Delsing 1988, Ritter 1988, 1991, Taraldsen 1990, Longobardi 1994, Chierchia 1998, among others).20 Abney’s original proposal has been later developed to include a number of other functional projections between DP and NP. For example, Ritter (1991) argues for the existence of a functional projection NumP (for Number Phrase); Zamparelli (2000) argues for a layered structure of DP, where the DP is split into three projections: SDP (Strong Determiner Phrase), PDP (Predicative Determiner Phrase), and KIP (Kind Determiner Phrase). Following much work on the topic, I assume that there is at
20
CHAPTER 2
least one intermediate functional projection in nominals; I use the label QP, typically used for this projection in Slavic literature. Furthermore, following Zamparelli (2000), I assume that strong and weak determiners are distinguished structurally. In addition, I depart somewhat from Abney’s (1987) analysis with respect to the structural position of attributive adjectives. Abney proposed that adjectives are heads taking NP complements (cf. also Kester 1993). Many alternative proposals have been put forward in the literature. For example, Delsing (1992, 1993) assumed, following Abney, that attributive adjectives are heads, but according to Delsing, the NP is in the right-hand specifier of the adjective rather than its complement (at least in the case of pre-nominal adjectives). Treating attributive adjectives as heads is also part of the analysis of Travis (1988), Sigurðsson (1993), and Sadler and Arnold (1994), but they argue the A head-adjoins to the N . Alternatively, attributive adjectives have been treated as phrasal categories, either as specifiers or as adjuncts. The specifier approach has been adopted by Holmberg (1993), who assumed that attributive adjectives are specifiers of N or n (a category analogous to Chomskian v ). A different take on the adjectives-as-specifiers analysis underlies Cinque’s view of attributive adjectives, which he argues to be in specifiers of functional projections (see Cinque 1994, cf. Cinque’s 1999 analysis of adverbs). Alternatively, attributive adjectives have been treated as adjuncts, either to N’, as in Santelmann (1993), or to NP, as in Svenonius (1993a, b), Duffield (1999), among others. Reviewing the arguments for and against each one of these views would take me too far afield, but for reasons discussed in detail in Pereltsvaig (2006b), I assume in this book that modifying adjectives can be either heads or specifiers of a special functional projection between DP and NP (in fact, between QP and NP). In this book, the focus is on nominal phrases with strong determiners (such as definite articles, pronouns and proper names), leaving nominals with weak determiners and those that are ambiguous between the two readings for future research; furthermore, I ignore structural and interpretative differences between definite descriptions, proper names and pronouns (for some discussion of these issues the reader is referred to Longobardi 1994 and Zamparelli 2000). Another important issue that arises in connection with Abney’s DP Hypothesis is whether the “size” of a nominal (i.e., whether it is a DP, or a bare NP, or something in between) correlates with is position as an argument or a predicate (i.e., a post-copular phrase, in the terminology adopted in this book). It has often been claimed that argument nominals are DPs (at least in languages with articles; see Section 2.3), whereas predicative nominals are bare NPs; cf. Longobardi (1994) and much subsequent work. But not everybody subscribes to this view; for instance, Julien (2005) in her work on Scandinavian nominals shows that in languages with articles predicative noun phrases need not be bare NPs and can be DPs. In her book, Julien dedicates a separate chapter predicative nominal phrases, where she develops three tests that can be used to tell predicative noun phrases from argumental ones; these tests include:
NOUNS AND THEIR EXTENDED PROJECTIONS
21
• the possibility of being a predicate in a small clause selected by consider-type verbs; • topicalization possibilities (discussed in Pereltsvaig 2001c and in Section 2.5 below); • ability to appear in pseudo-clefts. She then applies these tests to nominal phrases of various types, and concludes that nominals that pass the predicate tests (including definite nominals, nominals with possessors, nominals with possessors followed by numerals, and nominals with strong quantifiers and numerals) are not necessarily structurally smaller than argument nominal phrases. In other words, a nominal predicate (in Scandinavian) can be a full DP, and it can even contain a strong quantifier. Her overall conclusion is that “nominal predicates are not characterized by their syntactic category, but by their semantics: they have an intensional interpretation, whereas nominal arguments can have an intensional or an extensional interpretation” (p. 252). In the present book, I follow Julien in allowing the DP/NP distinction in the post-copular position, although the details of our analyses differ: whereas she places the burden of differentiating post-copular DPs and post-copular NPs on the Pred , I propose that post-copular DPs appear in a completely different, symmetrical structure. Before I proceed to review evidence for the DP/NP distinction in the postcopular position in Russian, in the next section I address the more general question of whether the DP projection is found at all in Russian (and other languages without articles). 2.3.
THE DP HYPOTHESIS IN RUSSIAN
Since Abney’s (1987) postulation of the DP Hypothesis, an important question arose as to whether noun phrases are uniformly DPs across languages (cf. Bowers 1991, Longobardi 1994), or whether some languages lack the functional projection(s) in the noun phrases (cf. Chierchia 1998, Baker 2003: 113). Those who adopt the latter view typically take the lack of DP to correlate with the lack of overt articles, which makes Russian a prime candidate for language-without-DP (the same is true for other article-less Slavic languages, including Ukrainian, Belarusian, Polish, Czech, Slovak, Sorbian, Serbian, Croatian, Bosnian and Slovenian).21 This led to a very intense debate in the Slavic syntax literature about the structure of noun phrases in these languages. On the one hand, a number of researchers (cf. Willim 1998, 2000, Trenkic 2004, Boškovi´c 2005) argued that article-less Slavic languages do not have the DP projection and that all nominals in these languages are NPs.22 This contention led to a number of controversial claims about the internal structure of Russian and similar languages (for a detailed discussion see Pereltsvaig 2006b); more importantly for the present discussion, under this view the contrast between copular sentences with the nom-nom pattern and those with the nom-instr pattern cannot be attributed to the contrast in the category of the post-copular phrase, as is the claim of this book.
22
CHAPTER 2
On the other hand, arguments have been brought forth in support of the alternative view, which I adopt in this book, namely that even article-less languages such as Russian have the DP projection. According to this view, at least some nominals in Russian are DPs, although not all nominals are DPs (cf. Pereltsvaig 2006a). Although a detailed examination of the arguments in favor on this view goes far beyond the scope of this book, in what follows I briefly review some of the core arguments. The issue of whether article-less Slavic languages have the DP projection was first addressed by Progovac (1998), who claimed that despite appearances to the contrary Serbo-Croatian has the DP projection. Her strongest evidence comes from noun/pronoun asymmetries, where the pronouns precede, and nouns follow, certain (intensifying) adjectives: (23)
a.
*Ni Marija sama u to ne veruje. neither Mary alone in that not believes b. Ni sama Marija u to ne veruje. neither alone Mary in that not believes “Not even Mary believes that.” c. Ni ja sama u to ne verujem. neither I alone in that not believe “Not even I believe that.” d. *?Ni sama ja u to ne verujem. neither alone I in that not believe
[Serbo-Croatian]
[Progovac 1998: 168]
Assuming that these adjectives occupy a fixed syntactic position, the conclusion must be that pronouns occupy a structurally higher position than nouns. Similar facts from Polish are considered in Rutkowski (2002) and the same conclusion is drawn there. Russian facts too are analogous. Rappaport (1998), Engelhardt and Trugman (1998), and Trugman (2005) argue for the existence of DP in Slavic languages (specifically in Russian) on the basis of their examination of noun phrases with two genitives: (24)
a.
tablica èlementov Mendeleeva table elements.gen Mendeleev.gen “the table of elements of Mendeleev” b. skripka Stradivari našej solistki violin Stradivarius.gen our soloist.gen “our soloist’s violin by Stradivarius”
[Rappaport 1998: 8]
Although the details of their analyses differ, they equally maintain that the DP projection is necessary in order to accommodate both genitive phrases. Similar facts from Polish are considered and the same conclusion drawn by Rappaport (2001). The idea that the internal structure of noun phrases provides evidence for the DP/NP distinction in Russian is further developed in Pereltsvaig (2006b), where the syntax of modifying adjectives, numerals, demonstratives and possessives is considered in great detail.
NOUNS AND THEIR EXTENDED PROJECTIONS
23
Alternatively, an investigation of the external distribution of different types of nominals, undertaken in Franks and Pereltsvaig (2004) and in more detail in Pereltsvaig (2006a), shows that even certain contrasts between different types of argument nominals are best accounted for if a distinction between DPs and NPs (or “Small Nominals”, in Pereltsvaig’s 2006a terminology) is postulated. In what follows, I show that post-copular nominals in Russian too exhibit a contrast between DPs and NPs. 2.4.
THE LOCUS OF REFERENTIALITY
Once the existence of a layered structure for nominals is established, the next obvious question is whether referentiality, the characteristic property of nominals, is associated with the lexical layer or with one of the functional layers, such as the DP. Both positions have their proponents. For instance, Baker (2003: 95–189) develops the analysis – originally proposed by Geach (1962) and later adopted by Larson and Segal (1995) and Chierchia (1998) – which is based on the idea that referentiality is the distinguishing property of nouns, and is therefore to be associated with the lexical N-layer. In contrast, Marantz (1997) – and implicitly Borer (2004) – have argued that referentiality must be associated with the functional layer since the lexical layer (for them) is category-neutral; similar argument is made in Stowell (1989) and Longobardi (1994), except that for them the lexical layer is not category-neutral. In this section, I briefly outline these two approaches and the implications they make for copular sentences and then argue for the StowellLongobardi-Marantz-Borer position. According to Baker (2003: 101–109), only nouns can bear a referential index (that allows them to be used in argument positions) because only nouns have Criteria of Identity, which is the standard that allows us to determine whether two things are the same or not. For example, knowing the meaning of a noun dog involves knowing whether X is the same dog as Y. Two different common nouns can have significantly different Criteria of Identity. Baker (2003: 101–102) argues that adjectives and verbs do not have such Criteria of Identity because they cannot fit in the frame “X is the same __ as Y”. One consequence of this claim is that only nouns and their projections (i.e., NPs) can participate in processes that require co-reference relations, such as anaphora, binding and movement, leading to the following well-known contrasts between genitive nominals and nationality adjectives:2324 (25)
a.
I was distressed by Italyi ’s invasion of Albania. Iti (i.e., Italy) should have known better. b. ??I was distressed by the Italian invasion of Albania. Iti (i.e., Italy) should have known better.
To summarize so far, Baker (2003) claims that referentiality, which determines the characteristic nominal behavior (such as the possibility of pluralization, participation in anaphoric relations, and possibility of argumenthood), is associated with the
24
CHAPTER 2
lexical category N and distinguishes nouns from other lexical categories. Importantly, he makes no distinction between the behavior of nominals with and without the determiner layer, whether as arguments or predicative nominals. The opposite position was first argued for in Chomsky’s (1970) “Remarks on Nominalization”; in recent years, this idea was “resurrected” most notably by Marantz (1997) and Borer (2004). According to this view, the distinctive properties of nominal and verbal projections (especially, their distribution) depend not on the lexical category of the root but on the functional structure that the root is embedded in. Moreover, Marantz and Borer claim that roots themselves are “category neutral”, √ √ for example, “roots like DESTROY and GROW are … neutral between N and V” (Marantz 1997: 15). Whether an item has a nominal or a verbal distribution depends on its insertion site: “when the roots are placed in a nominal environment, the result is a ‘nominalization’; when the roots are placed in a verbal environment, they become verbs” (Marantz 1997: 15). For example, if the root is inserted under a D node, it becomes a nominal, as in (26). (26)
the destruction of the city; the city’s destruction D √DESTROY
D
√DESTROY
the city
In contrast, if the root is inserted under one of the verbal functional heads (i.e., v ), it becomes a verb, as in (27); note that for Marantz v is a functional head, whereas under the analysis proposed in this book v is a lexical head. (27)
John destroyed the city. v
√DESTROY
v
√DESTROY
the city
Furthermore, a nominalization like John’s destroying the city is analyzed as a larger structure that involves a v-projection (i.e., a transitive verb structure) embedded under a determiner, which makes it a nominal: (28)
John’s destroying the city D D
v–1 v–1
√DESTROY
√DESTROY
the city
NOUNS AND THEIR EXTENDED PROJECTIONS
25
To recap, according to Marantz (and Borer), nominal projections differ from verbal ones not by the category of the root (which in this theory is not marked for category at all), but by the functional structure that the root is embedded in. For instance, the D-layer turns a category-neutral root into a nominal, while the v-layer the root into a verbal projection. Furthermore, two kinds of nominalizations – John’s destroying the city and the destruction of the city – are distinguished by the presence vs. absence of the verbal functional layer below the nominal functional layer. According to this view, it is the functional layer (namely, the D-layer) that is associated with referentiality. Note that Longobardi (1994) has made a similar claim that referentiality is associated with the functional D-layer and the presence/absence of the D-layer correlates with the argument/predicate distinction, even though he does not assume that the lexical layer is category-neutral. The choice between the two approaches – the one that associates referentiality with the lexical layer (as in Baker 2003) and the one that associates referentiality with the functional layer (as in Stowell 1989, Longobardi 1994, Marantz 1997 and Borer 2004) – has important ramifications for the study of copular sentences and Russian nominals more generally. Crucially, if the locus of referentiality is the lexical category N (as in Baker 2003), article-less languages such as Russian are truly article-less (and depending on one’s conclusions about the nature of demonstratives and possessives in Russian, perhaps also D-less), suggesting that Russian is very different from languages with (overt) articles, such as English or Italian. In contrast, if the locus of referentiality is in the D-layer (as in Stowell 1989, Longobardi 1994, Marantz 1997 and Borer 2004), a nominal must be minimally a DP in order to be an argument and cannot be a DP if it is a true predicate. This in turns implies that article-less languages like Russian involve a system of phonetically null determiners. In other words, this theory relegates the difference between an overt Italian-style article system and a non-overt Russian-style article system to the realm of morpho-lexical differences. As mentioned in the previous section, there are good reasons to believe that Russian exhibits a contrast between DPs and NPs, that is that both types of nominals are found in different contexts. In what follows, I largely adopt the Marantz-BorerStowell-Longobardi view that referentiality is associated with the D-layer and show that post-copular phrases come in two types DPs, on the one hand, and NPs/APs, on the other hand. 2.5.
POST-COPULAR DPS AND NPS IN RUSSIAN
In this section, I show that the case marking on post-copular phrases in Russian correlates neatly with the DP/NP distinction. Furthermore, I show that post-copular phrases headed by long-form adjectives in nominative or instrumental case pattern with phrases headed by nouns, while phrases headed by short-form adjectives (which as mentioned above are historically nominative forms) pattern not with nominativemarked nominal or long-form adjectival phrases but with instrumental-marked ones. Thus, the distinction in Russian is not between nominal and adjectival phrases, as
26
CHAPTER 2
one might suspect if one adopts the null-N analysis of long-form adjectives (as in Babby 1973, Bailyn 1994); rather, the distinction is between DPs, on the one hand, and bare NPs and APs, on the other hand. The first inkling that nominative- and instrumental-marked post-copular phrases are different in syntactic category comes from the coordination test, discussed in connection with post-copular phrases in Zamparelli (2000: 132) and Julien (2005: 273–277). As is well-known, coordination requires members that are alike in category (where the category is defined not just by the head of the projection but by its function so that a predicative PP and a predicative AP can be conjoined despite their having heads of different syntactic category: John is in love and very happy). As noted in Zamparelli and Julien, in English or Norwegian a post-copular definite DP of identity and a post-copular indefinite DP of predication cannot be conjoined: (29)
*Mark Twain var Samuel Clements og ein amerikansk forfattar.[Norwegian] M. T. was S. C. and a American writer “*Mark Twain was Samuel Clements and an American writer.” [Julien 2005: 275, her (7.51c)]
Note that not only the same is true in Russian, but more generally a nominativemarked post-copular phrase cannot be conjoined with an instrumental-marked one, whereas two nominative-marked or two instrumental-marked post-copular phrases can be conjoined: (30)
a. b.
c.
d.
*Mark Tvejn byl Samuèl Klements i amerikanskim pisatelem. M. T. was S.nom C.nom and American.instr writer.instr “*Mark Twain was Samuel Clements and an American writer.” *Aleksandr Porfir’eviˇc Borodin byl professor ximii i kompozitorom. A. P. B. was professor.nom chemistry and composer.instr intended: “Alexander Porfirevich Borodin was a professor of chemistry and a composer.” Aleksandr Porfir’eviˇc Borodin byl professor ximii i kompozitor. A. P. B. was professor.nom chemistry and composer.nom “Alexander Porfirevich Borodin was a professor of chemistry and a composer.” Aleksandr Porfir’eviˇc Borodin byl professorom ximii i kompozitorom. A. P. B. was professor.instr chemistry and composer.instr “Alexander Porfirevich Borodin was a professor of chemistry and a composer.”
Further support for my claim that the relevant contrast involves the distinction between DPs and NPs comes from the distribution of D-elements in Russian copular sentences. The crucial observation is that the alternation between nominative and instrumental post-copular phrases disappears when the post-copular phrase is headed by a demonstrative or a numeral. Such phrases can appear only in the nominative case and the instrumental is impossible. For example, consider demonstratives (cf. Pereltsvaig 2006b for arguments in support of its D-status). Post-copular phrases containing a demonstrative must appear in the nominative, as shown in (31a-b), although instrumental-marked phrases in an argument position containing a demonstrative are grammatical, as shown in (31c).25 Crucially, if the demonstrative is omitted, the instrumental is
NOUNS AND THEIR EXTENDED PROJECTIONS
27
grammatical. This strongly suggests that instrumental-marked post-copular phrases do not have the DP projection. (31) a. Ivanuška -duraˇcok byl [DP tot brat, kotoryj vsegda Ivanushka.nom-fool.nom was that.nom brother.nom which.nom always popadal v bedu]. got into trouble “Ivanushka the Fool was that brother who always got into trouble.” b. Ivanuška -duraˇcok byl [DP (* tem) bratom, kotoryj Ivanushka.nom-fool.nom was (*that.instr) brother.instr which.nom vsegda popadal v bedu]. always got into trouble intended: “Ivanushka the Fool was that brother who always got into trouble.” c. Oleg okazalsja [NP tem bratom, kotoryj vsegda Oleg turned-out-to-be that.instr brother.instr which.nom always popadal v bedu]. got into trouble “Oleg was proud of that brother that always got into trouble.”
Similarly, if the post-copular phrase contains a numeral, such as dva “two”, it must appear in the nominative, as in (32a), and cannot appear in the instrumental, as in (32b). Once again, if the numeral is omitted, the instrumental is grammatical. And again, (32c) shows that there is no problem with a phrase containing a numeral in the instrumental if the instrumental phrase is an argument of a verb that assigns “quirky” (i.e., inherent) case to its object. (32)
xorošix rabotnika]. Oleg i Ivan byli [DP dva Oleg.nom and Ivan.nom were two.nom good.gen workers.gen “Oleg and Ivan were the two good workers.” rabotnikami]. b. Oleg i Ivan byli [NP (* dvumja) xorošimi Oleg.nom and Ivan.nom were (*two.instr) good.instr workers.instr intended: “Oleg and Ivan were two good workers.” c. Oleg rukovodit [DP dvumja xorošimi rabotnikami]. Oleg supervises two.instr good.instr workers.instr “Oleg supervises two good workers.” a.
Finally, pronouns, which are often assumed to be D s, normally appear in the nominative case; this is compatible with the claim made here that instrumentalmarked post-copular phrases do not have the DP-layer. (33)
Èto byl on. this was he.nom “It was him.”
However, there are some examples where pronouns appear in the post-copular position in the instrumental case. Such examples, cited from Nichols (1981: 206), are suggested for the Jekyll-and-Hyde situation; note that in these examples, the
28
CHAPTER 2
pronoun does not have a characteristic referential interpretation, but “the meaning is one of function”, as Nichols (1981: 206) puts it. (34)
a.
Kogda ja byl im, to ja soveršal užasnye prestuplenija. when I was he.instr then I committed terrible crimes “When I was him, I committed terrible crimes.” b. Kogda ja byl samim soboj, byl uvažaemym cˇ elovekom. when I was myself.instr was [respected person].instr “When I was myself, I was a respected person.”
Initially, such examples appear as counter-evidence to the claim that instrumental post-copular phrases have no DP-layer. However, it is not clear if pronouns in Russian are necessarily merged in D ; rather, it appears that they are merged in N and move to D (along the lines of Longobardi’s proposal for proper names in Italian; cf. also Progovac 1998, Franks and Pereltsvaig 2004). Pronouns can be modified by adjectives and can be preceded by demonstratives: (35)
a.
Silnaja ja smogu èto preodolet’. strong I.nom will-manage this overcome “A strong me will manage to overcome this.” b. Ja ljublju togo tebja, kotorogo ja znaju. I love [that you].acc which I know “I love the you that I know.”
To recap, the data above show that DPs in post-copular position must appear in the nominative, and cannot appear in the instrumental. It is impossible to show that bare NPs appear only in the instrumental because of the possibility of null determiners. The argument goes as follows: there are reasons to believe that there is such a thing as a phonetically null article. For example, many researchers assume that in English the plural counterpart of the indefinite article a/an in Boys like to play with toys is phonetically null. The question is then whether such null articles exist not only as “fillers” in otherwise well-established paradigms, but also in languages where no overt articles exist at all (like in Russian). In languages with overt articles and bare NPs, such as Italian (as discussed in detail in Longobardi 1994 and his later work), bare NPs are very restricted in their distribution in argument positions; therefore, we can determine the set of contexts in which a null determiner appears and therefore the set of contexts in which the null determiner cannot appear. Then, it is possible to look at the distribution of bare NPs, namely, nominals that not only lack an overt determiner, but a null determiner as well. In contrast, in Russian it is impossible to distinguish empirically between nominals with a phonetically null determiner and those that lack a determiner altogether. However, based on the contrast observed above, I maintain that in post-copular position the distribution is as follows: DPs appear in the nominative case, and bare NPs appear in the instrumental. A further syntactic contrast between nominative- and instrumental-marked post-copular phrases in Russian comes from the inversion facts. Although both
NOUNS AND THEIR EXTENDED PROJECTIONS
29
sentences with the nom-nom pattern and those with the nom-instr pattern can be inverted so that the order XP1 was XP2 becomes XP2 was XP1, in what follows I show that the process that results in this word order permutation is different for the two types of sentences. The following examples illustrate the inversion for nominative-marked, instrumental-marked and short-form-adjective post-copular phrases: (36)
a.
Oleg byl moj luˇcšij drug. Oleg.nom was my.nom best.nom friend.nom “Oleg was my best friend.” b. Moj luˇcšij drug byl Oleg. my.nom best.nom friend.nom was Oleg.nom “My best friend was Oleg.”
(37)
a.
Oleg byl moim luˇcšim drugom . Oleg.nom was my.instr best.instr friend.instr “Oleg was my best friend.” b. Moim luˇcšim drugom byl Oleg. my.instr best.instr friend.instr was Oleg.nom “As for being my best friend, it was Oleg was.”
(38)
a.
Oleg byl dovolen moej rabotoj. Oleg.nom was satisfied.sf my.instr work.instr “Oleg was satisfied with my work.” b. Dovolen moej rabotoj byl Oleg. satisfied.sf my.instr work.instr was Oleg.nom “As for being satisfied with my work, Oleg was.”
As can be seen from the English translations, the inversion in sentences with the nom-nom pattern, on the one hand, and those with the nom-instr pattern or with a short-form adjective, on the other hand, does not lead to the same pragmatic results. What is even more telling is how the inversion works when anaphoric binding is introduced. The possessive anaphor svoj “self”s’ and the direct object anaphor soboj “himself” must be A-bound by an antecedent in the subject position, that is Spec-TP, as is the case for all reflexives in Russian (for a discussion of anaphoric binding in Russian, see Rappaport 1983, 1986). As can be seen from the following examples, inversion is grammatical for instrumental-marked and short-form-adjective postcopular phrases containing an anaphor, but not for nominative-marked post-copular phrases containing an anaphor. It is worth noting here that the distinction is not between nominal and adjectival phrases, as can be seen from the fact that (39b) patterns with (39c) and not with (39a), nor does the historically nominative shortform adjective pattern with the nominative-marked nominal post-copular phrases, which means that the relevant distinction is not purely morphological.
30 (39)
CHAPTER 2
a.
*Svoji luˇcšij drug byl Olegi . his(anaphor).nom best.nom friend.nom was Oleg.nom intended: “His own best friend was Oleg.” b. Svoimi luˇcšim drugom byl Olegi . his(anaphor).instr best.instr friend.instr was Oleg.nom “Oleg was his own best friend.” c. Dovolen soboji byl Olegi . satisfied.sf himself.instr was Oleg.nom “Oleg was satisfied by himself.”
How can these data be accounted for? As mentioned above, the possessive anaphor svoj “self ”s’ and the direct object anaphor soboj “himself ” must be A-bound by their antecedents, here Oleg. In order to be A-bound, the highest A-position in the anaphor chain must be c-commanded by the antecedent. In the above examples, I conclude that the inverted nominative-marked post-copular phrase appears in an A-position, such as Spec-TP, whereas the inverted instrumental-marked postcopular phrase and the inverted short-form-adjective post-copular phrase appear in an A -position, such as Spec-CP (or a specifier of one of the functional projections in the “exploded” Comp; cf. Rizzi 1997). In other words, the inversion in (36) and (39a) involves A-movement, whereas the inversion in (37), (38) and (39b-c) involves A’-movement.26 The relevant configurations are schematized below: (40)
a.
b.
TP DP
T'
T° …drug ‘friend.NOM’ …Oleg… byl
CP NP/AP
…drugom C° ‘friend.INSTR’ byl
C'
…Oleg…
How does this contrast in the type of movement relate to my claim that the moving category in the former case is a DP, whereas in the latter case it is a bare NP or AP? Here I follow much recent thinking (cf. Lavine 2000, Lavine and Freidin 2002, Bailyn 2004b) and assume that movement to Spec-TP is triggered by EPP rather than (abstract) Case or Topic/Focus features (as proposed for instance by King 1995). The EPP reduces to the presence of an uninterpretable [D]-feature on T , which needs to be checked before the interfaces. Therefore, only a DP can move into SpecTP to satisfy the EPP even though it need not be a subject/Agent.27 Thus, a postcopular DP (which according to the claim of this book is always nominative-marked) moves to Spec-TP to check the EPP/[D]-feature on T , whereas post-copular NPs and APs (which are morphologically instrumental-marked or short-form-adjective post-copular phrases) cannot do so. The only inversion option for the latter type of post-copular phrases is to move by A -movement (à la topicalization in English: Real Ukrainian borsch, I simply love!). I return to the question of the inversion in sentences with the nom-nom pattern (i.e., with the post-copular DP) in Section 3.4 below.
NOUNS AND THEIR EXTENDED PROJECTIONS
31
To summarize, in this section I have provided evidence in support of my claim that post-copular phrases in sentences with the nom-nom pattern are DPs, whereas postcopular phrases in sentences with the nom-instr pattern are bare NPs. Furthermore, I have shown that bare NPs pattern with APs (headed by short-form adjectives) rather than with DPs. I take this to support the idea that referentiality is associated with the D-layer rather than with the noun itself: it is referentiality that distinguishes DPs from bare NPs and APs in post-copular positions. In the next section, I provide further supporting evidence from other languages.
2.6.
CROSS-LINGUISTIC EVIDENCE
In the previous section I argued that some post-copular nominal phrases are DPs and others are bare NPs; more specifically, I proposed that nominative-marked postcopular phrases are DPs, whereas instrumental-marked post-copular phrases are bare NPs. In Russian this generalization is less than obvious due to the lack of overt articles in the language. In this section, I provide some supporting evidence from two languages that have overt articles (although their exact distribution is somewhat different): I show that (as in Russian) post-copular phrases which are identifiable DPs invert by moving into Spec-TP, an A-position, whereas post-copular phrases which are identifiable bare NPs invert by moving into Spec-CP, an A -position. First, consider Italian, which allows both DPs and bare NPs (lacking the functional DP projection) in the post-copular position, regardless of the number specification of the nominal; witness the singular bare count NP in (41b).28 (41)
a.
Maria è la figlia di un generale. Maria is the daughter of a general “Maria is the daughter of a general.” b. Maria è figlia di un generale. Maria is daughter of a general “Maria is a daughter of a general.”
[Longobardi 1994: 628, fn. 23]
As I have argued to be the case in Russian, post-copular DPs and post-copular NPs behave very differently; in fact, as I show below, the post-copular NP in (41b) behaves just like a post-copular AP, and not like the post-copular DP in (41a). First of all, the two sentences in (41) have different meanings. Here is how such meaning differences are described by Renzi (1988: 402–403; translation mine):29 In Giorgio è dottore [glossed as “Giorgio is doctor”] the NP dottore suggests that Giorgio has certain knowledge and abilities of a doctor, as one would say, with an adjective, that he is “bald” to say that he has the property of not having hair. In contrast, Giorgio è il dottore [glossed as “Giorgio is the doctor”] would serve instead to establish the identification of two individuals, defined one by a name (Giorgio) and the other with the profession (of doctor) in the same person.
32
CHAPTER 2
But more importantly for our present purposes, sentences with post-copular DPs and those with post-copular NPs differ syntactically; moreover, as is the case in Russian (see previous section), bare NPs pattern with APs. Recall from the previous section that post-copular DPs (which are nominative-marked in Russian) invert by A-movement into Spec-TP, whereas post-copular NPs (which are instrumentalmarked in Russian) and APs (which are headed by short-form adjectives in Russian) invert by A -movement into Spec-CP, as illustrated in (42) below. In what follows, I use two tests – embedding and cross-over-like effects – to show that the same is true in Italian.30 (42)
TP
a. DP
b. T'
la figlia di T° un generale …Maria… è
CP NP
C'
figlia di C° un generale …Maria… è
Consider the embedding test first. The widely-held assumption that I adopt here is that a TP can be embedded under the complementizer che “that”, but a CP cannot; in other words, Italian does not allow CP recursion. The following generalization can be made from the data in (43): if a sentence with a post-copular DP is inverted, it can be further embedded under che “that”, but the same is not possible when a sentence with a post-copular bare NP is inverted. Moreover, if a sentence with a post-copular AP is inverted, it cannot be embedded either; examples in (44) show that without embedding, such inversion structures are indeed grammatical.31 (43)
a.
Roberta pensa che [la figlia di un generale sia Maria]. Roberta thinks that the daughter of a general is.subj Maria “Roberta thinks that the general’s daughter is Maria.” b. *Roberta pensa che [figlia di un generale sia Maria]. Roberta thinks that daughter of a general is.subj Maria intended: “Roberta thinks that a daughter of a general is Maria.” c. *Roberta pensa che [alto sia Gianni]. Roberta thinks that tall is.subj Gianni intended: “Roberta thinks that it is Gianni who is tall. ” [Zamparelli 2000: 100]
(44)
a.
Figlia di un generale è Maria. daughter of a general is Maria “A daughter of a general is Maria.” b. Alto è Gianni. tall is Gianni “It is Gianni who is tall.”
From this I conclude that the bracketed string in (43a) is an TP and the bracketed strings in (43b) and (43c) are CPs. Thus, whether inversion of a copular sentence
NOUNS AND THEIR EXTENDED PROJECTIONS
33
is done by A- or A -movement depends on the category of the post-copular phrase, namely, whether it is a DP or an NP/AP. The second test involves cross-over-like effects. The relevant configuration is one that exhibits the possibility for a noun phrase contained within the inverted pre-copular XP (DP, NP, or AP) to bind the possessive pronoun in the post-copular phrase of an inverted copular sentence. Crucially, I assume here that cross-over applies only in the case of A -movement. The generalization emerging from the data below is that the inversion of a post-copular DP does not create cross-over effects, whereas corresponding inversion of an NP or an AP does. This once again strongly suggests that the inversion of an NP or an AP is an A -movement, whereas the inversion of a DP is an A-movement. Thus, these data further support the structure for inverted copular sentences proposed in (42) above. (45)
a.
La figlia di un generalei è la suai segretaria. the daughter of a general is the his secretary “The general’s daughter is his secretary.” b. *Figlia di un generalei è la suai segretaria. daughter of a general is the his secretary intended: “A daughter of a general is his secretary.” c. *Orgoglioso di Mariai (lo) è il suoi professore. proud of Maria lo is the her professor lit.: “Proud of Maria is her professor.”
To recap, Italian allows both DPs and NPs to appear in what looks like the same post-copular position. Yet, the two types of phrases do not pattern together, in particular in the context of copular inversion, suggesting that DPs and NPs do not appear in the same position in (41) above. Moreover, it is crucial that bare NPs pattern with APs in these constructions, and not with DPs. This strongly suggests that both APs and bare NPs have something in common that distinguishes them from DPs. In line with my proposal for Russian in the previous section, I propose here that the common characteristic of APs and bare NPs is their non-referentiality, which is what distinguishes them from DPs. The same contrast between post-copular DPs, on the one hand, and post-copular bare NPs and APs, on the other hand, in particular with respect to inversion, is found in Norwegian a well (similar Danish data are discussed in Mikkelsen 2002). Like Italian, Norwegian allows (definite) DPs, bare NPs (easily distinguishable in Norwegian from indefinite DPs), and APs; these types of post-copular phrases are illustrated below:32 (46)
a.
Peter er [DP den største beundreren av finnebiff]. Peter is the biggest fan of reindeer-meat “Peter is the biggest fan of reindeer meat.” b. Per er [NP lærer]. Per is teacher “Per is teacher.”
34
CHAPTER 2
c.
Per er [AP syk]. Per is sick “Per is sick.”
Even though all three types of sentences can be inverted (see below), the inversion does not happen in the same fashion in all cases: as is the case for Russian and Italian, post-copular DPs invert by A-movement into Spec-TP, whereas post-copular NPs and APs invert by A’-movement into Spec-CP (the copula always appears in second position because Norwegian is a verb-second language). (47)
er Peter. [DP Den største beundreren av finnebiff] the biggest fan of reindeer-meat is Peter “The biggest fan of reindeer meat is Peter.” b. [NP Lærer] er Per. teacher is Per “As for being a teacher, Per is.” c. [AP Syk] er Per. sick is Per “As for being sick, Per is.” a.
In what follows, I show three pieces of evidence in support of this claim, involving (i) anaphoric binding (similar to the Russian data in (39) and the Italian data in (45) above), (ii) embedding in no-CP-recursion contexts (similar to the Italian data in (43) above), and (iii) the position of negation ikke “not” and lower adverbs (e.g., alltid “always” and ofte “often”). The first piece of evidence involves anaphoric binding. The relevant configuration involves a reflexive anaphor (in this case, the possessive sin “self ”s’ contained in the inverted phrase. As schematized below, this configuration is grammatical if the inverted phrase is a bare NP or an AP; in contrast, it is judged as marginal if the inverted phrase is a referential DP.33 (48)
a.
??[DP Den største beundreren av sin kones kokekunst] var ikke Peter. the biggest admirer.def of self’s wife’s cooking was not Peter intended: “The biggest admirer of his own wife’s cooking wasn’t Peter.” kones kokekunst] var Peter ikke. b. [NP Offer for sin victim of self’s wife’s cooking was Peter not “A victim of his wife’s cooking, Peter wasn’t.” c. [AP Stolt av sin kones kokekunst] var Peter ikke. proud of self’s wife’s cooking was Peter not ≈ “As for being proud of his own wife’s cooking, Peter wasn’t.”
This is completely parallel to the Russian and Italian facts discussed above and receives the same explanation: the post-copular DP inverts into an A-position relevant for establishing the binding configuration, whereas the post-copular NP and AP invert into an A’-position and therefore are interpreted in their “reconstructed” positions, as far as anaphoric binding is concerned. Note that the data are
NOUNS AND THEIR EXTENDED PROJECTIONS
35
also compatible with the analysis where the inverted phrase appears in Spec-CP regardless of its referentiality (Norwegian being a verb-second language), but in the case of a DP it must pass through Spec-TP on its way to Spec-CP, whereas an NP or an AP would not pass through Spec-TP. The second piece of evidence involves embedding in no-CP-recursion contexts. When the matrix verb is one that does not allow embedded verb-second, the following generalization emerges with respect to the embedding possibilities: if the inverted phrase is a DP, the embedding is grammatical; in contrast, if the inverted phrase is a bare NP or an AP, the embedding is ungrammatical. This is illustrated below:34 (49)
a.
b.
c.
Advokaten nekter for at [DP det tjukkeste offeret ] (ikke) var Peter. lawyer.def denied that the fattest victim.def not was Peter “The lawyer denied that the fattest victim was(n’t) Peter.” (establishing/denying the identity of the fattest victim, not a property of Peter.) *Advokaten nekter for at [NP tyv ] var Peter (ikke). lawyer.def denied that thief was Peter not intended: “The lawyer denied that Peter was(n’t) a thief.” *Advokaten nekter for at [AP skyldig ] var Peter (ikke). lawyer.def denied that guilty was Peter not intended: “The lawyer denied that Peter was(n’t) guilty.”
Since the embedded constituent in this case must be a TP, we can conclude that the inverted post-copular DP appears in the Spec-TP, whereas the inverted post-copular NP and AP appear in the Spec-CP. The third and final piece of evidence involves the position of negation and lower adverbs.35 The generalization is the following: if the inverted post-copular phrase is a DP, the negation/adverb (in boldface) must precede the subject DP (in italics); in contrast, if the inverted post-copular phrase is a bare NP or an AP, the negation/adverb must follow the subject DP and appears in an otherwise unusual sentence final position:36 (50)
a.
[DP Den beste vennen min] var alltid Peter. the best friend.def my was always Peter “My best friend was always Peter.” (presupposes that I always had a best friend) b. [NP Tyv] var (*alltid) Peter (alltid). thief was (*always) Peter always ≈ “As for being a thief, Peter was always that.” c. [NP Syk] var (*alltid) Peter (alltid). sick was (*always) Peter always ≈ “As for being sick, Peter was always that.”
The widely-held assumption is that negation and adverbs are in the same position in all of these sentences; specifically, they mark the left edge of the small clause. Consequently, in sentences with an inverted DP, the subject (here, Peter) stays in the small clause, whereas in sentences with an inverted NP or AP, Peter appears outside
36
CHAPTER 2
the small clause, presumably in the Spec-TP. This difference can be explained if we assume that in (50a) the inverted DP moves into Spec-TP; as a result, Peter cannot move into this position. In senteces (50b-c), the inverted NP or AP must then appear in Spec-CP. Note that the position of the copula is determined by the verb-second effects. To summarize this section, I have shown that, just as is the case in Russian, in languages with overt articles which allow bare NPs in the post-copular position, such NPs pattern with APs and not with DPs. 2.7.
SUMMARY
In this chapter, I have argued for the following points: • The best defining criterion for lexical vs. functional categories is their association with thematic grids: only lexical categories are inherently associated with thematic grids, whereas functional categories are not (the only way a functional projection will have a thematic grid is if it inherits one from a lexical category; I return to this point in Chapter 4). • Nouns, much like adjectives, are predicative expressions which bear their own theta-grids (cf. Stowell 1989: 233); nouns are not bearers of referential indices (contra Baker 2003). • The burden of referentiality is to be placed on the D-layer, not the N-layer. Consequently, DPs are referential, whereas NPs are predicative in nature (as in Longobardi 1994, and contra Baker 2003). • Under inversion, post-copular DPs move into Spec-TP, an A-position, whereas post-copular NPs and APs move into Spec-CP, an A’-position. In the following chapters, these points are picked up in the following way: • In the next chapter: • the question of whether symmetrical structures like the one I propose for Russian copular sentences with the nom-nom pattern can be generated by syntax is considered in detail; • further attention is given to the issue of inversion in symmetrical copular sentences. • In Chapter 4: • the proposal is developed to the effect that an NP’s -role is discharged by -binding from the D (as in Higginbotham 1985); • post-copular NPs and APs are shown to be unable to discharge their -role by -marking their specifiers directly, without the “help” of a special head (following Baker 2003); • this special “helping” head is analyzed as a lexical rather than a functional category (contra Baker 2003).
NOUNS AND THEIR EXTENDED PROJECTIONS
37
• In Chapter 5: • it is argued that treating the two DPs in sentences with the nom-nom pattern as referential but not as argumental is advantageous in accounting for the distribution of case marking in copular sentences; • it is further argued that treating the head “helping” adjectives and nouns assign their -roles as a lexical and not as a functional category is likewise advantageous in accounting for the distribution of case marking in copular sentences.
This page intentionally blank
CHAPTER 3
PHRASE STRUCTURAL RELATIONS AND MERGE: SYMMETRY OR ANTISYMMETRY?
One issue raised by the proposal outlined in the previous chapter is whether symmetrical structures, such as those involved in sentences with the nom-nom pattern, can be generated in syntax. This issue is a very important one: it concerns one of the most fundamental properties of phrase structure. The choice one makes on this issue affects not only the types of structures that can be generated by the phrase structure component, but also the types of movement one’s theory allows or requires, and this will have a profound effect on all levels and components of the grammar. So are symmetrical structures to be allowed? Here I argue that the symmetrical structure proposed for sentences with the nom-nom pattern can be generated by Merge conceived in the true minimalist spirit. The major idea behind Chomsky’s Minimalist Program is that the grammar should be restricted to “virtual conceptual necessity” (Chomsky 1995: 169), retaining solely those mechanisms and principles that are absolutely necessary. In practice, however, it is not always easy to determine what is absolutely necessary in accounting for language data. But one thing is clear: syntax must include an operation that creates larger syntactic units out of smaller ones. In the Minimalist Program this operation is called Merge. This operation applies to two objects and and forms a new object, K. This new object K must be somehow composed of the two objects that constitute it, namely and . It is an agreed upon assumption that syntactic units are sets of features (see, for example, Jackendoff 1972: 21–22, Chomsky 1995: 244, 1998: 116, 2000: 126). Accordingly, Merge is an operation that calculates the output set of features from the two input sets of features. The crucial question is thus how the output set of features is calculated from the input sets of features and . In this chapter, I examine the claim of Chomsky (1995) and Kayne (1994) that Merge proceeds exclusively by asymmetrically projecting one or the other of the input constituents (Section 3.1) and discuss some of the problems raised by the Antisymmetry Hypothesis (Section 3.2). I argue that there is nothing that bans symmetrical structures from syntax a priori; therefore, the only way to exclude symmetrical structures is by introducing an unmotivated stipulation. Moreover, I argue that the empirical motivation for the Antisymmetry Hypothesis is undermined by the fact that the alleged empirical advantages of this theory are outweighed by its costs. Furthermore, I argue that the theoretical motivation for the Antisymmetry Hypothesis is problematic as well. Then, I conclude that there is no motivation for applying the LCA throughout the derivation. Furthermore, I will 39
40
CHAPTER 3
reject Moro’s (2000) proposal that all movement is triggered by the search for antisymmetry. Instead, I will adopt the view that neutralizing a point of symmetry is a possible consequence of movement, but is not required (and therefore cannot trigger movement). In Section 3.3, I consider the question of how symmetrical structures can be generated by Merge. In this respect, I will propose that if the two input constituents have the same feature compositions, Merge need not asymmetrically project one of the input constituents. In Section 3.4, I briefly discuss some constructions which may be analyzed as involving symmetrical structures. Finally, Section 3.5 summarizes the argumentation of this chapter.
