Noun Phrase in the Generative Perspective

≥
Studies in Generative Grammar 71

Editors
Jan Koster
Harry van der Hulst
Henk van Riemsdijk

Mouton de Gruyter
Berlin · New York
Noun Phrase in the
Generative Perspective

by
Artemis Alexiadou
Liliane Haegeman
Melita Stavrou

Mouton de Gruyter
Berlin · New York
Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

The series Studies in Generative Grammar was formerly published by

Foris Publications Holland.

앪
앝 Printed on acid-free paper which falls within the guidelines
of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data

Alexiadou, Artemis.
Noun phrase in the generative perspective / by Artemis Alexiadou,
Liliane Haegeman, Melita Stavrou.
p. cm. ⫺ (Studies in generative grammar ; 71)
Includes bibliographical references and index.
ISBN 978-3-11-017684-1 (cloth : alk. paper) ⫺
ISBN 978-3-11-017685-8 (pbk. : alk. paper)
1. Grammar, Comparative and general ⫺ Noun phrase. 2. Gram-
mar, Comparative and general ⫺ Syntax. I. Haegeman, Liliane
M. V. II. Stavrou, Melita. III. Title.
P271.A443 2007
415⫺dc22
2007015283

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

ISBN 978-3-11-017684-1 hb
ISBN 978-3-11-017685.8 pb
ISSN 0167-4331

쑔 Copyright 2007 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin.
All rights reserved, including those of translation into foreign languages. No part of this
book may be reproduced in any form or by any means, electronic or mechanical, including
photocopy, recording, or any information storage and retrieval system, without permission
in writing from the publisher.
Cover design: Christopher Schneider, Berlin.
Printed in Germany.
In memory of Tanya Reinhart
Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Aim of the book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Syntax of nominal projections and syntax of clauses . . . . . . . . . . . . . . . . . xvii
Organization of the book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xx
Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi

Part I
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1. Some parallelisms between clauses and nominal projections 2
1.1. Subjects and genitives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2. Functional structure: the DP hypothesis . . . . . . . . . . . . . . . . . . . . . . 3
1.3. Survey of this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2. The theoretical framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1. Levels of representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2. Syntactic structure: the X-bar format . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3. Lexical categories and functional categories . . . . . . . . . . . . . . . . . . 12
2.4. Lexical categories and argument structure . . . . . . . . . . . . . . . . . . . . 17
2.4.1. Verbs and arguments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.2. Decomposing the VP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4.2.1. Ditransitive verbs and binary branching . . . . . . . . . . . . . . . . . . . . . . 19
2.4.2.2. Extending the proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4.3. Nouns and arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.5. Functional projections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5.1. Evidence for functional projections . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5.2. Functional projections, movement and agreement . . . . . . . . . . . . . 31
2.5.2.1. Features and agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5.2.2. Types of movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.5.2.3. Features and movement in the nominal projection. . . . . . . . . . . . . 35
2.5.3. Challenging functional projections . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.5.3.1. AgrP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.5.3.2. Multiple specifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
viii Contents

2.6. Deriving variations in linear order . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

2.6.1. Cross-linguistic variation in linear order . . . . . . . . . . . . . . . . . . . . . 38
2.6.2. Antisymmetry and linear order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.6.2.1. Antisymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.6.2.2. Deriving OV-orders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.6.3.3. Snowballing in the nominal projection? . . . . . . . . . . . . . . . . . . . . . 49
3. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

Part II
The functional make up of the Noun Phrase . . . . . . . . . . . . . . . . . . . . . . . 51

Chapter 1
The emergence and the structure of DP. Articles and demonstratives 53
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2. The role(s) of the article. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.1. “Definiteness”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.2. The article as a subordinator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.3. The article and the concept of referentiality . . . . . . . . . . . . . . . . . . 66
2.4. Summary: the functions of the definite article . . . . . . . . . . . . . . . . 70
2.5. The article as a grammatical morpheme . . . . . . . . . . . . . . . . . . . . . . 71
3. The DP hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.1. Motivating ‘DP’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.1.1. Phrasal movement inside the nominal projection . . . . . . . . . . . . . . 79
3.1.2. PRO and the nominal projection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.1.3. Head movement inside the nominal projection . . . . . . . . . . . . . . . . 85
3.1.4. DP and the concept of definiteness . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2. Some challenges for the DP-Hypothesis. . . . . . . . . . . . . . . . . . . . . . 89
3.2.1. N incorporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.2.2. D incorporation to N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.2.3. Agreement in the extended projection of N . . . . . . . . . . . . . . . . . . . 92
4. Determiners, demonstratives and DP . . . . . . . . . . . . . . . . . . . . . . 93
4.1. The interpretation of demonstratives . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.1.1. A note on the historical connection between demonstratives
and the definite article. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.1.2. The deictic interpretation of demonstratives . . . . . . . . . . . . . . . . . . 98
4.1.3. The anaphoric use of demonstratives . . . . . . . . . . . . . . . . . . . . . . . 104
4.2. The syntactic representation of demonstratives . . . . . . . . . . . . . . 105
4.2.1. Demonstratives as maximal projections . . . . . . . . . . . . . . . . . . . . . 105
 Contents ix

4.2.2. Articles in D; demonstratives in SpecDP . . . . . . . . . . . . . . . . . . . . 108

4.2.3. A lower position for demonstratives . . . . . . . . . . . . . . . . . . . . . . . . 109
4.2.4. An alternative proposal: head movement and demonstratives . 116
4.2.5. Reinforcers as empirical evidence for a lower position . . . . . . . . 117
4.3. Interpreting the positions of the demonstrative in the DP.
The case of Greek . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.4. Splitting the DP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5. DP and CP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.1. DP as parallel to CP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.2. DP as a VP-like category: DP-shells . . . . . . . . . . . . . . . . . . . . . . . . 137
5.3. Topic Phrase and Focus Phrase in the DP: some proposals . . . . 140
6. The category D and Case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
7. Summary of the chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

Chapter 2
Determinerless Noun Phrases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
1. Introduction: D across languages . . . . . . . . . . . . . . . . . . . . . . . . . 159
1.1. Languages without (definite) articles . . . . . . . . . . . . . . . . . . . . . . . 159
1.2. The distribution of the definite article in languages that
have one . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
2. The functional head D and the configurationality of
noun phrases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
2.1. Setting the scene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
2.2. Gil 1987, Loebel 1989 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
2.3. Bare Plurals, Mass nouns, Proper nouns and Generic nouns . . . 172
2.3.1. The interpretation of bare plurals in English . . . . . . . . . . . . . . . . . 172
2.3.2. Mass Nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
2.3.3. Proper names. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
2.3.4. A note on the expletive article . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
2.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
3. Noun phrases as arguments and as predicates . . . . . . . . . . . . 190
3.1. A typological parameter: Chierchia (1998b) . . . . . . . . . . . . . . . . . 190
3.1.1. General presentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
3.1.2. The Nominal Mapping Parameter . . . . . . . . . . . . . . . . . . . . . . . . . . 191
3.1.3. Some applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
3.1.4. Chinese NPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
3.1.5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
3.2. Longobardi and N-to-D-Movement. . . . . . . . . . . . . . . . . . . . . . . . . 203
3.2.1. Bare Ns in Italian and null D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
x Contents

3.2.2. Proper Nouns in Italian and N to D movement . . . . . . . . . . . . . . . 206

3.2.3. Proper names in English and parametric variation . . . . . . . . . . . . 211
3.2.4. Bare common nouns in English . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
3.2.4.1. Bare indefinites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
3.2.4.2. Bare generics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
3.2.4.3. The role of morphological marking . . . . . . . . . . . . . . . . . . . . . . . . . 216
3.2.4.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
3.3. The syntax of determinerless languages . . . . . . . . . . . . . . . . . . . . . 218
3.3.1. Evidence for an empty D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
3.3.2. Extraction and determinerless languages . . . . . . . . . . . . . . . . . . . . 220
4. Further developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

