Functional Structure in DP and IP: The Cartography of Syntactic Structures [Volume 1 ] 0195148800, 9780195148800

This volume presents the first results of a long-term research project, funded by the Italian government, which aims at

483 121 2MB

English Pages 248 [242] Year 2002

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Functional Structure in DP and IP: The Cartography of Syntactic Structures [Volume 1 ]
 0195148800, 9780195148800

Citation preview

Functional Structure in DP and IP: The Cartography of Syntactic Structures, Volume 1

GUGLIELMO CINQUE, Editor

OXFORD UNIVERSITY PRESS

Functional Structure in DP and IP

OXFORD STUDIES IN COMPARATIVE SYNTAX Richard Kayne, General Editor Principles and Parameters of Syntactic Saturation Gert Webelhuth Verb Movement and Expletive Subjects in the Germanic Languages Sten Vikner Parameters and Functional Heads: Essays in Comparative Syntax Edited by Adriana Belletti and Luigi Rizzi Discourse Configurational Languages Edited by Katalin É. Kiss Clause Structure and Language Change Edited by Adrian Battye and Ian Roberts Dialect Variation and Parameter Setting: A Study of Belfast English and Standard English Alison Henry Parameters of Slavic Morphosyntax Steven Franks Particles: On the Syntax of Verb-Particle, Triadic and Causative Constructions Marcel den Dikken The Polysynthesis Parameter Mark C. Baker The Role of Inflection in Scandinavian Syntax Anders Holmberg and Christer Platzack Clause Structure and Word Order in Hebrew and Arabic: An Essay in Comparative Semitic Syntax Ur Shlonsky Negation and Clausal Structure: A Comparative Study of Romance Languages Raffaella Zanuttini Tense and Aspect: From Semantics to Morphosyntax Alessandra Giorgi and Fabio Pianesi Coordination Janne Bondi Johannessen Adverbs and Functional Heads: A Cross-Linguistic Perspective Guglielmo Cinque The Feature Structure of Functional Categories: A Comparative Study of Arab Dialects Elabbas Benmamoun Comparative Syntax of Balkan Languages Edited by María Luisa Rivero and Angela Ralli Functional Structure in DP and IP: The Cartography of Syntactic Structures, Volume 1 Edited by Guglielmo Cinque

Functional Structure in DP and IP The Cartography of Syntactic Structures, Volume 1

EDITED BY GUGLIELMO CINQUE

1 2002

3

Oxford New York Auckland Bangkok Buenos Aires Cape Town Chennai Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi São Paulo Shanghai Singapore Taipei Tokyo Toronto

Copyright © 2002 by Oxford University Press, Inc. Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 www.oup.com Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Functional structure in DP and IP : the cartography of syntactic structures, vol. 1 / edited by Guglielmo Cinque. p. cm. ISBN 0-19-514879-7; ISBN 0-19-514880-0 (pbk.) 1. Grammar, Comparative and general—Syntax. 2. Functionalism (Linguistics) I. Cinque, Guglielmo. P291 .F814 415—dc21

9

8

7

6

2002 2001036663

5

4

3

2

1

Printed in the United States of America on acid-free paper

Contents

Contributors

vii

1 Mapping Functional Structure: A Project, Guglielmo Cinque

Part I

3

DP Functional Structure

2 The Positions of Demonstratives in the Extended Nominal Projection, Laura Brugè 15 3 The Functional Structure of Noun Phrases: A Bare Phrase Structure Approach, Giuliana Giusti 54 4 Stacked Adjectival Modification and the Structure of Nominal Phrases, Gary-John Scott 91

Part II

IP Functional Structure

5 Clause Structure and X-Second, Anna Cardinaletti and Ian Roberts

123

6 Agreement and Tense as Distinct Syntactic Positions: Evidence from Acquisition, Maria Teresa Guasti and Luigi Rizzi 167 7 The Distribution of Functional Projections in ASL: Evidence from Overt Expressions of Syntactic Features, Carol Neidle and Dawn MacLaughlin 195 Subject Index

225

Language Index Name Index

228 230

This page intentionally left blank

Contributors

Laura Brugè University of Venice [email protected]

Dawn MacLaughlin McGill University [email protected]

Anna Cardinaletti University of Bologna and University of Venice [email protected]

Carol Neidle Boston University [email protected] Luigi Rizzi University of Siena [email protected]

Guglielmo Cinque University of Venice [email protected]

Ian Roberts Cambridge University [email protected]

Giuliana Giusti University of Venice [email protected]

Gary-John Scott School of Oriental and African Studies, London [email protected]

Maria Teresa Guasti University of Milan [email protected]

vii

This page intentionally left blank

Functional Structure in DP and IP

This page intentionally left blank

1

Mapping Functional Structure: A Project  

This volume presents the first results of a long-term research project aiming at the discovery and mapping out of the functional structure of natural language sentences.1 The impetus of the study of functional structure, after Chomsky (1986, 3) first suggested extending the X-bar format to the nonlexical categories of Inflection and Complementizer, can be traced back to some fifteen years ago, with Abney (1987) for the DP, and Pollock (1989) for the sentence, with subsequent modifications by Belletti (1990), Ouhalla (1991), and Chomsky (1993), among others. Since then, the amount of work produced in the study of functional structure,on a variety of languages, has grown to such an extent that a systematic mapping of the various functional heads found in the sentence and other phrases of the languages of the world has become appropriate, if not imperative. The idea is that we are at a point when it is finally possible to set the question in a systematic and empirical fashion, by having the various types of evidence considered in the literature interact and converge onto distinct functional hierarchies for each phrase. What makes the enterprise all the more interesting is the mounting evidence of the last several years that the distinct hierarchies of functional projections may be universal in the inventory of the heads they involve, in their number, and in their relative order (despite certain appearances). This is, at any rate, the strongest position to take, as it is compatible with only one state of affairs. It is the most exposed to refutation, and, hence, more likely to be correct, if unrefuted. In front of recalcitrant facts we might be led to a weaker position—one that allows languages to vary either in the inventory, or the number, or the order of the functional heads that they admit (or any combination thereof). Even if this position should eventually turn out to be right, methodologically it would be wrong to start with it, 3

4

FUNCTIONAL STRUCTURE IN DP AND IP

discarding the stronger position. That would only make us less demanding with respect to the facts and could lead us to miss more subtle evidence supporting the stronger position (a risk not present under the other option). To see how the systematic mapping of such universal and invariant hierarchies might be carried out, let me briefly review the principal kinds of evidence considered in the literature and also utilized in the chapters contained in this volume. One first source of evidence for positing functional projections other than the D(eterminer)P, the I(nflection)P, and the C(omplementizer)P recognized in the early work mentioned above was provided by word order considerations. One such case is Pollock’s (1989) classical argument for positing a nonlexical head higher than V, head of VP, and lower than I (or T), to which only finite verbs raise in French. His suggestion was that such word order alternations as (1)a–b, involving infinitives, could be accounted for directly if we take the infinitive to raise out of VP to a head higher than complètement (in the 1b. example), and lower than negation (and T).2 (1) a. (ne pas) complètement perdre la tête ‘not completely lose one’s mind’ b. (ne pas) perdre complètement la tête

This kind of consideration provides evidence (pace Chomsky 2001) for the existence of a nonlexical head category distinct from T but offers little insight into the exact label or function of that category and projection (Mitchell 1994, 9–10).3 Despite this indeterminacy, such considerations offer important clues to the abstract skeleton of the underlying functional structure—one that may one day be fleshed out on the basis of independent morphological evidence. More recent similar arguments based on word order considerations are Kayne’s (1998, 1999) W head and Pollock, Munaro, and Poletto’s (1999) G head. A second strand of evidence for functional structure is provided by the inventory of functional (more traditionally, “grammatical”) head morphemes of the languages of the world. This evidence offers a direct insight into the label of the functional categories which are expressed grammatically across languages, as well as into their number and order. The most insightful and influential proposal concerning the cross-linguistic order of functional heads is surely Baker’s (1985, 1988) Mirror Principle, which recognized the existence of a systematic relation between the order of bound morphemes in morphology and the order in which syntactic operations affect the corresponding free morphemes in the bottom-up syntactic derivation of the sentence: syntactic operations that correspond to affixes closer to the verb stem precede syntactic operations that correspond to affixes further away from the verb stem.4 For example, just as passivization in Italian applies after syntactic causativization has turned the subject of the embedded verb andare ‘go’ into the object of the complex predicate far andare ‘make go’ formed by the causative and the embedded verb (see (2)a), so the passive suffix -rare in Japanese is added after the causative suffix -(s)ase (see (2)b): (2) a. Gianni fu fatto andare. G. be. causativeV- go ‘G. was made to go.’

MAPPING FUNCTIONAL STRUCTURE

5

b. Hanako ga ik-ase-rare-ta. (see Miyagawa 1989, 151) H.  --- ‘H. was made to go.’

While the rough mirror relationship between the order of suffixes and the order of the corresponding free morphemes in “head-initial” languages (particles and auxiliaries), was also noted by other linguists (Bybee 1985; Dik 1989),5 Baker (1985, 1988) went further in suggesting a way to obtain it on the basis of independent syntactic principles (such as the Empty Category Principle subsuming Travis’s 1984 Head Movement Constraint). The idea, further refined in subsequent work by Pollock (1989), Belletti (1990), Ouhalla (1991), and Chomsky (1991), was essentially that affixal morphology is built into the syntax, subject to the same constraints that govern syntactic operations, each suffix being treated as a bound morpheme in head position in need of a host. Such a host is provided by the lower head, which raises and left-adjoins to it. In both “head-final” and “head-initial” languages, this has the effect of reversing the order of heads into their mirror image:6 (3)

TP

-ta

VoiceP

-rare

CauseP

-(s)ase

VP

ik-

Apparent difficulties for a literal derivation of both derivational and inflectional morphology in the syntax are provided by a number of irregularities, however. For example: Syncretism—when one morpheme encodes more than one functional notion, as with the Italian verbal suffix -v-, which encodes both imperfective aspect and past tense. Suppletion—failure of regular compositionality, as when an entirely different stem is selected in the presence of a certain functional feature: see went in place of go + ed in English. Multiple Exponence—when a single functional notion is apparently encoded in two, or more, morphemes, as with the Modern Greek selection of a particular stem in the presence of a particular functional notion (say pli-, instead of plen- or plin-, in the presence of nonactive voice, which is also expressed by the suffix th-. See Joseph and Smirniotopoulos 1993.7

6

FUNCTIONAL STRUCTURE IN DP AND IP

For a discussion of these and other irregularities, see Di Sciullo and Williams (1987), Noyer (1991,1994), Janda and Kathman (1992), Janda (1994, and references cited there). All of these irregularities are untypical of the syntactic component and would thus seem to favor an autonomous morphological treatment of derivation and inflection. In part as a consequence of such difficulties, much current research adopts a lexical treatment of derivational and inflectional morphology, which after Chomsky (1993) has come to be known as the “Checking Theory.” Chomsky (1995, 195–196) compares two possible alternatives, explicitly opting for the second: (1) a bare stem, inserted from the lexicon, “picks up” the features of distinct inflectional heads and is subsequently spelled out in PF as a single phonological word; and (2) a lexical item, comprising inflectional features, is inserted from the lexicon with such features, checked against the corresponding features of the inflectional heads through which the item raises. As noted by Beninca’ (1999, 13), the second option “could be too weak, if not supplemented by an independent principle, which at least specifies, for example, in what order the matching of features has to occur”(see also Potter 1996). Perhaps, the essence of the Mirror Principle can be better maintained under Chomsky’s first option if the checking of hierarchically arranged features follows the cycle, whereby no feature of an outer cycle can be checked before the feature of an inner cycle has been checked, and the relative brackets erased: [ [ [ [a] infl1] infl2] . . . infln]

If the features are then spelled out in PF, in a one-to-one fashion with morphemes, or many-to-one, or one-to-many, or if, in this spell-out, they are “tied to” a particular stem form, the irregularities mentioned earlier cease to be problematic. For example, one and the same morpheme (as in the syncretic case above) may “realize” two distinct features, checked in two separate heads (as long as such features are not separated by another which is spelled out as a distinct morpheme).8 A third source of evidence for functional structure is given by the the nature and order of certain nonlexical specifiers—that is, specifiers other than those hosting argument DPs in the (layered) VP. Rizzi (1997), for example, has proposed a particular articulation of the CP “space” for the positions of topic and focus phrases found in the left periphery of the sentence.9 Cinque (1994, 1999) has argued that APs and AdvPs should be viewed as specifiers of the functional projections of the DP and the sentence, respectively, so that their nature and order should provide evidence for the nature and order of these projections independently of (and, it is to be expected, in convergence with) the nature and order of the corresponding head morphemes.10 A fourth type of evidence for functional structure (specifically, functional heads), which in principle should overlap with that coming from the nature and order of the bound and free morphemes referred to earlier, is based on the idea, explored in Cinque (2000), that “restructuring” verbs are functional verbs, directly merged into the corresponding functional heads (for some preliminary related discussion, see Cinque 2001a, 2002).

MAPPING FUNCTIONAL STRUCTURE

7

It is to be hoped that these different types of evidence for functional structure will be systematically investigated and made to interact to eventually attain the complete map (and sequencing) of the functional elements of natural language phrases and sentences. The chapters contained herein focus in particular on aspects of the fine structure of DPs and IPs. Much work remains to be done concerning the functional structure of APs, PPs, AdvPs, and other, minor, phrases (QPs, etc.).

The Contributions The functional structure of nominal phrases has been the focus of much research in the 1990s, and this research has proved to be quite rich and articulated. The contributions by Brugè (chapter 2), Giusti (chapter 3), and Scott (chapter 4) begin with such findings and go on to reach new and important insights in this domain. Brugè presents compelling evidence from a variety of languages for the conclusion that Demonstrative Phrases are “base generated” in the specifier of a functional projection lower than those hosting attributive Adjective Phrases and (immediately) higher than NP (or the projection hosting the subject PP and possessive adjective, if different from the highest NP layer). Even those languages that obligatorily raise the Demonstrative Phrase to higher specifier positions (such as Spec, DP) show traces of the original lower “base generation” position in the possibility of stranding in that position locative reinforcers, which ordinarily form a constituent with demonstratives. She also singles out a few (interacting) parameters thus deriving the quite substantial variation shown by languages in this domain. Scott’s chapter constitutes a fundamental step forward in the identification of the finer grained functional structure underlying the DP internal space of Attributive Adjective phrases in object nominals. He shows the existence of a quite substantial number of distinct classes of attributive adjectives, all rigidly ordered with respect to each other. He also carefully considers broader theoretical issues stemming from this finding and points to some of the factors that may obscure the rigid order of the APs. Giusti’s contribution proposes a very simple and elegant minimalist system of principles to derive the wide variation among languages in the expression of definiteness. In particular, she argues for the last resort nature of articles, categorially heads, and for the phrasal nature of the elements (among which are null operators, demonstratives, possessives, and remnant NPs) that carry the referential import of the DP, ultimately as occupants of Spec, DP. The three contributions together attain a remarkable coverage of the various “spaces” of the functional structure of DPs, from the lowest (Bruge’s), to the intermediate (Scott’s), to the highest (Giusti’s). The remaining three contributions all deal with the internal articulation of the IP space of the sentence. Cardinaletti and Roberts’s (chapter 5) article is an updated version of a paper originally presented at the 1990 GLOW conference in Cambridge. It is one of the earliest works that exploits the heuristic potential of abstract functional projections, achieving the unification of various X-second (especially, Verb-second and “Clitic-second”) effects via the postulation of an additional Agr head between C

8

FUNCTIONAL STRUCTURE IN DP AND IP

and AGRs. Having being widely circulated, and quoted, in its earlier unpublished form, it is hosted here (in revised form) as the first regular publication of their results. Guasti and Rizzi’s contribution (chapter 6) presents and discusses quite impressive evidence from early English (around the third year), pointing to the conclusion that the functional head where agreement is checked is below C, and higher than the position to where, in negative contexts, auxiliary do raises, crossing over negation (presumably T). Their work thus appears to give evidence for a head intermediate between T and C, where agreement is checked (whether or not this is an AGR head). The contribution by Neidle and MacLaughlin (chapter 7) on American Sign Language (ASL) also gives evidence for positing agreement projections distinct from Tense. It is also of particular interest for the general issue of functional structure in that it provides evidence for its universality in language across modalities (spoken or signed) and for its universal hierarchical organization, with Tense higher than Aspect (functional information in ASL is marked nonmanually, with face and upper body, and runs parallel to manual signing). Noteworthy (from Cinque’s 1999 viewpoint) is also their observation that ASL has multiple aspect projections and that aspectual adverbs bear morphological inflections (movement modulations) that are of the same general kind as those found on verbs marked by the corresponding aspects. Notes 1. The project, originally conceived by Luigi Rizzi and myself, brings together different groups of researchers at the Universities of Ferrara, Florence, Milan, Padua, Siena, and Venice. It has been funded since 1998 (to 2004) by the Italian Ministery of the University and Scientific Research. Most of the articles included here were presented at the first meeting of the groups held in Venice in January 29 and 30, 1999. 2. Iatridou’s (1990) arguments against positing such extra heads are not cogent for this particular case of lexical infinitives. See Pollock (1997, 3.1) for discussion. 3. Pollock (1989) conjectured that such a category could be taken to be an abstract AGR head because of the lack of lexical verb raising to such a head in English and because English differs from French in being poorer in verbal agreement. But his argument does not depend on the exact label of the head. 4. A similar generalization, called there the Satellite Principle, appears sketched in footnote 4 of chapter 4 of Gerdts (1982, 193), within a relational grammar framework. 5. See also Foley and Van Valin (1984), who, however, refer to a previous version of Bybee (1985). 6. This works for suffixes. While Baker’s original discussion of the Mirror Principle applied to both prefixes and suffixes, much subsequent work (starting with Pollock 1989 and Belletti 1990) adopted a narrower interpretation, restricting attention to suffixes, which in “head-initial” languages systematically appear in an order that is the mirror image of the order of the corresponding free morphemes. This limitation is perhaps not accidental, nor is it undesirable, as prefixes appear to be a nonhomogeneous class (Cinque 1999, 68–70; see also Brody 2000, 34). Cinque’s (1999) discussion should be integrated for the case of those Bantu languages where prefixes show the same order of the corresponding functional heads but must be adjacent to each other and to the verb (i.e., cannot cliticize onto anything except other functional morphemes or the verb). The fact that nothing can intervene between them can be made sense of if we assume attraction of the (remnant) VP to the Spec of the prefix, followed

MAPPING FUNCTIONAL STRUCTURE

9

by head movement of the prefix itself to a Kaynean W head). This further requires that the AdvP corresponding to the functional head in question be in the Spec of a “small extended projection” of the functional head. 7. Possibly of the same general type is the apparent double expression of tense in the Australian language Wambaya (Nordlinger 1995). 8. A more careful investigation along these lines may indeed show that no weakening of the Mirror Principle of the sorts proposed in Noyer (1991), Speas (1991), Potter (1996), Alsina (1999), and others is really necessary. For a recent attempt at deriving the generalization expressed by the Mirror Principle, see Brody (2000). 9. Also see Beninca’ (2001), Beninca’ and Poletto (1999) for an interesting proposal about the order of the focus phrase with respect to different (and differently specialized) topic positions. Their finding, together with Cinque’s (2001b) that apparent multiple positionings of the same functional morpheme hide differently specialized functional heads, may suggest ruling out simple recursion from UG completely. 10. See also Scott (chapter 4, this volume).

References Abney, Steven (1987) “The English Noun Phrase in Its Sentential Aspect,” Ph.D. Diss., MIT. Alsina, Alex (1999) “Where’s the Mirror Principle.” Linguistic Review 16, 1–42 Baker, Mark (1985) “The Mirror Principle and Morphosyntactic Explanation.” Linguistic Inquiry 16, 373–416. Baker, Mark (1988) Incorporation: A Theory of Grammatical Function Changing. Chicago: University of Chicago Press. Belletti, Adriana (1990) Generalized Verb Movement. Turin: Rosenberg and Sellier. Beninca’, Paola (1999) “Between Morphology and Syntax: On the Verbal Morphology of Some Alpine Dialects,” in Lunella Mereu (ed.) Boundaries of Morphology and Syntax, 11–30. Amsterdam: John Benjamins. Beninca’, Paola (2001) “The Position of Topic and Focus in the Left Periphery,” in Guglielmo Cinque and Giampaolo Salvi (eds.) Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi, 39–64. Amsterdam: Elsevier. Beninca’, Paola, and Cecilia Poletto (1999) “Topic, Focus and V2: Defining the CP Sublayers.” Paper presented at the workshop “On the Cartography of Syntactic Positions and Semantic Types,” held at Certosa di Pontignano (Siena), November 25–26, 1999. Brody, Michael (2000) “Mirror Theory: Syntactic Representation in Perfect Syntax.” Linguistic Inquiry 31, 29–56. Bybee, Joan (1985) Morphology: A Study of the Relation between Meaning and Form. Amsterdam: John Benjamins. Chomsky, Noam (1986) Barriers. Cambridge, Mass.: MIT Press. Chomsky, Noam (1991) “Some Notes on Economy of Derivation and Representation,” in Robert Freidin (ed.) Principles and Parameters in Comparative Grammar, 417–454. Cambridge, Mass.: MIT Press. Reprinted in Chomsky 1995. Chomsky, Noam (1993) “A Minimalist Program for Linguistic Theory,” in Kenneth Hale and Samuel Jay Keyser (eds.) The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, 1–52. Cambridge, Mass.: MIT Press. Reprinted in Chomsky 1995. Chomsky, Noam (1995) The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, Noam (2001) “Derivation by Phase,” in Michael Kenstowicz (ed.) Ken Hale: A Life in Language, 1–52. Cambridge, Mass.: MIT Press. Cinque, Guglielmo (1994) “Evidence for Partial N-movement in the Romance DP,” in Guglielmo Cinque, Jan Koster, Jean-Yves Pollock, Luigi Rizzi, and Raffaella Zanuttini

10

FUNCTIONAL STRUCTURE IN DP AND IP

(eds.) Paths towards Universal Grammar: Studies in Honor of Richard S.Kayne, 85– 110. Washington, D.C.: Georgetown University Press. Cinque, Guglielmo (1999) Adverbs and Functional Heads: A Cross-linguistic Perspective. New York: Oxford University Press. Cinque, Guglielmo (2000) “Restructuring and Functional Structure.” Unpublished ms., University of Venice (to appear in Adriana Belletti (ed.) Structure and Beyond, New York: Oxford University Press). Cinque, Guglielmo (2001a) “ ‘Restructuring’ and the Order of Aspectual and Root Modal Heads,” in Guglielmo Cinque and Giampaolo Salvi (eds.) Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi, 137–155. Amsterdam: Elsevier. Cinque, Guglielmo (2001b) “The Status of ‘Mobile’ Suffixes,” in Walter Bisang (ed.) Aspects of Typology and Universals, 13–19. Berlin: Akademie Verlag. Cinque, Guglielmo (2002) “The Interaction of Passive, Causative, and ‘Restructuring’ in Romance,” in Christina Tortora (ed.) The Syntax of Italian Dialects, 50–66. New York: Oxford University Press. Dik, Simon (1989) Functional Grammar: Part I. The Structure of the Clause. Dordrecht: Foris. Di Sciullo, Anna Maria, and Edwin Williams (1987) On the Definition of Word. Cambridge, Mass.: MIT Press. Foley, William, and Robert Van Valin (1984) Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. Gerdts, Donna (1982) “Object and Absolutive in Halkomelem Salish,” Ph.D. Diss., University of California at San Diego. Iatridou, Sabine (1990) “About Agr(P).” Linguistic Inquiry 21, 551–577. Janda, Richard (1994) “Checking Theory, Syntactic Feature-Geometry, and the Structure of IP,” in Proceedings of the 12th West Coast Conference on Formal Linguistics, 319–337. Stanford: Stanford Linguistics Association. Janda, Richard, and David Kathman (1992) “Shielding Morphology from Exploded INFL,” in Papers from the 28th Regional Meeting of the Chicago Linguistic Society, 141–157. Chicago: Chicago Linguistic Society. Joseph, Brian, and Jane Smirniotopoulos (1993) “The Morphosyntax of the Modern Greek Verb as Morphology and Not Syntax.” Linguistic Inquiry 24, 388–398. Kayne, Richard (1998) “Overt vs. Covert Movement.” Syntax 1(2), 128–191. Kayne, Richard (1999) “Prepositional Complementizers as Attractors.” Probus 11, 39–73. Mitchell, Erika (1994) “Morphological Evidence for Syntactic Structure: The Finno-Ugric Languages and English,” Ph.D. Diss., Cornell University, Ithaca, N.Y. Miyagawa, Shigeru (1989) Structure and Case Marking in Japanese (Syntax and Semantics no. 22). New York: Academic Press. Nordlinger, Rachel (1995) “Split Tense and Mood Inflection in Wambaya,” in Proceedings of the 21st Annual Meeting of the Berkeley Linguistics Society, 226–236. Berkeley, Calif.: Berkeley Linguistics Society. Noyer, Rolf (1991) “The Mirror Principle Revisited: Verbal Morphology in Maung and Nunggubuyu,” MIT Working Papers in Linguistics 14, 195–209. Noyer, Rolf (1994) “Mobile Affixes in Huave: Optimality and Morphological Wellformedness,” in Proceedings of the 12th West Coast Conference on Formal Linguistics, 67–82. Stanford, Calif.: Stanford Linguistics Association. Ouhalla, Jamal (1991) Functional Categories and Parametric Variation. London: Routledge. Pollock, Jean-Yves (1989) “Verb Movement, Universal Grammar, and the Structure of IP.” Linguistic Inquiry 20, 365–424. Pollock, Jean-Yves (1997) “Notes on Clause Structure,” in Liliane Haegeman (ed.) Elements of Grammar, 237–279. Amsterdam: Kluwer.

MAPPING FUNCTIONAL STRUCTURE

11

Pollock, Jean-Yves, Nicola Munaro, and Cecilia Poletto (1999) “Eppur si muove! On Comparing French, Portuguese and Bellunese Wh-Movement.” Paper presented at the 25th Incontro di Grammatica Generativa, Università di Siena, February 25–27, 1999. Potter, Brian (1996) “Minimalism and the Mirror Principle.” Proceedings of the 26th North East Linguistic Society, 289–302. Amherst, Mass.: University of Massachusetts at Amherst, Department of Linguistics. Rizzi, Luigi (1997) “The Fine Structure of the Left Periphery,” in Liliane Haegeman (ed.) Elements of Grammar, 281–337. Amsterdam: Kluwer. Speas, Margaret (1991) “Functional Heads and the Mirror Principle,” Lingua 84, 181–214. Travis, Lisa (1984) “Parameters and Effects of Word Order Variation,” Ph.D. Diss., MIT.

This page intentionally left blank

P ART I

DP FUNCTIONAL STRUCTURE

This page intentionally left blank

2

The Positions of Demonstratives in the Extended Nominal Projection  

In Spanish, the demonstrative modifier can appear either in prenominal position, as in (1a–b), or in postnominal position, as in (1c–d): (1)

a. este/ese/aquel libro ‘this/that/that book’ b. estos/esos/aquellos libros ‘these/those/those books’ c. el libro este/ese/aquel1 the book this/that/that ‘this/that book’ d. los libros estos/esos/aquellos the books these/those/those ‘these/those books’

In this chapter I suggest a unified analysis for the two different word orders in (1). In section 1 I propose that in Spanish the demonstrative is always generated in a low position inside the extended nominal projection and that at PF it can be realized either in this position, (1c–d), or in the prenominal position, (1a–b). This second option is due to the movement of the demonstrative from its base position to [Spec, DP] before Spell-Out. Observing the position of the postnominal demonstrative with respect to the other elements internal to the extended nominal projection, I show, in sections 2, 3, and 4, that the demonstrative is generated in the specifier of a functional projection lower than all the other functional projections containing the different classes of adjectives 15

16

DP FUNCTIONAL STRUCTURE

and immediately superior either to the functional projection whose specifier is occupied by the postnominal possessive, if any, or to the NP projection. In section 5 I show that the demonstrative, even when it appears at PF in the low position, behaves, from a syntactic and semantic point of view, in the same way as when it appears in [Spec, DP]: the demonstrative in prenominal or in postnominal position provides a referential interpretation. On the basis of these facts, I propose that, if the movement of the demonstrative from its base position to [Spec, DP] is optional before Spell-Out, it is in any case obligatory at Logical Form. Moreover, I propose that this movement is motivated by feature checking: the demonstrative, which is specified for the Referential and Deictic features has to raise to [Spec, DP] in order to check the Referential feature which is in D° by Spec-Head Agreement. Finally, in section 6, I try to extend this analysis to other languages. I show that also in those languages in which the demonstrative always appears in [Spec, DP] at PF this element is generated in the same position we find in Spanish. I propose, then, that the cross-linguistic variation concerning the obligatoriness, the optionality, or the impossibility for the demonstrative to raise to [Spec, DP] before Spell-Out can be accounted for by suggesting that the Referential feature on the demonstrative has to be checked already at PF in the first case, either at PF or at Logical Form in the second case, and has to be delayed until Logical Form in the third case.

1. Theoretical assumptions 1.1. Demonstratives as specifiers With the aim of providing a unified analysis for the cases presented in (1), I base my proposal on the assumption that in Spanish the prenominal position (1a–b) and the postnominal position (1c–d) in which the demonstrative can appear are deeply related to each other. In section 5 I show empirical data in favor of this assumption. In this way, I discard the possibility that in Spanish, and in other languages in which we have the same phenomenon (cf. section 6.1.1.), there exist two (or more) different structural positions available for the demonstrative that are totally unrelated to each other. As for the categorial status of the demonstrative, I assume that this element is a maximal projection. I also assume that the demonstrative occupies the specifier position of a functional projection that belongs to the extended nominal projection. For this assumption I adopt Giusti’s (1993) proposal. Observing the syntactic behavior of the demonstrative in Romanian, Giusti claims that in this language the demonstrative is generated in the specifier of an AgrP that is immediately dominated by the DP projection and that it can be realized either in this position (2a) or, by raising to [Spec, DP], in the prenominal position (2b): (2)

a. b£iatul acesta (frumos) boy-the this (nice) b. acest (frumos) b£iat (frumos) this (nice) boy (nice) c. frumosul (*acesta) b£iat nice-the (*this) boy ‘this nice boy’

POSITIONS OF DEMONSTRATIVES

17

In (2a) the noun moves over the demonstrative to D° where the enclitic article appears. The fact that the demonstrative allows N-to-D movement shows that it can be conceived neither as a D° nor as an intermediate functional head.2 Therefore, the only possible solution is to assume that the demonstrative is in a specifier position. In this way, it could be explained why its presence blocks the movement of an adjective to [Spec, DP], as in (2c): a Minimality violation would prevent Spec-to-Spec movement of the adjective, which is otherwise possible in Romanian. 1.2. An antisymmetric approach In this analysis I adopt the antisymmetric approach proposed by Kayne (1994). Kayne’s antisymmetric hypothesis imposes, through the notion of asymmetric c-command, a rigid Specifier > Head > Complement order across languages. According to this order, a head that appears in the structure to the left of its specifier must have raised to a head position, asymmetrically c-commanding its trace and the specifier. Moreover, for the structure of nominals, I adopt the analysis proposed by Cinque (1994). According to this analysis, which is compatible with an antisymmetric approach, only the head noun moves to higher functional heads, while modifiers stay in their base positions unless they need to check some feature in a higher specifier.3 In this way, the order Noun > Demonstrative we observe in cases such as (1c–d) is obtained by the movement of the noun to a higher head position, as the structure in (3) shows:4 (3)

DP D' D˚ el

...... XP X' X˚ libroj

..... FP F'

Spec este F˚ tj

NP

tj

18

DP FUNCTIONAL STRUCTURE

We can assume that the low [Spec, FP] position the demonstrative occupies in (3) must be taken to be the base position. In those cases in which we have the order Demonstrative > Noun, (1a–b), I propose that this is obtained by the movement of the demonstrative to [Spec, DP] in order to check some feature that is present in this landing site, as (4) shows. This movement is triggered by feature checking, and the nature of this feature will be expressed in section 5. (4) [SpecDP estei [D° [. . . [XP [X° libroj] [FP ti [F° tj] [NP [N° tj] ] ] ] ] ] ]

1.3. Presence of the definite article In Spanish, as in many other languages, the prenominal demonstrative and the article are in complementary distribution: compare (1a–b) and the ungrammaticality of (5a). In contrast, when the demonstrative occurs postnominally, the DP projection is occupied by the article, (1c–d). In Spanish, as in Romanian, (2a), the presence of the article in D° is obligatory when the demonstrative follows the noun: compare (1c–d) and the ungrammaticality of (5b). Furthermore, the only possible form of the article is the definite one, as (1c–d) and the ungrammaticality of (5c) show: (5)

a. *este el libro this the book b. *libro este book this c. *un libro este a book this

The data in (5), compared with those in (1), raise the following two questions: 1. Why is it that the presence of the article is obligatory when the demonstrative appears in postnominal position, (5b), while its absence is necessary when the demonstrative appears in prenominal position, (5a)? 2. Why is it that the only accepted form of the article is the definite one when the demonstrative is postnominal, (5c)? As far as the first question is concerned, we assume that (5b) is ill formed for the same reasons for which sentences such as: *libro nuevo está en el despacho (lit.: ‘book new is in the office’) or *he comprado libro nuevo (lit.: ‘I-have bought book new’) are excluded. Sentences of this type are ungrammatical because the DP projection is not filled by some functional element after Spell-Out. In fact, Spanish requires that the DP is in general lexically overt. Moreover, in cases like these, D° can neither be interpreted as an empty category receiving existential interpretation (see Longobardi, 1994), due to the intrinsic properties of the head noun libro ‘book’ and to the presence of the demonstrative, as we will see in section 5. So I propose that the presence of the article in constructions such as (1c–d) is necessary to make the DP projection visible after Spell-Out because it cannot be interpreted as an empty category. A principled way to justify the well-formed cases in (1) and the ungrammatical constructions in (5a–b) has been proposed by Dimitrova-Vulchanova and Giusti

POSITIONS OF DEMONSTRATIVES

19

(1998). They assume that, in general, a functional projection is instantiated in order to realize some feature ϕ, and that this feature has to be “visible” in order to be interpreted at Logical Form. They also assume that the visibility condition imposed on functional features is the Spec-Head relation. Then, they propose the following two general principles: (6) “Economize Functional Heads.” ((42), p. 345) (7) “A Functional Projection must be visible at all levels of representation: by (a) making the Spec visible and/or (b) making the head visible.” ((44), p. 346)

Principle (6) establishes that when the specifier is lexically filled in order to realize some relevant feature, the functional head does not need (and by minimalist requirements cannot) be occupied by any lexical element. Principle (6) can account for cases such as (1a–b) and other cases in which, still inside the extended nominal projection, most functional heads are not lexically realized. Its violation is instead responsible for the ungrammaticality of cases such as (5a). Alternatively, the obligatory presence of the article in cases such as (1c–d), even if in contrast with principle (6), is justified by principle (7). On the basis of what we have anticipated, in (1c–d) the article has to be inserted as a “last resort.” In fact, only in this way can the relevant features in DP the demonstrative must check by raising to [Spec, DP] at Logical Form be rendered “visible,” either at PF or, crucially, at LF. This requirement also accounts for the ungrammaticality of (5b). Let us now consider the case of question (b). A tentative proposal to account for the obligatory presence of the definite form of the article in D°, (1c–d), is to assume that only the definite feature is compatible with the feature the demonstrative is specified for. We will call this feature Referential feature, as I will justify in section 5. In this way, if the referential feature has to be selected inside the DP projection in order to enable demonstrative movement at Logical Form, the insertion of an indefinite article in D° would show that this requirement has not been satisfied; and this would give rise to a clash of features between the head and its specifier. The obligatory movement of the demonstrative to [Spec, DP] at LF would be blocked and the construction ruled out (see (5c)). By contrast, if the definite form of the article is inserted in D° for the reasons just mentioned, there would be no clash of features between the head and its specifier. As it is assumed, in fact, referentiality implies definiteness.

2. Position of the postnominal demonstrative We have seen that also in Romanian, (2a), the demonstrative can be realized postnominally. Giusti (1993) proposes that in this case the demonstrative occupies the specifier of a functional projection (AgrP) immediately dominated by the DP. This assumption is justified by the fact that in Romanian no category (e.g., adjectives) can intervene between the noun raised to D° and the demonstrative itself (cf. *b£iatul frumos acesta [lit.: ‘boy-the nice this’]). In Spanish, contrary to what happens in Romanian, the postnominal demonstrative always follows the adjectives at PF:

20 (8)

DP FUNCTIONAL STRUCTURE

a. La reacción desinteresada esta/esa en realidad era preocupante. the reaction disinterested this/that in fact was worrying b. *La reacción esta/esa desinteresada . . . the reaction this/that disinterested ‘This disinterested reaction was in fact worrying.’ c. El chico alto este/ese vive cerca de casa. the boy tall this/that lives near home d. *El chico este/ese alto . . . the boy this/that tall ‘This/that tall boy lives near home.’

Examples (8a) and (8c) exemplify the unmarked word order. In cases like these a predicative analysis for the demonstrative alone is excluded because no intonational break between the adjective and the demonstrative itself is produced and because the demonstrative does not have a special stress. However, the ungrammatical sentences (8b) and (8d) become possible if there is an intonational break between the demonstrative and the adjective. Following Cinque (1994), we assume that in cases like these the structural representation should be different, given that the AP establishes a modification of a predicative type, namely a modification with respect to the entire DP projection.5 2.1. Postnominal demonstrative and adjectives Cinque (1994) proposes that there exists a specific unmarked serialization of the different classes of APs across languages. Each of these different classes is universally generated in the specifier position of functional projections, which are located between DP and NP. In Romance, the presence of the adjectives to the right of the noun is to be attributed to the raising of the noun to an intermediate functional head between N and D. The absence of N movement in Germanic is due to a parametric variation: Romance, contrary to Germanic, has strong features in the intermediate functional heads that need to be checked before Spell-Out (cf. Chomsky 1995). According to Cinque (1994), and given the data in (8), I propose that the position in which the demonstrative is generated is a very low position inside the extended projection of the noun (see Grimshaw 1991). Therefore, we can formulate the following generalization: (9) In Spanish the demonstrative is generated in the specifier position of a functional projection intermediate between the DP and the NP and lower than all the functional projections containing APs.

Let us observe, now, the behavior of the postnominal demonstrative with respect to the different classes of adjectives that can modify an “event” noun, according to Cinque’s terminology. The general internal structure that Cinque proposes for these nominals, and which we extend to Spanish, is presented in (10): (10) [DPD [XPposs.AP X [YPcard.AP Y [WPord.AP W [ZPspeaker-orientedAP Z [HPsubject-orientedAP H [NPManner/ThematicAP N ] ] ] ] ] ] ]6

Cinque shows that in cases like these the head N can move at most as far as the head H of the functional projection HP.

POSITIONS OF DEMONSTRATIVES

21

In Spanish, when the postnominal demonstrative co-occurs with a manner adjective, this adjective always precedes the postnominal demonstrative in the unmarked order and can never follow it, as the contrast in (11) shows: (11)

a. La reacción desinteresada esta/esa en realidad era preocupante. the reaction disinterested this/that in fact was worrying b. *La reacción esta/esa desinteresada . . . the reaction this/that disinterested ‘This disinterested reaction in fact was worrying.’

If in the same environment the postnominal demonstrative and thematic adjectives co-occur, we can observe the same results: thematic adjectives always precede the postnominal demonstrative: (12)

a. La reacción alemana esta/esa no nos convence. the reaction German this/that not us convinces b. *La reacción esta/esa alemana . . . the reaction this/that German ‘This German reaction does not convince us.’

Given the ungrammaticality of the constructions in (11b) and (12b), we can conclude that generalization (9) is correct. In addition, it is also possible to verify the correctness of this generalization by observing those nominal environments in which the head N is occupied by “objectdenoting” nouns. We extend to Spanish the general serialization of the different classes of adjectives that Cinque (1994) proposes for this class of nominals (for refinements, see Scott, chapter 4 in this volume): (13) [DPD [XPposs.AP X [YPcard.AP Y [WPord.AP W [ZPqualityAP Z [HPsizeAP H [LPshapeAP L [MPcolorAP M [OPnationalityAP O [NPN ] ] ] ] ] ] ] ] ] ]

Cinque shows that in these cases the head N can cyclically move at the most as far as the Z position. In Spanish, we can observe, again, that the postnominal demonstrative always follows, in the unmarked order, each of the adjectival classes which in (13) can appear postnominally at PF, as shown in (14) and (15): (14)

a. Las ciudades españolas estas/esas no son muy pintorescas. (nationality the cities Spanish these/those not are very picturesque b. *Las ciudades estas/esas españolas . . . the cities these/those Spanish ‘These Spanish cities are not very picturesque.’

(15)

a. El chico alto este/ese vive cerca de casa. (size ADJ.) the boy tall this/that lives near my home b. *El chico este/ese alto . . . the boy this/that tall ‘This tall boy lives near my home.’

ADJ.)

The ungrammaticality of the constructions in (14b) and (15b) shows that also in those cases in which the head noun belongs to the class of object-denoting nominals, the postnominal demonstrative occupies a structural position lower than those occupied

22

DP FUNCTIONAL STRUCTURE

by the APs, which can be crossed over by the noun. Therefore we conclude that generalization (9) is correct. All the data presented in this section can be considered a robust argument in favor of Giusti’s (1993) hypothesis that the demonstrative has an XP status and, for this reason, occupies the specifier of a functional projection. In fact, because in Spanish the postnominal demonstrative always appears after APs in the unmarked word order, we would expect that the head noun would not be able to move over the postnominal demonstrative if this element occupied a head position. Because the N movement is a cyclic head-to-head movement, the presence of the demonstrative in a low functional head would block its raising, and, as a consequence, all the APs would appear prenominally, which is contrary to fact. A second, less strong argument is that if we assume that the demonstrative occupies a specifier position, we expect that, also in Spanish, wh-movement of a genitive out of the nominal projection is impossible. In fact, if the demonstrative occupies the [Spec, DP] position, the trace left by the wh-movement of the genitive would not be licensed inside the nominal projection, as Giusti (1993) for Italian and Campbell (1993a, 1993b) for English have pointed out. This hypothesis is confirmed by the ungrammaticality of the following sentences:7 (16)

a. *De quién / de qué pintor han publicado esta/esa foto? of whom / of which painter have-they published this/that picture b. *De quién / de qué pintor han publicado la foto esta/esa? of whom / of which painter have-they published the picture this/that ‘Who/which painter have they published this/that picture of?’

To rule out cases such as (16b) we could propose that wh-movement of the genitive PP is blocked either before Spell-Out or at Logical Form. If we choose the first possibility, we have to assume that this movement is blocked at structural levels lower than [Spec, DP] (see Rizzi 1990): the genitive could not cross over the functional projection containing the demonstrative because its specifier position is filled by the demonstrative itself, giving rise to a violation of the Minimality Condition. If we choose the second possibility, we have to assume that also in those languages in which the demonstrative can or must appear in the base position the demonstrative must move to [Spec, DP] at Logical Form in order to check some particular feature(s) which is(are) present in DP. So, in (16b) the obligatory movement of the demonstrative to [Spec, DP] at Logical Form would be blocked by the presence, in this position, of the trace of the wh-genitive. In section 5 I will present independent data to support this second explanation. 2.2. Postnominal demonstrative and PPs of the noun Let us now see which position the postnominal demonstrative occupies in relation to the subject and the other complements of the noun. With event nouns, we observe that the internal complements always follow the demonstrative when this element appears postnominally, (17a). The realization of

POSITIONS OF DEMONSTRATIVES

23

the demonstrative after the complements of the noun always gives rise to ungrammatical structures, (17b): (17)

a. La reacción (alemana) esta/esa a las críticas preocupó a todos. the reaction (German) this/that to the criticisms worried everybody b. *La reacción (alemana) a las críticas esta/esa . . . the reaction (German) to the criticisms this/that ‘This (German) reaction to the criticisms worried everybody.’

We find the same word order in cases in which the genitive PP saturates the Agent q-role, namely, it is the subject of the NP: (18)

a. La reacción (imprevisible) esta/esa de Alemania . . . the reaction (unexpected) this/that of Germany b. *La reacción (imprevisible) de Alemania esta/esa . . . the reaction (unexpected) of Germany this/that ‘This/that (unexpected) reaction by Germany . . .’

With object-denoting nouns, we encounter, again, the same results, as (19) shows: (19)

a. La falda (corta) esta/esa de María está toda arrugada. the skirt (short) this/that of Maria is all wrinkled b. *La falda (corta) de María esta/esa . . . the skirt (short) of Maria this/that ‘This/that (short) skirt of Maria’s is all wrinkled.’ c. El cuadro este/ese de Juan de Picasso no es nada más que una copia mal hecha. the painting this/that of Juan(POSS.) of Picasso(Agent) is nothing more than a copy badly made d. *El cuadro de Juan este/ese de Picasso . . . the painting of Juan(POSS.) this/that of Picasso(Agent) e. *El cuadro de Juan de Picasso este/ese . . . the painting of Juan(POSS.) of Picasso(Agent) this/that ‘This/that painting by Picasso of Juan’s is nothing more than a badly made copy.’

According to the resulting structural order Demonstrative > PP presented so far, I propose for Spanish the following generalization: (20) In Spanish the demonstrative is generated in the specifier position of a nominal functional projection which immediately dominates the NP projection.8

2.3. Slightly modifying Cinque’s structure for nominals In the internal structure that Cinque (1994) proposes for event nominals, thematic and manner adjectives compete for the same position: the specifier of NP (see example (10)). In section 2.1, I showed that manner adjectives and thematic adjectives always precede the postnominal demonstrative, (11) and (12); while in section 2.2, I showed that the genitive PPs subject of the noun follow the postnominal demonstrative, (18), like all other complements of the noun. The different position the thematic adjectives and the genitive PPs occupy in relation to the postnominal demonstrative is quite surprising if we adopt the hypothesis that genitive PPs subject of the noun are also in [Spec, NP] (see Cinque 1995).

24

DP FUNCTIONAL STRUCTURE

A possible solution is to propose that thematic and manner adjectives do not actually occupy the [Spec, NP], contrary to (10), but, rather, a higher position. Nevertheless, if we want to maintain Cinque’s idea, we should have to assume that, in a Larsonian structure with more specifier positions inside the NP-shell, the demonstrative occupies the specifier position lower than the one occupied by thematic and manner adjectives. But this hypothesis would lead to two main theoretical problems that are difficult to treat. First and more significant, we would have to assume that the demonstrative is generated in a position internal to the NP projection. This assumption would be difficult to justify because the demonstrative does not establish any thematic relationship with the noun. Second, we would have to assume that the agentive q-role is assigned to two different positions inside the NP-shell: to a position higher than the demonstrative, when it is saturated by a thematic adjective, and to a position lower than the demonstrative, when it is saturated by a genitive PP, (18). Taking into account these theoretical problems, I propose that thematic and manner adjectives occupy a position external to the NP projection.9 At first glance, this proposal does not seem to cause theoretical consequences; rather, it seems to have the advantage of explaining in a natural way the fact that these adjectives agree in ϕ-features with the head noun, as do all other adjectives. Moreover, the fact that thematic adjectives absorb a q-role in a position external to the NP does not seem to posit any problem. In fact, also the prenominal possessive adjective, which is assumed to occupy a position inside a functional projection higher than NP (see Cinque 1994; Longobardi 1995), saturates a q-role assigned by the head noun.10 An independent argument to support the hypothesis I am defending is given by the following case: (21)

a. La reacción inmediata de Juan preocupó a todos. the reaction immediate(mannerADJ) of Juan(Agent ‘Juan’s immediate reaction worried everybody.’

PP)

worried everybody

The grammaticality of this sentence would be surprising if we assume that thematic and manner adjectives, which compete for the same position, occupy the [Spec, NP] position after Spell-Out. In fact, (21a) should be excluded because the manner adjective inmediata and the agentive PP de Juan would occupy the same position— that is, [Spec, NP]. Therefore, we conclude, slightly modifing Cinque’s proposal, that thematic and manner adjectives occupy a position external to the NP projection. We also propose that this position is the specifier of a functional projection very low in the structure, which dominates the functional projection containing the demonstrative.

3. The structure 3.1. Base position of the demonstrative The structures we propose for nominals are the following: (22) [DPD [XP . . . HPsubject-orientedAP H [LPManner/ThematicAP L [FPDemonstrativeP F [NPAgent/Exper.PP [N’ N ThemePP ] ] ] ] ] ]

POSITIONS OF DEMONSTRATIVES

25

(23) [DPD [XP . . . ZPqualityAP Z [HPsizeAP H [LPshapeAP L [MPcolorAP M [OPnationalityAP O [FPDemonstrativeP F [NPPossessorPP N [NPAgentPP [N’N Compl.PP ] ] ] ] ] ] ] ] ] ]

Example (22) represents the partial structure which corresponds to event nominals. Example (23), by contrast, represents the partial structure that corresponds to object-denoting nominals. Recall that we also assume that all genitive PPs occupy an internal NP position. For this reason, I propose for object-denoting nominals a Larsonian structure for the internal NP projection (see note 8).11 Finally, extending to Spanish Cinque’s proposal that the noun moves in Syntax to higher positions, we obtain for the two classes of nominals the following possible orders after Spell-Out, already presented in previous sections: (24)

a. la respuesta inmediata esta de Juan . . . [DPla [XP . . . [HP [H respuestai] [LP [APinmediata] [Lti] [FP [DemP esta] [Fti] [NP [PPde Juan] [N’ [Nti] ] ] ] ] ] ] ] the answer immediate this of Juan ‘this immediate answer by Juan . . .’ b. la conquista italiana esa de Libia . . . [DPla [XP . . . [HP [Hconquistai] [LP [APitalianaj] [Lti] [FP [DemPesa] [Fti] [NP [tj] [N’ [Nti] [PPde Libia] ] ] ] ] ] ] ] the conquest Italian that of Libya ‘that Italian conquest of Libya . . .’ c. el cuadro inglés ese de María . . . [DPel [XP . . . [MP [M cuadroi] [OP [APinglés] [Oti] [FP [DemPese] [Fti] [NP [PPde María] [N’ [Nti] ] ] ] ] ] ] ] the painting English that of María(Possessor) ‘that English painting of Maria’s . . .’

3.2. Internal structure of the demonstrative In Spanish, in constructions in which the demonstrative este ‘this’ appears, the locative form aquí ‘here’ can be simultaneously realized. The locative form has to be obligatorily preceded by the preposition de ‘of’, and the resulting sequence is este de aquí ‘this of here’.12 I propose that in these cases the locative has only the function of reinforcing the deictic value of the demonstrative for which we assume it is specified (see section 5). In fact, the interpretation of the DP does not change if the locative is present or absent: the information expressed by este libro ‘this book’ and by este libro de aquí ‘this book of here’ are the same. The sequence este de aquí can be realized either in postnominal position, as in (25a), or the demonstrative can be realized in [Spec, DP] at FP and the locative appear postnominally, as in (25b): (25)

a. El libro este de aquí está mal hecho. the book this of here is badly made b. Este libro de aquí está mal hecho.13 this book of here is badly made ‘This book here is badly made.’

26

DP FUNCTIONAL STRUCTURE

As for cases such as (25b), I propose that the position the locative occupies in the structure is the base position, namely the [Spec, FP] position ((22) and (23)). In fact, contrasting (26a) with (26b–c), we can observe that the locative must occur in the same position we found for the postnominal demonstrative: (26)

a. Este libro viejo de aquí de sintaxis / sobre la guerra está mal hecho.14 this book old of here of syntax / about the war is badly made b. *Este libro de aquí viejo de sintaxis / sobre la guerra . . . c. *Este libro viejo de sintaxis / sobre la guerra de aquí . . . ‘This old book here on syntax / about the war is badly made.’

I assume that the sequence demonstrative + locative has to be considered a constituent. Arguments that can support this hypothesis can be found in other languages in which this sequence can appear. One of these arguments is that in some languages the sequence demonstrative + locative can or must move together to [Spec, DP] before Spell-Out, as happens in Bosnian, in nonstandard English, and in Scandinavian: (27)

a. ona tamo (nova) knjiga (Bosnian) that there (new) book b. this here (nice) book (English) c. denne herre (flotte) bilen (Norwegian) this here (nice) car-the

Bosnian provides another piece of evidence that the sequence we are examining is a constituent, as noticed in Brugè and Giusti (1996). In fact, when the noun phrase modified by the sequence demonstrative + locative is fronted, a clitic in Wackernagel position can follow either the whole noun phrase, as in (28a), or the sequence onu tamo ‘that there’, as in (28b): (28)

a. [ [ [onu] tamo] knjigu] mi donesi that-ACC there book-ACC me(CL.DAT) give(IMP) b. [ [onu] tamo] mi knjigu donesi that-ACC there me(CL.DAT) book-ACC give(IMP) ‘Give me that there book.’

However, if in the same construction the head noun is modified by the demonstrative alone and an adjective, the clitic must appear immediately after the demonstrative, as in (29a), and it can never appear after an adjective, as in (29b): (29)

a. onu mi novu knjigu donesi that-ACC me(CL.DAT) new book-ACC give(IMP) b. *onu novu mi knjigu donesi that-ACC new me(CL.DAT) book-ACC give(IMP) ‘Give me that new book.’

The well-formed sentences in (28) and (29a), contrasted with the ungrammaticality of (29b), which is due to the fact that the demonstrative plus the adjective do not form a constituent, strongly supports the constituent status for the sequence demonstrative + locative. Let us now examine the internal structure of the constituent built by the demonstrative and the locative. As (25) and (26) show, we assume that the base position it

POSITIONS OF DEMONSTRATIVES

27

occupies in the structure is the same position we found for the postnominal demonstrative. To capture the strict relationship that is established between the two elements, we extend to this sequence the analysis that Kayne (1994) proposes for the deconstructions in French. We assume that the sequence este de aquí ‘this of here’ has a complex internal structure in which the element de ‘of’ occupies the head position of a maximal projection comparable to the CP projection.15 Este ‘this’ and aquí ‘here’ are considered as maximal projections dominated by the “preposition” de ‘of’, which establishes a predication relation between them, as in a small clause with an abstract copula. This predication relation can be further motivated by proposing that este and aquí must both agree, either for the positive value or for the negative value of the feature [± speaker]. This fact would also account for the impossibility of combining este ‘this’ with de allí ‘of there’ or aquel ‘that’ with de aquí ‘of here’ (see note 12). Keeping these considerations in mind, I suggest for the complex sequence demonstrative + locative the following internal structure:16 (30)

. . .FP F'

XP Spec. X˚ de

AGRP(Possessive)/NP



X' ZP DemP este

Z' Z˚

PP aquí

In this structure, este moves obligatorily to [Spec, XP], assuming that de in X° requires that its specifier be occupied by a lexical element. In this way, we obtain the order este de aquí and the orders ese de acá, ese de allí, aquel de allí, and aquel de allá in the other cases.17 The position to which the demonstrative raises, namely the [Spec, XP] position, gives, then, the demonstrative the possibility of moving to [Spec, DP] before SpellOut or, at the latest, at Logical Form, for reasons I will present later in section 5.18

4. Postnominal demonstrative and postnominal possessive adjective Let us observe, now, the behavior of the postnominal demonstrative with the postnominal possessive adjective to ascertain whether the structure in (30) is correct. In Spanish, there exist two paradigms for possessive forms: the clitic series and the strong series. The clitic series is morphologically poorer (i.e., mi ‘my’ / tu ‘your’ / su ‘his/her/their’) and differs from the strong series in syntactic behavior. In fact, while the clitic forms can appear only in prenominal position, in comple-

28

DP FUNCTIONAL STRUCTURE

mentary distribution with the article, the strong forms can appear in postnominal position and in predicative constructions. They can never appear in prenominal position. As for the prenominal possessive forms, we called them clitic forms because they undergo all the syntactic tests that characterize them as clitics, as many authors have pointed out. As for the position these elements occupy in the structure, I adopt Picallo’s (1994) hypothesis (see note 6), who proposes that, in Spanish, possessives raise to D° in Syntax. She suggests that possessives move as XPs, namely from Spec to Spec, as far as the functional projection immediately dominated by DP, which we can call AgrGenitivePhrase (AgrGP) following Siloni (1994) and Longobardi (1995). While in languages such as Italian and Catalan this position represents the last step for movement, in Spanish the same elements move further as X° elements to D°. This last step is justified by the clitic nature of possessives in Spanish and their complementary distribution with the definite article. Let us consider, now, the possessive in postnominal position. The only available forms that can appear in this position are the forms belonging to the strong series. Observing the unmarked word order inside the nominal extended projection, we can notice that the postnominal possessive undergoes the same restrictions that characterize the postnominal demonstrative.19 When it modifies an event noun, the only position the postnominal possessive can occupy is after a manner adjective–namely, the class of adjectives that occupy the lower functional projection in (22). (31)

a. La reacción desinteresada suya me parece sospechosa. the reaction disinterested his/her to-me(CLIT) seems suspicious b. *La reacción suya desinteresada . . .20 the reaction his/her disinterested ‘His/her disinterested reaction to me seems suspicious.’

The contrasts in (31) can be considered convincing evidence that in Spanish the postnominal possessive occupies a position lower than all the functional projections containing APs. As for the position the postnominal possessive occupies with respect to the complements of the nouns, we can observe in (32) and (33) that this category always precedes them: (32)

a. La reacción suya a las críticas preocupó a todos. the reaction his/her to the criticisms worried everybody b. *La reacción a las críticas suya . . . the reaction to the criticisms his/her ‘His/her reaction to the criticisms worried everybody.’ c. El descubrimiento suyo de América produjo mucha riqueza. the discovery his of America produced great wealth d. *El descubrimiento de América suyo . . . the discovery of America his ‘His discovery of America produced great wealth.’

(33)

a. El cuadro suyo de Picasso no es nada más que una copia mal hecha. the painting his/her of Picasso is nothing more than a copy badly made b. *El cuadro de Picasso suyo . . . the painting of Picasso his/her ‘His /her painting by Picasso is nothing more than a badly made copy.’

POSITIONS OF DEMONSTRATIVES

29

c. El cuadro chico suyo de Picasso . . . the painting little his/her of Picasso . . . d. *El cuadro chico de Picasso suyo . . . the painting little of Picasso his/her e. *El cuadro suyo chico de Picasso . . . the painting his/her little of Picasso ‘His/her little painting by Picasso . . .’

Finally, as for the relative order between the postnominal demonstrative and the postnominal possessive, we can notice that the postnominal demonstrative always precedes the postnominal possessive, (34). Therefore, we propose that the postnominal possessive occupies a position lower than the one occupied by the postnominal demonstrative. (34)

a. El libro (viejo) este suyo de sintaxis no me convence. the book (old) this his/her of syntax not me convince b. *El libro (viejo) suyo este de sintaxis . . . c. Este libro (viejo) suyo de sintaxis . . . ‘This (old) book on syntax of his/hers does not convince me.’

The data presented so far makes us wonder which position the postnominal possessive occupies inside the nominal extended projection. Recently, Picallo, in a brief footnote, says that in Catalan the postnominal possessives “do not behave like arguments, instead, they seem to behave like pseudoadjectives in some respects” (1992, 49). Because the postnominal possessive does not necessarily occupy a “marginalized” position inside the structure, as commented on in note 19 here and as the contrasts in (32) and (33) confirm (if it follows the APs, it always precedes the complements of the noun), I could suggest two tentative proposals. However, the validity of both of them cannot be easily checked through empirical data, given Cinque’s generalization (1980, 1981) concerning the realization of genitive arguments as possessives. The first proposal is that the postnominal possessive occupies either a unique position inside the NP—namely, the higher specifier of this projection—or different positions inside the same projection, according to the q-role it can saturate in the different cases: that is, it occupies the same positions occupied by the corresponding genitive PPs. The second proposal is to assume that there exists another functional projection external to the NP which immediately dominates it and whose specifier hosts the postnominal possessive after Spell-Out. This second proposal seems to me to be more plausible from a theoretical point of view. A piece of evidence that can support it is that the postnominal possessive shares the same properties of the adjectives rather than those of the complements of the noun. In fact, it agrees obligatorily in gender and number with the head noun, and it does not need the insertion of the preposition marking genitive case de ‘of’ in Spanish.21 Therefore, according to the resulting order in (34), and slightly modifying generalization (20), I propose that: (35) In Spanish the demonstrative is generated in the specifier position of a nominal functional projection which immediately dominates either the functional projection containing the possessive or the NP projection.

30

DP FUNCTIONAL STRUCTURE

5. Movement of the demonstrative and its intrinsic features Let us now come back to the main topic of our investigation, and let us observe the following data: (36)

a. El libro este/ese/aquel fue publicado en 1990. the book this/that/that was published in 1990 b. Este/ese/aquel libro fue publicado en 1990. this/that/that book was published in 1990 ‘This/that book was published in 1990.’

In the preceding sections we have proposed the insertion of another functional projection (FP) inside the extended nominal projection corresponding to the position where the postnominal demonstrative occurs (see (22) and (23)). Then, we have proposed that in Spanish the demonstrative is generated in the specifier of this FP (see (35)). According to this proposal, in (36a) the demonstrative appears, after Spell-Out, in the same position in which it is generated, while the noun raises before Spell-Out to a higher X° position. In (36b), in contrast, the prenominal position the demonstrative occupies is a derived position, due to the long movement of the demonstrative itself from [Spec, FP] to [Spec, DP] before Spell-Out. Moreover, given the grammaticality of both (36a) and (36b), I propose that in Spanish the movement of the demonstrative to [Spec, DP] is optional before Spell-Out. At this point, it becomes necessary to justify, in accordance with the Minimalist framework (Chomsky, 1995), what enables the demonstrative to move before Spell-Out. We can assume that a common noun modified by a demonstrative is interpreted as a referential nominal expression. In other words, we can assume that a common noun modified by a demonstrative designates directly the entity which it refers to, and, for this property, it can receive neither an existential nor a generic interpretation, just as happens with proper names and pronouns. Consequently, I propose that the demonstrative makes the common noun it modifies behave like a proper name or, more appropriately, like a pronoun. On the basis of these observations, I formulate the following assumption: (37) The demonstrative is an element specified for the features [+Referential] and [+Deictic].

From a syntactic point of view, an argument to support (37) is given by the fact that, as in the case of proper names and pronouns, a nominal modified by a demonstrative cannot be further modified by a restrictive relative clause, as shown in (38) and (39): (38)

a. Este libro, que publicó el año pasado, tuvo poco éxito. this book that he-published the year last had little success b. El libro este, que publicó el año pasado, tuvo poco éxito. the book this that he-published the year last had little success ‘This book that he published last year had little success.’

POSITIONS OF DEMONSTRATIVES

(39)

31

a. *Sólo comprarán estos libros que hayan sido publicados después del 90. only they-will-buy these books that have(SUBJ) been published after the ’90s b. *Sólo comprarán los libros estos que hayan sido publicados después del ’90. only they-will-buy the books these that have(SUBJ) been published after the ’90s ‘They will buy only these books that have been published after the ’90s.’

The sentences in (38) can be judged well-formed only if the relative clause is interpreted as appositive. The restrictive interpretation for the relative clause is excluded in both cases. This fact can also explain why the sentences in (39) are ungrammatical. In these cases the relative clause can be interpreted only as a restrictive relative clause because of the presence of the subjunctive mood in the relative clause itself.22 Furthermore, we assume that D° contains the Referentiality feature [± REF], the positive or negative value of which must be checked by some element in the structure, according to the following assumption: (40) All D positions are universally generated with an abstract feature ±R (suggesting “referential”), which must be checked with respect to at least one of its values. (Longobardi 1994, (97), p. 659)

Finally, we propose that in Spanish the [+REF] feature in D° must obligatorily be checked by the demonstrative on the basis of the proposals just made. We also propose that, in this language, such checking can take place either before Spell-Out or at Logical Form. If the movement of the demonstrative is carried out before Spell-Out, the [+REF] feature in D° is checked already at this level through the Spec-Head Agreement process. In this way, we will have constructions such as (36b). In these cases, the head D cannot be lexically filled for the reasons presented in section 1.3. Alternatively, if the demonstrative movement is not carried out before Spell-Out, as in (36a), the movement rule on the demonstrative must apply at Logical Form to satisfy the referential interpretation that the nominal must receive, which, also in these cases, can be obtained in [Spec, DP] through the Spec-Head Agreement process. In these cases, the head D has to be realized at PF for the reasons presented in section 1.3. Therefore, I propose, for the demonstrative in Spanish, the following assumption: (41) In Spanish the demonstrative can raise to [Spec, DP] optionally before Spell-Out, but it must raise to [Spec, DP] obligatorily at Logical Form.

This hypothesis can be argued for by the fact that the interpretation of the nominals modified by a demonstrative is always referential, even in cases in which the movement has not taken place before Spell-Out. One argument to support this hypothesis is provided by the interpretation of (38b) and by the ungrammaticality of (39b). Another argument is given by the ungrammaticality of the following two sentences: (42)

a. *Algunos estos libros tuvieron poco éxito. some these books had little success b. *Algunos libros estos tuvieron poco éxito. some books these had little success

32

DP FUNCTIONAL STRUCTURE

Example (42a) is excluded because the DP estos libros ‘these books’ can receive neither the existential reading nor the partitive case, either of which is required and assigned by the existential quantifier algunos ‘some’ to its complement (see Cardinaletti and Giusti 1993).23 I propose that this fact is due to the presence of the demonstrative, which gives the DP itself the referential interpretation.24 Example (42b) is ill-formed even if the definite article is absent (see the bare plurals algunos libros llegaron ayer ‘some books arrived yesterday’ and the modified bare plurals algunos libros interesantes llegaron ayer ‘some interesting books arrived yesterday’), and the demonstrative appears in postnominal position. This fact leads us to extend the considerations made to account for the ungrammaticality of (42a) to this case: in (42b) the presence of the demonstrative in postnominal position makes the DP incompatible with the existential reading and the partitive case. Given that the existential interpretation for nominals depends on the content of D° (see Longobardi 1994), we conclude that in (42b) this projection is specified for the [+REF] feature, and that at Logical Form the unique element in the structure that can satisfy the checking of this feature is the demonstrative itself. The cases in (42) support the hypothesis that there exists a strict relation between the low position in which I propose the demonstrative is generated and the position inside DP where the referential interpretation takes place. Consequently, the [Spec, DP] position is a derived position for the demonstrative in Spanish, to which it must raise in any case, at the latest at Logical Form. A third empirical argument to support the assumptions in (37) and (41) is represented by the following cases: (43)

a. *Este mi libro de sintaxis fue publicado hace dos años.25 this my book on syntax was published ago two years b. *Mi libro este de sintaxis fue publicado hace dos años. my book this on syntax was published ago two years ‘This book on syntax of mine was published two years ago.’

Example (43a) is excluded because in the DP projection, both the specifier and the head appear lexically filled, contrary to Dimitrova-Vulchanova and Giusti’s (1998) hypothesis (see section 1.3 here). The fact that also (43b) is ungrammatical even if only the head D is filled by the possessive at PF leads us to conclude, once again, that at Logical Form the demonstrative must be in any case interpreted in [Spec, DP]. In cases like these, the demonstrative, which must raise to [Spec, DP] in order to check its [+REF] feature by Spec-Head Agreement, cannot satisfy this requirement because the same feature has been already checked by the possessive which has moved to D° before Spell-Out.26 So, the [+REF] feature is no longer available for the demonstrative.27

6. Comparative approach Observing the syntax of the demonstrative in a wider cross-linguistic perspective, we could suggest that across languages there exists more than one position inside the DP projection in which this element is generated, in this way parameterizing different realizations of the demonstrative in the structure with respect to the head noun (see Giusti’s 1993 proposal for Romanian).

POSITIONS OF DEMONSTRATIVES

33

Nevertheless, my intention would be to propose that in all languages there exists a unique position in the structure in which the demonstrative is generated. It might be suggested that in all languages the unique position in which the demonstrative is generated is the one I have proposed and justified for Spanish. The languages will vary, however, as to their power to allow, oblige, or prevent the movement of the demonstrative to [Spec, DP] before Spell-Out. A way to make this hypothesis theoretically consistent is to find some empirical argument that can show that even in those languages which do not allow the demonstrative to appear in postnominal position at PF, this element starts from the same low position I have proposed for Spanish. One argument, which can be found in many languages, is represented by the position the locative (which functions as deictic reinforcer of the demonstrative) occupies in the structure (see section 3.2 here). In those languages in which the demonstrative obligatorily appears in [Spec, DP] and can co-occur with the locative, it can be argued that the demonstrative has moved before Spell-Out, leaving the locative in the base position, on a par with what happens for the sequence demonstrative + locative in Spanish, (30). Therefore, in these languages, the position of the locative itself will be taken to indicate the base position of the demonstrative, in much the same way as a floating quantifier marks the base position of the noun phrase it quantifies in Sportiche’s (1988) hypothesis. 6.1. Demonstrative in other languages 6.1.1. Languages in which the demonstrative can appear in postnominal position Let us begin our general overview by examining the behavior of the demonstrative in a language related to Spanish. In some areas of Catalonia, the demonstrative modifying a nominal can appear either in prenominal position or in postnominal position, as (44) shows: (44)

a. Aquest quadre és molt antic. this painting is very antique b. El quadre aquest és molt antic. the painting this is very antique ‘This painting is very antique.’