3.1.
THE ANTISYMMETRY HYPOTHESIS
In this section, I consider the hypothesis – proposed by Kayne (1994) and adopted by Chomsky in the Minimalist Program and later work – that Merge proceeds asymmetrically in all cases. As Chomsky (1995: 244) puts it: “the label must be constructed from the two constituents and . … Then the simplest assumption would be that is either: (51)
a. the intersection of and b. the union of and c. one or the other of , ”
However, Chomsky immediately excludes the first two possibilities because “the intersection of , will generally be irrelevant to output conditions, often null; and the union will be not only irrelevant but “contradictory” if , differ in value of some feature, the normal case.” This leaves him with only one option – that in (51c) – namely, asymmetric projection of either or . Therefore, Chomsky (1995: 246) concludes that “the operation Merge(, is asymmetric, projecting either or ”. This view is also adopted by Kayne (1994), who argues that all syntactic representations are asymmetrical in nature. The major motivation behind Kayne’s (1994) Antisymmetry Hypothesis is the idea that “the human language faculty is … rigidly inflexible when it comes to the relation between hierarchical structure and linear order” (p. xiii). In particular, “phrase structure … always completely determines linear order” through “asymmetric c-command invariably map[ing] into linear precedence” (p. 3). This is formulated as the Linear Correspondence Axiom (Kayne 1994: 5–6). (52) Linear correspondence Axiom (LCA) d(A is a linear ordering of T where A contains all pairs of nonterminals such that the first asymmetrically c-commands the second; d(X) is the set of terminals that X dominates; T is the set of terminals.
PHRASE STRUCTURAL RELATIONS AND MERGE
41
The LCA eliminates the distinction between a specifier and an adjunct and rules out the following three symmetric configurations, which, following Moro (2000), I will call “points of symmetry”: (53)
a.
XP YP
b. ZP
ZP X°
c. Y°
XP YP
XP ZP
XP
The structure in (53a) represents a combination of two maximal projections neither of which adjoins to the other (this is the structure that is of most interest for the discussion in this book); (53b) is a configuration in which neither of the two heads project, and (53c) is a multiple specifier configuration.37 Note also that even though there are two linear orderings defined on terminal nodes – precedence and subsequence – Kayne (1994: 35–36) claims that there are empirical reasons that force antisymmetric c-command to map into precedence rather than subsequence. For example, precedence defines the universal word order as SVO (or SOV; see the discussion in the next section), whereas subsequence defines it as OVS. A quick look at the world’s languages reveals that there is a sharp asymmetry between those two orders, the former one being much more common. Furthermore, the specifier of CP, “the typical landing site for moved whphrases, is visibly initial to an overwhelming degree” (p. 35). Overall, “although there may be some categories for which both orders [of specifier and head] are widespread, there are other categories where specifier-head order strongly predominates” (p. 35); furthermore, there are no categories for which head-specifier order is cross-linguistically predominant. Therefore, Kayne concludes that “specifier-headcomplement, and not the reverse, is the only order available to the subcomponents of a phrase” (p. 36), which in turn means that asymmetric c-command maps onto precedence, not subsequence. The choice of precedence as the relation underlying the relation between hierarchical structure and linear order means that both rightward adjunction and rightward movement are banned. Obviously, this rigid view of phrase structure compromises descriptive adequacy of the theory. In order to account for the phenomena previously accounted for by rightward adjunction or rightward movement (e.g., Right Node Raising, Heavy NP Shift, Extraposition and Right Dislocation), Kayne (1994) proposes alternative analyses that involve LCA-compatible structures and rely on massive leftward movement. Thus, a heavy load of descriptive adequacy is transferred from the Phrase Structure component to the Movement component. Crucially, Kayne (1994) envisages LCA as applying at all levels of syntactic representation, including “LF and … D-structure (or the closest counterparts to D-structure in Chomsky’s (1993) framework), for which one might think that linear order is not essential” (Kayne 1994: 48). This is so because Kayne conceives the LCA as “the source of all the major properties of phrase structure that have been attributed to X-bar theory”; therefore, he concludes that “the LCA does underlie the entire set of syntactic representations” (p. 49).
42
CHAPTER 3
To recap, Kayne (1994) proposes that hierarchical structure and linear order are related through the LCA, which bans symmetrical structures and rightward adjunction and rightward movement. Crucially, for Kayne (1994) the LCA applies at all levels of representation.
3.2.
CHALLENGING THE ANTISYMMETRY HYPOTHESIS
As noted by Chomsky (1995: 335), “Kayne offers two kinds of arguments for the LCA: conceptual and empirical”. Let us consider the empirical motivation for the LCA first. In order to evaluate a theory (relative to other theories) on empirical grounds, one needs to compare the relative “gains and pains” of the two theories. Even though a detailed discussion of the merits and the drawbacks of the Antisymmetry Hypothesis goes beyond the scope of this book, I argue here that it does not provide a superior account of many linguistic phenomena. The main empirical consequences of the LCA, as presented by Kayne (1994), include deriving SVO as the universal underlying order, and accounting for facts involving agreement with adpositions and relative clauses, to name just a few. Let us consider the claim that LCA derives SVO as the universal underlying order. Whether this is indeed so depends on whether objects are merged as complements or as specifiers of the verb. In the former case, LCA predicts that SVO is the universal underlying order, whereas in the latter case, the predicted universal underlying order is SOV. The latter situation is more in tune with the typological studies, which show that the majority of existing languages are actually SOV, not SVO. For example, Mallinson and Blake (1981) claim that 41% of languages in their sample are SOV, as opposed to only 35% of languages that are SVO. However, do such typological generalization play a role in defining syntactic theory? According to Chomsky’s conception of the grammar throughout the years, cross-linguistic variation defines the bounds of human language but not the statistical distribution of languages into different typological categories. For example, Universal Grammar allows question formation through fronting of the highest auxiliary to the pre-subject position (e.g., Is the boy who is playing happy?) but not of the linearly first auxiliary (e.g., *Is the boy who playing is happy?); yet, Universal Grammar has nothing to say about the number or proportion of languages that use auxiliary fronting for question formation (as opposed to question particles, reliance on intonation and other strategies). As Newmeyer (1998: 161) summarizes it, “grammars do not encode typological generalizations, either directly or indirectly… universal grammar tells us what a possible language is, but not what a probable language is”. Another conceptual problem with Kayne’s claim that LCA is justified by the predominance of SVO (or SOV) languages is that in the Antisymmetry framework any surface word order can be achieved from the underlying SVO (or SOV) order through a number of leftward movements (for a number of recent analyses that derive overt SOV order from the underlying SVO order and vice versa, see
PHRASE STRUCTURAL RELATIONS AND MERGE
43
Barbiers 2000, Brody 2000, Haegeman 2000, Haider 2000, Holmberg 2000, Hróarsdóttir 2000, Taraldsen 2000, among others). Therefore, it appears that languages that exhibit the surface SVO order (like English) or the surface SOV order (like Hindi) do so only accidentally. Thus, in Kayne’s analysis the fact that a large number of languages are overtly SVO (or SOV) is not supported by his claim that all languages are underlyingly SVO (or SOV). In fact, any tendency for a predominance of a certain surface word order is purely accidental, since any surface order is derivable from any underlying order with more or fewer movements. In this context, Kayne’s choice of precedence as the relation defining the mapping from hierarchical relations to linear ordering loses much of its attractiveness. Since the majority of world languages (65% according to Mallinson and Blake 1981) are not SVO, there is no reason to assume that SVO and not OVS is the universal underlying order. Therefore, the hypothesis that OVS is the universal underlying order is perfectly plausible in Kayne’s framework since all attested surface word orders can be derived from it. Just as a curious note, I will add here that the “OVS as underlying order” hypothesis has had its supporters throughout the history of linguistic thought. For example, Etienne Bonnot de Condillac, an eighteenth century French grammarian, cited in Lepschy (1998: 196), considered OVS as the “most natural order”: first the noun indicating the object one was talking about, then the verb indicating the operation one intended to carry out on that object: for example, fruit want; the subject of the verb came at the end of the whole series: for example, fruit want Peter. To recap, there are several empirical and conceptual problems with Kayne’s claim that it is an advantage of the Antisymmetry Hypothesis that it can derive SVO as the universal underlying word order. Other alleged empirical advantages of the Antisymmetry Hypothesis are problematic as well. For example, Kayne (1994) argues that dispensing with the distinction between adjuncts and specifiers has desirable empirical consequences. However, this claim has been challenged by Duffield (1999), who has argued that retaining the distinction between adjuncts and specifiers allows for a natural account of the variable freedom of attachment of AP modifiers between Semitic languages (Hebrew and Maltese), on the one hand, and Celtic languages (Irish), on the other hand. Furthermore, the exclusion of rightward adjunction and rightward movement leaves the Antisymmetry Theory with no elegant way of accounting for the facts concerning contrastive focus in Italian and Russian, as described by Samek-Lodovici (1996) and Pereltsvaig (2004).38 . To sum up so far, at least several of the alleged empirical advantages of the Antisymmetry Hypothesis are dubious. Moreover, the conceptual costs of this theory are quite high. More specifically, all constructions previously analyzed as involving points of symmetry or rightward adjunction/movement must be reanalyzed as involving numerous leftward movements, including remnant movements. In addition to the increased burden on the movement component of the grammar, a problem arises as to the motivation for these movements that a theory based on the
44
CHAPTER 3
Antisymmetry Hypothesis has to rely on. Nor is it apparent how such instances of (remnant) movement are restricted. In a way, this makes movement a fix-up solution for structures that do not come out the way the LCA predicts (cf. Moro’s 2000 view on movement in Dynamic Antisymmetry Hypothesis; see below). In the spirit of the Minimalist Program, where movement is motivated only by feature checking, the proponents of the LCA must introduce a large number of [–Interpretable] features in order to motivate movement and derive the correct word order. Even worse, they must rely on introducing a large number of functional heads that can host the moved phrases (including instances of remnant movement). Even though in some cases these heads may turn out to have semantic and/or phonetic content, many heads proposed in the framework of the LCA are semantically inert and phonetically null as well, their sole purpose being to host the moved phrases (and in essence to get the correct word order). In other words, these functional heads consist of [–Interpretable] features only and receive no support from either of the interfaces. But, according to Chomsky (1995: 349–355), such functional heads must be dispensed with. Indeed, for Chomsky (1995: 378) “the only functional categories are those with features that survive through the derivation and appear at the interfaces, where they are interpreted”. By this logic, fix-up functional categories the sole purpose of which is to host moved phrases are not allowed in the grammar. In essence, restricting the theory by disallowing certain (LCA-incompatible) structures results in the need to loosen the theory in a different aspect, namely, by allowing an unrestricted number of semantically inert functional categories. To sum up, the costs of the Antisymmetry Hypothesis seem to outweigh the empirical gains it might have. What about conceptual motivation for the LCA? According to Kayne (1994: 3), the main conceptual motivation for the Antisymmetry Hypothesis is that it “yield[s] a derivation of the essentials of X-bar theory”. In other words, X-Bar Theory is no longer a primitive but is rather derived from the LCA. One problem with this claim is that within the Minimalist framework the X-Bar Theory is abandoned in favor of the Bare Phrase Structure, where the essentials of the X-Bar Theory (which Kayne claims to be derived from the LCA) are for the most part derivable without the LCA (Chomsky 1995: 336). Yet, even if one does not adopt the Bare Phrase Structure, it is not clear if the LCA (as formulated by Kayne 1994) can indeed derive the basic properties of the phrase structure as defined by the X-Bar Theory. This point is investigated in detail in Akiyama (2000), who shows that only a subset of the essential properties of the X-Bar Theory can be derived from the LCA, whereas other important properties of the X-Bar Theory cannot be derived from LCA alone, which undermines the conceptual motivation for the Antisymmetry Hypothesis. Akiyama’s (2000: 43–44) list of principles that underlie the standard X-Bar Theory includes the following:39 • Lexicality: Every non-terminal (= phrase) is Xi , where X is a terminal and i >0. • Succession: Every rule rewriting some non-terminal Xn has a daughter labeled Xn−1 . • Uniformity: The maximum possible bar-level is the same for every category.
PHRASE STRUCTURAL RELATIONS AND MERGE
45
• Maximality: Every non-head daughter in a rule is a maximal projection. • Optionality: Non-heads are only optionally present. • Binary Branching: An Xn has at most one non-head daughter. As Akiyama proceeds to show, although some consequences of these principles can be derived from LCA, (e.g., two consequences of maximality: prohibition against doubly-headed structures and prohibition against non-maximal complements), it is not the case that all consequences of each of these principles can be derived from LCA. For instance, LCA cannot derive the principle of endocentricity, which guarantees that every non-terminal node has a head (i.e., a terminal node which it is a projection of) and that this head is of the same categorial type as the projection. Thus, LCA does not derive the contention – which fuelled so much research in generative syntax – that a clause (= S) must be a projection of some terminal (V or I or T). In standard X-Bar Theory, this contention is a consequence of the principle of lexicality; yet, LCA cannot ban the following structure (from Akiyama 2000: 51, his (15)): (54)
*
S(≠ IP or Vmax) S(≠ IP or Vmax)
NP N
I
VP
John
will
V laugh
In this structure, the ordering of terminals is total, transitive and antisymmetric, as required by the LCA: d(A) = {<John, will>, <John, laugh>, <will, laugh>}. Hence, as far as LCA is concerned, there is no reason to believe that S is the projection of Infl, V or any other category. Likewise, LCA has no straightforward way of excluding the following structure (from Akiyama 2000: 52, his (16)): (55)
VP
* N1
PP
assassination P
NP
of
N2 J.F.K.
The d(A) for this structure is {
, , }, which is once again total, transitive and antisymmetric, which yields no violation of LCA. For a further discussion of principles of X-Bar Theory which
46
CHAPTER 3
LCA cannot derive, the reader is referred to Akiyama’s work. Thus, unfortunately for the Antisymmetry Hypothesis, Akiyama (2000) concludes that LCA cannot derive all the properties of the X-Bar Theory. One way to retain some of the advantages of the Antisymmetry Hypothesis without the pitfalls mentioned above is to adopt a weaker version of the Antisymmetry Hypothesis, namely that the LCA does not apply at all levels of representation. Thus, it is possible that LCA is applicable only at the level where linearization process occurs, namely, at PF. Moreover, on the assumption that null categories are not visible at PF (in the relevant sense), they need not be linearized. Therefore, it is not clear why LCA should apply to them. This line of reasoning is entertained in Chomsky (1995: 337), who maintains that “there is no reason for LCA to order an element that will disappear at PF, for example, a trace”. The same idea is briefly entertained in Kayne (1994: 133, ch.2, note 3) and has been explored extensively in the work by Franks and Boškovi´c on Slavic clitics (cf. Franks 1999, Franks and King 2000, Boškovi´c 2001, Franks and Boškovi´c 2001).40 In addition, the hypothesis that LCA applies only at the level where linearization occurs has been explored by Moro (2000). Thus, Moro too subscribes to the idea that “the LCA, mapping hierarchy onto linear order, is active only when linear order is required by definition – that is, only when words are spelled-out” (Moro 2000: 28). Crucially for him, symmetrical structures can be generated by Merge, but this symmetry has to be neutralized by PF. The obvious way to neutralize a point of symmetry is to move one of the offending constituents. Movement saves the structure from a violation of the LCA (at PF) because it leaves a trace in the position of the moved element. According to Moro (2000: 28), by definition, traces are not visible in the linear sequence at PF. Thus, if one of the elements constituting the point of symmetry is a trace, no problem is expected to arise… In fact, Moro (2000) makes an even stronger claim. According to him, neutralizing a point of symmetry is not only a possible consequence of movement, it is its main motivation. Thus, Moro’s Dynamic Antisymmetry Hypothesis (p. 28) states that “movement is driven by the search of antisymmetry” rather than by morphological feature checking (as in Chomsky’s Checking Theory). In his book, Moro does not account for all types of movement; rather, he supports his hypothesis with an illustration of how movement neutralizes the three types of points of symmetry schematized in (53), repeated below. The configuration in (53a) is most relevant for this book since it obtains with small clause complements of the copula. The configuration in (53b) is relevant for the analysis of clitics; the configuration in (53c) occurs with wh-movement of objects. (53)
a.
XP YP
b. ZP
ZP X°
c. Y°
XP YP
XP ZP
XP
PHRASE STRUCTURAL RELATIONS AND MERGE
47
Even though there are some interesting results following naturally from the Dynamic Antisymmetry Hypothesis, there are numerous problematic gaps in the analysis. Most importantly, it is not clear how many of the paradigmatic cases of movement, such as A-movement constructions including passive, raising, unaccusatives, VPinternal subjects, ditransitive constructions, and Germanic-type object shift, can be accounted for within the Dynamic Antisymmetry framework.41 Only a brief footnote in Moro’s monograph (2000: 126) is devoted to A-movement, and only subject-to-subject raising and VP-internal subjects are discussed there. The analyses for both raising and VP-internal subjects are left very sketchy and many problems arise as to the details of these analyses (for a detailed discussion and critique of Dynamic Antisymmetry analyses of A-movement constructions, see Pereltsvaig 2001b). Thus, it is not obvious how the Dynamic Antisymmetry Hypothesis can provide an insightful analysis of A-movement constructions (and object shift in Germanic). It remains to be seen whether purely mechanistic analyses can be devised to account for these phenomena, or whether they present a real challenge for Moro’s theory. Therefore, I will not adopt the strong versions of the Antisymmetry Hypothesis or the Dynamic Antisymmetry Hypothesis in this book. Rather, I will develop the view that Merge can generate symmetrical structures which then need to be converted into LCA-compatible asymmetrical structures to be linearized at PF. Throughout the book, I argue that a theory that allows Merge to generate symmetrical structures (for example, in copular sentences with the nom-nom pattern in Russian) is preferable to Kayne’s (1994) Antisymmetry Hypothesis in that it allows us to account for a number of empirical facts concerning Russian copular sentences (including case marking, semantic interpretation, and syntactic properties such as binding and extraction) without relying on stipulations and principles that are not otherwise necessary. 3.3.
MERGING SYMMETRICAL STRUCTURES
In the previous section, I have put forward some arguments against the Antisymmetry Hypothesis of Kayne (1994). I argued instead that symmetry must be permitted throughout the derivation. But I have not yet addressed the question of how it is possible to generate points of symmetry in the first place. Recall that Kayne (1994) envisages the LCA as the substitute for X-Bar Theory, which applies at all levels of representation. However, Moro (2000) has proposed to depart from the view that Merge is necessarily asymmetric. According to him, Merge can generate symmetrical structures. I will agree with Moro that Merge is not necessarily asymmetric, but I will propose a different theory as to exactly how symmetrical structures can be generated. First, consider Moro’s proposal for generating symmetrical structures. He argues that in addition to asymmetrical projection – the only option allowed by Chomsky (1995: 244) – there is an option of projecting a node that has an empty label. Thus, Moro (2000: 33) claims that Merge can create three types of output:
48 (56) a. <>(or <> b. <, >(or <, > c. <>
CHAPTER 3
simple label – asymmetric projection complex label – adjunction empty label – symmetrical structure
The second option – complex label, created in the adjunction process – is not particularly relevant for the discussion in this book. Therefore, I will not talk about it here (for more discussion, the reader is referred to Chomsky 1995 and Moro 2000). The option that interests us here is the third one, namely, projecting an empty label. According to Moro (2000:33), this option satisfies the two conditions he identifies as underlying the Minimalist notion of Merge. In particular, this option “does not add extra information” and “there are no mixed labels” created as a result of this process (Moro 2000:33). However, there is a serious objection to this proposal. Recall that syntactic objects are sets of features; the label <>means that the object does not have any features, not even categorial features. However, a node with no features is not a legitimate syntactic object at all. In fact, it cannot be selected, and therefore cannot be a complement to a head. Moro himself does not specifically discuss the issue of selection, but according to him symmetrical small clauses (“bare small clauses”, in his terminology) are complements of, and presumably selected by, the copula. Furthermore, Moro (2000:71) proposes to analyze clauses (i.e., TPs) as symmetrical structures consisting of the subject DP and TP projection (the latter corresponds to an T in standard Government and Binding structures). Obviously, a clause, whether full or “small”, must be able to be selected, for example, by a copula, a complementizer or a verb (as in ECM constructions). Therefore, the issue of selection is very important here. Furthermore, a node with no features cannot project further, so that it cannot be a head of a larger structure either. In other words, syntactic computation can manipulate only features, and therefore, it cannot manipulate something that has no features. From the point of view of syntactic computation, something that does not have syntactic features does not exist at all. Thus, we do not want to generate symmetrical structures as empty nodes. The question is then what is a viable option for generating such symmetrical structures. At this point, recall that Chomsky (1995: 244) presents three options for Merge – union, intersection and asymmetrical projection. The first two options are immediately excluded because they create illegitimate syntactic constituents. However, Chomsky limits his discussion of union and intersection as non-viable ways for Merge to proceed to the “normal case”, namely, when the two input constituents and differ in some feature(s). Here, I would like to turn our attention to the “abnormal” case, namely such when the two input constituents and do not differ in features relevant for syntactic computation. I claim that in such cases, union and intersection are viable options for Merge to proceed (in fact, the results of union and intersection are indistinguishable from each other in such cases; cf. (59) below). Consider the following “toy” example: three features are relevant for syntactic computation, [±F], [±G] and [±H]; every syntactic constituent must be specified
PHRASE STRUCTURAL RELATIONS AND MERGE
49
for the first two of the features (they may be conceived as category features, for example, corresponding to [±N] and [±V]). The third feature – [±H] – is optional. In what Chomsky refers to as the “normal” case, the two input constituents and differ in features. In our example, assume that is specified as {+F, –G} and is specified as {–F, +G} (curly brackets are used here to indicate sets of features). In this case, the output of Merge can be one of the following (ignoring the adjunction possibility, which as mentioned above is largely irrelevant for the discussion in this book): (57)
a. b.
= {+F, –G} = {–F, +G}
[i.e., projects] [i.e., projects]
In (57a), asymmetrically projects; in other words, it is the features of that are taken to be the features of . In (57b), asymmetrically projects; in other words, it is the features of that are taken to be the features of . Now, consider what sets of features cannot be taken as the result of Merge of and defined above: (58)
a. * {+F, +G} b. * {–F, –G} c. * {+F, –G, +H} d. * {} e. * {+F, –G, –F, +G}
The set in (58a) is not a possible output of Merge because a mixed label is created: the value for [±F] is taken from and the value of [±G] is taken from . The set in (58b) is ruled out for the same reason; in this case the value for [±F] is taken from and the value of [±G] is taken from . The set in (58c) is ruled out because it violates the second condition for Merge identified by Moro (2000: 33) – it adds extra information not present in either or , namely, [+H]. The options in (58d) and (58e) are ruled out for the reasons discussed by Chomsky (1995: 244). In particular, in (58d), the output node is created by an intersection of features of and , and it has too few features for it to be identified as a legitimate syntactic constituent. In particular, it lacks the specifications for [±F] and [±G], which are defined as obligatory categorial features in our toy model of syntax. In (58e), the output node is created by a union of features of and , and it has too many features (which are also contradictory) for it to be identified as a legitimate syntactic constituent. Therefore, this option is also excluded as non-viable in the situation where the two input constituents differ in features. Now consider an “abnormal” case, namely, where the two input constituents have identical (values of) features. Assume that the two input constituents have the following specifications: both and are specified as {+F, –G}. In this case, the output of Merge can be one of the following: (59)
a. {+F, –G} b. {+F, –G}
union of and intersection of and
50
CHAPTER 3
Whether Merge proceeds by union or by intersection of and , the resulting output is the same, namely {+F, –G}. Furthermore, in this case asymmetric projection of is indistinguishable from the asymmetric projection of . In other words, whether Merge proceeds by asymmetrically projecting , asymmetrically projecting , union or intersection of the two, the results are the same. Therefore, I conclude that the simplest assumption about Merge is it can proceed in one of the three ways identified by Chomsky (1995: 244), but in certain cases – those where the two input constituents differ in features – union and intersection result in an illegitimate syntactic object, which would be discarded by the syntactic computation. On the other hand, if the two input constituents have the same features, the three ways for Merge to proceed identified by Chomsky all lead to the same result. Thus, the simplest theory of Merge contains no stipulations excluding the union and intersection options. Throughout this book, I argue that a symmetrical structure proposed here for Russian copular sentences with the nom-nom pattern is the best way to account for such sentences and for their contrast with copular sentences with the nom-instr pattern. 3.4.
SYMMETRICAL STRUCTURES
In the previous section, I have argued that symmetrical structures can be generated under the simplest assumptions about how the operation Merge proceeds. The next obvious question is whether such symmetrical structures are ever found in natural language. Since much of the recent research has concentrated on showing that asymmetry is the characteristic property of syntactic structure, at this point it is impossible to give a fully substantiated positive answer to the above question. However, a number of natural language constructions have been analyzed as instances of symmetrical structures. In addition to copular sentences of the relevant type (discussed here and in Moro 2000), other candidates for symmetrical structures include: • certain light verb constructions in Hindi/Urdu (cf. Butt and Ramchand 2001); • serial verb constructions in West African languages, such as Èdó (cf. Baker and Stewart 1999); • certain noun-noun phrasal “compounds” with conjunctive (dvandva) meanings in Vietnamese, which Noyer (1998) argues to be phrasal constituents rather than true compounds; • complex proper names, like Noam Chomsky; • appositives, such as Nicolas II, the last of the Romanoffs. For example, a symmetrical verbal structure is proposed by Butt and Ramchand (2001: 13) for what they call “Type 1 light verb constructions” in Hindi/Urdu. According to them, the event arguments of the two V s combine through the process of telic pair formation with the resulting meaning being that of completion, inception, benefaction, force, suddenness, etc. Similarly, Baker and Stewart (1999) identify three types of Serial Verb Constructions in Èdó that result from the application of Merge via union/intersection at
PHRASE STRUCTURAL RELATIONS AND MERGE
51
different stages of the derivation: covert coordination resulting from Merge via union/intersection of two VoicePs, as in (62a), consequential serial verb construction resulting from Merge via union/intersection of two vPs, as in (62b), and resultative serial verb construction resulting from Merge via union/intersection of two Vs, as in (62c). Crucially for Baker and Stewart (1999), these serial verb constructions are not simple adjunction structures, where one of the verbal projections adjoins to another one of the same kind; rather, these constructions are considered doubly headed. (60)
a.
Òzó [TP ghá [VoiceP [VoiceP gbé èwé] [VoiceP khièn ùhùmwùn érèn]]]. Ozo fut hit goat sell head its “Ozo will kill the goat and sell its head.” TP T°
VoiceP
VoiceP
b.
VoiceP
Òzó ghá [VoiceP [vP [vP gbé èwé] [vP khièn pro]]]. Ozo fut hit goat sell “Ozo will kill the goat and sell it.” VoiceP
νP
Voice°
νP
c.
νP
Òzó ghá [vP gbé [VP èwé [V [V tV ] [V wú]]]]. Ozo fut hit goat die “Ozo will strike the goat dead.” νP ν°
VP Spec
V°
V' V°
Even though I agree with Baker and Stewart (1999) that symmetrical structures can be merged, I disagree with them on a very important point: whether movement is permitted from a symmetrical structure. Baker and Stewart’s analysis is based on the idea that symmetrical structures – which Baker and Stewart call “doubly-headed structures” – are possible only if neither of the input constituents is attracted by a higher head. In other words, for Baker and Stewart, movement from a symmetrical
52
CHAPTER 3
structure is not possible. The tentative explanation that they propose is that the higher head has to find a unique element to attract, but in the case of a symmetrical structure there are two competing elements, and so a unique one cannot be selected. As a result, attraction fails, relevant features are not checked, and the derivation of the symmetrical structure crashes. Note that their analysis is incompatible with even the weaker version of LCA, discussed in the previous section. In contrast, I take the view that movement out of symmetrical structures is not only possible but also necessary. That is, as mentioned above, I adopt the weaker version of LCA, which places it at PF. In particular, in copular sentences of the relevant type (those with the nom-nom pattern in Russian) one of the two DPs moves into the pre-copular position. Two questions arise in this connection. • If there is no unique category attracted by a higher head, what determines which of the two input constituents to a symmetrical Merge (in the case of symmetrical copular sentences, the two DPs) will move? • In the structure below, can the top DP (i.e., DP1 move? If yes, under what conditions? (Numbers in the structure below are for ease of reference only). (61)
DP1 DP2
DP3
In order to answer these questions, we need to consider the issue of what makes the DP move to Spec-TP (i.e., into the position preceding the copula) in the first place. Recall from the previous chapter that I take the movement to Spec-TP to be triggered by the EPP (cf. Lavine 2000, Lavine and Freidin 2002, Bailyn 2004b, inter alia), which is reducible to the need to check an uninterpretable [D]-feature on T ; I take the latter to be related to the referential index (for more discussion of referential indices and their role in syntax see Section 4.3). Thus, only referential DPs, which have a referential index, have the [D]-feature and thus can check the [D]-feature of T . Consider the first of the above questions: if there is no unique category attracted by the higher head (here, T ), what determines which of the two DPs (namely, DP2 or DP3 moves into Spec-TP? In the syntax, either DP is a good candidate for movement: both DPs have a referential index and thus can check the D-feature of T . Furthermore, both DPs symmetrically c-command each other; hence, neither of the DPs is closer to T than the other. Thus, as far as syntax is concerned, the choice of the DP to move into Spec-TP is open. However, whichever DP moves into Spec-TP, it becomes the Topic (cf. also Moro 2000 on “mirror structures” where either of the symmetrical constituents can move in order to resolve symmetry). This is evidenced from the possibility of “flip-flopping” the order of the two DPs in such sentences; for example, in a discussion of a play, one can say either of the following: (62)
a.
Vysotskij byl Gamlet. Vysotsky.nom was Hamlet.nom “Vysotsky was Hamlet.”
b. Gamlet byl Vysotskij. Hamlet.nom was Vysotsky.nom “Hamlet was Vysotsky.”
PHRASE STRUCTURAL RELATIONS AND MERGE
53
Moreover, such sentences are symmetrical in the sense that both the pre-copular and the post-copular phrases are marked with the same morphological case, namely nominative. The only difference between those sentences is in the assignment of topic and focus: in (62a) Vysotskij is the topic and Gamlet “Hamlet” is the focus, whereas in (62b) it is the other way around. Note also that since orders DP copula DP are possible, it is clear that Topic is not a feature that is relevant for syntactic computation. If it were, the two input DPs (i.e., DP2 and DP3 would have to be both Topics or both Foci, given the discussion in the previous section. Clearly, this is not necessarily the case. This leads me to conclude that Topic/Focus is not a feature relevant for syntactic computation. It can be considered to be a feature somehow supplied at the interface level(s), as proposed in Pereltsvaig (2004). In other words, the choice between DP2 and DP3 as candidate for movement into Spec-TP is free in syntax but determines which of the DPs is interpreted as Topic and which as Focus. Now, let us consider the second question: whether the top DP (i.e., DP1 in (61) can move into Spec-TP. Since it is a referential DP just like its constituent parts – DP2 and DP3 – it should be able to check the D-feature of T . It could also be considered closer to T than either DP2 or DP3 . If this is true, then only DP1 should be able to move to Spec-TP blocking movement of either DP2 or DP3 . But this is clearly not the case. Thus, it appears that DP1 cannot move into Spec-TP. How can this be explained? Recall that DP1 is an output of symmetrical Merge (i.e., Merge via union/intersection) and thus a symmetrical structure which cannot be linearized at PF. Only if one of the two input DPs – either DP2 or DP3 – moves into Spec-TP can the structure in (61) be linearized at PF (recall from the discussion of Dynamic Antisymmetry in Section 3.2 above that a trace is invisible to linearization at PF). In other words, DP1 cannot move into Spec-TP because if it did, the structure would uninterpretable (or not linearizeable) at PF. From this, we have to conclude that the order DP DP copula, which is possible in Russian (as well as in Italian but not in Norwegian for V2 reasons) is derived by moving one of the input DPs into Spec-TP and then moving the other DP into Spec-CP.42 (63)
Frejd samyj umnyj byl. Freud.nom most.nom smart.nom was “Freud was the smartest one.”
Similarly, the order copula DP DP must be derived by moving one of the DPs into Spec-TP and then preposing the copula into C (note that like other verb-first clauses, these structures are restricted to the narrative genre; cf. Bailyn 2004b): (64)
Vse ego žaleli, potomu cˇ to byl on samyj nevrednyj i prostoj cˇ elovek. everybody him pitied because was he most benign and simple person “Everybody pitied him because he was the most benign and simple person.” [http://www.periscope.ru/pfin2.htm]
54
CHAPTER 3 3.5.
AGREEMENT IN SYMMETRICAL STRUCTURES
One important issue that arises from my claim that Russian copular sentences with the nom-nom pattern involve a symmetrical structure with the two DPs flanking the copula obligatorily having identical sets of features is this: what features are relevant for this matching relationship?43 It is fairly obvious that we can ignore phonetic and purely semantic features, since those do not play a role in syntactic computation. A more difficult question concerns morphosyntactic features, such as case and -features (including gender, number and animacy).44 Case, at least, does not present a particularly serious problem: since the two DPs appear both as nominatives, it is reasonable to assume that the syntactic specification of their case is the same. In fact, in Section 5.3 below I argue that both instances of the nominative in such sentences correspond to the lack of syntactic case specification and are thus instances of morphological default forms rather than having the nominative checked syntactically. In other words, the specification for syntactic case of the two DPs in such sentences is indeed the same: it is absent altogether. This leaves us with the issue of -features – gender, number and animacy – for which the possibility of a mismatch appear to exist. For example, gender (or declension class) features may be mismatched, for instance, if the pre-copular DP denotes a female individual: with certain nouns the post-copular DP in such cases can be either masculine or feminine; speakers exhibit certain variation as to their preferences, but none admit that either form is completely unacceptable:45 . (65)
Maša – {pobeditel’ / pobeditel’nica} konkursa. Masha winner.m / winner.f competition.gen “Masha is the winner of the competition.”
In some cases, where a distinct feminine form is unavailable, the masculine is the only possible variant:46 (66)
Valentina Ivanova – sekretar’ gorkoma i vernyj tovarišˇc po partii. [Valentina Ivanova].f secretary.m city-committee and loyal comrade at party “Valentina Ivanova is the secretary of the city committee and a loyal party comrade.”
Moreover, some nouns in Russian have only grammatically feminine forms (the so-called nouns of common gender, in Russian sušcˇ estvitel’nye obšcˇ ego roda; cf. Timberlake 2004: 165); these nouns can nevertheless be used as post-copular phrases with masculine pre-copular phrases:47 (67)
Ètot mal’ˇcik – takaja bol’šaja nerjaxa. this.m boy.m such.f big.f sloven.f “This boy is such a big sloven.”
[Comrie and Stone 1978: 77]
Finally, there are examples with a mismatch in grammatical gender, number and animacy of the two DPs:
PHRASE STRUCTURAL RELATIONS AND MERGE
(68)
55
Italjanskie studentki – narod vesëlyj. Italian.pl students.f.pl people.m.sg cheerful.m.sg “Italian (female) students are cheerful people.”
However, I argue here that the problem raised by the above examples is only apparent because what is mismatched in these examples is the grammatical features of the nouns, not the features of the DPs as a whole. I adopt the distinction made in Pereltsvaig (2001a, 2006a) between two types of features previously commonly referred to as -features, namely grammatical features of the head N and the -features of the noun phrase as a whole, that is the DP; cf. also Rappaport (2006), and elsewhere. In what follows, I reserve the term “-features” only for the latter, namely the features of the DP as a whole. These two types of features are relevant for different types of agreement: while agreement within a noun phrase is based strictly on the grammatical features of the head N , agreement of the noun phrase as a whole with elements outside of it (even with non-nominal predicates) operates on the basis of the -features (in the narrower sense defined here). Although the two sets of features often match, they do not always do so, which leads to the possibility of mismatches within a sentence between noun-phrase-internal and nounphrase-external agreement. The nouns involved in such mismatches are of the types mentioned above: nouns that are grammatically masculine but denote females and nouns of the common gender. Such nouns trigger agreement in grammatical gender on attributive adjectives but agreement in -gender on predicates (including verbal ones). This is illustrated below with attested examples involving brigadir “foreman” and skul’ptor “sculptor”, which are grammatically masculine but refer to females in these cases, and kinozvezda “movie star”, which is grammatically feminine but refers to a male here. Examples in (69a-b) are from newspapers, cited in Graudina et al. (1976: 97); example in (69c) is attested on the internet.48 (69)
a.
b.
c.
Naš brigadir naxodilas’ v dekretnom otpuske. our.m foreman(m) was.f in maternity leave “Our foreman was on maternity leave.” [“Komsomol’skaya Pravda”, 17 Feb. 1962] V sentjabre 1920 goda v Moskve pobyvala anglijskij skul’ptor, plemjannica in September 1920 year in Moscow visited.f English.m sculptor(m) niece ˇ cilja Cerˇ Klèr Šeriden. Churchill.gen Claire Sheridan “In September 1920 the English sculptor, Churchill’s niece Claire Sheridan visited Moscow.” [“Izvestija”, 19 March 1961] Filippinskaja kinozvezda Fernando Po mladšij … oficial’no zajavil… Philippine.f star(f) Fernando Po younger.m officially declared.m “The Philippine movie star Fernando Po, Jr. … officially declared…”
The same is true of agreement with relative clauses (in particular the form of the relative operator kotoryj “which”), as can be seen from the following example, where the notorious noun vracˇ (grammatically masculine but often referring to a
56
CHAPTER 3
female, as in the example here) shows masculine agreement on a noun-phraseinternal modifier, but feminine agreement on the relative operator and the verbal predicate inside the relative clause: (70)
Èto edinstvennyj vraˇc, kotoraja menja nastorožila. this sole.m doctor(m) which.f me alarmed.f “That was the sole doctor who alarmed me.” [http://forum.dolgopa.org/index.cgi?read=58835]
Such examples show that even though the subject noun phrases are headed by a noun of a particular grammatical gender, the noun phrase as a whole, that is the DP, need not trigger agreement in the same gender. The subject noun phrases in (69a-b), (70) have the feature specification in (71a) and the subject noun phrase in (69c) has the feature specification in (71b); for ease of presentation only gender and number features are listed here. (71)
DPs headed by nouns of the brigadir-type [-gender: fem] [-number: sg] [gram. gender: masc] [gram. number: sg] b. DPs headed by nouns of the zvezda-type [-gender: masc] [-number: sg] [gram. gender: fem] [gram. number: sg] a.
Note that under this approach grammatical features are specified for the N in the lexicon, while -features are introduced by the D˚, in conjunction with its referentiality index (which, as discussed in Section 4.3 below, is a feature as well; for a slightly different formal implementation of the distinction between two types of features, see Rappaport 2006). This resolves the apparent problem of mismatches in -features between the two DPs flanking the copula in examples in (66)-(68) because the mismatches are not really in -features after all, not in the narrower sense adopted here, but in grammatical features of the noun. Hence, the two DPs do have identical sets of features on the DP level, whereas the differences between them are on the level of NP, but do not percolate further. This can be confirmed by examples of copular sentences where one of the DPs (the one with the “wrong gender”, or another feature) is modified by a relative clause: here it appears that the agreement on the relative operator will be with the “wrong” DP, not the one modified by the relative clause: (72)
Ona vraˇc, kotoraja pri operacii zanimaetsja tem, cˇ to... she(f) doctor(m) which.f at surgery takes-care of such that “She is a doctor who during surgery takes care of...” [http://www.gradusnik.ru/rus/forum/showmsg?topic=forum&msg=3072]
PHRASE STRUCTURAL RELATIONS AND MERGE
57
To recap, the -features of the two DPs flanking the copula need to be identical for this symmetrical structure to be merged, and instances of apparent mismatches involve grammatical features of the nouns and not the true -features of the DPs. 3.6. S U M M A R Y
In this Chapter, I have argued for the following points: • The Antisymmetry Hypothesis does not apply to all levels of representations and all syntactic structures (contra Kayne 1994: 49). • Symmetrical structures are generated by an application of Merge via union (or intersection); contra Chomsky (1995: 246), Kayne (1994), and Moro (2000). • Merge via union (or intersection) is possible only if the two input constituents have identical features. • There are reasons to believe that a number of syntactic constructions can be analyzed as involving symmetrical structures, in particular copular sentences with the nom-nom pattern in Russian are to be analyzed in this way. • With such a conception of Merge, symmetrical structures can be excluded only by an extrinsic stipulation (such as LCA of Kayne 1994). In what follows it will be shown that the equative/identity interpretation of such sentences derives from the claim that such sentences can be merged only via union/intersection, and that for such Merge to apply the two input constituents must have identical features.
This page intentionally blank
CHAPTER 4
THEMATIC RELATIONS
This chapter is concerned with thematic relations in Russian copular sentences. Recall that the main claim of this book is that the two types of Russian copular sentences (identifiable most easily by the case marking on the post-copular phrase) differ in the category of the post-copular phrase and how it interacts with the copula and the rest of the structure. Also, as has been shown in the previous chapters, the relevant categorial type of the post-copular phrase does not depend on its ultimate lexical head, but on the presence or absence of the functional layer; thus, bare NPs pattern with APs and not with DPs, even though both NPs and DPs are projections of a noun. In the previous chapters, I also claimed that the defining property that singles out post-copular DPs is their referentiality, or being associated with a referential index, or being saturated expressions; these notions are further elaborated in the present chapter. Before I proceed with exploring how the two types of Russian copular sentences are interpreted, in Section 4.1 I outline some general issues pertaining to the theory of thematic relations and the assumptions that I make about thematic relations. In particular, I suggest that a theory that reduces thematic relations to structural configurations under Merge (as in Chomsky 1995) is insufficient; instead, I propose to locate thematic relations at the CI (conceptual-interpretative interface), or LF. In Section 4.2, I elaborate a richer theory of thematic relations, originally proposed by Higginbotham (1985) and adopted with certain modifications in this book; this theory – unlike many of its predecessors – allows for three modes of thematic discharge: -marking, -identification, and -binding. In Section 4.3, I examine thematic relations in sentences with the nom-nom pattern (or as argued in the previous chapters, sentences with post-copular DPs) and argue that the identity interpretation of such sentences derives from an interplay of two factors: (i) the requirement that in order for Merge via union/intersection to apply, the two input constituents must have the same feature compositions (as proposed in the previous chapter), and (ii) that referential indices are features. Furthermore, I will propose that DPs need not be -marked (i.e., involved in thematic relations with predicative expressions), but can enter into thematic relations with each other. Section 4.4 is dedicated to thematic relations in sentences with the nom-instr pattern, (or, as argued in the previous chapters, sentences with post-copular NPs and APs). Finally, in Section 4.5, I also argue that the analysis developed in this book accounts for the meaning differences between the two types of copular sentences better than the analyses that reduce the differences to the inherent vs. temporary 59
60
CHAPTER 4
property distinction, and characterize the interpretation of these sentences in terms of stage-level vs. individual-level properties. Section 4.6 summarizes the claims made in this chapter. 4.1.