Chapter 3
DP-internal functional projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
1. Introductory remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
2. Number and NumP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
3. Gender, Word Marker and ‘Gender Phrase’. . . . . . . . . . . . . . 235
3.1. What determines Gender? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
3.2. Gender in the syntax? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
3.3. Summary: Gender and syntax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
4. Distributional evidence for functional projections . . . . . . . . . 246
4.1. Noun Movement I: The construct state . . . . . . . . . . . . . . . . . . . . . 247
4.2. Noun Movement II: N-to-D movement . . . . . . . . . . . . . . . . . . . . . 253
4.3. How morphology comes into play . . . . . . . . . . . . . . . . . . . . . . . . . . 256
4.3.1. Nominal agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
4.3.2. Gender as trigger for N-movement (Bernstein 1993) . . . . . . . . . 258
4.4. The Germanic languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
4.4.1. German . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
4.4.2. West Flemish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
4.5. Greek. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
4.6. Conclusion: Terminal vowels, N-movement and Ellipsis . . . . . 272
5. Speculations on other functional categories in the DP . . . . . 273
5.1. Voice and Aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
5.2. Tense .......................................... 275
5.2.1. Semantic considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
5.2.2. Morphological considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
5.2.3. Nominal tense and possessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
6. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
 Contents xi

Part III
Modification relations inside the DP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Chapter 1
Adjectives in the DP. Problems of distribution and interpretation . 284
1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
1.1. Scope and organization of the chapter . . . . . . . . . . . . . . . . . . . . . . 284
1.2. Setting the scene: Cross-linguistic asymmetries . . . . . . . . . . . . . . 286
2. DP-internal adjectives in English . . . . . . . . . . . . . . . . . . . . . . . . . 289
2.1. The problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
2.2. The attributive-predicative dichotomy . . . . . . . . . . . . . . . . . . . . . . 290
2.3. The reductionist view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
2.3.1. Introducing the reductionist hypothesis . . . . . . . . . . . . . . . . . . . . . 292
2.3.2. Some complications for the reductionist view . . . . . . . . . . . . . . . 295
2.4. Interpretive contrasts between prenominal and postnominal
adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
2.5. One more dichotomy: intensional-extensional, or
intersective-non-intersective adjectives . . . . . . . . . . . . . . . . . . . . . 299
2.6. Non-intersective adjectives and deverbal nouns . . . . . . . . . . . . . . 302
2.7. Semantic classes and syntactic positions . . . . . . . . . . . . . . . . . . . . 305
2.8. Evidence from other languages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
3. Sequencing of adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
3.1. Hierarchical orders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
3.2. Absolute vs. non-absolute [or objective vs. subjective]
adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
3.3. Classifying adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
3.4. Patterns of adjectival modification. . . . . . . . . . . . . . . . . . . . . . . . . . 320
3.4.1. Direct vs. indirect modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
3.4.2. Parallel vs. hierarchical modification . . . . . . . . . . . . . . . . . . . . . . . 322
3.5. Hierarchical orders involving a finer subclassification
of adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
4. On the syntax of DP-internal adjectives. . . . . . . . . . . . . . . . . . . 326
4.1. General remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
4.2. Post and pre-nominal adjective meaning contrasts in the
Romance languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
4.2.1. The distinction restrictive – non-restrictive adjectives . . . . . . . . . 334
4.2.2. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
4.3. N-movement as a means of deriving the postnominal
position of the adjective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
xii Contents

4.4. Problems for the N-movement hypothesis . . . . . . . . . . . . . . . . . . . 340

4.5. The adjective-as-head hypothesis. . . . . . . . . . . . . . . . . . . . . . . . . . . 348
5. The clausal hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
5.1. The determiner complementation, or Relative Clause
Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
5.1.1. Kayne 1994 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
5.2. N- and D-shells 358
5.3. Demonte 1999 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
6. More on DP-internal phrasal movement . . . . . . . . . . . . . . . . . . 364
6.1. D-complementation and Greek polydefinite DPs . . . . . . . . . . . . . 364
6.1.1. Interpretive differences between a monadic DP and DS 366
6.1.2. Deriving the DS effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
6.2. Greek DS and adjectives in Romance . . . . . . . . . . . . . . . . . . . . . . . 373
7. Snowballing movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
7.1. Noun-initial DPs and the directionality parameter . . . . . . . . . . . . 377
7.2. Antisymmetry and mirror image word order in the DP. . . . . . . . 379
8. ‘Separationism’. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
9. Tying together the lines of inquiry . . . . . . . . . . . . . . . . . . . . . . . . 387
9.1. A mixed approach to pre-and postnominal adjectives . . . . . . . . 388
9.2. A note on classifying adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
10. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393

Chapter 2
Semi-functional categories: The N-of-N Construction and
the Pseudo-Partitive construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
1. Introduction: the data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
2. The Pseudopartitive Construction: cross-linguistic variation 400
2.1. Basic properties of the construction. . . . . . . . . . . . . . . . . . . . . . . . . 402
2.1.1. N1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
2.1.2. N2 and the relationship between N1 and N2 . . . . . . . . . . . . . . . . . 403
2.1.3. The relational character of N1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405
2.2. An excursus to the older accounts of PsP . . . . . . . . . . . . . . . . . . . 406
2.3. Juxtaposed pseudopartitives and the Monoprojectional
analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
2.3.1. Measure/Classifier phrases are like simplex Qs . . . . . . . . . . . . . . 410
2.3.2. The PsP is a unitary nominal projection with a single
referent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
2.3.2.1. Determiners, quantifiers and relative clauses
in the PsP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
 Contents xiii

2.3.2.2. Adjectival modification in the PsP . . . . . . . . . . . . . . . . . . . . . . . . . 417

2.3.2.3. Selection in the PsP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
2.3.2.4. Number agreement in the PsP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
2.4. The syntax of the juxtaposed PsP . . . . . . . . . . . . . . . . . . . . . . . . . . 425
2.4.1. PsP is not a DP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425
2.4.2. Properties of the PsP construction . . . . . . . . . . . . . . . . . . . . . . . . . . 431
2.5. N1: head or phrase? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
3. The Predicational approach to the PsP . . . . . . . . . . . . . . . . . . . 436
3.1. Evidence for the predicative relationship between N1 and N2 . 436
3.2. Similarities with the NoN construction . . . . . . . . . . . . . . . . . . . . . . 438
4. The NoN construction: the category PredP in the
noun phrase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
4.1. The Predicational analysis: a sketch . . . . . . . . . . . . . . . . . . . . . . . . 439
4.2. A-Properties of Predicate Inversion. . . . . . . . . . . . . . . . . . . . . . . . . 441
4.2.1. Raising . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
4.2.2. Extraction and quantification restrictions. . . . . . . . . . . . . . . . . . . . 442
4.2.3. The Spell-Out of the copula. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
4.3. The NoN construction as an inverted predicative construction . 444
4.3.1. Extraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
4.3.2. Predicate Inversion and the Spell-Out of the copula . . . . . . . . . . 445
4.3.3. ‘Nominal copulas’ and cross-linguistic differences in DP . . . . . 449
4.4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
5. The Predicate Inversion analysis of the English-type PsP . . 450
5.1. The basic proposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
5.1.1. The bar level of N1 in the non-juxtaposed PsP. . . . . . . . . . . . . . . 452
5.1.2. The Spell-Out of linking morphemes and the juxtaposed PsP . 455
5.2. A Predicational analysis of juxtaposed PsP. . . . . . . . . . . . . . . . . . 458
6. Comparing the Monoprojectional and the Predicational
approaches to PsP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
6.1. Summary of the discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
6.2. The container reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
6.2.1. The ambiguity of the PsP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
6.2.2. The container reading and the Predicational approach . . . . . . . . 462
6.2.3. The container reading and the Monoprojectional approach . . . . 465
6.2.4. Fully lexical nouns and DP projections . . . . . . . . . . . . . . . . . . . . . 467
7. Summary and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
xiv Contents