Putting aside those areas in which the phenomenon does not take place, in Catalan, the postnominal demonstrative appears in a position lower than all the functional projections containing adjectives. This happens both in the case in which the head noun belongs to the event nouns class, (45), and in the case in which the head noun belongs to the object-denoting nouns class, (46): (45)

a. Aquesta reacció desinteressada ens ha preocupat a tots. this reaction disinterested us(CL.) has worried all b. ?La reacció desinteressada aquesta . . . the reaction disinterested this c. *La reacció aquesta desinteressada . . . the reaction this disinterested ‘This disinterested reaction has worried all of us.’

34 (46)

DP FUNCTIONAL STRUCTURE

a. Aquest quadre rodó és molt antic. this painting round is very antique b. ?El quadre rodó aquest . . . the painting round this c. *El quadre aquest rodó . . . the painting this round ‘This round painting is very antique.’

With respect to the subject and the other complements of the noun, the Catalan postnominal demonstrative always appears preceding them in the unmarked order, as the contrasts in (47) with an event noun, and in (48) with an object-denoting noun show: (47)

a. La reacció aquesta d’Alemanya / a les crítiques no ha impressionat ningú. the reaction this of Germany / to the criticisms not has shaken anyone b. *La reacció d’Alemanya / a les crítiques aquesta . . . the reaction of Germany / to the criticisms this ‘This reaction by Germany / to the criticisms has not shaken anyone.’

(48)

a. El germà aquest d’en Joan és molt jove. the brother this of Joan is very young b. *El germà d’en Joan aquest . . . the brother of Joan this ‘This brother of Joan’s is very young.’

Finally, since the postnominal demonstrative also precedes the postnominal possessive, as we can observe by the contrasts in (49), we extend to this language the hypothesis proposed for Spanish (see (35)). In Catalan, the movement of the demonstrative from [Spec, FP] to [Spec, DP] in order to check its [+REF] feature is optional before Spell-Out but obligatory at Logical Form, (41). (49)

a. ??El llibre aquest seu de sintaxi no em convenç. the book this his on syntax not me convince b. *El llibre seu aquest de sintaxi no em convenç. the book his this on syntax not me convince ‘This book on syntax of his does not convince me.’

Also in Catalan, the demonstrative aquest ‘this’ can independently realize its [+Deictic] feature through the locative reinforcer d’aquí ‘of here’. As happens in Spanish, the complex sequence aquest d’aquí can appear in the same position we propose the demonstrative is generated in, (50a), or the demonstrative alone can raise and the locative stay “in situ,” as in (50b). (50)

a. ?El quadre (rodó) aquest d’aquí (de Picasso) està mal penjat. the painting (round) this of here (of Picasso) is badly hanged b. Aquest quadre (*d’aquí) (rodó) (d’aquí) (de Picasso) (*d’aquí) està mal penjat. this painting (*of here) (round) (of here) (of Picasso) (*of here) is badly hanged ‘This (round) painting here (by Picasso) is badly hanged.’

Another language in which the demonstrative shares the same syntactic behavior we have seen in Spanish and in Catalan is Bosnian. In this language, the demonstrative can appear, besides the prenominal position, in postnominal position, leaving

POSITIONS OF DEMONSTRATIVES

35

the DP projection empty, as in (51). Bosnian does not display a definite article at all. The realization of the demonstrative in postnominal position is acceptable only at a highly stylistic level.28 (51)

a. ovaj/onaj lijepi momak this(MASC.SING.NOM.)/that(MASC.SING.NOM.) nice boy b. ?lijepi momak ovaj/onaj nice boy this/that ‘this/that nice boy’

We extend to this language the general hypothesis proposed for Spanish and Catalan. Arguments that support this proposal are given by the contrasts in (52) through (55):29 (52)

a. ?nezainteresovana reakcija ova na kritike disinterested reaction this(FEM.SING.NOM.) on criticisms b. *nezainteresovana reakcija na kritike ova disinterested reaction on criticisms this(FEM.SING.NOM.) ‘this disinterested reaction to the criticisms’

(53)

a. ?suknja ova Marijina skirt this(FEM.SING.NOM.) Mary(GEN.) b. *suknja Marijina ova skirt Mary(GEN.) this(FEM.SING.NOM.) ‘this skirt of Mary’s’

In (52) and (53), the postnominal demonstrative can only precede a complement of the noun. At the same time, if a postnominal possessive is realized in the structure, the postnominal demonstrative always precedes it, (54): (54) knjiga (ova) tvoja (*ova) o sintaksi book (this) your (*this) of syntax ‘this book on syntax of yours’

Finally, since in this language the deictic feature of the demonstrative can be reinforced by a locative, the complex sequence, on the one hand, and the locative alone, on the other, can appear in the base position, (55):30 (55)

a. knjiga ova ovdie o sintaksi book this here of syntax b. ova knjiga ovdie o sintaksi (*ovdie) this book here of syntax (*here)

Therefore, we extend to Bosnian the hypothesis I have proposed for Spanish. We also extend to the sequence ova ovdie ‘this here’ the structure proposed in (30). However, we assume that in this language, as in English, (27b), Norwegian, (27c), and, as we shall see later, in others, the preposition that heads the XP projection and which selects the small clause is lexically empty. This case is comparable to those cases in which infinitival complement clauses are introduced by an empty complementizer.31 In Romanian the demonstrative can appear at PF either in prenominal position, (56a), or in postnominal position, (56b).32

36 (56)

DP FUNCTIONAL STRUCTURE

a. acest (frumos) b£iat (frumos) al Mariei this (nice) boy (nice) of Mary b. b£iatul acesta frumos al Mariei33 boy-the thisA (nice) of Mary c. b£iatul frumos (*acesta) al Mariei (*acesta) boy-the nice (*thisA) of Mary (*thisA) ‘this nice boy of Mary’s’

We can assume that, when the demonstrative appears postnominally, it occupies the specifier of a maximal projection immediately dominated by the DP, as the contrasts in (56b–c) show (see also the discussion in section 1.1). I suggest also that the position the demonstrative occupies in (56b) is a derived position and not the base one. The impossibility in Romanian of realizing the demonstrative in its base position (compare the ungrammaticality of (56c)), and the impossibility of realizing the locative reinforcer independently (for instance,*b£iatul acesta (de) aici / *acest b£iat (de) aici [lit.: boy-the thisA here / this boy here]) prevent us from seeing the base position in which the demonstrative is generated. Therefore, the null hypothesis is to assume that in Romanian the demonstrative starts in the [Spec, FP] as in the other languages—see (30)—and raises obligatorily to an intermediate position before Spell-Out. The hypothesis that the postnominal position in which the demonstrative appears at PF is an intermediate position can be supported by observing what happens in other languages in which the demonstrative can be realized at PF in more than two positions. One of these languages is Bosnian. In fact, in this language, besides the base position and the [Spec, DP] position (see (51) through (53)), the demonstrative can appear in what we can define an intermediate position, as (57) compared with the cases in (51) shows: (57) ?lijepi ovaj/onaj momak nice this/that boy ‘this/that nice boy’

Another language that also displays this intermediate position, besides the base one and the high position in [Spec, DP], is Modern Greek (cf. Giusti 1997), as (58) shows: (58) (afto) to oreo (afto) to vivlio (afto) tou Janis (this) the nice (this) the book (this) of Janis ‘this nice book of Janis’s’

On the basis of the data in Bosnian, (57), and in Modern Greek, (58), we can extend the hypothesis to Romanian. In this language, for reasons we can ignore at this point of the investigation, neither the demonstrative alone nor the sequence demonstrative + locative, nor the locative alone can be realized in their base position. The only available position in which the demonstrative can appear at PF is the intermediate position and the high [Spec, DP] position. So, I propose that in Romanian the demonstrative must obligatorily move to the intermediate position before SpellOut and can optionally raise to [Spec, DP] before Spell-Out, but, as happens in all other languages, it must obligatorily move to [Spec, DP] at the latest at Logical Form in order to check its referential feature.34

POSITIONS OF DEMONSTRATIVES

37

6.1.2. Languages in which the demonstrative must appear in prenominal position To extend our hypothesis to those languages in which the demonstrative must appear in [Spec, DP], it is necessary to find some empirical evidence that can justify that the demonstrative starts from the low [Spec, FP] position (see (22) and (23)). So, let us detect the position the locative reinforcer has in the nominal structure of these languages, given that, as we showed for Spanish, in the case in which the demonstrative raises to [Spec, DP] before Spell-Out, the locative reinforcer remains in its base position (see (25b), unless the movement before Spell-Out affects the entire complex, (27). Let us start by examining the behavior of the demonstrative in Italian. In Italian, the only position the demonstrative can occupy is the prenominal one—that is, [Spec, DP]. In this language it is possible to realize the locative element with the value of reinforcer of the deictic feature peculiar to the demonstrative. The two elements can appear string-adjacent only when the head noun is not lexically realized. The resulting combinatorial sequences, are as follows: questo qui ‘this here’, questo qua ‘this here’, quello lì ‘that there’, and quello là ‘that there’. When the demonstrative modifies a noun, the locative reinforcer always appears in postnominal position: (59) Questo libro qui non è molto interessante.35 ‘This book here is not very interesting.’

When the noun is modified by a “postnominal” adjective, the locative qui must follow the adjective itself in the unmarked word order. This is true either with both event nouns and object-denoting nouns, as (60) and (61) show: (60)

a. ?Questa risposta diplomatica qui non convincerà nessuno. this answer diplomatic here not will convince anyone b. *Questa risposta qui diplomatica non convincerà nessuno. this answer here diplomatic not will convince anyone ‘This diplomatic answer here will not convince anyone.’

(61)

a. ?Questo libro vecchio qui non deve essere venduto. this book old here not must be sent b. *Questo libro qui vecchio non deve essere venduto. this book here old not must be sent ‘This old book here must not be sent.’

Finally, as for the position the locative occupies in relation to the PP subject of the noun and the other complements of the noun, we find that in Italian, too, the locative reinforcer must always precede them, as (62) shows: (62)

a. Questa risposta qui del governo / al problema è una vera vergogna. this answer here of the government / to the problem is a real disgrace b. *Questa risposta del governo / al problema qui è una vera vergogna. this answer of the government / to the problem here is a real disgrace ‘This answer here by the government / to the problem is a real disgrace.’ c. Questo libro qui di sintassi / di Gianni è molto interessante. this book here on syntax / of Gianni is very interesting

38

DP FUNCTIONAL STRUCTURE

d. *Questo libro di sintassi / of Gianni qui è molto interessante. this book of syntax / of Gianni here is very interesting ‘This book here on syntax / of Gianni’s is very interesting.’

On the basis of the resulting well-formed constructions presented so far, we can observe that in Italian the locative reinforcer appears in the same position in which the postnominal demonstrative, the complex sequence demonstrative + locative or the locative alone appear in Spanish. Therefore, we conclude that also in Italian the demonstrative is generated in the low [Spec, FP] position (see (30)). Nevertheless, contrary to Spanish, in Italian the demonstrative must obligatorily move to [Spec, DP] already before Spell-Out in order to check its [+REF] feature. In French, too, where the demonstrative must appear prenominally at PF, it is possible to realize the locative reinforcer. In most areas of the country, its presence in the structure has the function of making the deixis explicit: namely, of expressing the different types of deictic relation between the speaker and the identified “object.” In French this relation is not morphologically expressed by the demonstrative alone because there exists a unique demonstrative form: ce.36 (63) ce livre-ci / ce livre-là 37 this book here / that book there

As for the position the locative reinforcer occupies in the structure, we can observe that in French the locative always follows the adjectives, (64), and always precedes the genitive PPs and the other complements of the noun, (65): (64)

a. ce livre rouge-ci / ce livre italien-ci this book red here / this book Italian here b. *ce livre-ci rouge / *ce livre-ci italien this book here red / this book here Italian ‘this red book here / this Italian book here’

(65)

a. ce livre-ci de Jean this book here of John b. ?*ce livre de Jean ci38 this book of John here ‘This book here of John’s’ c. ce reponses-ci au probléme this answer here to the problem d. *ce reponses au probléme ci this answer to the problem here ‘this answer here to the problem’

Because in French the locative reinforcer always appears in the same position in which the postnominal demonstrative, the complex sequence demonstrative + locative and the locative alone are realized in Spanish, we extend the hypothesis also to this language. We can observe the same results in German, as the cases in (66) show: (66)

a. dieses Buch hier this book here

POSITIONS OF DEMONSTRATIVES

39

b. dieses schöne Buch (hier) von Hans (*hier) this nice book (here) of Hans (*here) ‘this nice book here of Hans’s’

Among other languages, in Albanian we can observe the same results, as (67) and (68) show:39 (67)

a. ky libr-i këtu this(NOM) book-the(NOM) here ‘this book here’ b. ky libr-i (*këtu) i kuq (këtu) this book-the (*here) I red (here) ‘this red book here’ c. ky reagim-i (*këtu) i shpijtë (këtu) this reaction-the (*here) I quick (here) ‘this quick reaction here’

(68)

a. ky libr-i i kuq (këtu) mbi sintakën (*këtu) this book-the I red (here) on syntax-the(ACC) (*here) ‘this red book here on syntax’ b. ky reagim-i i shpijtë (këtu) ndaj problemit (*këtu) this reaction-the I quick (here) to problem-the(ABL.) (*here) ‘this quick reaction here to the problem’

So, we propose, that also in those languages in which the demonstrative must appear in [Spec, DP] at PF, the demonstrative is generated in the same position we have found for Spanish. The difference between Italian, French, German, Albanian, etc., on the one hand, and Spanish, Catalan, Bosnian, Russian, Romanian, Modern Greek, etc., on the other, is that in the first group of languages the demonstrative does not have the possibility of remaining in its base position at PF, or, in some cases, of appearing in an intermediate position; rather, it must move to [Spec, DP] already before Spell-Out. 6.1.3. Languages in which the demonstrative must appear in postnominal position In some languages the demonstrative must be realized in a postnominal position at PF. It can never raise to [Spec, DP] before Spell-Out. In these languages, the crucial empirical evidence for our hypothesis is the position the demonstrative occupies in the structure in relation to the other categories that belong to the extended nominal projection. The first language which belongs to this group is Hebrew, as (69) shows:40 (69)

a. (*ha-zot/ha-hi) ha-tšuva (ha-zot/ha-hi) (*this(FEM.)/that(FEM.)) the-answer(FEM.) (this(FEM.)/that(FEM.)) ‘this/that answer’ b. (*ha-ze/ha-hu) ha-sefer (ha-ze/ha-hu) (this(MASC.)/that(MASC.)) the-book(MASC.) (this(MASC.)/that(MASC.))

If, in the same structure, an adjective is realized, the demonstrative always follows this modifier, (70a, c), and can never precede it, (70b, d).

40 (70)

DP FUNCTIONAL STRUCTURE

a. ha-tšuva ha-meʕanyenet ha-zot/ha-hi the-answer(FEM.) the-interesting this(FEM.)/that(FEM.) b. *ha-tšuva ha-zot/ha-hi ha-meʕanyenet the-answer(FEM.) this(FEM.)/that(FEM.) the-interesting ‘this/that interesting answer’ c. ha-sefer ha-yafe ha-ze/ha-hu the-book(MASC.) the-nice(MASC.) this(MASC.)/that(MASC.) d. *ha-sefer ha-ze/ha-hu ha-yafe the-book(MASC.) this(MASC.)/that(MASC.) the-nice(MASC.) ‘this/that nice book’

The cases in (70) show that with respect to the adjectives the demonstrative appears in the same position we have found for Spanish. We have the same results as presented for Spanish when the demonstrative cooccurs with complements of the noun. In these cases, in fact, the demonstrative always precedes the subject and the other complements of the noun: (71)

a. ha-tšuva (ha-meʕanyenet) ha-zot šel Dani / la-seʔela41 the answer (the-interesting) this of Dan / to-the-problem b. *ha-tšuva (ha-meʕanyenet) šel Dani / la-seʔela ha-zot . . . the answer (the-interesting) of Dan / to-the-problem this ‘this interesting answer by Dan / to this problem’ c. ha-sefer (ha-yafe) ha-ze ʕal taxbir / šel Dani the-book (the-nice) this on syntax / of Dan d. *ha-sefer (ha-yafe) ʕal taxbir / šel Dani ha-ze the-book (the-nice) on syntax / of Dan this ‘this nice book on syntax / of Dan’s’

Furthermore, in Hebrew the demonstrative can be reinforced by a locative element. The resulting complex sequences ha-ze kan ‘this here’ and ha-hu sham ‘that there’ cannot be separated by any category, and they appear in the same position in which the demonstrative alone appears inside the nominal projection, as (72) shows: (72)

a. ha-sefer ha-yafe ha-ze (kan) šel Dani (*kan) the-book the-nice this (here) of Dan (*here) ‘this nice book here of Dan’s’ b. ha-sefer ha-yafe ha-hu (sham) šel Dani (*sham) the-book the-nice that (there) of Dan (*there) ‘that nice book there of Dan’s’

Another language, unrelated to Hebrew, in which the demonstrative must appear in a low position at PF is Irish. As (73) shows, in Irish the demonstrative forms seo ‘this’ and sin ‘that’must be realized in postnominal position:42 (73) an fear seo/sin the man this/that

an leabhar seo/sin the book this/that

Moreover, as happens in Hebrew, the demonstrative can never precede the adjective, and, at the same time, it can never follow a complement of the noun, as (74) shows. In Irish, the demonstrative appears in a structural position lower than the positions the adjectives occupy and immediately superior to the NP projection.

POSITIONS OF DEMONSTRATIVES

41

(74) an leabhar (*seo) nua (seo) faoi teangedaiocht (*seo) the book (*this) new (this) on linguistics (*this) ‘this new book on linguistics’

Therefore, I propose that in languages such as Hebrew and Irish the demonstrative is realized in its base position at PF. What distinguishes these two languages from languages such as Spanish and Italian is that in Hebrew and Irish the movement of the demonstrative to [Spec, DP] can be neither optional nor obligatory before SpellOut. Rather it must move to this position only at Logical Form in order to check its [+REF] feature. 6.2. The general hypothesis In section 5 I provided evidence to assume that in Spanish the demonstrative, even when it appears in postnominal position, is in any case interpreted in [Spec, DP], given that only in this position it can check, through Spec-Head Agreement, the [+REF] feature we showed it is specified for, (37). This requirement allows us to justify the existence of a strict relationship between the two positions in which the demonstrative can appear—[Spec, DP] and [Spec, FP]—and, consequently, to propose that in Spanish the demonstrative is generated in [Spec, FP]—(see (22) and (23)). The demonstrative raises to [Spec, DP] optionally before Spell-Out but obligatorily at Logical Form, (41). In section 6 I showed evidence to assume that cross-linguistically the demonstrative is generated in a low position inside the extended nominal projection, namely in the [Spec, FP] position. Moreover, because the interpretation of the demonstrative is the same across languages, I proposed that even in those languages in which this element can appear only in the base position it must in any case move to [Spec, DP] at Logical Form in order to check its [+REF] feature. According to Chomsky (1995), a type of variation across languages depends on whether a particular feature is strong (the checking process occurs already at PF) or weak (the checking process must be carried out only at Logical Form). If it is strong, and the checking process implicates a movement rule, the movement must take place before Spell-Out; otherwise, if the same feature is weak the movement must take place after Spell-Out, namely at Logical Form. Therefore, we formulate for the demonstrative the following parameterized principle: (75) Checking the [+REF] feature in Spec.DP is obligatory by Logical Form.43 a. The Demonstrative checks its [+REF] feature in [Spec, DP] before Spell-Out when this feature is strong. b. The Demonstrative checks its [+REF] feature in [Spec, DP] after Spell-Out when this feature is weak.

Principle (75) can account for the syntactic behavior of the demonstrative in all the languages we have examined until now. Parameter (75a) accounts for the behavior of the demonstrative we observe in Italian, French, German, and Albanian. In languages such as these, the demonstrative always appears in [Spec, DP] because, with its [+REF] feature being strong, it has to be

42

DP FUNCTIONAL STRUCTURE

checked already before Spell-Out. However, (75b) accounts for the behavior of the demonstrative in Hebrew and Irish. Since the [+REF] feature on the demonstrative is weak in languages such these, the checking via movement must take place only at Logical Form, and, for this reason, at PF the demonstrative must be realized in its base position. Finally, as for languages such as Spanish, Catalan, Bosnian, Russian, Romanian, and Modern Greek, we propose that the [+REF] feature on the demonstrative can be either strong or weak. The fact that these languages choose either of the two options in (75) can account for the optionality for the demonstrative to appear before SpellOut either in its base position or in [Spec, DP].44 7.

Conclusions

We have proposed that in Spanish there exists another functional projection, inside the extended nominal projection, in which the demonstrative is generated, and the position the demonstrative occupies is the specifier of this functional projection. We can start from the idea of providing a unified analysis for pairs such as (76) and (77): (76)

a. esta reacción al problema this reaction to the problem b. la reacción esta al problema the reaction this to the problem ‘this reaction to the problem’

(77)

a. este libro gordo de sintaxis this book big on syntax b. el libro gordo este de sintaxis the book big this on syntax ‘this big book on syntax’

Then, by adopting the fundamental assumptions of the antisymmetric hypothesis (Kayne 1994) and the hypothesis proposed by Cinque (1994) on the internal structure of nominals and the movement of the noun, we have seen that: A. The postnominal demonstrative is realized in the specifier of a functional projection lower than all the other functional projections containing the different classes of adjectives. B. The functional projection that contains the postnominal demonstrative immediately dominates either the functional projection containing the postnominal possessive, if any, or the NP projection, given that the postnominal demonstrative has to precede the postnominal possessive, the PP subject of the noun and all other PPs complements of the noun, (76b) and (77b). C. A locative element can optionally co-occur with the demonstrative. This element has the function of reinforcing the deictic value of the demonstrative itself. Demonstrative and locative establish a predication relation forming a constituent, and the complex sequence demonstrative + locative (for example, este de aquí ‘this of here’) or the locative alone (for example, de aquí ‘of here’) appear at PF in the same position occupied by the postnominal demonstrative.

POSITIONS OF DEMONSTRATIVES

43

D. A noun modified by a demonstrative behaves like a referential nominal expression, and a noun modified by a postnominal demonstrative is subject to the same referential interpretation that characterizes the noun when the demonstrative appears in [Spec, DP] at PF. Therefore, I propose that the demonstrative is specified for the features [+Referential] and [+Deictic]. Assuming that the referential interpretation takes place inside the DP (see Longobardi 1994), I also propose that the demonstrative must check its [+Referential] feature in [Spec, DP] through Spec-Head Agreement. These requirements lead me to claim that the demonstrative is generated in the [Spec, FP] of the functional projection whose position in the structure has been determined in points 1 and 2 and that the demonstrative must move from [Spec, FP] to [Spec, DP] for checking reasons. Therefore, the [Spec, DP] position is a derived position for the demonstrative. Since in Spanish the demonstrative displays the option of appearing either in its base position, (76a) and (77a), or in [Spec, DP] at PF, (76b) and (77b), I propose that the movement of the demonstrative can apply optionally before Spell-Out but obligatorily at Logical Form. If the demonstrative does not move to [Spec, DP] before Spell-Out, (76b) and (77b), the definite article must be realized in D° in order to show also at PF that this position contains some particular feature (that is, the [+REF] feature), which prevents it from being interpreted as existential. Finally, I have tried to extend my hypothesis for Spanish to other languages. I have examined the syntactic behavior of the demonstrative in three different groups of languages not necessarily related one with the other: (1) languages where the demonstrative can be realized in prenominal and in postnominal position: Catalan, Bosnian, Russian, Romanian, Modern Greek; (2) languages where the demonstrative has to appear in prenominal position at PF: Italian, French, German, Albanian; and (3) languages where the demonstrative has to be realized in postnominal position at PF: Hebrew, Irish (Celtic). I have shown that in all such groups either the demonstrative, or the complex sequence demonstrative + locative, or the locative reinforcer alone occupies the same position that the postnominal demonstrative or the complex sequence este de aquí ‘this of here’ or the locative reinforcer de aquí ‘of here’ occupies in Spanish. Therefore, since we have assumed that the demonstrative has to check its [+Referential] feature in [Spec, DP] in all languages, I have proposed that cross-linguistically the demonstrative is generated in the same position as found for Spanish: that is, [Spec, FP]. I have thus proposed the existence of a parameterized principle to account for the different positions the demonstrative can or must occupy at PF across languages. This principle is based on the weak/strong properties (in Chomsky’s 1995 terms) of the [+Referential] feature that is peculiar to the demonstrative itself. In those languages in which the demonstrative obligatorily raises to [Spec, DP] before Spell-Out, the [+REF] feature of the demonstrative is strong. In these cases, the base position from which the demonstrative strarts can be recovered by the optional realization of the locative reinforcer. Alternatively, in those languages in which the demonstrative does not move before Spell-Out, the same [+REF] feature of the

44

DP FUNCTIONAL STRUCTURE

demonstrative is weak. In these cases, the movement has to take place only at Logical Form. Finally, in those languages in which the demonstrative can optionally raise from its base position to [Spec, DP] before Spell-Out, as in Spanish, the [+REF] feature of the demonstrative has both the weak and strong properties. Notes This chapter is a short version of chapter 3 of my doctoral dissertation, “Aspetti della struttura dei nominali nelle lingue romanze” (1997), which appeared in published form as Categorie funzionali del nome nelle lingue romanze (2000). The central hypothesis was presented at the XXIIo Incontro di Grammatica Generativa (University of Bergamo, February 1996), at the Sexto Coloquio de Gramática Generativa (Valencia, March 1996), and at the 19th GLOW Colloquium (Athens, April 1996) in a paper presented in collaboration with Giuliana Giusti. I acknowledge the audiences of those conferences for helpful comments, and in particular I wish to thank Anna Cardinaletti, Guglielmo Cinque, and Giuliana Giusti for their constructive criticism. I am also grateful to Ignacio Bosque and Luis Sáez for their valid suggestions. None of these people should be held responsible, however, for any mistakes or misunderstandings herein. 1. In Spanish, nominals modified by a postnominal demonstrative are commonly used in colloquial speech. It seems that when the demonstrative appears in this position, the entire nominal expression receives a depreciatory meaning. This characteristic does not seem to hold true in all cases and for all native speakers, however. In this chapter I will leave open the question concerning this variation in the connotative meaning, assuming that it affects the pragmatic ground of the language. 2. In fact, the data in (2) show that there is no violation of the Head Movement Constraint, the principle that forbids a head from raising past another head (see Travis 1984; Baker 1988). 3. By these assumptions, we do not consider available a right-adjunction hypothesis such as the one proposed in (i) for the Spanish data in which the demonstrative follows the noun, (1c–d): (i)

*DP D' XP

D˚ el X'

Spec

..... NP

este

libro In a previous theoretical framework, a similar symmetric structure was proposed by Ernst (1992) for Irish demonstratives (see note 42 herein). As I will show, some theoretical arguments contribute to my rejecting this alternative. The most important one is that, since in Spanish and Irish the postnominal demonstrative always precedes the PP subject and all

POSITIONS OF DEMONSTRATIVES

45

the other PPs complements of the noun, we would be obliged to adjoin them in a very high position in the structure outside the NP, contrary to what is currently assumed. 4. In an antisymmetric framework, a competing structure for (3) has been defended by Bernstein (1997), who proposes the following derivation for the prenominal demonstrative in Romance: (i)

DP Demi

FP XPj [N+A]j

FP Spec. ti F˚ (deictic reinf.)

F' XPj

tj

In (i) the entire NP (XP) raises past the functional base projection of the demonstrative. We do not adopt Bernstein’s analysis because it would erroneously rule out (ii.a–b) and would erroneously predict the grammaticality of (ii.c–d): (ii)

a. la respuesta esta al problema the answer this to the problem ‘this answer to the problem’ b. la respuesta esta suya al problema the answer this his/her to the problem ‘this answer to the problem of his/hers’ c. *la respuesta al problema esta the answer to the problem this d. *La respuesta suya esta al problema the answer his/her this to the problem

In fact (i) cannot obtain in an antisymmetric way a constituent including the noun and the adjective or adjectives and excluding the subject of the noun, its complements (ii.a) and the postnominal possessive (ii.b), which in Spanish and other languages must follow the postnominal demonstrative. I will discuss these data in the next section. 5. An argument in favor of this hypothesis is given by the behavior of adjectives such as principal ‘main’, which can never be predicated: (i) *Este hecho es principal. this fact is main When this adjective occurs with a postnominal demonstrative, the demonstrative must follow the adjective, (ii.a). The alternative construction in which the postnominal demonstrative precedes this adjective, (ii.b), is always excluded, even with a sharp intonational break between the two elements, contrary to what happens in (8b, d).

46

DP FUNCTIONAL STRUCTURE

(ii)

a. Las dos razones principales esas no son suficientes para disculparle. the two reasons main those not are sufficient to excuse him/her b. *Las dos razones esas principales no son suficientes para disculparle. the two reasons these main not are sufficient to excuse him/her ‘Those two main reasons are insufficient to excuse him/her.’

6. In (10), as in (13), the prenominal possessive adjective appears in the specifier of a functional projection immediately dominated by DP, according to the syntactic behavior this element shows in languages such as Italian (la sua strana risposta [lit.: ‘the his/her strange answer’]) (see also Longobardi 1995). As for Spanish, we assume, adopting Picallo’s (1994) analysis, that this position is not the final landing site to which a genitive raises in Syntax. In Spanish, this category further moves to D°. 7. Notice that the sentences in (16) are not excluded because definiteness triggers opacity effects. In fact, in Spanish wh-extraction out of definite nominals is possible when the genitive PP is a theme, as (i) shows: (i) ¿De quién / de qué pintor han publicado la foto? of whom / of which painter have-they published the picture ‘Who/which painter have they published the picture of?’ 8. The generalization in (20) adopts a Larsonian structure for the internal NP projection (see Larson 1988): namely, a structure that creates more specifier positions inside the NPshell in order to host the possessor PP in the higher one and the Agent PP in the lower one (see (19c–e)). This hypothesis seems to us more plausible than the one defended by Picallo (1994), who proposes that the possessor PP occupies the specifier of a functional projection external to the NP and immediately superior to it. For further considerations to support this assumption, see section 5 in the text. 9. The idea that thematic adjectives occupy a position higher than [Spec, NP] seems to be confirmed also by Bulgarian, where the agent and the possessor can be realized by adjectival forms. See Dimitrova-Vulchanova and Giusti (1999). 10. To give thematic adjectives a derivation similar to the one proposed for possessive adjectives, we could tentatively suggest that they reach the specifier of the functional projection we singled out through XP-movement. Nevertheless, we leave the question about their derivation open here. 11. In this chapter I tentatively assume that nouns such as libro ‘book’, cuadro ‘painting’ (19c), and so on exclusively belong to the class of object-denoting nominals, even though they can assign an agentive q-role. In this way, the NP-shell in (23) can be justified. 12. In the text I use the sequence este de aquí ‘this of here’ as an illustration. The same combination demonstrative + locative is also possible with the other forms of demonstratives and locatives. So, we can have este de acá ‘this of here’; ese de acá ‘that of there’; ese de allí ‘that of there’; ese de allá ‘that of there’; aquel de allí ‘that of there’; aquel de allá ‘that of there’. Nevertheless, it is impossible to have in the same sequence a demonstrative form that, for example, expresses proximity to the speaker with a locative that expresses distance from the speaker, or vice versa, as in *este de allí/allá ‘this of there’ or *aquel de aquí ‘that of here’. A possible account for the impossibility of such combinations will be suggested later in the text. 13. In Spanish we can have cases in which only the locative can appear, as in (i): (i) El libro de aquí me gusta the book of here to-me(CLIT) likes ‘I like the book here.’

POSITIONS OF DEMONSTRATIVES

47

Such constructions can receive the interpretation of este libro (de aquì) me gusta ‘I like this book (of here)’ even if the demonstrative does not appear lexically realized. For cases like these, we propose that an abstract movement of the demonstrative to [Spec, DP] is involved at LF. An argument that can support this proposal is that, if in the same constructions a prenominal possessive is realized, the interpretation of the sentence with a covert demonstrative is impossible, as many speakers pointed out to me (for the incompatibility between the prenominal possessive and the demonstrative, see section 5): (ii)

su libro de aquí me gusta. his book of here to-me(CLIT) likes = *this his book of here to-me(CLIT) likes ‘I like his book here.’

In (ii) the locative cannot be interpreted as a deictic reinforcer of the demonstrative. Rather, it is interpreted as an adverbial PP “complement” of the noun, receiving the following reading: ‘his book which is here / which was published here’. This reading can be assigned also to sentences like (i), and it becomes more evident in those cases in which the noun is plural, given that it can receive a generic reading, or is a collective noun. This ambiguity in the interpretation, which we can observe in cases like (i), depends on the fact that in Spanish the locative must be preceded by the preposition de ‘of’ when it appears as reinforcer of the demonstrative. In fact, in languages such as Italian, where the locative reinforcer appears without a preposition –(for example, questo N qui ‘this N here’; see section 6.1.2.), the same ambiguity in interpretation is excluded, as the contrasts in (iii) show: (iii)

a. queste cartoline qui sono brutte these postcards here are ugly b. *le cartoline qui . . . c. le cartoline di qui . . . the postcards of here / of these places

In Italian, contrary to Spanish, an abstract movement of the demonstrative to [Spec, DP] at LF is excluded, (iii.b), probably because of the different properties the definite article has in the two languages. 14. In Spanish the deictic reinforcer de aquí cannot appear with the class of event nominals: (i) *Esta respuesta de aquí preocupó a todos. this answer of here worried everybody ‘This answer here worried everybody.’ Sáez (personal communication) points out that in (i) the presence of the locative is impossible because the locative cannot be interpreted without a real spatial content that must be restricted to the communicative context. We have the same results when the noun is an abtract noun, (ii): (ii)

a. *Estos modales de aquí these manners of here b. *Esta vez de aquì this time of here

As for (ii.a), we have to say that it is acceptable only if the locative is interpreted as an adverbial PP (see note 13) and not as deictic reinforcer: namely, if the construction receives the following interpretation: ‘these manners of these places’. Therefore we conclude that the ungrammaticality of (i) and (ii) is due to the incompatibility of the characteristics of the head noun with the intrinsic properties of the locative in Spanish. However, these properties can

48

DP FUNCTIONAL STRUCTURE

vary cross-linguistically. In fact, in Italian, where the locative can “extend” its spatial value, sentences such as (i) and (ii) are grammatical with the interpretation of the locative as deictic reinforcer, as we will show later. 15. Notice that when in some Italian dialects the sequence demonstrative + locative appears in a predicative construction, the locative is obligatorily preceded by the complementizer che ‘that’. This happens in the dialect of Montale (Florence), in which we have ‘kwesto ‘ke ‘kwi ‘this that here’, and in the dialect of Parma, in which we have kost ke ‘ki ‘this that here’ (see Poletto 1997). 16. The structure in (30) is a reduced and slightly simplified version of the structure that Kayne proposed during his Venice lectures (spring, 1995) to account for constructions such as qualcuno di interessante ‘someone of interesting’, and which appears also in Kayne (1994, pp. 106– 110), to account for constructions such as quelqu’un de célèbre ‘somone of famous’ in French. 17. In (30) we assign to the locative aquí ‘here’ the categorial status of a PP. This choice has been made according to Larson (1988), who proposes for Bare Adverbs either the status of NPs, since they can function as subjects (e.g., in copular sentences), or the status of PPs with an abstract preposition. 18. One could object that a structure such as (30) violates the Left Branch Condition. Even though in this case a violation indeed takes place, it is necessary to specify that its statement should be in any case reconsidered in the light of the Kayne’s antisymmetric approach. In fact, constructions such as (i), where the small clause is followed by another complement -i.e. the PP da tutti ‘by everybody’-, also violates the Left Branch Condition. (i)

XP

XP

Giannii era ritenuto ‘Giannii was considered

[ti [ti

inadatto] unable]

da tutti. by everybody

I will not say anything about this question here, leaving the topic open. 19. The postnominal position of the possessive does not necessarily imply a contrastive interpretation, neither in Spanish nor, for example, in Italian. This reading depends on the contexts in which the construction is used. However, it is obvious that, with the possessive having the possibility of also appearing prenominally, the postnominal position preferably receives this contrastive interpretation. In fact, the same interpretation is impossible to obtain in Spanish in the other available position, for the reasons just expressed in the text. It is possible that one associates the contrastive interpretation with a “marginalized” structural position of the postnominal possessive: namely, a position that does not correspond to the base position we can assume this element occupies in the cases that will be presented in the text. Nevertheless, we can assume that a contrastive interpretation is irrelevant to determine the position the postnominal possessive occupies in the structure. In fact, this interpretation can be obtained also in the base position and not necessarily in a “marginalized” position, as the word order of the examples in (31) through (33) shows and as, in other contexts, the use of a strong form instead of its clitic counterpart displays: for instance, vi a él ‘I-saw him’ vs. lo vi [lit.: him(CLIT) I-saw].

POSITIONS OF DEMONSTRATIVES

49

20. As for the relative order between the postnominal possessive and adjectives inside a nominal projection with an object-denoting noun, I refer the reader to the discussion in Brugè (1997, 2000). 21. This proposal is compatible with the Case Checking Principle proposed by Longobardi (1995). Its formalization, together with all its possible combinations, is presented on pp.15–16 of his paper. Because the author does not consider the case of the postnominal possessive, we can extend his hypothesis to it by suggesting that, besides the AgrGP and the base position inside the NP, there exists an intermediate functional projection inside the extended nominal projection in which genitive case is assigned. 22. Following Longobardi’s (1994) discussion on the semantic properties of proper names and pronouns (see pp.633–640 and the references cited there), we can see that a noun modified by a demonstrative also has the same characteristics. In fact, a noun modified by a demonstrative is never ambiguous between a de re/de dicto reading, contrary to what happens with definite descriptions, but it has always a de re reading, just like proper names and pronouns. Moreover, it has a rigid designation (see Kripke’s 1980 terminology) in the sense that, as Longobardi affirms for proper names, “[it appears] to designate the same object throughout all possible worlds (i.e., also in counterfactual situations)” (1994, 639). 23. For the incompatibility of partitive case with nonexistential nominal expressions, see the argumentations presented in Brugè and Brugger (1996). 24. Notice that if the complement of the existential quantifier is realized as a “definite partitive” PP (see Cardinaletti and Giusti’s 1993 terminology), the presence of the demonstrative is possible either in prenominal or in postnominal position, as (i) shows: (i)

a. Algunos de estos libros tuvieron poco éxito. some of these books had little success b. Algunos de los libros estos tuvieron poco éxito. some of the books these had little success

25. Brucart (1994 (9) p. 54) observes that in Spanish there are cases in which the demonstrative and the prenominal possessive can co-occur in the structure–estos mis hijos ‘these my(PL) sons’–just as in Catalan and Italian. Nevertheless, he suggests that this possibility is very limited and defective in the common use (compare the ungrammaticality of *todos estos mis hijos ‘all these my(PLU) sons’ and the grammaticality of todos estos hijos míos ‘all these sons my( PLU)(strong), and proposes that these cases must be considered residues of Medieval Spanish, when this co-occurrence was common. In fact, in Medieval Spanish definite article and prenominal possessive were not in complementary distribution (see Lapesa 1980, among others). 26. The hypothesis that in Spanish the prenominal possessive provides referential interpretation for the nominal expression it modifies is justified by the following syntactic behaviors: first, when a nominal modified by a prenominal possessive is further modified by a relative clause, the relative clause can only receive the appositive reading, as (i) shows: (i) Mi libro, que publiqué el año pasado, tuvo mucho éxito. my book that I-published the year last had great success ‘My book that I published last year had a great success.’ Second, a noun modified by a prenominal possessive cannot be further modified by a relative clause in subjunctive, as (ii) shows: (ii)

*Sólo irán a Barcelona sus estudiantes que hayan superado los parciales. only they-will-go to Barcelona his students who have(SUBJ.) got through the partial exams ‘Only his students who have got through the partial exams will go to Barcelona.’

50

DP FUNCTIONAL STRUCTURE

Third, a noun modified by a prenominal possessive can never appear in the domain of existential quantifiers, as (iii) shows: (iii)

*Muchos/algunos/tres mis libros de sintaxis fueron publicados hace años.

many/some/three my books on syntax were published ago years ‘Many/some/three books on syntax of mine were published years ago.’ The same behaviors do not affect the postnominal possessive, which we propose does not raise to [Spec, DP] at LF to check the [+REF] feature (see (34)). For a more detailed discussion of these cases, see Brugè (1997, 2000). 27. Notice that besides the main values the demonstrative can have in the language and that are described in (37), this category can also be used as a discourse anaphoric element, as in (i), or it can function as a “deictic am phantasma” (see Bühler 1934), as in (ii): (i)

(ii)

a. Juan es guapo, inteligente y simpático. Sin embargo, estas cualidades no hacen de él el hombre perfecto. ‘Juan is handsome, intelligent, and likeable. Nevertheless, these qualities do not make him the perfect man.’ b. Juan es guapo, inteligente y simpático. Sin embargo, las cualidades estas . . . ‘Juan is handsome, intelligent, and likeable. Nevertheless, the qualities these . . .’ a. Aquí me imagino estas casitas de estilo alpino . . . Here me(CLIT) I-imagine these small houses of Alpine style b. Aquí me imagino las casitas estas de estilo alpino . . . Here me(CLIT) I-imagine the small houses these of Alpine style ‘Here I imagine these small Alpine style houses . . .’

I propose that in cases such as these the demonstrative maintains its specifications for the feature [+Deictic], but loses its [+REF] value. Nevertheless, because in these cases the demonstrative displays the same main syntactic behaviors observed in the text until now, I consider it plausible to extend to them the movement hypothesis (see(41)). In these cases the demonstrative will check in [Spec, DP] some other particular feature, specific to the anaphoric interpretation and/or to the “deictic am phantasma” interpretation (see Brugè 1997, 2000). 28. I thank Ned0ad Leko for his helpful comments on the Bosnian data. 29. For independent reasons, in Bosnian, adjectives appear in prenominal position, apart from some marked cases. 30. Recall that in Bosnian the sequence demonstrative + locative can raise jointly to [Spec, DP] before Spell-Out. See example (27a). 31. As Michael Yadroff pointed out to me, we can extend our hypothesis also to Russian (see Brugè 1997, 2000). 32. In Romanian the demonstrative in prenominal position belongs to a very high stylistic level. Stylistically high, even if in a lower rank, is also the use of the form acesta ‘this’ in postnominal position. In the current language, the demonstrative appears in postnominal position and in its reduced forms, as the cases in (i) show: (i)

a. b£iatul £sta/£la boy(MASC.)-the(MASC.)this(MASC.)/that(MASC.) b. cartea £sta/£ia book(FEM.)-the(FEM.)this(FEM.)/that(FEM.)

33. For the value the invariable bound morpheme -a (which obligatorily appears on the demonstrative in cases such as (56b)) can have, see Brugè (1997, 2000). See also TasmowskyDe Ryck (1990).

POSITIONS OF DEMONSTRATIVES

51

34. I suggest that the movement of the demonstrative to the intermediate position is optional before Spell-Out for those languages such as Bosnian and Modern Greek which, as we have seen, also display this option. 35. As in Spanish, these constructions are commonly used in colloquial speech. 36. In French, the locative reinforcer is realized through the clitic form -ci ‘here’ instead of the unbound adverbial form ici ‘here’, (63). Coming back to the complex structure proposed in (30), I tentatively suggest that in French the “bare adverb” ici undergoes a cliticization process that turns it into a clitic element. For this reason it must cliticize to the lexical element that immediately dominates it in the structure–for example, the noun moved to higher position or an adjective. Probably, the same proposal could be extended to the other locative form là, even though in this case there is no morphological difference between the unbound adverbial form and its “clitic” counterpart. 37. Notice that in French the locative can also be absent: ce livre. In this case, the demonstrative ce receives the meaning of ‘this’ by default. The same is not true in some Lombard dialects of Switzerland, where a unique demonstrative form (quello ‘this/that’) exists. In fact, in these dialects the neutralization of the different types of deixis makes the presence of the locatives qui ‘here’ and lì/là ‘there’ obligatory. The locative forms always appear in their base position: quel libro *(qui) ‘this book’; quel libro *(lì/là) ‘that book’. Thanks to Cecilia Poletto for these data. 38. For some native speakers (65b) is acceptable even if the locative reinforcer follows the genitive PP. Nevertheless, in this case, the unique possible reading the construction can receive is with a strong contrastive stress on the entire DP ce livre de Jean ‘this book of John’, giving as resulting interpretation something like: ‘this book of John here, and not the other book(s) of John’. However, in (65a), excluding the possible intonational break between the locative and the genitive PP, no element receives a contrastive stress. Rather, the entire construction can be new information. 39. Thanks to Dalina Kallulli for her helpful comments on the Albanian data. 40. I am grateful to Ur Shlonsky for the data in Hebrew. 41. Also in construct state constructions the demonstrative occupies the same position with respect to adjectives and complements of the noun, as (i) shows: (i) beyt ha-more (*ha-ze) AP (ha-ze) PP (*ha-ze) home the teacher (*this) AP (this) PP (*this) 42. The Irish data come from Ernst (1992) and Ball (1993). Ernst, studying the syntactic behavior of the demonstrative in Irish, excludes any movement process and suggests that the demonstrative is located in a right branch adjoined to NP and immediately dominated by the genitive phrase, also right-adjoined to NP. As for why I do not adopt Ernst’s hypothesis, see note 3 herein. 43. Recall that this requirement agrees with what is assumed in Longobardi (1994). 44. From a theoretical point of view, it would be better to avoid resorting to both (75a) and (75b) in order to derive the optionality that these languages display. Nevertheless, we know that cases of redundancy exist in language and that the UG should be able to accommodate them.

References Baker, M. C. (1988) Incorporation: A Theory of Grammatical Function Changing. Chicago: University of Chicago Press. Ball, M. J. (ed.) (1993) The Celtic Languages. London: Routledge. Bernstein, J. (1997) “Demonstratives and Reinforcers in Romance and Germanic Languages.” Lingua 102, 87–113.

52

DP FUNCTIONAL STRUCTURE

Brucart, J. M. (1994) “Incompatibilidad entre posesivos y relativas especificativas,” in V. Demonte (ed.) Gramática del español, 51–86. Mexico City: El Colegio de México. Brugè, L. (1997) “Aspetti della struttura dei nominali nelle lingue romanze,” Ph.D. Diss., Università degli Studi di Padova. Brugè, L. (2000) Categorie funzionali del nome nelle lingue romanze. Milan: Cisalpino. Brugè, L., and G. Brugger (1996) “On the Accusative a in Spanish.” Probus 8, 1–51. Brugè, L., and G. Giusti (1996) “On Demonstratives.” Paper presented at the 19th GLOW Colloquium, April 17–18, Athens. Bühler, K. (1934) Sprachtheorie. Stuttgart: Gustav Fisher. Campbell, R. (1993a) “The Occupants of SpecDP.” GLOW Newsletter 30, 62–63. Campbell, R. (1993b) “Specificity Operators in DP.” Unpublished ms., Oakland University. Cardinaletti, A., and Giusti G. (1993) “Partitive ne and the QP-Hypothesis,” in E. Fava (ed.) Proceedings of the 17th Meeting of Generative Grammar, 121–141. Turin: Rosenberg and Sellier. Chomsky, N. (1995) The Minimalist Program. Cambridge, Mass.: MIT Press. Cinque, G. (1980) “On Extraction from NP in Italian.” Journal of Italian Linguistics 5, 47–99. Cinque, G. (1981) “Sulla nozione di soggetto di SN in Italiano.” Cultura Neolatina 41, 555– 570. Cinque, G. (1994) “On the Evidence for Partial N-movement in the Romance DP,” in G. Cinque, J. Koster, J.-Y. Pollock, L. Rizzi, and R. Zanuttini (eds.), Paths towards Universal Grammar, 85–110. Washington, D.C.: Georgetown University Press. Cinque, G. (1995) “Posizione del soggetto nel DP italiano,” in C. Lupu and L. Renzi (eds.) Studi Rumeni e Romanzi: Omaggio a Florica Dimitrescu e Alexandru Niculescu, 436– 463. Padua: Unipress. Dimitrova-Vulchanova, M., and G. Giusti (1998) “Fragments of Balkan Nominal Structure,” in A. Alexiadou and C. Wilder (eds.) Possessors, Predicates and Movement in the Determiner Phrase, 330–360. Amsterdam: John Benjamins. Dimitrova-Vulchanova, M., and G. Giusti (1999) “Possessors in the Bulgarian DP,” in M. Dimitrova-Vulchanova and L. Hellan (eds.) Topics in South Slavic Syntax and Semantics, 163–192. Amsterdam: John Benjamins. Ernst, T. (1992) “Phrase Structure and Directionality in Irish.” Linguistics 28, 415–443. Giusti, G. (1993) La sintassi dei determinanti. Padua: Unipress. Giusti, G. (1997) “The Categorial Status of Determiners,” in L. Haegeman (ed.) The New Comparative Syntax, 95–123. London: Longman. Grimshaw, J. (1991) “Extended Projection.” Unpublished ms., Brandeis University, Waltham, Mass. Kaplan, D. (1978) “Dthat,” in P. Cole (ed.) Pragmatics, 221–243 (Syntax and Semantics 9). London: Academic Press. Kayne, R. S. (1994) The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Kripke, S. (1980) “Preface” to Naming and Necessity. Cambridge, Mass.: Harvard University Press. Lapesa, R. (1980) Historia de la lengua española. Madrid: Gredos. Larson, R. (1988) “On the Double Object Construction.” Linguistic Inquiry 19, 335–391. Longobardi, G. (1994) “Reference and Proper Names: A Theory of N-Movement in Syntax and Logical Form.” Linguistic Inquiry 25, 609–665. Longobardi, G. (1995) “On the Typological Unity of Indoeuropean and Semitic Structural Genitive.” Unpublished ms., Università di Venezia. Picallo, C. (1992) “Catalan Possessive Pronouns: The Avoid Pronoun Principle.” Unpublished ms., Universitat Autónoma de Barcelona.

POSITIONS OF DEMONSTRATIVES

53

Picallo, C. (1994) “Catalan Possessive Pronouns: The Avoid Pronoun Principle Revisited.” Natural Language and Linguistic Theory 12, 259–299. Poletto, C. (1997) “Pronominal Syntax,” in M. Maiden and M. Parry (eds.) The Dialects of Italy, 137–144. London: Routledge. Rizzi, L. (1990) Relativized Minimality. Cambridge, Mass.: MIT Press. Siloni, T. (1994) “Noun Phrases and Nominalizations,” Ph.D. Diss., Université de Gèneve. Sportiche, D. (1988) “A Theory of Floating Quantifiers and Its Corollaries for Constituent Structure.” Linguistic Inquiry 19, 425–449. Tasmowski-De Ryck, L. (1990) “Les démonstratifs français et roumains dans la phrase et dans le texte.” Langages 97, 82–99. Travis, L. (1984) “Parameters and Effects of Word Order Variation,” Ph.D. Diss., MIT.

54

DP FUNCTIONAL STRUCTURE

3

The Functional Structure of Noun Phrases A Bare Phrase Structure Approach  

The goal of this chapter is to give a general overview of recent studies on the syntax of determiners, particularly in Romance, Germanic, and Balkan languages, which can shed some light on general theoretical issues. In so doing, I will reconsider some proposals of mine, as well as of other linguists in light of the recent framework of bare phrase structure as developed by Chomsky (1993, 1995, 1999). The essay is organized in three sections. In section 1, I present some general assumptions and some general hypotheses that will be motivated in the course of the chapter. Section 2 deals with articles. Section 3 deals with demonstratives, as well as other elements that may function as referential operators inside the noun phrase.

1. Basic assumptions and hypotheses I adopt a number of assumptions from the literature and add four more assumptions that I have proposed in previous work of mine and in collaboration with Mila Dimitrova-Vulchanova. The general framework is provided by Grimshaw’s (1991) theory of functional projections, which considers functional heads to be projected by the lexical head in a bottom-up fashion, without labels.1 In this particular regard, Grimshaw’s proposal is perfectly in line with Chomsky’s (1993, 1995) bare phrase structure approach. Notice, furthermore, that all the structures resulting from this work comply with Kayne’s (1994) antisymmetric hypothesis. 54

A BARE PHRASE STRUCTURE APPROACH

55

The proposals developed here make use of the following minimal assumptions to account for a wide range of syntactic phenomena that arise in relation to so-called determiners in a number of languages, including Romance, Germanic, and Balkan: (A1) The realization of a functional head is a last resort procedure. (A2) If a functional head is realized, then it is either a dependent morpheme or a weak2 (free) morpheme. (A3) All the functional heads of an extended nominal projection share the same f-features. (A4) The interpretation of a noun phrase at LF is done in its highest Specifier position (generally referred to as SpecDP, here referred to as SpecFPmax). Assumption (A1) captures the empirical observation that we certainly do not find as many overt functional heads as are generally assumed to be projected in the structure. We will see in section 3 that this assumption can be turned into the parameterized “Principle of Economy of Lexical Insertion” (see principle (43)) to account for a series of phenomena that can be described as the “doubly-filled XP filter.” Assumption (A2) describes a group of properties that have often been noticed for functional categories and which can be summarized as in (1), adapted from Abney (1987): (1) General properties of functional heads a. They constitute closed lexical classes. b. They are generally phonologically and morphologically dependent. c. They can be sisters only to one kind of category. d. They are usually inseparable from their sister projection. e. They lack substantive content.

None of the properties in (1) is necessary or sufficient to attribute functional status to a morpheme. Altogether they show a strong tendency, which is captured in (A2). The assumption in (A2) states a strict correlation between semantics and its morphosyntactic realization: A functional category is semantically “weak” in the sense that it only carries features such as number, gender, definiteness, and case, which are shared by all other elements of the same class. As a consequence, a functional category is also morphosyntactically “weak.” Given (A1), we expect that if and only if the conditions for the licensing of a covert functional head are not met, its realization will be morphologically minimal, either by an inflectional morpheme or by a free morpheme devoid of lexical content. Assumption (A3) captures the fact that in the largest majority of cases, functional heads in the nominal structure appear to trigger, on the modifiers of the noun, agreement for all the f-features present on the head noun (including gender, number, and case). If there were no sharing of such features, we would expect to find some modifiers agreeing for number only, some other for gender only, and other still for case only. Furthermore, if the functional structure were made of different functional projections, each one specified for a single different feature, we would expect the hierarchy on the adjectives to be mirrored in different agreement specifications on the single adjectives. But this is never the case in the empirical domain studied

56

DP FUNCTIONAL STRUCTURE

here. From this observation, I conclude that in natural languages, functional heads that trigger agreement of modifiers with these f-features are copies of the f-features on the head noun. Since these features are visible elsewhere, it is often the case that they do not need to be overt. Assumptions (A1) through (A3) are often implicit in the literature. Here they will be kept in the most general formulation. No difference will be made between Agr heads and other functional heads, including D. All of them will be labeled as F, a welcome result in a framework that aims to economy.3 Assumption (A4) is necessary if D (from now on Fmax) is maintained as devoid of substantive content, at the same time capturing the well-known facts that have led a large tradition of linguistic research to attribute semantic content to the article. The underlying idea is that the apparent effects of the referential properties of the article can be derived by the assumption, due to Campbell (1996), of an empty operator that functions like a demonstrative and is in the same position as a demonstrative when the definite article is in Fmax. The previous assumptions will allow the formulation of two very general hypotheses on the categorial status of determiners: (H1) Among determiners, only articles are functional heads (and appear in Fmax). (H2) Demonstratives as well as other maximal projections carrying referential features can/must check their referential features in SpecFPmax. Hypothesis (H1) derives from assumptions (A1) and (A2) generally extended to all functional categories in the noun phrase, including articles. Such an extension is welcome from the theoretical point of view in that it does not need to make any distinction among nominal functional heads; moreover, it can capture the empirical observation that articles display a very different syntactic behavior across languages. In particular, articles may be missing or, when present, may be inflectional morphemes (enclitic) or can be considered as dummies (proclitic or free morphemes with no semantic value). In section 2, I show that a definite article in some languages is inserted on syntactic grounds, regardless of the referential properties of the noun phrase. Cross-linguistic variation is found in cases in which insertion of the article (a last resort procedure) is necessary for a given language. According to (H2), demonstratives behave like modifiers of a particular kind. Differently than articles, they provide semantic referential features to the noun phrase. Notably, they are not the only elements that can contribute referential features to the noun phrase; a well-known example of this kind is the prenominal possessive in English. In section 3, I show that demonstratives are XPs across languages; they are not directly merged in FPmax but are merged lower in the structure and further moved to SpecFPmax to check their referential features. This property is shared by other modifiers of the noun, such as possessive adjectives, personal pronouns, and proper names, among others. Cross-linguistic variation is found in the stage of the derivation in which the referential modifier (whether a demonstrative or another element) moves to SpecFPmax.4 In the course of this essay, I argue that the term “determiner” is spurious. It is a jargon term to refer to elements that are often, but not always, found at the leftmost

A BARE PHRASE STRUCTURE APPROACH

57

position in the noun phrase, apparently in complementary distribution with one another in well-studied languages. In languages such as Modern Greek, Romanian, Hebrew, and Welsh (cf. Brugè and Giusti 1996; Brugè chapter 2 in this volume), articles and demonstratives are not (or not always) in complementary distribution, and the demonstrative is not (or not always) the leftmost element in the noun phrase. Elsewhere I have also argued against the unification of quantifiers with other determiners. In particular, I have argued that quantifiers are never in the position where determiners can be found. They are either lexical heads merged above FPmax or adjectival phrases in a functional specifier of the noun phrase. For a syntactic account of quantifiers in Germanic and Romance, see Giusti (1991a, 1991b, 1991c, 1993, 1997); in Bulgarian, see DimitrovaVulchanova and Giusti (1996); in Bosnian, see Giusti and Leko (1995). For reasons of space, I will limit my concern here to the study of FPmax. 2. 2.1.

Articles Phonologically dependent

In the languages displaying a morphologically free article, such as Romance (except Romanian) and Germanic (except Scandinavian), the form of this element is phonologically dependent on the context inside the noun phrase. This dependency is not displayed with respect to elements that are external to the noun phrase. I take this as evidence for the functional status of the article in the extended nominal projection, assuming that a functional head entertains a privileged relationship with an immediately higher functional head (which necessarily belongs to the same extended projection) and with the immediately lower specifier. The definite article in Italian is never enclitic on any element that is not part of the noun phrase. It is either proclitic on the following nominal element (2b) or enclitic on a preposition (3b): (2)

a. Mangi[a lo] scorfano. cf. *Mangi[allo] scorfano. ‘[(S)he] eats the scorpion fish.’ b. Mangia l’arrosto cf. *Mangia il/lo arrosto. ‘[(S)he] eats the roast [beef].’

(3)

a. Ha parlato [a lo]redana. ‘[(S)he] talked to Loredana.’ b. Ha parlato [allo] scolaro. ‘[(S)he] talked to the pupil.’

cf. *Ha parlato[allo]redana cf. *Ha parlato[alo] scolaro

In the Central Italian variety of the Marche (Ancona), “Raddoppiamento Sintattico”5 is very limited if there is any at all.6 In particular, no reduplication of the following consonant is found between the verb mangia and its object in (2a) and (2b) or between the preposition a and the following proper name in (3a). However, when the definite article is preceded by a preposition, the resulting form has a geminate consonant, suggesting that something different from prosodic rules has applied. Let’s assume that in (2b) and (3b) we have a case of head incorporation. In (2a) the article (a nominal functional head) cannot incorporate on the verb. In (3a) the proper name cannot incorporate onto the preposition because it is not a functional head, as I argue in section 3.4.

58

DP FUNCTIONAL STRUCTURE

This suggests that the definite article is a functional head in the extended projection of the noun phrase. It is the highest, unless a monosyllabic preposition is present.7 Although this is not the place to make a point on the functional versus lexical status of prepositions in Italian, it is plausible to assume that a subset of the monosyllabic prepositions in Italian have the function of case markers, as also argued for by Cardinaletti and Starke (1999). This function can also be claimed for the definite article.8 It is conceivable that the articulated preposition is not formed by incorporation of the article into a higher P, thus violating the Mirror Principle, but it is a unique element realizing case, as well as all other f-features of the noun in Fmax. 2.2.

Morphologically dependent

The strongest evidence in favor of the morphological dependency of the article is that in some languages the article is enclitic. This is found in some Balkan languages such as Romanian, Albanian and Bulgarian, and in all Scandinavian languages represented here by Norwegian:9 (4)

a. b. c. d.

b£iatul djali momceto gutten boy-the

(Romanian) (Albanian) (Bulgarian) (Norwegian)

The first studies on this topic (see Dobrovie-Sorin 1987 and Grosu 1988 for Romanian; Hellan 1985 and Taraldsen 1990 for Scandinavian) have taken for granted that the article is inserted in D and the N moves to D to obtain encliticization. A hypothesis of N-to-D movement at Spell-Out for all cases in (4) predicts that, in these languages, the presence of the enclitic article implies that the head noun is found at the leftmost side of the noun phrase when the enclitic article is inserted. But this is often not the case, as argued for in detail in Dimitrova-Vulchanova and Giusti (1998). In a late principles-and-parameters framework (Chomsky and Lasnik 1993), the obvious revision of that hypothesis was to assume that the noun, already inflected with the article, checks the D-features by movement to D. This was the position taken by Dimitrova-Vulchanova and Giusti (1998), who reduce the variation across languages to the different levels of representation where this movement applies, combined with the requirement that a specifier must be in a Spec-Head relation with a head that already contains the f-features of the noun. Here I argue that the enclitic article is not an independent head inserted in D, which for some language-specific feature is so “strong” as to trigger N movement all the way to D; but it is an inflection of the noun that can trigger only one-step movement. As a consequence, for the noun to host the enclitic article, it must be merged in the immediately lower functional head. Let us consider the contrasts arising when a modifier of the noun is inserted in a noun phrase in which the enclitic article is present. Notice that all examples in (5) through (8) present the same word order displayed in the parallel indefinite noun phrases where no enclitic article occurs:10 (5)

a. b£iatul frumos boy-the nice

cf. un b£iat frumos (Romanian) a boy nice

A BARE PHRASE STRUCTURE APPROACH

b. *frumos b£iatul c. frumosul b£iat d. *frumosul b£iatul (6)

a. djali i mirë boy-the ARTgood b. * i mirë djali c. %i miri djalë d. *i miri djali

59

cf. un frumos b£iat cf. një djalë i mirë a boy ART good

(Albanian)

cf. %një i mirë djalë

(7)

a. *momceto goljamo cf. *momce goljamo (Bulgarian) b. *goljamo momceto c. goljamoto momce cf. goljamo momce big-the boy [a] big boy d. *goljamoto momceto

(8)

a. b. c. d.

*gutten store cf. *en gutt stor(e) *store gutten *storen gutt(en) den store gutten cf. en stor gutt the big boy-the a big boy e. *den store gutt (OK in Danish!)

(Norwegian)

In (5–8a), we find the word order expected if N-to-D movement takes place before Spell-Out, in (5–8b) the word order expected if N-to-D procrastinates until LF. This is never found. In (5–8c) and (5–8d) N does not move at all. Statements (5–8c) show that the article can appear on a prenominal adjective. This is not allowed in Scandinavian (8c). There, in the presence of an adjective the article is a free morpheme. In (5–8d) the article is reduplicated. This is only allowed in Scandinavian, except for Danish where we find the parallel of (8e) which, mutatis mutandis, corresponds to (5–7c).11 Some general considerations are suggested by the data in (5) through (8). N-toD movement is only found in languages where N-movement independently occurs.12 This suggests that the presence of the enclitic article does not necessarily trigger such a movement. Furthermore, in all languages under consideration here, the article is the topmost head of the structure.13 In a bare structure framework where the structure is built bottom-up, this suggests that the features expressed by the article are merged last in the noun phrase. Thus, the article merges on the highest nominal element. In Romanian, the highest nominal element can be the head noun, which has moved across the specifier containing the adjective. But since the movement of the noun is not obligatory (depending on the adjective and on the stylistic choice),14 the enclitic article can appear on an adjective. In Albanian, N moves across all specifiers in the unmarked case, regardless of the presence of the enclitic article. Thus the article appears on the noun in the unmarked case. In Bulgarian, N does not move across any adjective and the article can appear on the N only if no specifier is inserted. The same is the case in Scandinavian.15 2.3. Inseparable from their sister projection A third piece of evidence for the dependent nature of the definite article is the fact that it cannot be used discontinuously from its sister projection (9a), similarly to demonstratives (9b) but contrary to quantifiers (9c):

60 (9)

DP FUNCTIONAL STRUCTURE

a. **Ragazzi (li/ne) conosco i. b. ?*Ragazzi (*li/*ne) conosco questi. c. Ragazzi, ne conosco molti. / I ragazzi li conosco tutti. boys [I] CL-GEN know many / the boys [I] Cl-ACC know all

Assuming that Move cannot break extended projections but can exclusively apply to (FPmax), we obtain that only elements external to FPmax can remain in place.16 In Giusti (1991, 1993, 1997), I have argued that those Qs which can appear in distant positions are external to the noun phrase and impose selectional restrictions on their sister (which is an FPmax), such as case requirements. In (9c), molti ‘many’ absorbs the accusative case assigned by the verb conosco and imposes a partitive case on the noun phrase, which is realized as morphological genitive on the clitic form ne. Tutti has the property to let its case percolate down to FPmax, this is why the clitic li displays morphological accusative features. On the contrary, demonstratives in (9b) and articles in (9a) are internal to FPmax and cannot be separated from the rest of the projection. The crucial difference between articles and demonstratives is the impossibility for articles to appear without an overt sister projection (10a), while this is possible with demonstratives (10b): (10)

a. **Ho comprato il/lo/la [I] bought the b. Ho comprato questo/a. ‘[I] bought this.’

This suggests that articles, parallel to inflectional morphology; are merged only if a lexical head (a noun or an adjective) is present, while demonstratives can occur with a covert sister projection. This property is directly derived from the morphophonologically dependent nature of the article, as seen in sections 2.1 to 2.2. Apparent counterevidence to this generalization can be found in various Germanic languages. I take German as a case study here. In (11b–c), the definite article appears in the same contexts as a personal pronoun: (11)

a. Hans hat eine Frau gesehen. Sie stand am Fenster. b. Hans hat eine Frau gesehen. Die stand am Fenster. ‘Hans saw a woman. She was standing at the window.’ c. Hans hat sie/die gesehen. ‘Hans saw her.’

It is possible to argue that the d-element in (11b–c) is not an article but a demonstrative. (see Passaler 1997). Prepositional phrases help us distinguish between the definite article and the d-pronoun. Example (12a) gives us the basic structure with an indefinite article, which is bisyllabic and never incorporates. The definite article obligatorily incorporates in (12b), while the d-element cannot do so in (12c): (12)

a. Wir treffen uns an einem Eingang des Bahnhofs. ‘We’ll meet at an entrance of the station.’ b. Wir treffen uns am Eingang des Bahnhofs. ‘We’ll meet at the entrance of the station.’ c. Wir treffen uns an dem Eingang des Bahnhofs dort drüben. ‘We’ll meet at that entrance of the station over there.’