THETA THEORY: AN OVERVIEW
The difficulty we have had in reaching agreement on just what a theory of thematic roles should look like is analogous to that of the blind men examining the elephant, each touching a different part of its body. –David Dowty, Thematic Proto-Roles and Argument Selection
In The Minimalist Program (Chomsky 1995), thematic relations are reduced to syntactic configurations; in other words, thematic relations are established through Merge (which is a very local operation). In this section, I argue that such a view of thematic relations is unduly simplistic and a much richer theory of thematic relations is appropriate. It is the goal of this chapter to develop such a theory. Consider first the question of the level at which thematic relations are established. Intuitively, thematic relations – being interpretative in nature – should be established at “the same level of representation as the anaphoric relations” (Williams 1989: 454) rather than be part of the “dumb computational system” (the way Chomsky envisages the syntax proper). This idea is not novel within the Minimalist framework either; it is explored at length in Neeleman and Weerman (1999). Here, I will not go through their arguments and the discussion of the consequences of adopting this assumption for the theory of movement; the reader is referred to Neeleman and Weerman’s work for a detailed discussion. The second issue to consider is the nature of thematic relations that the theory is to allow. In standard Minimalist framework (as in the preceding Government and Binding framework) there is only one type of thematic relation: the syntactic configuration that holds between heads and their complements and specifiers. Putting aside quasi-argumental elements such as the weather it, let us define predicative expressions and arguments as follows (cf. Rothstein 1983: 19): (73)
A predicative expression is a category (minimal or maximal) bearing a theta-grid which has not been fully discharged. An argument is a category in a complement or a specifier position of a predicative expression.
According to these definitions, whether an element is a predicative expression can be determined independently of its syntactic environment. On the other hand, whether an element is an argument can be determined only on the basis of its syntactic position; in other words, it is impossible to say whether a given NP or DP is an argument without considering its syntactic position. To recap, argumenthood is an inherently relational notion, but predicativity is not. This seems to be in accord with how these words have been used in the literature although actual definitions (explicit or implicit) may vary.
THEMATIC RELATIONS
61
Note further that according to the definitions above, predicative expressions and arguments are not in complementary distribution. In fact, a category can fall into one of four possible types: • a predicative expression but not an argument (e.g., the verb in The trolls kicked each other); • an argument but not a predicative expression (e.g., the subject DP in John laughed); • neither a predicative expression nor an argument (e.g., an expletive in It seems that Titania sleeps or a vocative in Waiter! There’s a fly in my soup);; • both a predicative expression and an argument. The latter two possibilities may seem peculiar at the first glance, but will be shown to exist in the following sections: in Section 4.3, I develop an analysis of Russian copular sentences with the nom-nom pattern according to which both DPs in such sentences are neither predicative expressions nor arguments and in Section 4.4, I argue that such post-copular NPs and APs are both predicative expressions (i.e., bearing thematic grids that have not been fully discharged) and at the same time also arguments of the copula. Consider now how predicative expressions and arguments can be licensed. Clearly, some principle is required in the grammar to rule out both “dangling predicative expressions” and “dangling arguments”. A predicative expression is considered “dangling” if one or more of its thematic positions are left undischarged, as in (74a):4950 (74)
a. *√ The trolls devoured. b. The trolls devoured the sausages.
An example of a “dangling arguments” is given below: √ (75) a. The troll laughed himself silly. b. * The troll laughed the gnome. Note that these sentences cannot be ruled out purely by Case Theory: himself in (75a) is usually analyzed as receiving a -role from silly (e.g., Schein 1995) and case from laughed. Therefore, the ungrammaticality of (75b) cannot be explained by Case Theory, namely, by the lack of case for the gnome. The obvious alternative is to rule this sentence out by Theta Theory, namely, by the lack of -role for the gnome. In the Government and Binding framework, both “dangling predicative expressions” and “dangling arguments” were ruled out by the Theta Criterion (cited below from Chomsky 1981: 36): the first half rules out “dangling arguments”, whereas the second half rules out “dangling predicative expressions”. (76)
Theta Criterion a. Each argument bears one and only one -role. b. Each -role is assigned to one and only one argument.
62
CHAPTER 4
However, as was noted already in Higginbotham (1985), this formulation of the Theta Criterion is too strong in some cases and too weak in others; I return to this point at the end of the following section, but before that I present a revised version of Theta Theory based on Higginbotham’s work and adopted in this book. 4.2.
A RICHER THEMATIC THEORY: HIGGINBOTHAM (1985) AND MODIFICATIONS
In this book, I adopt a revised version of Higginbotham’s (1985) theory of thematic relations; in what follows I review the main points of this richer theory of thematic relations. The first important point to note is that I take all lexical categories – including verbs, nouns, adjectives, and prepositions – to have theta-grids. In Chapter 2 above, I argued in favor of associating referential indices with the D-layer rather than with the N-layer; thus, N is taken to be predicational, that is associated with a theta-grid, much like a V or an A (as mentioned above, for the most I part ignore prepositions in this book).51 Furthermore, I assume, following Higginbotham that a thematic grid percolates in the syntactic structure along with all other categorial information (cf. cf. Williams’ 1989: 431, Speas 1990: 61 idea that “the external argument … is represented not only on the predicate itself but also on every projection of the predicate”). This brings us to the central issue of how thematic positions are discharged, that is how “open thematic positions in lexical items and in complex phrases” are eliminated (Higginbotham 1989: 475). As noted above, in the standard Minimalist framework there is only one way to discharge thematic positions, that is through Merge. I depart from this and adopt Higginbotham’s system of three basic modes of thematic discharge: -marking, -identification, and -binding.52 For Higginbotham, all three modes of discharge are restricted to the structural configuration of government; however, it turns out that these restrictions need not rely on the notion of government, which has no status in the Minimalist framework. In fact, the notion of government is not restrictive enough in defining the structural configuration for thematic discharge. If the range of configurations in which thematic discharge can occur were defined by government, it would include structures that are clearly not permitted (for a more detailed discussion of this, see below in connection with -binding). Let us now consider the three basic modes of thematic discharge more closely: first, -marking comes to replace -assignment; as a result of a -marking, the thematic position of the -marker is discharged. This process of -marking is instantiated, for example, by a transitive verb which -marks its direct object. Crucially, like other modes of thematic discharge, -marking is restricted to a local syntactic configuration, essentially, the maximal projection of the –marking head. The second basic mode of thematic discharge is -identification, which is involved in the interpretation of simple adjectival modification, as in red balloons.53 Following Higginbotham’s notation, -identification is indicated by
THEMATIC RELATIONS
63
a line connecting two thematic positions indicated by numbers (or s) in angled brackets. Following Higginbotham’s implicit assumption, I take it that -identification discharges the thematic position of the non-projecting element (here, of the adjective); this idea follows from the normal assumptions on the nature of syntactic projection (cf. also Speas 1990: 66–67).54 N' 〈1〉
(77)
A 〈1〉
N 〈1〉
Within the Minimalist framework, this mode of thematic discharge can be restricted in one of the following ways: (i) between two asymmetrically c-commanding heads, as in (78); or between two sisters (whether they are two heads, two maximal projections, or a head and a maximal projection). It is the latter restriction that will be relevant for the discussion of copular sentences later in this book. (78)
XP X°
YP Y°
ZP
The third and final mode of thematic discharge is -binding, which is also restricted to a relation of sisterhood. The core case of -binding considered by Higginbotham (1985: 560) is the DP-internal -binding; this process is at the core of the discussion of Russian copular sentences with the nom-nom pattern in the next section. Following Higginbotham, I assume that an N has a theta-grid, containing (at least) one thematic position which percolates up to the level where the D is merged into the structure. The D acts as the binder of the thematic position of the N ; note that a similar analysis was proposed by Longobardi (1994: 634), who treats the D as an operator that binds a variable, the range of which is defined by the common noun. Here and below, I use an asterisk to indicate -binding (departing from Higginbotham, who uses an asterisk to indicate a thematic position discharged by any of the three modes). The structure below is adapted from Higginbotham (1985: 560) in accordance with Abney’s (1987) DP Hypothesis, as discussed in Chapter 2 above.55 (79)
DP Spec D'〈1*〉 D°
NP〈1〉
the
N'〈1〉 N°〈1〉 dog
64
CHAPTER 4
The crucial property of this theory is that it allows us to distinguish between DPs, which are saturated expressions (i.e., all their thematic positions have been discharged), on the one hand, and non-saturated expressions such as bare NPs and APs, on the other hand. As argued in detail in Chapter 2 above, this is exactly the distinction we need to draw in order to account for the observations concerning copular sentences not only in Russian but also in Italian and Norwegian. Note also that this system is also different in a important way from the one proposed in Williams (1981 and his later work), where a noun’s external/referential -role is conflated with the referential index. According to the theory adopted in this book, a noun or an adjective (and their maximal projections, NP and AP) have thematic grids with undischarged thematic positions, whereas a DP does not have an undischarged thematic position, but has a referential index, as discussed in more detail in the following section. One important consequence of adopting such a richer system of thematic relations is that the original Theta Criterion can no longer be held valid. Higginbotham (1985) proposed a revised formulation of the Theta Criterion, which rules out “dangling predicative expressions” (cf. (74a) above). Yet, this principle is not sufficient to rule out “dangling arguments”, as in (75b) above. As has been mentioned at the end of the previous section, a condition requiring every argument to be assigned one and only one -role (as in the classical statement of the Theta Criterion in (76) above) is too strong; in fact, it is possible to have noun phrases that are -marked twice.56 One potential example of dual -marking involves resultative secondary predicates, such as in The trolls hammered the metal flat, where the noun phrase the metal can be said to receive two -roles: one from the verb hammer and the other from the secondary predicate flat.57 Thus, the possibility of a noun phrase that is -marked twice cannot be excluded.58 Yet, it appears that there is a need for a condition that excludes some noun phrases that are not -marked at all; however, this cannot apply to all noun phrases that are not -marked. Here, I will briefly discuss two constructions from colloquial Russian where noun phrases appear to be licensed although they are not -marked in any obvious way: Left-Dislocation and Pronoun-Doubling (discussed in detail in McCoy 1998).59 In both of these constructions, a noun phrase is licensed despite not being -marked appears to be “parasitic” on another noun phrase (almost always a pronoun), which in its turn receives a -role in the usual way. Some illustrative examples of these constructions are given below. (80)
Policejskie oni veli sebja užasno. Left-Dislocation Policemen.nom they.nom behaved self terribly “As for policemen, they behaved terribly.” b. Žizn’ ona voobš˘ce ne legkaja. Pronoun-Doubling life.nom it.nom usually not easy “Life is usually not easy.” [McCoy 1998: 234] a.
There are two important respects in which these two constructions differ from each other (noted first in McCoy 1998). First, the leftmost phrase in Left Dislocation
THEMATIC RELATIONS
65
(here policejskie “policemen”), but not in Pronoun Doubling (here žizn’ “life”), constitutes an independent prosodic unit (the intonational break is marked here with “”). Second, while the leftmost phrase in both constructions can appear in the nominative case, in Pronoun Doubling but not in Left Dislocation it can also agree in case marking with the intra-clausal “doubled” pronoun. For instance, if an accusative direct object is doubled/dislocated, the leftmost DP in the Pronoun Doubling construction can appear in the accusative case, but in the Left Dislocation construction it always appears in the nominative. (81)
/ *Viktora } ja ego uvažaju. Left-Dislocation Victor.nom / Victor.acc I.nom him.acc respect “As for Victor, I respect him.” {Viktora / Viktor } ego vse uvažajut. Pronoun-Doubling Victor.acc / Victor.nom him.acc everybody.nom respects “Everybody respects Victor.” [McCoy 1998: 235]
a. {Viktor
b.
Importantly for the present discussion, in both of these constructions the leftmost phrase is not -marked independently of the intra-clausal pronoun, but is dependent on the presence of a co-referential DP (usually a pronouns) elsewhere in the clause.60 Therefore, what is needed in the theory of thematic relations is a principle that would rule out non--marked DPs unless they are in a proper relation with another DP. To summarize so far, a need has been established for a condition that would rule out predicative expressions and DPs that do not establish appropriate thematic relations. Specifically, a predicative expression is licensed if and only if all its thematic positions are discharged through one of the following three modes: -marking, -identification or -binding. In other words, a thematic position must either be bound by an appropriate head or establish an appropriate thematic relation with another thematic position (as in -identification) or a DP (as in -marking). A DP is licensed (or can be interpreted) only if it establishes an appropriate thematic relation with a thematic position (when it is -marked) or another DP (which it is co-referential with, as in Left-Dislocation and Pronoun-Doubling in colloquial Russian, discussed above). These possibilities can be summarized into a general formulation of the revised Theta Criterion. (82) Theta Criterion (revised): Every DP and every (unbound) thematic position must receive an interpretation through establishing an appropriate relation with an appropriate element, whether another DP or another thematic position. This formulation rules out both “dangling” thematic positions and “dangling” DPs, and in this way it generalizes over both parts of the original Theta Criterion (as in (76) above). Note that this formulation of the Theta Criterion can be reduced to Full Interpretation; thus, I agree with Brody (1993: 2) that “the -Criterion holds at LF only to the extent required for meaningful interpretation”. Note further that the revised Theta Criterion covers three types of thematic relations:
66
CHAPTER 4
• between a DP and a thematic position, as in -marking; • between two thematic positions, as in -identification; • between two DPs, as in Left Dislocation and Pronoun Doubling constructions. To recap this section, I propose that the Theta Theory module of the grammar is located at LF and comprises the Theta Criterion, as formulated in (82). In informal terms, this Theta Criterion guarantees that there are no dangling -roles or dangling DPs. In the rest of this chapter, I show how these relations apply in copular sentences. In Section 4.3, I develop an analysis of copular sentences with the nom-nom pattern based on the notions of -binding and the thematic relations between two DPs, whereas in Section 4.4 I proceed to examine copular sentences with the nom-instr pattern and propose an analysis of those based on the concepts of -marking and -identification. 4.3.
INTERPRETATION OF COPULAR SENTENCES WITH THE NOM-NOM PATTERN
In this section, I investigate issues related to the interpretation of Russian copular sentences with the nom-nom pattern. The structure I proposed for these sentences is as follows: (83)
TP T'
DPi T°
FP ti
F' F° byt’ DP
DP DPi
In addition, recall from the previous section that DPs are saturated expressions with no undischarged thematic positions. As discussed in Section 2.1 above, being a functional category the copula byt’ “be” does not have a thematic grid in the first place. Thus, the interpretation of such structures cannot proceed through the usual -marking and predication relations. So how are such sentences interpreted? Furthermore, recall that it is the claim of this book that the category of the postcopular phrase (DP vs. NP/AP) determines whether the copula is merged as either functional category (in the case of DP) or as a lexical category (in the case of NP/AP). How is this achieved? In this section I address these issues in detail. Consider first what type of interpretation that such sentences receive. I maintain that a sentence of this type is true if and only if the referent of the pre-copular DP and that of the post-copular DP are identical. For example, the sentence in (84a) is true if and only if “there is someone who is the best cook in town and Griswold is that person” (Fiengo and May 1994: 22). In other words, this sentence asserts that the two DPs Griswold and the best cook in town are co-referential.
THEMATIC RELATIONS
(84)
67
[DP1 Griswold] is [DP2 the best cook in town].
The semantic representation of such copular sentences is given below (I ignore the issue of the representation of tense; the correct temporal interpretation is derived by quantification over or predication of the situation variable s. (85)
a. The king is the culprit. b. ∃s [ x (king (x s & y (culprit (y s & x = y]
The obvious next question is where the coreference comes from. Two approaches have been taken to this issue in the literature. One approach, adopted by Zamparelli (2000), derives the coreference between the two DPs from a last-resort operation in the semantic component. This operation applies to a symmetrical copular structure, like the one proposed in this book, co-indexes the two DPs and maps them onto the identity function (i.e., xy[x=y]). Even though this analysis gets the desired result, it is does not follow from anything else in the theory. Furthermore, it is not clear how semantic coreference relates to syntactic co-indexing, governed by the Binding Theory. Another approach is to derive the coreference between the two DPs on either side of the copula from the special meaning of the copular element itself. This approach has been taken (in various guises) by many previous studies, including such traditional grammarians as Benveniste (1966), Halliday (1967), Higgins (1973), Quirk and Greenbaum (1973), as well as more formal works of Montague (1973), Doron (1983), Bailyn and Rubin (1991), Fiengo and May (1994), Carpenter (1997), to name only a few. Informally, this approach maintains that the copula takes two argument DPs and denotes that they are co-referential. In this approach, a distinction is drawn between two lexical items corresponding to the copula in the two types of copular sentences; the two copulas are said to have distinct meanings. For reasons of parsimony, it is preferable not to split the copula into two distinct lexical items, wherever possible. Furthermore, this approach fails to account for the wide-spread homophony of the identity copula with its predicative counterpart. Finally, it is unclear if this approach can account for the distribution of case-marking in copular sentences – the main empirical problem investigated in this book (for an attempt to account for the case-marking problem within this approach, see Bailyn and Rubin 1991, and for criticism and discussion, see Chapter 5 below). The analysis of the identity interpretation developed in this section avoids the aforementioned problems: specifically, I propose to derive the coreference from syntactic co-indexing (as in Fiengo and May 1994), which is itself forced by the properties of the syntactic structure, in particular, the way it is merged. Therefore, the identity interpretation in my analysis is not assigned to these sentences in an ad hoc fashion, which is a significant advantage of this analysis over its rivals. Consider the distinction between co-indexing and coreference drawn originally by Fiengo and May (1994). Following their work, I take co-indexing as a syntactic relation which is subject to syntactic conditions (such as Binding Theory), and coreference as a semantic relation. Crucially, there is no one-to-one correspondence
68
CHAPTER 4
between co-indexing and coreference: lack of co-indexing does not entail lack of coreference because in addition to syntactic co-indexing, coreference can derive from extra-syntactic information. For example, coreference can be derived from pragmatics, as in the following example from Fiengo and May (1994: 3), originally due to Higginbotham (1985: 569–570). Suppose that Kim sees a man leaving the room, but she cannot see his face. She asks Sandy who that person is, and Sandy replies: (86)
I don’t know, but he put John’s coat on.
Here is an apparent paradox: according to Principle C of the Binding Theory (which requires R-expressions to be free, i.e., unbound), he and John in the above example cannot be coindexed (or in Evans’ 1980 terminology “referentially dependent”). However, in the reading that Sandy implies, given the extra-linguistic context, he and John refer to the same person (or to use Evans’ terminology, he and John must be “coreferential in the extensional sense”). However, this coreferentiality can be derived from the application of pragmatic principles: assuming that Sandy is being cooperative, Kim can deduce that first, in accordance with Grice’s (1975) Maxim of Quality, Sandy does not want to commit herself to a statement which she does not have sufficient evidence to support, namely that the person who has left the room is John, and second, assuming that Sandy obeys the Maxim of Relevance, one can deduce that the person who has left the room is John (otherwise, the information that the person who has left has put on John’s coat is irrelevant in the conversation). Thus, the distinction that Fiengo and May draw between co-indexing and coreference solves this apparent paradox: Principle C of the Binding Theory governs the syntactic relation of co-indexing but says nothing about coreference. (87)
Principle C: R-expressions must be free everywhere. Free = not bound
(88)
Binding (standard definition, to be changed below) binds iff (i) and are coindexed, and (ii) c-commands .
As for the copular sentences of the relevant type, Fiengo and May (1994) take the approach that distinguishes a special identity copula. Thus, for them the coreference between the two DPs flanking the copula (i.e., the identity interpretation) comes not from syntactic co-indexing but from the semantics of the copula itself. Although this allows to avoid any potential violations of Principle C, as well as makes such identity sentences informative (since the indices of the two DPs differ, the sentence asserts that X = Y and not X = X). There are, however, two serious objections to Fiengo and May’s analysis (in addition to those mentioned above in connection with the “identity copula” approach in general). First, if the copula is said to take the two DPs as its arguments, it must -mark them; however, as Rapoport (1987: 139–140) has noted, “it is not exactly clear which theta-roles the copula would assign, and how the identity relation would
THEMATIC RELATIONS
69
be derived from that assignment” (cf. also Heggie 1988). In essence, relegating the coreference requirement to the meaning of the copula does not explain why the identity copula appears with two DPs, whereas a different, non-identity copula appears when the post-copular phrase is an NP or an AP; this is exactly the generalization this book attempts to explain. Second, the “identity copula” approach requires the two DPs to be arguments of the copula, which means that the copula is a lexical category. Yet, in the structure I propose for these copular sentences, the two DPs are not arguments of the copula. In what follows, I adopt Fiengo and May’s distinction between co-indexing and coreference, but reject their “identity copula” analysis as far as Russian copular sentences are concerned. In contrast to Fiengo and May, I claim that coreference between the two DPs in these copular sentences is derived from syntactic co-indexing. Moreover, I show that an analysis that derives coreference in Russian copular sentences with the nom-nom pattern without introducing a separate lexical item (i.e., “identity copula”) is preferable to Fiengo and May’s analysis. Given my claim that the coreference between the two DPs flanking the copula in the sentences under consideration here derives from syntactic co-indexing, it is necessary to consider the relation of co-indexing more closely. Since co-indexing is a syntactic relation, referential indices must be present in syntax. But if we assume with Chomsky (1998: 116) that syntactic computation can deal only with features, we must conclude that a referential index is a feature.61 This is exactly what I claim a referential index to be: a feature which enters into syntactic computation from the lexicon through a numeration, much like all other features. Furthermore, I propose that the referential index is a feature of D ; this is a development of my earlier claim (cf. Chapter 2 above) that referentiality is associated with the functional D-layer rather than with the lexical N-layer. It should also be noted that referential index is understood here as an independent feature, rather than as a bundle of other features, such as [gender], [number], and [person], as is assumed in Pollard and Sag (1994); see also Svenonius (1993a). An interesting question that arises in this connection is about the number of possible referential indices. It appears that natural language would require a potentially infinite number of referential indices in order to express reference to a potentially infinite number of entities in the world. However, it is obvious that the lexicon cannot contain an infinite number of referential indices, for the obvious space limitation reasons. In order to resolve this paradox, I propose that the number of referential indices is curbed by processing limitations. It is never possible to set up an infinite number of referents in any given discourse; thus, the number of referential indices used in a given discourse is limited by processing and memory limitations. Therefore, there is no need for the lexicon to contain an infinite number of referential indices. Rather, the lexicon must contain a limited number of referential indices, corresponding to the maximal number of referential indices that can be used in any given discourse. In other words, the lexicon need not contain a separate index for each and every entity in the world; rather, the same index can be used for different entities in different discourses. For example, the same index
70
CHAPTER 4
can encode reference to John in one discourse and to Mary in another discourse. Thus, the number of referential indices in the lexicon is not infinite, though I will not venture to estimate their exact number. Interestingly, this analysis of referential indices as features in the syntax is further supported by studies of sign languages (such as ASL/American Sign Language and ISL/Israeli Sign Language). In sign languages, entities are represented in discourse through use of space. According to MacLaughlin (1997: 58–59), a referential entity can be associated with a location in the signing space, and this location is then used during the discourse to refer to that entity. These spatial loci are relevant to a variety of grammatical processes, including pronominal reference, possessive relations, and agreement. These spatial loci can be used for both manual and non-manual marking; the former includes pointing with an index finger or a thumb to encode pronominal reference, pointing with an open hand-shape to encode possessive reference, and modifying the starting point and the end point of movement of some verbal signs to encode agreement. The use of spatial loci with non-manual marking involves head tilt and eye gaze directed towards the relevant location. According to MacLaughlin (1997: 59; see also references cited therein), “spatial loci reflect the abstract grammatical feature person”. However, the use of spatial loci in sign language is different from the expression of the grammatical feature [person] in spoken languages in that the latter distinguish only among first, second, and third persons, whereas sign language is capable of distinguishing a much larger number of spatial loci. Instead, I suggest that spatial loci in sign language can be analyzed as overt realizations of referential indices. As with referential indices, the number of potentially distinguishable spatial loci is not infinite and is curbed by processing and memory limitations. Moreover, spatial loci serve to keep track of entities in the discourse, which is exactly what referential indices do. Another suggestive fact is that “only specific noun phrases may be associated with a location in space; non-specific noun phrases (which do not assert existence of any referent) cannot be established in space” (MacLaughlin 1997: 137).62 The use of spatial loci as expressions of referential indices provides further support for the claim that referential indices are present in the syntax (and therefore, in the lexicon), and against the alternative view that referential indices are present only in semantic representations (i.e., outside of the domain of computational syntax). Since spatial loci are involved in articulation of the index sign, they must be accessible to the PF component. According to Chomsky’s Inclusiveness Condition (1995: 228, 2000: 113–114), “no new features are introduced by CHL ” (Chomsky 2000: 113). Chomsky himself claims that the Inclusiveness Condition “rules out… indices”, but remarks that “questions arise if they enter into interpretation and function significantly within the computation” (Chomsky 2000: 114). My answer to this question is affirmative. Given that referential indices are accessible at PF (at least in sign language), they must be present in the syntax; then, they are available at both PF and LF interfaces. To recap,
THEMATIC RELATIONS
71
sign languages provide additional support for the claim that referential indices are features involved in the syntactic computation. If we assume that a referential index is a feature, we must conclude that it is subject to the same restrictions that apply to other features. Recall from Section 3.3 that in order for Merge via union/intersection to apply, the two input constituents must have identical features. Therefore, if an index is a feature, the two input DPs in a symmetrical structure must have identical indices (i.e., they must be coindexed). Thus, according to my analysis the following (co-)indexing obtains: (89)
DPi DPi
DPi
To summarize so far, I argued that the coreference between the two DPs (i.e., the identity interpretation) derives from two facts: (i) the two input constituents in a symmetrical structure, here DPs, must have identical features, and (ii) a referential index is a feature. This allows me to dispense with the idea that such sentences contain a special “identity copula” with a separate meaning, unrelated to that of the non-identity copula. However, the analysis that derives the coreference of the two DPs flanking the copula from syntactic co-indexing raises the question of potential violations of Principle C of the Binding Theory. As has been mentioned above, the standard definition of Principle C requires R-expressions to be free (i.e., not bound). However, if the two DPs in the symmetrical structure are coindexed, at least one of them is not free. If Principle C applies to lower copies (i.e., base positions), as in (90a), both DPs c-command and therefore bind each other. If Principle C applies to higher copies (i.e., surface positions), as in (90b), the pre-copular DP (which appears in Spec-TP position) binds the post-copular DP. (90)
a. Principle C applies to base positions: … FP F' DP
F°
byt’ [DP]1
[DP]1
b. Principle C applies to surface positions: FP [DPi]1 F°
F' DP
byt’ [DP]1
DP ti
72
CHAPTER 4
In order to avoid these Principle C violations, I propose to amend the definition of binding. The intuition behind this amendment is that binding is essentially an asymmetric relation. If the grammar is assumed to allow only asymmetrical structures (as is the case in Kayne 1994 and much subsequent literature), the traditional definition of binding suffices. However, if symmetrical structures are to be allowed in the grammar (as I have argued in chapter 3), these constructions need to avoid mutual binding by the two symmetrical phrases.63 (91)
Binding (revised definition) binds iff (i) and are coindexed AND (ii) at least one copy of c-commands at least one copy of AND (iii) the lowest copy of and the lowest copy of do not mutually c-command each other.
Note that this revised definition of binding does not undermine the empirical coverage of Principle C (or of the other Principles) in the standard Binding Theory, so there is no loss of empirical coverage. The only kind of structure where the standard and the revised definitions have different results is the one involving symmetrical structures, in which allowing apparent violations of Principle C is a desirable result. The following examples illustrate how the revised binding theory works. In asymmetric structures, binding is established as with the old definition of binding. For example, in John and Mary seem to each other to be stupid the anaphor each other (i.e., is bound by the subject John and Mary (i.e., because and are coindexed, at least one copy of the antecedent (here, the highest copy in the specifier of matrix TP) c-commands the anaphor, and the lower copies of both the anaphor and the antecedent (here, the copy in the small clause) do not mutually c-command each other. Thus, the necessary binding relations are established, and the sentence is grammatical. In contrast, a symmetrical structure, such as John is the culprit, contains two R-expressions which must be free. It is the claim of this chapter that the two DPs here are coindexed. Moreover, at least one copy of John (namely, the copy in Spec-TP) c-commands at least one copy of the other DP, the culprit. However, clause (iii) of the revised definition guarantees that there is no binding here. This is so because the lower copy of John and the lower copy of the culprit (namely, the copies in the position of merger) do mutually c-command each other. Therefore, John does not bind the culprit in this sentence. (Equally, the culprit does not bind John either.) Thus, an apparent violation of Principle C is avoided. Furthermore, consider anaphor binding in symmetrical structures. What makes sentences like *Himself is John ungrammatical? According to Principle C the R-expression John must be free, and it is: the anaphor himself and the R-expression John are coindexed, and at least one copy of himself (the one in the Spec-TP) c-commands at least one copy of John. However, the lower copies of himself and John do mutually c-command each other; therefore, himself does not bind John.
THEMATIC RELATIONS
73
As a result, John is free, thus satisfying Principle C. However, Principle A is not satisfied in this sentence, which is the cause of its ungrammaticality. According to Principle A, the anaphor himself must be bound. The only element it can be bound by in this sentence is John. Even though the anaphor himself and its potential antecedent John are coindexed and at least one copy of John c-commands at least one copy of himself (the one in the position of merger); clause (iii) of the revised definition of binding is not satisfied: the lower copies of himself and of John do mutually c-command each other which means that John cannot bind himself. As a result, the anaphor is not bound, violating Principle A. This results in the ungrammaticality of this sentence. The next obvious question involves the grammaticality of the following sentence: (92) John is himself. If this sentence is a “mirror image” of the ungrammatical sentence *Himself is John, it should be ungrammatical as well. Since the lower copies of John and himself mutually c-command each other, clause (iii) of the revised definition of binding is not satisfied and binding does not apply here, leaving the anaphor free, in violation of Principle A of the Binding Theory. So why is this sentence grammatical? My answer is that the sentence in (92) is not a copular sentence of the relevant type: it does not have a symmetrical structure and it does not have the identity reading asserting that John and himself refer to the same person. Rather, it has two possible predicational readings: (i) John is asserted to be (or behave like) his usual self, and (ii) John plays himself (as in a theatrical context). Two facts support this contention. First, as mentioned in Fiengo and May (1994: 35), the negated version of (92) – John is not himself – is not contradictory. Second, note that Russian counterpart of (92) is possible only with the instrumental marking on the anaphor.64 (93)
a.
Ivan byl samim soboj. Ivan was self.instr self.instr “Ivan was himself.” b. *Ivan byl sam {soboj / sebja}. Ivan was self.nom self.instr / self.acc intended: “Ivan was himself.”
Thus, I maintain that the grammaticality of John is himself does not contradict the revised definition of binding proposed here, which excludes the possibility of mutual binding in symmetrical structures. This revised definition relies on the intuition that binding is essentially asymmetric. Therefore, even though the two input DPs in symmetrical copular sentences are said to be coindexed in syntax (and not simply semantically coreferential), they do not violate Principle C of the Binding Theory. To summarize, I have proposed that Russian copular sentences with the nom-nom pattern are interpreted as follows: the thematic positions in the nouns’ grids are
74
CHAPTER 4
discharged by D s through -binding; DPs are saturated expressions, meaning that they have no undischarged thematic position. Instead, they have referential indices, introduced into the computation by D s; furthermore, I take referential indices to be features (evidence from sign language, discussed above, supports this claim). The claim that referential indices are features in conjunction with the requirement that the two input constituents for Merge via union/intersection have exactly the same features (as long as these features are relevant for computation, of courses) means that the two DPs flanking the copula must have identical referential indices (i.e., they must be syntactically coindexed). This co-indexing in turn leads to the observed semantic coreference, namely to the identity interpretation. Recall that this analysis does not require me to postulate a separate identity copula thus allowing me to avoid a host of problems associated with that approach. Instead, it derives the correct interpretation purely from independently necessary syntactic principles. In Section 4.5, I re-examine the differences in meaning between the two types of copular sentences in Russian and show that the analysis developed in this section for sentences with the nom-nom pattern and the analysis developed in the following section for sentences with the nom-instr pattern provide the best analysis for the often subtle meaning differences between the two sentence types. 4.4.
INTERPRETATION OF COPULAR SENTENCES WITH THE NOM-INSTR PATTERN
In this section, I develop an analysis of Russian copular sentences with the nominstr pattern. Recall from Chapter 2 that I propose that the post-copular phrase in such sentences is a bare NP or an AP, and the copula is merged as a lexical category v rather than a functional category F (although it raises to F like all verbs in Russian). The structure proposed for such sentences is repeated below. (94)
TP DPi
T' T°
FP ti
F' F°
vP
byt’ ti
v' v°
NP/AP
tbyt’
As discussed above, I assume that since nouns and adjectives are lexical categories, they have thematic grids and the positions in these grids need to be discharged (cf. Section 2.1 above). When an NP appears as a complement of a D in a DP, the
THEMATIC RELATIONS
75
NP’s thematic position is discharged through -binding (following Higginbotham 1985), but how is the thematic position of N (or A ) discharged in the absence of a D˚. In this section I propose the following: • nouns and adjectives are “defective -markers” in the sense that they fail to -mark their specifiers directly (cf. Baker and Stewart 1997 and Baker 2003); • the copula in such sentences is merged as a lexical head (i.e., it bears a theta-grid); • the thematic position of a noun or an adjective is discharged to the specifier of the small clause, with the “help” of the copula; • this process consists of two steps: -identification and -marking; • this is achived through distinguishing between two components of a thematic position: an argumental variable and a thematic index (following Samek-Lodovici 2003). These points are examined in detail below. 4.4.1.
Adjectives and Nouns as Deficient -Markers
First, consider the claim that nouns and adjectives are “defective -markers”. Baker and Stewart (1997) and Baker (2003: 23–94) proposed that it is the defining property of verbs that they can -mark their specifiers directly; other lexical categories, namely nouns and adjectives, cannot. Furthermore, they define nouns as the lexical category bearing a referential index; as explained in Sections2.4 – 2.6 and 4.3 above, I do not adopt this assumption but rather treat referential indices as features of D . Baker and Stewart’s main evidence comes from their investigation of adjectives with respect to unaccusativity diagnostics.65 In particular, Baker and Stewart show that adjectives differ from unaccusative verbs in Èdó (a Kwa language spoken in Nigeria) in that de-verbal nominalizations (derived from verbs by a productive ù-H!H-mwèn circumfix, where “H!H” indicates high-downstep-high tones on the root) allow an overt argument, whereas de-adjectival nominalizations (derived from adjectives by adding a vowel prefix) do not allow an overt argument (examples from Baker and Stewart 1997: 42):66 (95)
a.
b.
deverbal nominalization ù-mó!sé-mwèn óghé èmèrí nomin-beautiful-nomin of Mary “Mary’s beautifulness” deadjectival nominalization ì-mòsè (*óghé èmèrí) nomin-beautiful (*of Mary ) intended: “Mary’s beautifulness”
[Èdó]
Another set of evidence that Baker and Stewart (1997) and Baker (2003: 62–77) use to support their claim that adjectives are defective -markers involves more familiar unaccusativity diagnostics from several languages, including Italian, Mohawk, Hebrew, and Japanese.67 For instance, possessor datives in Hebrew have been used
76
CHAPTER 4
as a diagnostics for the unaccusative/unergative distinction: as has been first noted by Borer and Grodzinsky (1986), a dative phrase can be interpreted as a possessor of the subject of an intransitive verb only if the verb is unaccusative.68 For example, the verb in (96a) is unaccusative, and the dative clitic can be interpreted as identifying a possessor; in contrast, the verb in (96b) is unergative, and the dative clitic cannot be interpreted as identifying a possessor, resulting in the ungrammaticality of this sentence (examples from Borer and Grodzinsky 1986). (96)
a.
Ha-kelev neʔelam le-Rina. the-dog disappeared to-Rina “Rina’s dog disappeared.” b. *Ha-kelev ritrocec le-Rina. the-dog ran-around to-Rina intended: “Rina’s dog ran around.”
[Hebrew]
As with unergative verbs, in the case of an adjectival predicate, a dative noun phrase cannot be interpreted as a possessor. This is true of both simple adjectives and adjectival (i.e., lexically derived) passives (in contrast to verbal passives). (97)
a. simple adjective *Ha-simla hayta levana le-Rina. the-dress was white to-rina intended: “Rina’s dress was white.” b. adjectival passive *Ha-matana hayta munaxat le-Rina betox kufsa. the-present was placed.A-pass to-Rina inside box intended: “Rina’s present was placed inside a box.” c. verbal passive Ha-matana hunxa le-Rina betox kufsa. the-present place.pst.V-pass to-Rina inside box “Rina’s present was placed inside a box.”
[Hebrew]
Similarlarly, ne-cliticization in Italian – which is possible with unaccusative but not unergative verbs (cf. Burzio 1986) – shows that adjectives behave like unergatives.69 For instance, a verbal passive allows ne-cliticization, whereas an adjectival passive does not (the examples are from Burzio 1986: 30–31). (98)
a.
b.
verbal passive Ne sarebbero riconosciute molte. of-them would-be recognized many “Many of them would be recognized.” adjectival passive * Ne sarebbero sconosciute molte. of-them would-be unknown many intended: “Many of them would be unknown.”
[Italian]
77
THEMATIC RELATIONS
Furthermore, (most) simple adjectives do not allow ne-cliticization either:70 (99)
* Ne sono {intelligenti / rossi / bianchi} molti. of-them are intelligent / red / white many intended: “Many of them are {intelligent /red/white}.”
[Italian]
We can further supplement Baker and Stewart’s evidence by the following fact: verbs synonymous with simple adjectives do allow ne-cliticization (in contrast to adjectives that they are synonymous with). Although there is some interspeaker variation in this respect, most speakers accept the following sentences as grammatical:71 (100)
Ne sono {rosseggiati / biancheggiati} molti. of-them are being-red / being-white many “Many of them are {red / white}.”
[Italian]
On the basis of these data from Èdó, Hebrew and Italian, Baker and Stewart (1997) and Baker (2003) conclude that adjectives project their highest argument outside their maximal projection (similarly to unergative rather than unaccusative verbs); for them the argument is merged in the Spec-PredP, where Pred is the head hosting the copula (recall that I claim that the copula in such sentences is a lexical rather than functional head and label it v ). In what follows, I examine Russian data concerning various unaccusativity diagnostics and show that the conclusion reached by Baker and Stewart (1997) and Baker (2003), namely that the highest argument of adjectives is merged (and importantly, -marked) outside the maximal projection of the adjective, is warranted for Russian as well. Furthermore, I show that in this respect adjectives contrast sharply with synonymous verbs (some examples of such synonymous pairs oven triplets between verbs, adjectives and nouns are given in the Table 2 below). The goal of this comparison is to exclude the interference of the semantics of the predicates in question and to show that the differences in behavior with
Table 2. Synonymous stative predicates in Russian Intransitive verb
Short-form adjective
Noun
bolet’ “to be ill” bedstvovat’ “to be poor” pjanstvovat’ “be drunk” belet’ “to be/appear white” krasnet’ “to be/appear red” cˇ ernet’ “be black” želtet’ “be yellow” zelenet’ “be green” sinet’ “be blue” golodat’ “to be famished”
bolen “ill” beden“poor” pjan “drunk” bel “white” krasen “red” cˇ ëren “black” žëlt “yellow” zelen “green” – goloden “hungry”
bol’noj “patient” bednjak “pauper” pjanica “drunkard”
78
CHAPTER 4
respect to the unaccusativity diagnostics cannot be reduced to the differences in the thematic properties of the arguments compared. Finally, I also show that nouns pattern with adjectives in that they too fail to -mark their specifiers.72 Based on these observations, I draw the conclusion that structure for adjectival and nominal predicates is the one in (101a) rather than (101b). (101)
b. AP/NP
a. vP DP
v' v°
AP/NP
DP
A'/N' A°/N°
…
…A°/N°…
Borovikoff (2001) lists the following five unaccusativity diagnostics in Russian: • • • • •
genitive of negation; distributive po- with implicit or explicit odin ”one”; genitive initial sentences; first conjunction agreement; aspectual prefixation with na-, pere- and po-.
Due to its verbal nature, the last of these tests cannot be used for the purposes of the present discussion. To the other four tests, we can add possessor datives, which exist in colloquial Russian. Let us consider these tests more closely. The first test is the genitive of negation. As is well-known since Chvany (1975) and Pesetsky (1982), the genitive of negation, namely the rule whereby nominative/accusative nominals appear in the genitive in the scope of negation, applies to subjects of unaccusative verbs, as in (102), but not to subjects of unergative verbs, as in (103):73 (102)
a.
Griby zdes’ rastut. mushrooms.nom here grow “Mushrooms grow here.” b. Gribov zdes’ ne rastet. mushrooms.gen here not grow “No mushrooms grow here.”
(103)
a.
Mal’ˇciki zdes’ tancujut. boys.nom here dance “Boys dance here.” b. *Mal’ˇcikov zdes’ ne tancuet. boys.gen here not dance intended: “No boys dance here.”
If we compare adjectives with synonymous intransitive verbs, we observe that verbs allow the genitive of negation, while adjectives do not. From this we can conclude that these verbs are unaccusative, while adjectives are unergative.
THEMATIC RELATIONS
(104)
79
a. *(…ˇctoby) ni odnogo rebënka nikogda ne bylo {bol’ny / bol’no} that not one.gen child.gen never not was ill.3.pl / ill.3.sg.n intended: “…that no child would ever be ill” b. (…ˇctoby) ni odnogo rebënka nikogda ne bolelo that not one.gen child.gen never not be-ill.3.sg.n “…that no child would ever be ill” c. * Na gorizonte ne bylo belo ni odnogo parusa. on horizon not was white.3.sg.n not one.gen sail.gen intended: “Not a single sail was white on the horizon.” d. Na gorizonte ne belelo ni odnogo parusa. on horizon not was-white not one.gen sail.gen “Not a single sail was white on the horizon.” e. * V dorevoljucionnoj Rossii ne bylo bedno ni odnogo krest’janina. in pre-revolutionary Russia not was poor.3.sg.n not one.gen peasant.gen intended: “In czarist Russia no peasants were poor.” f. V dorevoljucionnoj Rossii ne bedstvovalo ni odnogo krest’janina. in pre-revolutionary Russia not was-poor not one.gen peasant.gen “In czarist Russia no peasants were poor.”