Part IV
DP-Internal Arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
Chapter 1
Argument Structure in Nominals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
2. Lexical vs. syntactic approaches to deverbal nouns . . . . . . . . 479
2.1. Williams (1981) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
2.2. Giorgi & Longobardi (1991) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 486
2.3. Thematic inheritance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
2.4. Distributed Morphology and Nominalization . . . . . . . . . . . . . . . . 491
3. An ambiguity in the nominal system: Complex Event
Nominals versus Result Nominals . . . . . . . . . . . . . . . . . . . . . . . . 495
3.1. Complex event nominals vs. result nominals . . . . . . . . . . . . . . . . 497
3.2. Simple event nouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
3.3. Grimshaw's analysis of deverbal nouns . . . . . . . . . . . . . . . . . . . . . 503
3.4. Syntactic approaches to complex event nominals and event
structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
3.4.1. Arguments for the presence of a VP in deverbal nouns . . . . . . . 507
3.4.2. Syntactic representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
3.4.2.1. The [DP [NP [VP]]] approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
3.4.2.2. van Hout & Roeper (1998) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
3.4.2.3. Items lacking category specification as input
to nominalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
4. Case assignment in derived nominals . . . . . . . . . . . . . . . . . . . . . 541
5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546

Chapter 2
Possessors and Genitives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
1. Introduction: aim and organization of this chapter . . . . . . . . 547
2. Possessor genitives and thematic complements . . . . . . . . . . . 551
2.1. Alienable and inalienable possession . . . . . . . . . . . . . . . . . . . . . . . 551
2.2. Possessors: complements or subjects? . . . . . . . . . . . . . . . . . . . . . . 552
2.2.1. Possessors as complements? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 552
2.2.2. Possessors as subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 556
3. The base position of the possessor . . . . . . . . . . . . . . . . . . . . . . . . 560
3.1. Light v and light n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
3.2. Two alternative proposals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
 Contents xv

4. Derived positions for possessors in the DP . . . . . . . . . . . . . . . . 565

4.1. Lexical possessors and DP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
4.2. Pronominal possessors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
4.2.1. Cross-linguistic variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
4.2.2. The typology of possessive pronouns . . . . . . . . . . . . . . . . . . . . . . . 568
4.3. DP-internal positions for possessors. . . . . . . . . . . . . . . . . . . . . . . . 570
5. Possessor extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
5.1. Possessors in SpecDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
5.2. Genitive extraction in Greek and the periphery of DP . . . . . . . . 580
5.3. Possessor extraction and the thematic hierarchy: comparative
data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583
5.4. Individual denotation vs. property denotation and extraction
from DP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
6. The possessor doubling construction. . . . . . . . . . . . . . . . . . . . . . 593
6.1. The data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
6.2. Possessor doubling and possessor extraction. . . . . . . . . . . . . . . . . 596
6.2.1. Extraction from a doubling construction? . . . . . . . . . . . . . . . . . . . 596
6.2.2. Arguments against a movement analysis for WF . . . . . . . . . . . . . 599
6.2.2.1. Extraction from subject. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599
6.2.2.2. Er-insertion and indefinite subjects . . . . . . . . . . . . . . . . . . . . . . . . . 600
6.2.2.3. Wh- islands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
6.2.2.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
6.3. Remote possessors and resumptive pronouns . . . . . . . . . . . . . . . . 605
7. More positions for possessors: some speculative remarks . . 608
7.1. Against a left dislocation analysis for WF possessor doubling . 608
7.2. A split DP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
8. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
Appendix. Inalienable possessors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
1. Types of inalienable constructions. . . . . . . . . . . . . . . . . . . . . . . . . . 613
2. Inalienable possessors as arguments.. . . . . . . . . . . . . . . . . . . . . . . . 615

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
Subject index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
Language index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
The authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665
Preface

Aim of the book

This book is a theoretically oriented, comparative study of the aspects of

morphosyntax of what is traditionally called the noun phrase (NP), i.e. the
projection of the noun. The goal of the book is to offer a survey of current
discussions on a number of key issues that have become prominent in re-
search on the syntax of nominal projections within the generative tradition.
The book is thus primarily intended for linguists interested in some as-
pect of the structure and morphology of the nominal projection. Although a
basic background in the generative tradition is presupposed, any crucial
theoretical assumptions adopted in the book will be elaborated at relevant
points. Hence the book should be accessible to advanced students as well as
to readers who are broadly familiar with generative syntax but who may not
be familiar with the precise implementations adopted in the book. As many
issues relating to the structure of the nominal domain are also relevant for
the analysis of the clause, and since we will often place the discussion
against the background of the development of the theory as a whole, the
syntactician whose main research interest lies outside the nominal domain
will hopefully also find areas of interest in this book.

Syntax of nominal projections and syntax of clauses

All current generative research on the syntax of the nominal projection has
been crucially motivated by the emergence of the ‘DP-hypothesis’, as ad-
vanced by the work of Abney (1987). In addition, as in any other area of
syntax, research on the nominal projection is obviously also influenced
continuously by the theoretical developments within generative grammar.
In research into the nominal domain as elaborated during the last twenty
years, a number of key areas of interest can be identified; we will briefly
introduce these here, though, obviously, the various domains of interest are
ultimately related and cannot be kept fully isolated.
The DP hypothesis postulates that, in the same way that the projection
of the verb is dominated by functional material, the projection of the noun
xviii Preface

is part of a larger functional complex, the DP. One of the central issues
with respect to the syntax of DPs arises from the fact that interesting paral-
lelisms can be observed between the nominal domain and the clause, that
is, the verbal domain. It is, for instance, tempting to compare the role of the
V head in the clausal domain to that of the N head in the nominal domain,
and while it is C, the complementiser position, that provides discourse an-
choring in the clause, the same role can be argued to be played by D, the
determiner, in the nominal domain.
The assumption that what used to be called NP should be reinterpreted
in terms of DP, that is a projection of D with a nominal complement, means
that the determiner has a central role in the nominal system. This in turn has
led to a number of questions concerning the status of the determiner ele-
ments found within the DP. In particular, questions have arisen about the
position and interpretation of definite and indefinite articles, of demonstra-
tive pronouns and of possessive pronouns in the languages that have them.
Equally, given the DP hypothesis and its core assumption that a NP is
dominated by a DP, questions arise as to how to analyse nominal projec-
tions without an overt determiner.
Another area of study concerns the assumption that in the same way that
clauses are basically V projections augmented with functional projections
(TP, AGRP, AspP etc), DPs are N projections augmented with functional
projections. This leads to obvious questions about the functional layering of
the DP: in addition to DP, are there other functional projections, how many
such projections are there, how can they be motivated, what are their inter-
pretative properties? Given that functional projections in the clause have
been tied in with the availability of morphological markers of Tense,
Agreement, Aspect etc, there has also been a renewed interest in the mor-
phological markers of the noun and their relevance for postulating func-
tional projections. Morphological issues related to the status of functional
categories include questions concerning the realisation and interpretation of
features such as agreement, case, gender (word marker/stem affix/inflection
class), in the nominal domain.
In the same way that the syntax of semantics of adverbial modifiers in
the VP has given rise to much discussion, the syntactic and semantic rela-
tionship of (primarily if not exclusively adjectival) modifiers to the noun
has received a lot of attention. This research ties in directly with that
concerning the status of functional projections in the NP and the question to
what extent the syntax of nominal modifiers (especially adjectives) can be
aligned with that of verbal modifiers (especially adverbs). A related question
 Preface xix