A BARE PHRASE STRUCTURE APPROACH

61

In (12c) the d-element appears with the locative da/hier which generally appears with demonstratives and is incompatible with articles in other languages, such as French in (13b' ) and Italian (13c'): (13)

a. das (Buch) da/hier (German) b. ce (livre) ci/la b'. *le livre ci/la (French) c. questo libro qui/quel libro lì c'. *il (libro) qui/lì (Italian) this (book) here / there

The co-occurrence of demonstratives and locatives is not just a matter of compatibility but one of selection, as argued in Brugè (1996 and chapter 2 in this volume). Therefore, the d-element should be analyzed as an XP, like a demonstrative. Brugè convincingly argues that the demonstrative and the locative are merged as a constituent. The demonstrative further moves to a higher projection, while the locative remains in place, thereby marking the basic position. Her analysis applies to d-elements in German, regardless of the presence of the locative. In (14a), I give the analysis of the relevant string of (12b), while (14b) corresponds to (12c): (14)

a. [FPmax am [FP2[Eingangi [FP1[des Bahnhofs] [NP [ti[. . .] ] ] ] ] ] ] b. [FPmax an [F3 demy [FP2[Eingangi [FP1[des Bahnhofs][DemP [ty (dort drüben) ] [NPti[. . .] ] ] ] ] ] ] ] ]

According to (H2), the demonstrative will further move for interpretive reasons to SpecFPmax, in compliance to (A4), as we will see in section 3.1.17 A further piece of evidence against a unification of the d-article in (15a) and the d-pronoun in (15b) is the fact that the two elements have a different dative plural form: (15)

a. mit den/*denen Freunden b. mit denen/*den

The fact that the d-element is partly homophonous with the definite article does not surprise us if we consider that in many Indo-European languages both articles and pronouns derive from demonstratives.18 The contrast in (12b–c) also shows that “articulated prepositions” like am are the conflation of two functional heads that are contiguous in the extended projection of the noun phrase. 2.4. Devoid of substantive content In some languages multiple occurrences of the article are found in one and the same noun phrase. This does not produce a multiple index interpretation: (16)

(17)

(18)

a. to oreo to vivlio b. to vivlio to oreo the good book ha-bxina ha-tedira shel ha-mismaxim the-examination the-frequent of the-documents ‘the frequent examination of the documents’ a. djali i mirë boy-the the-good

(Greek)

(Hebrew)

(Albanian)

62

DP FUNCTIONAL STRUCTURE

b. i miri djalë the good-the boy ‘the good boy’ (19)

a. b£iatul (cel) frumos boy-the (the) good b. frumosul b£iat nice-the boy ‘the nice boy’

(20)

den store gutten the good boy-the ‘the good boy’

(Romanian)

(Norwegian/Swedish)

In Greek (16) and Hebrew (17), the article on the demonstrative and the article on the adjective are apparently the same kind of element with respect to their morphology and their relation to the lexical head. Their occurrence is not limited to a single adjective but to every adjective present in the structure. In Albanian (18) and Romanian (19), we find two kinds of article that can appear on the adjective: one is the same as the nominal article (enclitic -i in Albanian (18b) and enclitic -ul in Romanian (19b)); the other is particular for the adjective. In both languages it is proclitic on the adjective, but it otherwise displays very different properties in the two languages.19 One apparent piece of evidence in favor of the referential value of the definite article is the fact that in Italian the repetition of the definite article introduces a different referential index, as argued in Longobardi (1994). In (21), where the article is repeated, the predicate must be plural, showing that the two articles in the singular have different indexes. This contrasts with (22) in which the article is not repeated and the interpretation of the subject can only be singular: (21)

a. *È arrivata la mia segretaria e la tua collaboratrice.20 has arrived the my secretary and the your assistant b. Sono arrivate la mia segretaria e la tua collaboratrice have arrived the my secretary and the your assistant ‘My secretary and your assistant arrived.’

(22)

a. È arrivata la mia segretaria e tua collaboratrice. has arrived the my secretary and your assistant b. *Sono arrivate la mia segretaria e tua collaboratrice. have arrived the my secretary and your assistant

However, the same data cannot be reproduced in Romanian, where the article is enclitic on the noun and cannot be missing. The sequence with the repeated articles in (23a) is ambiguous between the two possible interpretations, one with a single referential index and one with two indexes. The sequence in (23b) where the second noun is not inflected for the definite article is excluded: (23)

a. Directorul de departament „i presidentele de facultate a/au venit aici. director-the of department and president-the of faculty has arrived here b. *Directorul de departament „i presidente de facultate a/au venit aici. director-the of department and president of faculty has arrived here

Longobardi’s proposal to attribute a referential index to the definite article, therefore, can only account for the Italian free article and not for the Romanian enclitic

A BARE PHRASE STRUCTURE APPROACH

63

article. However, in most cases, the free versus dependent nature of the article is not crucial for the interpretive properties of the definite article in the two Romance languages under consideration. On the contrary, the Romanian enclitic article appears to have many features in common with the Italian free article, despite their different morphological status.21 It is welcome to give a unified account of the data in (21) through (23). This can be done by assuming that in neither language does the article have a referential index and that insertion of the article is always a last resort application of Merge. In Italian, the conjunction of two sequences of art + possessive adjective + N, as in (21), involves the conjunction of two FPmax’s. Both FPmax’s have a Specifier that is not empty but hosts a covert operator, which carries an index, à la Campbell (1996). The two indexes are interpreted as disjoint, according to principle C of the binding theory.22 If the insertion of the article is a last resort kind of operation, it cannot apply in the second conjunct in (22), where only one covert article is inserted. In Romanian, the conjunction of the sequence N + art + possessive noun phrase in (23) can either be interpreted as a conjunction of two FPmax’s parallel to Italian (21) or as a conjunction of two Fmax’s, parallel to Italian (22). In the former case, after each noun has checked its f-features in Fmax, a covert operator is merged in each Specifier. The result is the projection of two FPmax’s, which are then coordinated, yielding the interpretation with two different (disjoint) indexes. In the latter case, the enclitic article, because it is part of the morphological inflection of the noun, is part of each head N, without violating the last resort condition. N + art then checks the f-features in Fmax, the structure must get at least to the F2 level. At this stage, the two F2’s can be coordinated, yielding a recursive F2 node to which the covert operator is merged and obtaining an FP with a single SpecFPmax, which contains a single covert operator. So far, we have observed that in some languages, the co-occurrence of definite articles in one and the same noun phrase does not give rise to disjoint interpretation. In what follows, we observe that, conversely, the presence of the definite article is not always sufficient to trigger referential interpretation. In (24) the presence of a definite or an indefinite article is irrelevant for the interpretation of the noun phrase la/una segretaria di un onorevole, which is, in any case, interpreted as nonreferential, as shown by the subjunctive mood in the relative clause: (24) Scommetto che non troverai mai [FP la/una segretaria [PP di un onorevole che sia disposta a testimoniare contro di lui] ]. ‘I bet you’ll never find the/a secretary of a deputy who is-SUBJ willing to witness against him.’

I propose that the indefinite interpretation of the relevant noun phrase in (24) is due to the fact that the possessive prepositional phrase di un onorevole is moved to SpecFPmax at LF to give the same configuration as the English a deputy’s secretary. The indefinite interpretation then percolates from the possessive PP to the whole FPmax. In support of the covert movement of the indefinite possessor and pied-piping of its features into the entire FPmax is the fact that the indefinite possessor is incompatible with a demonstrative, which is a referential element in SpecFPmax, as in (25):

64

DP FUNCTIONAL STRUCTURE

(25) *Scommetto che non troverai mai [FP questa/quella segretaria [PP di un onorevole che sia disposta a testimoniare contro di lui] ]. ‘I bet you’ll never find this/that secretary of a deputy who is-SUBJ willing to witness against him.’

Example (24) also contrasts with (26) where the referentiality of the entire noun phrase is given by the referential possessive PP, the presence of which makes the noun phrase incompatible with the relative clause in the subjunctive mood: (26) *Scommetto che non troverai mai [FP la segretaria [PP di quell’onorevole]] che sia disposta a testimoniare contro di lui] ]. ‘I bet you’ll never be able to find the secretary of that deputy who is-SUBJ willing to witness against him.’

This is not the case in (27), which is introduced by an indefinite article that triggers an operator-variable interpretation and does not allow pied piping of the referential features of the possessor PP: (27) Scommetto che non troverai [FP una (sola) segretaria [PP di quell’onorevole] ] che sia disposta a testimoniare contro di lui ‘I bet you won’t be able to find a secretary of that deputy who is-SUBJ willing to witness against him.’

In (26) the possessive PP di quell’onorevole must (covertly) move to the Spec of the FP la segretaria for the whole FP to be interpreted, given that the definite article does not have an interpretive value. The referential interpretation of the possessor percolates to the entire FP, as argued here, and is therefore incompatible with the subjunctive mood, yielding an ungrammatical result. In (27), on the contrary, the interpretation of FP is that of the variable of a covert existential quantifier.23 This blocks movement of the PP containing the referential noun phrase di quell’onorevole. FP is compatible with the subjunctive relative, and the result is acceptable. A completely different type of evidence against the assumption of a referential index to the definite article is the fact that in several languages the enclitic article appears with the function of realizing nominal f-features. The Romanian examples in (28) are contrasted with the Italian counterparts to show that the Romanian definite article -ul is in all respects a feature marker of (at least) masculine singular parallel to the Italian morpheme -o; notice that it also contains case features (here nominative):24 (28) Romanian

Italian

Gloss

un(*ul) b£iat nici un(*ul) b£iat Am v£zut pe un*(ul) N-am v£zut pe niciun*(ul) un*(ul) a spus c£ Nici un*(ul) a spus c£ . . .

un(*o) ragazzo nessun(*o) ragazzo (ne) ho visto un*(o) non (ne) ho visto nessun*(o) un*(o) ha detto che un*(o) ha detto che

‘a boy’ ‘no boy’ ‘I saw one’ ‘I saw none’ ‘Somebody said that . . .’ ‘Nobody said that . . .’

In a number of languages articles appear in contexts in which they are not expected to realize case morphology. In (29a,b) we see a German generic mass noun that ap-

A BARE PHRASE STRUCTURE APPROACH

65

pears with no article in the accusative. But the noun phrase in dative in (29b) must have an overt definite article. The same happens when the noun phrase is in the genitive, as in (29c): (29)

a. Ich trinke gerne (*den) Kaffee. ‘I like drinking coffee.’ b. Ich ziehe (*den) Kaffee *(dem) Tee vor. ‘I prefer coffee to tea.’ c. die Zubereitung *(des) Kaffees ‘the preparation of coffee’

Example (29c) shows that the article in (29b) is not inserted to disambiguate the sentence. In (29c) no ambiguity would arise if the article were not present, but the article is still necessary. In this case it is necessary to make an oblique case (genitive) visible. So far, we have seen that the presence of the article is not sufficient to give referential interpretation, but it is necessary to realize nominal f-features such as gender, number, and case. Interestingly, the converse is also true: the presence of the article is not necessary for the referential interpretation of the noun phrase, and the article cannot be merged when the nominal f-features are realized in another way, as exemplified in prepositional phrases in Romanian. We have already noticed in note 21 that in Romanian noun phrases the article has the same distribution as in Italian generic and indefinite noun phrases. The same is also true for referential noun phrases, which must have a definite article, as in (30a). In prepositional phrases, however, in which case features are presumably recoverable from the preposition itself, the article must be missing, as in (30b), provided we have an unmodified noun phrase. But if the noun is modified either by an adjective, as in (30c), or by a complement, as in (30d), the article is necessary again: (30)

a. Am citit scrisoare*(a). ‘I read the letter.’ b. Îtçi multçumesc pentru scrisoare(*a). ‘I thank you for [the] letter.’ c. Îtçi multçumesc pentru scrisoare*(a) frumoas£. ‘I thank you for the beautiful letter.’ d. Îtçi i multçumesc pentru scrisoare*(a) de la Bucure„ti. ‘I thank you for the letter from Bucharest.’

In (30), gender and number features are recoverable from the morphology of the noun. The presence of the preposition makes the case features recoverable. These two conspirancies allow for the article to be absent in (30b).25 From the data presented in this section, we can conclude that the definite article is neither sufficient nor necessary to trigger referential interpretation on the noun phrase. This implies that the article is not the element that carries the referential index of the noun phrase at all. This is not an unwelcome result since it is well known that the distribution of articles is highly language-specific, while the distribution of semantic indexicals and operators such as demonstratives or quantifiers is rather uniform across languages.

66

DP FUNCTIONAL STRUCTURE

2.5. A bare phrase structure hypothesis So far, I have argued that the definite article, regardless of its morphology, is in all respects a functional head in the extended projection of the noun phrase. As such, it is expected to comply with the properties of functional heads assumed in assumptions (A1) to (A3): it is realized either as a free dummy or as an inflectional morpheme, it is merged as a last resort, and it shares all the f-features of the extended CHAIN. Being a functional head in the extended projection of the noun, it can only trigger incorporation of the lexical noun. Incorporation of other lexical heads is not expected under a strict interpretation of the Head Movement Constraint, which I maintain here. The apparent incorporation to adjectival heads will be treated in section 3.6. In all the languages observed here, the article appears to be the highest element of the noun phrase. In a bottom-up procedure, this is captured by assuming that the article is merged as the last functional head in the extended CHAIN. Here I follow a radical version of the bare phrase structure hypothesis under which merge and move are interacting procedures, each immediately applying on the resulting derivation of the other. All local movements, specifically head-movements, are reduced to a singlestep movement that is reiterated after every application of Merge. If this approach is correct, we expect the enclitic article to be merged in a configuration in which the immediately lower functional head is or contains the lexical head noun. This is actually the case, as Dimitrova-Vulchanova and Giusti (1998) have argued. Variation across languages is to be explained by independent properties of N-movement and its trigger. Let us focus on single cases. In Norwegian, N can move across a possessive adjective (see Taraldsen 1990): (31)

a. mitt hus b. huset mitt ‘my house’

Taraldsen shows that the word order in (31b) is derived by the basic order (31a) and a further application of N-movement triggered by the presence of the enclitic article. In Giusti (1995), I observed that Taraldsen’s proposal could not account for the different positions of the adjective and the possessive pronoun in noun phrases such as (32a) and (32b): (32)

a. mitt store hus b. det store huset mitt ‘my old house’

In (32a) the adjective follows the possessive, while in (32b) it precedes it. If (31a) were the base of (31b), we would expect the adjective in (32) to be in the same order with respect to the possessive pronoun. This criticism assumed Cinque’s (1994) hypothesis on noun phrase structure, which is set in an X-ba2 account in which all projections are labeled and present, even if not lexically filled, and each element is generated in a given position and may move to a different one to check features. In a bare phrase structure framework, however, it is the filler that labels the projection and the structure is projected bottom-up.

A BARE PHRASE STRUCTURE APPROACH

67

I maintain, following Brugè (chapter 2 in this volume), that possessives are (universally) merged very early in the structure: I take it here to be in SpecNP for expository reasons. The possessive carries a referential index. It is therefore sufficient to qualify the noun phrase as an argument, provided that it fills a functional Spec, according to assumption (A4). In (33a), the bare phrase structure of (31a), the possessive adjective is merged in SpecFP1. Alternatively, the possessive procrastinates movement to SpecFP1, so SpecFP1 must therefore be merged although empty. The last resort to project a functional structure is to merge the enclitic article triggering N-movement, as in (33b), parallel to (31b). In this case, SpecFP1 remains available for covert movement of the possessive pronoun, to yield the same interpretation as (33a) and (31a):26 (33) a.

FP1 F1'

mitt

NP

[φ]

N'

mitt

hus

b.

FP1 F1'

mitt

NP

hus-et mitt

N' hus

In other words, in (33) the possessive belongs to the same CHAIN in both structures. The only difference between the two structures in this respect is that in (33a) the upper link of the chain is phonetically realized, while in (33b) it is the lower link to be realized. In (33b), for the CHAIN to be interpretable, FP1 must be visible. Since SpecFP1 is silent, F1 must be overt. This is what forces merging of the enclitic article and further movement of the head N. On top of FP1, adjectives are merged recursively as specifiers of additional functional projections. This is the trigger for the FP2 projection in (34). Each functional head is projected copying all the f-features of the lexical head. The last application of Merge must create a functional specifier where the referential index of the noun phrase is checked. If these operations apply to (33a), the possessive pronoun moves to the highest Spec, thereby yielding (34a). If they apply to (33b), a free morpheme is inserted to fill the head of the highest functional projection, as in (34b), where the highest Spec remains available for covert movement of the possessive AP.27

68

DP FUNCTIONAL STRUCTURE

(34)

FP3 F3'

mitt

FP2

[φ]

F2'

store

FP1

[φ]

F1'

mitt FP3

NP

[φ] F3'

mitt

N'

mitt FP2

det

hus F2'

store

FP1

[φ] hus-et

NP mitt

N' hus

The heads containing f-features are present, although not overt, because the features are otherwise recoverable from the morphology in their specifier. This complies with assumption (A1).28 N-movement to F3 is excluded by the fact that the head did not raise to F2 in the cycle in which F2 was merged. In Norwegian, the realization of the enclitic -et on an N which is in a extended head CHAIN together with a free definite article, must be listed in the lexicon as a property of the inflectional morphology of nouns. The free morpheme det must be listed in the lexicon as the dummy to fill an Fmax in case N-et cannot reach Fmax before SpellOut. (A2) can be refined to explain the contrast in (35): (35)

a. huset b. *det hus c. *det huset ‘the house’

Example (35c) is already excluded by (A1). FP1 is already sufficient to satisfy all the interpretive requirements on the noun phrase, and no further head can be projected to host the dummy det. Assumption (A2), as it stands, cannot decide between

A BARE PHRASE STRUCTURE APPROACH

69

(35a) and (35b). In other words, we must introduce a hierarchy of “optimal realization,” according to which the enclitic article is preferred over the free article: (A2') If a functional head is realized, the optimal realization is the morphologically weakest.29

Danish is minimally different from Norwegian in that the enclitic article cannot appear in the same extended head CHAIN with a dummy. In other words, it cannot function as an inflectional morpheme in an intermediate projection. This implies that if it is inserted, its specifier must contain the covert operator with referential index, and no other application of Merge can apply, as in (36a). If adjectives are merged in the noun phrase, the only available functional head to realize the highest projection is the free det, as in (36b), for the same reasons as in Norwegian in (35), namely to comply with locality: (36)

a. [FP1 [hus-[F1et] ] [NP hus] ] b. [FP2 [F2’ det [FP1 store [NP hus] ] ] ]

The cross-linguistic variation between Danish and Swedish can be reduced to a stricter application of (A1) in Danish. Bulgarian (37) parallels Danish (36a). Different from Danish, however, the Bulgarian lexicon has nothing like a free article. So, there is nothing like (36b) in Bulgarian: (37) [FP1 [momce-[F1to] ] [NP momce] ]

When the adjective is merged, the noun does not move to the newly merged functional head, as in the indefinite noun phrase (38a). Merging of the enclitic article in FP2 does not trigger N-movement. Because head movement applies cyclically, N cannot reach F2 in (38b): (38) a. [FP1 goljamo F1 [NP momce] ] b. [FP2 -to [FP1 [AP goljamo] F1 [NP momce] ]

The alternative is for AP to move to SpecFP2. The details of this procedure will be considered in 3.6. Romanian and Albanian show the exact opposite situation. The noun is cyclically moved into the newly merged functional head. In Romanian, the noun can precede a descriptive adjective. In our framework, this means that it moved to F1 in the previous cycle, as in (39a). In (39b), we observe that the noun can further move across the demonstrative, provided that an enclitic article is merged to project the relevant portion of structure: (39) a. [FP2 acest [F2 b£iat] [FP1 [AP frumos] b£iat [NP b£iat] ] ] b. [FP3[ [b£iat-[F3ul] ][FP2acesta [ [F2b£iat][FP1[APfrumos][ [F1b£iat][NP b£iat] ] ] ] ] ] ]

In Albanian the noun must always move across the adjective:30 (40)

a. [FP4një [FP3[F3grua][FP2tjetër [ [F2 grua][FP1[APe bukur][ [F1grua][NP grua] ] ] ] ] ] ] b. [FP3[ [grua-[F3 ja] ] [FP2tjetër [ [F2grua][FP1[APe bukur] [ [F1grua] [NP grua] ] ] ] ] ] ]

The general observation made above that the enclitic article triggers N-movement only in those configurations in which the noun is in the head of the highest projection in the preceding cycle appears to be empirically true.

70

DP FUNCTIONAL STRUCTURE

Let us finally analyze the Romanian data in (30), where the article is not inserted if the noun is preceded by a preposition and has no modifier: (41)

a. [FP1 [F1 pentru] [NP scrisoare] ] b. [FP3 [F3 pentru] [FP2 scrisoarea [FP1 frumoas£ [NP scrisoare] ] ] ]

In both structures in (41), the preposition pentru fills Fmax. In (41a) the derivation is complete. In (41b), a modifier is inserted in FP1. For the head F1 to be in a CHAIN with the relevant f-features, the noun must move across it. This triggers insertion of the article in F2. Finally, the preposition is merged in F3. It is beyond the goal of this chapter to explain what makes it necessary for the head noun to move across modifiers in some languages and not in others.

3. Occupants of SpecFPmax In various cases in the preceding text, I assumed without discussion that possessive adjectives and demonstratives are maximal projections in specifier positions, even in languages where they appear in complementary distribution with the article. This complementary distribution between a lexical head in Fmax and an XP in SpecFPmax is derived by the interaction of assumption (A1), which disallows insertion of an overt element in a functional head if not necessary, and a general principle such as (42), proposed in Dimitrova-Vulchanova and Giusti (1998): (42) Principle of economy of lexical insertion: A functional projection must be licensed at all levels of representation by a. Making the specifier visible. b. Making the head visible.

Principle (42) is partly subsumed by the general theory of bare phrase structure, since a functional projection is built only if an XP is merged as its specifier or a functional element is merged in its head. Examples (42a) and (42b) can be either disjoint or conjoint; when (42a) and (42b) are disjoint, the principle of economy of lexical insertion derives the “doubly filled Comp Filter”; when (42a) and (42b) are conjoint, the principle of economy of lexical insertion accounts for verb-second structures and doubly filled Comp languages. The choice between the conjunction and the disjunction of the two conditions in (42) depends on the language, on the projection, and on the element in Specifier position. With respect to our topic, we will see that (42a) and (42b) are disjoint in cases in which the possessive adjective or the demonstrative is in complementary distribution with the article; they are conjoint in those cases in which the possessive adjective or the demonstrative precedes the article. We will see next that the demonstrative and the possessive adjective are not the only categories to appear in SpecFPmax. 3.1.

Demonstratives

In Giusti (1993, 1994a, 1997) I proposed, on comparative evidence, that demonstratives are maximal elements inserted in functional specifiers. For Romanian, I

A BARE PHRASE STRUCTURE APPROACH

71

argued for the maximal status of the demonstrative acest(a) on the ground that starting from a basic word order, as in (43a), the demonstrative can only be crossed over by the head noun, as in (43b), and not by an adjective that is more certainly than the noun a maximal element, as in (43c): (43)

a. acest b£iat frumos this boy nice b. b£iatul (acesta) frumos boy-the (this) nice c. frumosul (*acesta) b£iat nice-this (*this) boy ‘this nice boy’

I also argued there that its basic position is immediately lower than the highest functional projection, on the basis of its obligatory second position shown by the contrast in (44): (44)

a. b£iatul acesta frumos b. *b£iatul frumos acesta

However, Brugè (chapter 2 in this volume) and Brugè and Giusti (1996) show that demonstratives are very low specifiers in a wide range of languages. The second position of the Romanian demonstrative is also taken to be derived. But no motivation for this is given there. I will provide one in this section. Let’s compare Spanish and Romanian. In (45), the order of the modifiers in one language is neither parallel to nor a mirror image of the other: (45)

a. el (ultimo) cuadro redondo este (*el) suyo (Spanish) b. tabloul acesta rotund al s£u (Romanian) ‘this round painting of his’

In Spanish (45a), the order is Art (. . .) > N > descriptive AP > Dem > possessive AP. In Romanian (45b), the order is N + art > Dem > descriptive AP > possessive AP. There is a parallelism and two apparent differences. The parallelism is that the postnominal descriptive adjective is higher than the postnominal possessive adjective in both languages. The first difference is that the head noun in Spanish is in an intermediate position (it is preceded by the definite article and can be so by a prenominal adjective), while the head noun in Romanian is the first element in the noun phrase. The second difference is that the demonstrative in Spanish is lower than all specifiers except the possessive adjective, while in Romanian, it is the highest specifier, being the second element after the head noun inflected for the definite article. Recall that in the framework adopted here, descriptive adjectives check their f-features by Spec-Head Agreement with the local functional head. They have no need to move. The position of the descriptive adjective must be basic in the two languages. I assume that although functional projections are unlabeled, modifiers are merged in a universal order, as argued in Cinque’s (1994) seminal work and in much work inspired by it. I follow Brugè (chapter 2 in this volume) in taking the order found in Spanish (45a) and represented in (46a) to be the basic order across languages. Example (46b) presents a possible derivation: the FP including N + art and the demonstrative moves across the descriptive adjective. The resulting order

72

DP FUNCTIONAL STRUCTURE

is NP > poss.AP > dem.> descr.AP. This order is not found in Romanian, as shown in (46c): (46)

a. descr.AP > dem. > poss.AP b. [FP6 [FP4 [FP2 NP[FP1poss.AP [NP] [FP3 dem. [FP2NP[FP1poss.AP [NP] [FP5descr.AP [FP4 [FP2NP[FP1poss.AP [NP] [FP3 dem. [FP2NP[FP1poss.AP [NP] c. b£iatul s£u (**acesta frumos) boy-the his this nice

The alternative is (47a) where the demonstrative moves to an intermediate position before moving to SpecFPmax. In (47a), the head noun moves cyclically, first across the possessive AP, then across the descriptive AP. At this stage, the demonstrative must move to FP4, as shown by the grammaticality of (47b) and the ungrammaticality of (47c). At this point, the merging procedure can stop. In (47b) the demonstrative is in SpecFPmax, the noun phrase complies with (A4), and no article is merged in this structure. However, an enclitic article can be merged to create a further projection, as in the full structure in (47d). Merging of the article triggers further N-movement and the creation of a new Specifier position, which must remain available for further movement of the demonstrative at LF. Example (47e) is excluded by assumption (A1), according to which merging of the article is a last resort procedure. If dem is in SpecFPmax, the head Fmax needs not be overt and thus cannot: (47) a. [FPmax dem. [N + art [FP4 dem.[N [FP3 descr.AP [N [FP2dem. [N [FP1 poss.AP [N] ] ] ]]]]]]] b. acest b£iat frumos al s£u c. **b£iatul frumos acesta al s£u d. b£iatul acesta frumos al s£u e. **acest b£iatul (frumos al s£u)

Two open questions are left in the analysis of the pattern in (47): What makes movement of the demonstrative to the intermediate position necessary in (47d)? And what makes merging of the article in (47d) necessary? The former question is spurious in a bare phrase structure framework: demonstrative movement in (47b) is of the same kind as in (47d). Both are driven by (A4), namely by interpretive reasons. Romanian is parallel to Italian and different from Spanish, in that demonstrative movement cannot procrastinate. Romanian differs from Italian in the nature of the article. The Romanian enclitic article triggers N-movement, whereas the Italian free article does not. As for the second question, Tasmowski-De Ryck (1990) convincingly argues that the two positions of the demonstrative have a different discourse interpretation: the prenominal position is thematic, while the postnominal one is rhematic. The merging of the article in (47d) is therefore necessary to trigger N-movement and the consequent rhematic interpretation of the demonstrative. The unacceptability of the demonstrative in (43c) is still explained by the crucial assumption that the demonstrative must appear in SpecFPmax, at the latest at the LF level. In (43c) the adjective inflected for the article occupies this position, making it unavailable for checking of the demonstrative at LF. Bernstein’s (1997) proposal is an alternative to the present account in two respects: On the one hand, the complementary distribution of the demonstrative and

A BARE PHRASE STRUCTURE APPROACH

73

the article is taken as evidence that the two elements compete for the same position. The demonstrative is thus taken to cliticize onto D. On the other hand, it is proposed that such movement does not apply in the case of nondeictic demonstratives, as in (48): (48)

a. there’s this guy b. il y a ce gars

The demonstratives in (48), being nondeictic, would not be in the highest specifier. According to Bernstein, they are preceded by an empty functional projection. This predicts that there must be some word order differences between deictic and nondeictic demonstratives, at least in Romance languages where the noun moves across specifiers, contrary to facts.31 Furthermore, in Romance languages, indefinite singular noun phrases cannot lack an article, as is also the case for plural indefinite noun phrases in subject positions. In Bernstein’s framework, we would expect an indefinite article to precede the indefinite demonstrative. But this is never the case, and the nondeictic demonstrative appears to be in complementary distribution with the article, as well as with other determiners: (49)

a. C’è *(un) ragazzo nuovo che mi piace molto. b. C’è (*un) questo ragazzo nuovo che mi piace molto. ‘There’s this new guy who I like a lot.’

In the bare phrase structure analysis developed here, there is no way to differentiate between a high position preceded by an empty functional head and the topmost position. I think this is welcome, given that there is no syntactic effect to be noticed between the two positions. The different interpretation can be reduced to two different associations of lexicosemantic features to an element in a given language. In particular, the nondeictic interpretation is incompatible with a deictic reinforcer. A functional head status for demonstratives has been argued against in section 2, in comparison with the functional status of articles.32 In the framework developed here, the complementary distribution of the demonstrative and the article is dealt with by assuming a disjunction of the two conditions in (42). The assumption of the cliticlike status for demonstratives in languages such as Italian would not dispense with (42). Notice that in Italian demonstratives would be the only bisyllabic clitics and the only clitics to appear inside the noun phrase. Summarizing: The demonstrative moves to a high functional specifier to let its interpretive features percolate onto the noun phrase that contains it. This movement can procrastinate in some languages—for instance, Spanish among many others— but it must take place while constructing the structure in others, for example, Italian and Romanian. The second position of the demonstrative in Romanian is analyzed here as the result of such movement and a further application of Merge to allow for the noun to cross the demonstrative and derive a rhematic interpretation of it. The apparent second position of the demonstrative is not so language-specific as may appear at first sight. If we take functional prepositions to be part of the extended projection of the noun phrase, we must take demonstratives preceded by such prepositions as being in second position: cf. (14b) in German and the parallel case in Italian (50b):

74 (50)

DP FUNCTIONAL STRUCTURE

a. Ci vediamo all’entrata della stazione. b. Ci vediamo a quell’entrata laggiù. ‘Let’s meet at the entrance down there.’

At various points in the preceding discussion (see section 2.1 (3b) for Italian, section 2.3 (14b) for German, and section 2.4 (30b) for Romanian), I have hinted at the possibility of considering (at least some) prepositions to be merged in Fmax. This is also the position of the preposition in (50b). The demonstrative is necessarily in the immediately lower SpecFP. I assume that, according to (A4), it will covertly move to SpecFPmax. 3.2. Possessive adjectives Possessive adjectives are often considered to be determiners in Germanic languages. Even in these languages, however, there are cases in which the two different elements can co-occur. This happens only if the possessive is lower than the article. We have already observed and analyzed the Norwegian case in (31). Let us observe now the case of German in (51): (51)

a. (*die) meine Frage b. diese meine Frage

In (51a) we observe the unmarked case. In (51b) the noun phrase is emphatic. In (52a) the elliptic noun phrase consists of a possessive pronoun that displays strong inflection. In (52b) an article precedes the possessive adjective that displays weak inflection, which is typical of regular adjectives: (52) Deine Fragen wurden beantwortet, . . . a. meine jedoch nicht. b. die meinen jedoch nicht. ‘Your questions were answered, mine however were not.’

Up to this point I have maintained that if an element starts as an XP, it also moves as XP, unless it is a weak element (e.g., clitics or weak pronouns). The possessive in German cannot be weak, since it is bisyllabic. I take it to be an XP in all cases. The apparent complementary distribution with the article can be reduced to an either/or choice of principle (42).33 A SpecFPmax position can also be argued for in the case of possessive adjectives modifying kinship terms in Italian. In (53), it is shown that it must be a property of the noun that triggers the observed complementary distribution of possessive adjectives and articles, and not a property of the adjective itself, which remains morphologically the same in all cases: (53)

a. (*la) sua madre b. %(la) sua mamma. c. *(la) sua (amata) madre (adorata) ‘his/her dear mother’

In (53a) the article is ungrammatical when the possessive adjective modifies the noun madre. In (53b) the situation changes with the noun mamma with the same

A BARE PHRASE STRUCTURE APPROACH

75

meaning but more colloquial than madre. Here the presence of the article is allowed only in northern regional varieties, while it is excluded in other regional varieties and in the standard. In (53c) the noun madre is further modified by an adjective (regardless whether it is pre- or post-nominal), and the article becomes obligatory again.34 First, we must observe that this is a property of a closed class of lexical items that includes only some kinship terms, for example, madre, padre, fratello, sorella, zio/a, cugino/a, nonno/a, cognato/a, and suocero/a but not mamma, papà, fratellastro, sorellastra, patrigno, matrigna, consuocero/a, prozio/a, or bisnonno/a, as well as the diminutive counterparts of the former group of nouns, such as mammina (in the variant where mamma patterns with madre), sorellina, nonnetta, zietta, and so on: (54)

a. (*la) mia sorella b. *(la) mia sorellina

Second, it takes place only in case no modifier other than the possessive adjective is inserted, as shown in (53c). This latter property suggests that the structure of (53a) and (54a) is minimal—possibly limited to one functional projection. This may be due to a particular property of these nouns that assigns a particular q-role in SpecNP. Merging the possessive adjective in SpecFP1 satisfies every requirement of the noun phrase. But if a further adjective is inserted, as in (53c), the usual structure is projected to allow partial N-movement. In this case, the noun phrase has the same structure as other common nouns, and merging of an article is necessary. In the cases of other kinship terms that do not display the property of discharging the q-role in SpecNP, the result is the same as other common nouns. Possessive adjectives are not the only adjectives that can carry referential features and therefore be moved to SpecFPmax. Delsing (1993, 118–119) presents a list of referential adjectives in Scandinavian that trigger apparent optional article insertion when they modify certain nouns: (55) (den) franska revolutionen, (den) svenska kyrkan, Svarta Havet, Vita Huset (the) French Revolution-the, (the) Swedish Church-the, Black Sea-the, White House-the (det) sista paret, (den) tredje gången, (den) yttre/inre sängen, (den) västra sidan the last pair-the, the third time, the outer/inner bed-the, the western side-the

In the framework adopted here, the article cannot be optional, given that it is a last resort procedure. The noun phrases in (55) must have two different structures: one in which the article cannot be inserted, and one in which the article must be inserted. The former is parallel to the analysis proposed for Italian kinship terms modified by possessive adjectives; the other is the structure for common noun phrases. 3.3.

Personal pronouns

Postal’s (1969) seminal work drew attention to the possibility for personal pronouns to introduce noun phrases. In our framework we may wonder whether they are functional heads or specifiers. Here I argue for the latter hypothesis. In Italian (56a), the pronoun is in complementary distribution with an article, but this is not the case in Romanian (56b):

76 (56)

DP FUNCTIONAL STRUCTURE

a. Voi (*i) professori credete che sia facile ma per noi (*gli) studenti è difficile. b. Dumneavoastr£ profesori*(i) credetçi c£ e u„or, dar dup£ noi elevi*(i) e„te greu.35 ‘You professors believe that it is easy, but for us students it is difficult.’

It is crucial to establish whether we are dealing with two instances of the same construction and what this is. Cardinaletti (1994) argues for Italian, and this could be straightforwardly extended to Romanian, that pronouns are full noun phrases and that the following nominal is in a sort of predicate position. Example (57a) would be parallel to (57b), on the ground that the two noun phrases behave similarly with respect to some tests: (57)

a. noi/voi linguisti ‘us/you linguists’ b. i miei amici linguisti the my friends linguists ‘my friends [who are] linguists’

I follow Cardinaletti in taking the two strings in (57) as two instances of the same construction. I also agree with her to analyze strong pronouns as full noun phrases. But I depart from her line of reasoning where this is taken to be evidence for an adjunction construction, extending the parallelism to other cases of multiple occurrences of nouns or to proper names in a single noun phrase with a single referential index, such as those discussed in section 3.4. The framework developed here provides a straightforward explanation to the contrast between Italian and Romanian in (56), provided that we assume a rather costless stipulation: in Romanian, Merge of the pronoun is not sufficient to make the extended projection of the noun phrase visible, while in Italian it is. In the presence of a pronoun in SpecFPmax the two conditions in (42) must be conjoint in Romanian, while they must be disjoint in Italian.36 If pronouns carry the referential index of the noun phrase, they are expected to be in complementary distribution with demonstratives in the two languages, as in (58): (58)

a. *noi questi ragazzi b. *noi ace„ti b£ietçi / *noi b£ietçii ace„tia ‘we these boys’

Furthermore, if noun phrases are third person by default, we expect that Cardinaletti’s “Minimize structure” or our principle of “optimal realization” (A2') on p. 69 should prevent insertion of third-person pronouns, as is the case: (59)

a. *loro linguisti/e b. *essi/esse linguisti/e ‘they linguists’

3.4.

Proper names

Proper names have many properties in common with demonstratives. Longobardi (1994) derives (60b) from (60a) by head movement of the lexical noun into D. He does not discuss the case of complex noun phrases such as those in (60c–d): (60)

a. b. c. d.

il mio Gianni Gianni mio Gianni Rossi Francesco Maria Finzi Contini

A BARE PHRASE STRUCTURE APPROACH

77

I propose that (60c–d) are in all respects parallel to (60b). Reformulating Longobardi’s proposal in our bare phrase structure framework, we must face a problem: in (61) the lexical head Gianni projects a maximal projection in whose specifier the possessive adjective mio is merged. A functional projection is therefore merged to create at least the functional specifier, which is the locus of interpretation of the whole noun phrase. The functional head is filled by the definite article showing that the specifier is either empty or covert. The covert element cannot be the empty operator assumed for referential common nouns, since proper names have different interpretive properties. The assumption of two different covert operators would be quite stipulative and could not explain how speakers can differentiate between the two. We are left either with Longobardi’s proposal that the head noun covertly moves to Fmax (D in his framework) by substitution (which also appears rather stipulative) or with a natural alternative—namely, that the proper name is moved as NP into SpecFP2, covertly in (61a) and overtly in (61b). In the latter case the article is not inserted: (61)

a.

FP2 F'

NP Gianni

FP1

F2 il

F'

mio {tenero }

NP

F1

Gianni

b.

FP2 F'

NP Gianni

FP1

F2

F'

{Rossi mio } F1

NP Gianni

In (61a), the proper name is minimal and maximal at the same time, and it is labeled NP.37 A modifier is inserted in FP1, which can either be a further specification of the name or a possessive adjective (the lowest adjective in the hierarchy). Then the proper NP must raise to escape the cycle. A functional projection (FP2) is created by this application of Move. At this point in the derivation, nothing else is necessary: the highest specifier has an index, and that is sufficient to make the noun phrase

78

DP FUNCTIONAL STRUCTURE

interpretable at LF. Alternatively, the NP may remain in place and procrastinate movement when the modifier is the family name, obtaining Rossi Gianni. In this case, too, nothing else needs to be done, since the family name also carries a referential index. The extended projection would stop at FP1 in (61a). A possessive adjective in Italian cannot function as a referential operator, so either the NP moves across it, as in (61a), or a definite article is inserted to allow the NP to procrastinate this movement until later, as in (61b). I leave open the question of what prevents NP movement when a descriptive adjective is present.38 The case in (60d) is a cyclical application of the operations in (61a), as in (62): (62)

FP4 FP2

F4'

Francesco Maria

F4'

FP3

FP2

F3'

NP

F2'

F3

FP2

Finzi

F2

FP1

NP

F2'

NP

F1'

Francesco

F2

FP1

Contini

F1

NP

NP

F1

Finzi

Maria

F1

NP Francesco

In (62), first the complex first name (Francesco Maria) is created in FP2, then a complex family name (Finzi Contini) is merged with the same structure, which is labeled in italics to distinguish it from the structure of the first name. Finally the whole projection FP2 (in boldface) containing the proper name is merged to a higher specifier to give the referential index to the entire noun phrase, yielding Francesco Maria Finzi Contini. If this last movement does not apply, the referential index would be given by the last name, obtaining the string Finzi Contini Francesco Maria, which is typical of the bureaucratic style. If we reconsider the different results with respect to the conflation of article and preposition observed in (3), we can now explain why this does not take place with proper names (or demonstratives). Proper names and demonstratives are in specifier position, while this phenomenon only involves two contiguous functional heads in the extended projection. 3.5.

Apparent adpositions

Now we can go back to the problem presented by complex noun phrases such as (63a) pointed out in Cardinaletti (1994). I propose to extend the analysis given for complex proper names in (62) to all complex noun phrases in (63):

A BARE PHRASE STRUCTURE APPROACH

(63)

a. b. c. d.

79

i miei amici linguisti (my friends [who are] linguists) Giovanni il giardiniere (John the gardener) il dottor Gianni Rossi (‘doctor G. R.) (la) zia Vittoria (aunt V.)

An adposition analysis of the strings in (63) is excluded by the observation that they are not pronounced with comma intonation. The comma intonation is possible, but not with all strings. Furthermore, it blocks vowel deletion and, crucially, produces a rather different interpretation, as expected if the second noun phrase is a predicate of the first. Comma intonation is excluded for (64a, d). In (63c) the noun dottor is interpreted as an academic title, while in (64c) dottore is interpreted as the professional title “medical doctor” (as when no proper name is present at all) and does not delete the final vowel. Deletion of the epenthetic vowel of the article(i-) also takes place in fast speech in (63b) but not in (64b): (64)

a. b. c. d.

*Francesco, Maria, Finzi, Contini Giovanni, il giardiniere, il dottor*(e), Gianni Rossi, *la zia, Vittoria, / *zia, Vittoria

I take this to be evidence that the noun phrases in (63) are not formed by a predicative adposition, while those in (64) are.39 Example (64) shows that an adposition is only possible in a predication structure—namely, when the two noun phrases have two different indexes.40 The proposal developed here provides a framework to merge a number of maximal projections as specifiers of the head noun. Let us start from the cases that display a definite article, since I have argued earlier in this essay for a specific position inside the noun phrase structure universally—namely, the highest functional head in the extended projection of the noun phrase. In (63b), the proper name Giovanni must be a maximal projection in the specifier of the article. As a proper name it carries a referential index. This index percolates up the entire noun phrase. Examples (63c) and (63d) display exactly the opposite case. Here the lexical head of the extended projection is the proper name, and the common noun is a qualifying modifier. These proper names are therefore preceded by the definite article, parallel to proper names modified by a descriptive adjective, as in (65): (65)

a. il tenero Giacomo (sweet Jakob) b. l’antipatica Maria (obnoxious Mary)

In (65), the head of the construction is certainly the proper name. The head of the construction in (63) can be determined following the intuition of the speaker. In (63a), repeated as (66a), it is amici. It is possible to construct a parallel case with the head noun linguisti, obtaining a different word order, as in (66b). The pronominal position of the possessive is preferred when it refers to the head noun, as shown by comparing (66a) with the less acceptable (66c); while the postnominal possessive is preferred for the modifier noun phrase, as shown by comparing (66b) with the less acceptable (66d):

80 (66)

DP FUNCTIONAL STRUCTURE

a. b. c. d.

i miei amici linguisti (my friends [who are] linguists) i linguisti amici miei (linguists [who are] friends of mine) ?gli amici miei linguisti (my friends [who are] linguists) ?i linguisti miei amici (linguists [who are] friends of mine)

Other orders are unallowed. Example (67a) shows that the modifier noun phrase is merged very early in the structure even earlier than the possessive adjective. Example (67b) shows that the head noun must move across its noun phrase modifier: (67)

a. *gli amici linguisti miei (the friends linguists of mine) b. *i miei linguisti amici (the my linguists friends)

The early merging of the noun phrase modifier can be due to the slightly different properties of agreement for nouns and adjectives. For example, a modifier noun phrase may in marginal cases display gender features that are different from the gender of the head noun, as in la mia amica medico ‘my friend-FEM doctor-MASC’. In a bottom-up procedure, each newly projected functional head is identical to the lower one. Once the features triggering adjectival agreement are projected, no noun phrase modifier can be inserted. The noun phrase modifier must therefore be projected before the adjectival agreement projection is started. This excludes (67a) where the possessive is lower than and therefore merged before the modifier noun phrase. Example (67b) is excluded by the general principle that triggers partial N-movement in Italian. 3.6. Adjectives inflected for definiteness As we have observed in examples in (5) through (8), languages that display prenominal adjectives in indefinite noun phrases do so in definite noun phrases as well, regardless of the presence of the definite article. If the adjective is prenominal in the Balkan languages represented in (68), the article appears on the adjective: (68)

a. e bukura vajzë (Albanian) the nice-the girl b. goljamoto momce (Bulgarian) big-the boy c. frumosul b£iat (Romanian) nice-the boy

Grosu (1988) notices for Romanian that head-movement of the adjective would incorrectly predict that the adjective would precede its adverbial modifier, as in (69b). But the enclitic article on the adjective does not change the canonical word order with the adverbial preceding the adjective, as in (69a). The same is the case in Albanian (70) and Bulgarian (71), as observed in Dimitrova-Vulchanova and Giusti (1998): (69)

a. foarte frumosul b£iat (Romanian) very nice-the boy b. *frumosul foarte b£iat

(70)

a. shumë e bukura vajzë (Albanian) very the nice-the girl b. *e bukura shumë vajzë

A BARE PHRASE STRUCTURE APPROACH

(71)

81

a. mnogo goljamoto momce (Bulgarian) very big-the boy b. *goljamoto mnogo momce

Rephrasing the analysis proposed in Dimitrova-Vulchanova and Giusti (1998) in the framework developed here, I propose the structure in (72) where the leftmost adjective41 is merged in SpecFPmax, which is already inflected for the definite article:42 (72)

FNPmax FN' FAPmax

FN FAPmax

FA' AdvP

F

NP

AP

N' N

A' A˚ a. b. c.

foarte shumë mnogo

frumos-ul e bukur-a goljamo-to

frumos e bukur goljamo

¨ baiat vajzë momce"

This hypothesis makes two correct predictions. First, if the prenominal adjective can have a complement, the inflection still appears on the adjective. Second, if the prenominal element is a conjunction of adjectives, the inflectional morphemes appear in both. These two facts could not be captured in the alternative hypothesis, which assumes the enclitic article to be in FN. In Bulgarian, as in other Slavic languages, it is possible for a prenominal modifier to display a complement. Let’s compare the Bulgarian facts in (73) with the English Saxon genitive in (74): (73)

a. *[ [mnogo vernij na ©ena si] [-at [mu©] ] ] b. [ [mnogo vernij- at na ©ena si] [F° [mu©] ] ] very true -the to wife his man ‘the man very true to his wife’

(74)

a. [ [the man I met yesterday] [’s [hat] ] ] b. *[ [the man’s I met yesterday] [F° [hat] ] ]

In (73a) and (74a), the highest functional head of the noun phrase contains the relevant weak morpheme. In English, this yields a grammatical structure. In Bulgarian, this results in ungrammaticality. In (73b) and (74b), the weak morpheme is internal to the adjectival phrase. This is grammatical in Bulgarian and ungrammatical in English. This contrast can be explained by (A2), which favors dependent morphemes over free ones to realize functional heads. English has a very poor inflectional morphol-

82

DP FUNCTIONAL STRUCTURE

ogy. In other words, it is very poor of dependent morphemes. The English ‘s is phonologically dependent but morphologically free; it is merged in the genitive assigning functional projection, which is also FPmax in the English noun phrase, because the noun phrase in its specifier percolates its referential value up to the entire FPmax. In Bulgarian, the noun is not moved across its modifier; the only possible way to insert an inflectional morpheme to realize the referential features of the noun phrase is to do so inside the extended projection of the adjective. Coordination is a further argument in favor of the analysis in (72). If SpecFNPmax hosts a coordinated structure, (72) predicts that the inflectional morphology appears on both coordinated adjectives, as in (75a) and (76a); the alternative analysis places the inflectional morpheme in FN and wrongly predicts that the inflectional morpheme is only one and that it appears on the second conjoint adjective, as in (75b) and (76b):43 (75) a. frumoasele „i bunele fete (Romanian) beautiful-the and good-the girls b. *frumoase „i bunele fete c. *frumoasele „i bune fete (76)

a. e gjora dhe e vogla vajzë (Albanian) the poor-the and small-the girl b. *e gjora dhe e vogël vajzë c. *e gjorë dhe e vogla vajzë

3.7.

Conclusions

In this section, I have motivated the assumption formulated in (A4) according to which SpecFPmax is the locus of the interpretation of the referential value of the noun phrase at LF. This assumption grounds two claims made here. One is stated in (H2) and regards the categorial status of demonstratives as maximal projections. The other is complementary to (A1), according to which insertion of the article is a last resort. Insertion of the article is necessary if SpecFPmax must be projected, but it is either empty (because it must be available for movement of an XP at a later stage of the derivation) or covert (as in the case of the covert operator proposed by Campbell 1996). If SpecFPmax is filled, no article is needed. 4. Final remarks In this chapter I have argued against a unified treatment of determiners but in favor of a principled treatment of functional heads in the extended projection of the noun phrase. One of these functional heads is the article, cross-linguistically. I have proposed that functional elements are never inserted for semantic reasons, since they have no semantic content. As a consequence, their insertion is a last resort procedure. Demonstratives are different from articles and similar to other modifiers of the noun that provide referential features to the noun phrase. Among these, we can mention possessive adjectives, referential adjectives, possessive noun phrases, and proper

A BARE PHRASE STRUCTURE APPROACH

83

names. These elements contribute semantic content to the noun phrase and are maximal projections. The apparent complementary distribution of articles and these other elements is derived by the last resort nature of article insertion. If this study is correct in its essentials, it offers a more radical way to look at functional elements as the realization of features of the lexical head of their extended projection. In an economy framework, their realization is expected to be highly limited. Notes I thank Anna Cardinaletti and Guglielmo Cinque for comments and discussion, and the audience of the first meeting of the joint project “The Cartography of Syntactic Structures” held at the University of Venice on January 29–30, 1999, for helpful comments. I also thank Elisabeth Pearce and an anonymous reviewer for pointing out some faults in the submitted version. The usual disclaimers apply. 1. For the projection of the arguments of the noun, I assume Larson’s (1988) proposal according to which the elements that satisfy the selectional requirements of lexical heads, including q-role assignment, are merged in a shell built by recursion of the label of the lexical head (NP-shell in this case). I have nothing to contribute to the NP-shell theory here. My contribution is limited to the syntax of so-called determiners and will only be relevant to the functional part of syntactic structure. 2. “Weak” in the sense of Cardinaletti and Starke (1999). 3. In this respect, it is more general than the assumption made in Chomsky (1995, 240), which takes D together with C and T apart from other functional heads and attributes semantic content to them. I have nothing to say about T. In this essay I will not attempt any claim on C, either. However, I envisage a parallel treatment of D and C, as often implied in the literature. 4. I disregard the case of possessive clitics inside the noun phrase as spurious. Clitics or weak pronouns are special elements that have both maximal and minimal status. It is their maximal status that is accounted for in this essay. More work needs to be done to account for the position they end up in due to their the minimal status. Nothing here is against the hypothesis that they occupy a functional head as the result of movement. 5. For a phonological account of “Raddoppiamento Sintattico,” see Chierchia (1986). 6. I am not referring to the dialect spoken in that area, which tends to degeminate all consonants and would not show the relevant contrast in (3), but to the pronunciation of Standard Italian in that area. 7. See Rizzi (1988) for a detailed description of prepositions in Italian. 8. See Giusti (1993, chap. 2) for a general proposal in which the Romance and Germanic article is analyzed as a surrogate of the case morphology present in Latin and Germanic, respectively. See Giusti (1995) for a detailed account in German, and Giusti (2001) for an account in Romance. 9. Along the same line of reasoning, see Dimitrova-Vulchanova and Giusti (1998) for a more detailed account of the analysis of the Balkan noun phrase, and Giusti (1994b) for a comparative analysis of the Scandinavian languages. 10. The same word orders are also found with other determiners and demonstratives, so as to make the definite interpretation of the noun phrase irrelevant to the word order. 11. For an analysis of article reduplication in Scandinavian, see Delsing (1993), Santelmann (1993), Svenonius (1993), Giusti (1994b), Börjars (1995), and Kester (1996).

84

DP FUNCTIONAL STRUCTURE

12. In Romanian and Albanian, most APs follow the noun, even if this is not inflected with the enclitic article. Only in Romanian is there evidence that N + art can move across a specifier that cannot be crossed by an uninflected N; see Giusti (1997) and DimitrovaVulchanova and Giusti (1998). 13. In most Scandinavian languages the enclitic article is reduplicated on the noun, but this can be treated as a subcase of article reduplication, which will be discussed in section 2.1.4 as evidence for the nonsubstantive nature of the article. 14. Some adjectives are obligatorily postnominal (e.g., nationality adjectives); some others are preferably postnominal but may be prenominal (e.g., topicalized descriptive adjectives; see Cornilescu 1995); still others must be prenominal (e.g., biet ‘poor’ in the sense of ‘pitiful’, and ordinals). This situation is identical to Italian, as depicted in Cinque (1994) and Giusti (1993). This state of affairs confirms the total unrelatedness of the enclitic nature of the article with N-movement. 15. So far, I have assumed a generally accepted conception of how to build the lexical projection NP and its modifiers. I will stick to it. I will follow Cinque (1994) in assuming that adjectives are specifiers of functional heads. The presence of a functional head for each specifier captures in a direct way the redundant morphological agreement on adjectives. The possible alternative that considers adjectives as intervening heads, as proposed by Delsing (1988), Lamarche (1991), or a mixture of the two, as in Bernstein (1997), is incompatible with Grimshaw’s extended projection approach taken as a guiding line here. 16. An anonymous reviewer reminds me that VP preposing/topicalization appears to run against this assumption. As a matter of fact, VP preposing/topicalization is the only counterexample to such a claim that I can think of. The case of the position bound by the clitic ne is shown to be a full nominal projection and not an intermediate nominal projection in Cardinaletti and Giusti (1992) and in Giusti (1993), to which I refer the reader. Although this is not the place to entertain the attempt to show that VP preposing actually complies with the above restriction, I would like to suggest that what VP-preposing actually moves is a complete functional projection (possibly a vP phase in terms of Chomsky 1999). 17. An anonymous reviewer points out that in Norwegian a definite article can be reinforced by a locative, as in huset der borte (house-the over there). This is also possible in Spanish. Brugè (1996, n. 27) convincingly argues for Spanish that two possible analysis should be given, corresponding to an ambiguity in interpretation. One possible interpretation is that of a reduced relative (“the book [which is] over there”). Another possible interpretation is with an understood deictic (“[that] book over there”). This latter interpretation is taken to support the assumption of a covert demonstrative, which is licensed by the presence of the definite article. Interestingly, the deictic interpretation is not avaiIable if a prenominal possessive is present. This is explained by the observation that demonstratives are in complementary distribution with possessives in Spanish since the two elements compete for the SpecDP position. I see no reason not to extend Brugè’s analsis to Norwegian. 18. See Giusti (1995) for an analysis of the formation of the definite article from the demonstrative in German, and Giusti (2001) for the results of the Latin demonstrative ILLE as article and pronoun in Romance. 19. In Albanian, the preadjectival article, as e in e bukur, is part of the adjectival root, and it is present regardless of the definiteness of the noun phrase and of the pre- versus postnominal position of the adjective. In Romanian, it is optional, it can only appear on some classes of adjectives (e.g., thematic adjectives are excluded), and it can only appear when the adjective is postnominal. For a more detailed presentation of the data, see Giusti (1993, 73– 79). The analysis of “adjectival articles” is not directly relevant to the point to be made here, but more properties of this kind of articles are considered in section 3.6.

A BARE PHRASE STRUCTURE APPROACH

85

20. For some speakers the sentence is acceptable in the irrelevant reading with the ellipsis of the second predicate è arrivata. 21. In both languages, generic noun phrases must have a definite article, as shown in (i) and (ii): (i)

a. *(Le) brave ragazze sono sempre noiose. b. Fete*(le) cumînti sînt totdeauna plictisitoare. ‘Good girls are always boring.’

(ii)

a. Detesto *(le) brave ragazze. b. Detest fete*(le) cumînti. ‘I detest good girls.’

Indefinite noun phrases must have an indefinite article in subject position, while they may have a null article in object positions, as shown by the contrast between (iii) and (iv): (iii)

a. *(Delle) brave ragazze abitano di fronte. b. *(Ni„te) fete cumînti stau alaturi. ‘(Some) good girls live next door.’

(iv)

a. Conosco (delle) brave ragazze. b. Cunosc (ni„te) fete cumînti. ‘I know (some) good girls.’

22. I am assuming here that a coordinated structure is a governing category for the application of the binding theory, as appears to be the case: (i)

a. Sono arrivati [ [Gianni]i e [ [suo]j/i fratello] b. Sono arrivati [ [Gianni]i e [il fratello di [quel disgraziato]j/*i ] ].

(ii)

a. [ [John]i and [ [his]j/i brother] ] have just arrived. b. [ [John]i and [ [that bastard]j/*i ’s brother] ] have just arrived.

In (i.a) and (ii.a) the possessive adjective in the second conjunct can refer to the noun phrase Gianni, John in the first conjunct. In (i.b) and (ii.b) the R-expressions quel disgraziato, that bastard in the second conjunct cannot refer to the noun phrase in the first conjunct, yielding a principle C violation. This shows that in a coordination, a pronoun in the second conjunct can refer to a noun phrase in the first conjunct, while an R-expression must be free from it. 23. In this chapter I will not take stand on the nature of the indefinite article, but elsewhere I have argued that the indefinite article is a marker for the partitive case assigned by a covert quantifier (see Giusti 1995). 24. This phenomenon is very general and can be found with other quantifiers such as tot(ul) ‘all’, întreg(ul) ‘whole’, vreun(ul) ‘some’, alt(ul) ‘[an]other’, mult(ul) ‘much’, putçin(ul) ‘little’. 25. I refer to Giusti (1997) for an account of the obligatory presence of the article in (30c–d). 26. Procrastination of possessive movement is possibly related to the fact that N can move across it, yielding a configuration in which N not only agrees but also c-commands the possessive. These two phenomena go together in the languages considered here. 27. As a matter of fact, in (34a) movement of the possessive from SpecNP to SpecFP1 is vacuous and therefore irrelevant in a bare phrase structure approach. It is only reported to show the parallelism with (33a) on the one hand and (34b) on the other. The actual structure for (34a) should be rephrased as in (i) below:

86

DP FUNCTIONAL STRUCTURE

(i)

FP2 F2'

mitt

FP1

[φ]

F1'

store

NP

[φ] mitt

N' hus

28. The ungrammatical string in (i) is excluded if we assume, as suggested in note 26, that the possessive cannot procrastinate movement to SpecFPmax if N has not moved across it: (i)

*det store mit hus

29. Principle (A2') can be subsumed by a general “minimize structure” principle (cf. Cardinaletti 1994) if we extend Borer’s (1997) proposal on the syntax-morphology interface to functional elements. According to Borer, a complex event nominal is derived in the morphological component by projecting the verbal head and then incorporating it into a nominal head. This complex nominal head is inserted in the syntax with the internal VP still able to assign the q-roles to its arguments. But it behaves like other nominal heads in all other respects. In the case under consideration here, the free morpheme det is formed by incorporating a root d- to the clitic -et. This procedure takes place in the morphology and not in the syntax, since det behaves as a simple head. However, its insertion is less optimal than insertion of the enclitic -et since it has a more complex structure. 30. To simplify the structure, I assume that the indefinite marker një, which is optional, is in SpecFP3, but I have no claim on its actual position. 31. Bernstein reports that, except for French, Romance languages do not have nondeictic demonstratives. This is not correct for Italian, which makes massive use of them. In Italian, parallel to what Bernstein notices for French, they cannot occur with a reinforcer. The same is the case for Spanish, according to Brugè (chapter 2 in this volume). Both Italian and Spanish may be considered to have the position of the nondeictic demonstrative so high that it cannot be crossed by N-movement. Unfortunately, Romanian does not have the nondeictic demonstrative in either position. 32. This does not hold for French ce, which displays many properties in common with clitics. However, it is still necessary to make a difference between a clitic or a weak element and a functional element that fills a head in the extended projection. One important difference between these two categories is that the clitic is linked to an XP position with independent semantic properties, while the functional head is merged as a last resort and only realizes functional features. This difference is also present in Bernstein’s analysis, in that it allows for movement of the demonstrative to SpecDP in those languages (for example, Greek) in which the demonstrative co-occurs with an article in D. However, in these languages the demonstratives are not morphologically “heavier” than in Italian, for example. 33. This is not to deny that in some languages—for example, Spanish (see Picallo 1994)—possessive adjectives may be clitics, as in some respects they behave like heads.

A BARE PHRASE STRUCTURE APPROACH

87

However, if the possessive adjective is a clitic, it is a functional element that originates in a maximal projection in a Specifier position and further moves into a functional head of the structure and never a purely functional head. For possessive clitics, see also Cardinaletti (1999). 34. Recall that merging of the article was made necessary by the presence of an adjective in Romanian prepositional phrases in example (30). 35. The Romanian example is taken from Lombard (1974, 96). It is interesting to notice for the sake of the following discussion that, although Lombard places the nouns following the pronouns in commas, he explicitly remarks that this is not obligatory. 36. Hopefully, this stipulation can be accounted for by a deeper study of the morphologic realization of f-features and case features in articles and pronouns in the two languages. I think that the enclitic versus free nature of the article is irrelevant to this question, as was for the contrasts discussed in examples (21) through (23). 37. This is motivated by the fact that the proper name itself bears the R-relation (which, according to Higginbotham (1987), is in SpecNP in all noun phrases), and the assumption is made here that referential features are checked in SpecFPmax. 38. Given the XP status of the proper name, this prohibition reminds us of a relativized minimality effect. It is not clear, however, why the possessive does not give the same effect. Notice that in Longobardi’s proposal relativized minimality could not be invoked to rule out the movement of the proper name. No reason is given there to distinguish between the possessive and the descriptive adjective. 39. I will leave open here what the actual structure of the adpositional construction is. 40. This can also be checked by construing full predication structures from (64): (i)

(ii) (iii) (iv)

a. *Francesco è Maria b. *Francesco Maria è Finzi Contini cf. Francesco Maria è un Finzi Contini Giovanni è il giardiniere Il dottore è Gianni Rossi a. *la zia è Vittoria b. *zia è Vittoria cf. (la) zia si chiama Vittoria

41. Bulgarian is the only language in which the prenominal position of adjectives is the only possibility. Albanian hardly bears one prenominal adjective, as will be discussed in detail later in this section. Romanian, like other Romance languages, prefers postnominal adjectives. However, if two prenominal adjectives are merged, only the leftmost one bears the definite article: (i)

primele frumoase(*le) fete first-the beautiful girls

42. The AdvP is in SpecFAPmax only for expository reasons. The only claim on the internal structure of the extended projection of the adjectival phrase here is that the enclitic article is an inflectional morpheme of the adjective. It is therefore not in the extended projection of the noun phrase. 43. Bulgarian presents neither of the expected orders, indicated in (ii) and (iii). Instead we find (i), where the article appears on the first conjunct: (i)

nova-ta new-the (ii) *nova-ta (iii) *nova i

i interesna kniga and interesting book i interesna-ta kniga interesna-ta kniga

88

DP FUNCTIONAL STRUCTURE

The ungrammaticality of (iii) is crucial for our point here. Notice that the position of the enclitic article in the presence of a complement of the adjective in Bulgarian was our first piece of evidence in favor of (72). Example (i) can also be explained in terms of (72), with additional assumptions independently necessary of other conjoined structures. See Dimitrova-Vulchanova and Giusti (1998, 344) for a discussion of this property of Bulgarian.

References Abney, S. P. (1987) “The English Noun Phrase in Its Sentential Aspect,” Ph.D. Diss., MIT. Bernstein, J. (1997) “Demonstratives and Reinforcers in Romance and Germanic languages.” Lingua 102, 87–113. Borer, H. (1997) “The Morphology–Syntax Interface: A Study of Autonomy,” in U. Dressler, M. Prinzhorn, and J. R. Rennison (eds.) Advances in Morphology, 5–30 (Trends in Linguistics 97). Berlin: Mouton. Börjars, K. (1995) “Feature Distribution in Swedish Noun Phrases,” Ph.D. Diss., University of Manchester. Brugè, L. (1996) “Demonstrative Movement in Spanish: A Comparative Approach,” University of Venice Working Papers in Linguistics 6(1), 1–53. Brugè L., and G. Giusti (1996) “On Demonstratives.” Paper presented at the 19th GLOW Colloquium, April 17–18, Athens, Greece. Campbell, R. (1996) “Specificity Operators in SpecDP,” Studia Linguistica 161–188. Cardinaletti, A. (1994) “On the Internal Structure of Pronominal DPs.” Linguistic Review 11, 195–219. Cardinaletti, A. (1999) “Pronouns in Germanic and Romance Languages: An Overview,” in H. van Riemsdijk (ed.) Clitics in the Languages of Europe, 5–82. Berlin: Mouton de Gruyter. Cardinaletti, A., and G. Giusti (1992) “Partitive ne and the QP-Hypothesis: A Case Study,” in E. Fava (ed.) Proceedings of the 17th Meeting of Generative Grammar, Trieste, February 22–24, 1991, 121–142. Turin: Rosenberg and Sellier. Cardinaletti, A., and M. Starke (1999) “The Typology of Structural Deficiency: A Case Study of the Three Classes of Pronouns,” in H. van Riemsdijk (ed.) Clitics in the Languages of Europe, 145–233. Berlin: Mouton de Gruyter. Chierchia, G. (1986) “Length, Syllabification and the Phonological Cycle in Italian.” Journal of Italian Linguistics 8, 5–33. Chomsky, N. (1993) “A Minimalist Program for Linguistic Theory,” in K. Hale and Keyser (eds.) The View from Building 20: Essays in Honor of Sylvain Bromberger, 1–52. Cambridge, Mass.: MIT Press. Chomsky, N. (1995) The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. (1999) “Derivation by Phase.” Unpublished ms., MIT. Chomsky, N., and H. Lasnik (1993) “The Theory of Principles and Parameters,” in J. Jacobs, A. von Stechow, W. Sternefeld, and Th. Venneman (eds.) Syntax: An International Handbook of Contemporary Research, 506–509. Berlin: Mouton de Gruyter. Reprinted as chapter 1 in Chomsky (1995). Cinque, G. (1994) “On the Evidence for Partial N-Movement in the Romance DP,” in G. Cinque, J. Koster, J.-Y. Pollock, L. Rizzi, and R. Zanuttini (eds.) Paths towards Universal Grammar, 85–110. Washington D.C.: Georgetown University Press. Cornilescu, A. (1995) “Rumanian Genitive Constructions,” in G. Cinque and G. Giusti (eds.) Advances in Roumanian, 1–54. Amsterdam: John Benjamins. Delsing, L.-O. (1988) “The Scandinavian Noun Phrase.” Working Papers in Scandinavian Syntax 42, 57–79.

A BARE PHRASE STRUCTURE APPROACH

89

Delsing, L.-O. (1993) The Internal Structure of Noun Phrases in the Scandinavian Languages. Lund: University of Lund. Dimitrova-Vulchanova, M., and G. Giusti (1996) “Quantified Noun Phrase Structure in Bulgarian,” in J. Toman (ed.) Formal Approaches to Slavic Linguistics, 123–144. Ann Arbor: Michigan Slavic Publications. Dimitrova-Vulchanova, M., and G. Giusti (1998) “Fragments of Balkan Nominal Structure,” in A. Alexiadou and Ch. Wilder (eds.) Possessors, Predicates and Movement in the Determiner Phrase, 333–360. Amsterdam: John Benjamins. Dobrovie-Sorin, C. (1987) “A propos de la structure nominal en Roumain.” Rivista di Grammatica Generativa 12, 126–151. Giusti, G. (1991a) “The Categorial Status of Quantified Nominals.” Linguistische Berichte 136, 438–452. Giusti, G. (1991b) “La sintassi dei nominali quantificati in Romeno.” Rivista di Grammatica Generativa 16, 29–57. Giusti, G. (1991c) “L’ordine NPQ in lingue QNP,” in R. Dolci and G. Giusti (eds.) Quaderno del CLI 4, 47–64. Venice: Centro Linguistico Interfacoltà. Giusti, G. (1993). La sintassi dei determinanti. Padua: Unipress. Giusti, G. (1994a) “Heads and Modifiers among Determiners,” in G. Cinque and G. Giusti (eds.) Advances in Roumanian, 103–125. Amsterdam: John Benjamins. Giusti, G. (1994b) “Enclitic Article and Double Definiteness: A Comparative Analysis of Nominal Structure in Romance and Germanic.” The Linguistic Review 11, 241–255. Giusti, G. (1995) “A Unified Structural Representation of (Abstract) Case and Articles: Evidence from Germanic,” in H. Haider, S. Olsen, and S. Vikner (eds.) Studies in Comparative Germanic Syntax, 77–93. Amsterdam: Kluwer. Giusti, G. (1997) “The Categorial Status of Determiners,” in L. Haegeman (ed.) The New Comparative Syntax, 95–123. London: Longman. Giusti, G. (2001) “The Birth of a Functional Category: From Latin ILLE to the Romance Article and Personal Pronoun,” in G. Cinque and G. Salvi (eds.) Current Studies in Italian Syntax: Essays in Honor of Lorenzo Renzi, 157–171. Amsterdam: Elsevier. Giusti, G., and N. Leko (1995) “On the Syntax of Quantity Expressions in Bosnian,” in R. Benacchio, F. Fici, and L. Gebert (eds.) Determinatezza e indeterminatezza nelle lingue slave, 127–145. Padua: Unipress. Grimshaw, J. (1991) “Extended Projection.”Unpublished ms. Brandeis University, Waltham, Mass. Grosu, A. (1988) “On the Distribution of Genitive Phrases in Rumanian.” Linguistics 26, 931– 949. Hellan, L. (1985) “The Headedness of NPs in Norwegian,” in P. Muysken and H. van Riemsdijk (eds.) Features and Projections, 89–122. Dordrecht: Foris. Higginbotham, J. (1987) “On Semantics.” Linguistic Inquiry 16, 547–593. Kayne, R. S. (1994) The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Kester, E.-P. (1996) “The Nature of Adjectival Inflection.” OTS Dissertation Series. Utrecht: University of Utrecht. Lamarche, J. (1991) “Problems for N-Movement to NUM-P.” Probus 3(2), 215–236. Larson, R. (1988) “On the Double Object Construction.” Linguistic Inquiry 19, 335–391. Lombard, A. (1974) La Langue Roumaine: Une présentation. Paris: Klincksieck. Longobardi, G. (1994) “Reference and Proper Names: A Theory of N-Movement in Syntax and Logical Form.” Linguistic Inquiry 25, 609–665. Passaler, E. (1997) “Zur Syntax der d-Pronomina im Deutschen,” M.A. Thesis, Università di Venezia. Picallo, C. (1994) “Catalan Possessive Pronouns: The Avoid Pronoun Principle Revised.” Natural Language and Linguistic Theory 12, 259–299.