Note further that some intransitive verbs in Russian, which are “cognate” with adjectival (or nominal) predicates, have an agentive meaning “make oneself appear X”. Examples include umnicˇ at’ “be/play a wise guy” (cf. the adjective umën “smart”), važnicat’ “make oneself appear self-important” (cf. the adjective važen “important”), duracˇ it’sja “to monkey about” (cf. the noun durak “fool”). As expected, these agentive verbs have an external rather than an internal argument. Not surprisingly, none of these verbs allow the genitive of negation. (105)
a.
*Ni odnogo studenta ne umniˇcalo. not one.gen student.gen not played-wise-guy intended: “No student played a wise guy.” b. *Ni odnogo studenta ne važnicalo. not one.gen student.gen not made-self-appear-important intended: “No student made himself appear self-important.” c. *Ni odnogo studenta ne duraˇcilos’. not one.gen student.gen not monkeyed-about intended: “No student monkeyed about.”
Moreover, some intransitive verbs are ambiguous between stative/exitential and agentive readings. Two examples of such verbs are: • golodat’ – related to the adjective goloden “hungry” – which can mean either “be famished” (on a stative reading) or “fast” (on an agentive reading); • krasovat’sja – related to the adjective krasiv “beautiful” – which can mean either “appear beautifully” (on a existential reading) or “show off” (on an agentive reading). As expected, these verbs allow the genitive of negation only on the stative/existential reading, but not on the agentive one.74
80 (106)
CHAPTER 4
a.
(…ˇctoby) ni odnogo rebënka ne golodalo that not one.gen child.gen not hunger.3.sg.n “…that no child would be famished.” NOT: “… that no child would fast.” b. Ni odnogo buketika na stole ne krasuetsja. not one.gen small-bouquet.gen on table not appears-beautifully “Not a single little bouquet appears on the table.” [http://vanil-la.livejournal.com/2006/03/07/] c. *Ni odnogo studenta ne krasovalos’ pered devuškami. not one.gen student.gen not showed-off in-front girls intended: “No (male) student showed off in front of the girls.”
To recap, I have shown that agentive/unergative verbs in Russian do not allow the genitive of negation on their subjects, while stative/existential/unaccusative verbs do (whether they are related derivationally to adjectives or not). Furthermore, I have shown that adjectives pattern with unergative rather than unaccusative verbs, even when they are synonymous with the latter. What about NP predicates? As expected from the discussion so far, these predicates pattern with AP predicates and hence with unergative verbs rather than with unaccusative verbs: they do not allow the genitive of negation. (107)
a.
b.
*(…ˇctoby) ni odnogo rebënka nikogda ne bylo bol’nym. that not one.gen child.gen never not was ill.3.sg.instr intended: “…that no child would ever be {ill / patient}” *(…ˇctoby) ni odnogo rebënka nikogda ne bylo uˇcenikom. that not one.gen child.gen never not was pupil.instr intended: “…that no child would ever be a pupil.”
The second unaccusativity test involves the distributive preposition po, first mentioned as such by Pesetsky (1982) and discussed in great detail in Borovikoff (2001): unaccusative verbs allow subjects with the distributive po, while unergatives verbs do not:75 (108)
a.
Na každoj stene viselo po kartine. on each wall hung.sg.n distr picture.dat.sg “On each wall there was a picture hanging.” b. *Po každoj dorožke {bežalo / bežali} po sportsmenu. along each lane ran.sg.n / ran.pl distr sportsman.dat intended: “One sportsman was running along each lane.”
As with the genitive of negation, AP and NP predicates pattern with unergatives rather than with unaccusatives in that they do not allow subjects with the distributive po (the status of the ungrammatical examples below does not improve with changes in word order): (109)
a.
V každom klasse bolelo po rebënku. in each class were-sick.sg.n distr child.dat “One child in each class was sick.”
THEMATIC RELATIONS
b.
c.
81
*V každom klasse {bylo bol’no / byli bol’ny} po rebënku. in each class was sick.sg.n / were sick.pl distr child.dat intended: “One child in each class was sick.” *V každom klasse {bylo otliˇcnikom / byli otliˇcnikami} po rebënku. in each class was honor-roll-pupil / were honor-roll-pupils distr child.dat intended: “One child in each class was an honor roll student.”
This supports my claim here that AP and NP predicates pattern with unergative rather than unaccusative verbs in the sense that their subjects are merged and -marked outside the maximal projection of the predicative head. The third test involves the so-called genitive initial sentences, discussed in Babby (1980: 47–48), House (1982), Franks and House (1982), Franks (1995), Pereltsvaig (1998) and Borovikoff (2001). Once again, the construction is possible with subjects of unaccusative verbs but not with subjects of unergatives, and stative verbs pattern with unaccusatives: (110)
a.
Fil’mov idët vsego tri. films.gen go.sg.n only three “As for films, there are only three on show.” [Franks 1995: 191] b. *Ženšˇcin zagovorilo dve. women.gen.pl talked.sg.n two intended: “The number of women who started to talk was two.” [Crockett 1976: 359] c. Detej bolelo vsego troe. children.gen was-sick.sg.n only three “The number of children who were sick was only three.”
In line with the discussion above, AP and NP predicates pattern with unergatives and do not allow the genitive initial construction: (111)
a.
*Detej { bylo bol’no / byli bol’ny } vsego troe. children.gen was sick.sg.n / were sick.pl only three intended: “The number of children who were sick was only three.” b. *Mal’ˇcikov {bylo otliˇcnikom / byli otliˇcnikami} vsego troe. boys.gen was honor-roll-pupil / were honor-roll-pupils only three intended: “The number of boys who honor roll students was only three.”
The fourth test to be considered here is the first conjunct agreement, discussed in Babyonyshev (1996) and Borovikoff (2001): unaccusative verbs but not unergative verbs allow agreement with the adjacent first conjunct (examples from Borovikoff 2001).76 (112)
a. b.
Po poverxnosti {plavali / plavalo} brevno i suˇc’ja. on surface floated.pl / floated.sg.n log.nom.sg.n and branches.nom.pl “A log and some branches were floating on the surface.” V poljax {zreli /*zrel } kartofel’ i pomidory. in fields ripened.pl /*ripened.sg.m potatoes.sg.m and tomatoes.pl “In the fields potatoes and tomatoes were ripening.”
82
CHAPTER 4
As with the previous three tests, AP and NP predicates pattern with unergative rather than with unaccusative verbs with this respect as well: (113)
a. b.
Grippom {byli bol’ny / *byl bolen } mal’ˇcik i devoˇcka. Flu.instr were sick.pl / * was.sg.m sick.sg.m boy. sg.m and girl.sg.f “A boy and a girl were sick with flu.” { Byli otliˇcnikami / *Byl otliˇcnikom } mal’ˇcik i devoˇcka. were.pl honor-roll-pupils / *was.sg.m honor-roll-pupil boy.sg.m and girl.sg.f “A boy and a girl were honor roll students.”
The last test is possessor datives (as mentioned above, they have a colloquial flavor in Russian); this test too shows that there is a contrast in unaccusativity between adjectival predicates and intransitive verbs. As discussed for Italian above, Russian possessor datives are possible with unaccusative predicates but not with unergative predicates (note that the presence of the intransitivizing morpheme -sja does not play a role in determining the grammaticality of these examples). (114)
a.
Korzinka ej sovsem razvalilas’. basket her.dat altogether fell-apart “Her basket fell apart.” b. *Sobaka ej kusaetsja. dog her.dat bites intended: “Her dog bites.”
Once again, the contrast is observed between adjectival and nominal predicates, on the one hand, and (synonymous) intransitive verbs, on the other hand. (115)
a.
Deti ej nikogda ne boleli. children her.dat never not were-sick “Her children have never been sick.” b. *Deti ej nikogda ne byli bol’ny. children her.dat never not were sick intended: “Her children have never been sick.” c. *Deti ej byli studentami. children her.dat were students.instr intended: “Her children were students.”
The results of the unaccusativity tests discussed above are summarized in Table 3 below. To summarize so far, I have shown that adjectives project and -mark their highest argument externally to their maximal projections, whereas synonymous unaccusative verbs project and -mark the same argument internally to their maximal projections. Since these three types of predicates – verbal, adjectival and nominal – are very close in meaning, it is reasonable to assume that the thematic position discharged to their subjects is the same. For a want of a better term, I will call this thematic position Embodiment; for example, in the following examples Ivan embodies the state/property of being poor:
83
THEMATIC RELATIONS Table 3. Unaccusativity diagnostics in Russian for three types of predicates Diagnostics genitive of negation distributive po genitive initial first conjunct agreement possessor datives
(116)
a.
Unaccusative verbs √ √ √ √ √
Unergative verbs
AP predicates
NP predicates
* * * * *
* * * * *
* * * * *
Ivan bedstvoval. Ivan was-poor “Ivan was poor.” b. Ivan byl beden. Ivan was poor.sf “Ivan was poor.” c. Ivan byl bednjakom. Ivan was poor-person.instr “Ivan was poor.” OR: “Ivan was
verbal predicate
adjectival predicate
nominal predicate a pauper.”
The choice of the term is not as important as the differences between what I label Embodiment and various other proposals made in the literature. For example, my use of the term Embodiment is different from Williams’ (1981) R -role in two ways. First, unlike Williams’ R, Embodiment has no connection to referentiality since under my analysis referentiality is associated with the functional D-layer and not with the theta-grid of the noun (see Chapter 2 above). Second, unlike R, which is associated exclusively with nouns, Embodiment is associated with adjectives and stative verbs, as well as with nouns. Similarly, my use of the term Embodiment is different from Contreras’ (1995) use of the term Property: for him, Property is a -role that is assigned only by NP and PP predicates to their subjects, whereas verbal and adjectival predicates do not have the Property -role. Once again, my term Embodiment covers the -role of all stative predicates, regardless of their category. The term “Embodiment position” for stative verbs is similar to the Theme position of unaccusative verbs of change of state, such as arrive or die; yet, I will not use the term Theme for the thematic position of stative verbs, adjectives, and nouns in order to avoid a confusion between the Embodiment position which is of interests here and the Theme position inherited by derived nouns from the verbs (e.g., destruction inherits a Theme position from destroy and acquires an Embodiment position as a result of the nominalization).77 Another proposal that has been made in the literature regarding the thematic position of stative predicates is to call it Experiencer (cf. Chierchia and McConnell-Ginet 1990: 443). However, I prefer to restrict the use of the term Experiencer to sentient arguments of psych-predicates. To recap, three types of predicative expressions have the thematic position I call Embodiment: stative verbs, adjectives and nouns. However, as I have shown above with a range of unaccusativity diagnostics, these three types of predicative
84
CHAPTER 4
expressions fall into two categories with respect to the syntactic behavior of their Embodiment arguments. Stative intransitive verbs are unaccusative; their Embodiment position is discharged internally to the VP, much like the Theme position of unaccusative verbs of change of state. In contrast, adjectives and nouns pattern with unergatives; their Embodiment position is discharged externally to the AP or the NP, much like the Agent argument of transitive and unergative verbs (cf. Kratzer 1996). This is summarized in the Table 4 below. Table 4. Realization of arguments Type of predicative expression
Agent
Theme
Embodiment
transitive verb (e.g., kiss) unergative verb (e.g., laugh) unaccusative verb (e.g., arrive) Stative verb (e.g., Russian bedstvovat’ “be-poor”) adjective (e.g., Russian beden “poor”) noun (e.g., Russian bednjak “pauper”)
external external n/a n/a n/a n/a
internal n/a internal n/a n/a n/a
n/a n/a n/a internal external external
As can be seen from the shaded portion of this table, the Embodiment position is somewhat odd in that it can be discharged internally (in the case of stative verbs) or externally (in the case of AP and NP predicates). In that it differs from both the Agent position (which can be discharged only externally) and the Theme position (which can be discharged only internally). Recall that the reason for grouping the Embodiment arguments of stative verbs with those of AP and NP predicates is the close similarity and in some cases identity of meaning (in addition, an argument can be made that in some cases stative verbs are derived from adjectives/nouns or vice versa). Note that these facts provide a strong incentive for separating thematic relations from syntactic configurations. In this, I depart from a widely-adopted hypothesis that thematic relations reduce to structural configurations. This hypothesis goes back to Perlmutter’s (1978) Universal Alignment Hypothesis; it has been later formulated as the UTAH (Uniformity of Theta-Assignment Hypothesis) of Baker (1988: 46): (117) UTAH: Identical thematic relationships between items are represented by identical structural relationships between those items at the level of D-structure. The same idea also underlies the work of Hale and Keyser (see, for example, their 1987, 1991, 1993 papers), who have argued that there are no thematic roles. Instead, there are just the relations determined by the categories and their projections, and these are limited by the small inventory of lexical categories and by the principle of unambiguous projection. [Hale and Keyser 1991: 40, cited in Moro 1997: 229]
THEMATIC RELATIONS
85
Finally, Chomsky (1995: 313, see also Chomsky 2000: 103, 127) adopts the same approach to thematic relations. For him, a -role is assigned in a certain structural configuration; assigns that -role only in the sense that it is the head of that configuration … -relatedness is a property of the position of merger and its (very local) configuration. However, as can be seen from the behavior of the Embodiment arguments of stative verbs, on the one hand, and of those of AP and NP predicates, on the other hand, a configurational theory of thematic relations is insufficient. Here, I pursue Baker’s (2003) idea that even though the absolute position of the Embodiment is different in VPs and APs, the relative positions of the Embodiment with respect to Goal and Subject Matter are the same. To recap, we are faced with the following conundrum: the arguments of AP and NP predicates pattern semantically with the internal arguments of stative intransitive verbs, but syntactically with external arguments of transitive and unergative verbs. In what follows, I develop an analysis that resolves this paradox.
4.4.2.
Theta-Role Deconstruction and Non-Local Theta-Marking
So far, I have shown (following ideas in Baker and Stewart’s work) that the Embodiment position of AP and NP predicates is unlike the Embodiment position of stative intransitive verbs in that the former is discharged externally to the maximal projection of the predicative category, analogous to the way the Agent position of transitive and unergative verbs is discharged. Not only is this a problem for principles that relate thematic roles to structural configurations, such as UTAH, as mentioned at the end of the previous section, but it is also a from the point of view of locality restrictions on thematic relations: thematic relations have always been seen as very local. If, however, we allow certain thematic positions to be discharged outside the maximal projection of the predicative expression, it is very hard to restrict thematic discharge. Therefore, if at all possible, we should strive at keeping thematic relations strictly local; I will call this the Locality of -Marking Problem. In this section, I develop an analysis of predicative APs and NPs that avoids this locality problem. The starting point of this analysis is Kratzer’s (1996: 131) idea that the Agent of a transitive or unergative verb is “severed” from its verb and “is not really one of its arguments anymore”. Here I propose that the Embodiment argument of an AP or NP predicate is not its argument either, in the same sense. Thus the structure that I adopt based on the results of the unaccusativity diagnostics discussed in the previous section is parallel to the structure Kratzer proposed for the Agent arguments of transitive and unergative verbs:
86 (118)
CHAPTER 4
a. Kratzer’s structure for Agents:
vP
vP
DPEmbodiment
v'
DPAgent
b. My structure for Embodiments:
VP
v°
… V°…
v°
v' AP/NP …A°/N°…
However, adopting this structure for AP and NP predicates raises several problems. As mentioned above, assuming that the v in the structure in (118b) -marks the Embodiment argument in its specifier goes against the intuition that there is a connection between the Embodiment argument and the A /N head (a particularly troubling conclusion given their semantic similarity and synonymity, as discussed in the previous section). Thus, it appears that the Embodiment position “belongs” to the head of the predicate; I will call this the Predicate Head Problem. The second problem is that if the v discharges its own thematic position to its specifier, there is no reason a priori why it cannot take a DP complement rather than an AP or NP complement, possibly assigning it a -role as well. Thus, the structure in (118b) above does not exclude the possibility of having a copula with two DP arguments, that is having a copular sentence with the form DP was DP with the same structure as proposed here for copular sentences with the nom-instr pattern. In other words, nothing so far excludes the following structure: (119)
vP DPEmbodiment v°
v' DP …D° …
However, recall that the need to have distinct structures for the two kinds of copular sentences has been motivated by the data in Section 2.5 above. I will call this the DP Problem. Finally, assuming that v˚ -marks the Embodiment argument in its specifier makes it too parallel to the agentive v˚ proposed by Kratzer. Given that the latter is usually taken to host causative morphology, we expect more interchangeability between causative and copular elements than is found in natural language. For instance, as noted by Baker (2003: 53): many languages have causative morphemes that attach productively to verb roots, but not to nouns or adjectives… On the contrary, there is no similar category restrition on the periphrastic causatives formed with make in English. Some illustrative examples from Japanese and Chichewa (cf. Baker 2003: 54–55) are given below. In these examples, the underlined causative morpheme is compatible
87
THEMATIC RELATIONS
with verbal predicates (whether eventive, as in the Japanese example, or stative, as in the Chichewa example) but not compatible with adjectival or nominal predicates: (120)
a.
John-ga Mary-o ik-(s)ase-ta John-nom Mary-acc go(V)-caus-pst “John made Mary go.” b. *Taroo-ga heya-o hiro-sase-ta. Taro-nom room-acc wide(A)-caus-pst intended: “Taro widened the room.” c. *Hanako-ga Taroo-o sensei-sase-ta Hanako.nom Taro.acc teacher(N)-caus-pst intended: “Hanako made Taro a teacher.”
(121)
a.
Mwana akudetsa zovala. child is-dirty(V).caus.pres clothes “The child is making the clothes be dirty.” b. *Mbidzi zinakaliitsa mkango. zebras fierce(A).caus.pst lion intended: “The zebras made the lion fierce.” c. *Mbidzi zinafumuitsa mkango. zebras chief(N).caus.pst lion intended: “The zebras made the lion a chief.”
[Japanese]
[Chichewa]
I will call this the Copular-Causative Mismatch Problem. Even though this problem may be resolved by postulating lexically specified selectional properties for copular and causative elements, this is not the most elegant analysis. The analysis proposed in this book deals with this and the other problems mentioned above in a more straightforward way. To summarize so far, on the one hand we have the hypothesis that the Embodiment argument of adjectival and nominal predicates is -marked outside their maximal projection; this hypothesis runs into an obvious Locality of -Marking Problem. On the other hand, if we follow Kratzer in severing the Embodiment argument from the adjective/noun and assume that it is -marked (locally!) by v , we run into another set of problems, including the Predicate Head Problem, the DP Problem, and the Copular-Causative Mismatch Problem. This appears to be a fatal conundrum, which can be summarized as follows: the Embodiment argument of adjectival and nominal predicates appears to be “shared” between the head of the predicate (i.e., the A or the N˚) and the head of the small clause (i.e., the v ). In what follows, I will develop an analysis that will avoid all the aforementioned problems by taking the “argument sharing” idea seriously. This analysis is based on the distinction between two components of a thematic position, as proposed by Samek-Lodovici (2003); cf. also Speas (1990). The essence of Samek-Lodovici’s proposal is developing the idea that a lexical entry for an item must encode two kinds of information: (i) how many arguments a predicative expression must take in the syntax (i.e., the adicity of the predicate), and
88
CHAPTER 4
(ii) what the semantic content of each argument is.78 In particular, he deconstructs a thematic position (i.e., theta-role) into an argument-variable and a thematic index (the former is also called a-variable and the latter should not be confused with a referential index, cf. Section 4.3). In this book, I adopt Samek-Lodovici’s distinction and use “” to indicate an a-variable and a subscript (indicating the content of the thematic position) to indicate a thematic index. For example, the thematic grid of a transitive verb like kiss looks as follows:79 (122) kiss Ag Th Crucially for the analysis I propose here, both components of a theta-role are necessary for full interpretation; a-variables are “anonymous” (Samek-Lodovici 2003: 837) and therefore cannot be interpreted without thematic indices because only the latter introduce semantic information (cf. Speas 1990: 32). Samek-Lodovici uses this distinction between a-variables and thematic indices to account for two interesting facts concerning ata-nominalizations and nominalization-based complex predicates in Italian, illustrated below: the choice of the light verb (dare “give” vs. fare “do”) and the “resurrection” of the original external argument in nominalization-based complex predicates (examples from Samek-Lodovici 2003: 849). (123)
a.
Gianni ha fatto una risata. [Italian] Gianni has done a laughing “John laughed.” b. I ragazzi hanno dato una strizzata ai panni. the boys have given a wringing to-the clothes “The boys wrung the clothes.”
Specifically, Samek-Lodovici (2003: 837) proposes to analyze both light verb formation and thematic transfer as operations targeting thematic indices alone, while leaving a-variables untouched: light verb formation is analyzed as index-erasure, “which leaves an a-variable an orphan of its index and hence uninterpretable”, and thematic transfer as index-transfer, “which provides an unindexed a-variable with an index transferred from another variable”. (124) Light verb formation as index-erasure before index-erasure a. fare i j b. dare i j k
after index-erasure farelight v darelight
(125) Thematic transfer as index-transfer before index-transfer a. farelight + N1 (i (j b. darelight + N2 (i (j (k
after index-transfer farelight i j + N1 i j darelight i j k + N2 i j k
THEMATIC RELATIONS
89
In contrast, argument suppression is analyzed as an operation involving suppression of an a-variable without affecting the thematic index. Therefore, when a suppressed thematic position “resurrects” in nominalization-based complex predicate formation, the thematic index can be transferred to the light verb in the usual manner (suppression is indicated here by <>: (126)
(127)
Transfer with decomposition a. before index-transfer: b. after index-transfer:
farelight + N < >i k farelight i k +N< >i k
Transfer without decomposition a. before index-transfer: farelight – + N < > b. after index-transfer: farelight < > + N < >
In this book I use the distinction between a-variables and thematic indices to explain how an AP/NP predicate (which is a “deficient -marker”, as has been shown above) and a copula (which is a light verb) together -mark the Embodiment argument. In particular, I propose here that adjectival and nominal predicates have a complete theta-grid containing both an a-variable and a thematic index (here and below, “Emb” for Embodiment) but not the ability to -mark the specifier, whereas the copula v has a theta-grid which contains only an a-variable, but lacks a thematic index. In essence, a copula is a light verb, derived by index-erasure (cf. SamekLodovici’s 2003 analysis of light verb formation as index-erasure).80 The theta-grids of a copula v , an AP predicate, and an NP predicate are given below. (128)
a. vcopular b. AP Emb c. NP Emb
This lack of a thematic index places the copula between full lexical verbs (e.g., kiss in (122) above), which have a full theta-grid, and functional categories, which have no theta-grid at all (as discussed in Section 2.1 above).81 The copula’s lack of a thematic index also has two important consequences: first, it allows us to distinguish between a copula v and a causative (i.e., agentive) v (the former has a deficient theta-grid with only an a-variable, whereas the latter has a complete theta-grid containing both an a-variable and a thematic index); secondly, the lack of a thematic index forces the copular v to acquire an index from somewhere else so that it can be interpreted by virtue of having a complete theta role. The only way for the copular v to acquire a thematic index is through the process of index-transfer. Furthermore, I propose that index-transfer is done through -identification. As discussed in Higginbotham (1985) and Section 4.2 above, -identification encodes that two thematic positions are one and the same. Recall that as a result of -identification, only one of the original two thematic positions must be further
90
CHAPTER 4
discharged, whereas the other position (that of the non-projecting element) is considered discharged through -identification. Intuitively, -identification is restricted to thematic positions that have nondistinct thematic indices. Higginbotham (1985, 1989) himself discusses two cases of -identification: adjectival modification, and certain cases of adverbial modification: in the former case two arguments of the kind I call Embodiment are -identified, whereas in the latter case two Event arguments are -identified. Crucially, -identification cannot occur between two thematic positions that have distinct thematic indices, for example, a Theme and an Agent. However, -identification is possible between two thematic positions if one of them has no thematic index. Here, I propose that this is exactly how index-transfer in copular sentences happens, as schematized below: (129)
a. before index-transfer v'〈θ〉 v°〈θ〉 AP/NP〈θEmb〉 b. after index-transfer v ' 〈θEmb〉 v° 〈θEmb〉
AP/NP〈θEmb〉
Note that since -identification is restricted to the configuration of sisterhood (see Section 4.2 above), index-transfer is restricted to the same configuration. As in Higginbotham’s original proposal, -identification in (129) results in discharging the thematic position of the non-projecting element (here, the AP). Therefore, only one Embodiment position – that of the v – has to be further discharged by locally -marking the DP in Spec-vP, in accordance with the generalization made in section 4.2 above. Contrary to Baker’s (2003) proposal that v unilaterally “helps” its adjectival complement to discharge its thematic position, under my analysis both the copular v and its AP/NP complement “help” each other. On the one hand, -identification between the theta-grids of the AP/NP predicate and the copular v allows the AP/NP predicate to discharge its thematic position, and on the other hand, it allows the v to acquire a thematic index without which it cannot be interpreted. Consider how this analysis avoids the four problems outlined above. First of all, the Predicate Head Problem is resolved since the thematic index of the Embodiment position originates in the theta-grid of the predicate head (i.e., the A or the N ). Yet, the Locality of -Marking Problem is also avoided since the seemingly non-local process of -marking from the AP/NP predicate to the Spec-vP is now reduced to two independently necessary, strictly local processes: -identification (between the thematic positions of the copular v and its AP/NP complement) and -marking (from v to the Spec-vP). In other words, the Embodiment argument
THEMATIC RELATIONS
91
appears to be shared between the copula v (a local -marker) and the adjectival or nominal predicate (the semantic predicate) because it is shared between the two in the sense that the a-variable is supplied by the copula and the thematic index – by the A /N predicate. This is exactly what allows the Embodiment arguments of adjectival and nominal predicates to pattern semantically with internal arguments of stative intransitive verbs and syntactically with external arguments of unergative and transitive verbs. This analysis also resolves the Copular-Causative Mismatch Problem: the copular and causative v s are distinguished by the fact that the causative v has a complete theta-grid, whereas the copular v has an incomplete theta-grid lacking a thematic index. Therefore, there is no need to rely on selectional restrictions in order to capture the fact that each v must combine with their proper complements; these restrictions fall out from the analysis itself. Consider what would happen if each type of v were to take a wrong type of complement. If a causative v were to take an AP or an NP complement, the resulting structure would crash because the AP/NP would fail to discharge its thematic position since it cannot be dischanged within the AP/NP (due to A’s and N’s categorial deficiency). Nor can it be discharged through -identification with the thematic position in the theta-grid of the v because both elements have complete theta-grids with distinct thematic indices: the v has an Agent position, whereas the AP/NP has an Embodiment position. As discussed above, two distinct thematic positions – here, an Agent position and an Embodiment position – cannot undergo -identification. As a result, the AP/NP fails to discharge its thematic position, and remains a “dangling predicative expression” in violation of the Theta Criterion (see Section 4.1). (130)
v'〈θAg〉
*
v°〈θAg〉
AP/NP〈θEmb〉
undischarged thematic position
Similarly, a copular v cannot take a VP complement. Since a VP has no undischarged thematic positions (because all thematic positions in the theta-grid of the V being discharged internally to the VP), there is nothing for the copular v to undergo -identification with. Without -identification, the copular v lacks a thematic index and cannot be interpreted. (Here and below, stands for a fully discharged grid). (131)
* no thematic index
v'〈θ〉 v°〈θ〉
VP 〈 〉
To recap, the Copular-Causative Mismatch Problem is solved under the analysis proposed here because the causative v cannot occur with an AP or an NP complement and the copular v cannot occur with a VP complement. This is so because in both cases -identification fails to occur, resulting in an undischarged
92
CHAPTER 4
thematic position or an incomplete thematic position that cannot be interpreted. Thus, it is not necessary to postulate ad hoc selectional restrictions for these two v s. Finally, we come to the DP Problem. Crucially, I have argued in Chapter 2 that sentences with post-copular DP and sentences with post-copular APs/NPs must be associated with distinct syntactic structures; this means that a copular v cannot take a DP complement. How is a DP complement to a copular v ruled out? Since a DP has no undischarged thematic positions, -identification between the DP and the copular v is impossible (as with other modes of thematic discharge, -identification can apply only to undischarged thematic positions). As a result, the copular v cannot be interpreted for the want of a thematic index. This is schematized below: (132)
* no thematic index
v'〈θ〉 v° 〈θ〉
DP 〈 〉
not θ-marked
To summarize so far, I have shown how the analysis proposed here resolves the four problems identified earlier: the Locality of -Marking Problem, the Predicate Head Problem, the Copular-Causative Mismatch Problem and the DP Problem. Under this analysis, the thematic position which is discharged by the Spec-vP in rich copular sentences is in essence “shared” between the predicate head (i.e., the head of the predicative AP/NP) and the copular v : the a-variable comes from the v and the thematic index from the AP/NP (ultimately, from A /V ). Finally, it must noted that since the copular v takes a predicative expression (an AP or an NP) as a complement, it is in a good position to c-select the category of its complement. Thus, under the analysis proposed here nothing prevents the existence of a language with two copular v s: one selecting an NP complement and the other selecting an AP complement. Possible examples of a language like that include Èdó and Norwegian. According to Baker (2003: 40), there are two copular elements in Èdó: yé appears with adjectival complements, whereas rè appears with nominal complements (it is unclear from their description if rè appears with NPs or DPs or both). (133)
a.
Èmèrí *(yé) mòsèmòsè. Mary cop beautiful(A) “Mary is beautiful.” b. Úyì *(rè) òkhaèmwèn. Uyi cop chief(N) “Uyi is a chief.”
[Èdó]
In Norwegian small clauses, discussed by Eide and Åfarli (1999), som appears as the head of the small clause obligatorily only if its complement is an NP; with AP predicates som is obligatorily absent, as illustrated below with data from Eide and Åfarli (1999: 160).
93
THEMATIC RELATIONS
(134)
a.
Vi fant [Marit (*som) naken]. we found Mary (*som) naked “We found Mary naked.” b. Vi fant [Marit *(som) nervevrak]. we found Mary *(som) nervous-wreck “We found Mary a nervous wreck.”
[Norwegian]
A possible analysis of these data is to postulate two copular v : som and Ø, where the former c-selects an NP complement, whereas the latter c-selects an AP complement. Eide and Åfarli (1999: 162) propose a similar analysis: for them, som is syntactically present with both NP and AP complements, but its phonetic realization (overt or non-overt) depends on whether it functions as an abstract Case licenser, given the assumption that only nominal constituents require abstract Case (for a further discussion of case in copular sentences, see Chapter 5 below).82 A similar approach may also be adopted in dealing with the “present tense problem” in Russian copular sentences: even though instrumental-marked NPs (including long-form adjectives like umnym “intelligent.instr”) and short-form AP behave very similarly in all other respects (including their semantics), only the short-form adjectives are allowed with the phonetically null present tense copula; instrumental-marked NPs are ungrammatical in the same context: (135)
a.
Saša umën. Sasha intelligent.sf “Sasha is intelligent.” b. *Saša umnym. Sasha intelligent.instr intended: “Sasha is intelligent.” c. *Saša studentom. Sasha student.instr intended: “Sasha is a student.” [Bailyn and Rubin 1991: 121, fn.10]
There are three exceptions to this generalization where instrumental-marked NPs are grammatical in the present tense (with a null copula): • with certain nouns denoting occupations in the presence of a temporal or locative adverbial, such as zdes’ “here” or sejcˇ as “now”, as in (136a); • with a very small set of nouns meaning “cause, reason”, as in (136b); • in certain tautological constructions, as in (136c). (136)
a.
Saša zdes’ studentom. (cf. (135c) above) Sasha.nom here student.instr “Sasha is here as a student.” [Bailyn and Rubin 1991: 121, fn.10] b. Vsemu vinoj moja veˇcnaja neobdumannost’. all.dat fault.instr [my eternal rashness].nom “My perpetual rashness is to blame for everything.” [Turgenev, cited in Fowler 1997: 152]
94
CHAPTER 4
c.
Družba družboj, a služba služboj. friendship.nom friendship.instr but duty.nom duty.instr “Friendship is friendship, but duty is duty.” [Nichols 1981: 208]
These data can be accounted for by postulating two copular v s, one selecting an NP complement and the other one selecting an AP complement. The former but not the latter can appear in the present tense; in other words, there is a gap in the tense paradigm of the v that selects an NP complement, which results in the impossibility of sentences like (135b–c) above. Here, I will leave this solution as a tentative proposal because of the obvious disadvantage of increasing the number of lexical entries for copular elements in a language. For a more detailed discussion of these examples, the reader is referred to Nichols (1981: 125, 208, 266), Bailyn and Rubin (1991), Fowler (1997), and Geist (1999).
4.5.
MEANING DIFFERENCES: A REVIEW
In the preceding sections, I developed an analysis where Russian copular sentences with the nom-nom pattern are interpreted as equative (namely, asserting that the post-copular DP is interpreted as coreferential with the pre-copular DP), whereas sentences with the nom-instr pattern are interpreted as asserting that a property denoted by the post-copular NP/AP is ascribed to the individual denoted by the precopular DP. In this section, I examine whether this analysis matches the intuitions about what such sentences really mean.83 The problem of accurately describing the differences in meaning between the two case-marking patterns is an old one. Descriptive literature uses terms like “identity”, “characteristic”, “status”, “function”, “essential quality”, “appearance”, “concreteness”, “temporal” and the like to describe the meaning differences between sentences with the two case-marking patterns (for a good overview of descriptive literature on the subject, see Nichols 1973: 7–17). More recently, in the generative literature, the description relies on such notions as stage-level vs. individual-level properties/predicates. The most widely accepted generalization is that sentences with the nom-instr pattern denote transient, temporary, or changeable properties, whereas sentences with the nom-nom pattern denote characteristic, permanent, or non-changeable properties (cf. Peshkovskij 1914/1956, Jakobson 1936/1971, Rozental’ 1976: 37, Wierzbicka 1980, Bailyn and Rubin 1991, Fowler 1997, Smith 1999, Geist 1999, Timberlake 2004: 286–288, to name only a few). For example, Wierzbicka (1980: 119) characterizes the meaning differences as follows: the nominative case is used when the predicate nominal denotes a property seen as essential and inalienable; the instrumental case is used when the predicate nominal denotes a property which is seen as transient and inessential.
THEMATIC RELATIONS
95
For example, the nominative predicate nominal in (137a) denotes an inalienable property, whereas the instrumental predicate nominal in (137b) denotes a transient property, which is emphasized by the use of temporal adverbial modifier (the examples are from Bailyn and Rubin 1991: 120). (137) a.
Tatjana Iliniˇcna Ovsjanikova byla zenšˇcina vysokogo rosta. [Tatjana Ilinichna Ovsjanikova].nom was woman.nom [tall height].gen “Tatjana Ilinichna Ovsjanikova was a tall woman.” b. V prošlom godu Tatjana byla studentkoj. in last year Tatjana was student.instr “Last year Tatjana was a student.”
However, the use of the terms “stage-level” vs. “individual-level” to characterize the distinction expressed through case marking in Russian is problematic. In particular, properties that are clearly inherent and non-changeable can be expressed by instrumental phrases, as in the examples below (from Geist 1999: 7); similar examples are also found in Rozental’ (1976: 38) Comrie and Stone (1978: 118) and Timberlake (2004: 288). (138)
a.
Anna byla doˇcer’ju vraˇca. Anna was daughter.instr doctor.gen “Anna was a doctor’s daughter.” b. Sergej byl levšoj. Sergey was lefthanded.instr “Sergey was left-handed.”
Thus, the sentence in (138a) can be used truthfully and felicitously even if Anna never stopped being a doctor’s daughter (which would be possible only if her parent was no longer a doctor). Similarly, even though it is possible that Sergey – as many Russian children – was forced to acquire right-handedness (since left-handedness was thought to be a sign of mental retardation), the sentence in (138b) does not necessarily imply that. Note furthermore that the instrumental is possible even if the inherent nature of the property in question is expressed overtly in the sentence (the example from Geist 1999: 7): (139)
On byl { priroždennyj muzykant / priroždennym muzykantom}. he was [born musician].nom / [born musician].instr “He was a born musician.”
On the other hand, nominative post-copular phrases can denote properties which are not necessarily individual-level. This is particularly clear in the case of idiomatic post-copular noun phrases, which must appear in the nominative (cf. Geist 1999: 15), even if the temporary nature of the property is expressed explicitly in the sentence.84
96 (140)
CHAPTER 4
V detstve ona byla krov’ s molokom, no posle diety ona tak in childhood she was blood.nom with milk but after diet she still i ostalas’ koža da kosti. emph remained skin.nom and bones.nom “In her childhood, she was milk and roses, but after (being on) a diet she remained forever skin and bones.”
Examples of this sort can be multiplied, but what is important here is that the distinction between stage-level and individual-level properties/predicates is not adequate for describing the meaning differences between sentences with the nomnom and nom-instr patterns. Hinterhölzl (2001) proposed a modification to the analysis of “semantic constraints on case assignment” based not on a two-way distinction between stage-level and individual-level properties/predicates but on a three-way distinction in terms of boundedness. He focuses on depictives; in what follows, I show that this analysis does not extend easily to primary predicates that are the main concern of this book (I return to depictives in Section 5.6 below). Hinterhölzl (2001) uses the adverbs uže “anymore” and ešcˇ ë “yet” to distinguish three semantic classes of predicates: • bounded: having both an upper and a lower bound; e.g., bol’noj “sick”, pjanyj “drunk”, serdityj “angry”, etc.; • partially non-bounded: having either an upper or a lower bound but not both; e.g., molodoj “young”, naivnyj “naïve”, spelyj “ripe”, obrazovannyj “educated”, etc.; • unbounded: having neither an upper nor a lower bound: e.g., umnyj “intelligent”, glupyj “stupid”. (141)
a.
Kogda ja vstretil Ivana, on {ešˇcë / uže } ne byl pjanym. when I met Ivan he yet / anymore not was drunk.instr “When I met Ivan, he was not drunk {yet/anymore}.” b. Kogda ja vstretil Ivana, on { uže /*ešˇcë } ne byl molodym. when I met Ivan he anymore /*yet not was young.instr “When I met Ivan, he was not young {anymore / *yet}.” c. Kogda ja vstretil Ivana, on {ešˇcë/*uže } ne byl starym. when I met Ivan he yet /*anymore not was old.instr “When I met Ivan, he was not old {yet/*anymore}.” d. Kogda ja vstretil Ivana, on {*ešˇcë/*uže } ne byl umnym. when I met Ivan he *yet / *anymore not was intelligent.instr “*When I met Ivan, he was not intelligent {yet/anymore}.”
According to Hinterhölzl (2001: 104), bounded depictives appear alternatively in the agreeing case (which would correspond to nominative with primary predicates) or in the instrumental, partially non-bounded depictives appear only in the instrumental, and unbounded depictives are impossible at all. Can this three way distinction be applied to primary predicates? It appears that the answer is no. The
97
THEMATIC RELATIONS
examples in (141) above show all types of predicates can appear in the instrumental (the sentence in (141d) is grammatical if neither temporal adverb is included). As shown in (142) below, the same predicates can be used in the nominative as well: (142)
a.
Ivan byl pjanyj. Ivan was drunk.nom “Ivan was drunk.” b. Ivan byl molodoj. Ivan was young.nom “Ivan was young.” c. Ivan byl staryj. Ivan was old.nom “When I met Ivan, he was not old.” d. Ivan byl umnyj. Ivan was intelligent.nom “Ivan was intelligent.”
Another analysis that attempts to account for meaning differences between the two case-marking patterns in Russian copular clauses was proposed by Matushansky (2000). According to her, the two types of copular clauses in Russian differ in their aspectual properties, which are encoded structurally in her analysis: sentences with the nom-instr pattern contain an aspectual projection, AspP, which gives them a “perfective” flavor, whereas sentences with the nom-nom pattern lack this projection. She supports her proposal with the following evidence: when an explicit aspectual marker – such as a super-lexical aspectual prefix po-, as in (143a), or a habitual/repetitive suffix -(yv)aj-, as in (144a) – is present, only instrumental is possible (examples from Matushansky 2000: 113, her (19)–(20)): (143)
a. b.
(144)
a.
Ja pobyla { zavedujušˇcej / *zavedujušˇcaja } I was.perf manager.instr/ *manager.nom Ja byla { zavedujušˇcej / zavedujušˇcaja} I was manager.instr /manager.nom “I was a manager for two hours.”
Ja byvala { zavedujušˇcej / *zavedujušˇcaja}, I was.imperf manager.instr/ *manager.nom b. Ja byla { zavedujušˇcej / zavedujušˇcaja}, I was manager.instr / manager.nom “I have been a manager, but rarely.”
dva cˇ asa. two hours dva cˇ asa. two hours
no but no but
redko. rarely redko. rarely
The intuition that Matushansky’s analysis tries to capture, namely that the two types of copular sentences differ in their aspectual properties, corresponds well to the judgments of (non-linguist) native speakers; in what follows, I show that
98
CHAPTER 4
the analysis proposed in this book captures it too, in addition to other meaning differences between the two types of sentences. Consider the following contrast and the structures proposed in this book for these two types of sentences, repeated from above: (145)
(146)
a.