is also how the relative position of the noun with respect to the modifying
adjectives can be derived. For instance, in the same way that some posi-
tions of the verb in the clause have been argued to be derived by movement
of V to a functional position, it has been argued that the postnominal posi-
tion of the adjectives is due to N-movement across the adjective. However,
the N-movement hypothesis has not gone unchallenged and alternatives
have been elaborated. The assumptions that there is a rigid split between
lexical categories and functional categories have also come under scrutiny.
With respect to the clausal domain there have been proposals that certain
verbs belong to hybrid categories with both functional and lexical proper-
ties and the same proposals have also been made with respect to the status
of certain nouns.
A final area of research is centred on the parallelism between V as the
semantic head of the clause and N as the semantic head of the DP. In the
same way that lexical verbs have arguments with which they have thematic
relations in the clause, nominal heads may also be argued to have argu-
ments, with which they have thematic relations. The assumption that nouns
may have arguments seems particularly natural in the case of deverbal
nouns. In addition, possessor arguments are also typically found in nominal
projections. Assuming there are indeed arguments in the nominal domain,
then questions arise also with respect to their distribution, their relation to
the structure, in particular whether they have specifier or complement
status. It has further been argued that just like clauses (i.e. projections of
verbs) instantiate a predication relation, DPs contain evidence for predica-
tion relations. This line of enquiry has, among other things, led to new
analyses for possessor constructions and for pseudopartitive constructions.

In this book, we want to offer a discussion of the research areas in the do-
main of the syntax of the nominal projection outlined above, with special
attention for the parallelisms between the nominal projection and the clause.
In order to achieve this goal we will systematically relate phenomena rele-
vant for the nominal projection to other syntactic phenomena. For instance,
the syntax of possessive pronouns in the nominal projection is related to the
classification of pronouns which was elaborated to account for their distri-
bution in the clause, N-movement in the nominal domain is compared to V-
movement in the clause, the syntax of the genitive construction is related to
that of predicate inversion in the clause.
We also want to show how research into the nominal projection is un-
avoidably determined by developments in the theory. Often, we have at-
xx Preface

tempted to integrate earlier findings on the syntax of nominal projections

into newer theoretical proposals, casting new light on the empirical domain
at issue. In the various chapters, we will show how recent theoretical pro-
posals (distributed morphology, anti-symmetry, minimalism, cartography)
can cast light on aspects of the syntax of the DP and can enrich and refine
earlier analyses. We also indicate problems with the analyses that have
been proposed, whether they be inherent to the theories as such (e.g. what
is the trigger for movement in antisymmetric approaches) or to the particu-
lar instantiations. In the discussion of various issues, we apply the frame-
work that is most adequate to deal with problems at hand. We therefore do
not use the same theoretical approach throughout the book. As a conse-
quence, at various points in the book we will provide a brief introduction to
theoretical proposals which we adopt at that point.
We wish to underline that our book does not aim at providing the defini-
tive analysis of the syntax of noun phrases. We consider that this would not
be possible, given the current flux in generative syntax, with many new
theoretical proposals being developed and explored in parallel. Our goal is
to give the reader the background for research and to show how a number
of quite different proposals in the literature have been be applied in an in-
teresting way to the nominal domain. When relevant, we will point to re-
maining issues for further research. We also point out that, while we have
aimed at covering a wide range of areas, the book is not an exhaustive sur-
vey of the vast literature on noun phrases. And though proposals in the lit-
erature will be discussed when relevant, our aim is not to provide a critical
survey of the literature. We feel that such a critical approach to the litera-
ture would be guided by general theoretical choices rather than by issues
specific to the syntax of nominal constructions, which is the focus of our
book. Whenever we introduce proposals from the literature our goal is to
use them to cast light on the phenomena discussed.

Organization of the book

The book has four parts, each composed of a number of chapters. Part I is
a general introduction. Part II is concerned with the functional make up of
the nominal projection. Part III deals with DP internal modification rela-
tions. Part IV is concerned with the relation between a head N and other
DPs within the nominal projection.
Though there are obviously relations between the three parts of the book,
and between the various chapters, we have tried to make the main parts as
 Preface xxi

well as the chapters in them relatively freestanding. Each deals with one
specific aspect of the syntax of nominals and can be read on its own.
The book is comparative in its approach: as is standard practice in gen-
erative grammar, data from different languages will be examined, including
English, and the Germanic languages, the Romance languages, Slavic lan-
guages, Semitic languages and modern Greek. We do not systematically ex-
amine each of the languages discussed for all of the properties at stake, but
rather we will introduce data from those languages that seem particularly
telling for the point at issue.

Acknowledgements

The writing of this book started in 1999 as part of the research project “Offer
of career to Greek speaking researchers abroad” advertised by the Greek
Ministry of Development (General Secretary of Research and Technology)
which was awarded to Artemis Alexiadou and Melita Stavrou in 1998
(Number of project 97EL, filed by the Research Committee of the Aristotle
University of Thessaloniki, GR, as Project no. 09142 ) for four years. As
Liliane Haegeman also had the idea of writing a survey book on noun
phrases, the three authors decided to join forces.
The support of the General Secretary of Research and Technology is
hereby gratefully acknowledged. Artemis Alexiadou and Melita Stavrou
wish to thank Mrs Afroditi Zouridaki for her help on administrative and
practical matters throughout the four years of the duration of the project.
The completion of this book took much longer than planned or antici-
pated. During some of that period Artemis Alexiadou was a Heisenberg
fellow, and her research related to the book was financed by the DFG grant
AL 554/1-1, which is hereby acknowledged. Over the years, parts of the
book were presented in conferences, workshops and classes. Several people
had the chance to read draft chapters and commented on these. We wish to
say a big thank you to all those people who offered us stimulation and
valuable feedback.
We are particularly grateful to Henk van Riemsdijk for his valuable help
and support on various levels – practical matters, content-wise, biblio-
graphical matters. More specifically we thank him for his substantial and
critical comments on the chapter on pseudopartitives. We are grateful to
Héctor Campos for reading and commenting on an earlier draft of Part II,
chapter 2 of Part III and chapter 1 of Part IV. Giuliana Giusti for reading
xxii Preface

chapter 1 of Part II and helping us better understand what the issues are
with the crosslinguistic distribution of demonstratives. Guglielmo Cinque
read critically and meticulously chapter 1 of Part III and particularly com-
mented on questions pertaining to Romance adjectives. Florian Schäfer
helped us with the compilation of the literature and gave comments on
nearly all draft chapters. Björn Rothstein commented on large parts of the
book. Jennifer Pyroth checked the English of the manuscript. We are really
grateful to Ursula Kleinhenz and Anke Beck for their encouragement, un-
derstanding and most of all their patience.
We are also grateful to students in Potsdam, Berlin, Stuttgart, Amster-
dam (LOT Winter school 2004) and Thessaloniki for comments during the
writing up period of the book.
Part I

Introduction

This book is a theoretically oriented, comparative study of some aspects of

the morphosyntax of constituents that have been traditionally referred to as
noun phrases. The core empirical data addressed here are fairly straightfor-
ward. In the following examples the underlined strings are all noun phrases
of one type or another.

(1) a. Henry is hungry.

b. There is a grey cat waiting outside.
c. The cat probably wants to come in.
d. All our cats are very independent.
e. This big grey cat in the corner is Nelson.
f. The cat’s tail was moving energetically.
g. Rembrandt’s picture of Lulu was very detailed.
h. Cats are wonderful creatures.
i. Topsy loves fresh cream.

The semantic nucleus of the underlined constituents is a noun which may

be accompanied by other constituents of various categories. (2) provides a
very preliminary inventory of some of the components of the underlined
constituents in (1) with provisional category labels.