90

DP FUNCTIONAL STRUCTURE

Postal, P. (1969) “On So-called ‘Pronouns’ in English,” in D. Reible and S. Schane (eds.) Modern Studies in English, 201–224. Englewood Cliffs, N.J.: Prentice Hall. Rizzi, L. (1988) “Il sintagma preposizionale,” in L. Renzi (ed.) Grande grammatica italiana di consultazione, 507–531. Bologna: Il Mulino. Santelmann, L. (1993) “The Distribution of Double Determiners in Swedish: Den Support in D°.” Studia Linguistica 47(2), 154–176. Svenonius, P. (1993) “Selection, Adjunction and Concord in the DP.” Studia Linguistica 47, 198–220. Tasmowski-De Ryck, L. (1990) “Le démonstrative français et roumains dans la phrase et dans le texte.” Langages 97, 82–99. Taraldsen, K. T. (1990) “D-projections and N-projections in Norwegian,” in J. Mascaró and M. Nespor (eds.) Grammar in Progress: Essays in Honor of Henk van Riemsdijk, 419– 432. Dordrecht: Foris.

4

Stacked Adjectival Modification and the Structure of Nominal Phrases - 

This article is a preliminary attempt to look at adjectival ordering restrictions (henceforth AOR) and aspects of the syntax of stacked adjectives across several languages1 within a broad Principles and Parameters (P&P) framework. I will be examining the notion of “fixed order” and proposing that it be considered part of Universal Grammar (UG). As a working strategy, I will attempt to explore the hypothesis that adjectival distribution follows the pattern of adverbial distribution along the lines of Cinque’s Universal Hierarchy of Clausal Functional Projections (Cinque 1999, 106). I will argue for a framework that treats adjectives not as adjuncts, but as specifiers of distinct functional projections that are intrinsically related to aspects of their semantic interpretation. At present, attributive adjectives are usually analyzed as recursively stacked adjunct specifiers of extended A' /N' or even recursive NP projections. Yet if all adjectives had the same categorial status, the syntax, being unable to distinguish the ungrammatical from the grammatical strings, would inevitably generate fragments like *a red heavy good table. Following Zwarts (1992), this article will attempt to demonstrate how the standard lexical/functional dichotomy may be approached through a rich semantic ontology, and how the semantic universe may be studied via an independently motivated system of syntactic categories.

1. Problematic nature of stacked adjective research It is fair to say, both within “traditional” linguistics and generative grammar, that very few linguists have looked at the issue of AOR in any detail.2 Sproat and Shih 91

92

DP FUNCTIONAL STRUCTURE

(1988, 486) call it a “hoary problem”; Hetzron (1978, 175) considers it one of “the most delicate topics in linguistics.” There are various reasons for this. First, there is no clear agreement in the literature as to the fundamental questions of (1) just how many semantic categories of adjective there actually are; (2) what order these categories are generated in; and (3) whether or not that order is universal. Moreover, various other syntactic/semantic and perhaps even pragmatic criteria come into play when dealing with stacked adjectival ordering. For example, there is the issue of “comma” intonation:3 asyndetically coordinated APs exhibiting intonation that “breaks the adjectives up into separate intonational phrases” (Sproat and Shih 1988, 465) thereby destroying what Hetzron call the adjectives’ “cohesion.” All writers distinguish between those strings that exhibit comma intonation (as in a luxurious, hot, steaming bath) and those that are spoken “without a break” (such as a bright red balloon). Comma intonation is an example of Sproat and Shih’s (henceforth S&S) parallel direct modification.4 In the preceding example, luxurious, hot, and steaming all modify the head noun directly, yet they also all modify the noun in parallel—together at the same time. Confusingly, however, parallel direct modification as defined by S&S need not necessarily be an example of comma intonation, and, similarly, their hierarchical direct modification5 need not involve the semantic categories of adjectives that are normally considered to take part in AOR. Sproat and Shih (1991, 592) give the example of senatorial industrial investigations and industrial senatorial investigations from Levi (1975). In the first example, industrial directly modifies investigations and senatorial directly modifies the hierarchical structure [ [industrial] investigations]. As S&S note, both orderings have different meanings: [The first] refers to the senate’s investigations of industry whereas the second refers to industry’s investigation of the senate. These semantic differences follow, as Levi shows, from the differing assignment of the deverbal nominal head’s thematic roles to the modifying adjectives. . . . Hence these ordering facts are tied to the thematic structure in a way in which the kinds of ordering restrictions which we will be investigating are not. (S&S 1988, 467 n.1)6

In other words, “in these cases there is no basic ordering; rather the ordering chosen depends upon the intended interpretation” (S&S 1991, 592). Second, there is the issue of adjective combinations that would normally exhibit AOR, but where an adjective has been preposed for reasons of topic or focal stress, as in red big ball. Given that any XP can be preposed to TopicP/FocusP, it is hardly surprising that we find adjectives that have been similarly fronted. This article, then, will concern itself mainly with those cases of unmarked, hierarchical direct modification, as defined by S&S, without overt comma intonation as applied to result and simple, noncomplex event nominals. In other words, orderings that are unmarked and basic “in the sense that one uses the prescribed orders unless one intends a special interpretation” (S&S 1991, 592 n.2). Following is a short critique of past analyses. As I have argued previously (Scott, 1998a, chap. 1), a major problem with the majority of essays that have discussed AOR in the past is that either they take the hierarchical ordering restrictions proposed in the literature for granted or they illus-

STACKED ADJECTIVAL MODIFICATION

93

trate only those clear cases of ordering restrictions between a few easily definable categories. Thus Hetzron assumes an extended version of Dixon’s (1982) hierarchy, and while it is true that he does refine Dixon’s classification somewhat, he nevertheless fundamentally accepts both Dixon’s ordering restrictions and his semantic categories. S&S discuss only clear-cut cases of AOR: QUALITY > SIZE > SHAPE > COLOR > PROVENANCE. The “murkier” categories (TEMPERATURE, AGE, PATTERN) are not even mentioned. It is fair to say that none of the literature has questioned these generally accepted ordering hierarchies or even asked how many categories actually exist. Furthermore, all writers7 maintain that AOR can be adequately accounted for using broad “psychological” criteria, yet none of them are able to provide a convincing argument—which is, moreover, consistent with the data—for a psychological basis to AOR. Among these accounts, it is Hetzron’s whose is the least convincing. He states that “the major rule is to place the more objective and undisputable qualifications closer to the noun, and the more subjective, opinion-like ones farther away” (Hetzron 1978, 178), and a large part of his article is spent trying to give evidence in support of this claim. Yet, to give just two examples, the category COLOR comes before the category ORIGIN in linear ordering because judgments of color are supposedly open to more disagreement than are judgments of nationality. Yet can one seriously argue that this is the case? Surely, if I show you a [red Russian ball], the ball’s color is immediately more apparent than its origin. With those categories that require more subtle gradations of judgment, Hetzron’s arguments sound even slightly tenuous. To account for the order in [a long thin blade], Hetzron writes that “thickness requires more careful observation, and is hence more reliable as a judgement, than length, a dimension that is too easily perceivable and is therefore taken more lightly” (1978, 180). Presumably, one must conclude that in this latter example Hetzron is arguing there would be more disagreement about a blade’s length than its thickness. S&S’s cognitive account suffers from the same drawbacks as those found in Hetzron; as S&S state at the end of their 1988 article (p. 486), “we have not as yet, however, described the link between the cognitive basis of the ordering restrictions and their linguistic manifestation; why should more ‘apparent’ adjectives be ordered closer to the head than less ‘apparent’ adjectives? . . . Unfortunately, we can give no answers at the present time.” So the accusation still stands that it is impossible to show conclusively that it is indeed the case that the more apparent an adjective is, the further right in the adjective string it goes. In their 1991 article, S&S suggest that AOR can be accounted for by stating that restrictions must apply only over those hierarchical strings that directly q-mark their sisters; but, as they are forced to admit, this is an ad hoc solution since “it is not clear from what general principles of grammar such a stipulation would follow and it is not clear what phenomena besides AOR such a stipulation would account for” (S&S 1991, 591–592). In both articles, S&S similarly do not question the standard assumed ordering hierarchies. Indeed, like Androutsopoulou (1996), they take as given the fact that AOR occur, and they are mainly interested in proposing possible structural configurations for the various cases where stacked adjectives do or do not display AOR. Theoretically, S&S propose that stacked adjectival modification “involve[s] iterative adjunction of adjectives to Nx” (S&S 1988, 473). So, in English, where adjectives modify an NP, the structure would be:

94

DP FUNCTIONAL STRUCTURE

(1) [NP rapid [NP Russian [NP invasion of Afghanistan] ] ] (S&S 1991, 571)

If adjectives are adjoined iteratively to the head noun, how can S&S account for the fact that they display ordering restrictions? A syntactic analysis as in (1) will of course generate a rapid Russian invasion, but it will also generate fragments like *a new red bright dress. Although they are adequate as far as dealing with the differences between those cases where multiple stacked adjectives do and do not display AOR, the configurational structures presented in both S&S and Androutsopoulou are not able, as far as the cases that do display AOR are concerned, to account for and predictively generate the actual ordering restrictions themselves. 2. Stacked adjectives and X' -theory Stacked adjectives, like stacked adverbs, are generally analyzed in X' -theory as recursive adjunct projections. Such an analysis is not without its problems, however. To begin with, there is evidence that adjectives (like adverbs) are not adjuncts, since one of the basic criteria for judging adjuncts (Freidin 1992, 43) is that their ordering appears to be free: (2)

a. a book [about physics] [in German] b. a book [in German] [about physics] (Freidin 1992, 43)

Similarly, relative clauses also display free ordering: (3) the man [who paid the bill] [wearing a stripy scarf] [of medium build] [with a gold tooth] (Cook and Newson 1996, 145)

Example (3), containing two relative clauses and two PP adjuncts, can be ordered in any of its possible combinations without focusing any of its elements. This cannot be done with either stacked adjectives or stacked adverbs, since even if a nonstandard order is still grammatical, one of the elements will receive a definite focal/ contrastive stress: (4)

a. John repeatedly viciously attacked Bill. b. *John viciously repeatedly attacked Bill.

(5)

a. Mine’s the big red car. b. *Mine’s the red big car.

Here, the (b) examples are “ungrammatical” unless viciously/red is said with focus (i.e., emphatic or contrastively stressed) intonation. An analysis in which adjectives are adjoined iteratively to the head noun is inadequate because it is unable to account for the fact that adjectives display clear ordering restrictions. Moreover, this type of structural analysis is theoretically undesirable as it would make NPs unique in X'-theory: no other category allows recursive adjunction to an intermediate projection. Abney recognizes the same fact when he writes that such an analysis not only “espouses adjunction in the base, [but] furthermore, adjunction to a non-maximal category. An embarrassing question is why there are no elements adjoined to any other single-bar projection at d-structure: not V', P', A', I', C', etc.” (Abney 1987, 323).

STACKED ADJECTIVAL MODIFICATION

95

A problem with recent analyses such as Cinque (1994) and Giusti (1997) is that, even though in these works stacked adjectives are analyzed as specifiers of functional heads, they are not distinct functional heads. Apart from the lexical/functional difference, theirs are essentially the same type of adjunction analysis as the standard X’ treatment; consequently, of course, they will not be able to take into account the fact that adjectives occur in AOR. Moreover, Giusti’s proposed structural configuration (see Giusti 1997, 97) seems to suggest that whole AgrPs may be recursively projected. Under this analysis, adjectives (and presumably adverbs) would again be unique in X'-theory as they would be the only functional phrases capable of being iteratively stacked—DPs, for example, cannot further project into DPs. Indeed, such an analysis would seem to run counter to Grimshaw’s (1991) thesis of “extended projections.” 2.1. A short summary of Cinque (1999) Within the background of the current proliferation of research into functional projections (FPs), Cinque (1999) proposes a correlation between adverbials and semantically related FPs that are associated with tense, aspect, modal force, and so on. As Cinque himself states, the aim of his work is to justify the hypothesis that adverb phrases (AdvPs) are the unique, overt manifestations of the specifiers of distinct maximal projections (and not recursive IP/VP adjuncts) and to argue for the existence of a fixed universal hierarchy of clausal FPs. He does this by producing evidence for a functional-semantic hierarchy of AdvPs across a range of languages; by independently producing evidence for a hierarchy of clausal functional-semantic projections, again across a range of languages; and then, finally, by amalgamating both hierarchies and showing that there is a systematic one-to-one mapping between the AdvPs and the clausal functional heads to which they are associated. Cinque concludes: “The transparent semantic relation holding between each adverb class and the contiguous head morpheme to its right (when the two hierarchies are matched) has been taken to suggest that each AdvP is the specifier of the phrase projected by the corresponding functional head morpheme” (Cinque 1999, 140). Cinque’s hypothesis, then, is that each adverbial class must be occurring as the specifier of a phonologically null or “empty-headed” FP to which it is semantically related, the order of adjectives at S-structure/Spell-Out being a direct mapping from the ordering of the FPs themselves. As an example, and to closely paraphrase Zanuttini (1997, 228)—who discusses an earlier version of Cinque’s work in some detail with respect to negative markers and movement of nonfinite verbs in Romance—Cinque is able to show how adverbs belonging to the class “already” occur as specifiers of a temporally related projection that expresses the relationship between the event time and reference time; how adverbs of the class “no more” occur in an aspectual projection expressing perfectivity; and how past participles occur cross-linguistically in any head position made available by an FP that has an adverb as its specifier (i.e., in all the head positions labeled “e” in the following examples). For example: (6) [TP2already [T’ e [AspPperf no more [Asp’ e [AspPgen/prog always [Asp’ e ] ] ] ] ] ] (based on Zanuttini 1997, 228)

96

DP FUNCTIONAL STRUCTURE

Cinque suggests that there may be somewhere in the region of forty FPs that are present in all languages and proposes that, even where a particular projection is not overtly realized morphologically, it is still present structurally in the syntax. Finally, it should be noted that while correlations between adverbs and FPs associated with, for example, aspect and tense have been proposed before in the literature, Cinque’s work (and his Universal Hierarchy of Clausal Functional Projections in particular) stands out as the most comprehensive and clearly articulated cross-linguistic statement on the topic to date. 2.2. Extending Cinque’s thesis to adjectival projections If, as is generally accepted, “the structure of DP is essentially parallel to that of CP,” and that “NP is the nominal counterpart of VP, and DP is the nominal counterpart of CP” (Valois 1991, 12 and 34, respectively), it is a natural and logical step to go from looking at FPs associated with the clause to looking at FPs associated with the DP. Although there is little functional morphology in the DP, because adverbs are the sentential counterpart of adjectives, we might expect to find similar FPs hosting adjectives. Indeed, this notion has already been proposed in the literature8 but (1) either the nature of the FPs has been left unspecified (as in Cinque 1994, who proposes a general FP, “F”) or (2) the FPs are limited to certain types of nominal (i.e., event nominals). As noted earlier, positing an analysis in which stacked adjectives are the specifiers of FPs of particular semantic classes has one distinct advantage over the traditional adjunction analysis: it predicts ordering restrictions. However, there are other conceptual and empirical reasons for adopting the stance advocated here. A number of these are also given in Cinque (1994): 1. The linear ordering of each stacked adjective can be viewed as a direct mapping from the hierarchical ordering of the FP with which it is associated. 2. Why does there seem to be a limit to the number of adjectives found within the DP (according to Cinque 1994, 96, “apparently not exceeding six or seven”)? A priori there need be no such limit and, indeed, the adjunction hypothesis would predict that no such limit exists. The generation-in-Spec analysis at least in part predicts why: the number of adjectives is limited by the semantic possibilities for modification of the number of projections found between D and NP. 3. A theory allowing only Spec-Head relationships could presumably be said to be more “minimalist” in spirit than one that countenances both specifiers and adjuncts.9 4. Within the traditional adjunction hypothesis, the fact that APs appear to the left of their head has to be specifically stipulated in the phrase structure rules (and, of course, the existence of multiple Specs is allowed), whereas Kayne’s (1994) antisymmetric view of syntax predicts there is always only one single specifier per projection and that that specifier must be left-branching10— thus, under the present assumptions, Kayne’s abstract theory receives further empirical support.

STACKED ADJECTIVAL MODIFICATION

97

5. An analysis in which adjectives are the specifiers of FPs associated with their respective semantic classes leads to a tighter, more articulated correspondence between universal semantic properties and the syntax than in the recursive adjunction approach. 6. Cinque’s (1994) N-raising account suggests independent evidence that FPs exist within the DP, between DP and NP. 7. In many languages, stacked adjectives display case and f-features, so within a minimalist view of syntax (Chomsky 1995a), adjectives can only be in Spec positions: the Spec-Head relationship is the only structural configuration where they can get their features “checked” and erased11 since only specifiers are within the checking domain of a head; adjuncts are not. 8. (Tentatively) FPs are projected from closed lexical classes (e.g., determiners, pronouns, and classifiers); if adjectives are semantically related to the functional heads of the various FPs in the Spec of which they find themselves generated, then we might expect some “adjectival” interpretation or function to sometimes be realized as a closed class item. In some languages, at least, this does indeed seem to be the case. 9. Finally, if AOR are a direct and overt manifestation of the ordering of FPs, then conjectures as to the psycholinguistic motivation for AOR need not be posed: AOR fall out as a direct consequence of UG. Most of the preceding arguments, speak for themselves, I think. However, I would like to elaborate briefly on two points. First, as mentioned tentatively in point 8, “adjectives” do indeed seem to form a closed class in some languages. Dixon states that “not all languages have the major word class Adjective. Either they have no Adjective class at all, or else there is a small non-productive minor class that can be called Adjective”; he also notes that “languages that have only a limited class of adjectives show considerable similarity in the concepts that are expressed through adjectives” (Dixon, 1982, 2 and 3, respectively). So, for instance, he discusses the cases of Igbo with only eight adjectives; Hausa with about twelve; and the Australian aboriginal language Malak Malak with seven.12 Interestingly, the adjectives that these languages possess fall into (one or two members of) the same semantic categories—usually SIZE, LENGTH, AGE and COLOR. Dixon shows (1982, 7) that out of twenty languages he surveyed which had “closed” adjective classes, large occurred in all twenty languages; small in nineteen; short in fifteen; and long, old, bad, and white in fourteen. The second, minor, issue I wish to note is that, with regard to point 9, an obvious advantage of positing that AOR are part of UG is that the complexity of the learning procedure necessary for a child to master the restrictions is minimized (and the “logical problem of language acquisition” answered), and the linguist need not have recourse to such difficult-to-define psychological notions as S&S’s amount of comparison/computation (S&S 1988, 467–472) or absoluteness (S&S 1991, 588–591). Finally, due to the poverty of nominal functional morphology, the existence of functional heads related to the semantic classes associated with AOR is practically impossible to motivate independently within the DP (in the way that Cinque 1999 is able to look at clausal FPs related to AdvP classes independently of the classes themselves). It may be, however, that the augmentative/diminutive affixes found in the

98

DP FUNCTIONAL STRUCTURE

derivational morphology of adjectives in languages like Russian and Portuguese are overt manifestations of functional heads within the “AP.” In Russian we find such pairs as byeliy ‘white’ with its diminutive form byel-yen’k-iy ‘nice and white’ (the incorporated diminutives are in boldface italics in these examples) used to express endearment; and bol’shoy ‘big’ with its augmentative counterpart bol’sh-ooshch-iy ‘very large’ used to express awkwardness or ugliness.13 One could argue that these are overt manifestations of a SIZE-related functional head; similarly, classifier systems in those languages that have them may be the overt manifestations of a SHAPE-related FP. In a similar vein, Cinque (1994, 106 n.27) speculates whether adjectives like the Italian attuale ‘present’, passato ‘past’, and futuro ‘future’ are instantiations of FPs within the DP related to TenseP and AspectP. Hockett (1958, 238) describes the existence within the nominal system of Potawatomi, an Algonquian language of North America, of a morphologically overt manifestation of the category Tense or Aspect that is identical to that found within the Potawatomi verbal system: viz. nk©šat©s ‘I am happy’ (a verb) and nciman ‘my canoe’ (a noun) versus nk©šats©p©n ‘I was formerly happy (but am not now)’ and ncimanp©n ‘my former canoe (now lost, destroyed or stolen)’. And, of course, it is quite possible that there exist other languages in which such DPinternal functional heads are not empty.

3. Adjectives and semantic classes Since adjectives are the specifiers of FPs, what actual FPs are they the specifiers of? If adverbs are, as Cinque suggests, the specifiers of such FPs as Moodspeech actP, AspproximateP, T(Future)P—that is, of the FPs that reflect the semantic classes by which they pattern in linear ordering—it seems reasonable, and by Uniformity indeed theoretically desirable, to assume that adjectives, their nominal counterparts, are likewise treated as the specifiers of FPs that reflect the semantic classes according to which they are ordered. This entails looking at the classes of adjectives themselves, to research just which adjectives belong to which class; exactly how many classes of adjectives there are; whether certain adjectives can belong to more than one class (as is the case with adverbs); and finding out whether these classes and the ordering restrictions in which they occur are found cross-linguistically. A full discussion of all these questions is naturally out of the scope of an article this size: all I can hope to do is highlight some of the questions that need answering in future research. Because strings with more than three stacked adjectives at once (for example Hetzron’s 1978, 179 a happy young blind Belgian sheep dog) are comparatively rare and moreover quite long, they tend to make it difficult to judge which of the various possible orderings is the most common or most natural. Fragments of this sort allow greater freedom mathematically for elements to be preposed or focused: thus, a young happy blind Belgian sheep dog; a blind happy young Belgian sheep dog; ?a young blind happy Belgian sheep dog, and so on, are all possible. With this in mind, I shall therefore build a proposed hierarchy using two, maximally three, adjectives at a time. The AOR list I shall be using as a basis will be that of Kingsbury and Wellman:14

STACKED ADJECTIVAL MODIFICATION

(7)

99

> SUBJECTIVE COMMENT > SIZE > AGE > SHAPE > COLOR > NATIONALITY/ORIGIN > MATERIAL > COMPOUND ELEMENT > NOUN (1986, 40) DETERMINER

Kingsbury and Wellman’s (henceforth K&W) cline of categories is able to deal with orderings such as the following (where any other ordering in these languages is decidedly marked): (8)

a. suuri vanha musta englantilainen koira (Finnish) big old black English dog b. idiÇk akaan ndaidat ÇfÇng (Ibibio) ugly old red dress c. et∧k ekara okpokoro *ekara et∧k okpokoro (Ibibio) small round table round small table

Their cline of categories suggests a relatively rich functional hierarchy. Yet there is evidence that (1) some of these categories could be further decomposed and (2) there may well be other categories not even mentioned by them. For example K&W’s category cline is unable to account for the following ordering restriction (which appears to be SIZE > TEMPERATURE): (9) pitkä kuuma kesä long hot summer

*kuuma pitkä kesä15 hot long summer

(Finnish)

As I have previously suggested, it is necessarily out of the scope of an article this size to give full and adequate attention to all the questions posed here. Instead, I shall limit myself to illustrating how complex the issue is by attempting to formulate an answer to the following three questions: 1. Can the category SizeP be further decomposed? 2. Is there a SpeedP? 3. Is there a TemperatureP? Each of these three questions, if discussed extensively, could be the topic of a fulllength article on its own. Moreover, there are, of course, theoretically other possible categories (e.g., DEPTH, STRENGTH) that I am not discussing. Thus I repeat, I simply wish to highlight some of the complexities involved. 3.1. Refining the standard stacked adjective ordering hierarchy As Hinton and Marsden (henceforth H&M) note (1985, 80), it seems that the category SIZE can indeed be “split” into at least two more categories HEIGHT and WIDTH, with HEIGHT preceding WIDTH: (10)

a. pitkä lihava mies *lihava pitkä mies (Finnish) tall fat man fat tall man b. a short skinny woman *a skinny short woman c. a low narrow ceiling *a narrow low ceiling d. uyarnna kattiyulla bhithi *kattiyulla uyarnna bhithi high thick wall *thick high wall

(Malayalam)

100

DP FUNCTIONAL STRUCTURE

Within the umbrella category of SIZE, in addition to adjectives related to width and height, are also adjectives related to length and weight. These, too, seem to follow AOR. To begin with, it seems that LENGTH precedes all other categories: (11)

a. a long high cliff face *a high long cliff face b. dugacka uska ulica *uska dugacka ulica (Serbo-Croat) long narrow street narrow long street c. lyhyt ohut terä *ohut lyhyt terä (Finnish) short thin blade thin short blade d. a long heavy canoe *a heavy long canoe e. pitkä korkea laiva *korkea pitkä laiva (Finnish) long tall ship tall long ship f. haf hir poeth *haf poeth hir16 (Welsh) summer long hot summer hot long g. en lång vid kjol *en vid lång kjol (Swedish) a long wide skirt a wide long skirt

Similarly, it appears that WEIGHT follows all other categories; in particular, it seems to follow the category WIDTH:17 (12)

a. visoka teška statua tall heavy statue b. paksu painava kirja fat heavy book c. kanatha kanamkuranja thick heavy book d. a slim light volume

*teška visoka statua (Serbo-Croat) heavy tall statue *painava paksu kirja (Finnish) heavy fat book buk© *kanamkuranja kanatha buk© (Malayalam) *heavy thick book *a light slim volume

So it would seem that the FP SizeP can be further split into the following functional phrases: LengthP > HeightP > WidthP > WeightP.18 This hierarchy would consequently predict that a string such as a long tall narrow ship with any other ordering would produce a slight focal stress (and indeed this seems to be the case). Dixon (1982, 24) suggests that SPEED should be considered a separate adjectival semantic class and that it should be placed after the category he calls PHYSICAL PROPERTY (which includes hard, heavy, hot, rough) but before the categories HUMAN PROPENSITY (wicked, clever, happy) and AGE. Certainly it seems that adjectives of the category SPEED (fast, slow, quick) do take part in AOR: (13)

a. ÇsÇp ufa moto *ufa ÇsÇp moto (Ibibio) fast new car new fast car b. ein schnelles altes Pferd *ein altes schnelles Pferd (German) a fast old horse an old fast horse c. afon araf lydan *afon lydan araf (Welsh) river slow wide river wide slow d. ein langsamer breiter Fluss *ein breiter langsamer Fluss (German) a slow wide river wide slow river

But where in the ordering hierarchy should this category come? The following data from English may suggest a positioning: (14)

a. a long fast road b. a long slow river

*a fast long road *a slow long river19

STACKED ADJECTIVAL MODIFICATION

c. a tall fast ship d. a low fast plane

101

*a fast tall ship *a fast low plane

At first sight, it may seem that SPEED comes after the class SIZE. Note, however, the following: (15)

a. dugacka spora reka *spora dugacka reka (Serbo-Croat) long slow river slow long river b. hidas levä joki *levä hidas joki (Finnish) slow wide river wide slow river c. spori teški tenk *teški spori tenk (Serbo-Croat) slow heavy tank heavy slow tank d. tanc araf trwm *tanc trwm araf (Welsh) tank slow heavy tank heavy slow

Interestingly, it seems that the fact adjectives belonging to the class SPEED engage in AOR lends direct support to the hypothesis that the semantic class SIZE should be split. It appears, at least in the above languages and in English, that adjectives from the class SPEED come after adjectives of LENGTH and HEIGHT, but before adjectives of WIDTH and WEIGHT. If the semantic class SIZE were a single, homogenous class, then these facts would be difficult to explain. This then suggests a further refinement to the category cline: LengthP > HeightP > SpeedP > WidthP > WeightP.20 Finally, as the data in (9) indicate, there seem to be definite ordering restrictions with regard to adjectives from the class TEMPERATURE, at least as far as English and Finnish are concerned. This pattern seems to be repeated across other languages (see (11f) for Welsh): (16)

a. ein langer heiber Sommer *ein heiber langer Sommer (German) a long hot summer a hot long summer b. dugo toplo leto *toplo dugo leto (Serbo-Croat) long hot summer hot long summer c. niinda thanutha varshakâlam *thanutha niinda varshakâlam (Malayalam) long cold rainy season cold long rainy season d. en lång kall vinter *en kall lång vinter (Swedish) a long cold winter a cold long winter

Thus it seems as if one might wish to include a TemperatureP within the hierarchy of FPs. Yet where exactly would a TemperatureP come? Evidence from English (and the above languages) suggests after SUBJECTIVE COMMENT and adjectives related to SIZE, but before SHAPE: (17)

a. a beautiful long hot summer *a beautiful hot long summer b. ?a heavy hot round pan *a hot heavy round pan c. a cold square lump of ice *a square cold lump of ice

However, within K&W’s (and all other) category cline, the category AGE also comes between SIZE and SHAPE. This means that either the categories AGE and TEMPERATURE are interchangeable (as is suggested by Fowler 1992, 138) or these categories display AOR with respect to each other. In fact, the latter conclusion seems to be the case (although, probably for semantic reasons, it is uncommon for adjectives from the classes TEMPERATURE and AGE to come together):

102 (18)

DP FUNCTIONAL STRUCTURE

a. an extremely hot old oven *an extremely old hot oven b. a red-hot Victorian marble fireplace *a Victorian red-hot marble fireplace c. a cold young puppy *a young cold puppy

If AGE and TEMPERATURE had displayed interchangeable orderings that did not affect focus, then we could reasonably have expected them to be members of the same category (or of a category AGE /TEMPERATURE).21 Since this does not seem to be the case, I suggest that they do belong to separate FPs with the order AGE > TEMPERATURE. Thus our final, enriched (but still incomplete) 22 hierarchy of projections looks as follows: (19) A proposed universal hierarchy of AP-related functional projections (a first approximation) [DPthe/this [Subj.CommentPnasty/magnificent [SizePbig/small23 [LengthPlong/short [HeightPtall/high [SpeedPfast/slow [WidthPfat/thick/thin [WeightPlight/heavy [TemperaturePhot/cold [AgePEdwardian/modern [ShapePround/triangular [ColourPdark brown/white [Nationality/OriginPGerman/Parisian [MaterialProsewood/brass [NP . . .

3.2. Some theoretical consequences of this approach If we take a minimalist framework like Chomsky’s Bare Phrase Structure model (Chomsky 1995b), then we can assume that phrasal categories are not distinguished by their level and that heads and maximal projections are all the same category; the only difference being that a nonbranching X is a head and that “a category that does not project any further is a maximal projection[;] . . . any other is an X' “ (Chomsky 1995b, 396). Thus an adjective is an AP in the Spec of a semantically related FP; it is both an X° (a head) and an Xmax (a phrase). This has the intuitive advantage that adjectives can still be treated as fully lexical elements and so therefore have lexical meaning (consequently possessing all the ambiguity and exceptionality24 associated with the lexicon—see sections 4.1 and 5.1 for respective discussion on the adjectives cool and good in particular); at the same time, this model allows for the FPs, in whose Spec each adjective is generated, to determine both the adjectives’ hierarchical syntactic ordering and their combinatorial semantic interpretation. Such an analysis accords well with one of P&P theory’s fundamental tenets that “lexical heads provide the descriptive content; . . . functional heads determine the configurational geometry . . . thus contributing to the determination of the interpretation as well as the form of the linguistic expression” (Belletti and Rizzi 1996, 3). It also accords well with standard versions of the Minimalist Program, whereby all FPs are headed by a grammatical (in this case predominantly semantic) feature bundle and all elements are “generated” in lexical projections (LPs) and “licensed” in FPs (see Zwart 1997, 250): stacked adjectives are generated as lexical maximal projections (APs) and are licensed by hierarchically ordered FPs that are intrinsically related to aspects of their semantic interpretation.25 An important consequence of this approach is that adjectives can have a “core” meaning, but that same core meaning may be interpreted differently according to which FP the adjective has been generated under: thus one could argue an adjective like old has been generated alternately in an AGE-related FP in a fragment like the old

STACKED ADJECTIVAL MODIFICATION

103

man with the walking stick, in some kind of TEMPORAL-related FP in my old (=former) boss or an old (=ancient) joke, and even in an adjunction structure within a SUBJ. COMMENT-related FP in a phrase such as good old Fred (whereby it is adjoined to the adjective good). Likewise, an adjective may belong to a “core” semantic class, but which class it is interpreted as belonging to depends on the semantic interpretation of the head it is associated with (thus, the same adjective at different times seems to appear to belong to different classes). It has to be noted that an adjective is not the overt, lexical instantiation of an FP per se (this is also the case, respectively, with adverbs and clausal FPs for Cinque). FPs are not adjectival (or even lexical) in nature themselves: they are instantiations of bundles of “functional” (=syntactico-semantic) features. What is important is that an FP is a projection that can host in its Spec any element related to its semantic interpretation or field: APs, AdvPs, NPs, PPs, and, of course, morphological affixes (related to tense, aspectual features, etc.). A consequence of this is that it enables us to provide structures for the following: (20)

a. an off-the-cuff remark b. a ready-to-eat meal

To take [an [off-the-cuff] remark] as an example, there is clear evidence through coordination and modification phenomena that off-the-cuff behaves syntactically and semantically as an adjective: (21)

a. We found his remarks rather [AP off-the-cuff] and [AP uninteresting]. b. His remarks were both [AP off-the-cuff] and [AP weird]. c. His remarks were ridiculously/extremely/very/too [AP off-the-cuff].

Yet it also seems to behave like a PP, as only PPs can be modified by straight and right (i.e., straight/right down the road): (22)

a. b. c. d.

His remarks were straight [PP off-the-cuff] and [PP to-the-point]. His remarks were right [PP off-the-cuff]. *His remarks were straight [AP uninteresting]. *His remarks were right [AP uninteresting].

But in X' -theory Ds do not take PPs as complements so, in the case of [an offthe-cuff remark] and morphologically complex “adjectives” like it, for theoretical reasons we probably do not want to claim that this is a special or unique instance of a D being followed by a PP. If we assume for now that off-the-cuff somehow has dual AP/PP status, then structurally this AP/PP must nevertheless be in the Spec of some phrasal category—be it AP or PP: (23)

DP DP an

? XP ? X'

PP off-the-cuff ?X

NP remark

104

DP FUNCTIONAL STRUCTURE

Under Abney’s (1987) analysis of nominal modification, off-the-cuff, being headed by the preposition off has to be analyzed as a PP that somehow “inherits” AP status presumably from its NP complement. With the analysis presented in this article, together with Bare Phrase Structure assumptions, off-the-cuff remains a PP but receives its semantic interpretation from the FP it finds itself in the Spec of (presumably some kind of MannerP): (24)

DP DP an

MannerP PP off-the-cuff

MannerP' e

NP remark

In much the same way, we can also preserve the distinction between nominal modification and the adjectival category in such phrases as summer desserts, law school regulations, music teacher, and night bus in which the nominal can remain an NP and is simply an NP in the Spec of an adjectival FP from which it receives its interpretation.26 Finally, this approach has one further advantage over Abney (1987) who analyzes [red cars] as an AP. If we assume, as in current theory, that two elements from the Numeration are concatenated by a transformation Merge that projects the features of only one out of the two elements (Chomsky 1995b, 397), intuitively, I think, we would want the head/lexical features of the noun to be projected since this fragment denotes a set of entities. If the lexical head features of the adjective were projected as in Abney’s approach, this projection would incorrectly predict that the resulting larger structure be interpreted as a property—that is to say, a predicate. The structure in (25), however, merely states that the simple noncomplex noun in red cars has an extended projection into a non-event functional hierarchy. This analysis allows the adjective to be a head and an AP, but since it is the specifier of an FP, which is simply an extended projection of grammatical features from the noun, the noun itself can still be viewed as the head of the phrase because (unlike in Abney) it has not projected into a projection that is lexical: (25)

ColourP AP red

ColourP' e

Nationality/OriginP

STACKED ADJECTIVAL MODIFICATION

105

4. Semantic classes and DP-internal functional projections The comment made in passing during the previous section about ordering interchangeability and semantic class deserves some elaboration. As far as I am aware, this was first noticed by Dixon who writes that “there appears to be an underlying order between types but not within types” and who concludes that the interchangeability of [a brave clever man] with [a clever brave man] is proof that they belong to the same category (1982, 25).27 This fact was used here in conjunction with the AOR displayed by TEMPERATURE and AGE adjectives to show that they belong to different classes. Within a class, it does indeed seem that AOR do not occur: (26)

a. egy okos bátor ember a clever brave man

egy bátor okos ember (Hungarian) a brave clever man (Hetzron 1978, 171) (both COMMENT) b. a thick fat book a fat thick book (both WIDTH) c. a new modern kitchen a modern new kitchen (both AGE)

SUBJECTIVE

If it is the case that adjectives from the same class are interchangeable,28 this fact and the fact that adjectives are specifiers of the FPs of the semantic classes in which they group together make certain predictions: 1. Adjectives with the same orthography but which can occur in different positions must be able to be specifiers of more than one (i.e., different) FP. 2. Conversely, adjectives (like ancient and old) that appear to belong to the same semantic category but whose ordering facts suggest otherwise must therefore be specifiers of different FPs. 4.1. Same adjective, different FPs The first point highlights yet again the semantico-syntactic similarity between adjectives and adverbs, once more suggesting the same syntactic analysis for both classes. Take the following (now classic) facts from Jackendoff (1972, section 3.2): (27)

a. b. c. d.

John John John John

has answered their questions cleverly. cleverly has answered their questions. has cleverly answered their questions. has cleverly answered their questions cleverly/stupidly.

Cinque (1999, 19–22) shows how the adverb cleverly is base-generated in three different structural positions in the tree and receives different semantic interpretations, depending on where it has been generated.29 Cinque argues that since “a restricted theory should force a one-to-one relation between position and interpretation (i.e., one specific, and distinct interpretation for each position of ‘base generation’),” (1999, 20), it is empirically and theoretically preferable to locate each of the three cleverlys as the specifiers of distinct, semantically related FPs (so, for example, the cleverly in 27b would be the specifier of a Modability/permissionP). Applying this

106

DP FUNCTIONAL STRUCTURE

idea to adjectives would account for the fact that an adjective’s interpretation is a function of where in the tree it has been base-generated. For example, I would argue a phrase such as [that really cool long red dress!] has the following structure: (28)

DP D that

D' AdvP really

Subj. CommentP AP cool

Subj. CommentP' e

SizeP e

LengthP LengthP'

AP long e

ColourP ColourP'

AP red e

NP dress

Intuitively, cool is not the specifier of a TemperatureP here: it does not have the semantic reading not hot. Instead, as the specifier of a Subj.CommentP, it receives the interpretation great or excellent. Some syntactic evidence shows this to be the case:30 (29)

a. What a long cool red dress! [“ungrammatical” where cool receives the interpretation excellent] [“grammatical” where cool receives the interpretation not hot] b. What a cool long red drink! [“ungrammatical” where cool receives the interpretation not hot] [“grammatical” where cool receives the interpretation excellent]

If we follow the hierarchy of FPs suggested in (19), it becomes clear why cool in (29a) cannot receive the interpretation of great or excellent—it has, in fact, been generated in the Spec of a TemperatureP and therefore receives the interpretation “cool to wear.” Similarly, cool in (29b), having been generated in a Subj.CommentP position, is unable to mean “not hot” and is forced to take on a “subjective comment” reading. Thus this analysis provides a syntactic account for an area of the grammar that is usually viewed as the domain of the lexicon (the dual readings for cool) or even pragmatics, and it also illustrates clearly how the semantic universe may be studied via theoretical syntax.

STACKED ADJECTIVAL MODIFICATION

107

I would argue that the preceding analysis is also able to account for the following data (often discussed in the literature and seen as a “problem” for the syntax and so left to the semantic component of the grammar to deal with): (30)

a. a small large firm b. a large small firm

(31) he is a green young writer

It is interesting to note that there is no way that (30a) and (30b) can be instances of S&S’s parallel modification, since this would lead to a contradiction (??a firm which is small and large), so it must be that they are both instances of direct modification. Furthermore, the fact that in the fragment [a small large firm], small directly modifies [large [firm] ] is shown by the semantics: the firm is large, but small for a large firm. How can this reading be accounted for syntactically? The analysis here suggests an answer. In (30a), for example, large is generated closer to the head noun within SizeP and takes on a size-related interpretation, and small is generated further up the tree within Subj.CommentP and takes on a subject comment interpretation (and vice versa, of course, for [a large small firm]): (32)

DP D a

Subj. CommentP AP small

Subj. CommentP' e

SizeP AP large

SizeP' LengthP

e Spec

LengthP' e

HeightP

e

NP firm

Likewise with [a green young writer], according to the hierarchy in (19), green, when indicating COLOR is generated after AgeP; here, it has been base-generated before AgeP in Subj.CommentP and has the meaning inexperienced. Thus we are able to account for the difference in meaning between he is a young green (=color) Martian and he is a green (=inexperienced) young Martian and, similarly, between an ugly cold (=temperature) woman and a cold (=heartless) ugly woman. Furthermore, this suggests why the standard word order is he is a young black writer: he is a black

108

DP FUNCTIONAL STRUCTURE

young writer seems to give the impression that black means “depressive,” “macabre,” or even “bad” (black has been base-generated away from its usual position in the structure, higher up in Subj.CommentP). Thus I would suggest adjectives like cool and good are only genuinely “ambiguous” in the lexicon; the semantic reading they receive within the sentence does not depend on the grammar’s pragmatic component but is, rather, a direct function of where the adjective has been syntactically generated within the AOR structural hierarchy I have proposed. 4.2. Same apparent semantic class, different FPs The idea that adjectives are the specifiers of distinct FPs can also account for what would otherwise be the odd patterning behavior of adjectives such as ancient and old, which, although apparently belonging to the same semantic class AGE, have very different syntactic distributions: (33)

a. b. c. d.

He arrived in an ancient old car. *He arrived in an old ancient car. a tiny little man *a little tiny man

Why is the ordering in (33b) ungrammatical? If we assume that ancient can only be a specifier of Subj.CommentP, then, consequently, (33b) should only be grammatical with a stressed or contrastive reading (i.e., with “old” stressed). Why should old and ancient belong to different categories? As a tentative answer, it may well be that certain adjectives (young/old for AgeP; big/small for SizeP; heavy/light for WeightP) form the semantic “core” meanings of their respective FPs31 and that others, although related semantically, may either be found only in Subj.CommentP (as in ancient; bulky, ponderous; and probably tiny) or have “dual” status in that they may occur in their “core” FP and also in Subj.CommentP (although without taking a stress or contrastive reading).32 Thus, we would predict ancient and old are not interchangeable. And this prediction seems to be borne out: (34)

a. a long wide old river b. ??a long wide ancient river

[but, an ancient long wide river]

4.3. Same semantic class, interchangeable ordering The question now arises as to how one should analyze stacked adjectival phrases that really are composed of adjectives from the same category. For example: (35)

a. a boring famous book vs. a famous boring book (both Subj.CommentP) b. a new young recruit vs. a young new recruit (both AgeP)

One could propose that [a clever brave man] for example, consists of recursively stacked Subj.CommentPs. However, as we have seen, this analysis is not to be preferred since stacked adjectival phrases would then be the only complete XPs in X'syntax that are capable of being recursively stacked. One way, I suggest, is to consider cases of this sort as genuine adjuncts, recursively adjoined to intermediate projections of the head (as in the traditional analysis of stacked adjectives):

STACKED ADJECTIVAL MODIFICATION

(36)

109

DP D a

D' e

Subj.CommentP AP clever

Subj.CommentP' AP brave

Subj.CommentP' SizeP

e

e

NP man

Since they are projections of the same phrasal category, the adjectives here may be stacked hierarchically and recursively, with order being unimportant (a clever brave / brave clever man); and if, as is usual, we take it that the computational system proceeds in a “bottom-up” fashion, the adjunct (here, clever) will modify the structure to which it is adjoined, [brave [man] ]. This would also give theoretical credence to Dixon’s (1982) hitherto unmotivated observation that adjectives from the same category seem to appear in any order. One disadvantage of this analysis is that throughout this article I have argued extensively against an adjunction approach—and yet in this subsection I appear to be seeking to advocate exactly what I have been trying to disprove! Therefore an alternative way of analyzing the data in (35), and one that is in keeping with the generation-in-Spec analysis proposed throughout this work, is to see if the various “same categories” can themselves be further decomposed. Take, for instance [a boring famous book] and [a famous boring book], which, under the preceding analysis, are both simply instances of adjoined Subj.CommentPs. Such an analysis does not predict that the fragments can have the following semantic interpretations: (37)

a. Oh, it’s a really boring famous book. [“grammatical” with the interpretation “a book generally/universally considered famous but that I find boring”] [“ungrammatical” with the interpretation “a book I find famous but that is generally/universally considered boring”] b. Oh, it’s a really famous boring book. [“grammatical” with the interpretation “a book generally/universally considered boring but that I consider nevertheless famous as (or, for being) a boring book”] [“ungrammatical” with the interpretation “a book generally/universally considered famous but that I consider nevertheless boring as (or, for being) a famous book”]

It seems the category SUBJECTIVE COMMENT can be further decomposed into (a) some sort of Evidential or Universal Comment Phrase that indicates some kind of

110

DP FUNCTIONAL STRUCTURE

general, stable, or standard evaluation and (b) a phrase that can be considered more essentially “subjective comment” in that it indicates the speaker’s own evaluation or an evaluation that may be more transitory: (38)

DP D a

D' e

Subj.CommentP AP boring

Subj.CommentP' e

?EvidentialP/Universal CommentP AP famous

NP book

An interesting topic for further research would be to see if the preceding analysis could be extended to cover strings like a clever brave / brave clever man and other such “same category” combinations—for example, a thick fat / fat thick book (both WIDTH) and a modern new kitchen / a new modern kitchen (both AGE).33

5. General and specific topics for future research Like the final paragraph of the previous section, this section suggests some areas of possible future research vis-à-vis AOR. It seems that, unlike Cinque’s Universal Hierarchy of Clausal Functional Projections, the overall hierarchy of FPs found within the DP, while almost certainly being universal (at least with respect to hierarchical direct modification), may not be universal as far as the specific fine-grained ordering of the FPs is concerned. For instance, it could be that some languages (e.g., Finnish) may allow parametrization in the ordering hierarchy:34 (39)

a. an ugly long road

pitkä ruma tie long ugly road b. beautiful long red hair pitkät kauniit punaiset hiukset long beautiful red hair c. beautiful big red ball suuri kaunis punainen pallo big beautiful red ball

According to my informants, the orderings in (39) are the most natural and unmarked, whereas the orderings corresponding to the English seemed to them to bear focal stress. Thus in Finnish it may be either that LengthP/SizeP precedes Subj.CommentP in the FP hierarchy, or the adjective pitkä ‘long’, for example, is a Subj.Comment adjective in that language.35 The fact that there may be parametric

STACKED ADJECTIVAL MODIFICATION

111

variation within limited boundaries should come as no surprise since parametric variation within the IP has been noted by a number of linguists (for example Ouhalla 1991) and the very locus of cross-linguistic parameterization is now thought to be exclusively the domain of FPs. If parameterization exists in the hierarchies of DP, AgrP, and NegP, why not between the FPs found broadly within the “AP”? Similarly, Hetzron reports the phenomenon of what he terms, “big”-fronting in Amharic, Japanese, Spanish, and Romanian: the placing of the adjective big (but not usually little) before all other adjectives, as in the following example: (40) t©ll©q qong'o qäyy kwas (Amharic) big pretty red ball ‘pretty big red ball’ (Hetzron 1978, 172)

If this particular example is indeed direct hierarchical modification,36 then some languages may place SizeP before Subj.CommentP. Indeed, according to my own data (Scott 1998a, 72), it seems that Serbo-Croat may indeed be such a language. In either case, it appears that, just as in variation within the IP, these parameterizations are quite specific: only one or two categories vary (AgrP and TP within the IP, for example; and LengthP and Subj.CommentP within the DP). Another area for future research that has been mentioned at various points throughout this article is the number and nature of FPs found within the “AP.” Other areas include questions such as the following: Do groups of adjectives that belong to the same semantic class display similar overall syntactic behavior?37 Do we really need (for example) a category PHYSICAL DEFECT (blind, deaf), as Hetzron proposes? Is long in a long new book a LengthP or a Subj.CommentP? Should we only assign the central semantic terms to the semantic classes being proposed (heavy/light for WeightP; long/short for LengthP), so that the less “central” the term is, the more it is open to being seen as SUBJ. COMMENT (hefty, bulky, cumbersome, ponderous). And, of course, there are many classes of adjective that I have not had space to deal with: Dixon’s HUMAN PROPENSITY class (jealous, wicked), for example,38 and participles used as adjectives (freckled, lumbering, computerized, aged), to name but a few. 5.1. FPs and complex internal event structure Another issue not dealt with in this article is the question of adjectives displaying AOR that do not belong clearly to any of the semantic categories in (19). Valois (1991, chap. 4) for example, discusses how adjectives that have adverbial equivalents display distribution patterns cross-linguistically within event nominal projections that are identical with their adverbial counterparts: (41)

a. They [TP probably [VP completely invaded Jupiter] ]. b. *They completely probably invaded Jupiter. (Valois 1991, 149)

(42)

a. the probable complete invasion of Jupiter b. *the complete probable invasion of Jupiter (Valois 1991, 151)

According to Valois, these data are identical in French. He is led to conclude that “adjectives in event nominals, like adverbs in clauses, belong to different classes, and as such are generated at various places in the DP,” and he suggests the hierarchy

112

DP FUNCTIONAL STRUCTURE

is “probable-class > frequent-class > complete-class” (Valois 1991, 175, 165). As to what these classes are, what other members they have, and where they fit in the hierarchy I have proposed is, I can only say again, outside the scope of this article, but it is tempting to speculate that there may turn out to be a close correlation between Cinque’s Clausal Functional Projections and the FPs under which adjectives that display AOR group. In this chapter I have examined only AOR for simple and result nominals like dog and exam and not for complex event nominals like criticism, examination and criticizing; examining—that is to say, nominalizations in general.39 A question I posed earlier was why there should be fewer FPs in the DP than in the CP. I suggested this may be related to the simple/event nominal dichotomy: a beautiful big red ball is OK, but not *a beautiful big red invasion; similarly, to use Valois’s example, the probable complete invasion is again perfectly fine, but not *the probable complete ball. Since simple R argument nominals are not associated with an internal aspectual event structure, it is not surprising that much of the articulated aspectual and temporal FPs that characterize the CP are missing from projections that extend from nouns with R arguments. Does this mean there are two sets of functional hierarchies: one for simple/result nominals; the other for event nominals that incorporate an aspectual event projection? Certainly, according to Grimshaw, the external argument for complex event nominals is different from the external argument for simple and result nominals: “The possibility that the external argument of a complex event nominal is not R but something else offers potential insight into the differences between complex event nominals and the others. Since determiner and modifier systems are different for the two kinds of nouns, this can be attributed to their different external arguments” (Grimshaw 1990, 66). She calls the external argument associated with complex event nominals “Ev.” Does this difference in external argument between the two types of nominal force a different set of extended functional AP-related projections and therefore a different set of AOR: one for R- and one for E-type nominals?40 Grimshaw states, “no noun with R as its external argument can ever have event structure associated with it. . . . It is event structure that licenses modifiers like frequent and constant” (1990, 67). Can such projections be mixed (?the big frequent investigating; ?the ugly constant criticizing) or integrated into a single, all-encompassing functional hierarchy? Again, I can only point to these questions as worthwhile areas for future research.41 Two other sets of data also deserve to be mentioned. To begin with, in [the good good typist],42 where the sense is of a good (=fast, efficient) typist who is also good morally (as opposed to [the morally bad good typist]), the good nearest the head noun is constrained to receive the reading good in ability (=fast), whereas the good that modifies the structure [good [typist] ] can only receive the reading morally good. Thus the following reading is ungrammatical: *the good (in ability) good (in morality) typist; similarly, a bad good typist cannot mean “a typist who is bad at typing but, morally, a good person.” This data, I would argue, is further proof that an adjective generated in the hierarchy relates directly to the semantic reading it receives. Why else should it be that good in the sense of “morally good” can only be attributed to the first adjective in the linear series? Naturally this poses a further problem for the analysis presented here in that, in a good good typist, where do the two goods

STACKED ADJECTIVAL MODIFICATION

113

come in the hierarchical sequence? The “outer” good seems clearly to belong to Subj.CommentP, but what about the “inner” one nearest the head noun?43 It may be that it is generated in some kind of MannerP. 5.2. Adjectives and focus structures Throughout this article I have argued that it cannot be the case that adjectives are adjuncts that can simply be recursively stacked. Andrews uses the following data to argue (the traditional argument) that “modifiers are introduced by nesting rules expanding N' to N' plus a preceding or following modifier” (1983, 696). (43)

a. an alleged English baron b. an English alleged baron

What Andrews fails to notice, it seems, is that (a) and (b) differ quite clearly with respect to markedness: (b) is by far the more marked version, with English receiving contrastive and focal stress. A syntax that assigns both fragments the same structure is missing out on an important difference between the two.44 Within the analysis suggested in this article, (43a) would be assigned a structure as follows: (44)

a. [DP an [SpecSubj.CommentP alleged [ [SpecNationality/OriginP English [NP baron] ] ] ] ]

And (43b) would consequently have a derivation something like: (44)

b. [DP an [FocusP Englishj [SpecSubj.CommentP alleged [ [SpecNationality/OriginP tj [NP baron] ] ] ] ] ]

Since I have argued throughout this article that recursively stacked adjectives are not adjuncts but XPs, then, like any lexical XP, they should be capable of being preposed into some sort of FocusP.45 This analysis has the advantage of (1) treating adjectives like any other XPs (i.e., capable of being preposed) and (2) reflecting in a more articulated way the tight correspondence that exists between the syntax and semantics when it comes to AOR. Thus all of the following could be said to be instances of focus preposing: (45)

a. b. c. d.

It’s the greeni big ti chair that I want.46 It’s the oldj greeni big tj ti chair that I want. It’s the nice greeni big ti chair that I want. It’s the nice greeni oldj big tj ti chair that I want.

Note also: (46) Carol’s horrible six children made life miserable for her second husband. (Andrews 1983, 697)

To account for this last example, Andrews argues “it seems that QP is also introduced by a recursive rule expanding N' to QP N' .” Again, what Andrews seems to miss is that Carol’s horrible six children is a FocusP (hence the scope differences which he himself comments that “it is the group of six children rather than their cardinality that is horrible” [1983, 697]). Thus if Carol had twelve children, it would be quite possible to say, Carol’s horrible six children made life miserable for her second husband, but her nice six gave him great joy. To posit this as an example of preposing to FocusP is, I would argue, a much simpler and more elegant solution,

114

DP FUNCTIONAL STRUCTURE

which, moreover, has the advantage of providing an analysis that unifies this phenomenon with other Focus phenomena not involving QPs.

6.

Conclusion

In this article I attempted to look at the issue of AOR and the semantic classes into which adjectives displaying AOR group. Although the approach has been within a broad P&P framework, to integrate the linguistic data with the analysis outlined it has been necessary to also focus on aspects of the semantics of AOR. Consequently, I highlighted the extent to which the semantic universe may be studied via an independently motivated system of syntactic categories, and, in doing so, I stressed the fact that there exists a tighter interaction between the syntactic and semantic components of the grammar than is usually thought. In this article I argued that previous analyses of AOR were in general not sufficiently articulated and proposed the following refinement to the current ordering hierarchies found in the literature: (47) A proposed universal hierarchy of AP-related functional projections for noncomplex and result nominals (a second approximation) DETERMINER > ORDINAL NUMBER > CARDINAL NUMBER > SUBJECTIVE COMMENT > ?EVIDENTIAL > SIZE > LENGTH > HEIGHT> SPEED > ?DEPTH > WIDTH > WEIGHT > TEMPERATURE > ?WETNESS > AGE > SHAPE > COLOR > NATIONALITY/ORIGIN > MATERIAL > COMPOUND ELEMENT > NP

Furthermore, on the basis of cross-linguistic data, I suggested that both noncomplex and complex event nominal projections have a hierarchically articulated functional structure similar to that which Cinque (1999) proposed for the clause. The thesis presented—whereby adjectives are the specifiers of functional projections with which they are semantically associated—has been shown to explain many aspects of the cross-linguistic syntactic ordering restrictions, as well as semantic combinatorial interpretation of adjectival modifiers. Thus, in this chapter I explain how such an analysis can be extended to accommodate cases of modification by other XP categories (for example, NPs and PPs). Notes I would like to thank the following native speakers: Mika Kuusela, Minna Rajala, Ulla Siikanen, Hanna Vehkamäki (Finnish); Gloria Klein, Stefan Remmert (German); Rosemary Ekpo (Ibibio); Anili George (Malayalam); Jelena Radonjiè (Serbo-Croat); Anna-Lena Wiklund (Swedish); and Olwen Lintern (Welsh). Thanks to Guglielmo Cinque, Liliane Haegeman, Ruth Kempson, and Edwin Williams for comments, encouragement, and general support. Special thanks go to my parents, Violet and John Scott; Tushar Angré; and (of course!) Wynn Chao. 1. See Scott (1998a) for more cross-linguistic examples. 2. The main essays dealing with AOR are Dixon (1982), Hetzron (1978), the work of Sproat and Shih (1988 and 1991), and Androutsopoulou (1996). 3. Also referred to as “comma disjuncture” (Dixon 1982, 24). 4. See Sproat and Shih (1991, 568) for discussion.

STACKED ADJECTIVAL MODIFICATION

115

5. See Sproat and Shih (1991, 567–568). 6. This is related to Grimshaw’s (1990) simple versus event distinction for nominals, to which I return briefly in section 5.1. In her terminology, nouns divide into two classes: complex event nominals that have both an event and an argument structure, and others (noncomplex, simple event nominals and result nominals) that have neither. On an intuitive level, “nonargument-taking nouns refer to individuals and simple events and argument-taking nouns refer to complex events” (Grimshaw 1990, 57). In this article I primarily examine the modification of simple and result nominals: “simple” nominals (like computer and dog) and result nominals (like exam; see Grimshaw 1990, 49) that have an external nonthematic argument R, but no external thematic argument E (see Higginbotham 1985; Grimshaw, 1990, 63)—that is, nouns that are not linked to aspectual semantics. Levi’s (1975) example investigation is clearly a complex event nominal. 7. Excluding Androutsopoulou (1996) who—rightly, I think—argues that AOR must be part of UG. 8. See Cinque (1994) and Crisma (1990). Indeed, according to Cinque (1994, 97), Crisma has proposed at least seven Spec positions for APs: [DP D [XPpossessive X [YPcardinal Y [WPordinal W [ZPspeaker-oriented Z [HPsubject-oriented H [NP [thematic/manner] N]]]]]]]. 9. According to Zwarts (1992, 18 n.5), apparently Hoekstra (1991) argues that all specifiers are adjuncts, but adjuncts that agree with their head. Adjectives obviously seem to fit with this analysis. Indeed, it could be argued that, under this definition, adjectives would be prototypical specifiers. 10. Indeed, Kayne states that “adnominal adjective phrases that are not complements must then be specifiers, each of a separate head. . . . The same must hold for adverb phrases” (1994, 137 n. 31; see also idem, 100–101). 11. Or where the relevant features of the head noun can “percolate” up, depending on how one views the checking strategy: see Radford (1997, 158) for a percolation account. 12. See also Tallerman (1998, 34 and 44) for elementary discussion of the fact that adjectives form a closed class in many languages. In addition, there are languages where the semantic equivalents of “adjectives” are members of the verbal class, and it is debatable whether such languages have a syntactic category Adjective (i.e., lexical items with the feature matrix specification [+N,+V]) at all. In Japanese, for example, “adjectives,” being in reality stative verbs, not only display tense and aspectual morphology but also do not display AOR (this is the indirect modification patterning discussed by S&S; modification in such languages is typically the same as for relative clauses and appositives). Lakhota (Van Valin and LaPolla 1997, 28) is apparently another such language. 13. Data from Pulkina (n.d., 350–351). 14. I must stress that there is no particular reason why I have chosen to use this list other than it is relatively extensive to begin with and the category definitions are quite clear as to what they delineate. In general, the hierarchy lists found in the English-language teaching (ELT) literature are more fully articulated than those in the linguistics literature. For a full summary of the various hierarchy lists proposed throughout the linguistics and ELT literatures, see Scott (1998a, appendix 1). 15. The grammaticality judgments of many of the following examples are, in places, extremely subtle. The asterisk, *, does not necessarily indicate “ungrammaticality” as such but reflects the fact that some of the readings are simply more marked or focused. 16. Adjectives in VSO languages like Irish and Welsh follow English ordering (and are not, as one would expect, the mirror-image of English) because there is movement of the NP past the adjectives: presumably the NP crosses over the adjectives in a cyclic movement in order to assign case to the DP. See S&S (1988, 485) for a brief explanation. Head fronting is frequently used to explain various syntactic phenomena in the Celtic languages (for ex-

116

DP FUNCTIONAL STRUCTURE

ample, sentence-initial placement of the verb and the noun phrase placement of the nominal head). 17. Intuitions involving the category WEIGHT are, to me at least, probably the most difficult and subtle to judge. 18. As far as I am aware, this is the first time this ordering has been proposed. 19. Examples like a slow long river particularly highlight the difference between comma intonation (in which this fragment is grammatical: a slow [pause], long [pause], river) and non-comma intonation. 20. One of the categories also occurring within SizeP (which I will not be discussing) is probably a DepthP: a long deep trench (?a deep long trench); a deep wide ravine (*a wide deep ravine)—note that these facts seem to hold for Swedish as well. The question when proposing new categories, of course, is where does one stop? Thus the adjective wet displays definite cross-linguistic ordering restrictions (a cold wet day versus *a wet cold day); does this really mean one should go so far as to posit a “WetnessP” for example(!!!)? 21. See section 4 for discussion of adjective classes and ordering interchangeability. 22. Again I repeat that this is not the full hierarchy. A complete hierarchy would include a Cardinal/OrdinalP, a NumberP, and so on to account for orderings such as the last few years, the first three men, and the latest Paris fashions. See, for example, Ritter (1991, 50) whose work discusses the functional category NUMBER with respect to Hebrew. 23. Although (due to lack of space) they are not discussed in this section, it is clear that adjectives belonging to the core category SIZE (big, small) precede all other size-related categories: big fast, *fast big; big long, *long big; big tall, *tall big; big wide, *wide big; big hot, *hot big. 24. Chomsky (1995a, 235) refers to the lexicon as merely “a list of ‘exceptions.’” 25. Within a minimalist framework, it may be that the very fact that stacked adjectives are the specifiers of semantically related FPs falls out as part of the general Principle of Full Interpretation (Chomsky 1986, 98). Since each element of the sentence must play some role—whether syntactic, semantic, or phonological—stacked adjectives might be the lexical instantiations of functional, DP-related semantic information that has been projected into the syntax according to the subcategorization properties of the head noun. Incidentally, it should be noted that throughout this article I will assume (as Cinque does for his Hierarchy of Clausal Functional Projections) that the full number of FPs is always structurally present (just sometimes empty) in every sentence. 26. Jackendoff (1997, 230 n.9) criticizes Lieber (1992), who apparently classifies similar constructions as adjectives (Jackendoff gives the examples a Charles-and-Di syndrome; a let’sget-on-with-it attitude). He writes, “she [i.e., Lieber] presents these to be adjectival compounds. But she presents no evidence that they are A°s, and I suspect they are not, since they resist modification typical of adjectives such as ??I’ve never seen a more let’s-get-on-with-it attitude than yours, *John’s attitude is very let’s-get-on-with-it” (Jackendoff, 230 n.9). To me at least, both sentences are perfectly acceptable. Moreover, the analysis presented in this section would be able to deal with instances like [a more let’s-get-on-with-it attitude]: they would presumably be full IPs generated in the Spec of (like off-the-cuff ) a MannerP. 27. Of course, it may be that they are instances of S&S’s parallel modification. I leave this issue open. Hetzron (1978, 169; 171) illustrates the fragment a brave clever / clever brave man across a number of languages and concludes that Dixon’s statement seems to hold. 28. And it is by no means conclusive that they are: compare a small little trinket, *a little small trinket; an old Victorian kitchen, *a Victorian old kitchen. 29. This is a “pure manner” interpretation (John answered in a clever manner) in 27a; a “subject-oriented” interpretation (it was clever of John to answer their questions) in 27b; and an ambiguous reading (ambiguous between the two possible readings) in 27c.