Ivan byl xrabryj soldat. Ivan was brave.nom soldier.nom “Ivan was a brave soldier.” b. Ivan byl xrabrym soldatom. Ivan was brave.instr soldier.instr “Ivan was a brave soldier.” a. NOM-NOM pattern:
b.NOM-INSTR pattern
TP
TP T'
DPi T°
DPi T°
FP ti
T'
ti
F'
byt’ DP
F' F°
DP
F°
FP
DP
ti OR: ti
vP
byt’ ti
v' v°
NP/AP
tbyt’
According to the analysis developed in this book, in (145a, 146a) there is no verbal predication. The copula here is a functional category, meaning that according to the definitions given in Section 2.1, it does not have a thematic grid. Hence, in (145a, 146a) there is no verbal source for an event variable. Instead, if an event variable is present at all, it must come from the nominals themselves. Thus, past tense is interpreted as applying to the individuals denoted by the DPs Ivan and xrabryj soldat “brave soldier”. This explains the intuition that for this sentence to be used felicitously, Ivan must be dead.85 Furthermore, since the two DPs are co-referential, they denote the same individual, which results in the often-reported intuition that the nominative post-copular DP denotes an individual-level property (as discussed above). In contrast, the structure in (145b, 146b) contains v˚, which is lexical category bearing a thematic grid. One of the arguments in that grid is an event argument, which can be quantified over by aspectual morphemes, discussed by Matushansky (2000). Furthermore, tense marking in this structure applies to the predication itself indicating that Ivan’s having the property of being a brave soldier happened in the past. However, it does not limit the extent of the span in which the property of being a brave soldier applies to Ivan. Therefore, the sentence can be used felicitously
THEMATIC RELATIONS
99
even if Ivan was a brave soldier throughout his whole life (i.e., if the intended meaning is of the individual-level property), as mentioned above. By uttering these sentences, a speaker asserts that the individual denoted by the pre-copular DP had a certain property in some past situation, but makes no commitment as to whether s/he had that property prior or following the span of that situation. Thus, the effect of temporariness arises mostly from a meta-linguistic comparison with the nominative. However, the analysis proposed in this book goes beyond capturing the aspectual meaning differences between the two types of sentences. Not only can it account for the differences in syntactic behavior between post-copular DPs and NPs/APs (as discussed in Chapter 2 above), which Matushansky’s analysis cannot do, but my analysis also captures the intuition that nominative-marked post-copular phrases presuppose existence of an individual (named or characterized by that phrase), whereas instrumental-marked post-copular phrases have no such presupposition. My explanation is as follows: in a post-copular DP (which, according to the analysis proposed in this book, is always marked nominative) the head D introduces a referential index, whereas in a post-copular NP/AP (always instrumental-marked, according to my analysis) there is no referential index. Since it is the referential index that is responsible for the presupposition of existence, only DPs but not NPs or APs can be associated with such a presupposition.86 In other words, Russian copular sentences with the nom-nom pattern presuppose the existence of the individuals denoted by the two DPs flanking the copula and assert that those are one and the same individual. Thus, under the analysis proposed here sentences with the nom-nom pattern have an equative interpretation. However, the claim that copular sentences with the nom-nom pattern always have an equative interpretation is not a priori obvious. It is easier to see that this is the case with sentences like (147a) than (147b). (147)
a.
Mark Tven byl Samuèl Klements. Mark Twain.nom was Samuel Clements.nom “Mark Twain was Samuel Clements.” b. Piter byl doktor. Peter.nom was doctor.nom “Peter was {a/the} doctor.”
However, even sentences like (147b) pass the coordination test for an equative reading, from Holmberg (1993: 130); this test shows that the individual whose existence is presupposed by the post-copular DP cannot be equated to two other individuals. (148)
a. Peter is a teacher, and Lisa is a teacher, too. b. *Peter is the teacher, and Lisa is the teacher, too.
A similar contrast is found in Russian copular sentences.
100 (149)
CHAPTER 4
a.
Piter byl doktorom, i Andrej tože byl doktorom. Peter was doctor.instr and Andrew too was doctor.instr “Peter was a doctor, and Andrew was a doctor too.” b. *Piter byl doktor, i Andrej tože byl doktor. Peter was doctor.nom and Andrew too was doctor.nom intended: “Peter was the doctor and Andrew was the doctor too.”
Thus, sentences with the nom-nom pattern behave like equative sentences rather than as predicative sentences. This brings us to the issue of negation in Russian copular sentences; although a detailed discussion of negation goes beyond the scope of this book, it must be noted here that the two types of copular sentences do not behave in the same way with respect to negation. Since sentences with the nom-instr pattern (i.e., with a post-copular NP/AP) do not presuppose the existence of an individual, the negated sentence in (150a) is grammatical. In contrast, in sentences with the nom-nom pattern a post-copular DP presupposes the existence of an individual; hence, the negated sentence in (150b) is ungrammatical: this sentence both presupposes and negates the existence of my best friend, leading to semantic incongruity. (150)
a.
Nikto ne byl moim luˇcšim drugom. nobody not was my best friend.instr “Nobody was my best friend.” = “I had no best friend.” b. *Nikto ne byl moj luˇcšij drug. nobody not was my best friend.nom “Nobody was my best friend.” = “I had no best friend.”
The final meaning difference between the two types of copular sentences in Russian is highlighted by the data concerning verum focus on the copula (discussed in Geist 2002). Recall from the previous sections that according to the analysis proposed in this book, the copula in sentences with the nom-instr pattern serves to establish predication (it participates in thematic discharge), whereas in sentences with the nom-nom pattern the copula is just a functional head (it picks up the verbal tense morphology, but does not participate in thematic relations). As noted in Geist (2002), when a verum focus (i.e., contrastive focus that falls on the finite verb of the clause) falls on the copula, it has different effects in the two types of Russian copula sentences (here, verum focus is marked by CAPS). In particular, if the verum focus serves to deny (a negative) predication (i.e., to assert contrary to previously expressed proposition that a particular individual is indeed associated with a particular property), then only the sentence with the nom-instr pattern is acceptable and the one with the nom-nom pattern is incongruent. (# marks infelicity in context; note that Yury Solomin has died prior to the utterance of these sentences). (151)
A:
Ja ne sˇcitaju Yurija Solomina talantlivym aktërom. I not consider Yury Solomin talented.instr actor.instr “I don’t consider Yury Solomin a talented actor.”
THEMATIC RELATIONS
101
B:
Net, on BYL talantlivym aktërom. no he was talented.instr actor.instr “No, he WAS a talented actor.” B: # Net, on BYL talantlivyj aktër. no he was talented.nom actor.nom “No, he WAS a talented actor.” In contrast, if the verum focus serves to deny a previously expressed tense marking, both sentences are acceptable: the sentence with the nom-nom pattern implies that the individual denoted by the subject is dead, whereas the sentence with the nom-instr pattern implies that he no longer holds that property. This is exactly as predicted by the analysis discussed in this book. (152)
A:
Yurij Solomin dejstvitel’no talantlivyj aktër. Yury Solomin really talented.nom actor.nom “Yury Solomin is really a talented actor.” B: Net, on BYL talantlivym aktërom. no he was talented.instr actor.instr “No, he WAS a talented actor.” [and now he is no longer talented] B: Net, on BYL talantlivyj aktër. no he was talented.nom actor.nom “No, he WAS a talented actor.” [and now he is dead]
To summarize, post-copular phrases in sentences with the nom-nom pattern need not denote properties which are inherently individual-level, and post-copular phrases in sentences with the nom-instr pattern need not denote stage-level properties. The analysis proposed in this book allows us to account for the speakers’ intuitions about these sentences without relying on the distinction between changeable and non-changeable properties. Moreover, these intuitions and other intuitions about the subtle meaning differences between the two case-marking patterns are shown to follow from the independently required syntactic structures of these sentences. 4.6.
SUMMARY
In this chapter, I have discussed issues related to the interpretation of the two types of copular sentences in Russian and argued that the structures proposed in this book for these sentences account for their meaning differences. In particular, I proposed that the thematic module of the grammar is located at LF and consists of the Theta Criterion which bans dangling predicative expressions and dangling referential expressions. With respect to sentences with the nom-nom pattern, I made the following claims: • The interpretation of such sentences does not involve any thematic discharge. • A distinction is drawn between syntactic co-indexing and semantic coreference (cf. Fiengo and May 1994).
102
CHAPTER 4
• The two DPs in a such sentence are interpreted as co-referential because they are coindexed in the syntax (contra Fiengo and May 1994). • The requirement that the two DPs are coindexed is not ad hoc; it follows from the assumption that referential indices are features and the claim that two input constituents in a symmetrical structure must have the same feature compositions. Both of these points are independently motivated. • A-binding is an inherently asymmetric relation; accordingly, it is defined as involving only asymmetrical c-command. With respect to sentences with the nom-instr pattern, I made the following points: • Such copular sentences are interpreted on a par with non-copular sentences, through thematic discharge. • A distinction is drawn between a-variables and thematic indices (cf. SamekLodovici 2003). • The copular v has its own theta-grid, but its thematic position has only an a-variable and no thematic index. • The theta-index is acquired by v ’s thematic position as a result of thematic transfer (which is analyzed as index-transfer, cf. Samek-Lodovici 2003) through -identification. • Predicative NPs/APs discharge their Embodiment position through -identification with the theta-grid of v . • Consequently, copular v cannot take a DP as its complement, because a DP has no theta-grid, and -identification (and as a result, thematic transfer) would fail, resulting in an uninterpretable structure. Finally, I argued that the structures and analysis proposed in this book account for the meaning differences between the two types of copular sentences in the most straightforward way.
CHAPTER 5
CASE RELATIONS
Cases are one of the most irrational parts of language – Jespersen
In the previous chapters, I argued that the two case marking patterns in Russian copular sentences correspond to two distinct syntactic structures: the nom-nom pattern corresponds to a symmetrical structure where two DPs are merged together without a mediation of a functional or lexical head and the nom-instr pattern corresponds to an asymmetrical small clause structure where the post-copular NP/AP is a complement/sister of the lexical head v . These structures allow me to account for differences in syntactic behavior of post-copular DPs and NPs/APs. I also argued that the differences in meaning reported between these two case marking patterns derive from the structures I proposed for these sentences. In this chapter, I address the question of the case marking on post-copular phrases more directly, by showing how the observed morphological case follows from the structures proposed in this book. The chapter is organized as follows: after introducing the debate on the nature of case and Case Theory (in Section 5.1), I discuss the status of nominative case in Section 5.2 and propose to analyze some instances of it as morphological default which surfaces in the absence of syntactic case. Then in Sections 5.3 and 5.4 I discuss the two types of Russian copular sentences; for each type, I show that the structures proposed in this book (see Section 1.5) provide an elegant account for the distribution of case marking. Furthermore, I show that my analysis is superior to its alternatives in accounting for the distribution of case marking; in particular, I show that the structure proposed in this book for Russian copular sentences with the nom-instr pattern, namely the one involving a lexical head v , is superior to the one which involves a functional head Pred , as proposed by Bailyn and Rubin (1991). Moreover, I show that the symmetrical structure I propose for Russian copular sentences with the nom-nom pattern is preferable to an asymmetrical structure in which the two DPs are arguments of the identity copula. In addition, in Sections 5.5 and 5.6 I discuss case marking in sentences with copula-like verbs and in sentences with depictives, respectively. Finally, Section 5.7 concludes this chapter. 5.1.
CASE THEORY
Although orgininally the term “case” referred exclusively to morphological (specifically, inflectional) marking on nouns (and related elements), in Chomskian syntactic theory, and especially in Government and Binding, it has mutated into a more 103
104
CHAPTER 5
abstract notion of a licensing mechanism for nominals. The inability of overt nominals to appear in certain syntactic positions (e.g., *(For) John to speak such perfect Russian is incredible) has been explained by their failure to check/be assigned abstract syntactic case; this abstract syntactic case is found even in languages with impoverished or non-existent morphological case systems, such as English or Chinese. This separation between morphological case and abstract syntactic case (or “Case”, or even “Kase”) brings up the issue of how the two are related in languages with richer morphological case systems, such as Russian. As is witnessed by the wealth of literature on the topic (cf. especially the work of Babby and Franks), the relation between morphological and syntactic case in Russian is not straightforward. In this section, I address this issue and outline my assumptions on the matter. In standard Minimalist framework (as well as in its predecessor, Government and Binding) the focus is on the abstract syntactic case, which is taken to be a set of formal features “par excellence” because they are always uninterpretable and trigger movement, serving to license nominal phrases (cf. Chomsky 1995: 278–279, Chomsky 2000: 102, 119). However, since most attention has been paid to argumental nominal phrases, it is not clear whether the theory extends easily to predicative (e.g., post-copular) phrases. In the framework of Government and Binding, two views of case have developed: the original Case Filter of Rouveret and Vergnaud (1980) and Chomsky (1981, up to ch.6) rules out caseless noun phrases with phonetic content, regardless of their thematic status; however, this view was quickly replaced by Chomsky’s (1981: 336) alternative which derived the Case Filter from the Theta Criterion and applies to -marked noun phrases independently of their phonetic content. This revision of the Case Filter became known as the Visibility Condition since Case marking was treated as making a noun phrase visible for Theta marking. Note that the two versions of the Case Filter have different domains of application and, presumably, apply at different levels of representation: the original Case Filter applies at S-structure (or PF, in Minimalist terms) and the Visibility Condition has been interpreted as applying at LF. Even though the Visibility Condition has been widely accepted as the correct formulation of the Case Filter, some researchers challenged it as incorrect or insufficient, proposing to go back to the original Rouveret-Vergnaud formulation (case as licensing mechanism for all and only over noun phrases; cf. Lee 1989) or to adopt both (e.g., Franks 1995, ch.7, Falk 1997, and the references cited therein). One problem with the Visibility Condition is that, being an LF principle, it does not easily relate to the observable morphological case. Another related problem is that it is not relevant for case marking on post-copular phrases since presumably post-copular nominals are not -marked and thefore need not be marked with Abstract Case either, according to the Visibility Condition. Yet, at least in some languages with morphological case systems – including not only Russian but also Icelandic, Arabic, Korean, Finnish and others – post-copular phrases clearly bear morphological case marking, which under the system that includes the Visibility Condition remains unaccounted for. The solution proposed within the framework
CASE RELATIONS
105
of the Visibility Condition is to treat such case marking as a morphological default in the absence of syntactic case resulting from an independent morphological wellformedness requirement on nouns. However, this approach fails to account for two uncontroversial facts: first, some case-marking on post-copular phrases is not easily analyzed as morphological default (e.g., instrumental case in Russian, or accusative case in Arabic); second, at least some case-marking on post-copular phrases has been argued to be indicative of certain syntactic configurations (cf. Lee 1989, Maling and Sprouse 1995, Comrie 1997). This leads me to reject the Visibility Condition framework. Instead, I adopt a system that takes case – both syntactic and morphological – as a licensing mechanism for nominals: abstract syntactic case serves as a licensing mechanism in syntax (and has consequences for LF), while morphological case serves as a licensing mechanism at PF. The roots of these two licensing mechanisms are different: abstract syntactic case is necessitated by the need to identify arguments, that is – according to the definition discussed in Section 4.1 above – complements and specifiers of predicative expressions. As such, abstract syntactic case is a universal requirement, applicable in all languages regardless of their morphological case systems. This view of syntactic case is related to but not identical with the Visibility Condition or earlier theories of case as expressing semantic relations (cf. Panini’s karaka theory relating Sanskrit cases to “semantic relations holding between nouns and verb”; Jakobson’s 1936 theory of case; Fillmore’s 1968 view that there is relation between “underlying syntactic-semantic relationships”, which he called “cases”, and overt case marking; Anderson’s 1977 Localist Case Grammar and Starosta’s 1971, 1988 Lexicase model seeking to find connections between semantic roles and case marking). In recent years, a similar approach to syntactic case as identifier of arguments has been adopted by Falk (1997) and Neeleman and Weerman (1999): for the former, case identifies arguments at PF, and for the latter it identifies arguments in the syntax proper. Even though this approach to case as a identifier of arguments is not adopted in the classical Minimalist theory (cf. Chomsky 1995), it is perfectly compatible with the Minimalist reasoning about language: for instance, this approach is found in Martin (1999: 17–18), who proposes that the existence of uninterpretable features such as syntactic case is motivated by the needs of bare output conditions. According to him, “Case exists in the grammar to serve as the necessary label to make sets of +Interpretable features formally distinct”; in other words, “bare output conditions demand that the +Interpretable features of arguments in the same domain be formally distinguishable at LF, and it is Case that provides that solution”. It is important to note here that although the root of abstract syntactic case is in the need to identify arguments of predicative expressions, it is not always true that the case-checking head and the relevant predicative expression are one and the same. For instance, nominative agentive subjects (e.g. in John cooked the best borsch) are arguments of a causative v (following Kratzer 1996 and much subsequent work, as discussed in the previous chapter); yet, their syntactic nominative case is checked not by v but by T . In general, abstract syntactic case can be checked by a head
106
CHAPTER 5
either on its complement or on its specifier; the former is referred to as “inherent case”, while the latter is called “structural case” (and should not be confused with a more general notion of “syntactic case” that encompasses both structural and inherent case). In contrast to abstract syntactic case, morphological case marking is a languagespecific requirement found in some languages but not others. Russian is one language where such a requirement is found: nominal roots in Russian are bound so that all nouns in Russian regardless of their syntactic contexts (or even of whether there is one at all) need to bear an “ending” (in Russian okoncˇ anie) which encodes morphological case, along with gender/declension class and number; the dictionary form is nominative singular. This brings us to the issue of the nature of nominative case, considered in the next section. 5.2.
THE STATUS OF NOMINATIVE CASE
The standard Minimalist view of nominative treats it on a par with other structural cases (most commonly, accusative, and potentially also genitive and dative). For example, Chomsky (1995: 277) suggests that [assign nominative Case] is a formal feature on T , whereas [assign accusative Case] is a formal feature on the verb (in later literature, accusative became associated with v rather than V ). Others have treated case as an uninterpretable instantiation on a noun phrase of an otherwise interpretable feature. For example, Pesetsky and Torrego (2000) propose to analyze nominative case as an uninterpretable instantiation of Tense on the noun phrase, or uT on D in their terminology (this proposal is crucial to their explanation of the T-to-C asymmetry); furthermore, they speculate that accusative case is also uT on D (i.e., an uninterpretable Tense feature on D ), with “the choice of case morphology … taken to reflect the order in which the DPs enter into an attract relation with T” (cf. Chomsky 2000: 123). In contrast, Svenonius (2001), developing ideas in Pereltsvaig (forthcoming), proposes to analyze accusative case as a uAsp on D, that is an uninterpretable instantiation of Aspect on the noun phrase. This idea is taken a step further by Bailyn (2004a), who proposes that genitive case in Russian is a uQ on D, that is an uninterpretable instantiation of Quantifier on the noun phrase, while dative and instrumental are uC and uPred, respectively. However, even though some uses of nominative appear to be structural and thus related to a functional head, as in the abovementioned proposals, it is not clear that all nominative is structural. The idea that nominative has a special status is by no means new: it goes as far back as Aristotle and Panini, and appears time and again in the linguistic writings of Hjelmslev, Jakobson and others (for a discussion of pregenerative accounts of nominative as a special case, the reader is referred to Blake 1994: 31–32, 40, 65–67).87 More recently, a more radical version of this view has been argued for by Falk (1991, 1997), Bittner and Hale (1996), Weerman (1996), Neeleman and Weerman (1999), inter alia, who claimed that the special nature of the nominative case is not restricted to the morphological component, but is true of nominative in the syntax as well. Here, I will adopt the weaker position, namely,
CASE RELATIONS
107
that nominative is unmarked morphologically, rejecting the stronger position that nominative is also different from other cases from the syntactic point of view. Several pieces of evidence support this view. First, as noted in Blake (1994: 31), Dixon (1994: 56, 62), Weerman (1996: 8–11), Falk (1997: 5) and Neeleman and Weerman (1999: 64–67), there is a cross-linguistic correlation between nominative and the lack of overt morphemes: “if any case has zero realization (or a zero allomorph) it will be nominative” (Dixon 1994: 62). In fact, this generalization goes back to Greenberg’s (1963: 95) “Universal 38. Where there is a case system, the only case which ever has only zero allomorphs is the one which includes among its meanings that of the subject of the intransitive verb” (cited in Dixon 1994: 57). Neeleman and Weerman (1999: 64–67) argue that the so-called nominative morphemes are expressions of declension class, gender and number, rather than case. Therefore, they are typically found in fusional languages, like Icelandic, Latin or Russian, but not in agglutinative languages like Turkish (Japanese is one counterexample that comes to mind). Note that this argument bears on the morphological status of the nominative but does not necessarily support the claim that nominative has a special status in syntax, as has been argued by Falk (1997) and Neeleman and Weerman (1999). In Russian, nominative is morphologically unmarked in the sense that at least in some declension classes it is associated with a zero-morpheme (in the singular), while other cases have overt morphemes. Second, nominative is the citation form, namely, the form used when the nominal appears in no syntactic context (see Blake 1994: 31, Dixon 1994: 62, Falk 1997: 5).88 For example, nominative is the form used for nominals in the following contexts: • Left Dislocations (external topics), as in (153a);89 ˇ èto? “What is this?” and Kak po-russki nazyvaetsja X? “What’s • answers to Cto the Russian for X?” questions, as in (153b); • vocatives, except where a special form, such as Bože! “God!” (cf. Bog “God.nom”) or Mam! “Mom!” (cf. mama “Mom.nom”) is available, as in (153c);90 • tag questions, as in (153d); • some fragment-utterances and predicate-less sentences, as in (153d-e). (153)
Ljuda Koˇckina ne pripominaju takuju. Luda.nom Kochkina.nom not I-remember such “Luda Kochkina, I don’t remember her.” [Darya Dontsova, “Dama s kogotkami”] ˇ èto? – Moj ˇ b. Cto talisman, porosënok Cunja. what this my.nom talisman.nom piglet.nom Chunja.nom “What is this?” – “My talisman, piglet Chunja.” [Darya Dontsova, “13 nesˇcastij Gerakla”] c. Ezžajte, Darya, domoj. go Darya.nom home “Go home, Darya.” [Darya Dontsova, “Dama s kogotkami”] d. Gadost’, pravda? abomination.nom truth.nom “Disgusting, right?” [Darya Dontsova, “Domik tjetuški lži”] a.
108
CHAPTER 5
e.
Noˇc, ulica, fonar’, apteka… night.nom street.nom street-light.nom pharmacy.nom “Night. Street. Street-light. Pharmacy.” [Blok, “Noˇc, ulica, fonar’, apteka”]
Third, single-argument agreement is generally with a nominative nominal (cf. Falk 1991: 199–207, 1997: 6–7, Weerman 1996: 6–8, Neeleman and Weerman 1999: 63–64); the same does not apply to “multiple argument agreement” (cf. Falk 1997: 24–28) and participial agreement (as in Romance languages; cf. Neeleman and Weerman 1999: 64). Fourth, nominative appears to be special in relative clauses (as discussed in Weerman 1996: 14–15 and Neeleman and Weerman 1999: 68–69). In this construction, there are two case requirements: the requirement of the matrix clause on the case of the noun and the requirement of the embedded clause imposed on the relative operator. In a situation where the two requirements conflict languages differ in the strategy they employ. One strategy, found for instance in Russian, is to allow a mismatch in case between the noun and the relative operator. Other languages, however, require that one case wins over the other. As discussed by Neeleman and Weerman (1999: 69), if one of the cases happens to be nominative, it “never overrules any other case, while any case can overrule nominative”; thus, such languages treat nominative different from other cases. There is also an additional strategy, which is employed, for example, in colloquial Russian: the case mismatch is avoided by deleting – in the phonological sense – the relative operator, which would be marked for case, and pronouncing the complementizer cˇ to “that” instead. However, this strategy is not possible in all relative clauses: it is possible if the relative operator would be marked nominative, and it is impossible if the relative complementizer is marked dative, instrumental or genitive. Contrary to what is claimed by Pesetsky (1998: 369), the speakers I have consulted (and myself as well) do not accept cˇ to-relative clauses if the relative operator would be marked accusative. Such sentences are judged severely degraded even compared to cˇ to-relative clauses with the relative operator in the nominative (which are perceived as slightly degraded by many speakers as well). These judgments are schematized in (154) below, and the relevant examples are given in (155). The strikethrough symbolizes elements which are not pronounced. (154)
(155)
a. b. c. d.
… [CP which.nom that …] * … [CP which.acc that …] * … [CP which.dat that …] * … [CP which.instr that …]
a.
Ira govorila s mal’ˇcikom, kotoryj cˇ to govorit po-ispanski. Ira spoke with boy.instr which.nom that speaks Spanish “Ira spoke with the boy that speaks Spanish.” [Pesetsky 1998: 369]
[cf. Pesetsky 1998: 369]
CASE RELATIONS
109
b.
*My ne znali o knige, kotoruju cˇ to opublikovala naša firma. we not knew about book.prep which.acc that published our company intended: “We didn’t know about the book that our company published.” [Pesetsky 1998: 369; judgment mine] kotoromu cˇ to ty pomog, ljubit pel’meni. c. *Staryj professor, old professor.nom which.dat that you helped likes dumplings intended: “The old professor that you helped likes dumplings.” [Pesetsky 1998: 369] kotorym cˇ to on sejˇcas vladeet, ran’še nazyvalsja d. *Zavod, factory.nom which.instr that he now owns formerly was-called “Krasnyj Oktjabr’ ”. Red October intended: “The factory that he now owns was formerly called Red October.” Pesetsky’s explanation for the obligatoriness of pronouncing the relative operator when it is in the dative, instrumental or genitive case is the non-recoverability of such deletion. Contra Pesetsky’s claim the accusative patterns with other nonrecoverable cases and not with nominative, which is once again set apart by this construction, further supporting the claim about its special status. Fifth, nominative has a special status in child language acquisition and heritage language attrition: nominative is the first case to appear in first language acquisition; for example, nominative is acquired before accusative and dative in Japanese and German (cf. Eisenbeiss 2001). The same is also true of child acquisition of Russian: as noted in Slobin (1966: 134, 136), the first forms of nouns in acquisition are nominative singular, with accusative and genitive appearing next around 1;11–2;0, and dative and instrumental around 2;0–2;2 (prepositional case is acquired last). As a mirror image of child language acquisition, where nominative case marking is acquired first, in heritage language attrition, nominative is retained last. For instance, speakers of Heritage (American) Russian lose much of the case system and use nominative forms to replace the lost case forms (cf. Polinsky 1996, 1997, 1998).91 Specifically, Polinsky (1996: 43, 1997, 381) notes that American Russian develops a two-case system (nominative and accusative). While the nominative becomes the multifunctional case, the accusative is specialized as the case of the indirect object and in some instances is used to encode the direct object. In particular, speakers of American Russian tend to substitute nominative forms for accusative, genitive, dative or prepositional forms required by selecting prepositions (Polinsky 1996: 32, 1997: 379–380), as well as instead of accusative forms in the direct object position (especially in the presence of an indirect object):92
110 (156)
CHAPTER 5
a.
complement of P (from Polinsky 1996: 34; cf. Polinsky 1997: 379–380) American Russian: Moj deduška byl na mirovaja vojna. my.nom grandfather.nom was on world.nom war.nom Standard Russian: Moj deduška {voeval / byl} na mirovoj vojne. my.nom grandfather.nom fought / was on world.prep war.prep “My grandfather fought in World War [II].” b. direct object of ditransitive verb (from Polinsky 1996: 41; 1997: 380–381) American Russian: Ja pokazyvaju tebja moja sobaka. I.nom show.imperf.pres you.acc my.nom dog.nom Standard Russian: Ja pokažu tebe svoju sobaku. I.nom show.perf.fut you.dat self’s.acc dog.acc “I am going to show you my dog.”
Polinsky (1995: 95–96) also notes that loss of case morphology and substitution of nominative for other cases forms are also characteristic of reduced versions of Lithuanian, Polish, and Tamil. Yet, such “nominative spread” is characteristic not only of reduced language under attrition but of colloquial language as well; it has been pointed out as a characteristic of colloquial Russian by many researchers (cf. Sirotinina 1974, Lapteva 1976, 1999, inter alia). To sum up, evidence reviewed above shows that the nominative case has a special status in the grammar: it is the unmarked form in the nominal paradigms. Given that, the question is whether all nominative nominals are structurally case marked or whether nominative can correspond to syntactic lack of case as well. Here, I choose the second option. In other words, I maintain here that nominative nominals are not a homogeneous class: some nominative nominals are licensed by a certain syntactic configuration (as nominatives or as another syntactic case such as accusative), while others are not marked for syntactic case at all. The former includes nominative subjects, as well as nominative objects (in languages like Hindi, Finnish, Icelandic, Lithuanian and Old Russian); for a detailed discussion of the latter see Smith (1994), Maling and Sprouse (1995), Taraldsen (1995) and Lavine (1999, 2000). As for nominative nominals without syntactic case, as I claim in the following section, both noun phrases in Russian copular sentences with the nom-nom belong to this set. First, consider nominative nominals that have syntactic case features. However one defines the syntactic configuration in which nominative case features can be checked, it cannot be claimed that all morphological nominatives are licensed in that syntactic configuration. For example, it has been argued extensively in the literature that oblique (i.e., non-nominative) subjects in Icelandic occupy Spec-TP, the same position as as nominative subjects, which means that nominative objects that appear with such oblique subjects cannot be in the same position. Furthermore,
CASE RELATIONS
111
Lavine (1999, 2000) and Lavine and Franks (2005) argue that nominative objects in Lithuanian and Old Russian do not appear in the canonical subject position either. A solution for this can be reached if we assume that nominative can be checked by Agree, while Move is triggered by some other feature, so that the position of nominative objects is divorced from syntactic case; this has been the approach taken in much of the recent literature, for various reasons (cf. for instance Lavine and Freidin 2002, Bailyn 2004b, inter alia). Still, a special provision is needed for structures which manifest two nominatives (at PF): clearly, two nominals cannot occupy the same syntactic position or check the same uninterpretable feature on the head. For example, this includes Hindi transitive clauses with an imperfective/non-past verb and an non-specific/ indefinite object: (157)
rootii larkiyãã girls bread “The girls eat bread.”
khaat˜ı˜ı eating
hãi. be.3.pl [Falk 1997: 13]
Falk (1997) analyzes such nominative objects are syntactically accusative, but having this marked case replaced by the unmarked nominative in the discourse component (yet, the discourse conditions that allow for such case substitution are not well-defined). Additional support for the claim that some morphologically nominative nominals are syntactically marked for a different case, such as accusative, comes from Old Russian data involving coordinated nominative and accusative objects.93 (158)
I tobe bylo v{exavši v Kiev{ brata moego jati i syna and you.dat was.n entered in Kiev brother.acc my.acc to-seize and son.acc moego i žena moja, i dom{ moj vzjati. my.acc and wife.nom my.nom and house.nom/acc my.nom/acc to-take “It was for you, having entered Kiev, to seize my brother, my son and my wife and to take my house.” [Jakab 2003: 52]
Thus, it appears that while some morphological nominative corresponds to syntactic nominative, other instances of morphological nominative correspond to a different syntactic case, such as accusative. Yet, this does not cover the whole range of morphologically nominative phrases either. As has been mentioned above, some instances of morphological nominative appear outside any syntactic context, as with Left Dislocations, tag questions, answers to “What’s this?” and “What’s the name for X?” questions, vocatives and fragment utterances. In the following section, I argue that both DPs in Russian copular sentences with the nom-nom pattern also belong to this class. Furthermore, I also propose that the exceptional appearance of the nom-nom pattern with the verb nazyvat’sja “be called” can be accounted for in terms of the nominative being the citation form.
112
CHAPTER 5 5.3.
CASE MARKING IN COPULAR SENTENCES WITH THE NOM-NOM PATTERN
At the end of the previous section, I suggested that both DPs flanking the copula in sentences with the nom-nom pattern are syntactically caseless and that this syntactic caselessness maps into nominative case at PF. The question that arises in this connection is what allows these DPs to be syntactically caseless. In Section 5.1 above, I adopted the view that abstract syntactic case is a way of identifying arguments, that is complements of specifiers of predicative expressions. As discussed in Chapter 0 above, in Russian copular sentences with the nom-nom pattern neither the pre-copular phrase nor the post-copular phrase is an argument; instead, they are merged together and that combined DP is a complement of a functional head F hosting byt’ “be”: (159)
TP DPi
T' T°
FP F'
ti F°
DP
byt’ DP
DP
Since F itself is not a predicative expression (it does not bear a theta-grid; cf. Section 4.1 above), there are no nominals in this structure that need to be licensed through abstract syntactic case, and as a result they come out of the morphological component as nominatives. In what follows, I show that this analysis of both DPs in copular sentences with the nom-nom pattern as syntactically caseless is preferable to its alternative which I call the “agreement in case” approach. According to this approach, copular sentences with the nom-nom pattern have a normal transitive structure, with the copula being a transitive predicate that takes two arguments. Moreover, an exclusive agreement-in-case strategy applies to match the case of the post-copular DP with that of the pre-copular DP (the latter is nominative for the same reasons as normal transitive subjects are):94 (160)
byt’P DP.NOM byt’' byt’
DP.NOM
Under this approach, byt’ “be” in sentences with the nom-nom pattern is an identity copula distinct from the predicative copula (in sentences with the nominstr pattern); thus, the proponents of this approach assume (or explicitly argue
CASE RELATIONS
113
for) the “identity be” analysis for the semantics of such sentences, the problematic nature of which was discussed in Section 4.3 above. In what follows, I argue that this “agreement in case” approach is also inferior to the analysis developed in this book in terms of accounting for the case marking phenomena. This notion of “agreement in case” is typically justified by comparing it with agreement in other morphological features, such as gender and number: (161)
a.
Andrej – {pobeditel’ / * pobeditel’nica} konkursa. Andrew.m winner.m / * winner.f competition.gen “Andrew is the winner of the competition.” b. Andrej i Maša – {pobediteli / *pobeditel’ } konkursa. Andrew and Masha winners.pl / * winner.sg.m competition.gen “Andrew and Masha are the winners of the competition.”
However, the flaw of this comparison is that, as shown in Section 3.5 above, the pre-copular and the post-copular DPs do not necessarily agree in grammatical features such as gender (or declension class) and number. The relevant Russian examples are repeated below:95 (162)
a.
b.
c.
Valentina Ivanova – sekretar’ gorkoma i vernyj tovarišˇc po partii. [Valentina Ivanova].f secretary.m city-committee and loyal comrade at party “Valentina Ivanova is the secretary of the city committee and a loyal party comrade.” Ètot mal’ˇcik – takaja bol’šaja nerjaxa. this.m boy.m such.f big.f sloven.f “This boy is such a big sloven.” [Comrie and Stone 1978: 77] Italjanskie studentki – narod vesëlyj. Italian.pl students.f.pl people.m.sg cheerful.m.sg “Italian (female) students are cheerful people.”
Thus, the proposed “agreement in case” cannot be assimilated to agreement in gender and number. However, the main issue which, to my mind, has not been adequately addressed by the proponents of the “agreement in case” approach is why such agreement occurs only in this particular type of copular sentence and not in other transitive structures. In other words, if “identity be” is treated as a normal transitive verb (as Bailyn and Rubin 1991, inter alia, propose) the following contrast must be accounted for: (163)
a.
Andrej byl pisatel’. Andrew.nom was writer.nom “Andrew was a writer.” b. *Andrej videl pisatel’. Andrew.nom saw writer.nom intended: “Andrew saw a writer.”
Consider now how this “agreement in case” is implemented in technical terms. The simplest solution, adopted by Bailyn and Rubin (1991), Franks (1995), Comrie
114
CHAPTER 5
(1997), inter alia, is that byt’ “be” is somehow special in that it is the only head that allows “case transmission” from its specifier to its complement. This solution clearly lacks in generality; in a way, it restates the problem that byt’ is the only verb which allows such “agreement in case” between its two arguments rather than solves it. A more technical solution has been proposed by Fowler (1997: 158–159). Following Pollock’s (1989) ideas, the Infl node is split into two nodes, which Fowler dubs IAGR and ITNS ; both of these nodes bear the feature [±finite] which “is relevant to the case-licensing properties of both [nodes] … Crucially, each has the potential to license Nominative case, providing two landing sites in a double-Nominative sentence”. The structure proposed by Fowler (1997: 158) is the following (I use DP rather than NP for uniformity’s sake throughout the book):
(164)
IAGRP [Spec]
IAGR' IAGR
ITNSP [Spec]
ITNS' ITNS
VP
DP1
V' V
DP2
In addition to potential problems raised by using an AGR node (cf. Chomsky 1995: 346–355), this analysis does not provide a satisfactory answer to the question of why double-Nominative sentences are possible only with byt’ “be” but not with other verbs like videt’ “see” or pocelovat’ “kiss”, namely it does not explain the contrast in (163) above. Another alternative within the Minimalist framework has been proposed by Bailyn and Citko (1999): both DPs in sentences with the nom-nom pattern check their nominative case in essentially the same position, namely, Spec-TP. In order to make this analysis possible, Bailyn and Citko allow multiple specifiers, which are not themselves unproblematic. According to Bailyn and Citko’s analysis, the precopular phrase undergoes case-checking in overt syntax, whereas the post-copular phrase does so at LF. The resulting LF structure from Bailyn and Citko (1999: 32) is as follows (once again, I use DP in place of Bailyn and Citko’s NP for uniformity’s sake):
CASE RELATIONS
(165)
115
TP studentk TP ‘student.NOM’ DP
T' PredP
Ivani T° ‘Ivan.NOM’ [+ Nom] [+ pres]
ti
Pred' Pred° DP tk
Note that there is a potential technical problem with Bailyn and Citko’s analysis: T can attract only the closest nominal, namely, the pre-copular noun phrase. Bailyn and Citko avoid this problem by extending the minimal domain of T via head movement of Pred to T . However, this means that they place the copula in T , regardless of the usual adverb placement tests (cf. Section 1.5 above). Moreover, their analysis lacks an explanation for why such multiple case checking of nominative can happen when the predicate is byt’ “be” and not any other verb, say, videt’ “see”; in other words, they do not have an explanation for the contrast in (163) above and cannot rule out the following structure (for the sake of exposition, I ignore here the potential intervening PredP): (166)
* TP pisatel’k TP ‘writer.NOM’ DP
T'
Andreji T° ‘Andrew.NOM’ [+Nom] [+pst] videl ‘saw’
VP ti
V' V°
DP tk
To recap, while Bailyn and Rubin (1991), Franks (1995), Comrie (1997) and others simply assume case transmission by byt’ “be”, Fowler (1997) and Bailyn and Citko (1999) rely on double case checking – against two nominative-checking heads or in multiple specifiers of the same nominative-checking head, respectively. Yet, none of these analyses explain why the copula byt’ “be”, or more precisely the identity copula (as opposed to the predicative copula), is different from any other verb. The failure of the “agreement in case” approach to account for the ungrammaticality of sentences like (163b) is a serious problem with this analysis.
116
CHAPTER 5
Under the analysis developed in this book, this problem is avoided by assuming that whenever pre-verbal and post-verbal DPs are arguments (as in (163b)), they need to be marked with case. Therefore, the exceptional status of the pre- and postcopular phrases in sentences with the nom-nom pattern is explained by the fact that these phrases are not arguments (in other words, there are no lexical verbs in such sentences), whereas pre- and post-verbal phrases in normal transitive sentences are. 5.4.
CASE MARKING IN COPULAR SENTENCES WITH THE NOM-INSTR PATTERN
The instrumental case is for contemporary grammarians a source of great anxiety. –Vinogradov, Russkij Jazyk
So far, I proposed that nominative case on both DPs in Russian copular sentences with the nom-nom pattern is a morphological default in the sense that it satisfies the morphological requirement for case-ending on Russian nominals even though these nominals do not bear abstract syntactic case, which they do not need since they are not arguments of a theta-grid-bearing head. The question addressed in this section is where the instrumental case in Russian copular sentences with the nominstr pattern comes from. Recall from Section 2.5 that instrumental post-copular phrases are bare NP, that is they do not have a DP projection. Furthermore, such post-copular NPs are complements of the lexical head v . The structure proposed for such sentences in this book is repeated below: (167)
TP DPi
T' T°
FP ti
F' vP
F° byt’ ti
v'
v° NP/AP tbyt’ <θ> <θTh>
Recall that v bears a theta-grid, although its external theta-role has only an avariable and no thematic index; the latter is acquired through -identification (cf. Section 4.4 above). Therefore, according to the definition in Section 4.1, postcopular NPs are arguments of v and thus need abstract syntactic case. Once it is established that post-copular NPs bear abstract syntactic case, the question arises as to why this case is realized morphologically as instrumental
CASE RELATIONS
117
rather than, for instance, dative or genitive. One possible answer is to treat the choice of morphological case as an idiosyncratic property of the given item, in this instance the copula. This is the approach adopted by Bailyn and Rubin (1991); for them the post-copular phrase is a complement of Pred , a functional head (they do not assume that only lexical heads can assign/check -roles or case, as I do in this book). Furthermore, they propose that Pred assigns instrumental case when it is phonetically null; in his later work, Bailyn (2004a) generalized this to other functional heads such as Q , proposing that functional heads can check case only when phonetically null (yet it is not clear to me what this connection between phonetic content and case-checking ability derives from). Because for them instrumental is assigned to a post-copular phrase by a functional head Pred , Bailyn and Rubin cannot relate this instance of instrumental case marking to any other instrumental case marking, for example, by verbs. Here, I propose that the instrumental case marking on the post-copular NP originates with the v , which is a verb, albeit a deficient (i.e., light) one. In what follows, I draw a parallel between instrumental case marking on post-copular NPs and instrumental case marking on complements of verbs and argue that instrumental in Russian is the default syntactic case for complements of verbs (whether V or v˚), in contrast with the nominative, which, as has been argued in Section 5.2, is the default morphological case. 5.4.1.
Case Marking on Internal Arguments of Verbs
Recall from Section 5.1 that I draw a distinction between two types of syntactic case: structural and inherent. The former is checked by a functional or lexical head on a specifier, whereas the latter is checked by a lexical head on its complement at Merge (i.e., no movement ever applies to check inherent case). A good example of structural case is accusative case, which I assume (following Pereltsvaig 2000, Svenonius 2002, Borer 2005) to be checked by Asp , where the relevant kind of aspect is the so-called inner aspect (cf. Verkuyl 1972, 1993, Travis 1992; this kind of aspect is also known as lexical aspect – cf. Dahl 1985, Arad 1998, Cinque 1999 – or situation aspect, cf. Smith 1991), as opposed to the outer aspect (also known as grammatical aspect or viewpoint aspect). Note that under this analysis, the nominal appearing in Spec-AspP is an argument, which means that (under the assumptions made in this book) Asp is a lexical head. This is, however, not implausible since the projection of inner aspect, the AspP, is “sandwiched” between two lexical projections, vP and VP: (168)
vP Spec v°
v' AspP Spec
Asp'
Asp°
VP
118
CHAPTER 5
According to the Proper Head Movement Generalization (Li 1990), it is impossible to move a lexical category into a functional category and then back into a lexical category. Since movement from V to Asp to v is attested, it must be the case that Asp is not a functional category (note that v is a lexical category by definition: it introduces the external/Agent argument). This, in turn, means that a noun phrase generated in Spec-VP and moved into Spec-AspP is an argument of both the V and the Asp . Note that this type of multiple argumenthood is allowed by the revised Theta Criterion, as discussed in Section 4.1 above. Another example of structural case in Russian is dative. As has been noted in the literature, dative objects typically have the meaning of an Affected/Applicative participant, including goals (169a), benefactives (169b), experiencers (169c), possessors (169d), ethical datives/“affected speaker” (169e), interested hearer (169f), and reflexive datives (169g). (169)
a.
b.
c.
d.
e.
f.
g.