(2) Noun (N) proper name

Lulu, Henry, Nelson, Rembrandt,
Topsy
common noun cat, corner, creature, cream,
picture, tail,
Adjective (A) fresh, grey, wonderful
Determiner (D) definite the
indefinite a
Demonstrative (Dem) this
Quantifier all
2 Part I – Introduction

In this book we will be concerned with the distribution and function of the
components of nominal projections and with the various relations between
the noun and the other constituents in its projection. As a shorthand term
the labels Noun Phrase or NP are often used to refer to constituents headed
by a noun but, though there is indeed a need for this label to designate the
(lexical) projection of N, we will see that technically the underlined con-
stituents in (1) are more than projections of N, i.e. NPs. Following current
tradition in the generative framework (see Abney 1987) we will usually
refer to constituents such as those underlined in (1) as DPs.
The present chapter is an introduction to the book. We provide a survey
of some of the major areas of research in the domain of nominal syntax.
One prominent starting point of much research on the nominal projection
revolves around the similarities and differences between nominal syntax
and verbal syntax. To put it simply, comparisons are made between noun
phrases and sentences. As will be shown below, the way this issue is ad-
dressed is not independent of theoretical considerations.
In the introduction we provide first a discussion of the way in which the
nominal constituents seems to have certain properties in common with
clauses. These observations will be a basis for the remainder of the book, in
that we will examine to what extent proposals for the analysis of the clause
can be carried over to the analysis of the nominal constituent. In the second
section of the chapter we introduce the central theoretical concepts which
will be used in the book. This section is an introduction to some basic con-
cepts in syntactic literature. Readers familiar with the theoretical models
used here, namely the Government and Binding model, the Principles and
Parameters model and the recent Minimalist model, will not find much new
here and they can skip section 2 of the introduction.

1. Some parallelisms between clauses and nominal projections

1.1. Subjects and genitives

Many discussions concerning constituents headed by nouns will point out,

among other things, that in English the prenominal genitive seems to be to
the noun phrase what the subject is to the clause. This is especially clear in
the case of nominalizations. For instance, just as Caesar is the Agent of the
action denoted by destroy in (3a), it could be argued that the genitive Cae-
sar’s in (3b) denotes the Agent of the action expressed by destruction.
 Some parallelisms between clauses and nominal projections 3

(3) a. Caesar destroyed the city.

b. Caesar’s destruction of the city

Similarly, just as in (3c) what was the object of destroy has become the
subject due to passivization, in (3d) the Theme argument of destruction in
(3b) is now expressed by the genitive in (3d), suggesting that nominal pro-
jections, too, allow for argument changing, just like sentences do.

(3) c. The city was destroyed by Caesar.

d. the city’s destruction by Caesar

Finally, just as in (3e) the subject Caesar is the antecedent of the reflexive
himself and cannot be the antecedent of the pronoun him, in (3f) the genitive
Caesar’s is the antecedent of himself and cannot be the antecedent of him:

(3) e. Caesar described himself to him.

f. Caesar’s description of himself to him

These various subject-like properties of genitives may be taken as support

for postulating a large degree of parallelism between the syntax of noun
phrases and that of clauses. In generative approaches to syntax, this particu-
lar issue has been on the agenda at least since Chomsky (1970), who fo-
cused on the relation between clauses and the related nominalizations. One
specific question that arises is whether nominals such as those in (3b), (3d)
and (3f) can inherit the argument structure of the verbs they are derived
from, and if so, how this is achieved.

1.2. Functional structure: the DP hypothesis

The semantic nucleus of the clause is the verb, the semantic nucleus of the
nominal projection is the noun. In the same way that a clause can be shown
to be more than a mere projection of a verb, it has been argued that the so-
called Noun Phrase is more than the mere projection of a nominal head.
Clauses are extended projections (in the sense of Grimshaw 1991) of the
verb: the lexical projection, VP, is dominated by a number of functional pro-
jections, such as IP and CP, giving rise to the C-I-V hierarchy (Chomsky
1986b). In a similar way it has been proposed that the nominal projection is
dominated by functional projections, the first such projection being Deter-
miner Phrase or DP (Abney 1987; Horrocks & Stavrou 1987; among others).
4 Part I – Introduction

Much work in the late 1980s was devoted to establishing the correctness
of the so-called ‘DP-hypothesis’, i.e. the hypothesis that the determiner
heads the Det+Noun constituent, by bringing cross-linguistic facts to bear
on the issue. Two types of arguments were prominent in the discussion. On
the one hand there were arguments concerning the grammatical and dis-
tributional properties of determiners (e.g. Haider 1988 on German, among
many others). On the other hand, arguments concerning noun movement
can be seen to support postulating at least one functional projection above
NP. If one wishes to postulate that the nominal head moves within the pro-
jection of N one must assume that there is at least one additional head posi-
tion which can receive the moved N. The position of the determiner, D, has
been identified as just such a position. Consider for instance the distribution
of the noun casa in the Italian examples in (4) (Longobardi 1994, 1996):

(4) a. La mia casa è bella.

The my house is beautiful
b. Casa mia è bella.
c. *La casa mia è bella.
d. *Casa la mia è bella.

In (4a) the definite article la precedes the possessive pronoun mia. In (4b)
casa precedes mia and this order is incompatible with the presence of the
determiner (4c, d). The N-movement argumentation would go as follows:
Leaving aside a detailed analysis of the position of mia, one might say that
while in (4a) the noun head occupies the head position of the lexical projec-
tion of N, and D is the head of a functional projection dominating NP, in
(4b) N has moved to the position of the determiner.

(4) e. [DP [D casan] [ mia [NP [N tn]]]]

The moved constituent leaves a coindexed trace (tn) in its original position.
In Minimalist literature, such a coindexed trace has been replaced by a copy
(see section 2.5.2), so (4e) would be equivalent to (4f), where the crossed
out representation casa represents the copy of the moved noun casa.1

———–——————————
1
In this book we will use both the trace symbol (t) and the copy, but without
these notations implying any theoretical difference. When we use the symbol t
in a position, for trace, we understand this to mean that the relevant position is
occupied by a copy of a moved constituent and that this copy is not pronounced.
 Some parallelisms between clauses and nominal projections 5

(4) f. [DP [D casa] [ mia [NP [N casa]]]]

For a number of languages, the distribution of the noun with respect to

other constituents of the nominal projection has been interpreted in terms of
overt raising of N to D (cf. Delsing 1993a, 1998; Taraldsen 1990 on Scan-
dinavian; Ritter 1991 on Hebrew), an instance of head movement within an
extended projection paralleling verb movement to I or C. 2 We return to this
issue in section 2.3.

1.3. Survey of this book

The DP-hypothesis has achieved a broad consensus, not least since it allows
a conceptual unification of syntactic structure across categories. Without
the DP-hypothesis, the by now standard view of the extended projection
(Grimshaw 1991) as the basic constructional unit in natural language could
not have taken hold. Subsequent attempts to improve our understanding of
the D-N extended projection have been concerned with four main issues:

(5) a. the articulation of the D-N extended projection,

b. the status of arguments in DP,
c. the status of modifiers in DP,
d. the effects of head/phrasal movement inside the nominal projection
(NP and DP).