STACKED ADJECTIVAL MODIFICATION

117

30. These are the interpretations when both sentences are said with normal unmarked stress. It might be possible (although I think unlikely) for cool to receive the interpretation excellent, of course, if long in 29a is said with focal stress. 31. So heavy is much more of a prototypical WeightP adjective than, say, bulky. The notion of a certain X having more “X-ness” or being a much more representative “X” than a comparable and similar X is found right across cognitive science as a whole. To quote Fodor: “even concepts like ODD NUMBER, which clearly do have definitions, often have prototype structure as well. The number 3 is a ‘better’ odd number than 27. . . . The discovery of the massive presence of prototypicality effects in all sorts of mental processes is one of the success stories of cognitive science” (1998, 93). 32. It is hard to think of adjectives that might fit this description, however. 33. A related topic is how, and in what larger semantic groups, such FPs themselves cluster: so, for example, Cinque’s full Universal Hierarchy of Clausal Functional Projections (Cinque 1999, 106) shows how FPs within the clause group themselves (broadly!) into Mood-, Tense-, Modality-, and Aspect-related projections. A similar phenomenon might be found with AP-related FPs. For instance, the evidential-, evaluative-, and “subjective comment”-related FPs discussed in this section group themselves in the outermost projections furthest from the head (just as in Cinque’s analysis for the clause, Mood speech act-, evidential-, and evaluative-related FPs occur outermost). Within the DP, next come all the projections related to size and physical shape, perhaps paralleling clausal projections related to Aspect—almost as if within the DP, physical shape and size are manifested in projections hosting adjectives whose semantics reflect the “concrete” nature of the nominal (round, long) and, within the clause, temporal “shape” and “size” are manifested in Aspectual projections hosting adverbs whose semantics reflect the internal abstract “shape” of events (often, briefly). 34. An interesting question for further research is whether there are any other languages like Finnish. Hetzron (1978, 173) notes a similar phenomenon: the interchangeability of SIZE and SUBJ. COMMENT (in his terminology, EVALUATION) in Chinese (although the example he gives is of indirect, so-called, DE modification—see the work of S&S—so, following S&S, interchangeability is actually to be expected), Polish, and Chichewa. In Malayalam, according to Asher and Kumari (1997, 129), it seems that adjectives from the class AGE tend to precede those of the class SIZE. 35. Another interesting question is whether there are any other adjectives like pitkä (perhaps short?) in Finnish—if all “core” adjectives of the LENGTH class precede SUBJ. COMMENT, then we have good grounds for positing parameterization; if it appears that just pitkä does, then we are probably dealing with a lexical peculiarity. 36. As far as the Japanese and Spanish examples are concerned, it is not. Hetzron presents the following data for a pretty big red ball, neither of which, in the terminology of S&S, is hierarchical direct modification: (i) ookii kirei na akai booru (Japanese) big pretty red ball (ii) un balón grande rojo y lindo (Spanish) a ball big red and pretty (Hetzron 1978, 172) The Japanese ookii ‘big’ and akai ‘red’ are, in fact, stative verbs acting as relative clauses (i.e., “a ball which is big, which is red”) which correspond only semantically to their English adjectival equivalents; likewise, the na found after the (true) adjective kirei ‘pretty’ is a form of the copula obligatorily needed after adjectives in Japanese to modify the head noun, and as such it is directly equivalent to the Chinese DE modification discussed by S&S (1988 and 1991). As for the Spanish data, this seems to me to be an example of comma intonation fol-

118

DP FUNCTIONAL STRUCTURE

lowed by what I have termed “and-use” (see Scott 1998a, 7 n.8), a usage found across all Romance languages. 37. Pustejovsky (1995, 23) discusses Dixon’s AOR list but notes, for example, that, within Dixon’s class DIFFICULTY, both easy and difficult behave syntactically as tough-movement predicates, whereas within Dixon’s class QUALIFICATION, likely allows raising but probable and possible do not. He concludes that Dixon’s classification system based on the notion “semantic field” is unable to predict the syntactic behavior of its members. 38. Dixon (1982, 17) suggests they come after SPEED but before AGE. 39. After Chomsky (1970), the first type of nominal (criticism; examination) is termed derived nominalization; and the second (criticizing; examining), gerundive nominalization. Grimshaw (1990) shows how both types of nominal have an inherent “event” structure, which is an internal aspectual analysis: gerundive nominals always have a-structure, and derived nominals are ambiguous. Such nominalizations are further “verb-like” in that they take obligatory arguments; simple and result nominals do not (Grimshaw 1990, 47). Certain modifiers (for example, frequent) force an event interpretation onto the nouns they modify; they also modify the actual internal “event” referred to by the nominal. 40. In turn, this notion would have consequences for other parts of the grammar—for example, the Projection Principle: “Representations at each syntactic level are projected from the lexicon, in that they observe the subcategorization properties of lexical items” (Chomsky 1981, 29). In theory, then, R argument nominals would subcategorize for a different set of FPs to Ev argument nominals. 41. Another area of adjectival modification that I have not had space to deal with is the case of (1) languages where certain adjectives change their semantics, depending on whether they precede or follow the noun, and (2) the syntax/semantics of such adjectives when they are stacked; cf. the following French and Welsh data, respectively: un roi ancien ‘an ancient king’, un ancien roi ‘a former king’ (Napoli 1993, 182); and unig blentyn ‘an only child’, plentyn unig ‘a lonely child’. Presumably, an account of this phenomenon might involve combining some version of Cinque’s (1994) N-movement hypothesis with the FP analysis presented here and (perhaps) explaining movement in terms of (?semantic) feature checking. 42. These data are based (loosely) on Jackendoff (1997, 62). 43. The same problem is to be found with [an old old Etonian]—this fragment can only receive the interpretation an old (= in age) old (= former) Etonian and never *an old (= former) old (= in age) Etonian. 44. Also, in (43a) English is under the scope of alleged: as is well known [an alleged English baron] need be neither English nor an actual baron. In (43b), in contrast, alleged is within the scope of English: the person must be English but need not necessarily be a baron. 45. Although I am suggesting here that contrastively focused adjectives are preposed to “something like” a FocusP, the actual nature of the projection to which they are moved (whether it be FocusP, TopicP, ForceP, or other P) I will leave open. See Rizzi (1997) for discussion. 46. Liberman and Sproat use the sentence I want the red large ball, not the blue one to argue that AOR “are only preferences, operating in the absence of pragmatic reasons to choose a different order. It is perfectly possible to choose an alternative order . . . so it seems best to argue with the traditional view that these ordering preferences reflect some sort of natural continuum of psychological affinity, even though it seems difficult to define such a scale in a non-circular way” (1992, 165). Needless to say I disagree with this: arguments against the natural continuum of psychological affinity view are presented more fully in Scott (1998a); and just what could the “pragmatic reasons” be that AOR are “operating in the absence of ”?

STACKED ADJECTIVAL MODIFICATION

119

References Abney, S. (1987) “The English Noun Phrase in Its Sentential Aspect,” Ph.D. Diss., Cambridge, Mass.: MIT. Andrews, A. (1983) “A Note on the Constituent Structure of Modifiers,” Linguistic Inquiry 14, 695–697. Androutsopoulou, A. (1996) “The Licensing of Adjectival Modification,” in Proceedings of the Fourteenth West Coast Conference on Formal Linguistics, 17–32. Asher, R., and T. Kumari. (1997) Malayalam. London: Routledge. Belletti, A., and L. Rizzi (eds.) (1996) Parameters and Functional Heads: Essays in Comparative Syntax. New York: Oxford University Press. Chomsky, N. (1970) “Remarks on Nominalization,” in R. Jacobs and P. S. Rosenbaum (eds.) Readings in English Transformational Syntax, 184–221. Waltham, Mass.: Ginn and Co. Chomsky, N. (1981) Lectures on Government and Binding. Dordrecht: Foris. Chomsky, N. (1986) Knowledge of Language: Its Nature, Origin and Use. Westport, Conn.: Praeger. Chomsky, N. (1995a) The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. (1995b) “Bare Phrase Structure,” in G. Webelhuth (ed.) Government and Binding Theory and The Minimalist Program, 383–439. Oxford: Blackwell Cinque, G. (1994) “On the Evidence for Partial N-Movement in the Romance DP,” in G. Cinque, J. Koster, J.-Y. Pollock, L. Rizzi and R. Zanuttini (eds.) Paths towards Universal Grammar: Studies in Honor of Richard S. Kayne, 85–110.Washington, D.C.: Georgetown University Press. Cinque, G. (1999) Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York: Oxford University Press. Cook, V., and M. Newson. (1996) Chomsky’s Universal Grammar, 2nd ed. Oxford: Blackwell. Crisma, P. (1990) “Functional Categories inside the Noun Phrase: A Study on the Distribution of Nominal Modifiers,” M.A. Thesis, University of Venice. Dixon, R. (1982) Where Have All the Adjectives Gone? The Hague: Mouton. Fodor, J. A. (1998) Concepts. Oxford: Oxford University Press. Fowler, W. S. (1992) Synthesis Advanced. Walton-on-Thames, Surrey: Nelson. Freidin, R. (1992) Foundations of Generative Syntax. Cambridge, Mass.: MIT Press. Giusti, G. (1997) “The Categorial Status of Determiners,” in L. Haegeman (ed.) The New Comparative Syntax, 95–123. Harlow, Essex: Longman. Grimshaw, J. (1990) Argument Structure. Cambridge, Mass.: MIT Press. Grimshaw, J. (1991) “Extended Projection.” Unpublished Ms., Brandeis University, Waltham. Hetzron, R. (1978) “On the Relative Order of Adjectives,” in H. Seiler (ed.) Language Universals, 165–184. Tübingen: Gunter Narr Verlag. Higginbotham, J. (1985) “On Semantics.” Linguistic Inquiry 16, 547–593. Hinton, M., and R. Marsden. (1985) Options: Advanced English. Walton-on-Thames, Surrey: Nelson. Hockett, C. (1958) A Course in Modern Linguistics. Toronto: Macmillan. Hoekstra, E. (1991) “Licensing Conditions on Phrase Structure,” Ph.D. Diss., University of Groningen. Jackendoff, R. (1972) Semantic Interpretation in Generative Grammar. Cambridge, Mass.: MIT Press. Jackendoff, R. (1997) The Architecture of the Language Faculty. Cambridge, Mass.: MIT Press. Kayne, R. (1994) The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Kiefer, F., and K. Kiss (eds.) (1994) The Syntactic Structure of Hungarian (Syntax and Semantics 27). New York: Academic Press.

120

DP FUNCTIONAL STRUCTURE

Kingsbury, R., and G. Wellman. (1986) Longman Advanced English. Harlow, Essex: Longman. Levi, J. (1975) “The Syntax and Semantics of Nonpredicating Adjectives in English,” Ph.D. Diss., University of Chicago. Levine, R. (ed.) (1992) Formal Grammar: Theory and Implementation. Oxford: Oxford University Press. Liberman, M., and R. Sproat. (1992) “The Stress and Structure of Modified Noun Phrases in English,” in I. Sag and A. Szabolsci (eds.) Lexical Matters, 131–181. Stanford, Calif.: Center for the Study of Language and Information. Lieber, R. (1992) Deconstructing Morphology: Word Formation in Syntactic Theory. Chicago: University of Chicago Press. Napoli, D. (1993) Syntax: Theory and Problems. Oxford: Oxford University Press. Ouhalla, J. (1991) Functional Categories and Parametric Variation. London: Routledge. Pulkina, I. (n.d.) A Short Russian Reference Grammar. Moscow: Progress Publishers. Pustejovsky, J. (1995) The Generative Lexicon. Cambridge, Mass.: MIT Press. Radford, A. (1997) Syntactic Theory and the Study of English: A Minimalist Approach. Cambridge: Cambridge University Press. Ritter, E. (1991) “Two Functional Categories in Noun Phrases: Evidence from Modern Hebrew,” in S. Rothstein (ed.) Perspectives on Phrase Structure: Heads and Licensing, 37–62 (Syntax and Semantics 25). New York: Academic Press. Rizzi, L. (1997) “The Fine Structure of the Left Periphery,” in L. Haegeman (ed.) Elements of Grammar, 281–337. Dordrecht: Kluwer. Sag, I., and A. Szabolsci (eds.) (1992) Lexical Matters. Stanford, Calif.: Center for the Study of Language and Information. Scott, G.-J. (1998a) “Stacked Adjectival Modification and the Structure of Nominal Phrases,” M.A. Thesis, School of Oriental and African Studies, University of London. Scott, G.-J. (1998b) “Stacked Adjectival Modification and the Structure of Nominal Phrases.” SOAS WP in Linguistics and Phonetics, 8: 59–89. Scott, G.-J. (forthcoming) “The Syntax and Semantics of Adjectival Modification,” Ph.D. Diss., School of Oriental and African Studies, University of London. Sproat, R., and C. Shih. (1988) “Prenominal Adjectival Ordering in English and Mandarin,” in Proceedings of the Eighteenth Annual Meeting of the North Eastern Linguistics Society, 465–489. Sproat, R., and C. Shih. (1991) “The Cross-Linguistic Distribution of Adjective Ordering Restrictions,” in C. Georgopoulos and R. Ishihara (eds.) Interdisciplinary Approaches to Language, 565–593. Dordrecht: Kluwer. Tallerman, M. (1998) Understanding Syntax. London: Arnold. Valois, D. (1991) “The Internal Syntax of DP,” Ph.D. Diss., University of California at Los Angeles. Van Valin, R., and R. LaPolla. (1997) Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Zanuttini, R. (1997) “Negation and Verb Movement,” in L. Haegeman (ed.) The New Comparative Syntax, 214–245. Harlow, Essex: Longman. Zwart, C. J. (1987) Morphosyntax of Verb Movement: A Minimalist Approach to the Syntax of Dutch. Dordrecht: Kluwer. Zwarts, J. (1992) X2 -Syntax–X2 -Semantics: On the Interpretation of Functional and Lexical Heads. Utrecht: OTS.

P ART II

IP FUNCTIONAL STRUCTURE

This page intentionally left blank

5

Clause Structure and X-Second    

The purpose of this essay is to propose a unified analysis of a range of “secondposition” phenomena that have been attested in various languages, and in so doing to motivate a more elaborated theory of Nominative Case assignment. The proposal is that many languages, including a number of Germanic and Romance languages, have a projection that intervenes between Comp and the highest Infl-type projection (which, following Belletti 1990, we take to be AgrP). We refer to this projection as Agr1P, and we refer to the lower, “traditional” AgrP as Agr2P. Thus our claim is that in the languages in question there are two Agr-heads and two projections of Agr. These two Agrs are both “subject” Agrs; in this respect, our proposal is distinct from but not exclusive with Chomsky’s (1989) idea that, in addition to the standard “subject” Agr, there is also an “object” Agr. We show here that this proposal is of considerable empirical value in that it offers a new perspective on a range of second-position phenomena and allows us to connect “verb-second” effects with various kinds of “clitic-second” effects, known in traditional grammar as Wackernagel’s Law and the Tobler-Mussafia law.1 In fact, we suggest that the presence of Agr1P is fundamentally related to Nominative Case assignment, in that the basic property of Agr1° seems to be that of assigning Nominative Case; the other properties that we ascribe to it (e.g., attracting clitics and attracting the inflected verb) are intimately related to its Nominative-assigning property. In this sense, it may be best to think of Agr1P as NomP. As a working hypothesis, then, we assume that in languages that have both Agr1P and Agr2P, Agr2° is not an assigner of Nominative Case. Because our focus is on the interaction of clause structure with structural Case assignment, we concentrate almost exclusively on the processes and properties of S-structure. 123

124

IP FUNCTIONAL STRUCTURE

It is a classic tenet of generative grammar that inflectional affixes may be separate syntactic entities at pre-phonological levels of representation; see the analysis of the English auxiliary system in Chomsky (1957). At the turn of the 1990s, this idea received new impetus, beginning with Pollock (1989). Pollock’s proposal that the Infl-node of Chomsky (1981) should be split into its morphological components led to the working hypothesis in much recent research that any inflectional head which appears to be syntactically relevant heads its own maximal projection with the standard X-bar theoretic structure. Our proposal amounts to the claim that certain languages have a special X-bar projection for the assignment of Nominative Case; in terms of the connection with inflectional morphology, it may be possible to relate the existence of Agr1P to the possession of morphologically realized Nominative Case. The structure that we are proposing is as follows: CP Spec

C' Agr1P



Agr1'

Spec

Agr2P

Agr1˚

Agr2'

Spec Agr2˚

TP

The order of TP and Agr2° varies in the languages under consideration: in West Germanic TP is on the left of Agr2° (see Giusti 1986 for the proposal that IP is headfinal in German); in North Germanic and Romance, it is on the right. By contrast, Agr1°, like C°, precedes Agr2P in all the languages we discuss. The essay is organized as follows: in section 1 we analyze the phenomenon of embedded verb-second, basing ourselves largely on the best-known case of this type: Icelandic (although we also analyze both Old French and Yiddish in these terms). In section 2 we show how our system gives a natural analysis of the “clitic-second” phenomena in Germanic and Romance; in our terms the traditionally recognized Wackernagel position of Germanic languages is Agr1°, as is the clitic position in those Romance languages that obey the Tobler-Mussafia law. The last two sections deal with ways in which the properties of Agr1° vary parametrically: we discuss different modes of Nominative Case assignment in section 3, where we present and elaborate the recent proposals of Koopman and Sportiche (1991), and different kinds of null subjects in section 4. Finally, the appendix is devoted to the discussion of Stylistic Fronting. We have also added a postscript to this published version of the paper, explaining its rather unusual and protracted prepublication history.

CLAUSE STRUCTURE AND X - SECOND

125

1. Agr1 as a position for the inflected verb: Embedded V2 1.1.

Icelandic

The postulation of Agr1P allows us to account in a straightforward way for certain differences within the class of verb-second languages. While in many languages— for example, German, Dutch, and Mainland Scandinavian—verb-second is essentially a root phenomenon, it appears to be generalized to all types of embedded clauses in Icelandic. The following data (from Rögnvaldsson and Thráinsson 1990 and Thráinsson, personal communication) illustrate this, showing that in a variety of embedded clauses we have the order XP > V > subject:2 (1)

a. Ég held að þegar hafi María lesið þessa bók. I believe that already has M. read this book ‘I believe that Mary has read this book already.’ b. Ég harma að þegar hafi María lesið þessa bók. I regret that already has M. read this book ‘I regret that Mary has already read this book.’ c. Ég spurði hvort þegar hefði María lesið þessa bók. I asked whether already had M. read this book ‘I asked whether Mary had already read this book.’ d. sú staðreynd að þegar hefur María lesið þessa bók the fact that already had M. read this book ‘the fact that Mary had already read this book’ e. bókin sem þegar hefur María lesið book-the that already had M. read ‘the book that Mary had already read’

(factive)

(Wh)

(NP)

(relative)

The Icelandic situation, as illustrated in (1), contrasts with what we find in German. In German, embedded V2 is possible only in a limited class of embedded clauses, essentially the complements to “bridge verbs” of the type in (1a). Embedded V2 is excluded in all the contexts parallel to (1b–e):3 (2)

a. Ich glaube, gestern hat Maria dieses Buch gelesen. I believe yesterday has M. this book read ‘I believe Mary read this book yesterday.’ b. *Ich bedauere, (daß) gestern hat Maria dieses Buch gelesen. I regret (that) yesterday has M. this book read c. *Ich frage mich, ob gestern hat Maria dieses Buch gelesen. I ask myself whether yesterday has M. this book read d. *die Tatsache, gestern hat Maria dieses Buch gelesen the fact yesterday has M. this book read e. *das Buch, das gestern hat Maria gelesen the book which yesterday has M. read

For (2a), an analysis in terms of CP-recursion seems to be in order. As noted in Rizzi and Roberts (1996), the same class of verbs allows an otherwise root phenomenon— subject-aux inversion triggered by a negative-polarity item—in its complement in English: (3)

a. I believe that only in America could you do such a thing. b. *I wonder whether only in America could you do such a thing.

126

IP FUNCTIONAL STRUCTURE

It seems then that, independently of verb-second, the complements of bridge verbs are able to have root properties. We propose, still following Rizzi and Roberts (1996), that this is because bridge verbs allow CP-recursion in their complements. More precisely, we propose that bridge verbs select a C° which selects another C°; to avoid unlimited recursion at the C-level, we clearly must propose that the two C°s have different properties (e.g., that the first allows a “propositional” complement, while the second only allows a “predicational” complement in the terms of Rizzi 1990b). In German, the two C°s are different in form: the first is null, and the second is filled by the inflected verb, like the C° of a matrix clause. (In English, too, the C° that selects CP is different from other C°s: that in (3a) cannot be deleted, while other occurrences of that can be.) Adopting this analysis for (2a), we propose the following partial structure: (4) . . . glaube [CP O [CP gestern [C’ hat [ Maria . . . ] ] ] ]

Vikner (1995) proposes extending this analysis to embedded verb-second in Icelandic. This entails that CP-recursion is generalized in Icelandic, while it is limited to a specific class of complements in German, Mainland Scandinavian, and English. In other words, the property of selecting C° is available for all classes of C° in Icelandic. If this were true, however, then there would be no way to avoid unlimited recursion of C°, which is clearly an undesirable consequence. Instead, our proposal provides a straightforward account of the data in (1). These examples have the following structure (although, to the extent that the class of verbs which allows CP-recursion in German also allows it in Icelandic, (1a) may also have a structure like (4) with að in the higher C° and the inflected verb in the lower C°):4 (5) [CP C° [Agr1P TOP [Agr1’ Vi+Agr1 [Agr2P NPNom [Agr2° ti] . . . ] ] ] ]

As we will see in more detail in section 3, the special property of Icelandic is that SpecAgr1' is a topic position, while the usual superficial subject position is SpecAgr2'. If we do not assume a double-Agr structure, no other position would be available for the subject. Since there is no generalized CP-recursion, SpecC' is not available. Although we assume that the subject is base-generated in VP (see in particular Koopman and Sportiche 1991), the base-position of the subject is unavailable, at least for a definite NP, since it is not a position that can receive Nominative Case from Agr° (see section 4 for some evidence that indefinite NPs can appear in this position). This idea is confirmed by the fact that definite subjects always precede VPadverbs, as in …að þegar hefur María oft lesið þessa bók ‘that already has Mary often read this book’ (Thráinsson, personal communication) (see also Vikner 1995, 68). Furthermore, we follow Rizzi (1990a) in assuming that SpecT' is inherently an A'-position and, as such, is not a possible landing site for the subject. In fact, if we adopt Rizzi’s (1991) characterization of potential A-positions as either q-positions or specifiers of Agr, SpecT' must be an A'-position. Hence, SpecAgr2' is the position of the definite subject. Using this analysis of embedded clauses, we are not forced to treat matrix V2 in Icelandic as involving movement of the verb to C°. Movement to Agr1° would clearly suffice to derive the same orders (see Rögnvaldsson and Thráinsson 1990 and the references given there for recent discussion of similar proposals). At the same time, the data do not force us to reject a movement-to-C analysis.

CLAUSE STRUCTURE AND X - SECOND

127

One property of Icelandic, however, suggests that, in fact, matrix V2 should be handled in terms of verb-movement to Agr1° rather than to C°. Icelandic makes much more frequent use of declarative V1 orders than do the other (Modern) Germanic languages (aside from Yiddish). Declarative V1 is illustrated in the following example: (6) Hitti hann þá einhverja útlendingar. (Sigurðsson 1985, 1) met he then some foreigners ‘He then met some foreigners.’

We propose that in (6) the inflected verb has undergone structure-preserving topicalization (i.e., topicalization of an X° category to another head position Y°). This operation probably takes place for reasons connected to information structure, since the examples in question seem to be presentational sentences. We propose that the landing site of this operation is C°; these are the only declarative clauses in which the verb is in C° (as in other Germanic and Romance languages, the inflected verb is typically in C° in matrix interrogatives, imperatives, and hypotheticals). As we will see in section 2.2, this kind of “verb-topicalization” is not restricted to Icelandic but also is found in Medieval Romance languages (see also Beninca’ 1989; Alberton 1990). More generally, we expect this possibility to exist in all languages that realize V2 at the Agr1P-level since in such cases C° is freely available as a landing site for structure-preserving topicalization. Since the V2 requirement is satisfied elsewhere, SpecC' in such languages can remain empty. This analysis implies that declarative operators do not exist, since otherwise we would expect V1 declaratives to be generally possible on a par with V1 interrogatives, hypotheticals, and so on.5 One reason to favor an analysis of (6) in which the verb moves to C° over one in which the verb moves only to Agr1°, while the subject stays in SpecAgr2' , is that this kind of V1 is a root phenomenon. Since den Besten (1983), the simplest treatment of root phenomena has been to say that they involve movement to C°, a position available in principle in matrix clauses but unavailable in embedded clauses. If matrix clauses are Agr1P in Icelandic, this implies that verb second is not a unified phenomenon in the Germanic languages, at least in the sense that the landing site of the verb may vary cross-linguistically. We will see further evidence in favor of this conclusion as we proceed. See also Diesing (1988, 1990) and Santorini (1988, 1989) for a similar conclusion based on Yiddish evidence (and see section 1.3 here for some discussion of Yiddish). To sum up, following on from our “double-Agr” hypothesis about basic clause structure, the following conclusions emerge for Icelandic: (7)

a. b. c. d.

Agr1° can assign Nom under government (see section 3). SpecAgr2' is a subject position. SpecAgr1' can be a topic position. SpecC' is an operator position.

1.2. Old French The V2 nature of Old French (OF) is illustrated clearly by the examples in (8) (nonnominative clitics—e.g., en in (8b)—are effectively part of the inflected verb, and so do not “count” in the computation of the second position):

128 (8)

IP FUNCTIONAL STRUCTURE

a. Einsint aama la demoisele Lancelot. (Adams 1987b, 50) thus loved the lady Lancelot ‘Thus the lady loved L.’ b. Desuz un pin en est li reis alez. (Schulze 1888, 200) under a pine-tree of-it is the king gone ‘The king went underneath a pine tree.’ c. Quatre saietes ot li bers au costé. (Le Charroi de Nîmes, l. 20)6 four boats-of-war had the baron at-the side ‘The baron had four boats of war at his side.’

Adams (1987a,b) shows that V2 is possible in the complements to bridge verbs. The class of bridge verbs in question is comparable to the class which in V2 Germanic languages typically allows complements with matrix properties (see section 1.1). Here are some examples with null subjects: (9)

a. Or voi ge bien, plains es de mautalant. (Le Charroi de Nîmes, l. 295) now see I well full are (you) of bad-intentions ‘And now I see clearly that you are full of bad intentions.’ b. Je cuit plus sot de ti n’i a. (Adams 1987b, 17 (11a–c)) I think more stupid than you not there has (it) ‘I think that there is no one stupider than you.’

In neither of these examples is que present, suggesting that these are cases of German-style embedded V2 (this implies that “conjunctive discourse” is not specific to German; see note 3). The embedded sentences in examples like (9) can be analyzed as follows: (10)

CP1 CP2

C1˚

C2'

AdvP C2˚

AgrP

V+Agr NP

TP

It is clear that CP2 here is just like a matrix clause, and so V2 is possible, as expected. However, there are cases of V2 orders in Wh-clauses: (11)

a. Quant a eus est li rois venus, . . . (Dupuis 1989, 148 (40)) when to them is the king come ‘When the king came to them, . . .’ b. s’a la vostre bonté vousist mon pere prendre garde if against the your good-will wanted my father to-take precaution ‘if against your good will my father wanted to take precautions’ (Adams 1988b, (19c))

CLAUSE STRUCTURE AND X - SECOND

(12)

129

a. Por l’esperance qu’an lui ont, . . . for the hope which in him have (they) ‘For the hope which they have in him, . . .’ b. Et si ne sait que faire puisse. and so not knows what to-do can (he) ‘And so he doesn’t know what he can do.’

In terms of the standard assumption that the inflected verb cannot move to a [+wh] C° (see Rizzi and Roberts 1996 and Rizzi 1996 for an account of this), we are led to the conclusion that the verb is in Agr1° in these examples. So we assign the following structure to (11a), for example: (13)

CP XP

C' Agr1P

C˚ [+wh]

Agr1'

Spec

Agr2P

Agr1˚

Agr2'

Spec

quant

a eus

esti

li rois

Agr2˚

TP

ti

ti venus

Here the verb appears in Agr1° and assigns Case under government to the subject, li rois, in SpecAgr2'. What is the status of SpecAgr1' in OF? On the basis of examples such as (11), it appears to be a topic position of the Icelandic kind. However, clear examples of the type in (11) are not very frequent. According to Dupuis (1989, 151–152), this possibility is only attested with any real frequency in the Quatre Livres du Roi (a text from around 1170); in other texts (including some from the same period), there are very few cases of embedded V2 (in non-bridge complements) with overt subjects. There are, however, cases of embedded V2 order with null subjects in a range of twelfth-century texts, as well as in some thirteenth-century verse texts (see Hirschbuhler 1990 and section 4.2 here). Such examples are not clear cases of topicalization and can instead be treated as Stylistic Fronting. Stylistic Fronting is an operation found in Icelandic, Faroese, and Yiddish. The operation fronts some VP-constituent, usually an adverbial, a participle, or a complement (see Maling 1990 for details), to a position between C° and the inflected verb. The main condition on Stylistic Fronting is that the sentence must contain a subject gap, usually a trace of Wh-movement, but

130

IP FUNCTIONAL STRUCTURE

also possibly the trace of cliticization (Platzack 1988) or an NP-trace (Sigurðsson 1989). The following Icelandic example illustrates the application of StylisticFronting to a participle: (14) Þarna er konan sem kosin var / var kosin forsetí. there is woman that elected was / was elected president ‘There is a woman that was elected president.’

One test favors a Stylistic-Fronting analysis of examples like (12) over a topicalization analysis. Platzack (1988) shows that the subject-gap condition can be satisfied by a cliticized subject pronoun. This gives rise to the order Complementizer > Subject > XP > V, which cannot be a case of embedded verb-second. Such orders are found at the relevant period of OF: (15) quant il de ci departiront (Vance 1988, 89 (11)) when they from here will-leave ‘when they will leave here’

The existence of this kind of order, and the absence of clear embedded topicalization, at the relevant period of OF, suggests that apparently ambiguous examples of the kind in (12) are to be treated as involving Stylistic Fronting. For reasons of space, we make no proposal for the analysis of Stylistic Fronting in this essay, although we note that it does seem to correlate with the double-Agr structure in VO languages: Icelandic and OF both have both properties, as arguably so do Middle English and the Medieval Mainland Scandinavian languages (see Platzack 1988, 1994). In the appendix, we suggest a tentative analysis of this operation. The conclusion is that, with the possible exception of some of the very early texts, OF does not allow generalized embedded topicalization, although it does have Stylistic Fronting. We interpret this to mean that SpecAgr1' was not a topic position in OF but was a subject position. Nevertheless, movement of the inflected verb to Agr1° was general, and so the movement of the inflected verb to Agr1° does not necessarily imply that SpecAgr1' is a topic position (see section 3). In section 4.2, we will see that the distribution of null subjects in OF supports this conclusion. To sum up, OF had the following properties: (16)

a. Agr1° can assign Nominative under government. b. SpecAgr2' is a subject position. c. SpecAgr1' is also a subject position (see section 3).

1.3. Yiddish The other Germanic language that has been claimed to allow generalized embedded topicalization is Yiddish (see Diesing 1988, 1990; Santorini 1988, 1989). The following examples illustrate topicalization in [–wh] and [+wh] complements: (17)

a. Jonas bedoyert az dos bukh hob ikh geleyent. (Vikner 1995, 72) J. regrets that this book have I read ‘John regrets that I have read this book.’ b. Ikh veys nit far vos in tsimer iz di ku geshtanen. (Vikner 1995, 74) I know not for what in room is the cow stood ‘I don’t know why the cow has stood in the room.’

CLAUSE STRUCTURE AND X - SECOND

131

c. Ikh veys nit tsi in tsimer iz di ku geshtanen. (Santorini, personal communication) I know not if in room is the cow stood ‘I don’t know if the cow has stood in the room.’

Since “regret” is a non-bridge verb, (17a) is most likely not a case of CP-recursion. Examples (17b,c) are cases of topicalization in a [+wh] complement. Both Diesing (1988, 1990) and Santorini (1988, 1989) propose to analyze this kind of embedded V2 in Yiddish by treating SpecI' —the canonical subject position in a language like English—as a topic position and then taking the subject to be in its VP-internal base position. However, this analysis will not carry over straightforwardly in our terms. If we split IP into TP and AgrP, and assume that Nominative Case is assigned by Agr°, then the VP-internal subject position cannot receive Nominative Case. Instead, this position receives Partitive Case (see Belletti 1988) and therefore can only be occupied by indefinite NPs. Since we are assuming that SpecT' is an A'position (see section 1.1), no (nonscrambled) NP can occupy this position. Hence, the only position for a definite Nominative NP is SpecAgr'. For these reasons, the examples in (17) motivate an analysis in terms of the “double-Agr” structure. So (17a) would have the following structure: (18)

CP C˚

Agr1P Agr1'

NP

Agr2P

Agr1˚

Agr2'

NP

az dos bukh hobi

ikh

Agr2˚

TP

ti

ti geleyent

Further evidence in favor of embedded topicalization and against CP-recursion comes from sentences like the following where an argument undergoes long Whmovement: (19) Vos hot er nit gevolt az in shul zoln di kinder leyenen? (Santorini 1989, 59) what has he not wanted that in school shall the children read ‘What didn’t he want the children to read in school?’

As expected, the analogous long extraction of an adjunct is impossible (see Schwartz and Vikner 1996, 3.1). In general, extraction from complements with CP-recursion is impossible (see Rizzi and Roberts 1989). This, combined with the fact that CPrecursion under verbs of volition is otherwise unattested, argues that there is no CPrecursion here. Instead, then, this must be a case of embedded topicalization.

132

IP FUNCTIONAL STRUCTURE

However, there are a number of restrictions on embedded V2 in Wh-complements in Yiddish. Only a limited class of Wh-complements—for example, those in far vos ‘why’ and tsi ‘if’’—allow embedded topicalization. Compare (20) with (17b,c): (20)

a. *Ikh veys nit vu nekhtn iz di ku geshtanen. (Vikner 1995, 74) I know not where yesterday is the cow stood ‘I don’t know where the cow stood yesterday.’ b. *Ikh veys nit ven zayn khaver hot Moyshe getrofn. (Santorini, personal communication) I know not when his friend has M. met ‘I don’t know when Moyshe met his friend.’ c. *Ikh veys nit (far) vemen zayn khaver hot Moyshe forgeshtelt. (Santorini, personal communication) I know not (to) whom his friend has M. introduced ‘I don’t know who Moyshe has introduced his friend to.’

Short extraction of an argument Wh-element seems to be possible over a topicalized adjunct PP, however: Ikh veys nit vemen in restoran hot Moyshe getrofn ‘I don’t know whom in the restaurant has M. met’. This situation should be compared with that in Icelandic mentioned in note 2. On the other hand, topicalization in a Wh-complement apparently becomes generally possible where the subject is extracted or the subject is indefinite and apparently VP-internal: (21)

a. Zi iz gekumen zen ver frier vet kontshen. (Diesing 1988, 132) she is come see who earlier will finish ‘She has come to see who will finish earlier.’ b. Ikh veys nit vu nekhtn iz geshtanen a ku. (Santorini, personal communication) I know not where yesterday is stood a cow

Both contexts involve a gap in the canonical, preverbal subject position. As such, they appear to be cases of Stylistic Fronting (see section 1.2 and appendix). In relative clauses, verb-second order arises where there is a subject gap, but not otherwise. Relatives on the object are possible, provided that a resumptive pronoun is used (examples (22) from Lowenstamm 1977, cited in Santorini 1989, 56; example (23) from Lowenstamm 1977, cited in Diesing 1990, 63): (22)

(23)

a. nokh epes, vos oyfn hitl iz geven still something, that on the hat-DIM is been ‘something else that was on the little hat’ b. *Der yid vos in Boston hobn mir gezen iz a groyser lamdn. the man that in Boston have we seen is a great scholar ‘The man whom we saw in Boston is a great scholar.’ der yid vos in Boston hobn mir im gezen the man that in Boston have we him seen ‘the man that we saw (him) in Boston’

Here again, the possibility of embedded topicalization depends on the presence of a subject gap, confirming the idea that these are cases of Stylistic Fronting. The con-

CLAUSE STRUCTURE AND X - SECOND

133

trast between (22b) and (23) supports the idea that embedded topicalization is incompatible with any form of Wh-movement (with the exception noted after (20)). It seems, then, that Yiddish allows embedded verb second under rather limited conditions: in non-bridge complements and in Wh-complements where the Wh-element is not moved. Stylistic Fronting is possible in clauses where Wh-movement affects the subject. In other words, embedded topicalization creates an island for extraction in Yiddish Wh-complements (see Vikner 1995, 73–80). This seems to be a property specific to Yiddish, since embedded topicalization does not have this effect in Icelandic (or, perhaps, the effect is limited to extraction of adjuncts over adjunct topics; see note 2). In spite of this difference, Yiddish shows the same basic properties as Icelandic: it has a double-Agr clause structure, SpecAgr1' is a topic position, SpecAgr2' is a subject position, and Nominative Case can be assigned under government.

2. Agr1 as a position for clitics It has often been noticed, since the earliest work on Indo-European syntax (see Wackernagel 1892), that unstressed elements of various kinds tend to be found in the second position in the clause. In this section, we propose a general analysis of this salient fact about the syntax of a variety of languages in terms of the following general idea:7 (24) Agr1° is a position for clitics.

In the languages we discuss in this section, Agr1° is preceded by an element occupying its Spec. Since Agr1° is occupied by a clitic, the result is a “clitic-second” structure. We will see that there is cross-linguistic variation in these structures regarding the position of the inflected verb relative to the clitic; our system makes it possible to analyze this variation in a straightforward way. 2.1. The Wackernagel position in German A striking property of German that differentiates it from other Germanic languages is the fact that pronouns can occur between C° and the subject. The following examples illustrate this phenomenon for both embedded and matrix clauses. (In all our German examples the subject is unstressed; if the subject receives focal stress, the judgments can change, presumably owing to the fact that in this case the subject occupies a different position—a matter which we leave aside.) (25)

a. . . . daß es ihm der Johann gestern gegeben hat. that it him-DAT the J. yesterday given has ‘. . . that John gave it to him yesterday.’ b. Gestern hat es ihm der Johann gegeben. yesterday has it him-DAT the J. given ‘Yesterday John gave it to him.’

134

IP FUNCTIONAL STRUCTURE

In terms of our system, these examples have the following structure:8 (26)

C' Agr1P



{daß hat

Agr1' Agr2P

Agr1˚ NP

Agr2'

der J.

... gegeben ...

es ihm

As (26) shows, the subject (der Johann) can remain in SpecAgr2'; it is not required to move to SpecAgr1'. However, it is also possible to have the order in which the pronouns follow the subject. In our system, it is possible to say that in such cases the clitics are in the same position as in (26), with the subject occupying SpecAgr1': (27)

a.

C' C˚

{daß hat

Agr1P NP

Agr1' Agr1˚

Agr2P

der J.i

es ihm

Agr2'

NP ti

... gegeben ... b. . . . daß der Johann es ihm gestern gegeben hat. c. Gestern hat der Johann es ihm gegeben.

The post-subject position of the pronouns is not a VP-internal position since the VP-internal order of arguments is Dative-Accusative in German. The fact that an Accusative pronoun must appear before a Dative NP indicates that it is outside VP: (28) . . . daß der Johann es dem Hans gegeben hat. that the J. it the-DAT H. given has ‘. . . that John gave it to Hans.’

Moreover, a post-subject pronoun is typically unstressed and, like a pre-subject pronoun, has clitic properties (see Boschetti 1986). This suggests that post-subject es

CLAUSE STRUCTURE AND X - SECOND

135

occupies the same position as pre-subject es, rather than being in a scrambled position of some kind (although other pronouns aside from es may be scrambled; see note 9). As indicated in (26) and (27), the clitic position is Agr1°. A further argument in favor of our approach over a scrambling analysis comes from the absence of “clitic-splitting” in German. That is, we do not find sentences where one pronoun precedes the subject and one follows:9 (29)

a. *. . . daß ihm der Hans es/sie wahrscheinlich gegeben hat. that him-DAT the H. it/her probably given has b. *. . . daß es/sie der Hans ihm wahrscheinlich gegeben hat. that it/her the H. him-DAT probably given has

Example (29) is ungrammatical because there is only one position available for unstressed pronouns: Agr1°. Where there is more than one unstressed pronoun, they all appear in Agr1°. Furthermore, it is unlikely that (25) is a case of scrambling since many speakers reject scrambling to pre-subject position, as shown by (30): (30) ??. . . daß den Roman ihm Johann gegeben hat. that the-ACC novel him-DAT J. given has.

An important confirmation for this analysis of German comes from the fact that Dutch shows a pattern which differs minimally from German, something that we can treat very simply in our system.10 Zwart (1991) argues convincingly that Dutch has object clitics that occupy a VP-external position (see also Jaspers 1989). In fact, Zwart proposes an analysis very similar to ours, involving a head-initial functional projection whose head is to the right of the canonical subject position. This analysis provides a simple explanation for the order “subject > pronouns” in Dutch: (31) . . . dat Jan ‘t gisteren aan Marie gaf. that J. itcl yesterday to M. gave ‘. . . that John gave it to Mary yesterday.’

Example (31) is like the German examples in (27b,c), and we can analyze it in the same way: ‘t is in Agr1° and Jan is in SpecAgr1'. Dutch differs from German, however, in that the order in (25) is ungrammatical: (32) *. . . dat ‘t Jan aan Marie gaf.

This means that the structure in (26) is impossible in Dutch. Since (31) shows, in our terms, that the clitic can occupy Agr1°, (32) must be interpreted as indicating that the subject cannot be in SpecAgr2'. The minimal difference between German and Dutch lies in the possible positions of the Nominative subject. We conclude that, although in both languages SpecAgr1' is a subject position, SpecAgr2' is a position that receives Nominative only in German (see section 3). 2.2.

“Tobler-Mussafia” effects

A related phenomenon to the ones just discussed is found in all the Medieval Romance languages (Old French, Old Italian, Old Spanish) and, at least prescriptively, in European Portuguese (Beninca’ 1989; see also Galves 1991 for an analysis of the clitic systems of both European and Brazilian Portuguese in terms of the double-Agr

136

IP FUNCTIONAL STRUCTURE

clause structure). The phenomenon in question amounts to a ban on clitic-first orders; in constructions where a proclitic would appear in first position, enclisis is obligatory and proclisis is excluded (Mussafia 1983). This phenomenon is known as the “Tobler-Mussafia law” in the traditional literature. The following examples illustrate the operation of the Tobler-Mussafia law in Old French (OF): (33)

a. Toutes ces choses te presta Nostre Sires. (de Kok 1985, 74) all these things you lent our Lord ‘All these things our Lord lent you.’ b. Voit le li rois. (Le Charroi de Nîmes, l. 58) sees him the king ‘The king sees him.’ c. Fust i li reis, n’i oüssum damage. (Harris 1978, 240) were here the king, not here had (we) damage ‘If the king were here, we wouldn’t suffer any damage.’

In (33a), we have a regular V2 sentence (recall that OF was a V2 language; see section 1.2). The direct object toutes ces choses is topicalized to SpecC'; the finite verb appears in second position with the proclitic indirect-object pronoun te. Sentence (33b) is an example of “narrative V1” (see Hirschbuhler 1990). Here we see that the object clitic le follows the inflected verb; it is enclitic, not proclitic. Example (33c) is a verbinitial conditional clause where the same type of enclisis affects the locative i. Example (34) illustrates the Tobler-Mussafia law in matrix interrogatives (de Kok 1985, 82): (34)

a. Conois la tu? know her you ‘Do you know her?’ b. Et quex chevaliers i avra il . . . ? and which knight there will-be there ‘And which knight will be there . . . ?’

In (34a), the clitic la is enclitic on the inflected verb (note that the subject pronoun follows it). In (34b), the clitic i is proclitic to the inflected verb (and the subject pronoun still follows the verb). In (35) and (36), we illustrate the analogous effect in Old Italian (OIt) (all examples are taken from Alberton 1990): (35)

(36)

a. Poi vi trovò tanto oro e tanta ariente. (Novellino, LXXXIV) then there (he) found much gold and much silver ‘Then he found a lot of gold and silver there.’ b. Vogliolo sapere da mia madre. (Novellino, III) (I) want-it to-know from my mother ‘I want to know it from my mother.’ a. Hailo tu fatto per provarmi? have-it you done to try me ‘Have you done it to try me?’ b. Chi si potrebbe tener di piangere e di lagrimare in cotanto dolore? who self could (he) keep from crying and from weeping in such pain ‘Who could keep himself from crying and weeping in such pain?’

CLAUSE STRUCTURE AND X - SECOND

137

A comparison of (35) and (36) with the French examples in (33) and (34) shows that the basic phenomenon is the same: wherever proclisis would place the clitic in first position, the clitic follows the inflected verb. There are certain differences between OF and OIt, however; one important difference is that V1 declaratives are significantly rarer in OF than in OIt (Beninca’ 1989, 5). This may be due to the fact that the verb could remain in Agr1° in OIt matrix clauses (and so, to the extent that OIt was V2, V2 was realized at the Agr1–level), while in OF movement to C° was required (i.e., V2 was realized at the C-level). In this respect, OIt would be like Icelandic (see section 1.1), and OF differs from Icelandic (for other differences between OF and OIt; see Alberton 1990).11 We gloss over these differences here since our main goal is to account for the obligatory enclisis. Our account of the Tobler-Mussafia law runs as follows. Following Kayne (1991), we assume that clitics always occupy functional-head positions. In particular, the clitic is in Agr1° before verb-movement to C°. The difference between enclisis and proclisis lies in whether the clitic forms a complex with the verb, or whether the verb moves to C° independently of the clitic. Enclisis results from the latter situation: the verb moves to C°, while the clitic remains in Agr1°. This is what we find in yes/no questions like (34a) and (36a) and conditionals like (33c), as well as in examples where the verb undergoes structure-preserving topicalization to C°, as in (33b) and (35b). Proclisis, by contrast, involves the formation of the complex [Cl + V] in Agr1°, which may then be moved to C°, as clearly seen in Wh-questions such as (34b) and (36b)) (and in V2 sentences in which [Cl + V] moves to C°, as in OF examples like (33a)). A major advantage of this account is that it directly captures the fact that enclisis is a root phenomenon (Beninca’ 1989; Alberton 1990); like other root phenomena, enclisis involves the presence of an inflected verb in C°. In this respect, our analysis is more straightforward than analyses of the type proposed by Alberton and Beninca’, in which the clitic is in C° with both enclisis and proclisis in matrix clauses and topicalizes to SpecC' to produce enclisis. The clause structure we wish to propose for these languages makes it possible to assume that the topicalization of V is movement to C°, rather than movement to SpecC'. As we said earlier in this essay, this integrates the account of enclisis more directly with accounts of other root phenomena (e.g., subject-aux inversion in English or subjectclitic inversion in contemporary French, and other phenomena discussed in den Besten 1983). A further advantage of our account as compared to approaches of the type in Alberton and Beninca’ is that the topicalization rule is structure-preserving in that it moves the inflected verb, which is an X°, to C°. Our analysis leads to two questions: Why can the verb move to C° independently of the clitic in Agr1°? Why must the verb move to C° independently of the clitic in cases of topicalization and yes/no questions? There are two possible answers to the first of these questions: either the verb “excorporates” from Agr1° to C°, leaving the clitic behind in Agr1° (see Roberts 1991 on excorporation), or the verb is able—under the right conditions—to move from Agr2° to C° “skipping” Agr1°. Of these two possibilities, we will assume the second in what follows (see Roberts 1993b). For the second question, we see no alternative to the traditional idea that the languages that show this effect do not permit clitic-first orders, which is presumably a phonological constraint. Hence, the verb “skips” Agr1° exactly where the “regular” movement

138

IP FUNCTIONAL STRUCTURE

through Agr1°, picking up the clitic on the way, would give rise to a clitic-first order with [Cl + V] in C°. The question that now arises is, Why does V only “skip” Agr1° when moving through Agr1° would violate the ban on clitic-first orders? As we will suggest in section 3, the clitic in Agr1° attracts the inflected verb in the usual case (see note 13). The clitic thus seems to impose two distinct requirements: (a) it cannot appear first, and (b) it must combine with the inflected verb. Where some other element (e.g., a Wh-constituent) appears in first position for independent reasons, requirement (a) is automatically satisfied and requirement (b) can be most economically satisfied by V-movement to Agr1°, and, hence, it must be satisfied in this way, following Chomsky (1989)—although other requirements, e.g., Rizzi’s (1996) Wh-criterion, may lead to the [Cl + V] complex moving further. Alternatively, if no other element appears first, the two requirements imposed by the clitic are satisfied most economically if V moves to C°, skipping Agr1°, and the clitic left-adjoins to V (the latter operation possibly taking place in PF). Thus we see that the Tobler-Mussafia Law is a “lastresort” operation in the sense of Chomsky (1989). Concerning comparative questions, our analysis makes the prediction that a language which shows the Tobler-Mussafia effect has Agr1°. We can relate the possibility of independent movement of the verb with respect to the clitic to the existence of this position in the following way: in a system with both Agr1° and Agr2°, Agr2° contains the inflectional affixes that are required to form the inflected verb. Hence the verb-stem must move to Agr2° so that an inflected verb can be formed. However, verb-movement to Agr1° is not forced by such morphological factors (although the verb may be “attracted” to Agr1° independently of the need to pick up an inflectional affix; see section 3 and note 13). Contemporary French and Italian have only one Agr position, and this is the position that contains the finite verbal inflection and to which clitics are adjoined. Thus the finite verb always forms a unit with object clitics in these languages, and so, when the finite verb appears initially in C° in yes/ no questions (as in (37)) and certain types of conditionals (as in (38)), the clitic remains proclitic. We illustrate this with a conditional in Italian, as no overt form of subject-verb inversion is found in interrogatives; see Rizzi (1982) on this form of conditional inversion: (37) [C° [Agr° La connais ] ]—tu? her know you ‘Do you know her?’

(vs. (34a))

(38) [C° [Agr° L’avessi ] ] io saputo in tempo, . . . it-hadsubj I known in time ‘Had I known it in time, . . .’

Note that our analysis does not claim that clitics form a unit with the nonfinite verb. As Kayne (1991) shows, it is probably desirable to maintain that enclitics on nonfinite verbs in some Romance languages occupy a syntactic position separate from the verb—for example, in forms like Modern Italian farlo ‘to do it’. Moreover, Medieval Romance languages are like contemporary Germanic languages in that both groups tolerate clitic elements which can be independent of the finite morphology; this is completely impossible in most contemporary Romance

CLAUSE STRUCTURE AND X - SECOND

139

languages (but not in Romanian; see Motapanyane 1991 for an analysis of clitic placement in this language, which makes use of the same “double-Agr” system as that being proposed here). In our terms, both Medieval Romance and contemporary Germanic have clitics that occupy Agr1°, a functional-head position that is independent in principle of the position of finite morphology—Agr2°. Furthermore, our analysis captures the traditional idea that Tobler-Mussafia effects are related to Wackernagel’s Law; as we have seen, both sets of phenomena crucially involve the presence of an “autonomous” clitic in Agr1°. 2.3. V3 orders in Old English and Old High German In this section, we discuss one phenomenon of clitic-placement found in the older Germanic languages, Old English (OE) and Old High German (OHG). The data on OE are taken from van Kemenade (1987) and those on OHG from Tomaselli (1991). Both OE and OHG were V2 languages, although, as we shall see later in this discussion, this does not necessarily imply that CP was activated in all matrix clauses in these languages. In this section, we focus on one case of V3 word order that is found in both of these languages, in which the second element is a clitic. (In OHG, this order is restricted to subject clitics; in OE, complement clitics are found in this position, too.) This order is the same as that in the Medieval Romance languages in matrix declarative sentences: (39)

a. God him worhte þa reaf of fellum. (van Kemenade 1987, 114) God them wrought then garments of skin ‘God then made them garments of skin.’ b. Forðon we sceolan mid ealle mod & mægene to Gode gecyrran. therefore we shall with all mind and power to God turn ‘Therefore we must turn to God with all our mind and power.’ (van Kemenade 1987, 110) c. Dhes martyrunga endi dodh uuir findemes mit urchundin dhes heilegin chiscribes. his martyrdom and death we demonstrate with evidence of the holy writings ‘We demonstrate his martyrdom and his death with evidence from the holy scriptures.’ (Tomaselli 1991, 3)

There is a major difference with the Romance Tobler-Mussafia effects discussed in the previous section, however. In contexts where the verb is in C° and an XP of a particular class is in SpecC', the clitic follows the verb. In OE, these contexts involve a Wh-constituent, the negative element ne or the adverb þa ‘then’ in initial position, all of which can be plausibly treated as elements in SpecC' triggering movement of the inflected verb to C° (Tomaselli 1991, 6 implies that the same is true in OHG; cf. Tomaselli 1990): (40)

a. Ne geseah hine nan man nates-hwon yrre. (van Kemenade 1987, 114) not saw him no man so little angry ‘No one ever saw him so little angry.’ b. Hwæt sægest þu, yrþlincg? (van Kemenade 1987, 138–139) what saist thou, ploughman ‘What do you say, ploughman?’

140

IP FUNCTIONAL STRUCTURE

Parallel with our analysis of the Tobler-Mussafia effect in section 2.2, we propose that in examples like (39) the clitic occupies Agr1°. The verb can occupy this position, as the following OE case of verb-subject order shows (from van Kemenade 1987, 114): (41) Fela spella him sædon þa Beormas, ægþer ge of hiera agnum lande, . . . many stories him told the Permians, both of their own country ‘The Permians told him many stories, both about their own country, . . .’

We cannot provide the parallel evidence in OHG, since the clitic in Agr1° is always a subject clitic. However, OHG examples like (39c) can be analyzed as having the inflected verb in Agr1° together with the clitic. In Germanic matrix declaratives, then, the CP-level is not activated. The main verb can also appear in a position quite distant from that occupied by the clitic in embedded clauses and in the characteristically non-V2 second conjuncts in main-clause coordination (see Kiparsky 1994). Notice that the clitic can appear either before or after the subject in OE, as in Modern German (see section 2.1): (42)

a. . . . þæt him his fiend wæren æfterfylgende. (van Kemenade 1987, 113) that him his enemies were following ‘. . . that his enemies were chasing him.’ b. & se cyng him eac wel feoh sealde. (van Kemenade 1987, 113) and the king him also well property gave ‘and in addition the king gave him much property.’ c. . . . daz íh nîeuuánne necúme in conventicula haereticorum (Tomaselli 1991, 29) that I never not-come in the circle of-heretics ‘. . . that I never come into the circle of heretics.’

We take it that the inflected verb is in Agr2° in these examples. As in Medieval Romance languages, Agr2° is the position in which the inflected verb is formed by combining with the agreement affix, while Agr1° is the clitic position. The principal difference between these Germanic languages and the Medieval Romance languages discussed in section 2.2 is that in the Germanic languages the verb only moves to Agr1° in matrix declaratives, while in the Romance languages the verb moves to Agr1° both in matrix and in subordinate clauses, and, indeed, in any clause where the disallowed clitic-first order will not result from continued movement to C°. In Germanic interrogatives, and so on, where the inflected verb is clearly in C° (see (40)), the verb “skips” Agr1°, and so the clitic appears in third position here. This is consistent with the fact that verb-preposing is a root phenomenon in Germanic, while it is found also in embedded clauses in Romance. Our analysis captures the same range of facts as those proposed by van Kemenade (1987) for OE and Tomaselli (1991) for OHG. Compared to those analyses, ours has the advantage of greater generality, in that it relates the Germanic phenomena to the Tobler-Mussafia effects found in Romance. Also, the more elaborated clause structure that we propose makes it possible to distinguish the clitic position from the position of the inflected verb and to maintain that the inflected verb is always formed in the standard way in these languages, by head-movement into a position containing the agreement affix (see Pollock 1989). In contrast, neither Tomaselli nor van Kemenade allows for a position reserved purely for verbal inflection; in such a

CLAUSE STRUCTURE AND X - SECOND

141

framework, it is necessary to posit special rules for the formation of the inflected verb when it is “distant” from the clitic in examples like (42), an undesirable consequence that we are able to avoid.

3. Subject positions In this section, we show how the subject positions are determined in the various languages that we have considered (we limit our attention throughout to non-qsubject positions—i.e., those outside VP). A central part of our discussion is based on the approach to Nominative-Case assignment advocated in Koopman and Sportiche (1991). Koopman and Sportiche propose that UG makes available two mechanisms of Nominative-Case assignment. Nominative Case can be assigned either under government, illustrated in (43a), or under agreement, illustrated in (43b): (43)

a.

X' YP



Y'

NP b.

XP YP

X' X˚

Koopman and Sportiche consider that the choice among these two mechanisms is a “pure” parametric choice, in the sense that each language is in principle free to choose among these possibilities. We will suggest, however, that the choice of Nominativeassignment under agreement depends on other factors—in particular, on the nature of the Case-assigning head X°—and that the only pure parametric choice is that of selecting (43a) or not. In the context of the general assumption that Agr1° is the Nominative-Case assigner in all the languages we are concerned with (but see note 12 for a refinement of this assumption), if a language chooses the government option in (43a) then SpecAgr2' will be the subject position; and if a language chooses the agreement option in (43b), then SpecAgr1' will be the subject position. We further assume that the two options can be combined, making both SpecAgr1' and SpecAgr2' possible subject positions. We understand “subject position” to mean both Case-position—that is, a position to which Nominative is assigned—and A-position (although we are excluding q-positions here, as already mentioned here). The generalization that emerges from a consideration of the languages we have looked at is that, on the one hand, wherever Agr1° is a clitic position, its specifier is a subject position. On the other hand, where Agr1° is a “pure” verbal position (in the sense that the only element that ever appears there is the verb), its specifier is a topic position. Suppose, then, that it is the fact that Agr1° may host a clitic that makes it

142

IP FUNCTIONAL STRUCTURE

possible for it to assign Nominative Case under agreement. We will return to this idea later in this discussion. First, however, let us consider the evidence from the various languages that points to this conclusion. Consider first the situation in Icelandic (and Yiddish, see section 1.3). The evidence we presented in section 1.1 shows that in this language Agr1° is a position that always hosts the inflected verb and that SpecAgr1' is a topic (non-Case, A') position, while SpecAgr2' is a subject position. Schematically, then, we have the following: (44) [Agr1P TOP [ [Agr1° Vi ] [Agr2P Subj. [ [Agr2° ti ] . . . ] ] ] ]

This structure is common to main and embedded clauses. We can account for its basic properties in terms of Koopman and Sportiche’s system by saying that Agr1° assigns Nominative under government only. Note that Agr1° always contains the inflected verb and never contains a clitic in Icelandic; we propose that the lack of Case-assignment under agreement is related to this point. Now compare German, on the basis of the analysis given in 2.1. In main clauses, we have the following: (45) [CP TOP [ [C° V ] [Agr1P Subj. [ [Agr1° Cl ] [Agr2P Subj. . . . ] ] ] ] ]

As in standard approaches, we take it that the verb moves to C° in main clauses and that SpecC' is a topic position. Agr1° is the Wackernagel position and as such may contain one or more clitics. SpecAgr1' is a potential subject position in German and not a topic position (see section 2.1); this is the fundamental difference between German and Icelandic. SpecAgr2' is also a possible subject position in German, as it is in Icelandic. In terms of Koopman and Sportiche’s system, we can say that in German Agr1° assigns Nominative under both government and agreement. For this reason, SpecAgr1' is a subject position rather than a topic position. As mentioned earlier in this discussion, we believe that there is a nontrivial connection between the possible presence of a pronoun in Agr1° and the fact that SpecAgr1' is a subject position. There is a further complication concerning the topic positions in both German and Icelandic—that is, SpecC' in German and SpecAgr1' in Icelandic (this complication is probably common to all the V2 Germanic languages and to OF (see Roberts 1993b, 2.3). These positions also display properties of subject positions. Apart from being available for topicalization of any XP, they accept elements that are clearly not topics: lexical expletives and non-topic subjects (both pronominal and nonpronominal); see Cardinaletti 1990b, 1994 for further arguments for this analysis. We can treat this situation in terms of Rizzi’s (1991) proposal that any specifier that agrees in f-features with its head can be an A-position (and therefore a subject position; other A-positions are all structural complements, not specifiers). This agreement can naturally be thought of as a coindexation relation between the subject in SpecX' and X. We refer to this possibility henceforth as SPEC-HEAD COINDEXATION: this substantive relation is a subcase of the structural relation of spec-head agreement. Hence, precisely where a subject NP appears in SpecAgr1' in Icelandic, or in SpecC' in German, the position can be assigned Nominative since there will be agreement in f-features between the head and the NP. So, even in Icelandic,

CLAUSE STRUCTURE AND X - SECOND

143

Nominative can be assigned under agreement in these conditions.12 Nominativeassignment under spec-head coindexation can thus be distinguished in principle from Nominative-assignment under spec-head agreement, since in the latter situation it is possible that the head that assigns Nominative is not coindexed with its specifier (on this point we differ from the proposals made by Koopman and Sportiche 1991). In this sense, as mentioned, we do not take the possibility of Nominativeassignment under agreement given in (43b) to be an absolute parametric choice but, rather, a possibility related to the nature of the Nominative-assigning head. On the status of the possibility of Nominative-assignment under government, see the following discussion. Next, let us consider the more complex cases of OE and OF. Following the discussion in section 2.3, OE main clauses show the pattern in (46a), and OE embedded clauses show that in (46b): (46)

a. [Agr1P TOP [ [Agr1° Cl + Vi ] [Agr2P Subj. [ [Agr2° ti ] . . . ] ] ] ] b. [Agr1P Subj. [ [Agr1° Cl ] [Agr2P Subj. [ [Agr2° V ] . . . ] ] ] ]

Example (46b) clearly indicates that Agr1° can assign Nominative under either government or agreement in OE, and so both SpecAgr1' and SpecAgr2' are possible subject positions (see section 2.3). So, for embedded clauses in OE, no problem arises. (46a), which is motivated by the analysis of V3 orders in OE that we gave in section 2.3, leads to the following question: Why is SpecAgr1' a subject position in embedded clauses but a topic position in main clauses? The fact that SpecAgr1' is a subject position in embedded clauses follows from our proposal that the possible presence of a pronoun in Agr1° is correlated with the capacity to assign Case to SpecAgr1'. A pronoun can also occur in Agr1° in main clauses, as seen in (46a) (see section 2.3 for examples and discussion). However, Agr1° is also a position for the inflected verb in (46a). This must be the factor that makes the difference: where Agr1° is a position to which the inflected verb can move, its specifier is a subject position only where the two are coindexed (the same is true for C° in German, as we saw in the preceding discussion). If Agr1° hosts clitics but not inflected verbs, however, its specifier is uniquely a subject position (see the following discussion for a slight refinement). Finally, let us briefly consider the situation in OF, on the basis of the analyses proposed in sections 1.2 and 2.2 ((47a) is the structure of a main clause and (47b) that of an embedded clause): (47)

a. [CP TOP [ [C° Cl+Vi ] [Agr1P Subj. [ [Agr1° ti ] [Agr2P Subj. . . . ] ] ] ] ] b. [Agr1P Subj. [ [Agr1° Cl+V ] [Agr2P Subj. . . . ] ] ]

In both main and embedded clauses, Agr1° assigns Nominative under both government and agreement. Thus, both SpecAgr1' and SpecAgr2' are subject positions. There is an important difference between OF and what we saw for OE. In OE, the specifier of a head containing the combination [Cl + V] (SpecAgr1' in matrix clauses) is a topic position; in contrast, the seemingly identical head position—[Agr1° Cl + V]— has a subject-position specifier in OF embedded clauses, as (47b) shows.

144

IP FUNCTIONAL STRUCTURE

We propose that the difference between the two cases lies in the different status of V-to-Agr1 movement in OF as compared to OE, which we already noted in section 2.3. In OF, V moves to Agr1° in all embedded clauses and all root clauses where no violation of the ban on clitic-first orders will result, so this movement is clearly the unmarked option. In contrast, in OE V moves to Agr1° only in root clauses. Thus V-to-Agr1 is a more general process in OF than in OE, since the verb forms a unit with the clitic wherever it can. This is clearly not the case in OE. Since the clitic in Agr1° triggers V-movement to this position in OF, we can consider Agr1° to be essentially a clitic position; hence, its specifier is a subject position. Conversely, in OE the head [Agr1° Cl + V] is formed by an instance of verb-movement that is not triggered by the clitic. Because of this, we consider these cases of Agr1° to be verb positions, and so the account given earlier (after (46)) applies.13 To sum up, we can account for the different subject positions attested in the languages we have discussed. Icelandic allows Nominative-assignment only under government, while the other languages discussed here allow both possibilities. We arrived at these conclusions by holding constant the following two assumptions: (48)

a. Agr1° assigns Nominative Case (although it may not be the only Nominativeassigner; see note 12). b. The nature of a specifier position depends on the possible content of the head it specifies.

Of these assumptions, (48b) deserves further comment. What emerged from our discussion is that Agr1° (and perhaps also C°; see note 12) assigns Case under spechead agreement where (a) it contains an inflected verb that is coindexed with the NP in the specifier position or (b) it is a clitic position. This general conclusion is consistent with the spirit of Rizzi’s (1991) proposal that A-position specifiers are in a Spechead relation with Agr. As we mentioned at the beginning of this section, in turn, this suggests that the possibility of assigning Nominative under spec-head agreement is not a “pure” parameter but, rather, is related to the intrinsic properties of the Nominative-assigner. In contrast, the possibility of assigning Nominative under government may be a “pure” parametric choice. We saw evidence in section 2.1 that in Dutch Nominative Case can only be assigned under agreement (see (31) vs. (32)). In Dutch, Agr1° is clearly a clitic position, as Zwart’s (1991) data show; hence, SpecAgr1' is a subject position. However, SpecAgr2' cannot contain the subject, as (32) shows. We interpret this to mean that the government option is not chosen in Dutch. Where the verb raises to C°, SpecAgr1' remains a subject position, since Agr1° is not carried along by this movement; as in German, the verb “skips” Agr1° on its way to C°. So Agr1° continues to assign Nominative to its specifier in these conditions. Dutch shows the same evidence as German and Icelandic that SpecC’ may be a subject position (see Travis 1984); we can treat these facts exactly in the same way as we have treated the parallel German facts (see note 12). To see that this system has some generality, and is not limited to languages with the “double-Agr” structure, let us briefly consider the situation in languages with just one Agr projection. Here the same possibilities of Nominative-Case assignment are available, and they are determined in the same way in that Nominative-assignment

CLAUSE STRUCTURE AND X - SECOND

145

under agreement depends on properties of (the single) Agr while Nominativeassignment under government is a “pure” parametric choice. Modern Standard Italian is an example of a language in which Nominative is assigned under agreement only. This is consistent with our proposals in that Agr in Italian can contain both complement clitics and the inflected verb, as is well known. The possibility of Nominative-assignment under government does not exist as the ungrammaticality of the following example shows (see Roberts 1993b; Rizzi 1996): (49) *Ha Gianni fatto questo? has G. done this ‘Has John done this?’

(Here we are only concerned with the Case properties of Agr°; it is possible that “free inversion” involves Nominative-assignment by T° under government to a VP-internal subject position; see Rizzi 1990a.) Where Nominative-assignment under government is impossible in a system with only one Agr-projection, verb-second and other kinds of inversion around a nominal subject are impossible (see Rizzi and Roberts 1996 on French). More generally, it is probable that systems with just one Agr-projection must allow Nominativeassignment under agreement since they are unable to exploit the government configuration in many cases (this conclusion depends on the assumption that SpecT’ is an A'-position; see section 1.1). English, like Italian, has only one Agr-projection. Nominative-assignment under agreement is possible, since there is always a subject in SpecAgr', triggering the required spec-head coindexation. Nominative-assignment under government is also possible in English: (50) Has John done that?

The existence of Nominative-assignment under government in English shows that this possibility holds no necessary relationship to full verb-second. Tables 5-1 and 5-2 sum up the properties of the different grammatical systems that we have discussed up to now. Recall that we have proposed that the possibility of assigning NOM under agreement is not a “pure” parameter but, rather, a function of parametrized properties of the NOM-assigning head.

Table 5-1.

Variation among double-Agr languages

Language

SpecAgr1'

Agr1°

SpecAgr2'

Icelandic Yiddish Old French German Dutch Old English, Old High German

Topic Topic Subject Subject Subject Topic (main), Subject (embedded)

Governed NOM Governed NOM Governed/Agreeing Governed/Agreeing Agreeing NOM Governed/Agreeing

Subject Subject Subject Subject —a Subject

a. The position is never filled.

NOM NOM

NOM

146

IP FUNCTIONAL STRUCTURE

Table 5-2.

Parameters proposed

Parameter

Positive Value

Negative Value

Agr-recursion

See languages in table 5-1

V2

Germanic (other than English), Old French Germanic (other than Dutch), Old French

Mod. English, Mod. French, Mod. Italian Mod. Romance, English

Governed

NOM

Mod. Romance, Dutch

4. Null subjects In this section, we will show how the analysis developed in the previous sections can account for certain aspects of the distribution of null subjects in V2 languages. We adopt the conception of the pro-module outlined in Rizzi (1986). The most important aspect of this system is the fact that null subjects depend on Case-marking for formal licensing in the sense that a null subject must occupy a position of potential (although not necessarily actual) Case-assignment. Because of this, what we saw in the previous sections concerning the interaction of the double-Agr clause structure with Nominativeassignment has interesting consequences for the distribution of null subjects. 4.1. Expletive null subjects In Icelandic and German, there is an alternation between an overt expletive and a null expletive. This alternation is determined by several factors, most importantly the licensing condition for null subjects. In these languages, null subjects can only be licensed under government or under “pure” spec-head agreement (in the sense discussed in section 3). This gives rise to complementary distribution between the null expletive and the lexical expletive. Wherever it is possible for the expletive null subject to appear, it must appear, and so the lexical expletive is excluded (see Cardinaletti 1990a, 1994 for an account of this). Wherever the expletive null subject is not licensed, the overt expletive appears. This situation is illustrated by the following pairs of German, (51), and Icelandic, (52), matrix sentences:14 (51)

a. *(Es) wurde getanzt. it was danced b. Gestern wurde (*es) getanzt. yesterday was it danced ‘Yesterday there was dancing.’

(52)

a. *(Það) var dansað. it was danced b. I gær var (*það) dansað. yesterday was it danced ‘Yesterday there was dancing.’