Vanja poslal Kate pis’mo. Vanya sent Katya.dat letter.acc “Vanya sent Katya a letter.” Vanja spëk Kate pirog. Vanya baked katya.dat cake.acc “Vanya baked Katya a cake.” Vanja {poljubilsja / ostoˇcertel} Kate. Vanya caught-the-fancy / hundred-devils-ed Katya.dat “Vanya {caught Katya’s fancy / got on Katya’s nerves}.” Vanja slomal Kate cˇ asy. Vanya broke Katya.dat watch “Vanya broke Katya’s watch.” Vanja vošël mne posredi fil’ma. Vanya.nom entered me.dat in-the-middle-of film “Vanya entered in the middle of the film (and it annoyed me).” Ja tebe razmazal ego po stenke. I you.dat spread him on wall “(You see?!) I beat him up pretty good.” Vse begajut, volnujutsja, a on sidit sebe i raskladyvaet pas’jans. everybody run worry but he sits self.dat and plays solitaire “Everybody’s running around worried and he just sits around and plays solitaire.”
Since the various types of datives are incompatible with each other, as shown below, it is reasonable to assume that they all appear in the same structural position, the position where dative case is checked. Here, the incompatibility of different types of datives is illustrated by pairs of a goal and a possessor (170a), an ethical dative and an experiencer (170b), and a possessor and a reflexive dative (170c): (170)
a.
*Vanja poslal Maše mne podarok. Vanya sent Masha.dat me.dat gift.acc intended: “Vanya sent Masha my gift.”
CASE RELATIONS
b.
c.
*Vanja mne Vanya me.dat intended: “Vanya *Vanja sidit Vanya sits intended: “Vanya
119
poljubilsja Maše. caught-fancy Masha.dat caught Masha’s fancy (and it annoyed me).” mne sebe na divane. me.dat self.dat on couch just sits around on my couch.”
Many dative-object verbs (and only such verbs) can be paraphrased by a combination of a ditransitive verb and a noun corresponding to the root of the dative-selecting verb. For example, doverjat’ “trust” and pomogat’ “help” can be paraphrased with okazat’ doverie literally “render trust”, and okazat’ pomošcˇ literally “render help”, respectively; similarly, navredit’ “harm” can be paraphrased by nanesti vred “bring harm”.96 (171)
a.
Krasnyj Krest pomogaet postradavšim ot zemletrjasenija. Red Cross helps victims.dat from earthquake.gen “The Red Cross helps earthquake victims.” b. Krasnyj Krest okazyvaet pomošˇc postradavšim ot zemletrjasenija. Red Cross renders help.acc victims.dat from earthquake.gen “The Red Cross renders help to earthquake victims.”
Furthermore, as noted in Blauvelt (1980) and Fowler (1996), dative-object verbs do not passivize (unlike verbs that take instrumental or genitive objects). Thus, it has been proposed that dative objects do not occupy the same syntactic position as accusative or other oblique objects; however, there is no agreement in the literature as to whether dative objects are higher or lower than accusative ones. For example, Franks (1995) proposed that dative objects are higher than accusative objects, whereas Bailyn (1995) argued that dative objects are lower than accusative objects. In this book, I side with Franks (1995) on this issue, taking dative objects to be higher than accusative ones. Three pieces of evidence support the claim that dative objects (in what follows, Goals are illustrated) are higher than accusative ones: reciprocal binding, quantifier binding and ordering of wh-phrases in multiple wh-questions (for a discussion of the latter test, cf. Blond 1996).97 First, a dative object can bind a reciprocal anaphor inside the accusative object but not vice versa: (172)
a.
Vanya poslal vraˇcami pacièntov drug drugai . Vanya sent doctors.dat patients.acc each other’s “John sent the doctors each other’s patients.” b. *Vanja poslal pacièntovi vraˇcam drug drugai . Vanya sent patients.acc doctors.dat each other’s “*John sent each other’s doctors the patients.”
Second, a quantifier contained in the dative object can bind a variable inside the accusative object but not vice versa:
120 (173)
CHAPTER 5
a.
Vanja poslal každomu vraˇcui Vanya sent each doctor.dat “John sent each doctor his patients.” b. *Vanja poslal každogoi paciènta Vanya sent each patient.acc “*John sent his doctor each patient.”
egoi his
pacièntov. patients.acc
egoi his
vraˇcu . doctor.dat
Third, with multiple non-D-linked wh-phrases the dative one must precede the accusative one (the assumption here is that Superiority effects reflect an instance of Attact Highest): (174)
a.
Komu cˇ to Vanja dal? who.dat what.acc Vanya gave “Who(m) did John give what?” ˇ b. *Cto komu Vanja dal? what.acc who.dat Vanya gave “*What did John give who(m)?”
Therefore, I take dative objects to appear higher than accusative objects; although the exact label of the projection hosting dative objects has been debated in the literature, it is fairly clear that dative is checked in a specifier of that projection rather than in the complement position. Hence, dative is a structural case. Since the post-copular NP in sentences with the nom-instr pattern is a complement of v rather than a specifier, it cannot appear with a structural case; its only option is inherent case. Apart from the locative/prepositional case, which in Russian occurs only on complements of prepositions, there are two morphological cases that correspond to inherent syntactic case checked by verbs: genitive and instrumental. Here, I argue that the choice among genitive and instrumental is not as arbitrary as has been assumed so far. In this, my analysis differs crucially from that of Bailyn and Rubin (1991): instead of assuming that instrumental marking on post-copular phrases is an idiosyncratic property of the head that checks this inherent case, I propose that instrumental is chosen as the default inherent case for verbal complements. 5.4.2.
Inherent Case Marking in Russian
It has been traditionally assumed that the choice among inherent cases in Russian is idiosyncratic. However, it has also been noted (especially by Blauvelt 1980) that there are some regularities in the case checking possibilities for different verbs. Setting aside the dative-object verbs, discussed in the previous section, the only other possible inherent cases for verbal complements are genitive and instrumental. Let us consider them in turn. First, consider the inherent genitive case. As discussed in Neidle (1982: 400), there two kinds of verbs that take genitive objects: those that take either genitive or accusative objects (including xotet’ “want”, ždat’ “wait”, iskat’ “look for”, prosit’ “ask for”, trebovat’ “demand”, and several others; cf. (175a)) and others that take
CASE RELATIONS
121
genitive objects obligatorily (including dobivat’sja “achieve”, dostigat’ “attain, reach” (and its perfective counterpart dostignut’), želat’ “wish”, dožidat’sja “wait for”, and several others; cf. (175b)).98 (175) a. optional genitive
b.
Andrej išˇcet { sˇcastja / (svoë) sˇcastje }. Andrew looks-for happiness.gen / self’s happiness.acc “Andrew looks for (his) happiness.” obligatory genitive Pust’ grjadušˇcie pokolenija dostignut {sˇcastja / * sˇcastje}. Let future generations attain happiness.gen / * happiness.acc “‘Let future generations attain happiness.” [Chekhov, cited in Zolotova 1988: 35]
Following Neidle (1982), who in turn develops ideas in Quine (1960: 219–222), I maintain that when an object is marked genitive, it has a certain “quantificational force”, which is unavailable for accusative objects; furthermore, “intentional objects” of Quine (1960) always appear in the genitive case. Although there are some problems with Neidle’s (1982) analysis as she formulated it, her key generalization – that inherent genitive is related to quantificational force – appears to be on the right track.99 This generalization has been further explored in Pesetsky (1982), Pereltsvaig (1998) and Bailyn (2004a), inter alia, where the analysis has been extended to other instances of non-canonical (i.e., non-adnominal) genitives, including partitive genitive, large quantity genitive, non-argument bare genitive, and the genitive of negation. The main thrust of this approach to genitive case in Russian is that in all cases of apparently non-adnominal genitives, the genitive is assigned by a phonetically null quantifier, which is in its turn assigned accusative by the verb in the usual way. In other words, verbs never assign “inherent genitive” directly. The existence of a null quantifier accounts for the above-mentioned generalization that non-canonical genitives (including the so-called “inherent genitive”) correlate with quantificational force. There are some exceptions to this general correlation between inherent genitive and quantificational force, noted by Blauvelt (1980): in additional to verbs with intensional meaning, genitive-object verbs include some verbs that used to take ablative case in earlier stages of the language; these verbs have a meaning of removal or approach. Since the copula has neither the quantificational/intentional nor the ablative meaning, it is not surprising that post-copular NPs in Russian do not appear in the genitive. The “elsewhere” option for complements of verbs in Russian is the instrumental case. The most striking fact about verbs that appear with instrumental objects is that they do not belong to a unique lexical semantic class in the same way that verbs that take genitive or dative complements do. In addition to copula-like verbs (discussed in detail in the following section), verbs that take instrumental objects include verbs of the
122
CHAPTER 5
following classes (unless otherwise marked, only imperfective members of aspectual pairs are listed here and examples are from Darya Dontsova’s “Vynos Dela”): (176)
a.
b.
c.
d.
e.
f.
g.
h.
i.
verbs of management and control: dirižirovat’ “conduct (orchestra)”, komandovat’ “command”, pravit’ “govern”, rasporjažat’sja “handle”, rukovodit’ “supervise”, upravljat’ “manage”, vedat’ “manage”, vlastvovat’ “control”, zavedovat’ “manage”, etc.100 Lev Konstantinoviˇc Skorkin upravljal xozjajstvom krepkoj rukoj L. K. S. managed economy.instr strong hand “Lev Konstantinovich Skorkin managed the local economy with a strong hand.” verbs of interest: interesovat’sja “be interested in”, zanimat’sja “occupy oneself ” Pust’ zanimaetsja pustym rassledovaniem. let-him occupies-oneself empty.instr investigation.instr “Let him busy himself with a pointless investigation.” verbs of possession: izobilovat’ “replete with”, obladat’ “have”, raspolagat’ “have at one’s disposal”, vladet’ “own”, etc. Esli obladaete interesnoj informaciej – prixodite! if you-have interesting.instr information.instr come.imper “If you have interesting information, do come!” verbs of acquiring possession: obzavodit’sja “get hold of ”, poživit’sja “steal possession.perf”, razdobyt’sja “take possession.perf”, razživat’sja “get hold of ”, etc. Bol’šinstvo… obzavelis’ vtorym podborodkom, moršˇcinami ... most got-hold-of second.instr chin.instr wrinkles.instr “The majority (of them) got a second chin, wrinkles…” verbs of transferring possession: nadeljat’ “endow”, nagraždat’ “present (with gift, medal)”, obdeljat’ “not endow”, obespeˇcivat’ “provide”, obxodit’ “not endow”, odarivat’ “give as gift”, otplaˇcivat’ “pay”, platit’ “pay”, snabžat’“provide”, ssuždat’ “give as a loan”, etc. I vnov’ sud’ba odarila carskim podarkom. and again fate gave-as-gift royal.instr gift.instr “And again fate gave (her) a royal gift.” verbs of exchange: delit’sja “share”, menjat’sja “exchange”, obmenivat’sja “exchange”, perekidyvat’sja “bandy”, torgovat’ “trade”, etc. Devuška torgovala soboj na plošˇcadi u trëx vokzalov. girl traded self.instr on square next-to three train-stations “The girl traded in her own body on the square at the three train stations.” verbs of changing the object of possession: zamenjat’ “substitute”, etc. Ja zamenil sjurtuˇcok kurtkoj. I substituted surtout.acc coat.instr “I substituted a coat for the surtout.” [Turgenev, cited in Zolotova 1988: 247] verbs of smell and light emission: blestet’ “shine”, nesti “stink”, otdavat’ “smell”, otsveˇcivat’ “gleam”, páxnut’ “smell”, paxnút’ “smell.smlf”, tjanut’ “smell”,vejat’ “smell”101 Iz temnovatogo koridora paxnulo syrost’ju i xolodom. from darkish corridor smelled dampness.instr and cold.instr “From the darkish corridor it smelled of dampness and cold.” verbs of body-internal motion with no change of state: kaˇcat’ “dangle”, kivat’ “nod”, krutit’ “twist”, maxat’ “wave”, morgat’ “blink”, motat’ “waggle”,
CASE RELATIONS
j.
k.
l.
m.
n.
123
pobrjakivat’ “dingle-dangle”, požimat’ “shrug”, ševelit’ “move”, topat’ “stamp”, trjaxnut’ “shake up.smlf”, vertet’ “spin”, vzmaxnut’ “wave.smlf”, etc.102 Ja udruˇcënno pokaˇcala golovoj. I dejectedly shook head.instr “I dejectedly shook my head.” verbs of illness and suffering: bolet’ “be ill”, muˇcit’sja “suffer”, stradat’ “suffer”, zaražat’ “infect”, zarazit’sja “get infected.perf” Devoˇcka stradala blizorukost’ju. girl suffered myopia.instr “The girl suffered from myopia.” verbs of filling: napolnjat’sja “fill itself”, polnit’sja “fill itself”, usypat’ “fill (with grain, etc.)”, zabivat’ “stuff”, zabrasyvat’ “fill by throwing”, zasypat’ “fill (with grain, etc.)”, zavalivat’ “fill (with objects)”, etc. Komnata napolnilas’ tixoj muzykoj. room filled-itself quiet.instr music.instr “The room filled with quiet music.” verbs of explanation: obuslovlivat’ “condition”, obuslovlivat’sja “be conditioned”, ob”jasnjat’ “explain”, ob”jasnjat’sja “be explained”, opredeljat’ “determine”, opredeljat’sja “be determined”, etc. Jarostnye napadki na “dymoglotstvo” … deti ob”jasnjajut blagimi frenzied attacks on smoke-swallowing children explain best.instr poryvami. intensions.instr “My children justify their frenzied attacks on my smoking by their best intensions.” verbs of emotional treatment: brezgovat’ “strain at”, dorožit’ “cherish”, gordit’sja “be proud”, ljubovat’sja “admire”, naslaždat’sja “relish”, obol’šˇcat’sja “be beguiled”, pomykat’ “order about”, prenebregat’ “neglect”, tjagotit’sja “be born down”, vosxišˇcat’sja “admire”, xvalit’sja “brag”, xvastat’sja “brag”, etc. My posideli paru minut, naslaždajas’ tëplym aprel’skim veˇcerom. we sat couple minutes enjoying warm.instr April.instr evening.instr “We sat for a couple of minutes enjoying a warm April evening.” frighten-type psych verbs: divit’“marvel”, grozit’ “threaten”, poražat’ “amaze”, pugat’ “frighten”, radovat’ “gladden”, smešit’ “cause to laugh”, smušˇcat’ “confuse”, udivljat’ “surprise”, ugrožat’ “threaten”, utešat’ “console”, utomljat’ “tire”, vosxišˇcat’ “delight”, zabavljat’ “amuse”, etc. Širokij koridor radoval glaz “bodrjašˇcim” temno-zelënym kolerom sten. wide corridor gladdened eye invigorating.instr dark-green.instr color.instr walls “The wide corridor gladdened the eye with an “invigorating” dark-green color of its walls.”
Even a cursory examination of this list reveals that it is hardly possible to find a common denominator for all (or even most) subclasses of verbs that take instrumental objects. Instead, it appears then that instrumental is the “elsewhere” case marking for objects that get inherent case. The idea that instrumental case marking is somehow the default in Russian is not entirely new. For example, Jakobson (1958/1971) developed a theory of case features in which the instrumental is the second most unmarked case after nominative (although see Franks 1995 for a different system of case features); he used the minimal difference between nominative and
124
CHAPTER 5
instrumental in terms of the [± Marginal] feature to account for a number of contexts where nominative and instrumental appear to alternate, including not only the postcopular phrases in the center of the discussion in this book, but also Agents (which are nominative in actives and instrumental in passives), as in (177), and the socalled weather-subjects in impersonal sentences (which are nominative if the subject is construed as agentive/personified and instrumental if the subject is construed as non-agentive), as in (178); examples are from Zolotova (1988: 234–235).103 (177)
a.
Glavnyj arxitektor podpisyvaet projekt. chief.nom architect.nom sign.3.sg project.acc “The chief architect signs the project.” [Zolotova 1988: 235] b. Projekt podpisyvaetsja glavnym arxitektorom. project.nom sign.pass chief.instr architect.instr “The project is signed by the chief architect.”
(178)
a.
Veter sorval krysu. wind.nom tore-off roof.acc “The wind tore off the roof.” [Zolotova 1988: 234] b. Vetrom sorvalo krysu. wind.instr tore-off roof.acc “The roof was torn off by the wind.” [Babby 1980: 53]
A similar idea is entertained by Kilby (1986), who proposes to treat instrumental case marking in Russian as the case for all the “leftovers” that cannot be classified into well-defined semantic classes which are assigned other case markings. The same idea is developed in the framework of Relational Grammar (RG; see Channon 1987, Fowler 1987, 1988, 1997), where Russian instrumental is related to the notion of Chômeur. In RG, Chômeur is not a unified notion: an element with any grammatical function can become a Chômeur. For example, the by-phrase in English passives or the instrumental agent in Russian passives is a 1-Chômeur (i.e., Subject turned Chômeur), whereas the second DP in a double object construction (e.g., a book in I gave John a book) or the with-object in spray/load alternation (e.g., with butter in John spread the bread with butter) is a 2-Chômeur (i.e., a Direct Object turned Chômeur). Moreover, RG relates morphological case marking (and the choice of prepositions) to the grammatical function of the element at the final stratum; for example, in English 3 (i.e., Indirect Object) is marked by the preposition to. Therefore, the claim that instrumental in Russian is a case marking of Chômeurs is equivalent to saying that Instrumental marks everything that is not a Subject, a Direct Object, or an Indirect Object at the final stratum. Even though I do not take instrumental in Russian to be the default syntactic case in general, I do take it to be the default inherent case. This claim is not incompatible with the idea that accusative is the second most unmarked structural case after nominative (note that this idea is further supported by the facts concerning Argument Case Shift in American Russian; see fn. 92). The analysis proposed here that takes accusative on objects to be structural case checked in a specifier position
CASE RELATIONS
125
and instrumental on objects to be inherent case checked in a complement position implies that accusative objects are structurally higher than instrumental objects. This prediction is borne out, as shown by the following contrasts with respect to reciprocal binding and quantifier-variable binding: (179)
a.
Oni po ošibke nagradili soldat medaljami drug druga. they in error presented soldiers.acc medals.instr each other “They erroneously presented the soldiers with each other’s medals.” b. *Oni po ošibke nagradili drug druga medaljami soldat. they in error presented each other.acc medals.instr soldiers intended: “*They erroneously presented each other with the soldier’s medals.” or “They erroneously presented the soldiers with each other’s medals.”
(180)
a.
Oni nagradili každogo soldata ego medalju. they presented each.acc soldier.acc his medal.instr “They presented each soldier with his own medal.” b. *Oni nagradili každoj medalju eë soldata. they presented every.instr medal.instr its.acc soldier intended: “*They presented its soldier with ever medal.” or “They presented each medal to the soldier it belonged to.”
Returning to Russian copular sentences with the nom-instr pattern, the post-copular NP is a complement of v , but it does not fit into the lexical semantic classes of verbs which take genitive (or dative) objects (“quantitative or intentional objects” and “Goals/Affectees”, respectively). Therefore, it appears with the “elsewhere” inherent case marking, that is, the instrumental. Crucially, since the copula verb byt’ “be” is taken to be an instance of v , rather than a functional head Pred (as in Bailyn and Rubin’s 1991 analysis), it is possible to draw a parallel between its case checking properties and those of other verbs that check an inherent case. Thus, the instrumental case on the post-copular NP is not completely unexpected, but rather follows a more general patterns of inherent case checking in the language. To sum up, under the analysis developed in this book, there is no need for special stipulations about the case assigning properties of the head responsible for the instrumental case of the post-copular NP in the relevant type of copular sentences. 5.5.
CASE-MARKING WITH COPULA-LIKE VERBS IN RUSSIAN
In addition to the pure copula byt’ “be”, Russian has some verbs that appear to share the copula’s meaning; I refer to such verbs as “copula-like verbs” (in Russian, svjazocˇ nye glagoly lit. “linking verbs”). An exhaustive list of copula-like verbs in Russian is difficult to compile due to an unclear divide between copula-like and fully lexical verbs, but a list of most commonly cited copula-like verbs includes the following verbs (only imperfective counterparts are listed here):
126
CHAPTER 5
(181) videt’sja “look, appear”, voobražat’sja “imagine”, vygljadet’ “appear”, vyxodit’ “come out”, delat’sja “become”, dovodit’sja “be an X relative”, žit’ “live as, remain”, zvat’sja “be called”, igrat’ “play as”, kazat’sja “seem”, nazyvat’sja “be called, be termed”, stat’ “become”, obernut’sja “turn into”, obratit’sja “turn into”, okazyvat’sja “turn out”, ostavat’sja “remain”, pokazat’sja “come across as”, polucˇ at’sja “come out”, pocˇ itat’sja “be revered as”, predstavljat’sja “seem”, prevrašcˇ at’sja “turn into”, prikidyvat’sja “act as”, pritvorjat’sja “pretend to be”, rabotat’ “work as”, skazyvat’sja “claim to be”, slyt’ “gain the character of; have the reputation of”, služit’ “serve as”, smotret’sja “look like”, sostojat’ “work as” (archaic), stanovit’sja “become”, scˇ itat’sja “be reputed as”, ustraivat’sja “get a job as”, cˇ islit’sja “be itemized as, be considered as”, cˇ uvstvovat’ sebja “feel”, javljat’sja “be” (formal) Some examples of copula-like verbs in use (from Darya Dontsova’s “Lampa razyskivaet Alladina”) are given below: (182)
a.
b.
c.
d.
e.
f.
Otˇcego-to ja poˇcuvstvovala sebja {sˇcastlivoj / *sˇcastlivaja}. for-some-reason I felt self happy.instr / *happy.nom “For some reason I felt happy.” Situacija pokazalas’ {komiˇcnoj / *komiˇcnaja}. situation seemed comical.instr / *comical.nom “The situation appeared comical.” Esli oˇcen’ dolgo prikidyvat’sja {nesˇcastnoj / *nesˇcastnaja}, if very long act-as unhappy.instr / *unhappy.nom to v konce koncov {takovoj / *takovaja} i staneš. then in-end-of-ends such.instr / *such.nom emph you-will-become “If you act as an unhappy person for a very long time, at the end you will become one.” Sejˇcas popritvorjajus’ minut desjat’ {sotrudnicej / *sotrudnica}. now I-will-pretend-to-be minutes ten staff-member.instr / * staff-member.nom “I will pretend now to be a staff-member for about ten minutes.” Kvartira vygljadela {zapušˇcennoj / *zapušˇcennaja}. apartment looked unkempt.instr / *unkempt.nom “The apartment looked unkempt.” Ona okazalas’ {gluboko nesˇcastnoj ženoj / *gluboko nesˇcastnaja žena}. she turned-out deeply unhappy.instr wife.instr /* deeply unhappy.nom wife.nom “She turned out to be a deeply unhappy wife.”
As mentioned in Section 1.1 above and is illustrated in the above examples, copulalike verbs in standard (i.e., non-colloquial) Russian do not allow the nom-nom pattern and always appear with the nom-instr pattern (or with short-form adjectives as the head of the post-copular phrase); there are only two exceptions to this, to which I return shortly below.104 This is true even of the copula-like verb javljat’sja “be” (which appears to be synonymous with byt’ “be”, but stylistically different in that it is associated with a higher register): it too does not allow the nom-nom pattern (examples are again from Darya Dontsova’s “Lampa razyskivaet Alladina”):
CASE RELATIONS
(183)
a.
b.
c.
127
Alla javljaetsja {moej vnuˇckoj / *moja vnuˇcka}. A. is my.instr grand-daughter.instr/ *my.nom grand-daughter.nom “Alla is my grand-daughter.” Oni nikogda ne javljalis’ {moimi krovnymi det’mi / *moi they never not were my.instr blood.instr children.instr / *my.nom krovnye deti}. blood.nom children.nom “They never were my biological children.” Sous dlja zapravki salata ne javljaetsja {osnovnoj damskoj tajnoj sauce for dressing salad not is main.instr ladies’.instr secret.instr / / *osnovnaja damskaja tajna}. *main.nom ladies’.nom secret.nom “Salad dressing is not the main ladies’ secret.”
Consider how these facts can be explained under the analysis proposed in this book. Since copula-like verbs are lexical heads, the constituents that they select are arguments by definition. Therefore, copula-like verbs cannot take a symmetrical structure as the complement. Thus, the only option for a copula-like verb is to appear in a structure similar to that of the copular sentences with the nom-instr pattern, the same structure that leads to instrumental complements. The only two exceptions to the generalization that copula-like verbs always appear with the nom-instr pattern are the verbs stat’ “become” (which in colloquial Russian optionally appears with the nom-nom pattern) and nazyvat’sja “be called” (which optionally appears with the nom-nom pattern even in standard Russian). First, consider the verb stat’ “become”. In standard Russian it appears only in the nom-instr pattern (examples are once again from Darya Dontsova’s “Lampa razyskivaet Alladina”, judgments for standard Russian): (184)
a. Lëlja v odin god stala b.
{krugloj sirotoj / *kruglaja sirota }. L. in one year became complete.instr orphan.instr / *complete.nom orphan.nom “In one year, Lelja became a complete orphan.” Vsë tajnoe stanovitsja {javnym /*javnoe}. everything secret becomes visible.instr /*visible.nom “All secrets come out sooner or later.”
However, as noted in Bailyn and Rubin (1991: 121, fn.11), in colloquial Russian stat’ “become” allows the nom-nom pattern. (185)
a.
Saša stal delovym. S. became business-like.instr “Sasha became business-like.” b. Saša stal delovoj. S. became business-like.nom “Sasha became business-like.”
According to Bailyn and Rubin, “as opposed to (a), which indicates a true change in Sasha’s state, (b) implies that the world around has changed in such a way that
128
CHAPTER 5
Sasha is now inherently business-like”; this judgment is shared by my consultants and myself. Similarly, the sentence in (184a) below with the nom-instr pattern is appropriate in standard Russian, whereas in colloquial Russian the sentence in (184b) below with the nom-nom pattern is more appropriate since nothing has changed about Crimea itself, but rather the world around it has changed so that it became part of the Ukraine instead of Russia:105 (186)
a.
Krym stal Ukrainoj. Crimea became Ukraine.instr “Crimea became (part of) Ukraine.” b. Krym stal Ukraina. Crimea became Ukraine.nom “Crimea became (part of) Ukraine.”
In order to account for these data, I propose that in colloquial Russian stat’ “become” has become an aspectual variant of byt’ “be”, that is a true copula; in other words, it can be merged as a functional category (i.e., F ). The other exception to the generalization that copula-like verbs always appear with the nom-instr pattern is the verb nazyvat’sja “be called”, which sometimes allows the nom-nom pattern, depending on the word order. As illustrated below, if the new term introduced by the verb nazyvat’sja “be called” precedes the verb, the case on the noun phrase denoting the new term must be instrumental: (187)
a. Borom b.
nazyvaetsja les, v kotorom rastut xvojnye derevja. pinery.instr is-termed forest.nom in which grow.pl coniferous.nom trees.nom “A forest in which grow coniferous trees is termed a pinery.” *Bor nazyvaetsja les, v kotorom rastut xvojnye derevja. pinery.nom is-termed forest.nom in which grow.pl coniferous.nom trees.nom intended: “A forest in which grow coniferous trees is termed a pinery.”
However, if the new term introduced by the verb nazyvat’sja “be called” follows the verb both the instrumental and the nominative are possible, but the nominative is clearly the preferred option:106 (188)
a. b.
Les, v kotorom rastut xvojnye derevja, nazyvaetsja bor. forest.nom in which grow.pl coniferous.nom trees.nom is-termed pinery.nom “A forest in which grow coniferous trees is termed a pinery.” ?Les, v kotorom rastut xvojnye derevja, nazyvaetsja borom. forest.nom in which grow.pl coniferous.nom trees.nom is-termed pinery.instr “A forest in which grow coniferous trees is termed a pinery.”
As noted in Nichols (1981: 237–240) and confirmed by my consultants, this preference is particularly strong if the new term is “a foreign or exotic word”, “a title, epithet, or stock phrase”; it is also used when the new term is “an unfamiliar or new word, or when clarity is necessary, as when speaking to a child” (example in (187a) is from Darya Dontsova’s “Žaba s košel’kom”):
CASE RELATIONS
(189)
129
a. Ètot serviz b.
nazyvaetsja {“Otdyx v derevne”/ *“Otdyxom v derevne”}. this dinner-set is-called “Repose.nom in village” / *“Repose.instr in village” “This dinner-set is called ‘Repose in a village’.” Ètot serviz nazyvaetsja {??paradnyj / paradnym}. this dinner-set is-called ceremonial.nom / ceremonial.instr “This dinner-set is called a ceremonial one.”
In line with the above-mentioned observation, “place names have a much greater propensity to be in the nominative. Foreign and exotic names are normally in the nominative, even if declinable” (Nichols 1981: 23–7240). The following contrast highlights that novelty of the term introduced by the verb nazyvat’sja “be called” and the possibility of misconstruing it is at the core of the the use of the nominative: (190)
a. b.
Bljudo iz risa i mjasa s ovošˇcami nazyvaetsja {plov / *plovom}. dish from rice and meat with vegetables is-called pilau.nom / *pilau.instr “A dish made of rice with meat and vegetables is called pilau.” Plov iz risa i baraniny nazyvaetsja {?plov po-uzbekski / plovom pilau from rice and mutton is-called ?pilau.nom à la Uzbek / pilau.instr po-uzbekski}. à la Uzbek “A pilau made of rice and mutton is called pilau à la Uzbek.”
A similar pattern is found in the naming construction: (191)
Vobšˇce-to ego zovut Džon, no russkie druz’ja zovut ego Vanej. in-general him called John.nom but Russian friends call him Vanya.instr “In general, his name is John, but his Russian friends call him Vanya.”
The key to explaining these facts lies with the claim made in Section 5.2 above, namely that nominative is the default morphological case in Russian, or in Jakobson’s (1936/1984: 67) words, “vehicle of the pure naming function”; hence, it is the case chosen for citation forms. The more likely the term introduced by the verb nazyvat’sja “be called” to be new and unfamiliar to the hearer, the more the pressure to use the citation form, namely the nominative. The word order contrasts in (187)–(188) are explained as follows: it is generally the case that new information appears in Russian clause-finally, while old information appears clause-initially. Hence, when the term introduced by the verb nazyvat’sja “be called” follows the verb, as in (188), it is assumed to be unknown to the hearer; in order to ease the comprehension and to avoid ambiguity, the syntactic instrumental case checked by the verb is replaced by the citation form (i.e., the nominative form) in the morphological component (or the PF). In contrast, when the term introduced by the verb nazyvat’sja “be called” precedes the verb (as in (187)), it is assumed to be known to the hearer and therefore there is no need for the citation nominative form to override the syntactic case checked by the verb and the instrumental case surfaces. Note that this pragmatically-oriented analysis allows us to account for the fact that Nichols’ observations “are not absolute rules” (Nichols 1981: 240).
130
CHAPTER 5
This optional use of the citation form overriding the case checked by the selecting head is similar to the use of the citation form with foreign toponymics in Russian. Even though there is a fair amount of variation between different classes of nonnative toponymics in Russian, depending on their phonetic form, origin and frequency of use, most toponymics are used in the citation form (i.e., with the morphological nominative), even when selected by a preposition or a verb that requires a different case (for a detailed discussion, see Graudina et al. 1976: 143–149). For instance, even though the inflected forms of Osaka are also attested, in the following example the nominative case is used instead of the prepositional case normally assigned by the preposition v “in” (thus, the expected form would be v Osake “in Osaka.prep”, which is also attested though more rarely than the citation form).107 (192)
teatr kukol v Osaka theater.nom puppets.gen in Osaka.nom “puppet theater in Osaka”
[Graudina et al. 1976: 146]
Before concluding the discussion of copula-like verbs in Russian, consider again the verb javljat’sja “be”, which, as mentioned above, is fully synonymous with byt’ “be” (but belongs to a higher register) and yet it does not allow the nom-nom pattern. The question is why javljat’sja “be” cannot be a functional head V . My answer to this question has to do with the morpheme -sja, which functions as an intransitivizing morpheme creating passives, middles, unaccusatives, or unergatives. The most common approach to this morpheme in the literature analyzes it as a morpheme that operates on the thematic grid of the verb, either absorbing a -role or focusing a -role (cf. Isaˇcenko 1960, Babby 1975b, Dik 1983, and A. Williams 1993, inter alia). Under this approach a verb with -sja must have a theta-grid that -sja can operate on; therefore, by the definition of a functional head adopted in this book such a verb cannot be a functional head.108 Therefore, javljat’sja “be” cannot appear with the nom-nom pattern because it cannot be merged as a functional head due to the presence of -sja. 5.6.
CASE MARKING ON DEPICTIVES IN RUSSIAN
Although the main focus of this book is on primary predication in copular sentences, in this section I briefly review how the analysis proposed here can be extended to depictives. As mentioned in Section 4.5 above, depictives in Russian have two alternative case marking patterns: they can appear in the instrumental case or in the same case as the argument they are predicated of (such as nominative or accusative): (193)
a.
Maša tancevala goloj. Masha.nom danced nude.instr “Masha danced nude.: b. Maša tancevala golaja. Masha.nom danced nude.nom “Masha danced nude.”
– instrumental case
– “agreeing case”
131
CASE RELATIONS
(194)
a.
Ja našël Mašu I.nom found Masha.acc “I found Masha drunk.” b. Ja našël Mašu I.nom found Masha.acc “I found Masha drunk.”
pjanoj. – instrumental case drunk.instr pjanuju. drunk.acc
– “agreeing case”
According to Hinterhölzl (2001), the distribution of the two alternative case marking patterns with depictives depends on the semantic type of the predicate: bounded depictives (e.g., bol’noj “sick”, pjanyj “drunk”, serdityj “angry”, etc.) appear either in the “agreeing case” or in the instrumental, partially non-bounded depictives (e.g., molodoj “young”, naivnyj “naïve”, spelyj “ripe”, obrazovannyj “educated”, etc.) appear only in the instrumental, and unbounded predicates (e.g., umnyj “intelligent”, glupyj “stupid”, etc.) are impossible as depictives at all: (195)
a.
Ja videl Vanju {serditym / serditogo}. I saw Vanya.acc angry.instr / angry.acc “I saw Vanya angry.” b. Ja videl Vanju {molodym / *molodogo}. I saw Vanya.acc young.instr / *young.acc “I saw Vanya young.” c. Ja videl Vanju {*umnym / *umnogo}. I saw Vanya.acc *intelligent.instr / *intelligent.acc intended: “*I saw Vanya intelligent.”
Although Hinterhölzl himself focuses on depictives headed by long-form adjectives, his analysis can be used to explain Fowler’s (1997: 155–157) observation that “the only truly reliable examples of agreeing depictives are adjectives”, while depictives headed by nouns never exhibit agreeing case. In Hinterhölzl’s terminology, this is so because nouns do not typically denote bounded predicates. (196)
a.
On vernulsja {ugrjumym / ugrjumyj}. he.nom returned gloomy.instr / gloomy.nom “He returned gloomy.” b. On vernulsja {geroem / *geroj}. he.nom returned hero.instr / *hero.nom “He returned a hero.”
Although Hinterhölzl’s analysis goes a long way in accounting for the distribution of case marking in Russian depictives, it leaves much of the data unaccounted for. For instance, if the distribution of the two case marking patterns depends purely on the semantic class of the predicate, it is not clear why the syntactic position of the argument that the secondary predicate is associated with should matter; in particular, while both case marking patterns are possible if the secondary predicate is associated with a nominative or an accusative argument (cf. (191)–(192) above),
132
CHAPTER 5
only the “agreeing depictives” and not the instrumental ones can be associated with dative or genitive as well.109 (197)
a.
Ivan pomog Olegu pjanomu. Ivan.nom helped Oleg.dat drunk.dat “Ivan helped Oleg, who was drunk.” b. *Ivan pomog Olegu pjanym. Ivan.nom helped Oleg.dat drunk.instr intended: “Ivan helped Oleg, who was drunk.”
(198)
a.
Ivan dožidalsja Olega trezvogo. Ivan.nom waited-for Oleg.gen sober.gen “Ivan waited for Oleg, who would be sober.” b. *Ivan dožidalsja Olega trezvym. Ivan.nom waited-for Oleg.gen sober.instr intended: “Ivan waited for Oleg, who would be sober.” (OK on the reading: “Ivan, being sober, waited for Oleg.”)
An alternative analysis of case marking with depictives in Russian has been proposed by Bailyn and Citko (1999), who extend their “Multiple Case Checking” analysis (discussed in Section 5.3 above) from primary predication in copular sentences to depictives. Specifically, the instrumental is checked under Merge by Pred , as in (197a).110 In contrast, “agreeing case” (i.e., nominative or accusative) results from multiple case checking against a functional head into whose specifiers both the agreeing secondary predicate and the argument it is associated with move (overtly or at LF): in the case of a secondary predicate associated with a (nominative) subject, both the subject DP and the nominative secondary predicate check their nominative case in multiple specifiers of T , as in (197b), whereas in the case of a secondary predicate associated with an (accusative) object, both the object DP and the accusative secondary predicate check their accusative case in multiple specifiers of AgrO , as in (197c). Arrows here indicate case checking. (199)
a.
PredP
PRO
b.
Pred'
Pred° ∅
pjanym
TP
c.
pjanyj TP ‘drunk.NOM’ Ivan T'
‘drunk.INSTR’
‘Ivan.NOM’
T°
AgrOP
pjanogo AgrOP ‘drunk.ACC’ Ivana AgrO' …
‘Ivan.ACC’ AgrO°
…
Unlike Hinterhölzl’s (2001) analysis, Bailyn and Citko offer an explanation for the contrasts in (195)–(196) above: for them both dative and genitive objects occur lower than accusative objects (or nominative subjects for that matter), and from this low position they cannot control the PRO in the specifier of PredP, thus disallowing the structure in (197a).111
CASE RELATIONS
133
However, there are still several contrasts that Bailyn and Citko’s (1999) analysis fails to account for. First, as noted in Bailyn and Rubin (1991: 106–107, fn.5), preposing instrumental and agreeing depictives results in sentences with different status.112 According to Bailyn and Rubin (1991), instrumental depictives cannot be preposed, whereas agreeing ones can be. Further investigation of the data shows that instrumental depictives can in fact be preposed, but only if they function as a contrastive topic (as marked with $). There is no such requirements on preposing agreeing depictives (cf. also Nichols 1981: 359–360): (200)
a.
Golye, my tancevali. nude.nom we.nom danced “Nude, we danced.” b. $Golymi, my tancevali. nude.instr we.nom danced “As for being nude, we danced in that state.”
Second, as noted in Neidle (1982: 402–403), agreeing depictives are not subject to the same lexical restrictions that instrumental ones are (cf. also Nichols 1981: 122 and Schein 1995: 52–53): (201)
a.
Ivan cˇ itaet ugrjumyj. Ivan.nom reads gloomy.nom “Ivan reads gloomy.” b. *Ivan cˇ itaet ugrjumym. (cf. (194a)) Ivan.nom reads gloomy.instr intended: “Ivan reads gloomy.”
Third, there are certain meaning differences between the two constructions: in particular, an instrumental secondary predicate implies a connection between the event denoted by the primary predicate and the state denoted between the secondary predicate, whereas an “agreeing secondary predicate” make no such implication (cf. Timberlake 2004: 285). For example, in (202a) the instrumental secondary predicate implies that Ivan’s being healthy resulted from his being in the hospital (i.e., this sentences implies that he went into the hospital sick), whereas in (202b) the nominative secondary predicate has no such meaning; in fact, this sentence strongly suggests that Ivan went into the hospital healthy as well (note different translations of the secondary predicate). (202)
a.
Ivan vernulsja iz bol’nicy zdorovym. Ivan.nom returned from hospital healthy.instr “Ivan returned from the hospital cured.” b. Ivan vernulsja iz bol’nicy zdorovyj. Ivan.nom returned from hospital healthy.nom “Ivan returned from the hospital healthy.”
134
CHAPTER 5
Similarly, Nichols (1981: 156) provides the following minimal pair, and comments that the contrast is due to the fact that “ordinarily the police would have found the man drunk and brought him home, while his friends would have been drinking with him and would have seen him get drunk or gotten him drunk.”
(203)
a.
Milicija privela ego domoj pjanogo. police.nom brought him.acc home drunk.acc “The police brought him home drunk.” b. Druz’ja priveli ego domoj pjanym. friends.nom brought him.acc home drunk.instr “The friends brought him home drunk.”
Bailyn and Citko’s (1999) analysis has no easy way of accounting for these contrasts. Recall from Section 5.3 above that there analysis runs into problems with respect to primary predication as well. Here I propose to take a different line of analysis (suggested by Bailyn and Rubin 1991 in a footnote): I maintain that the so-called “agreeing depictives” are not depictives at all, in the technical sense. Instead, I take them to be appositive phrases. In effect, I extend the analysis of primary predicates in copular sentences developed in this book to depictives. Specifically, instrumental depictives are merged as complements of v , which checks the instrumental case at Merge (indicated by an arrow). The only difference between primary and depictives under my analysis is that in the former case v is the merger position for the copula byt’ “be”, whereas in latter case it is phonetically null. In fact, my analysis of true (i.e., instrumental) depictives is very similar to that of Bailyn and Citko (1999), except that I take the head of the secondary predication phrase to be v , a lexical head, rather than Pred , a functional head (as discussed above for copular sentences). (204)
vP PRO
v' v° pjanym ∅ ‘drunk.INSTR’
I differ more significantly in my analysis of the so-called “agreeing depictives”. As mentioned above, I take them to be appositive phrases, namely phrases merged with the argument that they are associated with in a symmetrical structure resembling the structure I argued for in this book for copular sentences with the nom-nom pattern:113
CASE RELATIONS
(205)
a.
DP
b.
DP
DP DP Maše pjanoj ‘Masha.DAT’ ‘drunk.DAT’
DP DP DP Mašu pjanuju ‘Masha.ACC’ ‘drunk.ACC’
DP DP Maša pjanaja ‘Masha.NOM’ ‘drunk.NOM’ c.
135
d.
DP DP DP pjanoj Maši ‘Masha.GEN’ ‘drunk.GEN’
These appositive phrases can be displaced, similarly to displaced adjectives in the split phrases (for a detailed discussion of which, see Pereltsvaig 2006b, 2006c): (206)
a.
Trudnuju on rešil vˇcera zadaˇcu. complicated.acc he.nom solved yesterday problem.acc “It was a complicated problem that he solved.” b. On rešil zadaˇcu vˇcera trudnuju. he.nom solved problem.acc yesterday complicated.acc “It was a complicated problem that he solved.”