The present book offers a survey of some of the literature on the issues
listed in (5) The book contains four major parts, which to a large degree can
be read independently, though there will obviously be some cross-refer-
ences. Part I, i.e. the current chapter, is a general introduction. Part II is
concerned with the functional make-up of the nominal projection. Chapter
1 of Part II deals with the category D, and will examine both the elements
that lexicalize D (in particular, articles and demonstratives) and the seman-
tic categories that D is currently taken to encode (in particular, definiteness
and reference). Chapter 2 of Part II surveys some of the various proposals
———–————————————————————————————
We will often prefer the trace notation to symbolize copies simply because using
multiple copies often gets in the way of clarity of presentation and ‘readability’.
2
Longobardi (1994) generalizes the proposal by arguing that N-raising to D occurs
covertly elsewhere. We discuss his proposal in chapter 2 of part II. See section
2.1. on covert movement.
6 Part I – Introduction

that have been advanced in order to account for articleless, or determiner-

less, noun phrases. Chapter 3 of Part II is concerned with functional catego-
ries within the nominal projection. Part III deals with DP-internal modifi-
cation relations. Chapter 1 of Part III is concerned with adjectival modifiers
in the nominal projection. It investigates the factors determining the distri-
bution of adjectives within the nominal projection (NP/DP) and it also ex-
amines to what extent a difference in distribution may correlate with a dif-
ference in interpretation. Chapter 2 of Part III deals with two constructions
that involve so-called semi-functional (or semi-lexical) categories: the N-of-
N construction and the pseudo-partitive construction. Part IV is concerned
with the relation between a head N and other DPs within the nominal pro-
jection. Chapter 1 of Part IV takes up the issue of arguments in nominals,
while Chapter 2 is concerned with the syntax of possession.
Before turning to the individual chapters Section 2 introduces our basic
theoretical background for the discussion to follow. Readers who are famil-
iar with generative literature will probably not find any new material in this
section. They can proceed immediately to Part II.

2. The theoretical framework

The book has been written against the background of what is usually re-
ferred to as generative syntax, the research program initiated by Noam
Chomsky in the 1950s. In particular, we shall be assuming the Principles
and Parameters framework as elaborated in the Government and Binding
model of the 1980s (see for instance, Haegeman 1994; Radford 1988) and
we will also be referring to theoretical proposals drawn from recent work in
syntax including (i) the Minimalist Program (Chomsky 1995; Radford
1997, 2004; Adger 2003; Lasnik, Uriagereka, Boeckx 2005), (ii) the anti-
symmetry approach to syntax (Kayne 1994), (iii) the Distributed Morphol-
ogy approach (Halle & Marantz 1993). Since Distributed Morphology will
only be relevant for Chapter 1 of Part IV we will not introduce the aspects
of the framework relevant for our discussion until section 2.4. of that chap-
ter. In this introductory chapter we will present only the broadest outlines
of the first two theories. When relevant, we will elaborate the specific im-
plementations in later chapters as they become relevant for a particular is-
sue or question.
 The theoretical framework 7

2.1. Levels of representation

A theory of syntax has to assume that language has two basic components,
the lexicon, which provides the elementary building blocks of the language,
and the syntax, a structure-building system which combines these primitive
elements into larger units.
Building on the generative tradition initiated in the 1950s, the Govern-
ment and Binding framework (based on Chomsky 1981) proposes that lexi-
cal items are inserted at a particular level of syntactic representation, called
D-structure. The syntax operates on this D-structure representation through
movement operations, leading to a second level of syntactic representation,
called S-structure. S-structure is the basis for both the interpretation of the
structure, Logical form (LF), and for its overt realization, Phonetic Form
(PF). S-structure results from various movement operations and is reflected
in the overt form of the sentence: the moved constituents are displaced. LF
is an interpretive level in which non-overt movements may have taken
place to encode semantic relations (scope, for instance). It is assumed that
any movement that can overtly take place before S-structure may also apply
covertly to generate LF-relations.
Thus, we obtain what has been referred to as the T-model of grammar
with its three interface levels D-structure, PF and LF. S-structure mediates
between these levels. A representation of this model is given in (6a).

(6) a. (lexicon)
!
D-structure
!
S-structure
?

PF LF
(sound) (meaning)

The geometrical relations between the various levels represented in (6a) are
not accidental. Specifically, because the path between S-structure and PF is
different from the path from S-structure to LF, whatever (movement) op-
erations mediate between S-structure and LF will not affect the phonetic
form of a structure. Similarly, manipulations of S-structure which apply on
the path to PF will not have any impact on the interpretation (LF).

Let us illustrate this point with a very much simplified example. Consider
(7a) and (7b):
8 Part I – Introduction

(7) a. John has met Mary

b. Who has John met?

In (7a) the nominal constituent Mary is an argument of the verb meet. It

occupies the canonical object position. In (7b), on the other hand, the direct
object of meet is the interrogative pronoun who, which does not occupy the
canonical object position. However, it is clear that in (7b) too who is an
argument of meet. In order to represent the relation between meet and who
in (7b) we propose that the sentence is derived in two steps: (i) first the
object of meet, i.e. who, is inserted into the VP, and (ii) then undergoes
movement to a sentence-initial position. The moved constituent preserves
its relation with the original object position, or, to put it differently, in (7b)
who still counts as the object of meet.
In addition to the interrogative pronoun, the inflected auxiliary has also
moves to a position to the left of the subject. Again we assume it is inserted
into the position in which we find the auxiliary in (7a) and then it moves
leftrd. To represent this we use the trace3 notation. The indices i and j are
used to show which trace relates to which moved constituent: ti is the trace
of who, tj is the trace of has.

(7) c. [Whoi hasj [John tj met ti]]

The question arises why this movement has taken place. Probably the an-
swer must be that to signal interrogative force we need to use the left edge
of the clause. We could propose that the left edge of the clause is the area
that encodes illocutionary force (among other things). Thus who is obliged
to move, since, being interrogative, it needs to end up in the layer of the
clause that can express interrogative force. On the other hand, not being
interrogative, Mary has no need to move to that zone. Since there is no rea-
son to move up, the object Mary stays where it has been inserted, in the

———–——————————
3
As pointed out above, the trace notation has been replaced by the copy notation
in the minimalist literature. Thus (7c) would be represented as (i), where the
strikethrough notation is used to indicate the copies of moved constituents:
(i) [Who has [John has met who]]
In the trace notation, the link between the trace and the moved constituent is
indicated by coindexation, as shown in (7c). It is obvious that in the copy nota-
tion coindexaion has become superfluous, since from the strikethrough notation
it is clear which constituent the copy is related to.
 The theoretical framework 9

canonical object position. Constituents only move if there is a need for

them to move. Or, to put it in technical terms, movement takes place as a
last resort.
Now consider the following French examples:

(8) a. Qui as-tu rencontré?

Who have-you met
‘Who did you meet?’
b. Tu as rencontré qui?
You have met whom?
‘Who did you meet?’

In (8a) the same pattern is to be found as in the English counterpart, again

the interrogative pronoun and the auxiliary have moved:

(8) c. [quii asj [tu tj rencontré ti]]

Now consider (8b). This example has the same interpretation as (8a) it is a
question about the object.4 In this example the object has not moved. How-
ever, the object is an interrogative pronoun. If interrogative force is inter-
preted also on the left edge in French, then we must assume that to be fully
interpretable qui (‘who’) in (8b) should actually also end up on the left
edge. One proposal that has been adopted is that there IS indeed movement
of qui to the left edge, but this movement only takes place at the interpreta-
tive level. Hence the movement will also lead to a structure such as (8c)
above, but the movement to derive this structure will not be associated with
a visible displacement. Thus (8d) would be the LF representation of (8b):

(8) d. [quii asj [tu tj rencontré ti]]

The Minimalist Program (Chomsky 1993, 1995, 2000, 2001) reconsiders

the role of these levels of representation. It is argued that the only concep-
tually necessary levels of representation are those related to external sys-
tems, i.e. the level which is related to the so called articulatory-perceptual
———–——————————
4
Obviously pairs such as the French examples in (8a) and (8b) pose a problem
for the hypothesis that movement is a last resort operation, since the very fact
that (8b) is grammatical makes us wonder what could be the motivation for
movement in (8a). We will not dwell on this issue here.
10 Part I – Introduction

system (i.e. the level of ‘sound’ or PF) and the level which is related to the
conceptual-intentional system (i.e. the interpretative level of LF). The levels
D-structure and S-structure in (6a) are completely internal to the structure-
building system. Since there is no independent direct evidence for their
existence, the minimal assumption is that these levels do not exist. Within
the Minimalist framework, it is therefore assumed that the only two levels
of representation are the interface levels PF and LF. The lexicon is taken to
provide the building blocks of the sentence.
The starting point for the construction of a sentence is a set of lexical
elements (the so-called Numeration). The syntax builds up the structure by
combining the elements drawn from the Numeration, according to certain
principles and until the Numeration is exhausted. At some point during this
derivation the information contained in the structure built up so far and
which is relevant to PF is fed to the PF component. This point is called
Spell Out. All syntactic operations carried out before Spell-Out are re-
flected in the PF output. After Spell out, additional non-overt processes
may apply to the structure to derive the semantic representation (the LF
interface). These additional processes, which apply AFTER Spell Out, do
not have any repercussion on the overt representation of the sentence. The
Minimalist type of grammar can be represented as follows:

(6) b. Lexicon
!
!
? -----> PF ‘sound’
!
covert syntax
#
? -----> LF ‘meaning’

Returning to our examples: in English (7b) as well as in French (8a) the

movement of the interrogative object (who, qui) takes place before Spell-
Out, in the overt syntax, thus producing a visible displacement. In (8b) there
is no movement of the interrogative constituent before Spell-Out. Movement
in (8d) takes place after Spell-Out, in the covert syntax.
In what follows we outline the internal working of syntax, i.e. the com-
putational system that builds structure. We will be combining Minimalist
insights with more traditional insights from the Government and Binding
tradition. The difference between the two traditions will be highlighted
when relevant.
 The theoretical framework 11

2.2. Syntactic structure: the X-bar format

There are a number of assumptions that seem to be relatively constant

across the various incarnations of the generative framework. One is that all
syntactic structure is endocentric: syntactic units are organized around a
head. Each head, X, projects a larger syntactic unit (a phrase, XP), and each
phrase, XP, must have one head. This assumption captures the traditional
intuition that the head of a verb phrase, for instance, is a verb.
In one precise implementation of this idea, all syntactic constituents
have the same format, which can be represented as in (9a) below, where X
indicates the head of the constituent XP. The head X combines with a con-
stituent, here ZP, which itself is built according to the format in (9a). ZP is
referred to as the complement of X. The combination of X and its comple-
ment is referred to as X’, the intermediate projection of X. This projection X’
combines with another constituent, YP, referred to as a specifier, to form XP,
the maximal projection. Again, YP itself is also formed according to (9a).

(9) a. X-bar format

XP [MAXIMAL PROJECTION]

YP X’ [INTERMEDIATE PROJECTION]
[SPECIFIER]
X ZP
[HEAD] [COMPLEMENT]

It is sometimes proposed that phrases can be added to XP through what is

called adjunction. Adjunction of WP to XP creates an additional projection
of the same category. In (9b), the phrase WP is adjoined to the base XP
giving rise to another XP projection. The base XP and the XP created by
adjunction are sometimes identified by numbers (XP1, XP2)

(9) b. X-bar format with adjunction

XP2 [MAXIMAL PROJECTION1]

WP XP1 [MAXIMAL PROJECTION1]

YP X’ [INTERMEDIATE PROJECTION]
[SPECIFIER]
X ZP
[HEAD] [COMPLEMENT]
12 Part I – Introduction

However, the concept of XP-adjunction is not generally accepted (cf. Kayne

1994; Cinque 1999).
Observe that both (9a) and (9b) contain only binary branching structures
(cf. Kayne 1984), that is to say: from each point there are at most two
downward branches. This format is generally assumed in current work in
generative syntax both in the Principles and Parameters tradition and in the
Minimalist tradition.
The basic structural configurations used to express geometrical relations
between different elements in the structure in (9) are dominance and c-
command. (i) Dominance refers to a relation in which one node is higher in
the structure than another node. XP in (9a), for instance, dominates all the
other nodes (YP, X’, X and ZP); X’ dominates the nodes X and ZP; YP, X
and ZP do not dominate any other node. (ii) C-command expresses a rela-
tion between a node α and a node β in which the node α does not dominate
the node β, but in which every node that dominates α also dominates node
β. In (9a), YP c-commands X’, X and ZP; X c-commands ZP. In (9b) WP
c-commands XP1, X’, X, ZP and YP. (9) provides the blueprint for syntac-
tic structure, which is then realized by various categories. We turn to the
concrete realization of this format presently.
The format in (9) also constrains movement: basically a constituent of
the type head (i.e. X) must move to another position of the type head, while
a constituent of the type XP must move to another position of the type XP.
In (9a), for instance, we might imagine that ZP moves to the position YP,
but not that X moves all by itself to the position YP.

2.3. Lexical categories and functional categories

It is also often assumed that there is a clear-cut distinction between two

types of heads: lexical heads and functional heads.5 Lexical heads are the
‘content words’ of traditional grammar: they contribute directly to the ‘de-
———–——————————
5
As we will see in Chapter 2 of Part IV, in the recent literature (see e.g. work
within the framework of Distributed Morphology and Borer (2005)) this position
has been challenged. Several researchers thus claim that sentence elements, such
as noun, verb, adjective have no universal significance and are essentially de-
rivative from more basic morpheme types. Specifically, the different ‘parts of
speech’ can be defined as Roots which combine with a set of functional heads
that determine category. See Embick & Noyer (2004), Embick & Halle (to ap-
pear) for further discussion.
 The theoretical framework 13

scriptive content’ of the sentence, i.e. the description of the event or state of
affairs expressed in the sentence. Lexical categories ‘link’ the language with
the non-linguistic world, in that they ‘denote’ entities, properties, activities,
etc which are as such non-linguistic. For instance, in (10) the lexical heads
are cat, drink, milk.

(10) a. The [N cat] [V drinks] the [N milk].

There are four kinds of lexical heads: in addition to N and V, illustrated

above, there are adjectives (A) and prepositions (P).

(11) a. The cat is [A thirsty].

b. The cat is [P under] the table.

Functional heads do not contribute directly to the description of the event.

One of their purposes is to encode grammatical relationships, i.e. relation-
ships among linguistic entities. For instance, functional categories will be
involved in expressing the relation of agreement. Consider as an example
the agreement between a subject and a verb in English as in (10a): the cat is
singular and the verb drinks has the ending -s, which matches the number
of the noun. The -s morpheme on drink is not an inherent part of V. The
verb eat does not always come with the ending -s. Rather, it is a functional
morpheme which is added to V for third person singular agreement in the
present tense. This agreement morpheme links a singular subject with the
verb but it does not modify the event described in the sentence. In a sense,
then, the ending -s on the verb as such does not contribute to the interpreta-
tion of the clause. In Minimalist terms the agreement ending on the verb is
said to be [–interpretable].
Now consider (10b):

(10) b The [N cats] [V drink] the [N milk].

Here we find an ending -s on the N cat. Again the -s ending is not an intrin-
sic part of the noun, in (10b), for instance, there is no such ending. The -s
ending is added to the noun to encode plural. Though this ending expresses
Number, and is a functional element added to the lexical head, the number
ending on the N is not uninterpretable: informally speaking, cat differs
from cats in that the former denotes one entity with the relevant properties
to qualify as a ‘cat’ and the latter denotes a plurality of such entities. So
14 Part I – Introduction

while, in Minimalist terms, the agreement ending on the verb is [–inter-

pretable] that on the N will be [+interpretable].
Functional elements need not be bound morphemes, free morphemes
may also be functional. For instance, inserting the modal auxiliary will in
(10c) also does not modify the event depicted by the sentence.

(10) c. The cat will drink the milk.