In the (a)-examples, the expletive is in topic position (although it is not a topic, see section 3 and the following discussion). Since this is a position where null subjects

CLAUSE STRUCTURE AND X - SECOND

147

are not licensed, the expletive pro cannot appear, and hence the overt expletive is required. Null subjects cannot be licensed in this position since this is a configuration in which Nominative is assigned under spec-head coindexation (where the head is C° in German and Agr1° in Icelandic); as we will see in more detail later in this discussion, Spec-head coindexation does not license a null subject. In the (b)-examples, in contrast, the expletive appears in a position where null subjects are licensed, and hence the lexical expletive is ungrammatical. In Icelandic, the situation in matrix clauses just illustrated is also found in all types of embedded clauses, as long as the thematic subject is indefinite and is as such able to remain in VP where it receives Partitive Case (in the following examples, we use enginn ‘nobody’). As the examples show, the null expletive is in SpecAgr2' (we are grateful to Hoski Thráinsson for providing us with these examples): (53)

a. Ég harma að þegar skuli pro enginn hafa lesið þessa bók. I regret that already should nobody have read this book ‘I regret that nobody should have read this book already.’ b. Ég spurði hvort þegar hefði pro enginn lesið þessa bók. I asked whether already had nobody read this book ‘I asked whether anybody had read this book already.’ c. sú staðreynd að þegar hefur pro enginn lesið þessa bók the fact that already has nobody read this book ‘the fact that nobody has read this book already’ d. bókin sem þegar hefur pro enginn lesið book-the that already has nobody read ‘the book that nobody has read already’

As we said, overt expletives cannot appear where a null expletive is possible (Thráinsson, personal communication): (54)

a. *Ég held/harma að þegar skuli það María/enginn hafa lesið þessa bók. I believe/regret that already should it M./nobody have read this book b. *Ég spurði hvort þegar hefði það María/enginn lesið þessa bók. I asked whether already has it M./nobody read this book c. *sú staðreynd að þegar hefur það María/enginn lesið þessa bók the fact that already had it M./nobody read this book d. *bókin sem þegar hefur það María/enginn lesið book-the that already had it M./nobody read

In (53), the null expletive occupies SpecAgr2'. As we saw in section 3, this is a subject position in the sense that it is a position that receives Nominative Case. Hence, the licensing condition on pro is satisfied here, and the null expletive is present, making the overt expletive impossible as in (54). The reverse situation is found in SpecAgr1'. This can be seen in examples like the following (Thráinsson, personal communication): (55)

a. Ég harma að *(þaði) skuli ti enginn hafa lesið þessa bók. I regret that (it) should nobody have read this book ‘I regret that nobody should have read this book.’ b. Ég spurði hvort ?(þaði) hefði ti enginn lesið þessa bók. I asked whether (it) had nobody read this book ‘I asked whether anybody had read this book.’

148

IP FUNCTIONAL STRUCTURE

c. sú staðreynd að *(þaði) hefur ti enginn lesið þessa bók the fact that (it) has nobody read this book ‘the fact that nobody has read this book’ d. bókin sem *(þaði) hefur ti enginn lesið book-the that (it) has nobody read ‘the book that nobody has read’

(There is a complication concerning (55b), which improves this example in comparison with the others; see the following discussion.) Moreover, when there is no subject, for example, in impersonal passives, það is able to appear in SpecAgr1' (example from Rögnvaldsson 1984, 17–18): (56) Ég veit að *(þaði) er ti ekið vinstra megin í Astralíu. I know that (it) is driven on the left in Australia ‘I know that people drive on the left in Australia.’

In both (55) and (56), we see that the overt expletive is required, and pro is impossible. The null subject is ungrammatical in SpecAgr1' because the only way in which Nominative Case can be assigned to this position is by spec-head coindexation and not by “pure” spec-head agreement or by government. In general, this implies that spec-head coindexation is not adequate for licensing pro. This approach is confirmed by the behavior of expletives in interrogatives where the inflected verb is raised to C°. In such examples, the overt expletive is impossible and the null expletive is required: (57) Var (*það) dansað? was (it) danced ‘Was there dancing?’

Here the presence of the verb, containing Agr1°, in C° makes it possible to assign Nominative to SpecAgr1' under government. Hence pro is possible. In (55), as in (53), there is a definiteness effect. See the ungrammaticality of the parallel examples where the thematic subject is definite (Thráinsson, personal communication): (58)

a. *Ég harma að það skuli María hafa lesið þessa bók. I regret that it should M. have read this book b. *Ég spurði hvort það hefði María lesið þessa bók. I asked whether it had M. read this book c. *sú staðreynd að það hefur María lesið þessa bók the fact that it has M. read this book d. *bókin sem það hefur María lesið book-the that it has M. read

We attribute the ungrammaticality of (58) to the fact that there are two NPs that compete for Nominative Case: það and María. In contrast, the indefinite enginn in (53) and (55) is able to receive Partitive Case (see Belletti 1988; Vikner 1995), making Nominative Case available for það.15 Given these considerations, it is clear that no definite NP can appear lower than SpecAgr2'. However, it is clear from the discussion in section 1.1 that there is no definiteness effect with embedded V2 in Icelandic. Therefore, this consideration further confirms that we cannot adopt an analysis of Icelandic embedded V2 of the type pro-

CLAUSE STRUCTURE AND X - SECOND

149

posed by Rögnvaldsson and Thráinsson (1990), and by Diesing (1988, 1990) and Santorini (1989) for Yiddish embedded V2 (moreover, given the general similarities between Yiddish and Icelandic noted in section 1.3, this analysis probably does not hold for Yiddish, either). This type of analysis relies on the idea that the standard SpecI'—the canonical subject position of English—is able to be a topic position and that the subject can remain in an “internal” position. However, the contrasts between (53) and (54) and between (55) and (58) show that the “internal” position is subject to a definiteness effect. All this supports our analysis of embedded V2 in terms of the double-Agr structure, and so supports the postulation of that structure as a general possibility. Icelandic also allows an empty SpecAgr1' in embedded clauses. This is only possible where C° is able to license an expletive null subject and only where the subject is indefinite. The conditions under which C° can license an expletive null subject are rather difficult to understand; however, they correspond closely to those where in French Stylistic Inversion an expletive null subject may appear, being essentially cases where C° is [+wh] or subjunctive (see Kayne and Pollock 1978; Pollock 1986) (examples (59a–c) from Rögnvaldsson and Thráinsson 1990, 31; example (59d) from Rögnvaldsson 1984, 17–18; example (60) from Kayne and Pollock 1978): (59)

(60)

a. *Ég spurði hvort pro hefði María/hún lesið þessa bók. I asked whether had M./she read this book b. ?Ég spurði hvort proi hefði ti enginn lesið þessa bók. I asked whether had nobody read this book ‘I asked whether anybody had read this book.’ c. Hann spurði hvort proi hefðu ti komið margir gestir. he asked whether had come many guests ‘He asked whether many guests had come.’ d. Ég veit að proi væri ti ekið vinstra megin í Astralíu. I know that is driven on the left in Australia ‘I know that people drive on the left in Australia.’ J’exige que pro soit procédé au réexamen de cette question. I require that be proceeded to-the reexamination of this question. ‘I require that this question be reexamined.’

As the examples in (59) indicate, the definiteness effect here is due to the fact that pro is base-generated in SpecAgr2' and raises to SpecAgr1'. The trace of pro in SpecAgr2' blocks movement of a definite subject into that position. One aspect of sentences of this kind may seem problematic, given what we said above: in at least some cases, null and overt expletive subjects are apparently not in complementary distribution. This can be seen by comparing examples like (55b) with (59b); we repeat (55b) for convenience here: (55)

b. Ég spurði hvort þaði hefði ti enginn lesið þessa bók. I asked whether it had nobody read this book ‘I asked whether anybody had read this book.’

This noncomplementary distribution of það and pro is not inconsistent with the basic idea that an overt expletive cannot appear wherever a null expletive can be licensed. The crucial point here is that there are two different potential licensers for expletive pro: Agr1° and C° (where C° is [+wh] or subjunctive). Suppose that each of these

150

IP FUNCTIONAL STRUCTURE

heads is only able to license pro under government, so Agr1° licenses pro in SpecAgr2' and C° licenses pro in SpecAgr1'. When Agr1° licenses pro in SpecAgr2', það cannot appear in this position and so must move to SpecAgr1'. When C° licenses pro in SpecAgr1', það cannot appear there for exactly the same reason. This is what we see in (59). So we arrive at the result that the only expletive possible in SpecAgr2' is the null one, while in SpecAgr1' either það or pro is possible, depending on which head— Agr1° or C°—is the licenser. A similar situation occurs in French with Stylistic Inversion; alongside examples like (60), the comparable case with the overt expletive il is possible: (61) J’exige qu’il soit procédé au réexamen de cette question. I require that it be proceeded to-there-examination of this question ‘I require that this question be reexamined.’

The preceding paragraphs show that our proposed clause structure for Icelandic can account for the distribution of null and overt expletives in both matrix and embedded clauses. A major advantage of our account is that it explicitly relates the phenomenon of embedded verb-second to the distribution of these expletives in embedded clauses. In German, the situation regarding null expletives is largely comparable to what we have just seen for Icelandic. A null expletive is possible in SpecAgr1', as shown by (51b) and the following: (62)

a. . . . daß pro es ihm ein Mann gegeben hat. that it to-him a man given has ‘. . . that a man has given it to him.’ b. . . . daß pro es ihm dieser Mann gegeben hat. that it to-him this man given has ‘. . . that this man has given it to him.’

In (62a) the thematic subject, ein Mann, is indefinite and so able to receive Case in VP independently of Agr1°. Hence pro can appear in SpecAgr1'. In (62b), the thematic subject is definite, and, as its position relative to the clitics shows, it occupies SpecAgr2’. In this example, then, pro must occupy SpecAgr1'. Note that here pro does not receive a Case; nevertheless it is licensed since it occupies a potential Caseposition (in the sense that Nominative can be assigned under spec-head agreement here; see section 3). As we saw for SpecAgr1' in Icelandic, no null subject can be licensed by spechead coindexation. In German the same situation holds at the level of SpecC' (it does not hold at SpecAgr1' as this position receives Nominative Case under “pure” spechead agreement in German). This is illustrated by both (51a) and the following: (63) Esi hat ti ihm ein/dieser Mann das Buch gegeben. it has him a/this man the book given ‘A/This man has given him the book.’

In this section we have shown how the proposals we made in previous sections regarding both the clause structure and the modes of Nominative assignment in German and Icelandic can provide a straightforward account of the distribution of overt and null expletives, a matter that has long been regarded as problematic. This account de-

CLAUSE STRUCTURE AND X - SECOND

151

pends crucially on the assumption that pro can be licensed in configurations of potential Case assignment, with the exception of those that depend on spec-head coindexation. 4.2.

Referential null subjects

We next consider the interaction of our proposals with the analysis of referential null subjects, concentrating on Old French. As we saw in section 1.2, OF was a verbsecond language, and it is well known that the distribution of null subjects interacted with the movement of the verb (see Thurneysen 1892; von Wartburg 1934; Price 1971; Beninca’ 1986; Foulet 1982; Beninca’ 1983–84, 1989; Vanelli, Renzi, and Beninca’ 1985; Adams 1987a,b). Null subjects are frequently found with verb-second in OF. The following lines from Le Charroi de Nîmes illustrate this: (64)

a. Muetes de chiens font avec els mener. (Compl V) troups of dogs make (they) with them bring ‘They have troups of dogs brought with them.’ b. Par Petit Pont sont en Paris entré. (PP V) by Petit-Pont are (they) in Paris come ‘They entered Paris by the Petit-Pont.’ c. Li cuens Guillelmes fu molt gentix et ber. (Subj V) the count G. was very kind and good ‘Count Guillelmes was very kind and good.’

Beginning with Vanelli, Renzi, and Beninca’ (1985), various authors have made an analytic connection between the verb-movement required for verb-second and the licensing of referential null subjects in this language. We concur with this basic approach, and, following Roberts (1993b), we interpret the connection in terms of the relationship between Case theory and formal licensing. So the connection between null subjects and verb-movement in OF is due to the fact that verb-movement creates the configuration in which pro can be formally licensed. To the extent that verb-second is cross-linguistically a root phenomenon, this approach leads to the expectation that OF null subjects are a root phenomenon. However, there is evidence (from Adams 1988a,b,c; Dupuis 1988, 1989; Hirschbuhler and Junker 1988; Hirschbuhler 1990; and Vance 1988) that null subjects were possible in a range of embedded clauses, including, most importantly, Wh-complements (they are also possible in complements to bridge verbs, which we argued in section 1.1 involve CP-recursion). Adams gives the following examples of V1 and V2 orders with null subjects in Wh-complements (1988b, 5–6 (9), (11); 1988c, 10–11 (9), (11)): (65)

a. Je sui le sire a cui [volez parler]. I am the lord to whom (you) wish to-speak ‘I am the lord to whom you wish to speak.’ b. L’espee dont [s’estoit ocis] the sword by-which himself-was killed ‘the sword that he killed himself with’ c. quant [vit le roi] when (he) saw the king ‘when he saw the king’

152

IP FUNCTIONAL STRUCTURE

Continuing to assume that the inflected verb cannot move to a [+wh] C°, we are led to the conclusion that the verb is in Agr1° in these examples. If this is so, we need to explain where the null subject is and how it is licensed. The same question emerges from a consideration of the null subjects combined with Stylistic Fronting that we saw in section 1.2 (see (12)). Hirschbuhler (1990) presents evidence that referential embedded null subjects are found in just one phase of the OF period. He studies twelfth- and thirteenth-century prose and verse and concludes that there are essentially two systems at work, according to the nature of the text. One system, which we call the “conservative” system, allows null subjects in embedded clauses fairly freely, as in (65). This system is typical of twelfth-century prose texts and for both twelfth- and thirteenth-century verse. In contrast, in thirteenth-century prose texts, embedded V1 orders with null subjects are limited to two types of rather fixed expressions. Call this the “advanced” system. Here are some representative examples of the advanced system: (66)

a. se ne fu chiés le Roi Mehaigniée (Hirsehbuhler 1990, 7 (14)) if not were “chez” King M. ‘if it were not at the court of King M.’ b. quant vint a cele hore que . . . (Hirschbuhler 1990, 7 (15a)) when came to that time that ‘when the time came that . . .’

These are clearly cases of expletive null subjects. Moreover, they seem to represent an unproductive option. We propose, on the basis of Hirschbuhler’s data, that the “double Agr” system was at work in the more conservative texts. So, for instance, we assign the following structures to the embedded clause in (65a): (67)

CP XP

C' Agr1P

C˚ [+wh]

Agr1'

Spec

Agr2P

Agr1˚

Agr2'

Spec

a cui

proi

volezk

ti

Agr2˚

TP

tk

tk parler

Here the verb appears in Agr1° and licenses the null subject, the pro in SpecAgr1' , under agreement. The verb moves from Agr2° to Agr1° to license the subject posi-

CLAUSE STRUCTURE AND X - SECOND

153

tion in SpecAgr1'. So SpecAgr1' is a subject position. As we saw in sections 1.2 and 3, SpecAgr2' is also a subject position. Putting together what we have just seen with the results of section 1.2, our analysis of OF runs as follows. There may be a very early period, essentially represented by the Quatre Livres du Roi, in which OF is like Icelandic in that SpecAgr1' was a topic position, and embedded topicalization was possible. However, most of the texts of “conservative” OF do not give sufficient evidence for embedded topicalization, as opposed to Stylistic Fronting, to support this analysis. For conservative OF (i.e., leaving aside the system that may be instantiated by the Quatre Livres du Roi), we propose that SpecAgr1' was a subject position. So Nominative Case could be assigned to this position under spec-head agreement. On this view, “conservative” OF is just like German with respect to the nature of both SpecAgr1' and SpecAgr2'; the differences between the two languages are due to the independent existence of both Stylistic Fronting and referential null subjects in OF. In main clauses at this period, the well-known connection between verb-second and null subjects is retained since V is in C° and null subjects must be in SpecAgr1'. Third, there is the “innovative” system of thirteenth-century prose. Here, only expletive null subjects are allowed in embedded contexts. The obvious analysis is that Agr1° is no longer able to license a referential null subject. We think this is because Agr1° is not present at all at this stage. This leaves Agr2° and C° as potential licensers of null subjects. Of these, Agr2° is unable to license referential null subjects alone (i.e., without moving to C°). As a result, C° is the only potential licenser in embedded clauses, and so only expletive null subjects are available in these OF texts. Agr(2)° (i.e., the sole Agr after the thirteenth century) is unable to license a null subject under agreement in embedded clauses, but it continues to license null subjects under government when moved to C° in main clauses, giving rise to the clear asymmetries between main and embedded clauses found in thirteenth-century prose. We suggest that Agr(2)° is unable to license pro under agreement as the inflectional morphology it contains is too “poor”; for licensing under agreement genuinely rich morphology is required, as, for example, in Italian. French verbal morphology has been relatively impoverished since roughly the twelfh century (see Foulet 1935–1936, Roberts 1993b for discussion). This difference in the licensing properties of Agr(2)° may be related to the suggestion made in section 3 that relations that hold under spechead agreement depend on the content of the head, while relations that hold under government are “pure” parametric choices. Note that this idea is consistent with the proposal—which we have seen several times already—that spec-head coindexation does not suffice for licensing null subjects. Finally, we need to say something about why and how Agr1P was lost. Two, possibly related, factors are relevant to this question. First, there appears to be a relation between the loss of Agr1°, instantiated as the loss of embedded referential null subjects, and the loss of Tobler-Mussafia effects (see section 2.2). De Kok (1985, 93) gives early examples of clitic-first orders in yes/no questions in exactly the thirteenth-century prose texts that Hirschbuhler (1990) uses to establish the development of the “advanced” system of OF null subjects. This diachronic correlation is in need of further investigation and documentation, but, to the extent that it holds up, it can be accounted for straightforwardly in our terms as the loss of Agr1P.

154

IP FUNCTIONAL STRUCTURE

A second development that may be relevant is the loss of the morphological case system. It is well known that OF had a morphological case system that distinguished nominative from non-nominative in NPs headed by (most) masculine nouns. This system was lost between the twelfth and the fourteenth centuries (Foulet 1982, 32–33). Now, as we mentioned in section 3, the presence of Agr1P makes Nominative-assignment under government possible independently of movement to C° and, hence, allows inversion to be a non-root phenomenon. Suppose, as seems intuitively reasonable, that non-root Nominative-assignment under government is linked to generalized nominative morphology (i.e., nominative morphology not restricted to the pronominal system). One way to think of this is by taking Agr1P to be really NomP, a projection of a head whose sole function is to assign Case. Then, the loss of the morphological case system triggers the loss of Agr1P/NomP. This proposal has the virtue of tying the loss of Agr1P/NomP to a very well-known development in the history of French. The Germanic languages offer other examples of a possible relation between a morphological case system and Agr1P/NomP: German and Icelandic both have both, while English and the Mainland Scandinavian languages have neither. (Dutch is unusual in that it has very little morphological Case and Agr1P/NomP.) In conclusion, the preceding analysis of OF accomplishes two things: first, it allows us to retain the idea that verb-movement, either to C° or to Agr1°, was intimately connected with the licensing of referential null subjects in this language; second, we can account for a range of the word orders attested at different periods of OF (although up to now we have given no analysis of Stylistic Fronting; see the appendix). A further point is that our approach allows us to explain the restrictions on null subjects that developed during the OF period and to relate them to changes in wordorder possibilities, to changes in clitic positions, and, possibly, to the loss of morphological case.

5.

Conclusion

In this essay, we have proposed that some languages have a more complex clause structure than has usually been supposed, in that there are two Agr-projections available. These two projections locate properties of Agr° in different positions: in particular, clitics can be thought of as occupying a position which is distinct from that of the finite verbal agreement morphology. By means of this simple assumption, we have been able to give a straightforward account of a range of phenomena involving the interaction of verb-movement, clitic-placement, and Nominative-Case assignment. On the empirical level, it emerges from our analysis that certain phenomena are characteristic of the “double-Agr” system. These are (i) generalized embedded topicalization; (ii) Tobler-Mussafia effects; (iii) phenomena involving second-position clitics more generally (e.g., the V3 orders of OE and OHG, as well as the “Wackernagel effects” of Modern German). If the suggestion made at the end of the previous section that the existence of Agr1P is closely connected to the existence of generalized morphological nominative case is correct, then we make the further prediction that properties (i–iii) correlate with the presence of morphological nominative case.

CLAUSE STRUCTURE AND X - SECOND

155

On the theoretical level, we have developed a theory of subject positions. Taking as our starting point the proposals in Koopman and Sportiche (1991), we have suggested that the inventory of subject positions in a given language is a function of the possible modes of Nominative-assignment. We elaborated Koopman and Sportiche’s proposal that Nominative Case can be assigned under either government or spec-head agreement. First, we suggested that spec-head coindexation, although a subcase of spechead agreement, should be distinguished from spec-head agreement. Spec-head coindexation may allow Nominative-assignment to positions that are otherwise A'positions (see Rizzi 1991). Also, this idea has consequences for the theory of null subjects, since spec-head coindexation never formally licenses a null subject. Second, we proposed that Nominative-assignment under agreement is not a “pure” parametric choice but, instead, is determined by the content of the Nominative-assigning head (which may be subject to parametric variation). In contrast, the possibility of Nominative-assignment under government appears to be a parametric choice. Of course, various questions remain open. The most pressing of these concerns the factors that determine the movements of the inflected verb to the various functional-head positions. In terms of the distinction between L-related and non-L-related heads (see Chomsky 1995), since both C° and Agr1° would be non-L-related heads, while Agr2° would be L-related, the question becomes: What determines verbmovement to non-L-related heads? We hope that certain approaches to answering this question have emerged in the foregoing. Appendix: Stylistic Fronting and embedded topicalization Throughout this essay, and especially in section 1, we have kept Stylistic Fronting distinct from embedded topicalization. However, Thráinsson (1986) proposes that Stylistic Fronting is to be identified with embedded topicalization; he suggests that the “subject-gap” requirement for Stylistic Fronting is illusory and that Stylistic Fronting is simply topicalization of a non-subject where the subject is unavailable. In contrast, Vikner (1995, 117) gives two reasons to keep the two constructions separate. First, Stylistic Fronting typically affects non-nominal elements (participles, adverbs, and adjectives), while topicalization preferentially applies to NPs and PPs (see Maling 1990). Note that, for participles at least, it is not even clear that the moved category is a maximal projection. Second, topicalization from clauses containing an embedded topicalization is impossible in Icelandic, while topicalization from clauses where Stylistic Fronting has taken place is fine (examples from Vikner ibid.): (A1) a. *Maríu veit ég að þessum hring lofaði Olafur. Mary-DAT know I that this ring-ACC promised Olaf-NOM ‘I know that Olaf promised this ring to Mary.’ b. Þessi maður held ég að tekið hafí út peninga úr bankanum. this man think I that taken has out money from bank-the ‘I think that this man has taken money out of the bank.’

Furthermore, our discussion of OF in section 1.2 and of Yiddish in section 1.3 suggests that the two processes are distinct, since these languages seem to allow Stylistic Fronting in cases where they do not allow embedded topicalization.

156

IP FUNCTIONAL STRUCTURE

Nevertheless, the two operations are very similar to each other, in both their effects and their cross-linguistic distribution, and so it is worth briefly exploring a way of relating them to each other. In this way, too, we can provide a concrete analysis of Stylistic Fronting, something we did not attempt in the main part of the article. Let us begin by considering again a minimal contrast between topicalization and Stylistic Fronting from Yiddish (Santorini, personal communication): (A2) a. *Ikh veys nit vu nekhtn iz di ku geshtanen. I know not where yesterday is the cow stood ‘I don’t know where the cow has stood yesterday.’ b. Ikh veys nit vu nekhtn iz geshtanen a ku. I know not where yesterday is stood a cow

Example (A2a) has the same structure as parallel cases of embedded topicalization in Icelandic: nekhtn is in specAgr1' and di ku is in specAgr2', while the inflected verb is in Agr1°. As we saw in section 1.3, this structure is ruled out in Yiddish for reasons connected to constraints on extraction from clauses involving topicalization. What is the structure of (A2b)? The grammaticality of this kind of sentence depends on the presence of a subject gap, as is well known. The thematic subject is indefinite and appears to occupy someVP-internal position. We propose to explain this well-known requirement that Stylistic Fronting involves a subject gap in the following way: Stylistic Fronting is topicalization via SpecAgr2'. In (A2b), nekhtn moves to specAgr1' via specAgr2'. Hence, specAgr2' must not contain the subject in order for Stylistic Fronting to be possible, since it must contain the trace of the fronted element (for a very similar idea, see Vikner 1995, 118). Where Stylistic Fronting takes place, specAgr2' is nevertheless a subject position in the sense of the discussion in section 3. This means that the movement is a case of fronting of a non-nominal to an A-position. This makes it possible for the trace in this position to be Case-marked (although the Case-marking is not obligatory; hence, where some NP undergoes Stylistic Fronting, there is no Case clash). In languages in which specAgr1' is a topic position (e.g., Icelandic and Yiddish), the further movement from specAgr2' to specAgr1' is formally an instance of topicalization—that is, A'-movement. In languages in which specAgr1' is not a topic position (e.g., Old French), this movement could be a further instance of “Amovement” of a non-nominal. The “A” character of this movement means that (a) in OF, the fronted element can be moved to a non-topic position and (b) in Yiddish, the “A”-movement to specAgr2' is not subject to the constraints that apply to Wh-movement. Where the precondition on Stylistic Fronting is satisfied by extraction of the subject, what we have said implies that the subject trace is not in specAgr2' but is in a lower position, possibly the “internal” subject position in VP. This analysis is analogous to the analysis of postverbal subject-extraction often proposed for Italian (see Rizzi 1982, ch. 4). Further support comes from the fact that postverbal subjects are found in the languages that have Stylistic Fronting. This is illustrated in the following examples from Icelandic (Platzack 1987, 378), Yiddish (Santorini 1989, 101), and Old French (Dupuis 1989, 163), respectively:

CLAUSE STRUCTURE AND X - SECOND

157

(A3) a. Það munu kaupa þessa bók margir stúdentar. it will buy this book many students ‘Many students will buy this book.’ b. Es hot ongerufen mayner a khaver. it has on-called my-NOM a friend ‘A friend of mine called.’ c. . . . que li soudans de Coine oï dire que si faitement avoient fait li Francois. that the Sultan of Coine heard tell that so in this way had done the French ‘. . . that the Sultan of Coine heard tell that the French had done in this way.’

Postscript The ideas presented in this chapter were first developed in 1989–1990. The material was submitted as an abstract to the GLOW Colloquium held at Cambridge University in 1990, and was accepted as an alternate paper. The essay was completed in its present form in 1990–1991, with a view to submission to the proceedings of the 1990 GLOW Colloquium. However, this volume failed to materialize, and so the paper, despite becoming somewhat well known, has never been published until now. The reason for publishing this essay now, despite the fact that it uses assumptions and technology that are at least ten years old, is that it was one of the first attempts at exploring the “structural cartography” of the clause at the CP-level. In fact, one of the goals of the essay was to extend the results of Pollock’s (1989) “split-Infl” analysis of English and French clause structure to the structural level immediately above the traditional IP. In this respect, then, we are very pleased that the paper has been accepted for publication in the first volume devoted to the structural cartography of the clause. The essay deals with four principal empirical areas, corresponding to each of the four main sections: (embedded) verb-second; clitic-second; subject positions; and, finally, null subjects. In this postscript, we briefly indicate some of the developments in each of these areas since this essay was written. First, however, we should say something about its basic idea: that there exists an Agreement projection occupying a structural position in between the traditional C and the traditional I, and that the head of this projection has the typical properties of such elements (it can attract the inflected verb, clitics and subjects; and it can assign Nominative Case). This proposal combines two strands that have emerged as distinct trends since the essay was written; we can call these the “split-Agr” idea and the “split-C” idea. Let us briefly look at these, in turn. Numerous proposals for splitting Agr into categories such as Person, Number, and Gender began to emerge very quickly after Pollock’s initial split-Infl proposal. Shlonsky (1989) proposed a split Agr on the basis of Hebrew data. Most recently, the idea that “subject agreement” may correspond to up to four separate projections has been defended in great detail on the basis of data from Italian dialects by Poletto (2000) and Manzini and Savoia (2000). A distinct proposal was made in Chomsky (1991), cited in the early paper as Chomsky (1989) and later reprinted as chapter 2 of Chomsky (1995): that subject-agreement and object-agreement should be distinguished. Despite the conceptual argument made against the postulation of Agree-

158

IP FUNCTIONAL STRUCTURE

ment projections in Chomsky (1995, 4.10) there seem to be strong empirical reasons to postulate such positions in many languages, at least as hosts for clitics, as the work by Poletto and by Manzini and Savoia shows. The idea of a “split-Comp” of one kind or another was proposed as early as Chomsky (1977) but has received greater impetus in recent theoretical work. Shlonsky (1992, 1994) proposed an Agr-C position with many properties similar to the Agr1 position argued for here; Shlonsky exploits this position, in particular, to account for the intricate distribution of clitic and non-clitic pronouns in West Flemish (see also Haegeman 1995). The most influential recent proposal is Rizzi (1997). According to this proposal, the traditional C is split into Force and Finiteness, the former being the category that interfaces with discourse in matrix clauses and with a selecting predicate in embedded clauses, and the latter being the category that interfaces with the Isystem. It is thus natural to think that Force is the structurally highest C-projection and that Fin is the structurally lowest, because it is the category that selects IP. Rizzi provides evidence that intervening between Force and Fin are a Focus category and a potentially unlimited number of Topic Phrases. Finally, Rizzi also proposes that interspersed among the Focus and Topic Phrases there may be Agreement Phrases of various kinds. It is clear, then, that Rizzi’s system shares many features with the proposals made here, although it is considerably richer and is largely aimed at a different empirical domain (the nature of “left-periphery” elements in Italian and English and their interactions with subject extraction). Tentatively, one can identify Rizzi’s Fin position with our Agr1; note that ascribing the ability to assign Nominative Case to this position creates a symmetry with the Case-properties of T in the I-system. Turning now to the topics of the specific sections, a prevalent view of the nature of embedded verb-second of the type found in Yiddish and Icelandic has been that this construction involves V2 at the IP-level, with SpecIP acting as the position able to host any fronted XP (i.e., as an A'-position). This idea was proposed by both Diesing and Santorini for Yiddish (see section 1.3 and the references given there). This kind of analysis has also been widely assumed in discussions of “transitive expletive” constructions in Icelandic (see Bobaljik 1995; Chomsky 1995, ch. 4; Bobaljik and Jonas 1996, etc.), although the considerations raised here in section 4 suggest that this assumption may not be warranted. More generally, the analysis proposed here makes it possible to take the traditional SpecIP to be always and only an A-position (in more recent technical terms, a position to which DPs bearing agreement features are attracted). Clitic-second effects have been often treated as phonological phenomena. Many of the contributions in Halpern and Zwicky (1995) explicitly propose this, often appealing to a PF repair strategy such as Prosodic Inversion to place clitics, which otherwise would be illicitly in first position at PF, in the required second position. The most thorough and convincing rebuttal of this approach to second-position clitics in Slavic is Starke (1993). Madeira (1993) proposed an analysis of enclisis in European Portuguese in terms that are largely compatible with the proposals made here. The approach advocated here makes appeal to phonological processes in secondposition clitic placement, as in other cases of clitic-placement, unnecessary. During the 1990s a great deal of important work was done on the typology of weak and clitic pronouns, considerably improving our understanding of this area. The most impor-

CLAUSE STRUCTURE AND X - SECOND

159

tant publications in this connection are Cardinaletti and Starke (1996, 1999); see also the chapters in van Riemsdijk (1999). It is natural that the proposal to split IP into various component projections should have led to the postulation of various subject positions. Cardinaletti (1997) pursues the ideas put forward here, arguing in detail that nonpronominal subjects in Italian and a number of other languages appear in a higher position than do various pronominal and null subjects. McCloskey (1996) argues that the subject position in Irish, a VSO language, is lower than the Tense position. Haider (1995), following Grewendorf (1989), proposes that German has no designated subject position; subjects are able to remain in their VP-internal position. This possibility is not considered here and would have a number of implications for the analyses of German that are proposed. Regarding null subjects, there has been a tendency with the advent of the minimalist program to reconsider the status of empty categories. Chomsky’s (1993, 1995) proposal that traces are actually copies of the moved material implies that the earlier traces have no status in the current theory. This, in turn, casts doubt on the existence of the pronominal empty categories PRO and pro. Two important recent proposals for theories of control that do away with PRO are Hornstein (1999) and Manzini and Roussou (2000). Regarding pro, Borer (1986) proposed that this element may be eliminated in favor of taking Infl itself to act as a subject. This idea has been developed recently by Alexiadou and Anagnostopoulou (1998) and Manzini and Savoia (2000). Precisely how the proposal made here would fare in such a system is matter for future research. Notes We would like to thank Paola Beninca’, Guglielmo Cinque, Cecilia Poletto, Luigi Rizzi, Sten Vikner, and an anonymous reviewer for their comments; Beatrice Santorini for help with Yiddish data; and Hoski Thráinsson for help with Icelandic data. Versions of this material were presented at the Incontro di Grammatica Generativa in Pisa, 1990; at the Seminario di Ricerca in Venice, 1990; and at a Workshop on Comparative Syntax in Venice, 1990. Thanks to the audiences for their comments. For the specific concerns of the Italian Academy, Anna Cardinaletti is responsible for 1.1, 1.3, 2.1, 3, 4.1, Appendix, and Ian Roberts is responsible for 1.2, 2.2, 2.3, 4.2, Postscript. The references have been updated to incorporate later published works of the original papers and dissertations and to include references to authors in the postscript. 1. We use “verb-second” (V2) as a pretheoretical descriptive term, without prejudicing how to analyze it theoretically. 2. Although, as our data show, there is no restriction on the class of clauses which allows verb-second in Icelandic, there are some independent restrictions on the interaction of Wh-movement and topicalization. It seems that an adverbial Wh-constituent cannot introduce a verb-second clause where the topicalized element is also adverbial, as the following contrasts illustrate: (i) Ég spurði hvar henni hefðu flestir aðdáendur gefið blóm. I asked where her had most fans given flowers ‘I asked where the most fans had given her flowers.’ (Thráinsson 1986, 186; cited in Santorini 1989, 67)

160

IP FUNCTIONAL STRUCTURE

(ii) bókin sem þegar hefur María lesið (cf. (1e)) book-the that already had M. read ‘the book that Mary had already read’ (iii) *Ég veit ekki hvar í gær hefur kýrin staðið. (Vikner 1995, 74) I know not where yesterday has the-cow stood Example (i) has an adverbial Wh-element and an argumental topic; (ii) shows the opposite pattern. Both examples are grammatical as cases of embedded topicalization. Example (iii), by contrast, involves an adverbial Wh-element and an adverbial topic and is ungrammatical. Vikner (ibid.) proposes that the impossibility of (iii) should be attributed to failure of antecedent government. However, more needs to be said (at least given standard accounts of antecedent-government) to distinguish (iii) from (i) and (ii). 3. In German, the complementizer must be absent from the embedded clause for V2 to be possible; if daß is included in (2a), the sentence becomes ungrammatical. This is not true in Mainland Scandinavian, where the equivalent of daß is required (see Vikner 1995, 84). Example (2d) is improved if the verb is subjunctive, but not to the point of full grammaticality: (i) ??die Tatsache, gestern habe Maria dieses Buch gelesen the fact yesterday havesubj M. this book read However, deverbal nouns seem to accept a V2 complement of this kind more readily: (ii) die Behauptung, er wäre in Frankfurt the statement he weresubj in Frankfurt ‘the statement [that] he was in Frankfurt’ (iii) die Behauptung, gestern wäre er in Frankfurt the statement yesterday weresubj he in Frankfurt ‘the statement [that] yesterday he was in Frankfurt’ 4. Cecilia Poletto (personal communication) raises the question of double topicalization. Why do we not find structures with both CP-recursion and Agr-recursion, which would give rise to orders of the following kind (where the inflected verb is either in the second C° or in Agr1°): (i) C° [ TOP [ C°/V [ TOP [ Agr1°/V [ SUBJECT . . . ] ] ] ] ] This structure would be manifested by double topicalization. In general, double topicalization is not possible, due to relativized minimality, since the lower topic prevents the higher one from antecedent-governing its trace. 5. An apparent problem, pointed out to us by Guglielmo Cinque (personal communication), concerns the possibility of generating VOS orders in V1 declaratives. If, as we are proposing, the verb is in C° in such sentences, what prevents the object from moving to SpecAgr1'? This would give rise to VOS order, something we do not find (with a definite subject) in Icelandic. Notice that the problem is not limited to V1 declaratives, but, to the extent that they involve V-to-C movement, holds also of interrogatives, imperatives, and similar clause types in languages with Agr-recursion. A solution comes from the adjacency condition on Caseassignment; McCloskey (1991) shows that in a systematically V1 language like Irish, the verb must be adjacent to the subject for Nominative-assignment to take place and suggests that this is true wherever Nominative-assignment takes place under government. Adopting McCloskey’s proposal, we suggest that when the verb moves to C° the subject must move to SpecAgr1', blocking movement of the object to this position. 6. We have used Le Charroi de Nîmes as a primary source because it is a representative text from the relevant period (twelfth century) of Old French.

CLAUSE STRUCTURE AND X - SECOND

161

7. The idea that Agr1° can be a clitic position leads naturally to an extension of our system to account for the “reduplication of agreement” found in many Gallo-Italian dialects with rich systems of subject clitics. See Roberts (1993a) for a version of this idea. Also, our system may provide a straightforward account of inflected complementizers of the type found in West Flemish (see Haegeman 1992) and Bavarian (see Bayer 1984–1985). These phenomena are illustrated in (i): (i)

a. (La Maria) la parla. (Northern Italian dialects) (the M.) she talks ‘Mary talks.’ b. . . . da-n-k goan. (West Flemish) that-1sg-I go-1sg ‘. . . that I go.’ c. . . . da-ø-me goan. (West Flemish) that-1pl-we go-1pl ‘. . . that we go.’ d. . . . wenn-st kummst. (Bavarian) when-2sg come-2sg ‘. . . when you come.’

8. In German, V-to-C movement can bypass the Agr1 head. In principle (i.e., according to general conditions on locality), head-movement to C can skip Agr. In the languages where it cannot—for example, Icelandic—we take Agr1 to act as an attractor for V. 9. In fact, the situation is more complicated than (29) indicates in that sentences like (i) are possible on condition that at least the second pronoun is stressed: (i) ?. . . daß es/sie der Hans ihm wahrscheinlich gegeben hat. that it/her the H. him-DAT probably given has We interpret this fact not as a counterexample to the analysis being put forward in the text, but as an example of scrambling of a (stressed) pronoun. When pronouns occupy some position other than the Wackernagel position, they have undergone scrambling in the same way as full NPs. 10. Another important confirmation comes from Penner’s (1990) work on the acquisition of Bernese Swiss German. Penner shows that acquisition of this dialect proceeds in phases that can be neatly accounted for by assuming that Agr2P is acquired before Agr1P and Agr1P before CP. 11. This connection between matrix V1 and embedded topicalization was noticed by Santorini (1989, 64). According to her, the languages that have both properties include Icelandic, Yiddish, and Old French. We capture this correlation in our system by saying that such languages allow matrix clauses to be Agr1Ps, with “V°-topicalization” to C° giving matrix V1. This is, in fact, what we said for Icelandic in section 1.1. The relative infrequency of matrix V1 in OF as compared to OIt suggests that this is not the right approach for OF, at least not for the “core” period where matrix declaratives are CPs (see sections 1.2 and 3). It is possible that this analysis will turn out to be correct for the early period of OF where embedded topicalization is possible (see section 1.2). 12. This proposal appears to lead to a problem in German. We have been operating under the assumption that Agr1° is the head that assigns Nominative. However, the analysis in section 2.1 clearly implies that V moves to C°, “skipping” Agr1°. So, the Nominative-assigning head does not appear to be moved to C°. To deal with this, we tentatively suggest that C° in German independently has the capacity to assign Nominative when it

162

IP FUNCTIONAL STRUCTURE

contains a verb that agrees with its specifier. Evidence in favor of this comes from es-clauses with a definite subject, for example: (i)

?Esi hat ti dieser Mann angerufen. it has this man called ‘This man has called.’

It is hard to avoid the conclusion that there are two Nominative NPs here: es and dieser Mann. In terms of our analysis, C° assigns Nominative to es in SpecC' under agreement, and Agr1° assigns an independent Nominative feature to dieser Mann in SpecAgr2' under government. SpecAgr1' is occupied by the trace of es, see Cardinaletti (1990a, 1994). This proposal captures the frequently made observation that the definiteness effect is less rigid in German than in other languages. 13. It is possible that the distinction that Rizzi and Roberts (1989) draw between morphologically selected and free head-to-head movement may be relevant here. Since Rizzi and Roberts use the notion of morphological selection only for cases where affixes attract stems, rather than for cases of cliticization, their system should be modified to be able to apply to cases where clitics simply impose a requirement of attachment to words of a certain class. As we saw in section 2.2, the best treatment of the Tobler-Mussafia effect involves saying that the clitic must combine with the inflected verb, but we cannot state whether this requirement is one of enclisis or proclisis. In terms of Kayne’s (1991) proposal that enclisis involves a configuration where the clitic is structurally lower than the element it cliticizes to, it is not clear how the mechanism of morphological selection can be adapted since selection always requires that the selecting element govern the selected element. 14. As in Cardinaletti (1994), we are assuming that es and það are essentially like the expletives of non-V2 languages. They have no special affinity with the sentence-initial position (SpecC' and SpecAgr1', respectively), and they are not “fillers” of this (or any other) position. 15. According to the Partitive Hypothesis, an indefinite subject can receive Partitive Case in a position inside VP. If we assume that the base position of the subject is inside the projection of the main verb, the position assigned Partitive cannot be the base position of the subject, since this NP can precede an auxiliary that selects the main verb (see Vikner 1995 for further discussion).

References Adams, M. (1987a) “From Old French to the Theory of Pro-Drop.” Natural Language and Linguistic Theory 5, 1–32. Adams, M. (1987b) “Old French, Null Subjects and Verb Second Phenomena,” Ph.D. Diss., University of California at Los Angeles. Adams, M. (1988a) “Embedded pro,” in J. Blevins and J. Carter (eds.) Proceedings of Nels 18, 1–21. Amherst: University of Massachusetts. Adams, M. (1988b) “Les effets V2 en ancien et en moyen français.” Revue québécoise de linguistique théorique et appliquée 7 (special issue on Aspects de la syntaxe historique du français), 13–40. Adams, M. (1988c) “Word Order and Null Subjects: Contributions from Old French.” Unpublished ms., University of California at Los Angeles. Alberton, S. (1990) “Enclise du pronom objet en français et en italien antique ou la loi ToblerMussafia,” Mémoire de Licence, Université de Genève. Alexiadou, A., and E. Anagnostopoulou (1998) “Parametrizing AGR: Word Order, V-movement and EPP-checking.” Natural Language and Linguistic Theory 16(3), 491–539. Bayer, J. (1984–1985) “COMP in Bavarian Syntax.” Linguistic Review 3, 209–274.

CLAUSE STRUCTURE AND X - SECOND

163

Belletti, A. (1988) “Unaccusatives as Case Assigners.” Linguistic Inquiry 19, 1–34. Belletti, A. (1990) Generalized Verb Movement: Aspects of Verb Syntax. Turin: Rosenberg and Sellier. Beninca’, P. (1983–84) “Un’ipotesi sulla sintassi delle lingue romanze medievali.” Quaderni Patavini di Linguistica, 3, 3–19. Beninca’, P. (1989) “L’ordine delle parole nelle lingue romanze medievali.” Paper presented at the 19th International Congress of Linguistics and Romance Philology, Santiago de Compostela, Spain. Bobaljik, J. D. (1995) “Morphosyntax: The Syntax of Verbal Iinflection,” Ph.D. Diss., MIT, Cambridge, Mass. Bobaljik, J. D., and D. Jonas (1996) “Subject Positions and the Roles of TP.” Linguistic Inquiry 27(2), 195–236. Borer, H. (1986) “I-Subjects.” Linguistic Inquiry 17, 375–416. Boschetti, L. (1986) “Zur Syntax des Pronomens Sich,” M.A. Thesis, Università di Venezia. Cardinaletti, A. (1990a) “Es, pro and sentential arguments in German.” Linguistische Berichte 126, 135–164. Cardinaletti, A. (1990b) “Subject/Object Asymmetries in German Null Topic Constructions and the Status of SpecCP,” in J. Mascaró and M. Nespor (eds.) Grammar in Progress: GLOW Essays for Henk van Riemsdijk, 75–84. Dordrecht: Foris, Cardinaletti, A. (1994) La sintassi dei pronomi: Uno studio comparativo delle lingue germaniche e romanze. Bologna: Il Mulino. Originally cited as: (1990c) “Pronomi nulli e pleonastici nelle lingue germaniche e romanze: Saggio di sintassi comparata,” Ph.D. Thesis, Università di Padova e Venezia. Cardinaletti, A. (1997) “Subjects and Clause Structure,” in L. Haegeman (ed.) The New Comparative Syntax, 33–63. London: Longman. Cardinaletti, A., and M. Starke (1996) “Deficient Pronouns: A View from Germanic—A Study in the Unified Description of Germanic and Romance,” in H. Thráinsson, S. D. Epstein, and S. Peter (eds.), Studies in Comparative Germanic Syntax, vol. 2, 21–65. Dordrecht: Kluwer. Cardinaletti, A., and M. Starke (1999) “The Typology of Structural Deficiency: A Case Study of the Three Classes of Pronouns,” in H. van Riemsdijk (ed.) Clitics in the Languages of Europe, 145–233. Berlin: Mouton de Gruyter. Chomsky, N. (1957) Syntactic Structures. The Hague: Mouton. Chomsky, N. (1977) “On wh-movement,” in A. Akmajian, P. Culicover, and T. Wasow (eds.) Formal Syntax, 71–132. New York: Academic Press. Chomsky, N. (1981) Lectures on Government and Binding. Dordrecht: Foris. Chomsky, N. (1989) “Some Notes on Economy of Derivations and Representations,” in I. Laka and A. Mahajan (eds.) MIT Working Papers in Linguistics 10. Cambridge, Mass.: MIT Press. Reprinted in Chomsky 1995, ch. 2. Chomsky, N. (1991) “Some Notes on Economy of Derivation and Representation,” in R. Freidin (ed.) Principles and Parameters in Comparative Grammar, 417–454. Cambridge, Mass.: MIT Press. Chomsky, N. (1993) “A Minimalist Program for Linguistic Theory,” in K. Hale and S. J. Keyser (eds.) The View from Building 20, 1–52. Cambridge, Mass.: MIT Press. Chomsky, N. (1995) The Minimalist Program. Cambridge, Mass.: MIT Press. de Kok, A. (1985) La place du pronom personnel régime conjoint en français: Une étude diachronique. Amsterdam: Rodopi. den Besten, H. (1983) “On the Interaction of Root Transformations and Lexical Deletive Rules,” in W. Abraham (ed.) On the Formal Syntax of the Westgermania, 47–131. Amsterdam: John Benjamins.

164

IP FUNCTIONAL STRUCTURE

Diesing, M. (1988) “Word Order and the Subject Position in Yiddish,” in J. Blevins and J. Carter (eds.) Proceedings of NELS 18, 124–140. Amherst: University of Massachusetts. Diesing, M. (1990) “Verb Second in Yiddish and the Nature of the Subject Position.” Natural Language and Linguistic Theory 8, 41–80. Dupuis, F. (1988) “Pro-drop dans les subordonnées en ancien français.” Revue québécoise de linguistique théorique et appliquée 7 (special issue on Aspects de la syntaxe historique du français), 41–62. Dupuis, F. (1989) “L’expression du sujet dans les subordonnées en ancien français,” Ph.D. Diss., Université de Montréal. Foulet, L. (1935–1936) “L’extension de la forme oblique du pronom personnel en ancien français,” Romania 61, 257–315, 401–463; Romania 62, 27–91. Foulet, L. (1982) Petite syntaxe de l’ancien français (3rd ed.). Paris: Editions Champion. Galves, C. (1991) “V-Movement, Levels of Representation and the Structure of S.” Unpublished ms., Universidade Estadual de Campinas, Campinas, Brazil. Giusti, G. (1986) “On the Lack of Wh-Infinitivals with zu and the Projection of COMP in German,” Groninger Arbeiten zur germanistischen Linguistik 28, 115–169. Grewendorf, G. (1989) Ergativity in German. Dordrecht: Foris. Haegeman, L. (1992) Generative Syntax: Theory and Description—A Case Study from West Flemish. Cambridge: Cambridge University Press. Haegeman, L. (1995) Theory and Description in Generative Syntax: A Case Study in West Flemish. Cambridge: Cambridge University Press. Haider, H. (1995) Derivations and Representations in German (Arbeitspapiere des Sonderforschungsbereichs 340). Stuttgart/Tübingen: University of Stuttgart/ Tübingen. Halpern, A., and A. Zwicky (1995) Approaching Second. Stanford, Calif.: Center for the Study of Language and Information. Harris, M. (1978) The Evolution of French Syntax: A Comparative Approach. London: Longman. Hirschbuhler, P. (1990) “La légitimation de la construction V1 à sujet nul dans la prose et le vers en ancien français.” Revue québécoise de linguistique 19, 32–55. Hirschbuhler, P., and M.-O. Junker (1988) “Remarques sur les sujets nuls en subordonnées en ancien et en moyen français.” Revue québécoise de linguistique théorique et appliquée 7 (special issue on Aspects de la syntaxe historique du français), 63–84. Hornstein, N. (1999) “Movement and Control.” Linguistic Inquiry 30(1), 69–96. Jaspers, D. (1989) “A Head Position for Dutch Clitics or: Wilma, Wim and Wackernagel,” in D. Jaspers, W. Klooster, and Y Putseys (eds.) Sentential Complementation and the Lexicon, 241–253. Dordrecht: Foris. Kayne, R. (1991) “Romance Clitics, Verb Movement and PRO.” Linguistic Inquiry 22, 647– 686. Kayne, R., and J.-Y. Pollock (1978) “Stylistic Inversion, Successive Cyclicity and Move-NP in French.” Linguistic Inquiry 9, 595–621. Kiparsky, P. (1994) ‘Origins of Germanic Syntax,” in A. Battye and I. Roberts (eds.) Clause Structure and Language Change, 140–170. New York: Oxford University Press. Koopman, H., and D. Sportiche (1991) “The Position of Subjects.” Lingua 85, 211–258. Le Charroi de Nímes (1972) Ed. J. L. Perrier. Paris: Champion. Lowenstamm, J. (1977) “Relative Clauses in Yiddish: A Case for Movement.” Linguistic Analysis 3, 197–216. Madeira, A. (1993) “Clitic-second in European Portuguese.” Probus 5, 155–174. Maling, J. (1990) “Inversion in Embedded Clauses in Modern Icelandic,” in J. Maling and A. Zaenen (eds.) Modern Icelandic Syntax: Syntax and Semantics, vol. 24, 71–91. New York: Academic Press. Originally cited as: (1980) Islenskt mál og almenn málfræði 2, 175–193.

CLAUSE STRUCTURE AND X - SECOND

165

Manzini, M. R., and A. Roussou (2000) “A Minimalist Theory of A-Movement and Control.” Lingua 110, 409–447. Manzini, M. R., and L. Savoia (2000) “I dialetti italiani.” Unpublished ms., University of Florence. McCloskey, J. (1991) “Clause Structure, Ellipsis and Proper Government in Irish.” Lingua 85, 259–302. McCloskey, J. (1996) “On the Scope of Verb Movement in Irish.” Natural Language and Linguistic Theory 14(1), 47–104. Motapanyane, V. (1991) “Theoretical Implications of Complementation in Rumanian,” Ph.D. Diss., Université de Genève. Mussafia, A. (1983) Scritti di filologia e linguistica. Ed. A. Daniele and L. Renzi. Padua: Antenore. Penner, Z. (1990) “The Acquisition of the Syntax of Bernese Swiss German: The Role of Propositional Elements in Restructuring Early Grammar.” Paper presented at the Fifteenth Annual Boston University Conference on Language Development, October. Linguistic Institute, University of Berne. Platzack, C. (1987) “The Scandinavian Languages and the Null Subject Parameter.” Natural Language and Linguistic Theory 5, 377–401. Platzack, C. (1988) “The Emergence of a Word-Order Difference in Scandinavian Subordinate Clauses,” McGill Working Papers in Linguistics: Special Issue on Comparative Germanic Syntax, 215–238. Platzack, C. (1994) “The Loss of Verb Second in English and French,” in A. Battye and I. Roberts (eds.) Clause Structure and Language Change, 200–226. New York: Oxford University Press. Poletto, C. (2000) The Higher Functional Field: Evidence from the Northern Italian Dialects. New York: Oxford University Press. Pollock, J.-Y. (1986) “Sur la syntaxe de en et le paramètre du sujet nul,” in M. Ronat and D. Couquaux (eds.) La Grammaire Modulaire, 211–246. Paris: Editions de Minuit. Pollock, J.-Y. (1989) “Verb Movement, UG and the Structure of IP.” Linguistic Inquiry 20, 365–424. Price, G. (1971) The French Language: Present and Past. London: Edward Arnold. Rizzi, L. (1982) Issues in Italian Syntax. Dordrecht: Foris. Rizzi, L. (1986) “Null Objects in Italian and the Theory of pro.” Linguistic Inquiry 17, 501–557. Rizzi, L. (1990a) Relativized Minimality. Cambridge, Mass.: MIT Press. Rizzi, L. (1990b) “Speculations on Verb Second,” in J. Mascaró and M. Nespor (eds.) Grammar in Progress: GLOW Essays for Henk van Riemsdijk, 375–385. Dordrecht: Foris. Rizzi, L. (1991) “Proper Head Government and the Definition of A-Positions.” GLOW Newsletter 26, 46–47. Rizzi, L. (1996) “Residual Verb Second and the Wh Criterion,” in A. Belletti and L. Rizzi (eds.) Parameters and Functional Heads: Essays in Comparative Syntax, 63–90. New York: Oxford University Press. Originally cited as: (1991a) Technical Reports in Formal and Computational Linguistics, No. 2. Geneva: University of Geneva. Rizzi, L. (1997) “The fine structure of the left periphery,” in L. Haegeman (ed.) Elements of Grammar, 281–337. Dordrecht: Kluwer. Rizzi, L., and I. Roberts (1996) “Complex Inversion in French,” in A. Belletti L. Rizzi (eds.) Parameters and Functional Heads: Essays in Comparative Syntax, 91–116, New York: Oxford University Press, 1996. Originally cited as: (1989) Probus 1, 1–30. Roberts, I. (1991) “Excorporation and Minimality.” Linguistic Inquiry 22(1), 209–218. Roberts, I. (1993a) “The Nature of Subject Clitics in Franco-Provençal Valdôtain,” in A. Belletti (ed.) Syntactic Theory and the Dialects of Italy, 319–353. Turin: Rosenberg and Sellier. Roberts, I. (1993b) Verbs and Diachronic Syntax. Dordrecht: Kluwer.

166

IP FUNCTIONAL STRUCTURE

Rögnvaldsson, E. (1984) “Icelandic Word Order and það-Insertion.” Working Papers in Scandinavian Syntax 8, 1–21. Rögnvaldsson, E., and H. Thráinsson (1990) “On Icelandic Word Order Once More,” in J. Maling and A. Zaenen (eds.) Modern Icelandic Syntax: Syntax and Semantics 24: 3– 40. New York: Academic Press. Santorini, B. (1988) “Against a Uniform Analysis of All Verb-Second Clauses.” Unpublished ms., University of Pennsylvania. Santorini, B. (1989) “The Generalization of the Verb-Second Constraint in the History of Yiddish,” Ph.D. Diss., University of Pennsylvania. Schulze, A. (1888) Der altfranzösische direkte Fragesatz. Leipzig: S. Hirzel Schwartz, B. D., and S. Vikner (1996) “The Verb Always Leaves IP in V2 Clauses,” in A. Belletti and L. Rizzi (eds.) Parameters and Functional Heads: Essays in Comparative Syntax, 11–62. New York: Oxford University Press. Shlonsky, U. (1989) “The Hierarchical Representation of Subject-Verb Agreement.” Unpublished ms., University of Haifa. Shlonsky, U. (1992) “The Representation of Agreement in Comp.” Geneva Generative Papers 0, 39–52. Shlonsky, U. (1994) “Agreement in Comp.” Linguistic Review 11, 351–375. Sigurðsson, H. (1985) “Subordinate V/1 in Icelandic: How to Explain a Root Phenomenon.” Working Papers in Scandinavian Syntax 18, 1–58. Sigurðsson, H. (1989) “Verbal Syntax and Case in Icelandic,” Ph.D. Diss., University of Lund. Starke, M. (1993) “En deuxième position en Europe Centrale,” Mémoire de Licence, University of Geneva. Thráinsson, H. (1986) “V1, V2, V3 in Icelandic,” in H. Haider and M. Prinzhorn (eds.) Verb Second Phenomena in the Germanic Languages, 169–194. Dordrecht: Foris. Thurneysen, R. (1892) “Die Stellung des Verbums im Altfranzösischen.” Zeitschrift für Romanische Philologie 16, 289–371. Tomaselli, A. (1990) La sintassi del verbo finito nelle lingue germaniche. Padua: Unipress. Tomaselli, A. (1991) “Cases of V3 in Old High German,” Groninger Arbeiten zur germanistischen Linguistik, 33, 93–127. Travis, L. (1984) “Parameters and Effects of Word Order Variation,” Ph.D. Diss., MIT. Vance, B. (1988) “L’évolution du pro-drop en français medieval.” Revue québécoise de linguistique théorique et appliquée 7 (special issue on Aspects de la syntaxe historique du français), 85–112. Vanelli, L., L. Renzi, and P. Beninca’ (1985) “Typologie des pronoms sujets dans les langues romanes.” Actes du XVIIe Congrès International de Linguistique et Philologie Romanes (Aix-en-Provence, 29 August–3 September 1983), vol. 3, 162–176. Aix-en-Provence: Université de Provénce. van Kemenade, A. (1987) Syntactic Case and Morphological Case in the History of English. Dordrecht: Foris. van Riemsdijk, H. (ed.) (1999) Clitics in the Languages of Europe. Berlin: Mouton de Gruyter. Vikner, S. (1995) Verb Movement and Expletive Subjects in the Germanic Languages. New York: Oxford University Press. Originally cited as: (1990) “Verb Movement and the Licensing of NP-Positions in the Germanic Languages,” Ph.D. Thesis, Université de Genève. von Wartburg, W. (1934) Evolution et structure de la langue française. Paris: Larousse. Wackernagel, J. (1892) “Über ein Gesetz der indogermanischen Wortstellung.” Indogermanische Forschungen 1, 333–435. Zwart, J.-W. (1991) “Clitics in Dutch: Evidence for the Position of INFL.” Groninger Arbeiten zur germanistischen Linguistik 33, 71–92.

6

Agreement and Tense as Distinct Syntactic Positions Evidence from Acquisition      The aim of this chapter is to contribute to the ongoing debate on the clausal architecture by presenting new relevant evidence based on language acquisition. The theoretical context is the study of the functional structure of the clause, which is a major focus of classical and current syntactic research. In the early days of the Principles and Parameters approach, much progress on the clausal structure was made possible by the assumption that clauses are headed by an inflectional node expressing morphosyntactic specifications of tense and agreement (among others); this is a principled development of an intuition dating back to Chomsky’s (1957) Syntactic Structures. The uniqueness of inflection was challenged by Pollock’s (1989) comparative analysis of verb movement, which forcefully argued for the postulation of separate functional heads for agreement and tense; by the late 1980s, the Split-Infl approach was generally adopted; in particular, it led to refined explorations of the functional structure of the sentence, analyzed as a system of inflectional heads and projections, each expressing an elementary morphosyntactic property; this trend is now fully systematized in Cinque (1999). Meanwhile, Chomsky (1995) has proposed a limited but significant step in the opposite direction from this trend, denying the status of autonomous head to the subject agreement specification and assuming that agreement features are associated to T and checked by a DP in the specifier of T. In this chapter we argue that tense and agreement features are licensed in distinct syntactic positions in English, with agreement higher than tense. As the relevant facts are found in the English spoken in the third year of life, and certain developmental properties are crucial for our argument, this chapter also reinforces the view, shared by much recent acquisition literature, that language acquisition 167

168

IP FUNCTIONAL STRUCTURES

and development can provide critical evidence bearing on central issues of syntactic theory. Throughout the essay, we characterize the position in which subject agreement features are licensed as Agr. In fact, our evidence is consistent with both the classical view that Agr is an independent functional head and the alternative view that agreement features are associated with a higher substantive functional head, say M(odal), or the like (a view distinct from Chomsky’s 1995 proposal, but consistent with his guidelines). The crucial point is that the head where subject agreement features are checked is independent from, and higher than, tense. The basic pattern to be discussed is the following. During the third year of life, learners of English typically produce negative sentences with third-person subjects and uninflected do, as in (1a). Such forms alternate for some time with the regularly inflected forms (1b) and then disappear: (1)

a. Robin don’t play with pens. (Adam 28, 3;4)1 b. So Paul doesn’t wake up. (Adam 28, 3;4)

This alternation is hardly surprising, as in early English inflected and uninflected verbal forms seem to freely alternate with lexical verbs as well: (2)

a. Robin break it # your pen. (Adam 28, 3;4) b. And the motor comes out. (Adam 28, 3;4)

So, (1a–b) seem to instantiate one of the many apparent cases of optionality involving inflectional morphology and, more generally, functional elements that child language allows. Surprisingly, the optionality in (1a–b) does not carry over to the interrogative use of do: while (1a) is robustly attested in natural production corpora, (3a) is virtually absent as a form alternating with (3b): (3)

a. (#)Do he go? b. Does dis write? (Adam 28, 3;4)

The contrast between (1) and (3) is not a trivial artifact of the later emergence of the auxiliary do in questions (for a discussion of Aux-less questions, see Stromswold 1990; Guasti and Rizzi 1996, among others): in the same period in which (1a) freely alternates with (1b), (3b) represents the overwhelming majority of the relevant questions. In this article, we claim that this asymmetry is due to the different positions that interrogative and negative do fills in the structure, respectively higher and lower than agreement (for an analogous approach to other kinds of early uninflected clauses, see also Phillips 1995, 1998). Our analysis makes crucial use of a clausal architecture involving a structural layer in which subject agreement is checked, a layer higher than TP. Thus, we claim that the observed acquisition pattern provides evidence for an articulated view of the inflectional system, as in the research trend initiated by Pollock (1989). Our evidence is fully consistent with the view that the morphologysyntax interface is maximally transparent—that is, distinct functional heads are needed to license distinct morphosyntactic features. This chapter is organized as follows: first we argue for the existence of the asymmetry (1)–(3) in Early English on the basis of a quantitative analysis of natural production corpora; next we introduce a principle applying on the morphological side

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

169

of the syntax-morphology interface, and we motivate it on the basis of a comparative analysis of subject agreement and past participle agreement; we then show that this approach naturally extends to early uninflected negative do. We then address related kinds of evidence bearing on our system, with particular reference to the behavior of negative questions and to the case properties of subjects in clauses with non-agreeing do, the early uninflected clauses, and we conclude with a discussion of an asymmetry between (negative) do and be in early systems.

1.

Methods

We have investigated the production of seven English-speaking children (from the CHILDES database; see MacWhinney and Snow 1985; MacWhinney 1991): Adam, Sarah, Eve (Brown 1973), Nina (Suppes 1974), Peter (Bloom 1970), Shem (Clark 1982), and Ross. Using the COMBO program, we have extracted all the sentences containing a form of do and have retained only those with an overt third-person subject (be it a pronoun or a nominal expression). For Peter, we have counted some questions that featured subject-auxiliary-inversion but that were not marked with a question mark in the transcripts. Thus, our counts are based on negative declaratives and on positive yes/no and wh-questions featuring subject-auxiliary-inversion. We have not included in the counts a small number of negative questions, which are analyzed separately in section 6, and some yes/no questions without inversion (see Stromswold 1995). In addition, we have eliminated so-called double-tense structures (Does it broke?) (see Stromswold 1990 and references cited there); sentences including the symbol [?] (the best guess) in front of words relevant for our analysis; and sentences with contracted auxiliaries, like it’s don’t, where the ‘s may be an inflection or more likely it’s is a variant of it (see Brown 1973). We have excluded from further analysis the production of two children—Eve and Shem—because their use of do in negative sentences was adult-like from the start. The files used in this investigation are reported in table 6-1, where we have also indicated the age of our subjects. Table 6-2 reports the first occurrence of do, does, and did in negative and interrogative sentences containing an overt subject for the five children investigated. Negative occurrences of do and morphological variants are reported in the first five

Table 6-1. Sources of data (CHILDES databases: MacWhinney and Snow, 1985) Child

Files

Adam Sarah Nina Ross Peter

01–40 01–138 01–56 20–53 01–20

Age range 2;3 2;3 1;11 2;6 1;9

3;11 5;1 3;3 4;6 3;1

170

Age (File)

NEG don’t +3 (first occurrence) NEG don’t +3 (last occurrence) NEG don’t –3 NEG doesn’t +3 NEG didn’t ±3 INT do +3 (first occurrence) INT do +3 (last occurrence) INT do –3 INT does +3 INT did ±3

Adam

Sarah

Nina

Peter

Ross

2 yrs. 4 mos. (3)

3 yrs. 0 mos. (42)

2 yrs. 1 mo. (12)

2 yrs. 6 mos. (15)

2 yrs. 8 mos. (24)

3 yrs. 5 mos. (33)

5 yrs. 0 mos. (137)

3 yrs. 2 mos. (51)

2 yrs. 9 mos. (18)

6 yrs. 3 mos. (50)

2 yrs. 5 mos. (5) 2 yrs. 8 mos. (11) 2 yrs. 11 mos. (19) 3 yrs. 4 mos. (28)

2 yrs. 3 mos. (1) 3 yrs. 2 mos. (50) 3 yrs. 2 mos. (49) 4 yrs. 5 mos. (107)

2 yrs. 1 mos. (11) 2 yrs. 1 mo. (12) 2 yrs. 4 mos. (23) None

2 yrs. 3 mos. (10) 1 yr. 9 mos. (1) 2 yrs. 4 mos. (12) 3 yrs. 1 mo. (20)

3 yrs. 0 mo. (33) 2 yrs. 6 mos. (20) 12 yrs. 9 mos. (25) 3 yrs. 7 mos. (43)

3 yrs. 10 mos. (49) 3 yrs. 6 mos. (65) 3 yrs. 3 mos. (53)

2 yrs. 3 mos. (16) 2 yrs. 2 mos. (15) 2 yrs. 0 mos. (5)

2 yrs. 3 mos. (10) 2 yrs. 4 mos. (12) 3 yrs. 1 mo. (20)

2 yrs. 6 mos. (20) 2 yrs. 8 mos. (24) 2 yrs. 7 mos. (22)

3 yrs. 5 mos. (33) 2 yrs. 9 mos. (14) 2 yrs. 9 mos. (13) 2 yrs. 6 mos. (7)

Note: do +3 and don’t +3 indicate occurrences with third person; do –3 and don’t –3 indicate all others, collapsed. See text.

IP FUNCTIONAL STRUCTURES

Table 6-2. Age at first (and last) occurrences of different forms of “do” in negative sentences and interrogatives

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

171

rows and are indicated with the shorthand NEG followed by the morphological form of do. We have distinguished do occurring with third-person subjects, notated, in table 6-2, with do +3 or don’t +3, and have collapsed all the others, notated with do –3 or don’t –3. For don’t occurring with a third-person subject, we have indicated the first and the last occurrence. In the fifth row, we report the first occurrence of didn’t with any person. The last five rows report occurrences of do in interrogatives (INT). For occurrences of bare do with third-person subjects, we have reported the first and the last occurrence. This row includes data from Adam, who produced three such occurrences; Sarah, Peter, and Ross produced only one occurrence of bare interrogative do with third-person subjects, and Nina produced none. For each child we have indicated the age and the number of the file (between parentheses) in the CHILDES database in which the relevant production was found.

2.