Consider how the contrasts described above can be accounted for under this analysis. First, Hinterhölzl’s (2001) generalization still holds: only bounded predicates can appear as appositives. Second, appositive phrases are structurally freer than true depictives because the former do not involve a PRO subject that must be controlled as do the latter; this explains the contrast in (195)–(196). Third, displacement of appositives is freer than displacement of true depictives: the former derives from argument movement (as discussed in detail in Pereltsvaig 2006c), which is freer than predicate movement. Forth, appositives are not subject to lexical restrictions, whereas true depictives are. Finally, true depictives are contained in a vP, the head of which is associated with a theta-grid containing an event position; therefore, a connection can be established between the event denoted by the primary predicate and the state denoted between the true secondary predicate. In contrast, appositives introduce no event variable; hence, the difference in meaning illustrated in (200)– (201) above. This analysis is further supported by the fact that sentences with postposed agreeing phrases are most felicitous in contexts where such phrases are focused, which is in accordance with the general observation that focused constituents tend to appear clause-finally (cf. Pereltsvaig 2006c). (207)
On skazal èti slova pjanyj, potom gor’ko ob ètom požalel. he.nom said [these words].acc drunk.nom then bitterly about that regretted “He said these words drunk and then bitterly regretted it.” [Nichols 1981: 133]
To summarize, I propose to extend the analysis developed in this book for copular sentences to depictives and similar constructions: while instrumental depictives are merged as complements of v , the so-called “agreeing depictives” are not predicates
136
CHAPTER 5
at all, but rather appositive DPs merged in a symmetrical structure and possibly displaced in a way analoguous to split phrases. 5.7. S U M M A R Y
In this chapter, I drew a distinction between morphological case (i.e., overt case marking) and abstract syntactic case (i.e., uninterpretable formal features). Furthermore, I distinguished between two types of syntactic case: structural case (i.e., case checked in a specifier position) and inherent case (i.e., case checked in a complement position, at Merge). Furthermore, I proposed to analyze the Russian case system as follows: nominative is the morphological default case appearing in the absence of syntactic case (in addition to structures where syntactic nominative is checked); instrumental is the default inherent case appearing on a complement of a verb (whether V or v ) unless the verb is specified for checking a different inherent case. Moreover, I adopted the view that only syntactic arguments, namely phrases merged as complements and specifiers of lexical heads, need to check abstract syntactic case. From this, I deduced that the two DPs merged together into a symmetrical structure (as in Russian copular sentences with the nom-nom pattern) need not check abstract syntactic case since they are not syntactic arguments. In the absence of syntactic case, they appear in the morphological default case, namely nominative. In addition, I drew a parallel between instrumental case on post-copular phrases in Russian and other instances of instrumental case on complements of verbs. Thus, a more unified view of instrumental case has been achieved. Finally, I showed how the analysis of case marking in Russian copular sentences developed here extends to sentences with copula-like verbs and to sentences with depictives (and their look-alike constructions).
CHAPTER 6
CONCLUSION
In this book, I investigated Russian copular sentences and distinguished between two kinds of such sentences, depending on the case marking on the post-copular phrase. Furthermore, I proposed that the differences between the two types of copular sentences are not purely morphological, but rather syntactic. In particular, I proposed that post-copular phrases in sentences with the nom-nom pattern are DPs, whereas post-copular phrases in sentences with the nom-instr pattern are bare NPs. Furthermore, I showed that post-copular bare NPs behave like APs and not like DPs. Moreover, I proposed that the pre-copular and the post-copular DPs in sentences with the nom-nom pattern are merged together into a symmetrical structure, which I argued to be possible under the most minimalist conception of Merge (contra Kayne 1994). In contrast, I analyzed post-copular NPs in sentences with the nom-instr pattern as complements of v where the copula byt’ “be” is merged. Thus, the copula can be merged either as a functional head F (in sentences with the nom-nom pattern) or as a lexical head v (in sentences with the nom-instr pattern). The former can be derived from the latter by a process of semantic bleaching which eliminates the theta-grid of a lexical copula. Thus, no special “identity copula” with special semantics and/or syntax needs to be postulated. This analysis is shown to account for syntactic, semantic, and morphological properties of the two types of copular sentences. Although the discussion in this book is couched in the Minimalist framework, I proposed some revisions to the standard Minimalist set of analytic tools. In particular, I argued against the strong version of the Antisymmetry Hypothesis (cf. Kayne 1994), which takes symmetrical structures to be completely impossible at any stage in the derivation. Instead, I showed that symmetrical structures can be merged under certain well-defined conditions, which leads to a more parsimonious analysis of Russian copular sentences. Furthermore, I argued against the strong version of UTAH (i.e., the hypothesis that there is a one-to-one mapping between theta-roles and structural positions). Specifically, I showed that the highest arguments of stative intransitive verbs, on the one hand, and of nominal and adjectival predicates, on the other hand, are not merged in the same structural position with respect to their predicates, despite their being the same semantically. From this, I concluded that a more flexible theory of thematic relations is required and developed a theory along these lines. Moreover, I argued against the widely accepted “agreement in case” analysis of Russian copular sentences with the nom-nom pattern. Instead, I proposed that both nominative DPs are non-argumental, and as such need not be marked with syntactic 137
138
CHAPTER 6
case at all. As a result, they appear with nominative case, the morphological default. One of the biggest advantages of this analysis is that it allows us to avoid postulating a separate lexical item for the “identity copula” in sentences with the nom-nom pattern. Rather, the copula in both types of copular sentences can be subsumed under the same lexical item, and the differences between them are shown to result from the distinction between a lexical and a functional status. One of the conclusions emerging from this analysis of the copula is that verbs can be divided into three categories: (i) full lexical verbs whose theta-grids contain both an a-variable and a thematic index for each thematic position; (ii) lexical light verbs derived by index erasure (their theta-grids lack a thematic index); and (iii) functional “verbs” that have no theta-grids at all. Most copula-like verbs in Russian are shown to belong to the second category, whereas byt’ “be” in all varieties of Russian and stat’ “become” in colloquial Russian are true functional copulas. Given that the functional copula (in sentences with the nom-nom pattern) does not have a theta-grid, the interpretation of such sentences cannot be derived through normal processes of theta-assignment and predication. Rather, I proposed that the appropriate interpretation for such sentences comes from a syntactic coindexing between the two DPs flanking the copula. This coindexing in turn is shown to derive from the symmetrical structure which requires that both DPs have the same feature composition. Furthermore, it is argued that a referential index is indeed a feature relevant for syntactic computation. Even though this book focuses on copular sentences in Russian, I showed that the proposals made here are relevant for syntactic analysis of other languages, such as Italian and Norwegian.
ACKNOWLEDGEMENTS
This book is a revised version of my doctoral dissertation. Many fine linguists have provided insightful comments, suggestions and criticisms during my work on this book. Among them are: David Adger, Len Babby, John Bailyn, Mark Baker, Joanna Blaszczak, Jonathan Bobaljik, John Bowers, Marcel den Dikken, Steven Franks, Ljudmila Geist, Stephanie Harves, Marit Julien, Richard Kayne, James Lavine, Andrea Moro, Barbara Partee, Vieri Samek-Lodovici, Lisa Travis, and Roberto Zamparelli. The material in this book was presented at the LAGB biannual meetings at the University College London, University of Durham and University of Leeds, at FASL 12 at the University of Ottawa and FASL 13 at the University of South Carolina, at CASTL Conference in Tromsø, and at the FDSL 6 in Potsdam, as well as in guest talks at Concordia University, University of Sheffield, University of York, CUNY, Princeton and Cornell; I thank the audience at these events for sharing their questions, comments and ideas. I am also deeply grateful to my Russian consultants who contributed their judgments and discussed the data with me: Andrey Antonenko, Olga Babko-Malaya, Eva Bar-Shalom, Julia Belopolsky, Olga Borik, Vladimir Borschev, Georgiy Bronnikov, Natalia Cherepashenets, Elena Chernishenko, Denis Chesnokov, Lora Dubinskaya, Marina Dyakonova, Anna Feldman, Natasha Fitzgibbons, Anna Geisherik, Dmitry Gitlin, Lydia Grebenyova, Malvina and Yakov Guralnik, Tanya Ionin, Misha Ishenko, Olga Komarova, Ioulia Kovelman, Maxim Kryachko, Julia Kuznetsova, Denis Lyakin, Ella and Sophia Malamud, Vita Markman, Ora Matushansky, Galya Minevich, Olga Oganesyan, Sonia Osminkin, Iouri Pereltsvaig, Yana Pugach, Zhenia Romanova, Nina Rojina, Marina Samsonova, Elena Schmitt, Vladimir Segal, Irina Sekerina, Alla Smyslova, Natalia Strakhov, Sergey Tatevosov, Yakov Testelets, and Yehor Tsedrik. Last but not least, I am indebted to my father Michael Pereltsvayg and my grandmother Rakhil Savinskaya , and to John C. Simeone 3a . This book is dedicated to the memory of my late grandfather Yakov Savinsky and my late mother Freyda Pereltsvayg.
139
NOTES
1. Some non-verbal predicatives, such as nužno “need” take a nominative post-copular phrase, but the overall pattern is dat-nom: (i) Nam nužna kniga. us.dat need.3.sg.f book.nom ‘We need a book.’ Note that in colloquial Russian this nominative is often replaced by accusative, as in (ii). In this case, the verb does not agree with the nominative phrase but appears with the default agreement (i.e., 3rd person singular neuter). (ii) Nam nužno knigu. us.dat need.3.sg.n book.acc ‘We need a book.’ 2. For a detailed discussion of the notion “copula” vs. “copula-like”, see Lekant (1995). 3. In English and many other languages, the copula be is also used in clefts (e.g., It is garbage that I am talking about) and pseudo-clefts (e.g., What I am talking about is garbage). For this reason, clefts and pseudo-clefts have often been investigated together with copular sentences, and similar analyses have been proposed for both types of constructions (e.g., Heggie 1988, Heycock and Kroch 1999, inter alia). However, Russian does not have a true cleft construction (cf. Junghanns 1997: 187): Russian èto-“clefts” do not allow a copula to intervene between the pronoun èto and the focused constituent (Junghanns 1997: 188, fn.4). (ii) a. Èto (*est’ /* bylo) Boris vypil vodku. èto is / was Boris.nom drank vodka.acc intended: ‘It was Boris who drank the vodka.’ b. Èto (*est’ / *bylo) vodku Boris vypil. èto is / was vodka.acc Boris.nom drank intended: ‘It was the vodka that Boris drank.’ In this book, I will not consider cleft or pseudo-cleft constructions. For further discussion of Russian counterparts of English clefts, see Grenoble (1998: 200–204), Junghanns (1997), and the references cited therein. 4. The term “semantic bleaching” comes from work on complex predicate formation and usually refers to the process of light verb formation. As a result of this process, light verbs have “either a completely empty, or merely an incomplete, argument structure” (Butt 1995: 144). For further discussion of semantic bleaching and the nature of light verbs, see Grimshaw and Mester (1988), Rosen (1989), Butt (1995) and Samek-Lodovici (2003); also Section 4.4.2 below. A similar concept of “desemanticization” is discussed by Babby (1980: 128–146) in connection with lexical verbs that can appear in what he calls “Negated Existential Sentences”. 5. A post-copular phrase can also be a noun phrase headed by a noun in the genitive case. In this construction the genitive noun head has the meaning of a qualitative evaluation, as in (i), or possession, as in (ii). In the former case, adjectival modification is obligatory (cf. Rozental’ 1976: 39). I will not discuss these construction in this book. (i) Voda byla *(olovjannogo) cveta. water.nom was tin.gen color.gen “The water was of tin color.”
141
142
6.
7. 8.
9.
10.
11.
12.
13. 14. 15.
16. 17.
18.
19.
NOTES (ii) Dom ètot – Meškova. house.nom this.nom Meshkov.gen “This house is Meshkov’s.” [Rozental’ 1976: 39] For a further discussion of Russian adjectives, see Kozyreva and Khmelevskaja (1972), Babby (1975a), Siegel (1976), Crockett (1976), Comrie (1987), Bailyn (1994), Spencer (2000), Pereltsvaig (2001a) and the references cited therein. For a more extensive discussion of pre-generative-grammar theories of the copula, see Moro (1997: 248–258). A similar distinction is drawn by Goossens (1992), who distinguishes between three different kinds of be: (i) a predicational verb be, which denotes existence, as in I think; therefore, I am, corresponding to Halliday’s class 1, (ii) a semi-predicational verb be, as in be a nurse, be nice, corresponding to Halliday’s class 0, and (iii) a minimally predicational verb be, which is also called “ascriptive be” and includes the identifying be, corresponding to Halliday’s class 2. There are different views of spell-out within the Minimalist literature: for instance, Uriagereka (1999) proposes that spell-out is a rule which, like any other rule, can apply multiple times in the course of a derivation. Alternatively, it has been proposed that spell-out is the single output of syntactic derivation (cf. Single Output Syntax, Bobaljik 1995: 349–350). According to this view, “the syntax produces a single output representation from a given input, and this output representation is then interpreted by semantic and morpho-phonological components”; therefore, “Spell-Out is after LF movement operations”. The exact nature of Spell-Out is not crucial for the discussion in this book. Throughout this book, I will continue to use labels like X , X and XP as convenient labels, following common practice. However, it is to be understood that the distinction between X’ and XP has no theoretical importance. As noted in Bailyn and Rubin (1991: 123), these judgments come from speakers who accept extraction from cˇ to-complements (most Russian speakers do not). That this is indeed a matter of extraction is evidenced by the fact that the wh-phrase is perfectly grammatical regardless of case marking if it stays in situ (as with echo questions). Not everybody agrees that the “subject-predicate” sequences form a constituent (i.e., a small clause) in the first place (e.g., Bresnan 1978, 1982, E. Williams 1980, 1983, 1994, Schein 1995, and many others do not accept the small clause analysis). Here, I will not discuss these latter approaches. For other proposals that fall under this approach, see Kayne (1985), Hornstein and Lightfoot (1987), Moro (1988), and Bailyn and Rubin (1991). Depending on one’s definition of a clause, this may not be a clause at all; here and below, I use the term “small clause” in a loose sense. Another combination approach to small clauses based on the category of the post-copular phrase is that of Contreras (1995), for whom the difference depends on the specification for the [±V] feature. He adopts the small clause analysis for adverbial and verbal predicates (i.e., [+V] categories), but not for nominal and prepositional predicates (i.e., [–V] categories); for the latter type of predicates he proposes a version of the complex predicate analysis modeled on Larson’s (1988) structure for double object constructions. Throughout this book, I assume for ease of exposition that all tense, modality, mood and (outer) aspect morphemes are generated in T. Calling the head of the small clause a v rather than Pred is in line with Eide and Åfarli’s (1999: 176) claim that “predication operator [i.e., Pred ] is internal to the verb, or in other words that the verb is a lexicalization of the operator”. Another distinctive property, proposed by Fukui and Speas (1986), is that functional categories can have a maximum of one specifier position, whereas lexical categories can have any number of specifiers; however, this proposal has been challenged by various researchers (cf. Speas 1990: 47–48, Koizumi 1995, among others). Possessor phrases are generally assumed to be generated in the specifier of NP; they may or may not move overtly to the specifier of DP (e.g., they move in English, but not in Hebrew, cf. Ritter 1988, 1991); for a discussion of possessor phrases in Russian see Trugman (2004).
NOTES
143
20. Note that not everybody agrees with the DP Hypothesis being universal. For example, Börjars (1998) argues that Scandinavian noun phrases are uniformly NPs (she remains agnostic as to whether the DP Hypothesis is incorrect or just not applicable to Scandinavian). I will come back to this question in section 2.5 below. 21. The remaining two South Slavic languages − Bulgarian and Macedonian − do have overt articles. 22. Trenkic (2004) claims that article-less languages, such as Serbo-Croatian, lack the DP projection on the basis of Second Language Acquisition data; in particular, she shows that adult Serbo-Croatian learners of English have many problems with the English article system. Specifically, her study shows that these speakers often omit articles and are more prone to this kind of mistake if the target nominal involves adjectival pre-modification. Trenkic’s conclusion is that articles and adjectives in the learners’ interlanguage compete for the same (presumably, modifier) position. Yet, crucial data are missing from her study: she considers only omission mistakes but not substitution mistakes. If it turns out that her subjects make not only more omission mistakes but also more substitution mistakes in phrases with adjectival pre-nominal modification, an alternative explanation based on processing load would suggest itself. Additionally, one would need to look at cases where putative article omission can be explained as substitution of a null article (e.g., dogs may be a substitution mistake to be analyzed as Ø dogs rather than just an omission mistake). 23. For the sake of exposition, I use a single referential index rather than an ordered pair, as does Baker. 24. A possible counterexample was brought to my attention by Jonathan Bobaljik (p.c.): (i) I found the Jakobsoniani perspective on case intriguing. Hei was certainly a brilliant linguist. 25. The ungrammaticality of (31b) is not caused by the nominative case of the relative pronoun (instead of instrumental agreeing with the case of the head noun), as can be seen from the grammaticality of (31c). In Russian, a relative pronoun satisfies the requirements of the embedded (i.e., relative) clause rather than the matrix clause; therefore, the noun-head of the relative and the relative pronoun may appear with different case markings. 26. For reasons that I will not go into here, the copula moves into C in A'-inversion. Note that the present claim that different types of copular sentences involve either A- or A'-inversion harmonizes well with Bailyn’s claim that there are two types of scrambling in Russian: A- and A'-scrambling (cf. Bailyn 2002, 2003a, b). 27. But see Bailyn (2004b) for a discussion of locative inversion in Russian where a PP moves to Spec-TP to satisfy the EPP, according to his analysis. What PPs and DPs have in common is that they are fully saturated, whereas bare NPs and APs are not. 28. Here I am focusing on data involving definite DPs because the data involving indefinite DPs are rather controversial. 29. The interpretation of copular sentences with NPs, APs and DPs in post-copular position is discussed in more detail in Chapter 4 below. 30. The discussion below has been heavily inspired by Moro (1997, 2000). Discourse-driven word order in Italian more generally is discussed in Antinucci and Cinque (1977), Benincà (1988), Benincà and Salvi (1988), Samek-Lodovici (1996), Rizzi (1997), Frascarelli (2000), Pereltsvaig (2004), and the references cited therein. 31. Note that the judgments given here are for plain intonation only. The sentence (43c) is grammatical for some of the speakers I have consulted with a heavy stress on either alto or Gianni. It is also marginally acceptable for some of my consultants on a pseudocleft-like reading that can be paraphrased as “tall is like what Gianni is” (i.e., Gianni is the definition of what it is to be ‘tall’, or how much height one must have to count as ‘tall’). This reading, however, is irrelevant for the argument made here. I thank Roberto Zamparelli for first bringing this reading to my attention. 32. There are also bare NP arguments in Norwegian; see Borthen (2001) and the reference cited therein. 33. There is some variation among speakers with respect to the configuration in (48a) in the text: judgments vary from “worse than (48b-c)” to “completely ungrammatical”. 34. Interestingly, the same generalization about embedding of inverted copular sentences is observed in (Austrian) German as well. Thus, embedding in no-CP-recursion contexts is possible only if the
144
35. 36. 37. 38. 39. 40.
41.
42.
43. 44. 45.
46.
47.
NOTES inverted phrase is a DP, but it is ungrammatical if the inverted phrase is a bare NP or an AP. The data are from Czinglar and Köhler (2001); judgments are theirs. (i) a. dass [DP die Ursache für den Skandal] dieses Bild ist that the cause for the scandal this picture is ‘… that the cause for the scandal is this picture.’ b. ?*dass [NP Lehrer] ein Kollege ist that teacher a colleague is ‘… that a colleague is a teacher.’ c. ?*dass [AP krank] keine Kinder sind that ill no children are ‘… that no children are ill.’ Mikkelsen (2002) also discusses data involving NPI-binding in Danish. I was not able to replicate her results in Norwegian. For some speakers the contrasts are much clearer with adverbs like alltid “always” than with ikke “not”. I have no explanation for this inter-speaker variation. Note that (53c) but not (53a-b) can be merged by asymmetric projection. Note that Cecchetto (1999) and Frascarelli (2000) have argued for LCA-compatible analyses of these facts. Yet, they cannot account for all the facts (see discussion in Pereltsvaig 2004). Note that Succession, Uniformity and Maximality are not relevant under Bare Phrase Structure. The hypothesis that traces are not visible for linearization and therefore are not subject to the LCA is the starting point for Nunes (1999), but he develops an analysis (within the copy theory of movement) where traces are subject to the LCA. There is a disagreement in the literature as to whether the Germanic-type object shift is an A- or A -movement. Here, I ignore this controversy and mention Object shift on a par with other cases of A-movement, because this type of movement presents a problem for the Dynamic Antisymmetry Hypothesis, just like other cases of A-movement. Sentences of this type also serve as the basis for a very productive type of split phrases in colloquial Russian (for a detailed discussion of split phrases in colloquial Russian see Pereltsvaig 2006c): (i) Do vojny èto redˇcajšee bylo životnoe. before war this.nom rarest.nom was animal.nom “Before the war, this was the rarest animal.” [Lapteva 1976: 224] This section is inspired by discussions with Andrea Moro and Marcel den Dikken. See also den Dikken (2006: 266, n. 9). Animacy is not a purely semantic feature in Russian as it plays a role in defining morphological case paradims of certain nouns. See Franks (1995), Timberlake (2004: 165–171), inter alia. Note that a similar phenomenon is also attested in English (and possibly other languages as well). For example, at the 2001 Emmy Awards, the following was attested from one of the journalists: Julia Roberts is a great actor (not a great actress) Interestingly, the feminine form sekretarša “secretary” corresponding to the masculine sekretar’ “secretary” is “used only for a secretary in an office, a shorthand-typist; the secretary of, for instance, a Party committee would be sekretar’, irrespective of sex” (Comrie and Stone 1978: 166). This is part of a general pattern where feminine forms are reserved for low prestige occupations and masculine forms are used for high prestige occupations irrespective of the actual gender of the individual. The noun tovarišˇc “comrade” “resolutely refuse[s] to allow derived feminine form” (Comrie and Stone 1978: 163). For more on nouns of this type, see Timberlake (2004: 163–165). This sentence has a somewhat colloquial ring to it, and some prescriptive grammars of Russian condemn it, recommending instead the following:
NOTES
48.
49. 50.
51. 52.
53.
54. 55.
56. 57.
58.
59.
145
(i) Ètot mal’ˇcik – takoj bol’šoj nerjaxa. this.m boy.m such.m big.m sloven.f ‘This boy is such a big sloven.’ The nouns of the common gender are particularly susceptible to variation in this respect. For example, in Darya Dontsova’s “Dama s kogotkami” just a few pages apart we find the following examples (referring to the same male person), one with masculine agreement on the predicate and another with feminine agreement: (i) a. V subbotu pozvonil “zvezda komp’juterologii” i soobšˇcil… in Saturday called.m star(f) computer-science and informed.m ‘On Saturday the “star of computer science” called (me) and informed (me)…’ [Darya Dontsova, “Dama s kogotkami”, p. 269] b. Zvezda pominutno popravljala padajušˇcie oˇcki i slegka zaikalas’. star(f) every-minute rearranged.f falling glasses and slightly stuttered.f ‘The star again and again rearranged (his) falling glasses and stuttered slightly.’ [Darya Dontsova, “Dama s kogotkami”, p. 258] In this book, I use the terms -role and thematic position (in a theta-grid) interchangeably. Note that an alternative would be to explain the ungrammaticality of such examples by violations of s-selection (for “semantic selection”). According to such an approach, devour selects for a particular semantic type of a complement – an individual (rather than, say, a proposition) – and is ungrammatical if this required complement is not present. Yet, it has been suggested that s-selection is none other than -assignment (cf. Svenonius 1994: 35). In this book, I also remain largely agnostic as to whether all verbs and adjectives (or only eventive ones, or only verbs, or only stage-level predicates) have an event argument in their thematic grids. The assumption that there exists only one mode of thematic discharge (i.e., of semantic composition) is challenged in Chung and Ladusaw (2004), who adopt the notion of predicate restriction, a mode in which “the property argument is interpreted as a restrictive modifer of the predicate” (p. 5). This notion of predicate restriction is similar to Higginbotham’s -identification; Chung and Ladusaw have no counterpart to Higginbotham’s -binding. An additional mode of thematic discharge proposed by Higginbotham (1989: 475) is autonomous theta-marking, “where the value assigned to the open position in the theta-marker is the attribute given by its sister constituent”; this mode is designed to account for cases of modification in which the property denoted by the adjective is not absolute, but are relative to the choice of attribute (e.g., an expensive meal is interpreted as a meal whose price is high for a meal, but not for instance for an automobile). This mode of thematic discharge is largely irrelevant for the discussion in this book. For a more detailed discussion, the reader is referred to Higginbotham (1985, 1989) and Speas (1990: 69-70). Note that Higginbotham’s assumptions about the syntactic structure of adjectival modification are different from those adopted in this book. See chapter 2 above for discussion. “Expletive determiners”, discussed by Longobardi (1994) and Zamparelli (2000: 158), are vacuous in the sense that they do not -bind the open thematic position of the NP, which then percolates up to the level of the DP. Such determiners are largely irrelevant for the discussion in this book. For the sake of exposition, I will focus on nominal arguments, largely ignoring clausal and prepositional arguments. Another potential (although to my mind more dubious) example of double -marking is the construction of the sort John rolled down the hill discussed by Jackendoff (1972: 32–34), who claimed that in this sentence the subject John receive two -roles: Theme and Agent (for further examples of potentially doubly -marked phrases, see Jackendoff 1972: 35–36). An additional example of double theta-marking arises if control is treated as raising. Note that if the theory admits a possibility of a noun phrase being -marked twice, it is impossible to claim that -roles are in one-to-one correspondence to either case forms or syntactic positions (as according to UTAH). Another construction that raises the same problem is the so-called sounds-like-construction in colloquial English, discussed in Heycock (1994) and Pereltsvaig (2001c). I will not discuss it here.
146
NOTES
60. There are several other intriguing issues regarding semantic and pragmatic properties of these three constructions; however, these issues go beyond the scope of this book. 61. This idea is not entirely novel: Fiengo and May (1994: xii) propose to “understand indices to be a structural part of the ‘feature component’ of a category in a phrase marker”. 62. Interestingly, an index sign can express more than just a referential index. In these cases, its articulation is modified in a variety of ways. For example, “the path that the index finger traces as it moves to its final location can vary, conveying distance and ‘route’ information about the location” (MacLaughlin 1997: 118). This is very much in accordance with the generalization that sign language is 3-dimensional (rather than linear, as spoken language is) and can express several bits of information simultaneously (cf. MacLaughlin 1997 and Talmy 2000). 63. Here, I assume the copy theory of movement; therefore, the amended definition of binding refers to copies of and . 64. The strengthening anaphor sebja has no nominative form. Furthermore, it is impossible to examine the distribution of the reciprocal drug druga “each other” in this connection because it has no nominative form either. It is possible that the reciprocal has no nominative form exactly because the relevant structure where it would be used is ruled out syntactically. I will not pursue this issue any further in this book. 65. Note that Baker and Stewart do not consider nouns in this connection; I return to the unergativity of nouns below. Throughout this book, I follow Baker and Stewart in assuming that unaccusativity is a syntactic distinction (contra Dowty 1991). 66. As expected on the basis of a cross-linguistic generalization, adjectival predicates in Èdó require the presence of the copula yé, whereas verbal predicates (including stative verbs synonymous with adjectives) are incompatible with it (cf. Baker and Stewart 1997: 33). 67. Additional unaccusativity diagnostics in Italian, not considered by Baker and Stewart (1997) and Baker (2003), include possessor datives and wh-extraction from subject; they too can be used to show that adjectives pattern with unergatives rather than unaccusatives; since the focus of this book is on Russian, I will not discuss the Italian tests here detail (cf. Pereltsvaig 2001c for a more detailed discussion). 68. Possessor datives in Hebrew are distinct from ethical datives: the former but not the latter can be a full dative noun phrase, as well as a dative clitic. Note also that a reflexive dative cannot be a full dative noun phrase either. For more discussion of various dative constructions in Hebrew, see Berman (1981) and Landau (1999). 69. It must be noted here that the validity of ne-cliticization as an unaccusativity diagnostic has been challenged (cf. Lonzi 1985, Saccon 1992, and Levin and Rappaport Hovav 1995: 275–277). For example, it has been claimed that ne-cliticization does not match up with auxiliary selection, another alleged unaccusativity diagnostics; yet, it is not completely clear if auxiliary selection is a good unaccusativity diagnostic in first place (for a detailed discussion of auxiliary selection as an unaccusativity diagnostic, see Sorace 2000). Thus, it appears that even though auxiliary selection and ne-cliticization select slightly different sets of verbs, it is the ne-cliticization which indicates whether the verb has an external (i.e., Agent) argument or not. The data reported in this book are from speakers for whom ne-cliticization is clearly associated with unaccusativity (i.e., the absence of an external argument) rather than discourse conditions (as proposed by Lonzi); these speakers do not accept Lonzi’s crucial examples where verbs traditionally thought of as unergative allow ne-cliticization. 70. This has been first noted by Burzio (1986: 74, fn.13); however, he maintains that there are no unaccusative adjectives at all. Yet, as discussed in detail by Guasti (1988: 325) and Cinque (1990), certain adjectives pattern with unaccusatives; the list includes but is not limited to certo “certain”, chiaro “clear”, oscuro “obscure”, noto “(well-)known”, probabile “probable” and evidente “evident”. Still, these adjectives are the exception rather than the rule. 71. Crucially, there is a strong negative correlation between the acceptability of ne-cliticization by a given speaker for a given verb and the acceptability of the progressive periphrasis with the same verb (progressive periphrasis is a standard diagnostic for stativity in Italian): speakers who accept the progressive with a given verb (i.e., those who treat this particular verb as eventive rather than
NOTES
72.
73.
74.
75.
76. 77. 78.
79. 80.
81
82
83
147
stative) do not allow ne-cliticization, and vice versa. Thus, those speakers who treat these verbs consistently as statives allow ne-cliticization with them. I leave the question of possessor phrases outside the scope of investigation here. Clearly, even if possessor phrases are -marked in the specifier of NP, this position is still unavailable for subjects of copular sentences. In fact, the compatibility of subjects in copular sentences with possessor phrases (as in Monica is Bill’s best friend) has been taken as an argument against the analysis that generates the subject of the copular sentence as the specifier of NP, as proposed by Stowell (1983). For a detailed discussion, see Heycock (1994: 98–99). However, Chvany (1975) did not use this terminology. A similar though not identical point is made in Babby (1980); for a list of verb classes that take genitive subjects under negation in Russian, see Babby (1980: 131–133). A similar contrast is discussed in Pesetsky (1982) in connection with plavat’ which can mean either “swim” (on an agentive reading) or “float” (on an existential reading). However, plavat’ “swim, float” is not derivationally related to an adjective. As shown in Borovikoff (2001) there are two distinct constructions involving the distributive po: one involving overt numerals other than odin “one” and the other with explicit or implicit odin “one”. Of the two, only the latter construction serves as a reliable unaccusativity diagnostics, while the former construction does not distinguish unaccusative and unergative verbs (examples from Borovikoff 2001); cf. also Kuznetsova (2005): (i) a. Po desjat’ studentov v každoj gruppe proˇcitali roman “Lolita”. distr ten.acc students.gen in each group read.pl novel Lolita “In each group ten students read the novel ‘Lolita’.” b. *Po studentu v každoj gruppe {proˇcitalo / proˇcitali} roman “Lolita”. distr student.dat.sg in each group read.sg.n / read.pl novel Lolita “In each group one student read the novel ‘Lolita’.” See also Timberlake (2004: 354–355). Note that Baker and Stewart (1997) and Baker (2003) refer to what I call Embodiment argument of adjectives as Theme. Compare this to a distinction between Predicate Argument Structure (or theta-grid) and Lexical Conceptual Structure in the work of Levin and Rappaport (1986), Hale and Keyser (1986, 1987), Jackendoff (1987), and Speas (1990). Here and below, I ignore the event position in verbs’ theta-grids for clarity of exposition. A similar idea has been entertained by Eide and Åfarli (1999: 176), who maintain that “the copula is a verb with a very sparse attributive content which functions as little more than a pure lexicalization of the predication operator. Ordinary main verbs, on the other hand, have full attributive content, including Theta properties”. Potentially, other “light” verbs belong to this intermediate category as well. Some candidates that come to mind are “functional verbs” of Cinque (2004), “semi-lexical motion verbs” in Romance and Germanic discussed by Cardinaletti and Giusti (2000), as well as få “get” in Norwegian passives discussed by Taraldsen (1995). A detailed examination of these constructions in light of their “light verb” status goes beyond the scope of this book. Another language that may fall under the same rubric is Polish, where nominal and adjectival predicates pattern differently with respect to case marking and other morphosyntactic factors. The details of the data are rather complex and for this reason I will not consider Polish here in any great detail. The reader is referred to Rothstein (1986), Citko (2006a, b) and the references cited therein. It must be noted here that the distribution of the two types of copular sentences in Russian has been changing over the last two hundred years, but the process is far from being simple: in the standard language instrumental is “pushing the nominative out” (as part of the more than 200-year old history of East and West Slavic where predicative instrumental “encroaches” on the nominative; cf. Nichols 1973 for a detailed discussion), whereas in colloquial language nominative often replaces the instrumental (as it does with other cases in a wide range of constructions). The discussion in this section focuses on modern standard Russian.
148
NOTES
84 So far, I have no explanation for the fact that idiomatic post-copular noun phrases must appear in the nominative, even after copular-like verbs that otherwise always take instrumental complements (see examples in the main text). One possible explanation is that idiomatic expressions need to appear in the citation form (see discussion in Section 5.5 below). I will leave this issue open for further research. 85 Alternatively, the temporally restricted universe of discourse must be bound. This property of such sentences is often exploited as a narrative tool. For example, in (i) below – from Darya Dontsova’s “Domik tëtuški lži” – although one would assume that a person is young at some point in their life, Miss Marple as we know her from Agatha Christie’s novels was always an older woman: (i) [Dasha Vasilyeva, being offended by a comparison with Miss Marple:] Mis Marpl byla staruxa! miss Marple was old-woman.nom “Miss Marple was an old woman!” 86 Note that although post-copular DPs may be associated with a presupposition of uniqueness, they not always are. 87. Curiously, some grammars of classical languages (such as Woodcock’s 1959 otherwise thorough grammar of Latin) omit nominative from their discussions of the case system of the language altogether. 88. As a curious note, it should be added here that the most attrited speakers of Heritage (American) Russian and Heritage (American) Polish use genitive as the citation form (cf. Polinsky 1994: 10). No explanation has been found for this fact. For a further discussion of case marking in Heritage Russian, see in the main text below. 89. This generalization is true not only of Russian but of other languages with case morphology as well. For example, Left Dislocated phrases in Hebrew appear in the nominative case as well, regardless of the case of its associate pronoun within the clause. (i) (*Et) ha-šotrim, ani lo ohevet otam. acc the-policemen I not like them “As for the policemen, I don’t like them.” 90. For a more detailed discussion of vocatives in Russian, see Yadroff (1996); for a discussion of how vocatives fit into syntactic structure of sentences (and whether they do at all), see McCawley (1989: 117–119). 91. Heritage (American) Russian is to be distinguished from Emigré Russian: the former is the language acquired as the first language by the second generation of emigrants (who were born outside of Russia or emigrated as small children) and used in a limited fashion, mostly in home, whereas the latter is the language of first generation Russian immigrants who came to America as adults and continue using Russian as their primary language. For a more detailed discussion of the differences between American Russian and Emigré Russian, the reader is referred to Polinsky’s studies. 92. Note also that the Heritage (American) Russian data in Polinsky’s study further support the case feature system argued for by Neeleman and Weerman (1999). According to their system, nominative is unmarked, accusative bears the feature [+case] and dative is [+case, +dependent]. Thus, nominative is unmarked with respect to accusative and accusative is unmarked with respect to dative. Polinsky (1996, 1997) describes Heritage (American) Russian data as an Argument Case Shift whereby dative forms are replaced by accusative and accusative forms by nominative. In other words, in Heritage (American) Russian each case form is replaced by the next less marked one. See example (156b) in the main text. 93. In this example, the case marking on dom{ moj lit. “house my” is underdetermined between nominative and accusative due to syncretism in the nominal paradigm. 94. It is irrelevant for the argument made here whether byt’ “be” is taken by proponents of the “agreement in case” approach to be a functional or lexical head, given that they do not assume the definition of lexical heads as the only argument-taking categories, as is in fact assumed in this book. 95. This phenomenon is not restricted to Russian. For relevant Italian examples, see Salvi (1988b: 232–233) and for similar Norwegian examples and further discussion, see Pereltsvaig (2001c).
NOTES
149
96. Note that the verb blagodarit’ “thank” has undergone a shift in case selection in the last 100 years or so: as noted in Comrie and Stone (1978: 103, it “is often found with the dative case in nineteenth-century literature; in the current usage only the accusative is standard”; however, the deverbal preposition blagodarja “thanks to” still takes dative. Conversely, the verb l’stit’ “flatter” switched from accusative objects, which can still be observed in the semi-idiomatic (ia) to dative objects, as in (ib); cf. Comrie and Stone (1978: 105). (i) a. l’stit’ sebja nadeždoj b. l’stit’ Maše flatter self.acc hope.instr flatter Masha.dat “flatter oneself with hope’ ‘flatter Masha” 97. Since reflexives in Russian are notoriously subject-oriented, they cannot be used as a test here. 98. Comrie and Stone (1978: 103) note that verbs bojat’sja “fear” and doždat’sja “wait for” are often attested with accusative objects in colloquial Russian (the latter only with animate objects). This is part of a more general shift from inherent case marking on verbal complements to either accusative or PP complements, which “has operated, slowly, throughout the history of Russian and other Slavonic languages” (Comrie and Stone 1978: 105). 99. The only exception to the correlation between quantificational force and genitive case marking is the verb kasat’sja “touch; concern”, which obligatorily takes a genitive object even when no quantificational force is present. I know of no account for this phenomenon. Note, however, that in the nineteenth century this verb “also occurred with the preposition do [‘up to’] (and less commonly with k [‘toward’], or with a prepositionless dative)” (Comrie and Stone 1978: 105). 100. Interestingly, some of these verbs, including rukovodit’ “supervise” and dirižirovat’ “conduct (orchestra)” have acquired the instrumental-assigning ability in the last century, whereas in the nineteenth century they “used to take the accusative” (Comrie and Stone 1978: 104). This spreading of inherent instrumental is not unlike the spreading of dative marking on objects in Icelandic (the so-called “dative sickness”; cf. Smith 1994, Svenonius 2001 and the references cited therein). Also, it must be noted that this “instrumental sickness” goes against the general trend “from oblique objects towards accusative or prepositional objects” (Comrie and Stone 1978: 104; see also fn. 99 above). 101. Cf. Levin and Rappaport Hovav (1995: 91). 102. Cf. Levin and Rappaport Hovav (1995: 226–227). Note that Levin and Rappaport Hovav (1995: 138, 226–227) analyze verbs of smell and light emission and verbs of body-internal motion as unergative. 103. For a further discussion of the latter construction, see Salnikov (1977), Green (1980), Babby (1980: 53–54, 1994), Lavine and Freidin (2002), Timberlake (2004: 336), and the references cited therein. 104. Nichols (1981: 218) cites one example of nominative complement of javljat’sja “be”, and attributes it to “an illiterate worker who attempts to imitate formal speech and uses the verb inappropriately and with the incorrect case”; furthermore, she also cites three examples of delat’sja “become” with a nominative complement and attributes them all to “highly colloquial speech” (p. 222). Rozental’ (1976: 37–38) mentions that the verbs stat’ “become” and sˇcitat’sja “be reputed as” are attested with nominative complement in colloquial speech and poetry, but not in standard literary language. This is a likely consequence of a general tendency in colloquial Russian to replace inherent cases with nominative; more on the verb stat’ “become” below in the main text. 105. As suggested to me by Steven Franks, this use of stat’ may be related to its inchoative use, as in Vanja stal prygat’ na odnoj nožke. “Vanya began to jump on one leg.” 106. It appears that instrumental is most felicitous on a new term that follows the verb nazyvat’sja “be called” if the sentence contains the so-called verum focus, associated with the emphatic particle i, which can be roughly translated as “in fact”. Nichols (1981: 164–165) provides the following minimal pair (note that, according to Nichols 1981: 165, instrumental here is ungrammatical rather than marginal; I do not agree with her judgment): (i) a. Èto i nazyvaetsja velosiped. this.nom emph is-called bicycle.nom “This is (what’s) called a bicycle.”
150
107.
108.
109.
110. 111. 112.
113.
NOTES b. Èto nazyvaetsja {velosiped / ? velosipedom}. this.nom is-called bicycle.nom / ? bicycle.instr “This is called a bicycle.” Similar facts are found with respect to the appositional construction like v romane “Evgenij Onegin” ‘in the novel Eugene Onegin’ and constructions with nomer “number”, as in Chekhov’s short story title Palata N˚6 ‘Ward number 6’. See Comrie and Stone (1978: 112–117). An alternative approach, adopted by Levin (1985) and others , takes -sja to be a case checking morpheme, which in virtue of this property “absorbs” the accusative case normally checked by the verb. However, arguments have been made to show that this “case absorption” is a result of the morpheme’s ability to manipulate the -grid of the verb. See references cited in the main text above for a detailed discussion. If a secondary predicate is associated with an instrumental argument, it is impossible to tell whether we are dealing with an instrumental or with an agreeing pattern; however, given the generalization in the text above, it is reasonable to assume that this too is an “agreeing secondary predicate”: (i) On napilsja cˇ ajem gorjaˇcim. he drank-to-the-full tea.instr hot.instr ‘He filled himself with tea hot.’ The specifier position of the PredP is occupied by a PRO controlled by the argument that the instrumental secondary predicate is associated with. Recall from Section 5.4.1 above that there are good arguments to analyze dative objects as being higher than accusative ones, rather than vice versa, as assumed by Bailyn and Citko (1999). Interestingly, Bailyn and Rubin (1991: 106–107, fn. 5) propose to treat “agreeing depictives” as appositives – in this book I adopt a similar line of analysis, as discussed below in the main text – but in his later work with Citko, Bailyn rejects this analysis without providing any arguments for doing so. Note the syncretism between the genitive and dative forms of the adjective; this is a purely morphological issue which does not affect the argument made in the main text.