Results

All the five children use don’t with third-person subjects in their negative sentences. At the same time, they also employ the adult form doesn’t. For Adam and Sarah the first recorded occurrence of don’t with third-person subject precedes the first recorded occurrence of doesn’t; for Ross and Peter it is the other way around, and for Nina the two forms are produced for the first time in the same file. By contrast, this alternation is not found in interrogative sentences. Except for some rare errors, in which does is employed with non-third-person subjects, the correct form does is always used with third-person subject. These findings are summarized in table 6-3 where the distribution of do and does in interrogative and in negative sentences is displayed. This table includes only a subset of the data produced by children in the period investigated. The criteria for inclusion are the following. We started to count from the point at which don’t and doesn’t appeared at least once in each child’s negative sentences, although not necessarily in the same file; we stopped when don’t was used for the last time. The rationale for this was that we wanted to make sure that the comparison involved a period of genuine overlap between the two forms. Table 6-3. Do vs. does in negative and interrogative sentences for the five children investigated. The difference between do and does in interrogatives and negatives is significant for all children, except for Peter. Do Child (files)

Adam (11–33) Sarah (50–137) Ross (24–50) Nina (12–51) Peter (15–18) Totals

Negative

12 40 20 65 7 144

Does

Interrogatives

3 1 1 0 0 5

Negatives

Interrogatives

8 55 72 65 20 220

78 76 51 62 3 270

172

IP FUNCTIONAL STRUCTURES

In the section where we discuss the individual children, we indicate the files included in the counts. The asymmetry between negative and interrogative sentences is not a consequence of the fact that do in questions shows up somewhat later than in negative declaratives. For each child, there is a clear period of overlap during which does alternates with do in negative sentences, but not in interrogatives, as shown by the graph in figure 6-1. This graph displays for each child the periods, indicated by the three columns, during which don’t and doesn’t are used in negative sentences and does is employed in interrogative sentences. On the Y-axis we reported the age of children in months. The period of overlap is quite stable and can last several months, except for Peter. The fact that interrogative sentences are virtually error free while negative sentences display a number of errors cannot be due to chance or to the lack of knowledge of the syntax of do-support. On the contrary, the asymmetry indicates that children are sensitive to the morphosyntactic properties of do-support and of verb movement in general. 2.1. 2.1.1.

Individual subjects Adam

The first occurrence of don’t in declaratives with third-person subjects is in file 3 (only one occurrence in this file), and the second is in file 9; the phenomenon is then attested through file 33. The first occurrence of doesn’t with third-person subjects is in file 11. So, the two options overlap from file 11 to file 33 (age range 2;8 to 3;5; i.e., for about ten months). In this period, Adam produces twelve occurrences of don’t (or 57.9 percent) and eight occurrences of doesn’t (or 42.1 percent). Following is the exhaustive list of all the occurrences of don’t and doesn’t through file 33. (4) Adam don’t wear wear shoe. (Adam 3) Rin+tin+tin don’t fight me. (Adam 9) Cowboy # don’t fly. (Adam 9) Rin+tin+tin don’t fly # Mommy. (Adam 9) He don’t want some money. (Adam 19) Dis one don’t fit. (Adam 22) Because why de tape recorder don’t lie it. (Adam 25) Dis don’t have a hole in it. (Adam 26) He don’t have a bag. (Adam 28) Robin don’t play with pens. (Adam 28) Robin don’t play with dat (Adam 28) Mommy # he don’t have a baseball. (Adam 28) An(d) dis don’t have a wheel on it. (Adam 32) It don’t know how to get out. (Adam 32) Fish don’t roll. (Adam 33) Daddy don’t wear dese glasses. (Adam 33) Saggy baggy doesn’t eat a [?] all up. (Adam 11) Trailer doesn’t # fit in (th)ere. (Adam 13) Dis doesn’t fit. (Adam 23) So Paul doesn’t wake up. (Adam 28) He doesn’t + . . . (Adam 30)

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

173

Figure 6-1. Children’s uses of “do/does” in negative and of “does” in positive interrogative sentences during the period investigated

Dis doesn’t work. (Adam 32) It doesn’t + . . . (Adam 32) This doesn’t be straight. (Adam 33)

From file 36 to file 40, no occurrence of don’t is found but fifteen occurrences of doesn’t are, which suggests that the adult pattern is acquired at this point. Interest-

174

IP FUNCTIONAL STRUCTURES

ingly, in the ten months from file 11 to file 33, the two forms appear to freely alternate, even in individual files: in file 28, there are four occurrences of don’t and one occurrence of doesn’t; in file 32, there are two occurrences of don’t and two occurrences of doesn’t; in file 33, there are two occurrences of don’t and one occurrence of doesn’t. As for interrogatives, the first occurrence of the correct form does with a third-person subject is in file 13 (2;9 to 4); if we consider the period between file 13 and file 33 (age 2;9 to 3;5), Adam produces 78 occurrences of does; in the same period he produces only three occurrences of incorrect do (or 3.6 percent). In conclusion, throughout a period of ten months from file 13 through 33, Adam shows a free alternation of inflected and uninflected do in negative sentences, while his interrogative do is virtually always inflected. This sharp contrast is shown in table 6-3. In this table, we have included only the occurrences of do up to file 33—that is, up to the point where there is overlap between do and does in negative sentences and does in interrogatives. 2.1.2.

Sarah

Sarah’s first occurrence of don’t with third person in declaratives is in file 42, while doesn’t shows up in file 50. Throughout the whole period investigated up to age 5;1, Sarah uses the two forms interchangeably, from file 50 through file 126 (age range 3;2 to 4;9). In the latter files (from 127 to 137), Sarah produces five negative sentences with don’t but no instances of doesn’t. It is likely that the absence of doesn’t in these later files is due purely to chance. In contrast to what we have found in negative sentences, there is only a single instance of the nonadult do with third person in interrogatives (file 107), against seventy-six occurrences of does. The first occurrence of does appears in file 65 and shortly after the second in file 70. Up to file 99, we find twenty-three instances of does; they double from file 103 to file 138: precisely, we find fifty-five occurrences of does. Hence, there is a sharp contrast between interrogative and negative sentences with a long period of overlap from file 65 to file 137 (age range 3;6 to 5;1) in which do and does freely alternate in negatives, but in which does is the only option in interrogatives. This is shown in table 6-3. 2.1.3.

Ross

Unlike Sarah and Adam, Ross’s first relevant occurrence is doesn’t with third person in negatives (file 20), but at file 24 he also employs the nonadult form don’t. Ross oscillates between these two forms from file 24 up to file 50 (4;3). By contrast, as the other children do, Ross only uses does in interrogatives starting from file 24. A single occurrence of do at file 43 is found. For a long period starting from file 24 through file 50 (age range 2;8 to 4;3), do and does alternate in negatives, but not in interrogatives, where the only form found is does. This discrepancy is shown in table 6-3. 2.1.4.

Nina

Nina begins to use do and does with third person in negative sentences beginning at file 12. However, for a period, from file 13 till 23, she only uses do. From file 28 till 37 do and does alternate (age range 2;5 to 2;10): in this time span we find thirty-one

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

175

occurrences of do and twenty-four of does. We still find two occurrences of do in file 44 and 51, but from file 38 does is by far the more common form: there are thirtynine eight instances of does against two occurrences of do (i.e., 4 percent of errors). In interrogatives, does is used from file 15, and this is the only form used in the whole recorded period. As for the other children, there is a long period, from file 12 to file 51, during which do and does alternate in negatives but only does shows up in interrogatives. Even confining the comparison to files 28 to 37, where the alternation in negatives is more consistent, we find a sharp contrast: against thirty-one do and twenty-four does in negative sentences, we find fifteen does in interrogative sentences and this is the only type of form found, as shown in table 6-3. 2.1.5.

Peter

Despite the dearth of relevant data, Peter displays a pattern very similar to the other children. He starts to use the adult form does with third person in negatives in file 1; there are two other occurrences in file 10. The form do with third person is employed for the first time in file 15 and for a short period, it is in free alternation with the form does from file 15 to 18 (age range 2;6 to 2;9). In interrogatives, only does is used from file 12. In the period where do and does alternate in negatives (seven do and twenty does), only does is present in interrogatives, although the number of instances is low (three instances). The distribution of do and does in negative and interrogative sentences is reported in table 6-3. Notice that Peter’s grammar is almost adultlike from the start and does is used in the overwhelming majority of the cases, whatever the type of sentence (negative versus interrogative). However, we may observe that the trend is similar to that of other children. 3. A principle of the morphology-syntax interface Much work over the last ten years has shown the existence of strong form-position correlations, both across languages and in language development. One rough crosslinguistic generalization, which different notions of feature strength have tried to capture, is that syntactic movement of the relevant lexical head to a functional head seems to be favored by the richness of the morphological specification expressing the feature content of the functional head. So the verb moves to the inflectional system in the overt syntax in French or Italian but not in contemporary English. English (and Mainland Scandinavian) lost overt verb movement concomitantly with the loss of a (richer) morphological specification of agreement (Roberts 1993). This generalization will become potentially relevant later on, in connection with the properties of the acquisition of be in English. A related but independent generalization, which will be of immediate relevance here, is that the overt morphological realization of a feature seems to depend in part on whether the feature has been checked in the overt syntax. Let us first illustrate this generalization on the basis of some comparative evidence. Consider the agreement alternations that preverbal and postverbal subjects show in some languages but not in others. A rather stable generalization is the following. When the subject DP occupies a surface position in the higher parts of the inflec-

176

IP FUNCTIONAL STRUCTURES

tional system, typically higher than the inflected verb—hence, presumably in the Spec of agreement or higher (we continue to call “Agr” the functional head where agreement features are checked)—the morphological expression of agreement is compulsory (provided that the language has the relevant morphology). If the subject DP is left VP-internal or in the lower part of the inflectional system, typically lower than the inflected verb—hence, presumably lower than the agreement layer—then languages may go both ways: some express morphological agreement with the DP, others do not (whether or not the language fills the Spec of Agr with an overt expletive). (5)

a. DP Agr . . . Compulsory morphological expression of Agr b. . . . Agr . . . DP . . . Variable morphological expression of Agr

For instance, while both English and French display obligatory subject agreement with preverbal subjects, in existential and presentational constructions, a postverbal subject triggers agreement in English but not in French: (6)

a. Three girls are in the garden. b. There are three girls in the garden. c. There come three girls.

(7)

a. Trois filles sont arrivées. three girls are arrived-FEM-PL ‘Three girls arrived.’ b. Il est arrivé trois filles it is arrived-MASC-SG three girls ‘Three girls arrived.’

The different behavior of postverbal subjects is a common pattern, often differentiating closely related grammatical systems and extending to languages that use null expletives for the preverbal subject position (or, possibly, no expletive at all). For instance, among Null Subject Languages, standard Italian and some northern Italian dialects pattern with English; other dialects pattern with French. The former case is illustrated by standard Italian examples as in (8) (see Belletti 1999 on these cases of inversion); the latter is illustrated by Fiorentino (examples in (9) adapted from Brandi and Cordin 1989), by Trentino, and, among the varieties lacking subject clitics, by the variety spoken in Ancona, according to Cardinaletti (1997) (examples in (10) taken from this work). (8)

a. Le tue sorelle sono venute. (Standard Italian) the your-FEM-PL sisters are come-FEM-PL b. Sono venute le tue sorelle. are come-FEM-PL the your-FEM-PL sisters ‘Your sisters came.’

(9)

a. Gl’è venuto le tu’ sorelle (Fiorentino) it is come the your-FEM-PL sisters b. Le tu’ sorelle le son venute the your-FEM-PL sisters they are come-FEM-PL ‘Your sisters came.’ (Examples adapted from Brandi and Cordin 1989)

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

(10)

177

a. Questo, i bambini lo fanno sempre. (Anconetano) this, the children it do-3PL always b. Questo, lo fa sempre i bambini this, it always do-3SG the children

Similarly, Standard Arabic patterns with French showing agreement alternations depending on the position of the subject, but several Arabic dialects—for example, Lebanese Arabic and Moroccan Arabic—pattern with English and Italian in manifesting agreement irrespective of the position of the subject. The following examples are taken from Aoun, Benmamoun, and Sportiche (1994). (11)

a. * ʔAl-ʔawlaad-u naama. (Standard Arabic) the children slept-3MASC-SG b. ʔAl-awlaad-u naamuu. the children slept-3MASC-PL c. Naama l-ʔawlaad-u. slept-3MASC-SG the children d. *Naamuu l-ʔawlaad-u. slept-3MASC-PL the children

(12)

a. *lE-wlaad neem. (Lebanese Arabic) the children slept-3SG b. lE-wlaad neemo. the children slept-3PL c. *Neem lE-wlaad. slept-3SG the children d. Neemo lE-wlaad. slept-3PL the children

As the preverbal subject position presumably manifests a position in which the subject checks agreement features (in classical terms, Spec AgrS), while the postverbal position is lower in the tree (under Kayne’s 1994 Antisymmetry), the conclusion suggested by this pattern is the following: when agreement features are checked in the overt syntax, they are expressed in the morphology (if the language has the appropriate morphological form in the paradigm for the item involved); when agreement features are not checked in the overt syntax, but only in the syntax of LF (in terms of the model of Chomsky 1995), languages may go both ways: either express them in the verbal morphology or leave the verb uninflected (or inflected with the unmarked specification) for agreement.2 The morphological realization of features left unchecked in the overt syntax is a property that can vary between closely related systems and which, in fact, may be unstable within the same system, as is suggested by alternations like the following in (colloquial) English, some varieties of colloquial Italian, and French: (13)

a. There are / there’s many people in the garden. b. Ci sono / c’è molte persone in giardino. c. Ce sont/est des linguistes. it are/is some linguists ‘They are linguists.’

178

IP FUNCTIONAL STRUCTURES

Similarly, varieties of spoken Brazilian Portuguese that have retained the singular/ plural distinction in the verbal paradigm require agreement with preverbal subjects, not with postverbal subjects (Carlos Mioto, personal communication): (14)

a. Dois meninos chegaram/*chegou. two boys came-PL/came-SG b. Chegaram/chegou dois meninos. came-PL/came-SG two boys

In some such cases, normative grammars make decisions that may well vary from grammar to grammar: normative English, Italian, and Brazilian Portuguese demand agreement in (13a–b) and (14). Normative Standard Arabic demands lack of agreement in (11). In both cases, the existence of close varieties and registers manifesting the opposite choice is a clear symptom of instability: it seems to be the case that when a feature is not checked in the overt syntax, UG makes it possible to leave its morphological realization fluctuating. Extending the pattern from agreement to feature licensing in general, we may express the observed state of affairs through the following principle:3 (15) If a feature is checked in the overt syntax, then it is expressed in the morphology.

Principle (15) is meant to operate on the way in which the morphology reads off the output of syntax, manifesting the invariant property—obligatoriness of morphological expression—when a feature is checked in the overt syntax. The system is asymmetric in that it says nothing about the case in which a feature is left unchecked in the overt syntax and is to be checked in covert syntax, according to the model of Chomsky (1995). Whether a feature is morphologically expressed or not in this case is a property of the language-specific system of morphological rules: in the absence of UG guidance, a particular grammar may include a morphological rule requiring expression of the unchecked feature, but it does not have to— whence the variation between closely related systems, the instability, and the room for normative intervention. Notice that the system says nothing about the syntactic parameterization involved in the overt or covert checking of a given feature. For concreteness, we can continue to assume some notion of feature strength, say through one of the options explored in Chomsky (1995): for example, if a feature is syntactically strong, it involves movement of the category to ensure checking before Spell-Out; if it is syntactically weak, checking will be ensured by LF feature movement. Whatever the syntactic parameterization, principle (15) expresses an invariant element on the morphological side of the syntax-morphology interface. The application of the proposed system to the oscillations with subject agreement still needs important refinements: for instance, it does not express the fact that the possibility of verbal agreement with a VP internal subject seems to depend in part on the nature of the expletive; for instance, in French, agreement is optional with ce, as in (13c), and impossible with il, as in (7b). This may well be a nonarbitrary fact, as there seem to be cross-linguistic subgeneralizations, depending on certain characteristics of the expletive. In particular, Cardinaletti (1997) argues that expletives morphologically marked with nominative case such as French il in general do

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

179

not admit agreement with the VP internal subject, whereas expletives that do not overtly express nominative (such as French ce, German es, and English there) may admit or require agreement. Let us briefly address this point, basically following Cardinaletti’s analysis. We continue to assume, with much previous and current literature, that nominative case is assigned by Agr, not by tense. In fact, in the relatively rare cases in which agreement and tense are dissociated in a grammatical system, we can directly see that Agr, not T, is responsible for nominative: so, agreeing infinitives (+Agr, –T) in Portuguese determine nominative case marking on their subjects (Raposo 1987), while tensed but nonagreeing verbs in singular concord constructions in Belfast English (–Agr, +T; see Henry 1996) are incompatible with nominative (for relevant discussion, see also Schuetze and Wexler 1996; Schuetze 1997; Giorgi and Pianesi 1997). So, if the expletive is overtly marked as nominative, as in French, it checks nominative case and agreement in the syntax and, in compliance with (15), agreement with the expletive must be morphologically realized (and thus agreement with the associate is not possible). If the expletive is not overtly marked as nominative, syntactic agreement with the associate is possible, perhaps compulsory, but agreement and nominative checking take place only at LF (through expletive substitution, or feature movement, as in Chomsky 1995), so that the morphological realization of agreement is not required by principle (15) and may fluctuate (in closely related grammatical systems, give rise to normative intervention), as we have observed.

4. Past participle agreement in Romance Let us consider another case illustrating the system involving principle (15). Agreement between a DP and a verbal element involves two steps: Spec-head and headhead in local configurations (basically, in configurations respecting Relativized Minimality; see Rizzi 2000). An agreement feature is licensed on a functional head through a local Spec-head configuration by a DP carrying the same specification; then it is licensed by the functional head on the verbal element moved to the functional head, in a local head-head configuration. The examples considered so far illustrate how the system functions in a case of feature licensing by a specifier. As for feature licensing by a head, relevant cases illustrating the fluctuations allowed by the system are more difficult to find because of the relative rarity of alternations in head movement comparable to alternations in movement to a Spec position such as in (6) and (7) (alternations in the positions of subjects and other arguments are governed by such properties as focus, definiteness, and specificity that are not relevant for heads). Familiar cases of apparent optional movement of the untensed functional verb in English and French are not relevant because the untensed paradigm does not show any Agreement morphology that could allow us to observe a morphological alternation. Nevertheless, we believe that the operation of the system in cases of head movement can be observed, at least indirectly, by comparing closely related languages. Take past participle agreement triggered by object cliticization in Italian and French:

180 (16)

IP FUNCTIONAL STRUCTURES

a. La macchina, l’ha messa/*messo in garage. b. La voiture, il l’a mise/mis dans le garage. the car he it-FEM-SG has put-FEM-SG/put-MASC-SG into the garage ‘The car, he put it into the garage.’

In Italian, agreement is fully obligatory, while in spoken French, in spite of a strong normative pressure for the agreement option, agreement is often omitted. It is not important here to decide whether the appropriate analysis of spoken French is that agreement is optionally expressed, or that two systems coexist, one with obligatory agreement and the other with no agreement (for discussion, see Friedemann and Siloni 1997). Either way, it is fair to characterize this agreement option as oscillating. Why is Italian different? Under principle (15), it is tempting to relate this difference to an independent difference between the two systems—that is, the fact that Italian past participles move to a higher position than French past participles, as shown by their positions with respect to floated quantifiers and other adverbial material: (17)

a. Gianni ha (*tutto) capito (tutto). b. Jean a (tout) compris (*tout). ‘Gianni has (all) understood (all).’

The traditional analysis of this pattern was that French has a rule moving tout and other similar quantifiers to the left (Kayne’s 1975 “Left Tous”), while Italian lacks such a process. However, more recent work has clearly shown that Italian tutto also moves to the left from the object position in (17a). For instance, it can naturally precede unstressed bene (well), a word order that is marginal for an object DP; moreover, although in a marked order tutto can precede particles like via (away), which don’t tolerate a preceding direct object: (18)

(19)

a. Gianni ha fatto tutto bene. ‘Gianni has done everything well.’ b. Gianni ha fatto bene il lavoro / ?il lavoro bene ‘Gianni has done well the work / the work well.’ a. Gianni ha messo tutto via. ‘Gianni has put everything away.’ b. Gianni ha messo via il lavoro / *il lavoro via. ‘Gianni has put away the work / the work away.’

If tutto moves as well, it is plausible to assume that it occupies the same surface position as French tout, as is suggested by the fact that tutto/tout appear in the same relative ordering with respect to adverbial positions (see Cinque 1999). Then the asymmetry in (17) is to be attributed to a different surface position of the verb in the two languages, as proposed by Belletti (2001). Following her analysis, we assume that the participial verb moves up to the relevant Agr head in Italian, thus bypassing the position filled by tutto/tout (possibly Spec of the participial head), while it stops in a lower position (possibly the participial head) in French.4 (20)

a. Gianni ha [AGRP capito [PartP tutto [VP t] ] ] b. Jean a [AGRP [PartP tout compris [VP t] ] ] ‘Gianni has (all) understood (all).’

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

181

If this is correct, then the obligatory agreement in Italian follows from principle (15): the relevant agreement feature is licensed syntactically on the participle; therefore, it must be expressed. By contrast, because the French participle does not check agreement in the overt syntax, principle (15) is irrelevant, and the morphological expression of agreement can only be demanded by a language-specific morphological rule, a rule required in normative French but which appears to fluctuate in spoken French. Neither French nor standard Italian permits past participle agreement if the object is unmoved: this is the basic fact that Kayne’s (1985) classical analysis intended to capture. Interestingly, a number of Romance varieties (including a very archaic-sounding variety of formal Italian, as described for instance in Fornaciari 1974; see Kayne 1985) admits agreement in this case, too: (21) Gianni aveva già presa la sua decisione. Gianni had already taken –FEM–SG the his –FEM decision –FEM

In the terms of our system, here UG does not enforce the morphological expression of agreement, as the feature is unchecked in the overt syntax (because the object has not overtly moved to the relevant spec); however, nothing excludes expression, if the language has a specific morphological rule to this effect. Morphological expression is compulsory under (15) if the object has overtly moved to (or through) the relevant spec and if the verb has also overtly raised to (or through) the relevant functional head. Both conditions are met with clitic movement of the object in Standard Italian, but not with clitic movement in French (the verb has not overtly moved to the relevant functional head) or with in situ objects in Standard Italian (the object has not overtly moved to the relevant Spec). When the conditions for compulsory morphological expression of the feature are not met, the language may still choose to express the feature via a language-specific morphological rule. Alternations in the morphological expression of agreement known in the literature as anti-agreement has been given an explanation similar to ours by Phillips (1998). In a number of languages, subject agreement is marked in declarative clauses, but it may disappear in subject extraction environments, depending on several factors. According to Phillips, anti-agreement occurs when the verb does not need to move to or through Agr and thus does not check agreement feature overtly. For example, Breton and Berber display anti-agreement in positive subject extraction questions but not in negative ones. This difference is traced back to the fact that negation (NegP) is higher than AgrP in these languages. While in positive subject extraction questions the verb does not need to raise to Agr, in negative subject questions the verb needs to move to Neg. In the former case, agreement is not overtly checked and thus need not be morphologically expressed; in the latter, the verb must move through Agr on its way to Neg and thus overtly checks agreement features, which must be morphologically realized. 5. Asymmetry between interrogative and negative do in Early English We are now in a position to explain the asymmetry in (1) through (3) in Early English. The idea is that interrogative and negative do occupy two distinct positions in English, the first higher and the second lower than Agr:

182

IP FUNCTIONAL STRUCTURES

(22) DoInt . . . Agr . . . DoNeg . . .

Agreement is checked syntactically on interrogative do because on its way to the C system it must transit through Agr, where agreement features are licensed; principle (15) then applies, making the morphological expression of agreement compulsory. By contrast, if negative do can remain in a position lower than agreement in the syntax, its features will be checked only at LF and principle (15) is inoperative, whence the fluctuation in the expression of agreement observed in the early system. That interrogative do occurs in a position higher than the whole inflectional system, including the site where agreement is checked, is uncontroversial: the inflectional material must move to the C system in main questions to satisfy the well-formedness condition that triggers interrogative inversion (the Wh Criterion, as in Rizzi 1996, or the checking of the Q feature in C). Less obvious is that negative do (or any other functional verb) may be lower than Agr. A fairly standard assumption is that English functional verbs move to the highest inflectional head in the overt syntax, much as their French or Italian (functional or lexical) counterparts (Pollock 1989; Belletti 1990). Now, that functional verbs move higher than lexical verbs in English is uncontroversially shown by the fact that they must precede the negative marker not. But there are good reasons to reject the standard assumption that they must move to the highest inflectional head: the systematic possibility of adverb interpolation in cases like (23), under some reasonably principled approach to adverb positions (Laenzlinger 1998; Cinque 1999), strongly suggests that even functional verbs in English, while moving higher than lexical verbs, do not have to move as high as the highest inflectional head, hosting the subject in its Spec (Henry 1996; Kayne 1989). (23) He probably does not (doesn’t) know the answer.

Let us assume for concreteness the order of projections proposed in Chomsky (1991), with a negative phrase lower than T, in turn lower than Agr, and with the adverb in the spec of T (or of some other functional head in between Agr and T): (24)

AgrSP AgrS TP

DP

T He

NegP

AdvP

Neg

T Probably

Not

VP

Does

The functional verb does, starting from its own functional VP will have to move through Neg at least as far as T, in order to give rise to the correct word order does not, with the specifier of NegP not optionally cliticizing onto it to give rise to the

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

183

contracted form doesn’t.5 We can now go back to the acquisition pattern observed in section 2. Negative do does not need to raise as far as Agr in the overt syntax, principle (15) is irrelevant, and the morphological expression of agreement is a matter of a language-specific morphological rule, one that must be learned under no special UG guidance. We thus expect the observed fluctuation between do and does for a fairly long period. Interrogative do, instead, must proceed to a head of the C system due to some structural property inherent in the interrogative construction (say the Wh-Criterion). Due to Relativized Minimality, do cannot skip the Agr position: (25)

CP C AgrSP AgrS C Does

TP

DP He

T

AgrS

VP T

But then agreement features are always checked syntactically with interrogative do, and principle (15) is operative, enforcing the morphological manifestation of Agr.6 The asymmetry expressed by table 6-3 is thus explained by the system based on principle (15) and Relativized Minimality, which ultimately traces it back to the different structural positions of interrogative and negative do, one above and the other below Agr.

6. Negative questions A priori, a very relevant additional testing ground for our hypothesis is provided by negative questions. We expect the following pattern: (26)

a. b. c. d.

He don’t go. He doesn’t go. *Why don’t he go? Why doesn’t he go?

Uniniverted and inverted occurrences of negative do should then provide minimal pairs, with only the former showing agreement alternations: inverted negative forms should pass through agreement in their way to C—hence, they should systematically manifest agreement, if our hypothesis is correct. In fact, this prediction is not easy to test in a fully reliable way because only a very small number of negative questions are attested in the corpus. Problems with the mastery of negative questions in child language have been observed independently; in particular, children are re-

184

IP FUNCTIONAL STRUCTURES

luctant to apply the required I to C movement in this construction (Stromswold 1990). This reluctance, possibly related to the need of simultaneously satisfy the Wh-Criterion and the Neg-Criterion, as suggested by Guasti, Thornton, and Wexler (1995), may well be the cause of the limited attestation of the construction in our corpus (the relevant list of the negative questions in our corpus is given in the appendix). Nevertheless, insofar as negative questions are attested in the corpus, they permit a very direct comparison between inverted and uninverted negative questions. So, the agreement pattern observed in the few examples available in the period in which do and does alternate is expressed in table 6-4. While the nine occurrences of noninverted negative do may or may not be inflected, all fourteen occurrences of inverted negative do are inflected, as we expect. The significance of this distribution is obviously limited, because of the very small number of the relevant occurrences and because the fourteen inverted structures are produced by two children, Sarah and Ross, and six of them are tag-like, quasi-formulaic expressions. Nevertheless, the fact is worth mentioning that in the period in which they produce the fourteen occurrences of inverted doesn’t, Sarah and Ross are well within the phase of don’t/doesn’t alternation in noninverted positions. In conclusion, to the limited extent to which negative questions can be brought to bear on our hypotheses, the observed pattern definitely goes in the expected direction. These facts suggest that do cannot be simply a host for negation. If it were, it would have such function both in uninverted and in inverted positions, and we would expect to find examples like (26c), along with the attested cases in (26a).

7. Case and nonagreeing do Our analysis claims that clauses with nonagreeing negative do in Early English are full finite clauses, except that the agreement morphology may fail to be expressed on do as a consequence of its structural position. As for the tensed character of these structures, the very presence of do strongly supports this conclusion: negative do is limited to tensed environments and is sharply excluded from infinitives and gerunds: (28)

a. to (*do) not go b. for not going / *for doing not go

In addition to T, our analysis claims that an abstract syntactic specification of agreement is present in the relevant structures, except that the concrete morphological specification of agreement fails to be expressed. Can syntactic agreement be detected independently? Remember that we have assumed, following Cardinaletti (1997), Table 6-4. Don’t vs. doesn’t in negative questions as a function of inversion

–INV +INV

Do

Does

3 0

6 14

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

185

Schuetze (1997), and Giorgi and Pianesi (1997), among many other references, that agreement is responsible for the assignment of nominative case. We would then expect to observe pronominal subjects overtly marked with nominative case in our construction. In fact, such cases are typically found: (29) He don’t have a baseball.

(Adam 28, 3;4)

These cases provide the most straightforward evidence for the completeness of the relevant structures, with do directly manifesting the presence of T and the nominative subject indirectly detecting the presence of Agr.7 The distribution of nominative and non-nominative case with third-person pronominal subjects of uninflected negative do in the corpus we have considered is expressed by table 6-5. The case-agreement configurations relevant for this table are the following: (30)

a. b. c. d.

She Her She Her

don’t go. don’t go. doesn’t go. doesn’t go.

As the table shows, nominative marked subjects with uninflected do (configuration (30a)) are very robustly attested, thus granting the conclusion that such structures can be complete finite clauses, as we have argued. No case of non-nominative subject is found to co-occur with inflected do (configuration (30d)), which confirms Schuetze and Wexler’s (1996) observation that structures expressing agreement always require nominative subjects in Early English. Case (30c) is target-consistent, hence unproblematic. What remains to be interpreted is the set of cases with uninflected do and non-nominative subjects (configuration (30b)), which are the minority (13/59) of the cases with third-person pronominal subject and uninflected do attested. If we take non-nominative subjects to be a reliable manifestation of the lack of agreement from the syntactic representation (in a language like English, in which nominative is not the default case), we could interpret such cases as involving truncated structures (in the sense of Rizzi (1994) above T (and therefore including do), but under Agr (and therefore excluding nominative marking of the subject). This would then instantiate a case of truncation allowed in principle by the general approach, but which had not been discussed so far in the literature on truncation. (After

Table 6-5. Distribution of third-person nominative and non-nominative subject pronouns as a function of the presence of don’t or doesn’t +Nom

–Nom

46 79

13 0

Don’t Doesn’t X2 = 19.21, p < .005

186

IP FUNCTIONAL STRUCTURES

having completed this article we had access to Ingham 1998, which provides clear evidence for truncation immediately above T on the basis of a case study.) This possible line of analysis of case (30b) is theoretically interesting and not implausible. Before firmly adopting and developing it, though, it is necessary to consider some peculiarities of the thirteen cases attesting this configuration. First, all thirteen examples were produced by the same child, Nina. No other child ever produced a non-nominative subject with don’t in our corpus. Second, Nina produced the thirteen examples in four consecutive files over a short period of four days during which she was taped daily, from age 2;5,25 through 2;5,28 (files 28 to 31). The exhaustive list of examples is given next: (31) Her don’t want go in the bathtub. (Nina 28) Her don’t have a paw. (Nina 29) Her don’t cry. (Nina 29) Her don’t want a cup. (Nina 31) Her don’t want. Uh # her don’t want take a bath # Mommy. Her don’t want a bath. No # her don’t bubbles. Her don’t. Her don’t want it in her eyes. Her don’t want it in her eyes. Her don’t want soap in her + + . . . Her don’t want [//] her want clothes on.

The non-nominative subject her is the only non-nominative subject used by Nina in that period. For masculine subjects, Nina employs the adult form he, which is used both with don’t and doesn’t, as with other verbal forms. Apart from the don’t context, Nina occasionally uses her with bare forms and with some finite forms. Some examples are given below: (32) Her cried. Her nipped me. Her was cried. Her cried. Cause her cried. (Nina 28) Her stand it up. (Nina 29) Her wanna sleep on another bed.

(Nina 30)

In the same period of four days, the nominative subject she is almost never employed. While there are sixty-one sentences (negative and non-negative) with her subjects, there are just five sentences with the subject she. Table 6-6 summarizes Nina’s various occurrences of her and she with different types of verbs. Given the rarity of she subjects in the relevant period, and the use of her also in finite clauses, it is conceivable that the case alternation she/her was not productive in Nina’s grammar at that point. If this is correct, then the thirteen cases illustrating configuration (30b) are irrelevant for drawing theoretical conclusions on the nature of the syntactic representations involved. The question of whether or not this configuration is attested and represents a genuine grammatical option for the child (e.g., with truncation in between Agr and T) awaits further empirical inquiries over larger

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

187

Table 6-6. Nina’s subjects according to verb forms (files 28 to 31) don’t

V-finite

V bare

1 13

2 4

2 44

She Her

production corpora. In any case, our results firmly establish the attestation of the Nom + don’t configuration (30a), which clearly argues for the hypothesis that structures with uninflected negative do can be full finite clauses with missing morphological expression of agreement in Early English.

8. An asymmetry between do and be Under standard assumptions on the structural positions of different verb types in English, finite functional verbs occupy a higher position than lexical verbs in the inflectional system; a less uncontroversial but widely held assumption is that all functional verbs occupy the same position(s) in the inflectional system of the English clause. If this were correct, we would expect alternations of the don’t/doesn’t kind with other functional verbs in Early English. To test this prediction, we should immediately discard modals, which do not manifest morphological alternations with respect to agreement. Potentially relevant would be the have/has alternation; but, the perfective have + past participle construction is too rare in early production corpora to allow us to test the prediction, and possessional have generally functions as a main verb, as in many adult varieties of English. We are then left with copular and progressive be: Do we find a be/is alternation parallel to the do/does alternation? The answer clearly is no, as table 6-7 shows. This table includes all the occurrences of be and of is with third-person subjects during the whole period

Table 6-7. Occurrences of be and of is with with third-person subject in declarative and interrogative sentences during the period investigated Child

Is

Be

Adam Ross Nina Peter Sarah Total

751 614 650 285 720 3,020

3 11 1 0 4 19

188

IP FUNCTIONAL STRUCTURES

investigated. It is evident that while children manifest the do/does alternation, the uninflected form be does not alternate with the inflected form is in third-person contexts: be is used in only 0.6 percent of the cases (19 of 3,039) (see the appendix for an exhaustive list). This then raises the question of why the mechanism allowing do with third-person subjects does not extend to be in child grammars. Why do we have the following asymmetry? (33)

a. b. c. d.

Daddy doesn’t go. Daddy don’t go. Daddy is here. *Daddy be here.

A straightforward possibility is that the child interprets the form be as explicitly marking lack of finiteness, so that, quite independently from the surface position it fills, the form is always excluded from finite contexts (moreover, (33)d cannot arise as a genuine root infinitive because this construction is generally inconsistent with functional verbs; see Rizzi 1933/1994; Wexler 1994; among others).8 A different possibility to account for the asymmetry in (33) is that do and be do not occupy the same position in the inflectional system, the finite occurrence of be being forced to raise to a higher position, possibly because of the unique richness of its morphological agreement paradigm in Modern English. In fact, be is the only English verb with a fully developed paradigm of person distinctions in the present singular and with person distinctions in both present and past. Of potential relevance here is Vikner’s (1995) attempt to precisely define the morphological conditions on syntactic V movement to Agr. His conclusion, based on a wide range of synchronic and diachronic evidence in Romance and Germanic languages, is that V movement occurs in the overt syntax when the verbal paradigm shows morphological distinctions of person in all synthetic tenses. Now, be is the only verb in Modern English meeting Vikner’s criterion, as morphological distinctions for person are found in both present and past paradigms (am, are, is; was, were). If we now interpret (a suitable adaptation of) Vikner’s criterion as applying to individual verbs, we have a principled reason forcing finite occurrences of be to raise to Agr in the syntax. Then the agreement features on be would always be licensed in the overt syntax, and principle (15) would enforce their morphological expression, thus banning (33d). That finite be occurs in a higher position than other functional verbs has also been proposed by Giorgi and Pianesi (1997) on the basis of considerations completely independent from Vikner’s criterion.9 If these proposals are on the right track, the asymmetry be / negative do can be explained in a way parallel to the asymmetry between negative and interrogative do.10

9.

Conclusion

Around the age of three years, children acquiring English typically produce negative sentences with do uninflected for agreement; such uninflected forms freely alternate with the inflected forms for several months in some children’s productions; at the same time, children almost always inflect interrogative do in the C system. The

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

189

sharp contrast between the two kinds of do has led us to develop an analysis based on the different syntactic positions that they fill: interrogative do is moved to the C system, so it is certainly higher than the structural layer in which subject agreement is checked; negative do, while being moved to the higher part of the IP on a par with other functional verbs, does not have to move in the overt syntax to the highest IP layer, as properties of adverbial distribution suggest; so, arguably, it does not have to move to the position in which subject agreement is checked. To capitalize on this positional difference, we have introduced a principle determining the way in which morphology reads and expresses syntactic specifications: if a morphosyntactic feature is checked in the overt syntax, it is expressed by the morphology (principle (15)); if a feature is left unchecked in the overt syntax (and will be checked in the covert syntax through the devices introduced in Chomsky 1995), then UG offers no guidance as to its morphological expression: whether it is realized or not is a matter of a language-specific morphological rule, and this property may vary across closely related systems and fluctuate within the same system. The optionality of agreement with negative do in Early English is a manifestation of this fluctuation, which may remain stable for a longish period in development, as we may expect for a language-specific rule that is not enforced by the core system of UG principles and parameters. On the contrary, interrogative do must move overtly to the C system because of the familiar well-formedness constraints on question formation. It must then pass through the agreement position in the overt syntax due to Relativized Minimality; therefore, agreement is checked syntactically and its morphological expression is enforced by principle (15) in Early (as well as in Adult) English. If negative do (and other functional verbs) moves in the overt syntax at least as far as T (as is shown by its fixed order with negation, among other things), then the observed pattern provides evidence that the position in which subject agreement is checked is distinct from and higher than tense.

Appendix Negative questions including “do” or “does” with third-person subjects in the period in which don’t and doesn’t alternate Why it don’t work? (Adam 26) Why he [/] don’t know how to pretend? (Adam 29) He don’t need it? (Adam 31) Why he doesn’t? (Adam 29) Why dis doesn’t work? (Adam 32) Why it doesn’t stay on? (Adam 32) Mama # he doesn’t stand up real # huh? (Sarah 91) This doesn’t open? (Nina 28) It doesn’t come out? (Nina 29)

190

IP FUNCTIONAL STRUCTURES

Why doesn’t this soldier have a pee+pee? (Ross 36) Why doesn’t Mommy get any sleep? (Ross 40) Why doesn’t he? (Ross 43) Why doesn’t he? (Ross 44) Why doesn’t the big boy get the big package? (Ross 44) He worries about the world # doesn’t he? (Ross 48) Doesn’t Bozo talk? (Sarah 93) Spells dolly # doesn’t it? (Sarah 97) Doesn’t it? (Sarah 102) Doesn’t it # Dad? (Sarah 102) Doesn’t it? (Sarah 102) Doesn’t she look like Baby+Boo? (Sarah 110) Doesn’t she look graceful like that to you? (Sarah 110) Doesn’t it? (Sarah128) Sentences with be How # tiger be so # healthy # and fly # like kite? (Adam 11) Robin always be naughty # when he break pens. (Adam 28) When he be naughty # he break pencil # an(d) you put him in de chair? (Adam 28) The Hulk # Doctor David Banner take his shirt off and be the # be the Hulk. (Ross 24) Because that why he be a penguin. (Ross 30) Yeah # because that why he be the penguin. (Ross 30) Because I want to because that why he be a penguin. (Ross 31) He be nice. (Ross 31) You be the witch and him be the black cat and you be the bat and I be the pumpkin. (Ross 32) You be the witch and him be the black cat and I be the pumpkin. (Ross 32) You be the black cat and you be the witch and her be the bat and I be the pumpkin! (Ross 32) You be a bat and you be the witch and him be the black cat and I[!] (Ross 32) You be the big papa bear # and Mommy be the mommy bear. (Ross 42) And Marky be the big sister. (Ross 44) That be funny? (Nina 33) (a)n(d) # this be in it? (Sarah 71) On here # so it be nice. (Sarah 111) Y(ou) mean like dis # so it be easy? (Sarah 120) Easter be coming too. (Sarah 120)

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

191

Notes 1. The number after the name indicates the file in the CHILDES database from where the example is taken. “Adam 28” means that the sentence was produced by Adam in file 28. 2. In the system of Chomsky (1998) this generalization would be expressed as follows: if a feature is checked as part of the complex operation “move” (“agree” plus “second merge”) it must be overtly expressed in the morphology; if it is checked by “agree” only, individual languages may or may not express it in the morphology. 3. Of course, if a language does not have the appropriate morphological paradigm, this system applies vacuously. For example, in Mainland Scandinavian, Agr features are presumably checked syntactically, determining nominative case on the subject, but are not expressed on V because the verbal morphology of the language lacks this option; the same conclusion holds for English modals and past forms of the auxiliary “have.” 4. Agreement of the past participle in passives is fully obligatory in French as well: (i) La voiture a été mise/*mis dans le garage the car-FEM-SG has been put-FEM-SG/put-MASC-SG in the garage This suggests that agreement in passive past participles occupies a different and lower position than agreement in perfect past participles, perhaps a position immediately higher than a Voice head (see Cinque 1999) and lower than the aspectual head of Belletti (1990). It is arguable that this position is always reached by French passive participles, with the morphological expression of agreement compulsory under (15). The obligatoriness of agreement in (i) is significant as it shows that the contrast between Italian and French in (17) is not a trivial consequence of the phonetic weakness of French past participle agreement in comparison to Italian. Interestingly, first- and second-object clitics in Italian differ from third-person object clitics in that they trigger agreement only optionally, as displayed by the contrast below: (ii) Gianni vi ha visto/visti. Gianni you-pl has seen-MASC-SG/seen-MASC-PL (iii) Gianni li ha visti/*visto. Gianni them-PL has seen-MASC-PL/seen-MASC-SG First- and second-person clitics always are positioned to the left of third-person clitics in Italian. If this linear order constraint corresponds to a hierarchical difference, the optionality of agreement in (ii) may be directly amenable to our analysis through the assumption that first- and second-person clitics trigger agreement on a higher projection than do third-person clitics, a projection that is not (necessarily) reached by past participles in the overt syntax. 5. If n’t is not the clitic form of not, but the head of a higher NegP, as is suggested by the possible co-occurrence of the two negative markers in examples like the following (from Kayne 1989), “He couldn’t not have accepted,” the analysis remains essentially the same, except that does raises at least as far as the head of this higher NegP, placed in between Agr and T, to pick up n’t. On the possible co-occurrence of multiple NegPs in different languages, see Zanuttini (1997). 6. Our analysis of nonagreeing do in EE immediately carries over to other bare verbal forms in this language for example, the cases in (2). Uncontroversially, lexical verbs do not raise to the highest inflectional heads in English (if they move at all from the VP-internal position), so that Agr is unchecked in overt syntax, principle (15) is inoperative, and the marking of agreement is a matter of a language-specific morphological rule. We thus expect a certain instability in the ‘s marking, which is witnessed by the adult varieties of English that do not take this option (Labov 1985); we also expect instability in acquisition, with a longish period in which ‘s may be omitted, as, in fact, we observe in acquisition corpora.

192

IP FUNCTIONAL STRUCTURES

This has immediate consequences for the analysis of uninflected verbal forms in Early English. Granting that such forms may arise as the English specific variety of Root Infinitives (Wexler 1994), our analysis leads us to conclude that they may also arise as full finite structures with the verbal inflection that fails to be morphologically expressed. This dual structural possibility leads us to expect a nonhomogeneous behavior of uninflected verbal forms in Early English, which may help explain many peculiarities of such forms with respect to genuine Root Infinitives in other early languages (e.g., the fact that uninflected verbal forms in Early English continue to be produced well after the end of the early null subject period, as originally pointed out by Ingham 1982, whereas genuine root infinitives in other languages never survive after the end of early null subjects). We will not explore the consequences of this extension of our analysis in this article. 7. Notice that this case is not naturally amenable to the analysis of uninflected structures in Early English by Schuetze and Wexler (1996) and Schuetze (1997). According to this analysis, clauses with verbs lacking the ‘s marking can arise through the omission of either T or Agr (or both); but in (29) both layers should be syntactically present, if our assumptions are correct. 8. Strictly speaking, in adult English the form be is not limited to finite contexts because it occurs in subjunctive clauses such as “I demand that he be released.” But it is conceivable that in such cases be is not the subjunctive form of the verb but, rather, the infinitival form selected by a null subjunctive modal (the null counterpart of should in “He should be released”) (Emonds 1976; Roberts, 1985). 9. Examples like “John probably isn’t ready” would then involve movement of the subject to a specifier position higher than the agreement layer (for discussion about the location of the agreement projection in English, see Cinque 1999). 10. This structural analysis clearly is more appealing than the one based on the fact that be inherently expresses finiteness; if we take Vikner’s observation as criterial for V-raising to Agr, however, the structural analysis may lead us to expect be/is oscillations before agreement distinctions of be in the present and past paradigms are mastered. Not knowing the full developmental course of acquisition of the paradigm of be, we prefer to leave the room open for both alternative analyses of the be/do asymmetry.

References Aoun, J., E. Benmamoun, and D. Sportiche (1994) “Agreement, Word Order and Conjunction in Some Varieties of Arabic.” Linguistic Inquiry 25, 195–220. Belletti, A. (1990) Generalized Verb Movement. Turin: Rosenberg and Sellier. Belletti, A. (2001) “Agreement Projections,” in M. Baltin and Ch. Collins, Handbook of Syntactic Theory, 483–510. Oxford: Blackwell. Belletti, A. (1999) “Inversion as Focalization.” Unpublished ms., University of Siena. Bloom, L. (1970) Language Development: Form and Function in Emerging Grammars. Cambridge, Mass.: MIT Press. Brandi, L., and P. Cordin (1989) “Two Italian Dialects and the Null Subject Parameter,” in O. Jaeggli and K. Safir (eds.) The Null Subject Parameter, 111–162. Dordrecht: Kluwer. Brown, R. (1973) A First Language. Cambridge: Harvard University Press. Cardinaletti, A. (1997) “Agreement and Control in Expletive Constructions.” Linguistic Inquiry 28, 521–533. Chomsky, N. (1991) “Some Notes on Economy of Derivation and Representation,” in R. Fridein (ed.) Principles and Parameters in Comparative Grammar. Cambridge, Mass.: MIT Press. Chomsky, N. (1995) The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. (1998) “Minimalist Inquiries: The Framework.” Unpublished ms., MIT.

AGREEMENT AND TENSE AS DISTINCT SYNTACTIC POSITIONS

193

Cinque, G. (1999) Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York: Oxford University Press. Clark, E. (1982) “Language Change during Language Acquisition,” in M. E. Lamb and A. L. Brown (eds.) Advances in Child Development, vol. 2, 171–195. Hillsdale, N.J.: Erlbaum. Emonds, J. (1976) A Transformational Approach to English Syntax: Root Transformations, Structure-Preserving and Local Transformations. New York: Academic Press Fornaciari (1974) Sintassi italiana. Florence: Sansoni. Friedemann, M.-A., and T. Siloni (1993) “Agrobject Is Not Agrparticiple.” Linguistic Review 14, 69–96. Giorgi, A., and F. Pianesi (1997) Tense and Aspect: From Semantics to Morphosyntax. New York: Oxford University Press. Guasti, M.T., and L. Rizzi (1996) “Null Aux and the Acquisition of Residual V2,” in A. Stringfellow, D. Cahana-Amytay, E. Hughes, and A. Zukowski (eds.) Proceedings of the 20th Boston Conference on Language Development, 284–295. Sommerville, Mass.: Cascadilla Press. Guasti, M. T., R.Thornton, and K. Wexler (1995) “Negation in Children’s Questions: The Case of English,” in D. MacLaughlin and S. McEwen (eds.) Proceeding of the 19th Boston Conference on Language Development, 228–239. Sommerville, Mass.: Cascadilla Press. Henry, A. (1996) Belfast English and Standard English. New York: Oxford University Press. Ingham, R. (1992) “The Optional Subject Phenomenon in Children’s English: A Case Study.” Journal of Child Language 19, 133–151. Ingham, R. (1998) “Tense without Agreement in Early Clause Structure.” Language Acquisition 7, 51–81. Kayne, R. (1975) French Syntax. Cambridge, Mass.: MIT Press. Kayne, R. (1985) “Notes on English Agreement.” Unpublished ms. Graduate Center, City University of New York. Kayne, R. (1989) “Romance Clitics, Verb Movement and PRO.” Linguistic Inquiry 22, 647– 686. Kayne, R. (1994) The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Labov, W. (1995) “The Case of Missing Copula: The Interpretation of Zeroes in AfricanAmerican English,” in L. R. Gleitman and M. Lieberman (eds.) An Invitation to Cognitive Science: Vol. 1. Language, 23–56. Cambridge, Mass.: MIT Press. Laezlinger, C. (1998) Comparative Studies on Word Order Variation: Adverbs, Pronouns, and Clause Structure in Romance and Germanic. Amsterdam: John Benjamins. MacWhinney, B. (1991) The CHILDES Project: Tools for Analyzing Talk. Hillsdale, N.J.: Erlbaum. MacWhinney, B., and C. Snow (1985) “The Child Language Data Exchange System.” Journal of Child Language 12, 271–296. Phillips, C. (1995) “Syntax at Age Two: Cross-linguistic Differences,” in C. T. Schutze, J. B. Ganger, and K. Broihier (eds.) MIT Working Papers on Language Processing and Acquisition 26, 325–382. Cambridge, Mass.: MIT Press. Phillips, C. (1998) “Disagreement between Adults and Children,” in A. Mendikoetxea and M. Uribe-Extebarria (eds.) Theoretical Issues on the Morphology-Syntax Interface, 359– 396. San Sebastian, Spain: ASJI. Pollock, J-Y. (1989) “Verb Movement, Universal Grammar and the Structure of IP.” Linguistic Inquiry 20, 365–424. Raposo, E. (1987) “Case Theory and Infl-to-Comp: The Inflected Infinitive in European Portuguese.” Linguistic Inquiry 18, 85–109. Rizzi, L. (1933/1994) “Some Notes on Linguistic Theory and Language Development: The Case of Root Infinitives.” Language Acquisition 3, 371–393.

194

IP FUNCTIONAL STRUCTURES

Rizzi, L. (1996) “Residual Verb Second and the Wh Criterion,” in A. Belletti and L. Rizzi (eds.) Parameters and Functional Heads, 63–90. New York: Oxford University Press. Rizzi, L. (2001) “Relativized Minimality Effects,” in M. Baltin and Ch. Collins, Handbook of Syntactic Theory, 89–110. Oxford: Blackwell. Roberts, I. (1985) “Agreement Parameters and the Development of English Modals.” Natural Language and Linguistic Theory 3, 21–58. Roberts, I. (1993) Verbs and Diachronic Syntax: A Comparative History of English and French. Dordrecht: Kluwer. Schuetze, C. T. (1997) “INFL in Child and Adult Language: Agreement, Case and Licensing,” Ph.D. Diss., MIT, Cambridge, Mass. Schuetze, C. T., and K. Wexler (1996) “Subject Case Licensing and English Root Infinitives,” in A. Stringfellow, D. Cahana-Amitay, E. Hughes, and A. Zukowski (eds) Proceeding of the 20th Annual Boston University Conference on Language Development, 670–681. Sommerville, Mass.: Cascadilla Press. Stromswold, K. (1990) “Learnability and the Acquisition of Auxiliaries,” Ph.D. Diss., MIT, Cambridge, Mass. Stromswold, K. (1995) “The Acquisition of Inversion and Negation in English: A Reply to Deprez and Pierce.” Unpublished ms., Rutgers University. Suppes, P. (1974) “The Semantic of Children’s Language.” American Psychologist 29, 103– 114. Vikner, S. (1995) Verb Movement and Expletive Subjects in the Germanic Languages. New York: Oxford University Press. Wexler, K. (1994) “Optional Infinitive, Head Movement and Economy of Derivation,” in N. Horstein and D. Lightfoot (eds.) Verb Movement, 305–350. Cambridge: Cambridge University Press. Zanuttini, R. (1997) Negation and Clausal Structure. New York: Oxford University Press.

7

The Distribution of Functional Projections in ASL Evidence from Overt Expressions of Syntactic Features     Signed languages provide a unique type of evidence of the architecture of functional projections. As in other signed languages, ASL, the focus of this article, has nonmanual expressions of many of the major syntactic features postulated to occur in functional heads. These expressions take the form of particular gestures on the head and upper body that occur potentially over phrasal domains, in parallel with manual signing. The distribution and intensity of these grammatical markings provide evidence of hierarchical structure and phrasal boundaries. In this article, we summarize the major findings of our recent research on the syntax of ASL relevant to the issue of the mapping of functional projections. For further detail and discussion of the structure of the noun phrase and the clause in ASL, see especially Neidle, Kegl, MacLaughlin, Bahan, and Lee (2000),1 where we have argued for the hierarchical structures shown in figures 7-1 and 7-2. Here we focus, in particular, on the projections of tense, aspect, agreement, and negation. We begin with some background information about the characteristics of ASL as a signed language and about its basic grammatical organization. Then we consider the distribution of nonmanual markings associated with syntactic features, using negative and wh-constructions as illustrations of the essential generalizations governing their distribution. Finally, we examine in greater detail the evidence for the relative hierarchical arrangement of tense, agreement, aspect, and negation shown in figure 7-1. 195

196

IP FUNCTIONAL STRUCTURE

Position to which wh-phrases may move

CP C'

Surface subject

TP

Spec C

T' T

NegP Neg' Neg

AspP Asp'

Asp

AGR SP

[perfect]

AGR S' AGR S

AGR P O

AGR ' O

AGR

O

VP

Figure 7-1. Syntactic structure of the ASL clause

1. Background information about ASL American Sign Language is the natural language used by the Deaf community2 in the United States and other parts of North America. ASL is distinct from English, as well as from other signed languages, such as British Sign Language (BSL), French Sign Language (LSF), and so on. While American and British spoken English are mutually intelligible (for the most part), ASL and BSL are quite different. In contradistinction, ASL is historically related to LSF because of contacts between American and French Deaf people. Contrary to one popular misconception, there is no universal sign language. Signed languages evolve naturally everywhere there are sufficient numbers of Deaf people, and signed languages partake in the same kinds of linguistic variation (i.e., formal and informal registers, dialectal variations) and change as spoken languages. 1.1.

Phonology and morphology

Despite the difference in modality, signed languages have essentially the same structural organization as spoken languages. The main difference is that the basic discrimi-

FUNCTIONAL PROJECTIONS IN ASL

197

DP DP i

D' D

Agr o P

[+agr]

Agr o '

[+definite]

POSS

Agr o

… NP

[+agr]

ti

N' N

Figure 7-2. Structure of a possessive DP

natory articulatory units are produced by the hands and body and are perceived visually. However, such units, comparable to phonemes, do exist.3 Morphemes are composed of and distinguished by particular handshapes, hand orientations, places of articulation relative to the body, and movement characteristics. For example, as shown in figure 7-3, the sign FATHER is produced with a particular open handshape, with the palm oriented toward the nondominant side, and with the thumb contacting the forehead. The sign FINE involves the same handshape and orientation, but the point of contact is the chest. There are analogs for the types of phonological processes that exist in spoken language, based on the fundamental physical nature of articulation. Just as there are allophones of phonemes in spoken languages, so there are in signed languages. For example, the “B” handshape may be articulated with either straight or curved fingers (as illustrated in figure 7-4), depending, in part, on the context in which the phoneme occurs. There are also processes of assimilation (with respect to handshape, hand orientation, and place and manner of articulation). Just as in spoken languages, signs may be articulated less precisely as speed of production increases. There are instances of contractions (as described by Kegl and Poizner 1994, for example). There is an analog to whispering (using a small signing space hidden from view of unintended “overlookers”) and an analog to louder speech (involving larger signing space that can be perceived more easily at a distance). ASL phonologists have made a convincing case for the existence of syllable structure. According to Perlmutter (1992), for example, movements correspond to vowels, and the positions held immediately before or after a movement are equivalent to consonants.

198

IP FUNCTIONAL STRUCTURE

Figure 7-3. FATHER vs. FINE (From Carol Neidle, Judy Kegl, Dawn MacLaughlin, Benjamin Bahan, and Robert G. Lee [2000] The Syntax of American Sign Language: Functional Categories and Hierarchical Structure. Cambridge, Mass.: MIT Press, © MIT Press)

ASL also has a productive morphology. There are derivational affixes, such as an agentive suffix that can be added to a verb like TEACH to form the sign for TEACHER. In addition to affixal inflections, there are also inflections expressed through modification of the articulation of the stem. 1.2. Syntax: The state of the art It is only within the last forty years or so that ASL has been recognized as a language worthy of study. Whereas the phonology and morphology of ASL have been fairly well described, the syntax of ASL has received less careful attention. In fact, there is still very little agreement about some of the most basic aspects of ASL sentence structure. Most of the disagreements about ASL syntax involve not only differences in analysis but also strikingly different claims about data, even for the simplest constructions.

Figure 7-4. Allophonic variants of “B” handshape (From American Sign Language Linguistic Research Project [ASLLRP], © ASLLRP 1999)

FUNCTIONAL PROJECTIONS IN ASL

199

1.2.1. Particular challenges for linguistic research on ASL ASL poses special challenges for linguistic research that may not be immediately obvious. One complicating factor is the need to have access to video data, since written representations of ASL used to designate example sentences in linguistics articles or presentations provide too little information about what was actually signed. There is no universally accepted written representation to capture the phonological and morphological aspects of the language. According to the standard conventions, an English near-equivalent word written in capital letters is used to gloss each ASL sign (although, of course, there is no one-to-one correspondence between English words and ASL signs). Nonmanual expressions (of the face and upper body) that occur in parallel with manual signing are notated by a labeled line that extends over the signs with which the markings are coarticulated. Despite the inadequacies of this gloss notation, almost all of the linguistic claims about ASL refer to data presented solely in gloss form, and access to video exemplars of the constructions under discussion is normally not provided. This makes it impossible to evaluate differing claims about apparently similar constructions. Our research group has made it a point to make video data corresponding to the constructions we discuss publicly available. The majority of the examples in this article are available as digitized QuickTime™ movies from our Web site (http://www.bu.edu/ asllrp) and on CD-ROM. One factor contributing to the difficulties of analyzing video-based language data has been the lack of tools to facilitate such analysis. For example, if one is to analyze the distribution of grammatical markings such as eyebrow raises, eye gazes, or head movements, careful frame-by-frame transcription is required. This work is extremely time-consuming and almost impossible to carry out by hand using standard videotapes. Even if such detailed transcriptions are available, the problem of searching through handwritten transcriptions and finding the segments of videotape appropriate for testing hypotheses is daunting. For this reason, we have been developing SignStream™, a software application that facilitates precise coding and fast retrieval of video-based information (Neidle and MacLaughlin 1998; MacLaughlin, Neidle, and Greenfield 2000; Neidle 2001; see also http://www.bu.edu/asllrp/SignStream). A picture of the SignStream interface is shown in figure 7-5. Our understanding of the functioning of nonmanual marking has been enhanced by our ability to take advantage of this kind of tool. Most important, however, the process of ASL data collection is fraught with hazards because of the unusual sociolinguistic setting in which signed languages are used. It is worth a brief digression to discuss some of the sociolinguistic peculiarities of signed languages that have consequences for linguistic research. 1.2.2. Sociolinguistic considerations From a sociolinguistic perspective, signed languages present certain unique characteristics (as discussed in more detail in Neidle et al. 2000, chap. 2, and references contained therein). Although it has been shown that Deaf children born to Deaf signing families acquire their native language in essentially the same way as hearing children born to hearing parents, only about one deaf child in ten is born

200

IP FUNCTIONAL STRUCTURE

Figure 7-5. Sample SignStream video (top) and gloss windows (bottom) (From American Sign Language Linguistic Research Project [ASLLRP], © ASLLRP 1999)

into a family that provides exposure to signed language from birth (Schein and Delk 1974). Many deaf children are not exposed to signed language until later in life, and in such cases, they may never acquire native proficiency. Thus, in ASL-signing Deaf communities, the native signers constitute a minority of about 10 percent or less. Astonishingly, relatively few schools for the Deaf use signed language as a primary means of instruction; many impose artificial signed systems that are more English-like and that are not, in fact, natural languages (see Lane, Hoffmeister, and Bahan 1996 for discussion). In such cases, Deaf children frequently learn ASL from

FUNCTIONAL PROJECTIONS IN ASL

201

native signing peers outside of the classroom, while ASL is stigmatized in formal educational settings. This situation presents certain challenges for non-native signing linguists who engage in ASL data collection. First and foremost, it is essential to collect data from those signers who have acquired ASL natively. It is also important to minimize potential interference from English-based signing systems that even native signers may use with non-native signers, particularly in formal and educational settings. These may be some of the reasons for the fact that progress in the area of ASL syntax has been exceedingly slow. There is very little consensus on even basic issues. We have made every effort to ensure the integrity of the data collection and analysis, however, by having active involvement by Deaf native signing researchers. We have also made the data on which we base our conclusions public.

2. Mapping the functional architecture of ASL Before proceeding to our major findings on the functional architecture of ASL, we first provide some syntactic background. 2.1.

Basic sentence structure

The basic sign order is SVO. A variety of surface word orders is found, given the prevalence of left dislocations, topicalization, null arguments, pronominal right dislocations, and sentence-final tags. Despite earlier suggestions that signed languages involve essentially free word order,4 more careful analysis has revealed that ASL and other signed languages have essentially the same type of hierarchical syntactic structure as other languages. 2.2.

Nonmanual markings

One interesting source of evidence for hierarchical structure is the distribution of particular nonmanual grammatical markings that occur over phrasal domains. Nonmanual markings are movements of the head and upper body that have several different functions within ASL. Some of these functions are strictly linguistic, while others are not. There is a fundamental distinction between affective facial expressions (used universally by both Deaf and hearing people) and the grammatical use of facial expressions that is unique to signed languages. Neurolinguistic research on signed languages has demonstrated that affective expressions are processed in the right hemisphere, while linguistic expressions are processed in the left hemisphere. Thus, impairment affects linguistic and affective processes differentially (see Kegl and Poizner 1991, 1997; Poizner and Kegl 1992). Further evidence for the distinction between the two types of nonmanual markings comes from both first- and second-language acquisition studies, which have revealed differences in the acquisition of similar facial gestures used for linguistic and affective purposes (McIntire and Reilly 1988; Reilly, McIntire, and Bellugi 1990; Reilly

202

IP FUNCTIONAL STRUCTURE

and Bellugi 1996). In sum, while signers can and do use facial expressions affectively, such uses are distinguishable from linguistic uses. There are several types of nonmanual markings used for linguistic purposes. Lexical and adverbial markings co-occur with individual signs, while syntactic markings potentially occur over larger, phrasal domains. Our research has shown that nonmanual syntactic markings are frequently associated with syntactic features (e.g., +neg, +y/n, +wh, f-features) located in the heads of functional projections and that their distribution is characterized by the following generalizations: • The nonmanual marking may spread over the c-command domain of the node with which it is associated (reflecting relations that hold at Spell-Out).5 This spread is optional if manual material is available locally (e.g., in head or specifier position). However, if no manual material is available locally, then the marking spreads obligatorily so that it may be coarticulated with manual material. • The intensity of the nonmanual marking is greatest at the node of origin and decreases as distance from the source increases (Baker 1996; Neidle et al. 2000). We now illustrate these generalizations with respect to the nonmanual markings for negation and wh-questions. The nonmanual expression of negation, labeled “neg,” consists of a cluster of gestures, including furrowed brows, squinted eyes, and a side-to-side headshake.6 This combination of behaviors may occur concurrently with the signing of the manual negative expression, glossed as NOT, as shown in (1). neg (1)

JOHN [NOT]Neg BUY HOUSE

‘John is not buying a house.’

Alternatively, this nonmanual marking can spread over the c-command domain of Neg, as shown in (2).7 (2)

neg [ [NOT]Neg [BUY HOUSE] VP]NegP ‘John is not buying a house.’ JOHN

In this case, the expression is strongest at the source of the negative feature, that is, with the articulation of the sign NOT. The expression diminishes in intensity as distance from that node increases. This can be most easily discerned by examining the change in the headshake. Before the onset of the movement, the head assumes a starting position that will enable it to rotate over the maximal arc as the sign NOT is articulated.8 The headshake continues, with each subsequent rotation covering a smaller arc, as illustrated schematically in figure 7-6. It also possible, in ASL, to have a negative sentence that does not contain a manual sign of negation. In this case, the nonmanual expression must spread over its c-command domain (since otherwise there would be no local manual material available with which it could be articulated). (3)

neg *JOHN [+neg]Neg

BUY HOUSE

FUNCTIONAL PROJECTIONS IN ASL

203

Figure 7-6. Intensity of negative headshake (From Benjamin Bahan [1996] “Nonmanual Realization of Agreement in American Sign Language. Ph.D. diss., Boston University. © Benjamin Bahan)

(4)

neg [+neg ]Neg BUY HOUSE ‘John is not buying a house.’

JOHN

In (4), as in (2), the intensity of the negative marking is greatest at the node containing the negative feature and diminishes toward the end of the sentence. The same generalizations also account for the distribution of yes/no question marking, wh-question marking, and rhetorical question marking. Here we provide illustrations of wh-constructions. We have argued that, in ASL, wh-phrases move rightward to a sentence-final specifier position (which we have assumed to be [Spec, CP]), where the +wh feature is checked. On such an analysis, we can explain, straightforwardly, the distribution of wh-marking9 in questions of several types. ASL has questions with both moved and in situ wh-phrases. Under certain circumstances, the wh-marking may occur over the wh-phrase itself or it may spread over the entire IP. In other cases, spread over IP is required. The possibilities of spread are correctly predicted by the generalizations just presented. When the wh-phrase moves to [Spec, CP], spread of the wh-marking over the c-command domain of the head containing the +wh feature is optional, since even when spread does not occur, there is manual material available locally with which the wh-marking can be articulated.10 Such optional spread is illustrated by examples (5) through (8). Examples (5) and (6) illustrate wh-phrases that have moved from subject position, while (7) and (8) illustrate moved wh-objects. If the wh-phrase remains in situ, as in (9) and (12), however, then the spread of the wh-marking over the entire IP is obligatory, since, otherwise, there is no manual material with which the marking may be coarticulated. Thus, (10), (11), (13), and (14) are ungrammatical. With a wh-phrase in [Spec, CP], the spread of nonmanual marking is optional: wh (5)

[ti

HATE JOHN]IP WHOi

(6)

wh [ti HATE JOHN ]IP WHOi ‘Who hates John?’

(7)

[JOHN SEE ti

(8)

wh [JOHN SEE ti YESTERDAY]IP WHOi ‘Who did John see yesterday?’

wh YESTERDAY]IP WHOi

204

IP FUNCTIONAL STRUCTURE

With a wh-phrase in situ, the spread of nonmanual marking is obligatory: wh (9)

[WHO

HATE JOHN]IP

wh (10) *[WHO HATE

JOHN]IP

wh (11) *[WHO HATE JOHN]IP ‘Who hates John?’ wh (12) [JOHN

SEE WHO YESTERDAY]IP

wh (13) *[JOHN

SEE WHO YESTERDAY]IP

wh (14) *[JOHN SEE WHO YESTERDAY]IP ‘Who did John see yesterday?’

Here we have considered only simple wh-constructions. Further details of our analysis are contained in Neidle, Kegl, Bahan, Aarons, and MacLaughlin (1997); Neidle, MacLaughlin, Lee, Bahan, and Kegl (1998a, 1998b); and Neidle et al. (2000). In those works, we discuss more complex constructions involving multiple wh-phrases associated with a single argument. We also address possible alternative analyses of these data consistent with Kayne’s (1994) antisymmetry framework. In addition, we argue against a very different proposal by Petronio and Lillo-Martin (1997). In the case where wh-marking has spread over the entire IP, the intensity of the marking is greatest at the end of the sentence. As we have seen, both with negative and wh-constructions, the intensity of the nonmanual marking provides evidence for the position of the head containing the relevant feature. In negative constructions, the intensity of the nonmanual marking diminishes as the rest of the sentence is signed, indicating that the +neg feature is contained within a Neg node that is to the left of its complement. In contrast, nonmanual wh-marking increases in intensity as the sentence is signed, indicating that the head containing the +wh feature is located to the right of its complement. For both negative and wh-constructions, the domain over which the marking spreads provides evidence of the boundaries of the functional projection in question. We now turn to the functional structure of the clause and the noun phrase, specifically focusing on agreement, tense, negation, and aspect. We show that syntactic agreement may be expressed nonmanually, both in the noun phrase and in the clause; study of the distribution of these nonmanual markings provides evidence as to the boundaries of the various functional projections. 2.3.

Functional structure

There have been differing claims about the hierarchical arrangement of functional categories in the clause. In particular, considerable discussion has revolved around the relative position of projections of tense and agreement. Pollock (1989, 1997) has

FUNCTIONAL PROJECTIONS IN ASL

205

proposed that tense and agreement constitute distinct functional projections, with TP dominating Agr(S)P. Belletti (1990) and Chomsky (1993) have proposed that AgrSP dominates TP. Others have suggested that the relative ordering of AgrSP and TP may differ from language to language (Ouhalla 1991), or even within a single language (Campbell 1991). More recently, Chomsky (1995) and Baker (1996) have suggested that there are no agreement projections at all. We argue here that in ASL, there is evidence to support the postulation of agreement projections and that such projections are dominated by TP. Evidence comes from the distribution of manual and nonmanual realizations of agreement. In the subsections that follow, we consider first tense, then agreement; finally, we briefly discuss aspect and examine how it fits into the functional hierarchy. 2.3.1.

Tense

In ASL, the three-dimensional signing space can be used to represent temporal information. In general, the space in front of the signer’s body is used to indicate future time, while the space behind the signer is associated with past time. Relative distance from the body indicates greater or lesser distances in time. For example, the articulation of the adverbial “in the future” may be varied to express distinctions between near future and distant future, as illustrated in the top and middle rows of figure 7-7. (For each row, the picture on the left represents the starting point of the sign and the picture on the right represents the end point.) For the sign expressing distant future, the dominant hand extends further up and forward than for the sign expressing near future, which involves a shorter movement. It had generally been assumed that ASL and other signed languages lack grammatical tense but make use exclusively of time adverbials to convey temporal information. However, we have argued that this is incorrect and that ASL does encode tense grammatically (Aarons, Bahan, Kegl, and Neidle 1995). The evidence for the existence of lexical tense markers, occurring in the canonical position—that is, the head of TP—is, we believe, straightforward. Part of the prior confusion on this issue may have arisen from the similarity in the articulation of many lexical tense markers and the corresponding adverbials from which they presumably derived. The sign used to express future tense, glossed here as WILL, is similar to the adverbial just described. One crucial difference, however, is that there is a frozen articulation, as shown in the bottom row of figure 7-7. There is a fixed path length used to articulate future tense. This sign cannot be modulated to distinguish near future from distant future. This is the sign that occurs in a sentence like (15): (15)

JOHN WILL GO

‘John will go.’

The same is true of the articulation of other lexical tense markers that are related to temporal adverbials. In addition, the distribution of tense markers differs from that of adverbials, with tense markers occurring in T—that is, between the subject and the negative element, if there is one11—and temporal adverbials occurring generally at the beginning or end of the clause. Interestingly, some recent additions to the repertoire of lexical tense markers do not have corresponding adverbial forms. For example, the prefix “ex” as in “ex-foot-

206

IP FUNCTIONAL STRUCTURE

Figure 7-7. Articulation of adverbial vs. tense marker: adverbial: in the (distant) future (top); adverbial: in the (near) future (middle); TENSE marker: WILL (frozen articulation) (bottom) (From Carol Neidle, Judy Kegl, Dawn MacLaughlin, Benjamin Bahan, and Robert G. Lee (2000) The Syntax of American Sign Language: Functional Categories and Hierarchical Structure. Cambridge, Mass.: MIT Press, © MIT Press)

ball player” was borrowed from English into ASL. It subsequently evolved into an independent tense marker in ASL, glossed as #EX,12 which conveys the meaning of habitual past, somewhat like “used to” in English. Although #EX thus conveys both tense and aspect information, it occurs with the same distribution as other tense markers.13 An example is given in (16). (16)

#EX LIKE CHOCOLATE ‘John used to like chocolate.’ JOHN

FUNCTIONAL PROJECTIONS IN ASL

207

There seems to be some dialectal variation among signers as to exactly which tense markers are used. #EX is used as a tense marker by some, but not all, native signers. There are also modals in ASL that occur in complementary distribution with tense markers. We have assumed that both modals and tense markers occur in T. (In a more elaborate system with multiple tense, aspect, and modality projections, such as that proposed by Cinque 1999, a different explanation for the cooccurrence restrictions between tense markers and modals would be required.) Modals, like tense markers, appear after the subject and before a negative element, if there is one. The modal SHOULD is illustrated in (17) and (18). (17)

JOHN SHOULD GO

‘John should go.’ neg (18)

JOHN SHOULD NOT GO

‘John should not go.’

Both tense markers and modals often have contracted negative forms, as well, as illustrated by (19) and (20). neg (19)

JOHN SHOULD^NOT GO

‘John shouldn’t go.’ neg (20)

JOHN WILL^NOT GO

‘John won’t go.’

When the signs WILL and NOT are contracted, instead of a distinct handshape and movement for each sign, there is a single movement involving a handshape that combines the two individual signs’ handshapes. While there are overt tense markers, such as future and past, sentences in the present tense normally do not contain a lexical tense marker, as shown in example (21). (21)

JOHN EAT

‘John eats/is eating.’

It is also not uncommon to find sentences involving future or past interpretations that lack overt tense markers. This is similar to what occurs in English or French in sentences like (22) and (23). (22) Tomorrow John goes to Venice.

(English)

(23) Demain, je pars pour Paris. (French) ‘Tomorrow, I leave for Paris.’

The absence of tense markers in comparable constructions may explain, in part, why there was the presumption that grammatical tense is not employed in ASL. 3.3.2.

Agreement

In this section, we discuss agreement within the clause and the noun phrase. Agreement may be realized manually through morphological inflection; it may also be

208

IP FUNCTIONAL STRUCTURE

expressed nonmanually. We first present some background information on the way that the three-dimensional signing space is used for the instantiation of f-features. Then we address agreement within the DP and the clause. We show that there are significant parallelisms in the expression of agreement in the two domains. 3.3.2.1. Spatial representation of f-features In ASL (and signed languages generally), the three-dimensional signing space is used for referential purposes. A referent may be associated with a particular location in space. That location can then be accessed in a variety of ways to express agreement and accomplish subsequent reference.14 For example, a signer who is talking about a man and a woman might establish the man in one location and the woman in another. To refer to “the man,” the signer might use the definite determiner, which is articulated by pointing the index finger to the location in space associated with the referent. This definite determiner sign, glossed as IX, is illustrated in figure 7-8.15 A subsequent pronominal reference would involve exactly the same pointing gesture. The fact that the definite determiner and the pronominal in ASL have exactly the same articulation supports proposals that pronominals are essentially determiners (Postal 1969; Abney 1987; among others). To express a possessive relation, a possessive marker sign, glossed as POSS, is used; this sign involves pointing to the location in space associated with the possessor, but with an open palm instead of the index finger, as shown in figure 7-9. Reflexives and emphatics are expressed by pointing to the location associated with the referent with a different handshape, as shown in figure 7-10. Finally, subject and object agreement are expressed manually by beginning or ending the agreeing sign in the location associated with the agreeing referent. This is illustrated in figure 7-11, which shows the start and end points of the sign GIVE. In fact, ASL has several different morphological classes of verbs, not all of which allow for overt manual realization of subject and object agreement (see Padden 1983, 1988). So-called agreeing verbs, like GIVE, generally involve starting the verb’s arti-

Figure 7-8. Determiner/pronominal (from Benjamin Bahan, Judy Kegl, Dawn MacLaughlin, and Carol Neidle (1995) “Convergent Evidence for the Structure of Determiner Phrases in American Sign Language,” in Leslie Gabriele, Debra Hardison, and Robert Westmoreland (eds.) FLSM VI: Proceedings of the Sixth Annual Meeting of the Formal Linguistics Society of Mid-America, 1–12. Bloomington: Indiana University Linguistics Club. © Benjamin Bahan et al.)

FUNCTIONAL PROJECTIONS IN ASL

209

Figure 7-9. Possessive marker (From Carol Neidle, Judy Kegl, Dawn MacLaughlin, Benjamin Bahan, and Robert G. Lee (2000) The Syntax of American Sign Language: Functional Categories and Hierarchical Structure. Cambridge, Mass.: MIT Press, © MIT Press.

culation at the point in space associated with the subject16 and ending the articulation at the point in space associated with the object.17 A sentence containing an agreeing verb (SHOOT) is presented in (24).18 (24)

JOHNi iSHOOTj FRANKj

‘John shoots Frank.’

In contrast, a verb like LOVE, sometimes called a “plain” verb, does not allow modification of its start and end points to reflect subject and object agreement. The verb is articulated in the same way, regardless of the identity of its subject or object.19 (25)

JOHN LOVE MARY

‘John loves Mary.’

In sum, particular spatial locations are accessed in ASL in precisely the same set of contexts where f-features are postulated to be relevant cross-linguistically. We have analyzed these spatial locations as instantiations of f-features. In what follows, we focus, in particular, on person agreement features (by restricting attention to definite, singular referents).20 Notice, though, that this kind of agreement is different in one important respect from what is generally found in spoken languages. Unlike the typical distinction

Figure 7-10. Reflexive / emphatic (From Benjamin Bahan [1996] “Non-manual Realization of Agreement in American Sign Language. Ph.D. diss., Boston University. © Benjamin Bahan 1996)

210

IP FUNCTIONAL STRUCTURE

Figure 7-11. Manual expression of verb agreement: beginning and end point of GIVE (From Benjamin Bahan [1996] “Non-manual Realization of Agreement in American Sign Language. Ph.D. diss., Boston University, © Banjamin Bahan)

among first, second, and third persons, ASL allows further subclassification of thirdperson referents, and in fact does not make significant grammatical distinctions between second and third persons. The primary grammatical distinction is between first and non-first person; however, agreement is sensitive to referential information as well as person information, as different second- and third-person referents can be distinguished.21 One consequence of this is that pronominal reference in ASL is essentially unambiguous. 3.3.2.2. Nonmanual expressions of f-features within DP Like the syntactic features +neg and +wh, f-features have nonmanual expressions (see Bahan 1996, MacLaughlin 1997). To express f-features nonmanually, there are two available devices in ASL: the head may tilt toward the same location in space involved in all of the agreement phenomena just described, and the eyes may gaze toward that same location. It should be noted that nonmanual expressions of agreement are not required for grammaticality, although they occur with great frequency. We consider first the distribution of these nonmanual expressions of f-features within DP. In a simple definite noun phrase like “the old man,” the index finger (expressing the definite article) points to a specific location associated with the DP’s referent. Concurrently with this pointing sign, the head may tilt toward that same location or the eyes may gaze toward that same location. Finally, it is possible to have both head tilt and eye gaze toward that location. See examples (26) to (28). Alternatively, these nonmanual markings associated with the determiner may spread over the rest of the DP, as shown in (29) to (31).

FUNCTIONAL PROJECTIONS IN ASL

Without spread

With spread

hti (26) [IXi OLD

MAN]DP ARRIVE

head tilti (29) [IXi OLD MAN]DP

ARRIVE

egi (27) [IXi OLD

MAN]DP ARRIVE

eye gazei (30) [IXi OLD MAN]DP

ARRIVE

hti egi (28) [IXi OLD MAN]DP ARRIVE ‘The old man is arriving.’

211

head tilti eye gazei (31) [IXi OLD MAN]DP ARRIVE ‘The old man is arriving.’

Thus, the distribution of the nonmanual agreement markings within DP follows from the generalizations presented earlier, if we postulate that the agreement features associated with these nonmanual markings occur in the head of DP, in the same position where the definite determiner is also found. Since there is manual material in the relevant node, the spread of these expressions over the c-command domain of D would be predicted to be optional. As with other nonmanual markings, these markings reach their maximal intensity concurrently with the signing of the lexical item in the head of the node with which the corresponding features are associated. With respect to head tilt, for example, the head achieves its maximal tilted position at the same time that the index finger reaches its maximal extension. This provides support for the claim that the determiner is (at least at Spell-Out) in the same node in which the agreement features expressed by the head tilt reside. Just as with negation, the manual sign corresponding to the definite determiner is not required. The agreement information may be carried solely by the nonmanual marking. In such cases, the nonmanual marking obligatorily spreads over the rest of the DP, since otherwise it would have no manual material with which to be expressed. This is the case with a construction like (32). head tilti eye gazei (32) [ OLD MAN]DP ARRIVE ‘The old man is arriving.’

It is particularly illuminating to examine the manifestation of nonmanual expressions of agreement within possessive DPs, since such DPs contain two distinct sets of f-features: those associated with the possessor, which is in “subject” position of the DP, and those associated with the possessee. In ASL possessive constructions, the possessive marker (illustrated in figure 7-9) manually expresses the agreement features of the possessor. We have argued that this lexical possessive marker occurs in D (see figure 7-2). In possessive constructions containing nonmanual agreement markings, head tilt is used to express “subject” agreement, while eye gaze is used for “object” agreement. Consider the predictions for the spread possibilities. While the head tilt should

212

IP FUNCTIONAL STRUCTURE

spread optionally over its c-command domain (since there is manual material, namely POSS, with which it may be articulated even if spread does not occur), the spread of eye gaze is predicted to be obligatory, as there is no manual material in the relevant agreement head. This is exactly what we find. hti egj (33) [JOHN [POSSi]D [+agrj]AgrO FRIEND]DP ‘John’s friend is arriving.’

ARRIVE

hti egj (34) [JOHN [POSSi]D [+agrj]AgrO FRIEND]DP ‘John’s friend is arriving.’

ARRIVE

As with the definite determiner, the possessive marker, POSS, also reaches its maximal extension at the same time that the head tilts to its maximal position. This supports the claim that POSS occurs in the same node in which the subject (possessor) agreement features reside. To summarize, agreement features within the DP can be expressed nonmanually by head tilt and eye gaze pointing toward locations in space associated with the relevant f-features, the same locations that are accessed manually for manual expression of those f-features. In non-possessive DPs, either head tilt or eye gaze, or both, may constitute the nonmanual expression of the single set of agreement features contained within the DP. Spread of these markings is optional when there is a manual determiner sign in D. In possessive constructions, where there are two distinct sets of agreement features, head tilt expresses those associated with the subject (or possessor) and eye gaze expresses those associated with the object (or possessee). There is one significant difference in the spread of head tilt and eye gaze in possessive constructions. While head tilt spreads optionally, the spread of eye gaze is obligatory. This difference is a consequence of the availability of manual material in the head of the Agr projection in one case, but not in the other. We now turn to the nonmanual expression of agreement features in the clause, where we find parallelism in the expression of subject and object agreement features by head tilt and eye gaze. 3.3.2.3. Nonmanual expressions of f-features within the clause Within the clause, f-features postulated to occur in the heads of agreement projections have the same nonmanual expressions just discussed. These expressions spread over the c-command domain of the Agr head. However, in the clause, this spread is obligatory because there is no manual material in Agr (as the verb does not raise overtly to Agr in ASL; see Neidle et al. 2000). We first present examples of intransitive constructions. As in the case of DPs that have a single set of agreement features, either head tilt or eye gaze or both may be used to express the agreement features associated with the subject.

FUNCTIONAL PROJECTIONS IN ASL

(35)

(36)

213

head tilti [+agri]Agr ARRIVE eye gazei b. JOHNi [+agri]Agr ARRIVE head tilti eye gazei c. JOHNi [+agri]Agr ARRIVE ‘John is arriving.’ a.

JOHNi

head tilti [+agri]Agr BATHE eye gazei b. JOHNi [+agri]Agr BATHE head tilti eye gazei c. JOHNi [+agri]Agr BATHE ‘John is bathing.’ a.

JOHNi

In transitive constructions, however, agreement with the subject is expressed by head tilt, while agreement with the object is expressed by eye gaze (if such nonmanual realizations are used). It is significant to note that these nonmanual expressions of subject and object agreement may be found with all morphological classes of verbs: not only “agreeing” verbs that have manual inflection for subject and object agreement, but also “plain” verbs, which do not. The next two examples illustrate transitive sentences with nonmanual markings of subject and object agreement, the first involving the agreeing verb BLAME, and the second involving the plain verb LOVE. head tilti eye gazej (37)

[+agri]AgrS [+agrj]AgrO ‘Ann blames Mary.’ ANNi

iBLAMEj MARYj

head tilti eye gazej (38)

[+agri]AgrS [+agrj]AgrO ‘John loves Mary.’ JOHNi

LOVE MARYj

Figure 7-12 contains pictures illustrating head tilt and eye gaze co-occurring with the agreeing verb GIVE. Sentences such as (38) provide evidence of syntactic agreement features occurring in the head of functional projections independent of inflectional agreement features present on verbs. In fact, we have argued in more detail elsewhere (Neidle et al. 2000) that the distribution of nonmanual correlates of syntactic features, in general, provides evidence to support a feature-checking approach (such as that proposed in Chomsky 1993) involving features present both as part of inflected lexical items and in the heads of functional projections (over earlier approaches whereby lexical items acquire inflection in the course of the derivation). As already mentioned, within the clause, the spread of nonmanual expressions of subject and object agreement is obligatory, which follows from the postulation that

214

IP FUNCTIONAL STRUCTURE

Figure 7-12. Nonmanual expressions of subject agreement (head tilt) and object agreement (eye gaze) with verb GIVE (From Benjamin Bahan [1996] “Non-manual Realization of Agreement in American Sign Language. Ph.D. diss., Boston University. © Benjamin Bahan)

there is no manual material in either AgrS or AgrO. Interestingly, the timing of head tilt and eye gaze are nonetheless different in intransitive and transitive constructions. In sentences like (35) and (36), if both head tilt and eye gaze are used to express the agreement features postulated to occur in AgrS, they are realized simultaneously. However, in (37) and (38), if head tilt expresses features within AgrS while eye gaze expresses features within a separate agreement projection, the onset of the head tilt generally (slightly) precedes the onset of eye movement. Further support for the claim that these nonmanual markings are, in fact, expressions of agreement comes from their ability to license null arguments. ASL allows null subjects and objects, with both plain and agreeing verbs. With verbs that have overt agreement morphology, null subjects and objects are licensed regardless of whether the optional nonmanual agreement markings are present. Consider examples (39) and (40). Since SHOOT is overtly inflected for both subject and object agreement, a null subject or object is allowed. (39) proi iSHOOTj FRANKj ‘(He/she) shoots Frank.’ (40)

proj ‘John shoots (him/her).’ JOHNi iSHOOTj

With a plain verb, the only possible overt manifestation of agreement is the nonmanual expression of subject and object agreement. With verbs that do not have overt manual agreement inflection, a null argument is licensed only if the nonmanual expression of agreement is present.22 Thus there is a contrast in the grammaticality of sentences (41) and (42), on the one hand, which include nonmanual agreement markings, and sentences (43) and (44), on the other, which do not. As shown by (45), it is not the case that nonmanual markings of agreement are generally required for grammaticality.

FUNCTIONAL PROJECTIONS IN ASL

215

head tilti eye gazej (41) proi [+agri]AgrS [+agrj]AgrO ‘(He/she) loves Mary.’

(42)

[+agri]AgrS [+agrj]AgrO ‘John loves (him/her).’ JOHNi

(43) *pro

head tilti eye gazej LOVE proj

LOVE MARY

(44) *JOHN (45)

LOVE MARYj

LOVE

pro

JOHN LOVE MARY

To summarize, head tilt and eye gaze constitute the nonmanual expressions of the f-features found in the heads of agreement projections, both within the clause and the noun phrase. The parallelisms in the expression of agreement in transitive clauses and possessive DPs on the one hand and between intransitive clauses and possessorless DPs on the other are summarized in table 7-1. This is in keeping with the growing body of cross-linguistic evidence for such parallels between clauses and DPs with respect to case and agreement morphology.23 With respect to the hierarchical relationships between tense and agreement, sentences such as (46) demonstrate that tense occurs in a position distinct from agreement and that TP (with WILL in head position) dominates AgrSP. head tilti eye gazej

neg (46)

[+agri]AgrS [+agrj]AgrO ‘John will not visit mother.’ JOHNi WILL NOT

VISIT MOTHERj

Therefore, the ASL data provide evidence to support the postulation of agreement projections (distinct from tense) and the relative ordering of tense and agreement shown in the tree in figure 7-1.

Table 7-1. Parallelisms between transitive/intransitive noun phrases and clauses (From Dawn MacLaughlin [1997] “The Structure of Determiner Phrases: Evidence from American Sign Language,” Ph.D. diss., Boston University © Dawn MacLaughlin) Transitive

Clause

Noun phrase

head tilti eye gazej DPi [ ]Agr [ ]Agrj V DPj i head tilti eye gazej DPi [POSS]Agri [ ]Agrj NPj

Intransitive

head tilti eye gazei DPi [ ]Agri V head tilti eye gazei [IX]Agri NPi

216

IP FUNCTIONAL STRUCTURE

3.3.3. Aspect Aspectual information in ASL may be marked through the use of lexical aspect markers or through morphological inflection. There is a lexical perfect marker that may occur, as appropriate, after negation but before nonmanual markings of agreement and before the verb, as shown in examples (47) and (48). head tilti eye gazej (47)

JOHNi FINISH

[+agri]AgrS [+agrj]AgrO ‘John has visited mother.’

(48)

JOHNi NOT FINISH READ BOOK

VISIT MOTHERj

neg ‘John has not read the book.’

We analyze FINISH as occurring in the head of a perfect aspect projection located above AgrSP but below NegP (see figure 7-1). In addition to the perfect aspect marking, other kinds of aspectual information in ASL may be expressed through inflection. See Klima and Bellugi (1979) for details of some of the semantic and articulatory distinctions; the aspects they describe include iterative, distributional, habitual, durational, incessant, frequentative, and resultative. For example, durational aspect may be expressed through a modulation of the verb stem involving repeated (and rhythmic) circular movements. Iterative aspect is generally expressed through reduplication of the verb stem. These aspectual distinctions are similar to those included by Cinque (1999) in his proposed universal hierarchy of clausal functional projections; we assume, consistent with his approach, that such aspectual information in ASL is contained within the heads of aspectual projections.24 It is also interesting to note that adverbs that convey similar semantic information (e.g., about iteration or duration) may bear similar kinds of morphological inflection. Thus, the adverb CONTINUOUSLY (shown in figure 7-13) may be articulated with the same kind of repeated circular motion just described for verbal aspectual inflection of duration. Analogously, frequency adverbs such as AGAIN (shown in figure 7-14) exhibit reduplicated articulations. The occurrence of the same kind of morphological inflections on aspectually marked verb forms and the semantically related adverbs is particularly significant in light of Cinque’s proposal that such adverbs occur in the specifier positions of the corresponding aspect phrases.

Figure 7-13. CONTINUOUSLY (From American Sign Language Linguistic Research Project [ASLLRP] © ASLLRP, Boston University, 1999)

FUNCTIONAL PROJECTIONS IN ASL

217

In addition to the similarities in inflections on adverbs and aspectually marked verbs, we have begun to investigate parallels in aspectual inflections occurring within both the clausal and nominal domains. Our work thus far has focused on the inflection that expresses distributive aspect.25 Distributive aspectual marking on the verb distinguishes a sentence meaning ‘John gave each of the boys books’ from a sentence with the meaning ‘John gave the boys books’. This inflection takes the form of reduplication of the verb stem (similar to the general expression of iterative aspect) that is expressed simultaneously with plural verb agreement marking (see MacLaughlin, Neidle, Bahan, and Lee 2000).26 This same kind of distributive inflectional marking is found within the nominal domain, on determiner and pronominal elements, suggesting that a functional projection for distributive aspect occurs in the extended noun phrase projection as well as in the clause.27 This is yet another case where data from ASL provide evidence to support parallel structures within the nominal and clausal domains.

4.

Conclusion

We have argued for a particular functional architecture of the ASL clause and noun phrase, not unlike what has been proposed for other (spoken) languages. Thus, the structure of ASL does not differ dramatically from that of spoken languages. An important source of evidence for hierarchical structure comes from the distribution of nonmanual syntactic markings in ASL, a kind of evidence that is not available in spoken languages. In particular, the ASL data provide evidence for the following conclusions, with respect to the following: • The status of agreement features. Agreement features are like other syntactic features in having nonmanual correlates that obey the same generalizations about distribution and intensity. • The status of agreement projections. Agreement features head functional projections that are distinct from tense and aspect. • The dual representation of agreement features. There is evidence for morphological agreement features associated with lexical items (manifested by manual agreement inflection), as well as agreement features in functional heads.

Figure 7-14. AGAIN (From American Sign Language Linguistic Research Project [ASLLRP] © ASLLRP, Boston University, 1999)

218

IP FUNCTIONAL STRUCTURE

• Parallels in the expression of agreement in transitive/intransitive DP and TP. With respect to the expression and distribution of nonmanual agreement markings, possessive DPs pattern with transitive clauses while nonpossessive DPs pattern with intransitive clauses. The one difference between DP and TP is that DP may have manual material in its highest agreement head, resulting in optional spread of the associated nonmanual agreement marking. • The distribution of aspect projections within the clause and the instantiation of aspectual information. There is evidence for multiple aspect projections within the clause. Similar inflections occur on aspectually marked verbs and semantically related adverbs. • Parallels related to aspect in DP and TP. Distributional aspect inflection can be found on elements in both domains. • The relative ordering of functional projections in the clause. In particular, TP dominates the agreement projections in ASL. Notes The research reported on here has been conducted as part of the American Sign Language Linguistic Research Project (ASLLRP), in collaboration with other members of our research team, especially Benjamin Bahan (who has generously contributed important insights relevant to this essay) and Robert G. Lee. We are grateful to the National Science Foundation for providing funding for this project (grants SBR-9410562, SBR-9729010, IIS-9528985, IIS-9912573, and EIA-9809340 to Boston University). We would also like to thank Norma Bowers Tourangeau, Lana Cook, Ginger Leon, and Kim Hand Arrigo for useful discussions and help with various aspects of this work. Thanks also to DawnSignPress for allowing us to use the video illustrated in figure 7-5. This article is an outgrowth of a presentation given at the workshop on the mapping of functional projections, held at the University of Venice in January of 1999, and we are grateful to the workshop organizers and participants for their questions and comments. 1. Many of the ASLLRP publications are available in electronic format at our Web site, http://www.bu.edu/asllrp, where it is also possible to download digitized video examples, as signed by native signers, corresponding to many of the constructions we discuss. 2. Following Woodward 1973, the term “Deaf” with a capital D has been used to refer to those individuals who are linguistically and culturally deaf (members of the “Deaf-World,” in the sense of Lane, Hoffmeister, and Bahan 1996); the term “deaf” describes those who have some degree of hearing loss. 3. For more about ASL phonology, see, for example, Padden and Perlmutter (1987); Sandler (1989); Perlmutter (1992); Coulter (1993); Brentari (1998). 4. See, for example, Tervoort (1968), or, for a more recent claim along the same lines, Bouchard and Dubuisson (1995) (but see also the reply by Kegl, Neidle, MacLaughlin, Hoza, and Bahan 1996). 5. Crucially, nonmanual spread reflects c-command relations that hold at Spell-Out (within the model of Chomsky 1993 or 1995, chapter 4), or at s-structure (within earlier models), but neither before nor after. We are not adopting recent proposals that features that must be checked have moved by Spell-Out. For example, proposals such as Bobaljik (1995), Brody (1995), Groat and O’Neil (1996), and Pesetsky (1997) would eliminate the need for covert movement after Spell-Out. We assume, with Chomsky (1995), that features (at least) do move covertly after Spell-Out.

FUNCTIONAL PROJECTIONS IN ASL

219

6. For description of the nonmanual marking of negation, see Stokoe (1960); Bellugi and Fischer (1972); Baker (1976, 1980); Baker and Padden (1978); Baker and Cokely (1980); Liddell (1980); Baker-Shenk (1983); Veinberg and Wilbur (1990). 7. Petronio (1993) and Petronio and Lillo Martin (1997) have made different claims about the distribution of negative marking, insisting that it must spread over the entire IP (including the subject); they thus dispute the grammaticality of examples such as (2). It is possible that Petronio and Lillo-Martin are misinterpreting the anticipatory portion of the head movement (see note 8), as discussed in Neidle et al. (2000). Crain and Lillo-Martin (1999, 282, examples (3a) and (3b)) not only report examples such as (2) to be grammatical but use them to illustrate the properties of nonmanual grammatical marking in ASL. In any case, the spread reported here is consistent with previous claims in the literature (see Veinberg and Wilbur 1990; McIntire, Reilly, and Anderson 1994). 8. Head movements like the negative headshake and the affirmative headnod generally involve an anticipatory motion to a starting position that will allow a maximal initial movement. Thus, the head normally raises before a headnod and moves sideward before a headshake. 9. Wh-marking consists of furrowed brows, squinted eyes, a slight rapid headshake, head tilted back. See, for example, Baker-Shenk (1983); Aarons (1994). 10. [Spec, CP] is sufficiently local to the +wh feature to enable expression of the whmarking with manual material in that position. This may be attributable to the fact that the +wh feature in C is shared by the Spec through Spec-Head agreement (Rizzi 1991). 11. Tense markers are also found in the T position of tag constituents, which may occur sentence-finally, as illustrated in (i), where “hn” represents a head nod that is characteristic of the tag construction and is found generally with clausal constituents containing null verbal material (Liddell 1980). hn (i)

JOHN WILL GO, WILL

‘John will go, he will.’ See Aarons et al. (1995) and Neidle et al. (2000) for discussion of the tag construction (and of ways in which this construction has been misanalyzed by others in the literature). There are also other cases where T ends up in sentence-final position as a result of leftward movement of other constituents (Aarons 1994; Neidle et al. 2000). 12. The # is used to indicate fingerspelled loan signs. 13. While tense markers precede negation, aspect markers such as FINISH (which marks perfect aspect) follow negation. This is discussed further in section 3.3.3. 14. There is interesting systematic variation in the nature of the location in space that is associated with the referent, depending on its definiteness and number properties. For discussion, see MacLaughlin (1997); MacLaughlin et al. (2000); Neidle et al. (2000). For simplicity, in this article we focus on definite singular referents. 15. The existence of determiners in ASL (and signed languages generally) is a somewhat controversial issue. We have argued that ASL has both definite and indefinite determiners; see MacLaughlin 1997 and Neidle et al. (2000). One possible source of confusion may be the similar articulation of the locative adverbial. Analogous to what we saw earlier with the distinction between tense markers and time adverbials, the form corresponding to the determiner has a frozen articulation, while the adverbial may vary in its articulation to distinguish greater or lesser distance in space. 16. There is also an alternative form of subject agreement not discussed here. For further details, see Bahan (1996); Neidle, Bahan, MacLaughlin, Lee, and Kegl (1998); Bahan, Kegl, Lee, MacLaughlin, and Neidle (2000); and Neidle et al. (2000).

220

IP FUNCTIONAL STRUCTURE

17. With a few verbs, the so-called backward verbs like INVITE (see Friedman 1976; Brentari 1988, 1998; Padden 1988), subject agreement is expressed instead by a suffix while object agreement is expressed by a prefix. 18. Subscripts used on verbs, determiners, head tilt, and eye gaze mark spatial information about the start and end points of these elements. Subscripts on the corresponding nouns are intended to convey agreement information only and do not necessarily imply that the noun was signed in the spatial location designated by the subscript (although some nouns may, in fact, be articulated in that location, as discussed in note 19). 19. Other classes of signs also express agreement through modification of the location properties of the stem. For example, some nouns can be articulated in different locations to reflect the f-features of the DP. An agreeing noun like HOUSE, for example, may be articulated in a nonneutral location in space, thereby conveying referential information, while a plain noun like BOY cannot (see MacLaughlin 1997). 20. ASL also expresses number agreement. For discussion, see, for example, Klima and Bellugi (1979); Padden (1983, 1988, 1990); MacLaughlin, Neidle, Bahan, and Lee (2000). 21. The ASL literature contains several different claims about person distinctions. For example, Friedman (1975) suggests that ASL distinguishes among first, second, and third person. In contrast, Lillo-Martin and Klima (1990) claim that there are no person contrasts in ASL. Meier (1990) provides strong arguments for a grammatical distinction between first person and non-first person. (Lillo-Martin 1995 adopts Meier’s view.) What we have proposed is that, while there is a primary distinction between first and non-first persons (consistent with Meier), nonfirst person can be further subclassified into many distinct person values. 22. This analysis differs from previous claims by Lillo-Martin (e.g., 1986, 1991) that sentences containing plain verbs lack agreement projections and therefore that null arguments in such sentences must be licensed by Topic, as has been proposed for Chinese (Huang 1982, 1984). See Bahan et al. (2000) for arguments against the topic-licensing account for ASL. 23. For discussion of similar parallels between agreement marking in the clause and in DP in spoken languages, see, for example, Abney (1987); Szabolcsi (1987); and Ouhalla (1991). With respect to case, Bittner and Hale (1996, 60) report that many languages that use ergative case do so both for the subject of a transitive VP and for the possessor in a possessive noun phrase. 24. Certain aspectual markings may occur in combination (although restrictions on cooccurrence of these aspect markings, as well as their possible co-occurrence with the perfect aspect marker, FINISH, have not been fully investigated). Within the clause, we have not yet determined the position of the various aspectual heads. However, assuming some version of Baker’s (1985) Mirror Principle, we have found some evidence at least to suggest that there is an iterative aspect projection that is higher in the tree than the durative, as there are cases where iterative inflection involves reduplication of a verbal form that is inflected for durative. 25. The expression of distributive aspect interacts with the expression of number, as a plural referent is inevitably involved with the notion of distributivity. For discussion, see MacLaughlin, Neidle, Bahan, and Lee (2000) and references therein. 26. It is interesting that the use of distributive aspect implies that the event was repeated with respect to each member of the set that constitutes the relevant argument. Thus it is not coincidence that reduplication is used both for iterative and distributive aspect inflections. As we discuss in MacLaughlin, Neidle, Bahan, and Lee (2000), reduplication is also used to mark plurality of both events and nouns. (There are also other aspectual inflections involving reduplication that do not involve plurality, however.) 27. For a similar proposal of a distributive aspect projection just above DP in Spanish, see Sánchez (1996).

FUNCTIONAL PROJECTIONS IN ASL

221

References Aarons, Debra (1994) “Aspects of the Syntax of American Sign Language,” Ph.D. Diss., Boston University. Aarons, Debra, Benjamin Bahan, Judy Kegl, and Carol Neidle (1995) “Lexical Tense Markers in American Sign Language,” in Karen Emmorey and Judy S. Reilly (eds.) Language, Gesture, and Space, 225–253. Hillsdale, N.J.: Lawrence Erlbaum. Abney, Steven P. (1987) “The English Noun Phrase in Its Sentential Aspect,” Ph.D. Diss., MIT. Bahan, Benjamin (1996) “Non-manual Realization of Agreement in American Sign Language,” Ph.D. Diss., Boston University. Bahan, Benjamin, Judy Kegl, Robert G. Lee, Dawn MacLaughlin, and Carol Neidle (2000) “The Licensing of Null Arguments in American Sign.” Linguistic Inquiry 31(1), 1–27. Bahan, Benjamin, Judy Kegl, Dawn MacLaughlin, and Carol Neidle (1995) “Convergent Evidence for the Structure of Determiner Phrases in American Sign Language,” in Leslie Gabriele, Debra Hardison, and Robert Westmoreland (eds.) FLSM VI: Proceedings of the Sixth Annual Meeting of the Formal Linguistics Society of Mid-America, 1–12. Bloomington: Indiana University Linguistics Club. Baker, Charlotte (1976) “What’s Not on the Other Hand in American Sign Language,” in C. Mufwene and S. Steever (eds.) Papers from the 12th Regional Meeting of the Chicago Linguistic Society, 24–32. Chicago: University of Chicago Press. Baker, Charlotte (1980) “Sentences in ASL,” in Charlotte Baker and Robbin Battison (eds.) Sign Language and the Deaf Community: Essays in Honor of William C. Stokoe, 75–86. Silver Spring, Md.: National Association of the Deaf. Baker, Charlotte, and Dennis Cokely (1980) American Sign Language: A Teacher’s Resource Text on Grammar and Culture. Silver Spring, Md.: T. J. Publishers. Baker, Charlotte, and Carol A. Padden (1978) “Focusing on the Nonmanual Components of American Sign Language,” in Patricia Siple (ed.) Understanding Language through Sign Language Research, 27–57. New York: Academic Press. Baker, Mark (1985) “The Mirror Principle and Morphosyntactic Explanation.” Linguistic Inquiry 16, 373–415. Baker, Mark (1996) The Polysynthesis Parameter. New York: Oxford University Press. Baker-Shenk, Charlotte (1983) “A Micro-Analysis of the Nonmanual Components of Questions in American Sign Language,” Ph.D. diss., University of California at Berkeley. Belletti, Adriana (1990) Generalized Verb Movement. Turin: Rosenberg and Sellier. Bellugi, Ursula, and Susan D. Fischer (1972) “A Comparison of Sign Language and Spoken Language: Rate and Grammatical Mechanisms.” Cognition 1(3), 173–200. Bittner, Maria, and Ken Hale (1996) “The Structural Determination of Case and Agreement.” Linguistic Inquiry 27(1), 1–68. Bobaljik, Jonathan (1995) “Morphosyntax: The Syntax of Verbal Inflection,” Ph.D. Diss., MIT, Cambridge, Mass. Bouchard, Denis, and Colette Dubuisson (1995) “Grammar, Order and Position of wh-Signs in Quebec Sign Language.” Sign Language Studies 87, 99–139. Brentari, Diane (1988) “Backward Verbs in ASL: Agreement Re-opened,” in D. Brentari, G. Larson, and L. MacLeod (eds.) Chicago Linguistics Society: Volume 24, Part 2. Agreement in Grammatical Theory, 16–27. Chicago: Chicago Linguistics Society. Brentari, Diane (1998) A Prosodic Model of Sign Language Phonology. Cambridge, Mass.: MIT Press. Brody, Michael (1995) Lexico-Logical Form: A Radically Minimalist Theory. Cambridge, Mass.: MIT Press.

222

IP FUNCTIONAL STRUCTURE

Campbell, Richard (1991) “Tense and Agreement in Different Tenses.” Linguistic Review 8, 159–183. Chomsky, Noam (1993) “A Minimalist Program for Linguistic Theory,” in Kenneth Hale and Samuel Jay Keyser (eds.) The View from Building 20, 1–52. Cambridge, Mass.: MIT Press. Chomsky, Noam (1995) The Minimalist Program. Cambridge, Mass.: MIT Press. Cinque, Guglielmo (1999) Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York: Oxford University Press. Coulter, Geoffrey R. (ed.) (1993) Phonetics and Phonology: Current Issues in ASL Phonology. Vol. 3. New York: Academic Press. Crain, Stephen, and Diane Lillo-Martin (1999) An Introduction to Linguistic Theory and Language Acquisition. Malden, Mass.: Blackwell. Friedman, Lynn A. (1975) “Space, Time and Person Reference in ASL.” Language 51, 940– 961. Friedman, Lynn A. (1976) “The Manifestation of Subject, Object, and Topic in the American Sign Language,” in Charles N. Li (ed.) Subject and Topic, 125–148. New York: Academic Press. Groat, Erich, and John O’Neil (1996) “Spell-Out at the LF Interface,’ in Werner Abraham, Samuel Epstein, Höskuldur Thráinsson, and C. Jan-Wouter Zwart (eds.) Minimal Ideas, 113–139. Amsterdam: John Benjamins. Huang, C.-T. James (1982) “Logical Relations in Chinese and the Theory of Grammar,” Ph.D. diss., MIT, Cambridge, Mass. Huang, C.-T. James (1984) “On the Distribution and Reference of Empty Pronouns.” Linguistic Inquiry 15, 531–574. Kayne, Richard (1994) The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Kegl, Judy, Carol Neidle, Dawn MacLaughlin, Jack Hoza, and Benjamin Bahan (1996) “The Case for Grammar, Order and Position in ASL: A Reply to Bouchard and Dubuisson.” Sign Language Studies 90, 1–23. Kegl, Judy, and Howard Poizner (1991) “The Interplay between Linguistic and Spatial Processing in a Right Lesioned Signer.” Journal of Clinical and Experimental Neuropsychology 13, 38–39. Kegl, Judy, and Howard Poizner (1994) The Phonetics of Contraction in American Sign Language. Boston: Linguistic Society of America. Kegl, Judy, and Howard Poizner (1997) “Crosslinguistic/Crossmodal Syntactic Consequences of Left-Hemisphere Damage: Evidence from an Aphasic Signer and His Identical Twin.’ Aphasiology 11(1), 1–37. Klima, Edward S., and Ursula Bellugi (1979) The Signs of Language. Cambridge, Mass.: Harvard University Press. Lane, Harlan, Robert J. Hoffmeister, and Benjamin Bahan (1996) A Journey into the DeafWorld. San Diego, Calif.: DawnSignPress. Liddell, Scott K. (1980) American Sign Language Syntax. The Hague: Mouton. Lillo-Martin, Diane (1986) “Two Kinds of Null Arguments in American Sign Language.” Natural Language and Linguistic Theory 4, 415–444. Lillo-Martin, Diane (1991) Universal Grammar and American Sign Language. Dordrecht: Kluwer. Lillo-Martin, Diane (1995) “The Point of View Predicate in American Sign Language,” in Karen Emmorey and Judy S. Reilly (eds.) Language, Gesture, and Space, 155–170. Hillsdale, N.J.: Lawrence Erlbaum. Lillo-Martin, Diane, and Edward S. Klima (1990) “Pointing out Differences: ASL Pronouns in Syntactic Theory,” in Susan D. Fischer and Patricia Siple (eds.) Theoretical Issues in

FUNCTIONAL PROJECTIONS IN ASL

223

Sign Language Research: Vol. 1. Linguistics, 191–210. Chicago: University of Chicago Press. MacLaughlin, Dawn (1997) “The Structure of Determiner Phrases: Evidence from American Sign Language,” Ph.D. Diss., Boston University. MacLaughlin, Dawn, Carol Neidle, Benjamin Bahan, and Robert G. Lee (2000) “Morphological Inflections and Syntactic Representations of Person and Number in ASL.” Language et surdité (Recherches linguistiques de Vincennes 29), 73–100. MacLaughlin, Dawn, Carol Neidle, and David Greenfield (2000) SignStream User’s Guide, Version 2.0. Report No. 9, American Sign Language Linguistic Research Project. Boston: Boston University. McIntire, Marina L., and Judy Snitzer Reilly (1988) “Nonmanual Behaviors in L1 and L2 Learners of American Sign Language.” Sign Language Studies 61, 351–375. McIntire, Marina L., Judy S. Reilly, and Diane Anderson (1994) “Two Forms of Negation— or NOT.” Paper presented at the Eighteenth Boston University Conference on Language Development, Boston, November 1993. Meier, Richard P. (1990) “Person Deixis in American Sign Language,” in Susan D. Fischer and Patricia Siple (eds.) Theoretical Issues in Sign Language Research: Vol. 1. Linguistics, 175–190. Chicago: University of Chicago Press. Neidle, Carol (2001) “SignStream™: A Database Tool for Research on Visual-Gestural Language,” in P. Boyes Braem, T. Hanke, B. Bergman, and E. Pizzuto (eds.) Databases, Transcription and Tagging Tools (Journal of Sign Language and Linguistics 4½), 193– 204. Neidle, Carol, Benjamin Bahan, Dawn MacLaughlin, Robert G. Lee, and Judy Kegl (1998) “Realizations of Syntactic Agreement in American Sign Language: Similarities between the Clause and the Noun Phrase.” Studia Linguistica 52(3), 191–226. Neidle, Carol, Judy Kegl, Benjamin Bahan, Debra Aarons, and Dawn MacLaughlin (1997) “Rightward Wh-Movement in American Sign Language,” in D. Beerman [sic], D. LeBlanc, and H. Van Riemsdijk (eds.) Rightward Movement, 247–278. Philadelphia: John Benjamins. Neidle, Carol, Judy Kegl, Dawn MacLaughlin, Benjamin Bahan, and Robert G. Lee (2000) The Syntax of American Sign Language: Functional Categories and Hierarchical Structure. Cambridge, Mass.: MIT Press. Neidle, Carol, and Dawn MacLaughlin (1998) “SignStream™: A Tool for Linguistic Research on Signed Languages.” Sign Language and Linguistics 1(1), 111–114. Neidle, Carol, Dawn MacLaughlin, Robert G. Lee, Benjamin Bahan, and Judy Kegl (1998a) “The Rightward Analysis of Wh-movement in ASL: A Reply to Petronio and Lillo-Martin 1997.” Language 74, 819–831. Neidle, Carol, Dawn MacLaughlin, Robert G. Lee, Benjamin Bahan, and Judy Kegl (1998b) Wh-Questions in ASL: A Case for Rightward Movement. Report No. 6, American Sign Language Linguistic Research Project. Boston: Boston University. Ouhalla, Jamal (1991) Functional Categories and Parametric Variation. London: Routledge. Padden, Carol A. (1983) “Interaction of Morphology and Syntax in American Sign Language,” Ph.D. Diss., University of California at San Diego. Padden, Carol A. (1988) Interaction of Morphology and Syntax in American Sign Language. New York: Garland. Padden, Carol A. (1990) “The Relation between Space and Grammar in ASL Verb Morphology,” in Ceil Lucas (ed.) Sign Language Research: Theoretical Issues, 118–132. Washington, DC: Gallaudet University Press. Padden, Carol A., and David M. Perlmutter (1987) “American Sign Language and the Architecture of Phonological Theory.” Natural Language and Linguistic Theory 5(3), 335– 375.

224

IP FUNCTIONAL STRUCTURE

Perlmutter, David (1992) “Sonority and Syllable Structure in American Sign Language.” Linguistic Inquiry 23, 407–442. Pesetsky, David (1997) “Optimality Theory and Syntax: Movement and Pronunciation,” in Diana Archangeli and D. Terence Langendoen (eds.) Optimality Theory: An Overview, 134–170. Malden, Mass.: Blackwell. Petronio, Karen (1993) “Clause Structure in American Sign Language,” Ph.D. Diss., University of Washington at Seattle. Petronio, Karen, and Diane Lillo-Martin (1997) “WH-Movement and the Position of SpecCP: Evidence from American Sign Language.” Language 73(1), 18–57. Poizner, Howard, and Judy Kegl (1992) “The Neural Basis of Language and Motor Behavior: Perspectives from American Sign Language.” Aphasiology 6, 219–256. Pollock, Jean-Yves (1989) “Verb Movement, UG, and the Structure of IP.” Linguistic Inquiry 20, 365–424. Pollock, Jean-Yves (1997) “Notes on Clause Structure,” in Liliane Haegeman (ed.) Elements of Grammar, 237–279. Dordrecht: Kluwer. Postal, Paul M. (1969) “On So-Called ‘Pronouns’ in English,” in David A. Reibel and Sanford A. Schane (eds.) Modern Studies in English: Readings in Transformational Grammar, 201–224. Englewood Cliffs, N.J.: Prentice Hall. Reilly, Judy Snitzer, and Ursula Bellugi (1996) “Competition on the Face: Affect and Language in ASL Motherese.” Journal of Child Language 23(1), 219–239. Reilly, Judy Snitzer, Marina L. McIntire, and Ursula Bellugi (1990) “Faces: The Relationship between Language and Affect,” in V. Volterra and C.J. Erting (eds.) From Gesture to Language in Hearing and Deaf Children, 128–141. New York: Springer-Verlag. Rizzi, Luigi (1991) Residual Verb Second and the Wh-Criterion. Technical Reports in Formal and Computational Linguistics. Geneva: Université de Genève. Sánchez, Liliana (1996) “Syntactic Structure in Nominals: A Comparative Study of Spanish and Southern Quechua,” Ph.D. Diss., University of Southern California. Sandler, Wendy (1989) Phonological Representation of the Sign: Linearity and Nonlinearity in American Sign Language. Dordrecht: Foris. Schein, J. D., and M. T. Delk (1974) The Deaf Population of the United States. Silver Spring, Md.: National Association of the Deaf. Stokoe, William C. (1960) “Sign Language Structure: An Outline of the Visual Communication Systems of the American Deaf.” Studies in Linguistics Occasional Papers, vol. 8. Buffalo, N.Y.: University of Buffalo, Department of Anthropology and Linguistics. Szabolcsi, A. (1987) “Functional Categories in the Noun Phrase,” in I. Kenesei (ed.) Approaches to Hungarian: Vol. 2. Theories and Analyses, 167–189. Szeged: JATE. Tervoort, B. T. (1968) “You Me downtown Movie Fun?” Lingua 21, 455–465. Veinberg, Silvana C., and Ronnie B. Wilbur (1990) “A Linguistic Analysis of Negative Headshake in American Sign Language.” Sign Language Studies 68, 217–244. Woodward, J. (1976) “Deaf Awareness.” Sign Language Studies 2, 57–60.

Subject Index

Adjective, 91, 114 attributive, 91 manner, 21, 23–24, 28 thematic, 21, 23–24, 46nn. 9, 10 Adjunct PP, 94 Adverb aspectual, 8 AdvP, 95 again, 217 Agreement, 179, 180, 182, 191, 207–215 alternations, 177, 181, 187 anti-, 181 features, 167, 183 in DP, 210–212 in TP, 212–215 past participle, 179, 191 subject, 176, 178 American Sign Language (ASL) adverbials in, 216–217, 219 aspect, 206, 216–217, 219 definite determiner, 208, 219n. 15 DP, 210–212 glossing, 199 modals, 207 morphology, 198, 208–209, 216–217 negation, 202–203 225

nonmanual markings, 201–204 null arguments, 214–215 phonology, 186–197 reflexive, 208 sociolinguistic issues, 199–201 spatial representation in, 208–210 tags, 219n. 11 tense, 205–207 wh-questions, 203–204 word order, 201 Antisymmetric Hypothesis, 17, 42, 45n. 4 Article, 57–70, 85n. 25 adjectival, 80–82, 84n. 19 definite,18–19, 43, 47n. 13 indefinite, 18–19 Augmentative, 97 Bridge verbs, 125, 126, 128, 131, 133 British Sign Language, 196 Checking Theory, 5 Classifier, 97, 98 Clitic article, 58–59 possessive, 83n. 4, 85–86nn. 26, 27, 28 splitting, 135 pronouns, 60

226

SUBJECT INDEX

Comma intonation, 92 continuously, 216 CP-recursion, 125, 126, 131, 151

Infinitive root, 188, 192 I-to-C movement, 184

Deictic feature, 16, 25, 30, 34, 37, 42, 43, 50n. 27 Demonstrative, 15–51, 61, 70–74, 75, 83n. 10 and locative, 25–27, 33, 34, 35, 36, 38, 40, 42, 43, 46n. 12, as XP, 16–17, 22, 42 base position of, 24–25, 33, 36, 39, 41, 42 intermediate position of, 36, 39, 51n. 34 movement of, 15, 17, 18, 19, 22, 27, 30– 32, 34, 36, 38, 39, 41, 42, 43, 47n. 13, 50n. 27 movement before Spell-out, 15, 16, 17, 18, 27, 30–32, 33, 34, 36, 37, 38, 39, 41, 43, 44 movement at LF, 16, 19, 22, 27, 31, 34, 36, 41, 42, 43, 44, 47n. 13 prenominal, 15, 16, 18, 30, 33, 34, 35, 43, 45nn. 4, 5, 50n. 32 postnominal, 15, 16, 17, 18, 20–23, 27– 29, 33, 34, 35, 36, 38, 39–41, 42, 43, 44nn. 1, 3, 45n. 4, 50n. 32 Determiner, 56–57, 58, 73–74, 83n. 10 Diminutive, 97–98 Direct modification hierarchical, 92 parallel, 92

Locative as deictic reinforcer, 25–27, 33,

Empty Category Principle, 5 Event Nominal, 20–21, 22–23, 24, 25, 28, 33, 34, 37, 47n. 14 Expletive, 178 Extended projection, 95 French Sign Language (LSF), 196 Focus, 113 Functional categories, 3–9, 55–57, 58, 67– 70, 77–78, 81, 91–118, 168, 182, 204– 218 verbs, 187, 189 Genitive PPs, 23, 25, 42, 44n. 3 Head Movement Constraint, 5

34, 35, 36, 37, 38, 40, 42, 43, 45n. 4, 46–47nn. 13, 14, 47n. 14, 48n. 17, 51nn. 36, 38 Minimality relativized, 183, 189 Mirror Principle, 4, 8, 9, 58, 220n. 24 Morphology-Syntax Interface, 175, 178 Multiple exponence, 5 Narrative V1, 136 Negation, 181–184, 191, 202–203 Nominative case, 124, 126, 131, 133, 141– 146, 147, 148, 150, 153, 154, 155, 157, 158, 160n.5, 161n.12, 162n. 12, 179, 185 Nominals, internal structure, 20–21, 24–25, 42 Noun Phrase, 54–88 Noun movement, 17, 19, 20, 25, 30, 42, 51n. 36 Null subjects, 128, 129, 146, 157, 159 expletive, 146–151, 152 referential, 151–154 Object denoting nominals, 21–22, 23, 25, 33–34, 37, 49n. 20 Partitive case, 131, 147, 148, 162n.15 Principle of Economy of Lexical Insertion, 55, 70 Principle of Optimal Realization, 69–70 Pronoun personal, 75–76 Possessive adjective, determiner, pronoun, 74–75, 85nn. 26, 27 marker, 208 prenominal, 24, 27–28, 46nn.6, 10, 47n. 13, 49nn. 25, 26 postnominal, 16, 27–29, 34, 35, 42, 45n. 4, 49nn. 20, 21 PP adjunct, 14 Proper name, 57, 76–78

SUBJECT INDEX

Questions negative, 183, 189 Referential feature, 16, 19, 30, 31, 32, 34, 36, 38, 41, 42, 43, 50nn. 26, 27 Relational grammar, 8 Restructuring, 6 Satellite Principle, 8 Scrambling, 135, 161n. 9 SignStream, 199 Spec-head coindexation, 142, 143, 147, 150, 151, 155 Split-Infl, 167 Stylistic fronting, 124, 129, 130, 132, 133, 152, 153, 154, 155–157 Stylistic inversion, 149, 150, 152 Suppletion, 5 Syncretism, 5

227

Tense features, 167 TP dominating AgrSP, 215 Truncation, 185, 186 Tobler-Mussafia law, 123, 124, 135–139, 140, 153, 154, 162n. 13 Topicalization embedded topicalization, 130, 131, 132, 133, 153, 154, 155–157, 161n. 11 V1 declaratives, 127, 137, 151, 152, 160n. 5, 161n. 11 Voice head, 191 Wackernagel’s law, 123, 124, 133–135, 139, 142, 154, 161n. 9

228

LANGUAGE INDEX

Language Index

Albanian, 39, 42, 43, 58–59, 61, 69, 80–82, 84 nn. 12, 19, 86n. 30 American Sign Language (ASL), 8, 195–224 Amharic, 111 Anconetano, 176–177 Arabic Lebanese, 177 Moroccan, 177 Standard, 177, 178 Bantu, 8 Bavarian, 161n. 7 Berber, 181 Bernese Swiss German, 161n. 9 Bosnian, 26, 34–35, 36, 42, 43, 50n. 28, 51n. 34 Brazilian Portuguese, 135 Breton, 181 British Sign Language, 196 Bulgarian, 46n. 8, 58–59, 69, 80–81, 87nn. 41, 43 Catalan, 28, 29, 33–34, 39, 42, 43, 49n. 25 Chichewa, 117 Chinese, 117, 220n. 22 228

Danish. See Scandinavian Dutch, 125, 135, 144, 145, 146, 154 English, 5, 8, 22, 26, 35, 92–94, 98–113, 115, 116, 118, 125, 126, 131, 145, 146, 149, 154, 157, 158, 167, 175, 176, 177, 182, 196, 206, 207 Belfast, 179 Early (child), 167–192 Middle English, 130 Old English, 139, 140, 143, 144, 145, 154 European Portuguese, 135, 158 Faroese, 129 Finnish, 99, 101, 117 Fiorentino, 176 French, 4, 8, 38, 39, 41, 43, 51n. 36, 73, 86n. 32, 111, 118, 137, 138, 145, 146, 149, 150, 152, 153, 157, 160n. 6, 175, 176, 177, 178, 179, 180, 181, 191, 207 Old, 124, 127–130, 135, 136, 137, 142, 143, 144, 145, 146, 151, 152, 153, 154, 155, 156, 161n. 11 French Sign Language, 196

LANGUAGE INDEX

Gallo-Italian dialects, 161n. 7 German, 38–39, 42, 43, 60–61, 65, 74, 100, 125, 126, 128, 133–135, 140, 142, 143, 144, 145, 146, 147, 150, 153, 154, 159, 160n. 3, 161nn. 8, 12, 162n. 12, 179 Old High German, 139, 140, 145, 154 Greek Modern, 5, 36, 39, 42, 43, 51n. 34, 57, 61 Hausa, 97 Hebrew, 38–40, 41, 42, 43, 51n. 41, 116 Hungarian, 105 Ibibio, 99, 100 Igbo, 97 Icelandic, 124, 125–127, 129, 130, 133, 137, 142, 144, 145, 146, 147, 148, 149, 150, 154, 155, 156, 158, 159n. 2, 161nn. 8, 11 Irish, 40–41, 42, 43, 44n. 3, 51n. 42, 115, 159, 160n. 15 Italian, 4, 5, 22, 28, 37–38, 40, 41, 43, 47n.13, 48n. 14, 60, 62, 63–64, 74, 76–80, 85nn. 20, 21, 22, 87n. 40, 138, 145, 146, 147, 153, 158, 159, 175, 176, 177, 178, 179, 180, 181, 191 Old Italian, 135, 136, 137 Dialects of, 48n. 15

229

Malayalam, 99, 100, 101, 117 Malek, 97 Norwegian, 26, 35 (see also Scandinavian) Polish, 117 Portuguese Brazilian, 178 European, 98, 179 Potawatomi, 98 Romance, 20, 45n. 4 Romanian, 16–17, 18, 19, 32, 35–36, 39, 42, 43, 50n. 32, 57, 58–59, 62, 64, 65, 69–70, 80–82, 84n. 12, 85nn. 21, 24, 87n. 35, 111, 139 Russian, 39, 42, 43, 50n. 31, 98 Scandinavian, 26, 58–59, 75, 66–69, 83n. 11, 84nn. 13, 17, 86n. 29, 175, 191 Serbo-Croatian, 100, 101, 111 Spanish, 15–51, 71, 86–87n. 33, 111, 117, 220n. 27 Old Spanish, 135 Swedish, 100 (see also Scandinavian) Trentino, 176

Japanese, 4, 111, 115, 117

Wambaya, 9 Welsh, 100, 101, 115, 118 West Flemish, 158, 161n. 7

Lakhota, 115 Lombard dialects of Switzerland, 51n. 37

Yiddish, 124, 127, 129, 130–133, 142, 145, 148, 149, 155, 156, 158, 161n. 11

230

NAME INDEX

Name Index

Aarons, D., 204, 205, 219 Abney, S., 94, 104, 208, 220 Adams, M., 128, 151 Alberton, S., 127, 136, 137 Alexiadou, A., 159 Alsina, A., 9 Anagnostopoulou E., 159 Anderson, D., 219 Andrews, A., 113 Androutsopoulou, A., 93, 94, 114, 115 Aoun, J., 177 Asher, A., 117 Bahan, B., 195, 200, 204, 205, 210, 217, 218, 219, 220 Ball, M., 51n. 42 Baker, C., 202, 219 Baker, M., 4, 5, 8, 44n. 2, 205, 220 Bayer, J., 161n. 7 Belletti, A., 3, 5, 8, 102, 123, 131, 148, 176, 180, 182, 194, 205 Bellugi, U., 201, 202, 216, 219, 220 Beninca’, P., 6, 9, 127, 135, 137, 151 Benmamoun, E., 177 230

Bernstein, J., 45n. 4, 72, 73, 84, 86 Bittner, M., 220 Bloom, L., 169 Bobaljik, J. D., 158, 218 Borer, H., 159 Börjars, K., 83 Boschetti, L., 134 Bouchard, D., 218 Brandi, L., 176 Brentari, D., 220 Brian, J., 5 Brody, M., 8, 9, 218 Brown, R., 169 Brucart, J., 49n. 25 Brugè, L., 7, 26, 48n. 20, 49n. 23, 50nn. 26, 27, 31, 33, 57, 61, 67, 71, 84, 86 Brugger, G., 49n. 23 Bybee, J., 5 Bühler, K., 50n. 27 Campbell, R., 22, 56, 82, 205 Cardinaletti, A., 7, 32, 49n. 24, 58, 76, 78, 83, 84, 86, 87, 142, 146, 159, 162n. 11, 162n. 14, 176, 178, 179, 184 Chierchia, G., 83

NAME INDEX

Chomsky, N., 20, 30, 41, 43, 54, 58, 83, 84, 97, 102, 104, 116, 118, 123, 124, 138, 155, 167, 168, 178, 179, 182, 189, 191, 205, 213, 218 Cinque, G., 6, 8, 9, 17, 20, 21, 23, 24, 25, 29, 42, 66, 71, 83, 84, 91, 95, 96, 97, 103, 105, 112, 114, 115, 116, 117, 118, 167, 180, 182, 191, 192, 207, 216 Crain, S., 219 Cokely, D., 219 Cook, V., 94 Cordin, P., 176 Cornilescu, A., 84 Coulter, G. R., 218 Crisma, P., 115 Delk, M. T., 200 Delsing, L.-O., 75, 83, 84 de Kok, A., 136, 153 den Besten, H., 127, 137 Diesing, M., 127, 130, 131, 132, 149, 158 Dik, S., 5 Dimitrova-Vulchanova, M., 18–19, 32, 46n. 9, 57, 58, 66, 70, 80, 81, 83, 84, 88 Dixon, R., 93, 97, 105, 114, 116, 118 Di Sciullo, A.-M., 6 Dobrovie-Sorin, C., 58 Dubuisson, C., 218 Dupuis, F., 128, 129, 151, 156 Emonds, J., 192 Ernst, T., 44n. 3, 51n. 42 Fisher, S., 219 Fodor, J., 117 Foley, W., 8 Fornaciari, R., 181 Foulet, L., 151, 153, 154 Fowler, W. S., 101 Freidin, R., 94 Friedemann, A., 180 Friedman, L., 220 Galves, C., 135 Gerdts, D., 8 Giorgi, A., 179, 185, 188 Giusti, G., 7, 16, 18–19, 22, 26, 32, 36, 46n. 9, 49n. 24, 57, 60, 66, 70, 71, 80, 81, 83, 84, 85, 88, 95, 124

231

Greenfield, D., 199 Grewendorf, G., 159 Grimshaw, J., 20, 54, 84, 95, 112, 115, 118 Groat, E., 218 Grosu, A., 58, 80 Guasti, M. T., 8, 168 Haegeman, L., 158, 161n. 7 Haider, H., 159 Hale, K., 220 Halpern, A., 158 Harris, M., 136 Hellan, L., 58 Henry, A., 179, 182 Hetzron, R., 92, 93, 98, 105, 111, 114, 116, 117 Higginbotham, J., 87, 115 Hinton, M., 99 Hirschbühler, P., 129, 136, 151, 152, 153 Hoekstra, E., 115 Hoffmeister, R. J., 200, 218 Hornstein, N., 159 Hoza, J., 218 Huang, J., 220 Iatridou, S., 8 Ingham, R., 186, 192 Jackendoff, R., 105, 116, 118 Janda, R., 6 Jaspers, D., 135 Jonas, D., 158 Junker, M.-O., 151 Kallulli, D., 51n. 39 Kayne, R., 4, 9, 17, 27, 42, 48nn. 16, 18, 54, 96, 115, 137, 138, 149, 162n. 13, 177, 180, 181, 182, 191, 204 Kegl, J., 197, 201, 204, 205, 218, 219 Kester, E.-P., 83 Kinsbury, R., 98–99, 101 Kiparsky, P., 140 Klima, E., 216, 220 Koopman, H., 124, 126, 141, 142, 143, 155 Kripke, S., 49n. 22 Kumari, T., 117 Labov, W., 191 Laenzlinger, C., 182 Lamarche, J., 84

232

NAME INDEX

Lane, S., 200, 218 Lapesa, R., 49n. 25 LaPolla, R., 115 Larson, R., 46n. 7, 48n. 17, 83 Lasnik, H., 58 Lee, R., 195, 217, 219, 220 Leko, N., 50n. 28, 57 Levi, J., 92, 115 Liberman, M., 118 Liddell, S., 219 Lieber, R., 116 Lillo-Martin, D., 204, 219, 220 Lombard, A., 87 Longobardi, G., 18, 24, 28, 32, 42, 46n. 6, 49nn. 21, 22, 51n. 43, 62, 76, 77 Lowenstamm, J., 132 MacLaughlin, D., 8, 195, 199, 204, 210, 217, 218, 219, 220 MacWhinney, B., 169 Madeira, A., 158 Maling, J., 129, 155 Manzini, M. R., 157, 158, 159 Marsden, R., 99 McCloskey, J., 159, 160n. 5 McIntire, M., 201, 219 Meier, R.P., 220 Mitchell, E., 4 Miyagawa, S., 5 Motapanyane, V., 139 Munaro, N., 4 Mussafia, A., 136 Napoli, D., 118 Neidle, C., 8, 195, 199, 202, 204, 205, 212, 213, 217, 219 Newson, M., 94 Nordlinger, R., 9 Noyer, R., 6, 9 O’Neil, J., 218 Ouhalla, J., 3, 5, 111, 205, 220 Padden, C. A., 208, 218, 219, 220 Passaler, E., 60 Penner, Z., 161 n.10 Perlmutter, D., 197, 218 Pesetsky, D., 218 Petronio, K., 204, 219

Phillips, C., 168, 181 Pianesi, F., 179, 185, 188 Picallo, C., 28, 29, 46nn. 6, 8, 86 Platzack, C., 130, 156 Poizner, H., 197, 201 Poletto, C., 4, 9, 48n. 15, 51n. 37, 157, 158 Pollock, J.-Y., 4, 5, 8, 124, 140, 149, 157, 167, 168, 182, 204 Postal, P., 75, 208 Potter, B., 6, 9 Price, G. , 151 Pulkina, I., 115 Pustejovsky, J., 118 Radford, A., 115 Raposo, E., 179 Reilly, J. S., 201, 219 Renzi, L., 151 Ritter, E., 116 Rizzi, L., 6, 8, 22, 83, 102, 118, 125, 126, 129, 131, 138, 142, 144, 145, 146, 155, 156, 158, 162n. 13, 168, 179, 182, 185, 188, 219 Roberts, I., 7, 125, 126, 129, 131, 137, 142, 145, 146, 151, 153, 161n. 7, 162n. 13, 175, 192 Rögnvaldsson, E., 125, 126, 148, 149 Roussou, A., 159 Sáez, L., 47n. 14 Sánchez, L., 220 Sandler, W., 218 Santelmann, L., 83 Santorini, B., 127, 130, 131, 132, 149, 156, 158, 159n. 2, 161n. 11 Savoia, L., 157, 158, 159 Schein, J. D., 200 Schuetze, C., 179, 185, 192 Schulze, R., 128 Schwartz, B.D., 131 Scott, G.-J., 7, 9, 21, 92, 111, 114, 115, 118 Shih, C., 91, 92, 93, 94, 97, 114, 115, 117 Shlonsky, U., 51n. 40, 157, 158 Shomswold, K., 168, 184 Sigurðsson, H., 127, 130 Siloni, T., 28, 180 Snow, C., 169 Speas, M., 9

NAME INDEX

Sportiche, D., 33, 124, 126, 141, 142, 143, 155, 177 Sproat, R., 91, 92, 93, 94, 97, 114, 115, 117, 118 Starke, M., 58, 158, 159 Stokoe, W. C., 219 Suppes, P., 169 Svenonius, P., 83 Szabolcsi, A., 220 Tallerman, M., 115 Taraldsen, T., 58, 66 Tasmowski-De Ryck, L., 50n. 33, 72 Tervoort, B. T., 218 Thráinsson, H., 125, 126, 149, 155, 159n. 2 Thurneysen, R., 151 Tomaselli, A., 139, 140 Travis, L., 5, 44n. 2, 144 Valois, D., 96, 111, 112 Vance, B., 130, 151

233

Vanelli, L., 151 van Kemenade, A., 139, 140 van Riemsdijk, H., 159 van Valin, R., 8, 115 Veinberg, S. C., 219 Vikner, S., 126, 130, 131, 132, 133, 148, 155, 156, 160n. 2, 162n. 15, 188, 192 von Wartburg, W., 151 Wackernagel, J., 133 Wellman, G., 98–99, 101 Wexler, K., 185, 188, 192 Wilbur, R. B., 219 Williams, E., 6 Woodword, J., 218 Yadroff, M., 50n. 31 Zanuttini, R., 95, 191 Zwart, C. J.-W., 135, 144 Zwarts, J., 91, 115 Zwicky, L., 158