REFERENCES
Abney S (1987) The English noun phrase and its sentential aspect. Dissertation, MIT Akiyama M (2000) On the LCA-based derivation of the X-Bar Theory. Sophia Linguistica 44/45:43–61 Anderson J M (1977) On case grammar: Prolegomena to a theory of grammatical relations. London: Croom Helm Antinucci F, Cinque G (1977) Sull’ordine delle parole in italiano: l’emarginazione. Studi di grammatica generativa 6:121–146 Arad M (1998) VP-structure and the Syntax-Lexicon Interface. Dissertation, University College London, distributed by MIT working papers in linguistics Babby L (1973) The Deep Structure of adjectives and participles in Russian. Language 49:349–360 Babby L (1975a) A Transformational Grammar of Russian adjectives. The Hague, Mouton Babby L (1975b) A Transformational analysis of transitive –SJA verbs in Russian. Lingua 35:297–332 Babby L (1980) Existential sentences and negation in Russian. Ann Arbor, MI, Karoma Babby L (1994) A Theta-theoretic analysis of Adversity Impersonal Sentences in Russian. In: Avrutin S et al (eds) Annual workshop on formal approaches to Slavic linguistics. The MIT meeting 1993. Ann Arbor, Michigan Slavic Publications, pp 25–67 Babyonyshev M (1996) Structural connections in syntax and processing. Studies in Russian and Japanese. Dissertation, MIT Bailyn J (1994) The syntax and semantics of Russian Long and Short Adjectives: An X’-Theoretic account. In: Toman J (ed) Annual workshop on formal approaches to Slavic linguistics. The Ann Arbor meeting: Functional categories in Slavic syntax. Ann Arbor, MI, Michigan Slavic Publications, pp 1–30 Bailyn J (1995) A configurational approach to Russian ‘free’ word order. Dissertation, Cornell Bailyn J (2002) Inversion, dislocation and optionality in Russian. In: Zybatov G et al (eds) Current issues in formal Slavic linguistics (FDSL 3). Frankfurt am Main, Peter Lang, pp 280–293 Bailyn J (2003a) Does Russian scrambling exist? In: Karimi S (ed) Word order and scrambling. Oxford, Blackwell, pp 156–176 Bailyn J (2003b) A (purely) derivational account of Russian scrambling. In: Partee B et al (eds) Annual workshop on formal approaches to Slavic linguistics 11. Ann Arbor, MI, University of Michigan Press Bailyn J (2004a) The Case of Q. In: Arnaudova O et al (eds) Annual workshop on formal approaches to Slavic linguistics 12. The Ottawa meeting 2003. Ann Arbor, MI, University of Michigan Press, pp 1–36 Bailyn J (2004b) Generalized Inversion. Natural Language and Linguistic Theory 22(1):1–50 Bailyn J, Citko B (1999) Case and agreement in Slavic predicates. In: Dziwirek K et al (eds) Annual workshop on formal approaches to Slavic linguistics. Ann Arbor, MI, Michigan Slavic Publishers, pp 17–38 Bailyn J, Rubin E (1991) The unification of instrumental case assignment in Russian. In: Toribio A et al (eds) Cornell working papers in linguistics 9. Department of Modern Languages and Linguistics, Cornell University, Ithaca, NY, pp 99–126 Baker M (1988) Incorporation: A theory of grammatical function changing. Chicago, University of Chicago Press Baker M (2003) Lexical categories. Verbs, nouns, and adjectives. Cambridge, UK, Cambridge University Press
151
152
REFERENCES
Baker M, Stewart OT (1997) Unaccusativity and the adjective-verb distinction: Èdó evidence. Proceedings of the North East Linguistic Society 27:33–47 Baker M, Stewart OT (1999) On double-headedness and the anatomy of the clause. Ms., Rutgers University Barbiers S (2000) The right periphery in SOV Languages: English and Dutch. In: Svenonius P (ed) The derivation of VO and OV. Philadelphia, John Benjamins, pp 181–218 Barwise J, Cooper R (1981) Generalized Quantifiers and Natural Language. Linguistics and Philosophy 4:159–219 Benincà P (1988) L’ordine degli elementi della frase e le costruzioni marcate. In: Renzi L, Salvi G (eds) Grande grammatica italiana di consultazione, vol I. Bologna, Il Mulino, pp 115–119; 129–194 Benincà P, Salvi G (1988) L’ordine normale degli elementi nella frase semplice. In: Renzi L, Salvi G (eds) Grande grammatica italiana di consultazione, vol I. Bologna, Il Mulino, pp 119–129 Benveniste E (1966) Être et avoir dans leurs fonctions linguistiques. In: Benveniste E (ed) Problèmes de linguistique générale. Paris, Gallimard, pp 187–207 Berman R (1981) Dative marking on the affected in Modern Hebrew. In: Borer H, Aoun Y (eds) Theoretical issues in the grammar of Semitic languages. MIT Working Papers in Linguistics 3:150–179 Bhat DNS (1994) The adjectival category. Criteria for differentiation and identification. Amsterdam, John Benjamins Bittner M, Hale K (1996) The structural determination of Case and Agreement. Linguistic Inquiry 27:1–68 Blake B (1994) Case. Cambridge, Cambridge University Press Blauvelt Y (1980) Russian verbal government. Dissertation, Ohio State University Bobaljik J (1995) Morphosyntax: The syntax of verbal inflection. Dissertation, MIT Borer H (2004) Structuring sense. An exo-skeletal trilogy. Vol 1: In name only. Oxford, Oxford University Press Borer H (2005) Structuring sense. An exo-skeletal trilogy. Vol 2: The normal course of events. Oxford, Oxford University Press Borer H, Grodzinsky Y (1986) Syntactic cliticization and lexical cliticization: the case of Hebrew dative clitics. In: Borer H (ed) Syntax and Semantics 19. New York, Academic Press, pp 175–217 Borovikoff N (2001) Unaccusativity and movement in Russian: integrating formal syntax and discourse functions. Dissertation, Indiana University Bloomington Borthen K (2001) Treating predicative and nonpredicative bare singulars as a unitary phenomenon. Paper presented at the Predication conference, Trondheim, October 2001 Boškovi´c Ž (2001) On the nature of the syntax-phonology interface: cliticization and related phenomena. North-Holland Linguistic Series, vol 60, Amsterdam, Elsevier Boškovi´c Ž (2005) On the locality of left branch extraction and the structure of NP. Studia Linguistica 59(1):1–45 Bowers J (1991) The syntax and semantics of nominals. In: Moore S, Wyner AZ (eds) Cornell working papers in linguistics 10. Ithaca, NY, Cornell University, pp 1–30 Bowers J (1993) The syntax of predication. Linguistic Inquiry 24:591–656 Bresnan J (1978) A realistic Transformational Grammar. In: Halle M et al (eds) Linguistic theory and psychological reality. Cambridge, MA, The MIT Press, pp 1–59 Bresnan J (1982) Control and complementation. Linguistic Inquiry 13:343–434 Brody M (1993) -Theory and arguments. Linguistic Inquiry 24(1):1–23 Brody M (2000) Word order, restructuring and Mirror Theory. In: Svenonius P (ed) The derivation of VO and OV. Philadelphia, John Benjamins, pp 27–44 Burzio L (1986) Italian syntax. Dordrecht, Reidel Butt M (1995) The structure of complex predicates in Urdu. Stanford, CA, CSLI Publications Butt M, Ramchand G (2001) Complex aspectual structure in Hindi/Urdu. In: Jensen B et al (eds) Oxford working papers in linguistics 6. Oxford, Oxford University, pp 1–30 Börjars K (1998) Feature distribution in Swedish noun phrases. Oxford, Blackwell Publishers Cardinaletti A, Giusti G (2000) “Semi-lexical” motion verbs in Romance and Germanic. University of Venice working papers in linguistics 10(2):1–43
REFERENCES
153
Cardinaletti A, Guasti MT (1995) Small Clauses: Some controversies and issues in acquisition. In: Cardinaletti A, Guasti MT (eds) Syntax and semantics 28: Small Clauses. New York, Academic Press, pp 1–23 Carpenter B (1997) Type-logical semantics. Cambridge, MA, The MIT Press Cecchetto C (1999) A comparative analysis of Left and Right Dislocation in Romance. Studia Linguistica 53(1):40–67 Channon R (1987) A function of the instrumental case in Russian. In: Channon R, Shockey L (eds) In honour of Iste Lehiste. Dordrecht, Foris, pp 339–355 Chierchia G (1998) Reference to kinds across languages. Natural Language Semantics 6:339–405 Chierchia G, McConnell-Ginet S (1990) Meaning and grammar. Cambridge, MA, The MIT Press Chomsky N (1970) Remarks on nominalization. In: Jacobs R, Rosenbaum PS (eds) Reading in English Transformational Grammar. Waltham, MA, pp 184–221. (Also in Studies on semantics in Generative Grammar. The Hague, Mouton, pp 11–61.) Chomsky N (1981) Lectures in Government and Binding. Dordrecht, Foris Chomsky N (1993) A Minimalist Program for linguistic theory. In: Hale K, Keyser SJ (eds) The view from Building 20. Cambridge, MA, The MIT Press, pp 1–52 Chomsky N (1995) The Minimalist Program for linguistic theory. Cambridge, MA, The MIT Press Chomsky N (1998) Some observations on Economy in Generative Grammar. In: Barbosa P et al (eds) Is the best good enough? Cambridge, MA, The MIT Press, pp 115–127 Chomsky N (2000) Minimalist inquiries: The framework. In: Martin R et al (eds) Step by step: Essays on Minimalist syntax in honor of Howard Lasnik. Cambridge, MA, The MIT Press Chung S, Ladusaw W (2004) Restriction and saturation. Cambridge, MA, MIT Press Chvany C (1975) On the syntax of BE-sentences in Russian. Cambridge, MA, Slavica Cinque G (1990) Ergative adjectives and the Lexicalist Hypothesis. Natural Language and Linguistic Theory 8(1):1–40 Cinque G (1994) Evidence for partial N-Movement in the Romance DP. In: Cinque G et al (eds) Paths towards Universal Grammar. Studies in honor of Richard S. Kayne. Washington, DC, Georgetown University Press, pp 85–110 Cinque G (1999) Adverbs and functional heads: A cross-linguistic perspective. Oxford, Oxford University Press Cinque G (2004) “Restructuring” and functional structure. Ms., Università di Venezia Citko B (2006a) Small Clauses: not so small and not all alike. Ms., University of Washington Citko B (2006b) Copula sentences reconsidered. In: Lavine J et al (eds) Annual workshop on formal approaches to Slavic linguistics 14: The Princeton meeting. Ann Arbor, MI, Michigan Slavic Publications Comrie B (1987) Russian. In: Comrie B (ed) The world’s major languages. Oxford, Oxford University Press, pp 329–347 Comrie B (1997) The typology of predicate case marking. In: Bybee J et al (eds) Essays on language function and language type dedicated to T. Givon. Amsterdam, Benjamins, pp 39–50 Comrie B, Stone G (1978) The Russian language since the Revolution. Oxford, Clarendon Press Contreras H (1995) Small Clauses and complex predicates. In: Cardinaletti A, Guasti MT (eds) Syntax and semantics 28: Small Clauses. New York, Academic Press, pp 135–152 Crockett D (1976) Agreement in Contemporary Standard Russian. Columbus, OH, Slavica Publishers Czinglar C, Köhler K (2001) (How) Do children distinguish between subjects and predicates? Paper presented at the Predication conference, Trondheim, October 2001 Dahl Ö (1985) Tense and aspect systems. Oxford, Blackwell Publishers Delsing LO (1988) The Scandinavian noun phrase. Working papers in Scandinavian syntax 42:57–80 Delsing LO (1992) On attributive adjectives in Scandinavian and other languages. In: Holmberg A (ed) Papers from the workshop on the Scandinavian noun phrase. Umeå, University of Umeå, pp 20–44 Delsing LO (1993) On attributive adjectives in Scandinavian and other languages. Studia Linguistica 47(2):105–125 Dik S (1983) On the status of verbal reflexives. Communication and Cognition 16(1/2):39–63
154
REFERENCES
Dikken M den (2006) Relators and linkers. The syntax of predication, predicate inversion and copulas. Cambridge, MA, MIT Press Dixon RMW (1994) Ergativity. Cambridge, Cambridge University Press Dobrovie-Sorin C (1987) A propos de la structure du groupe nominal en roumain. Rivista di grammatica generativa 12:123–152 Doron E (1983) Verbless predicates in Hebrew. Dissertation, University of Texas at Austin Dowty D (1991) Thematic proto-roles and argument selection. Language 67:547–619 Dowty D, Wall R, Peters S (1981) Introduction to Montague semantics. Synthese Language Library, Dordrecht, Reidel Duffield N (1999) Adjectival modifiers and the specifier-adjunct distinction. In: Adger D et al (eds) Specifiers: Minimalist approaches. Oxford, Oxford University Press Eide K, Åfarli T (1999) The syntactic disguises of the predication operator. Studia Linguistica 53(2): 155–181 Eisenbeiss S (2001) The role of asymmetric relations between arguments in the acquisition of Case. Paper presented at the Asymmetry Conference, UQAM, Montreal Engelhardt M, Trugman H (1998) D as a source of adnominal genitive in Russian. In: Boškovi´c Ž et al (eds) Annual workshop on formal approaches to Slavic linguistics. The Connecticut meeting 1997. Ann Arbor, MI, Michigan Slavic Publications, pp 114–133 Epstein SD, Hornstein N (eds) (1999) Working Minimalism. Cambridge, MA, The MIT Press Evans G (1980) Pronouns. Linguistic Inquiry 11(2):337–362 Falk Y (1991) Case: Abstract and morphological. Linguistics 29(2):197–230 Falk Y (1997) Case typology and Case theory. Ms., Hebrew University, Jerusalem. Also available from Fiengo R, May R (1994) Indices and identity. Linguist Inquiry Monograph, Cambridge, MA, The MIT Press Fillmore C (1968) The case for case. In: Bach E, Harms RT (eds) Universals in linguistic theory. London, Holt, Reinhart & Winston, pp 1–88 Fowler G (1987) Generalizing the notion of ascension in Russian. University of Chicago working papers in linguistics 3:75–85 Fowler G (1988) Grammatical relation 0. University of Chicago working papers in linguistics 4:90–101 Fowler G (1996) Oblique passivization in Russian. Slavic and East European Journal 40(3):519–546 Fowler G (1997) Toward a rapprochement between form and intuition: Approaches to the Russian double nominative construction. In: Lindseth M, Franks S (eds) Annual workshop on formal approaches to Slavic linguistics. The Indiana meeting of 1996. Ann Arbor, MI, Michigan Slavic Publications, pp 144–165 Franks S (1995) Parameters of Slavic morphosyntax. Oxford, Oxford University Press Franks S (1999) Optimality Theory and clitics at PF. In: Dziwirek K et al (eds) Annual workshop on formal approaches to Slavic linguistics. Ann Arbor, MI, Michigan Slavic Publications, pp 101–122 Franks S, Boškovi´c Ž (2001) An argument for multiple Spell-Out. Linguistic Inquiry 32:174–183 Franks S, House R (1982) Genitive themes in Russian. CLS 18:156–168, Chicago Linguistic Society, Chicago Franks S, King TH (2000) A handbook of Slavic clitics. Oxford, Oxford University Press Franks S, Pereltsvaig A (2004) Functional categories in the nominal domain. In: Arnaudova O et al (eds) Annual workshop on formal approaches to Slavic linguistics 12. The Ottawa meeting 2003. Ann Arbor, MI, University of Michigan Press, pp 109–128 Franks S, Yadroff M (2002) The origin of prepositions. In: Junghanns U et al (eds) Current issues in formal Slavic linguistics (FDSL 3). Peter Lang, Frankfurt am Main Frascarelli M (2000) The syntax-phonology interface in focus and topic constructions in Italian. Dordrecht, Kluwer Fukui N, Speas M (1986) Specifiers and projection. MIT working papers in linguistics 8:128–172 Geach PT (1962) Reference and generality: An examination of some medieval and modern theories. Ithaca, NY, Cornell University Press
REFERENCES
155
Geist L (1999) Russisch byt’ (‘sein’) als funktionale und/oder lexikalische Kategorie. ZAS papers in linguistics 14:1–39 Geist L (2002) Russisch byt’: zwei Kopulae, zwei Kasus, ein integrierter Ansatz. In: Kosta P, Frasek J (eds) Current approaches to formal Slavic linguistics. Contributions of the Second Conference on Formal Description of Slavic Languages (FDSL II). Potsdam University, November 20-22, 1997. Frankfurt am Main, Lang, pp 169–178 Goosens L (1992) Graded predicationality, semantization and BE in a Functional Grammar of English. Belgian Journal of Linguistics 7:53–73 Graudina LK, Ickovich VA, Katlinskaja LP (1976) Grammaticheskaja pravil’nost’ russkoj rechi. Opyt chastotno-stilisticheskogo slovarja variantov. Moscow, Izdatel’stvo Nauka Green MC (1980) On the syntax and semantics of impersonal sentences in Russian: a study of the sentence type ‘vetrom uneslo lodku’. Dissertation, Cornell University Greenberg JH (1963) Some universals of grammar with particular reference to the order of meaningful elements. In: Greenberg JH (ed) Universals of Language. Cambridge, MA, The MIT Press, pp 73–113 Grenoble L (1998) Deixis and information packaging in Russian discourse. Pragmatics and beyond new series 50. Amsterdam, John Benjamins Grice HP (1975) Logic and conversation. In: Cole P, Morgan JL (eds) Syntax and semantics, vol 3: Speech acts. New York, Academic Press, pp 41–58 Grimshaw J, Mester A (1988) Light verbs and -marking. Linguistic Inquiry 19(2):205–232 Guasti MT (1988) La struttura interna del sintagma aggettivale. In: Renzi L, Salvi G (eds) Grande grammatica italiana di consultazione, vol II. Bologna, Il Mulino, pp 321–340 Guéron J, Hoekstra T (1995) The temporal interpretation of predication. In: Cardinaletti A, Guasti MT (eds) Syntax and semantics 28: Small Clauses. New York, Academic Press, pp 77–107 Haegeman L (2000) Remnant movement and OV order. In: Svenonius P (ed) The derivation of VO and OV. Philadelphia, John Benjamins, pp 69–96 Haider H (2000) OV is more basic than VO. In: Svenonius P (ed) The derivation of VO and OV. Philadelphia, John Benjamins, pp 45–68 Hale K, Keyser SJ (1986) Some transitivity alternations in English. Lexicon Project working papers 7, MIT Hale K, Keyser SJ (1987) A view from the middle. Lexicon project working papers 10, MIT Hale K, Keyser SJ (1991) On the syntax of argument structure. The lexicon project monograph series. Cambridge, MA, The MIT Center for Cognitive Science Hale K, Keyser SJ (1993) On argument structure and the lexical expression of syntactic relations. In: Hale K, Keyser SJ (eds) The view from Building 20. Cambridge, MA, The MIT Press, pp 53–109 Halliday M, Kerkwood A (1967) Notes on transitivity and theme in English. Journal of Linguistics 3:37–81 Heggie L (1988) The syntax of copular structures. Dissertation, University of Southern California Heycock C (1994) Layers of predication: The non-lexical syntax of clauses. New York, Garland Publishing Heycock C, Kroch A (1999) Pseudocleft connectedness: implications for the LF interface level. Linguistic Inquiry 30(3):365–398 Higginbotham J (1985) On semantics. Linguistic Inquiry 16:547–593 Higginbotham J (1989) Elucidation of meaning. Linguistics and Philosophy 12:465–518 Higgins FR (1973) The pseudo-cleft construction in English. Dissertation, MIT. Published by New York, Garland Press, 1979 Hinterhölzl R (2001) Semantic constraints on case assignment in secondary adjectival predicates in Russian. ZAS papers in linguistics 22:99–112 Holmberg A (1993) On the structure of predicate NP. Studia Linguistica 47(2):126–138 Holmberg A (2000) Deriving OV order in Finnish. In: Svenonius P (ed) The derivation of VO and OV. Philadelphia, John Benjamins, pp 123–152 Hornstein N, Lightfoot D (1987) Predication and PRO. Language 63:23–52 House R (1982) The use of genitive-initial sentences for the specification of quantity in Russian. Dissertation, Cornell University
156
REFERENCES
Hróarsdóttir Þ (2000) Parameter change in Icelandic. In: Svenonius P (ed) The derivation of VO and OV. Philadelphia, John Benjamins, pp 153–180 Isaˇcenko AI (1960) Grammatiˇceskij stroj russkogo jazyka v sopostavlenii s slovackim. Bratislava, SAV Jackendoff R (1972) Semantic interpretation in Generative Grammar. Cambridge, MA, MIT Press Jackendoff R (1977) X-Bar-Syntax: A study of phrase structure. Cambridge, MA, MIT Press Jackendoff R (1987) The status of thematic relations in linguistic theory. Linguistic Inquiry 18:369–411 Jakab E (2003) Asymmetry in case: Finnish and Old Russian nominative objects. In: Di Sciullo A-M (ed) Asymmetry in grammar. Vol 1: Syntax and semantics. Linguistik Aktuell/Linguistics Today 57. Amsterdam, John Benjamins, pp 51–84 Jakobson R (1936/1971) Beitrag zur Allgemeinen Kasuslehre. Gesamtbedeutungen der Russischen Kasus. In: Selected Writings 2. The Hague, Mouton, pp 23–71 Jakobson R (1958/1971) Morfologicheskie nabludenija nad slavjanskim skloneniem. In: Selected Writings 2. The Hague, Mouton, pp 154–183 Julien M (2005) Nominal phrases from a Scandinavian perspective. Amsterdam, John Benjamins Junghanns U (1997) On the so-called ÈTO-cleft construction. In: Lindseth M, Franks S (eds) Annual workshop on formal approaches to Slavic linguistics. The Indiana meeting of 1996. Ann Arbor, MI, Michigan Slavic Publications, pp 166–190 Kahn CH (1973) The verb ‘be’ in Ancient Greek. Dordrecht, Reidel Kayne RS (1985) L’accord du participe passé en français et en italien. Modèles linguistiques 7:73–89 Kayne RS (1994) The antisymmetry of syntax. Linguistic Inquiry Manuscript, Linguistic Inquiry Monograph. Cambridge, MA, The MIT Press Kerstens J (1993) The syntax of number, person and gender: A theory of phi-features. Berlin, Mouton de Gruyter Kester E-P (1993) The inflectional properties of Scandinavian adjectives. Studia Linguistica 47(2): 139–153 Kilby D (1986) The instrumental in Russian. In: Brecht RD, Levine JS (eds) Case in Slavic. Columbus, OH, Slavica Publishers, pp 323–337 King TH (1995) Configuring topic and focus in Russian. Stanford, CSLI Publications Koizumi M (1995) Phrase structure in Minimalist syntax. Dissertation, MIT Kozyreva TG, Khmelevskaja ES (eds) (1972) Sovremennyj russkij jazyk: Imja prilagatel’noe, imja chislitel’noe, mestoimenie. Minsk, Vyshejshaja Shkola Kratzer A (1996) Severing the external argument from its verb. In: Rooryck J, Zaring L (eds) Phrase structure and the lexicon. Dordrech, Kluwer, pp 109–138 Kuznetsova J (2005) Against the Russian distributive PO-construction as a diagnostic for unaccusativity. In: Franks S et al (eds) Annual workshop on formal approaches to Slavic linguistics 13: The South Carolina meeting. Ann Arbor, MI, Michigan Slavic Publications, pp 170–180 Landau I (1999) Projection and bounding in possessor raising. In: Shahin K et al (eds) Proceedings of the WCCFL XVII. Stanford, CA, CSLI Publications, pp 405–419 Lapteva OA (1976) Russkij razgovornyj syntaksis. Moscow, Nauka Lapteva OA (1999) Živaja russkaja reˇc’ s teleèkrana. 2nd edn, Moscow Larson R (1988) On the double object construction. Linguistic Inquiry 19:589–632 Larson R, Segal G (1995) Knowledge and meaning: An introduction to semantic theory. Cambridge, MA, The MIT Press Lavine J (1999) Subject properties and ergativity in North Russian and Lithuanian. In: Dziwirek K et al (eds) Annual workshop on formal approaches to Slavic linguistics. Ann Arbor, MI, Michigan Slavic Publications, pp 307–328 Lavine J (2000) Topics in the syntax of nonagreeing predicates in Slavic. Dissertation, Princeton Lavine J, Franks S (2005) On nominative objects. In: Franks S et al (eds) Annual workshop on formal approaches to Slavic linguistics 13: The South Carolina meeting. Ann Arbor, MI, Michigan Slavic Publications, pp 195–206 Lavine J, Freidin R (2002) The Subject of defective T(ense) in Slavic. Journal of Slavic Linguistics 10(1/2):253–289
REFERENCES
157
Lee J-S (1989) Case assignment to predicate nominals and the Case Filter in Korean. Harvard studies in Korean linguistics III, pp 277–287 Lekant PA (1995) Funkcii svjazok v russkom jazyke. Russkij jazyk v shkole 3:90–95 Lepschy G (ed) (1998) History of linguistics. Renaissance and early modern linguistics. London, Longman Levin B (1985) Case theory and the Russian reflexive affix. In: Goldberg J et al (eds) Proceedings of the West Coast Conference on Formal Linguistic 4. Stanford, CSLI Publications, pp 178–191 Levin B, (1986) What to do with theta-roles. Lexicon project working papers 11, MIT Levin B, Rappaport Hovav M (1995) Unaccusativity: At the syntax-lexical semantics interface. Linguistic Inquiry Monograph, Cambridge, MA, The MIT Press Li Y (1990) X˚-binding and verb incorporation. Linguistic Inquiry 21:399–426 Longobardi G (1994) Reference and proper names: A theory of N-movement in syntax and Logical Form. Linguistic Inquiry 25(4):609–665 Lonzi L (1985) Pertinenza della struttura tema-rema per l’analisi sintattica. In: Stammerjohann H (ed) Theme-rheme in Italian. Tübingen, Gunter Narr, pp 99–120 MacLaughlin D (1997) The structure of determiner phrases: Evidence from American Sign Language. Dissertation, Boston University Maling J, Sprouse RA (1995) Structural Case, Specifier-Head relations, and the Case of predicate NPs. In: Haider H et al (eds) Studies in comparative Germanic syntax. Dordrecht, Kluwer, pp 167–186 Mallinson G, Blake BJ (1981) Language typology: Cross-linguistic studies in syntax. Amsterdam, North-Holland Manzini R (1983) Restructuring and reanalysis. Dissertation, MIT Marantz A (1997) No escape from syntax: don’t try morphological analysis in the privacy of your own lexicon. Penn working papers in linguistics, University of Pennsylvania Martin R (1999) Case, the Extended Projection Principle, and Minimalism. In: Epstein SD, Hornstein N (eds) Working Minimalism. Cambridge, MA, The MIT Press, pp 1–26 Matushansky O (2000) The instrument of inversion: Instrumental case and verb raising in the Russian copula. In: Billerey R et al (eds) Proceedings of WCCFL 19. Somerville, MA, Cascadilla Press, pp 101–115 McCawley J (1989) Individuation in and of syntactic structures. In: Baltin MR, Kroch A (eds) Alternative conceptions of phrase structure. Chicago, The University of Chicago Press, pp 117–138 McCoy S (1998) Individual-level predicates and Pronoun Doubling in Colloquial Russian. In: Boskovic Ž et al (eds) Annual workshop on formal approaches to Slavic linguistics. The Connecticut meeting 1997, Ann Arbor, MI, Michigan Slavic Publications Mikkelsen L (2002) Specification is not inverted predication. In: Hirotani M (ed) Proceedings of NELS 32:403–422. University of Massachusetts, Amherst, GLSA Montague R (1973) The proper treatment of quantification in English. Reprinted in Montague R (1974) Formal philosophy: Selected papers of Richard Montague. New Haven, Yale University Press Moro A (1988) Per una teoria unificata delle frasi copulari. Rivista di grammatica generativa 13:81–110 Moro A (1997) The raising of predicates: Predicative noun phrases and the theory of clause structure. Cambridge, Cambridge University Press Moro A (2000) Dynamic antisymmetry. Linguistic Inquiry Monograph. Cambridge, MA, The MIT Press Neeleman A, Weerman F (1999) Flexible syntax: A theory of Case and arguments. Dordrecht, Kluwer Neidle C (1982) Case agreement in Russian. In: Bresnan J (ed) The mental representation of grammatical relations. Cambridge, MA, The MIT Press, pp 391–426 Newmeyer F (1998) The irrelevance of typology for grammatical theory. Syntaxis 1:161–197 Nichols J (1973) The Balto-Slavic predicate instrumental: A problem in diachronic syntax. Dissertation, University of California Berkeley Nichols J (1981) Predicate nominals: A partial surface syntax of Russian. Berkeley, University of California Press Noyer R (1998) Vietnamese ‘morphology’ and the definition of word. In: Dimitriadis A et al (eds) University of Pennsylvania working papers in linguistics. Philadelphia, University of Pennsylvania, Department of Linguistics, pp 65–89
158
REFERENCES
Nunes J (1999) Linearization of chains and phonetic realization of chain links. In: Epstein SD, Hornstein N (eds) Working Minimalism. Cambridge, MA, The MIT Press, pp 217–250 Partee B (1986) Ambiguous pseudoclefts with unambiguous BE. In: Berman S et al (eds) Proceedings of NELS 16. University of Massachusetts, Amherst, pp 354–366 Partee B (1987) Noun phrase interpretation and type-shifting principles. In: Groenendijk J et al (eds) Studies in discourse representation theory and the theory of generalized quantifiers. Dordrecht, Foris, pp 115–143 Partee B (1999) Copula inversion puzzles in English and Russian. In: Dziwirek K et al (eds) Annual workshop on formal approaches to Slavic linguistics. Ann Arbor, MI, Michigan Slavic Publishers, pp 361–396 Partee B (ed) (1976) Montague grammar. New York, Academic Press Pereltsvaig A (1998) Non-Argument Bare Genitive Construction in Russian and the autonomy of syntax. Ms., McGill University Pereltsvaig A (2000) On accusative adverbials in Russian and Finnish. In: Alexiadou A, Svenonius P (eds) Adverbs and adjunction. Linguistics in Potsdam 6, pp 155–176 Pereltsvaig A (2001a) Syntactic categories are not primitive: Evidence from Short and Long Adjectives in Russian. In: Franks S, King TH (eds) Annual workshop on formal approaches to Slavic linguistics. Ann Arbor, MI, Michigan Slavic Publishers Pereltsvaig A (2001b) Review of Moro 2000 Dynamic antisymmetry. Linguist List 12.1030 Pereltsvaig A (2001c) On the nature of intra-clausal relations: A study of copular sentences in Russian and Italian. Dissertation, McGill University Pereltsvaig A (2004) Topic and focus as linear notions: Evidence from Italian and Russian. Lingua 114(3):325–344 Pereltsvaig A (2006a) Small nominals. Natural Language and Linguistic Theory 24(2):433–500 Pereltsvaig A (2006b) On the universality of DP: A view from Russian. Studia Linguistica (in press) Pereltsvaig A (2006c) Split phrases in Colloquial Russian: A corpus study. In: Goledzinowska M et al. (eds.) Annual workshop on formal approaches to Slavic linguistics 15: The Toronto meeting. Ann Arbor, MI, Michigan Slavic Publications (in press) Pereltsvaig A (forthcoming) What does American Russian morphology tell us about syntax? To appear: Journal of Slavic Linguistics Perlmutter DM (1978) Impersonal passives and the Unaccusative Hypothesis. In: Proceedings of the Fourth annual meeting of the Berkeley Linguistics Society. Berkeley Linguistics Society, University of California, Berkeley, pp 157–189 Pesetsky D (1982) Paths and categories. Dissertation, MIT Pesetsky D (1998) Some Optimality principles of sentence pronunciation. In: Barbosa P et al (eds) Is the best good enough? Cambridge, MA., The MIT Press, pp 337–383 Pesetsky D, Torrego E (2000) T-to-C movement: Causes and consequences. To appear in Kenstowicz M (ed) Ken Hale: A life in language. Cambridge, MA, The MIT Press Peshkovskij AM (1914/1956) Russkij sintaksis v nauchnom osveschenii. 7th edition. Moscow, Uchpedgiz Polinsky M (1994) The grammar of case loss: American Russian. Paper presented at AATSEEL, San Diego Polinsky M (1995) Cross-linguistic parallels in language loss. Southwestern Journal of Linguistics 14(1/2):1–45 Polinsky M (1996) American Russian: An endangered language? Ms., University of California San Diego. Available at Polinsky M (1997) American Russian: Language loss meets language acquisition. In: Browne W et al (eds) Annual workshop on formal approaches to Slavic linguistics. Ann Arbor, MI, Michigan Slavic Publishers, pp 370–406 Polinsky M (1998) A composite linguistic profile of a speaker of Russian in the U.S. In: Kagan O, Rifkin B (eds) The learning and teaching of Slavic languages and cultures, pp 437–465 Pollard C, Sag I (1994) Head Driven Phrase Structure Grammar. Chicago, IL, The University of Chicago Press
REFERENCES
159
Pollock J-Y (1989) Verb movement, Universal Grammar, and the structure of IP. Linguistic Inquiry 20(3):365–424 Progovac L (1998) Determiner phrase in a language without determiners. Journal of Linguistics 34: 165–179 Quine WVO (1960) Word and object. Cambridge, MA, MIT Press Quirk R, Greenbaum S (1973) A university grammar of English. London, Longman Rapoport TR (1987) Copular, nominal and Small Clauses: A study of Israeli Hebrew. Dissertation, MIT Rappaport GC (1983) On anaphora and control in Russian. In: Harbert W (ed) Cornell University working papers in linguistics 4, pp 201–221 Rappaport GC (1986) On anaphor binding in Russian. Natural Language and Linguistic Theory 4(1): 97–120 Rappaport GC (1998) The Slavic noun phrase. Position paper presented at the Comparative Slavic morphosyntax: State of the art, Indiana University, Bloomington Rappaport GC (2001) The geometry of the Polish nominal phrase: Problems, progress, and prospects. In: Banski P, Przepiorkowski A (eds) Generative linguistics in Poland: Syntax and morphosyntax. Warsaw, Polish Academy of Sciences, pp 173–89 Rappaport GC (2006) Toward a theory of the grammatical use of lexical information. In: Lavine J et al (eds) Annual workshop on formal approaches to Slavic linguistics 14: The Princeton meeting. Ann Arbor, MI, Michigan Slavic Publications, pp 293–318 Renzi L (1988) l’articolo. In: Renzi L (1988) Grande grammatica italiana di consultazione, vol I. Bologna, Il Mulino, pp 357–424 Ritter E (1988) A Head-movement approach to construct state noun phrases. Linguistics 26:909–929 Ritter E (1991) Two functional categories in noun phrases: Evidence from Modern Hebrew. In: Rothstein S (ed) Syntax and semantics 25: Perspectives on phrase structure. New York, Academic Press, pp 37–62 Rizzi L (1997) The fine structure of the Left Periphery. In: Haegeman L (ed) Elements of grammar. Dordrecht, Kluwer, pp 281–337 Rosen ST (1989) Argument structure and complex predicates. Dissertation, Brandeis University Rothstein R (1986) Equation vs. ascription: The nominative / instrumental opposition in West Slavic. In: Brecht RD, Levine JS (eds) Case in Slavic. Columbus, OH, Slavica Publishers, pp 312–322 Rothstein S (1983) The syntactic forms of predication. Dissertation, MIT Rothstein S (1995) Small Clauses and copular constructions. In: Cardinaletti A, Guasti MT (eds) Small Clauses. Syntax and semantics 28. New York, Academic Press, pp 27–48 Rouveret A, Vergnaud J-R (1980) Specifying reference to the Subject: French causatives and conditions on representations. Linguistic Inquiry 11:97–202 Rozental’ DE (1976) Sovremennyj russkij jazyk. Vol 2: Syntax. Moscow, Vysshaja shkola Russell B (1919) The philosophy of mathematics. London, Allen & Unwin Rutkowski P (2002) Noun/pronoun asymmetries: evidence in support of the DP hypothesis in Polish. Jezikoslovlje 3(1/2):159–170 Saccon G (1992) VP-internal arguments and locative subjects. In: Proceedings of NELS 22:383–397 Sadler L, Arnold D (1994) Prenominal adjectives and the phrasal/lexical distinction. Journal of Linguistics 30:187–226 Salnikov N (1977) Bezlichnye predlozhenia tipa ‘kryshu sorvalo vetrom’. Russian Linguistics 3:271–292 Salvi G (1988a) Le frasi copulative. In: Renzi L, Salvi G (eds) Grande grammatica italiana di consultazione, vol II. Bologna, Il Mulino, pp 163–189 Salvi G (1988b) L’accordo. In: Renzi L, Salvi G (eds) Grande grammatica italiana di consultazione, vol II. Bologna, Il Mulino, pp 227–244 Samek-Lodovici V (1996) Constraints on Subjects: An Optimality Theoretic analysis. Dissertation, Rutgers University Samek-Lodovici V (2003) The internal structure of arguments and its role in complex predicate formation. Natural Language and Linguistic Theory 21(4):835–881 Santelmann L (1993) The distribution of double determiners in Swedish: DEN support in D˚. Studia Linguistica 47(2):154–176
160
REFERENCES
Schein B (1995) Small Clauses and predication. In: Cardinaletti A, Guasti MT (eds) Syntax and semantics 28: Small Clauses. New York, Academic Press, pp 49–76 Sekerina I (1997) The syntax and processing of scrambling constructions in Russian. Dissertation, CUNY Siegel M (1976) Capturing the Russian adjective. In: Partee B (ed) Montague grammar. New York, Academic Press, pp 293–309 Sigurðsson H Á (1993) The structure of the Icelandic NP. Studia Linguistica 47(2):177–197 Sirotinina OB (1974) Sovremennaja razgovornaja reˇc’ i ee osobennosti. Moscow, Prosveschenie Slobin D (1966) The acquisition of Russian as a Natural Language. In: Smith F, Miller GA (eds) The genesis of language: A psycholinguistic perspective. Cambridge, MA, The MIT Press, pp 129–148 Smith CS (1991) The parameter of aspect. Dordrecht, Kluwer Smith H (1994) “Dative sickness” in Germanic. Natural Language and Linguistic Theory 12:675–736 Smith MB (1999) From instrumental to irrealis: Motivating some grammaticalized senses of the Russian instrumental. In: Dziwirek K et al (eds) Annual workshop on formal approaches to Slavic linguistics. Ann Arbor, MI, Michigan Slavic Publishers, pp 413–433 Sorace A (2000) Gradients in auxiliary selection with intransitive verbs. Language 76:859–890 Speas M (1990) Phrase structure in Natural Language. Dordrecht, Kluwer Sportiche D (1995) French predicate clitics and clause structure. In: Cardinaletti A, Guasti MT (eds) Syntax and semantics 28: Small Clauses. New York, Academic Press, pp 287–324 Starke M (1995) On the formal for Small Clauses. In: Cardinaletti A, Guasti MT (eds) Syntax and semantics 28: Small Clauses. New York, Academic Press, pp 237–269 Starosta S (1971) Some lexical redundancy rules for English nouns. Glossa 5:167–201 Starosta S (1988) The case for lexicase. London, Pinter Stowell T (1981) Origins of phrase structure. Dissertation, MIT Stowell T (1983) Subjects across categories. The Linguistic Review 2:285–312 Stowell T (1989) Subjects, specifiers and X-Bar Theory. In: Baltin M, Kroch A (eds) Alternative conceptions of phrase structure. Chicago, The University of Chicago Press, pp 232–262 Svenonius P (1993a) Selection, adjunction, and concord in the DP. Studia Linguistica 47(2):198–220 Svenonius P (1993b) The structural location of the attributive adjective. In: Duncan E et al (eds) The Proceedings of the Twelfth West Coast Conference on Formal Linguistics, Stanford, CA, CSLI, pp 439–454 Svenonius P (1994) Dependent nexus: Subordinate predication structures in English and the Scandinavian languages. University of California at Santa Cruz Svenonius P (2001) Case and event structure. In: Zhang N (ed) ZASPIL 26 (Zentrum für allgemeine Sprachwissenschaft Papers in Linguistics) Svenonius P (2002) Case is uninterpretable aspect. In: Proceedings of Perspectives on Aspect. Utrecht Talmy L (2000) Spatial structuring in spoken language and its relation to that in sign language. Paper presented at UQAM, Montreal Taraldsen KT (1990) D-projections and N-projections in Norwegian. In: Nespor M, Mascaró J (eds) Grammar in progress. Dordrecht, Foris, pp 419–431 Taraldsen KT (1995) Participle-based Small Clause complements of FÅ ‘get’ in Norwegian. In: Cardinaletti A, Guasti MT (eds) Syntax and semantics 28: Small Clauses. New York, Academic Press, pp 207–233 Taraldsen KT (2000) V-movement and VP movement in derivations leading to VO-order. In: Svenonius P (ed) The derivation of VO and OV. Philadelphia, John Benjamins, pp 97–122 Timberlake A (2004) A reference grammar of Russian. Cambridge, UK, Cambridge University Press Travis L (1988) The syntax of adverbs. McGill working papers in linguistics. McGill University, pp 280–310 Travis L (1992) Inner aspect and the structure of VP. Cahiers de linguistique de l’UQAM 1(1):130–146. Université de Quebec à Montréal Trenkic D (2004) Definiteness in Serbian/Croatian/Bosnian and some implications for the general structure of the nominal phrase. Lingua 114:1401–1427
REFERENCES
161
Trugman H (2004) More puzzles about postnominal genitives. In: Kim J-Y et al (eds) Possessives and beyond: Semantics and syntax. UMOP 29. University of Massachusetts occasional papers in linguistics. Amherst, MA, GLSA Publications Trugman H (2005) Syntax of Russian DPs, and DP-internal agreement phenomena. Dissertation, Tel-Aviv University Uriagereka J (1999) Multiple Spell-Out. In: Epstein SD, Hornstein N (eds) Working Minimalism. Cambridge, MA, The MIT Press, pp 251–282 Verkuyl HJ (1972) On the compositional nature of the aspects. Dordrecht, Reidel Verkuyl HJ (1993) A theory of aspectuality: The interaction between temporal and atemporal structure. Cambridge, UK, Cambridge University Press Weerman F (1996) Asymmetries between nominative, accusative and inherent Case. OTS Wierzbicka A (1980) The case for surface case. Ann Arbor, MI, Karoma Williams A (1993) The argument structure of SJA-Predicates. Journal of Slavic Linguistics 1(1):167–190 Williams E (1980) Predication. Linguistic Inquiry 11:203–237 Williams E (1981) Argument structure and morphology. Linguistic Review 1:81–114 Williams E (1983) Against Small Clauses. Linguistic Inquiry 14:287–308 Williams E (1989) The anaphoric nature of -roles. Linguistic Inquiry 20(3):425–456 Williams E (1994) Thematic structure in syntax. Cambridge, MA, The MIT Press Willim E (1998) On the DP-hypothesis in Polish, an articless language. In: Stalmaszczyk P (ed) Projections and mapping: Studies in syntax. Lublin, FOLIUM, pp 137–158 Willim E (2000) On the grammar of Polish nominals. In: Martin R et al (eds) Step by step: Essays on Minimalist syntax in honor of Howard Lasnik. Cambridge, MA, The MIT Press, pp 319–346 Woodcock EC (1959) A new Latin syntax. London, Methuen Yadroff M (1996) Modern Russian vocatives: A case of substractive morphology. Journal of Slavic Linguistics 4(1):133–153 Zamparelli R (2000) Layers in the Determiner Phrase. New York, Garland Publishing Zolotova GA (1988) Sintaksiˇceskij slovar’. Moscow, Nauka