The activity referred to in (10c) remains the same as that in (10a). Will is a
functional element, it does not fundamentally contribute to the description
of the state of affairs expressed by the sentence. The function of the auxil-
iary will is to shift the temporal reference of the event into the future. Un-
like the case for the agreement morpheme -s on the verb in (10a), we can-
not say that will in (10c) does not contribute to the interpretation of the
clause and that it is [-interpretable]. Will does have an impact on the tempo-
ral interpretation, but it does not alter the state of affairs depicted by the
clause. Temporal and modal morphemes are also functional elements be-
cause, though certainly not meaningless, they do not have any impact on
the event expressed by the sentence.
The third person bound morpheme -s in (10a) and the modal will in
(10c) are functional elements associated with verbs. Extensive research has
postulated additional functional categories related to the verb/clause, in-
cluding a range of aspectual markers, modal markers etc (see Cinque 1999
for a maximally rich array of functional heads associated with the clause).
There also exist functional elements associated with nouns. We have
already come across the example of the number ending on N. In the exam-
ples above, the functional element the is associated with the N cat and also
with the N milk. The is a definite article or a definite determiner. Once again,
inserting the definite article will not directly contribute to the description of
the entity denoted by the nominal constituents: a cat and the cat both denote
a certain type of animal. However, these functional elements, too, are inter-
pretable in that, despite lacking descriptive content, they contribute to the in-
terpretation of the DP. Articles or determiners play a role in the referential
properties of the DP: the choice of the definite article in association with an
N indicates that we are dealing with entities (‘cat’ on the one hand, ‘milk’
on the other) which are not mentioned for the first time; the definite deter-
miner signals that the referents of the DPs are already accessible in the dis-
course, we know which cat and which milk we are talking about. By using
the indefinite article a in (10d) we introduce a novel cat into the discourse.
 The theoretical framework 15

(10) d. A cat was eating crisps under the table.

Functional categories, whether they are associated with the clause (and ulti-
mately with V) or with the N, share a number of properties (see also Abney
1987: 64f):

(i) They constitute closed classes.

(ii) They are generally phonologically and morphologically dependent, and
stressless. Often they are clitics or affixes and sometimes they are pho-
nologically null.
(iii) They are usually inseparable from their complement.
(iv) They lack descriptive content. (See also Ouhalla 1991; Giusti 1997 for
further elaboration of these basic properties.)
(v) Functional heads (usually) do not have arguments.

It is a standard assumption that there exists a clear-cut opposition between

functional heads and lexical heads and that categories are either lexical or
functional. However, as van Riemsdijk has shown (see van Riemsdijk
1998; Corver and van Riemsdijk 2001 for discussion and references), the
question arises whether such a clear-cut dichotomy is tenable. In fact, it has
been pointed out that in certain cases there seem to be categories which (at
least in certain environments/constructions) have properties both of lexical
heads and of functional heads (van Riemsdijk 1998).
One case in point concerns motion verbs. Consider for instance the
American English use of go in (12a) (see among others Jaeggli and Hyams
1993; Pollock 1994):

(12) a. John will go buy bread.

This construction, in which the verb go appears followed immediately by a

bare infinitive, is subject to a number of restrictions. For instance, the verb
may not appear inflected:

(12) b. *John goes buy bread

There is a very restricted number of verbs in English that enter this pattern,
namely come, go, run. The fact that we are dealing with a closed class may
lead us to think that these verbs are functional, rather than lexical. This is
confirmed also by the fact that in this use go cannot associate with a Goal
argument:
16 Part I – Introduction

(12) c. *John will go to the store buy bread.

d. John will go to the store to buy bread.

Typically, lexical heads are associated with arguments/thematic roles, while

functional categories are not associated with thematic roles. Thus it appears
as if the verb go has two uses in American English: it is either a lexical
verb, the ‘normal’ use, in which case it can take arguments, or it has acquired
functional properties, as in the examples illustrated in (12a–c). The analo-
gies of go in other languages too, display ‘mixed’ properties, as shown, for
instance, by Haegeman (1990) for West Flemish, by Schoenenberger and
Penner (1995) and van Riemsdijk (2002) for Swiss German, and by Cardi-
naletti and Giusti (2001) for Southern Italian dialects and for Swedish.
Elements like go illustrated above seem to belong to a hybrid category, in
that they are partly lexical and partly functional. Often they are referred to
as semi-lexical or semi-functional categories. For further illustration of
hybrid categories see also the papers in Corver and van Riemsdijk (2001).
One of the goals of this book is to provide an inventory of the functional
categories that have been identified in relation to the nominal domain. We
will also examine to what extent they correspond to matching functional
categories in the clause. In the next section we survey some of the functional
categories associated with the clause level. Once we have established the
functional structure of clauses, we can investigate to what extent nominal
projections are similar to or different from clauses in the course of this
book. Once we decide that clauses contain functional projections such as
TP or AspP, we will try to determine to what extent such projections are
valid for the nominal projection. This will be discussed in Part II of this
book.
Given the evidence for the semi-lexical categories in the clausal domain
(see van Riemsdijk 1998, Corver and van Riemsdijk 2001), we may ask
ourselves if the same is true for the nominal projection. The answer seems
to be positive. Van Riemsdijk (1998), for instance, discusses partitive con-
structions like (13) from Dutch:

(13) a. een plak kaas

a slice cheese
‘a slice of cheese’
b. een snee brood
a slice bread
‘a slice of bread’
 The theoretical framework 17

As the English translations suggest, the two juxtaposed nouns are in a parti-
tive relation. Van Riemsdijk shows that in spite of there being two nouns in
the constituent, the behavior of the containing nominal constituent is that of
a projection of a single head. In Chapter 2 of Part III we will return to the
issue of the presence of semi-lexical heads in the nominal domain, building
mainly on van Riemsdijk’s work (1998).

2.4. Lexical categories and Argument structure

2.4.1. Verbs and arguments

As mentioned already, lexical heads contribute directly to description of the

event or state of affairs expressed in the sentence. Typically, the element
which plays the major semantic role in this is the lexical verb. Consider the
examples in (14). Depending on the choice of lexical verb, the sentences
contain a different number of arguments: (14a) has two arguments, Topsy
and the milk, (14b) has one argument, Topsy, and finally (14c) has three
arguments, we, Topsy and the milk. In the traditional literature we will say
that drink is a transitive or an intransitive verb, yawn is an intransitive verb
and give is a ditransitive verb.

(14) a. Topsy drank the milk

b. Topsy yawned.
c. We gave Topsy the milk.

The number of arguments in a given clause is determined by the type of

predicate, here the verb. The predicate (here the verb) assigns a number of
thematic roles associated with the participants involved in the event or state
described.6
There is a one-to-one correspondence between theta roles and argu-
ments within a given clause. In the Government and Binding framework,
this property of the grammar was expressed in terms of the Theta Criterion,
which required that (i) each theta role of a predicate is assigned to one and
only one argument, and (ii) each argument is assigned one and only one
———–——————————
6
For instance, among transitive verbs some are associated with an Agent and a
Theme (ia), others with an Experiencer and a Theme (ib):
(i) a. I picked up the cat.
b. I liked the cat.
18 Part I – Introduction

theta role. Thus, the Theta Criterion determines the number of arguments
which are required and allowed within a clause. For example, the activity
described by a verb like yawn only involves one participant, the Agent of
the action, the verb yawn therefore assigns one thematic role and it only
requires one argument, in (14b) realized as Topsy. The verb drink involves
two participants, hence it assigns two thematic roles, Agent and Theme, the
entity affected by the action. A verb like drink therefore requires two argu-
ments. Finally the verb give is associated with three participants, the Agent,
the Receiver (or Goal), and the Theme.
There is a vast literature on the matching of argument structure with
syntactic structure and in this introduction we cannot hope to do justice to
all the various approaches. The reader is referred to Baker (1997), Levin &
Rappaport Hovav (2005) and Borer (2005) for discussion. We will limit
ourselves only to those aspects that will become relevant for the discussion
on the presence of argument structure in nominals.
With respect to verb syntax, two approaches to the question of argument
structure can be identified. On the one hand, concentrating on the lexical
semantics of a verb and the syntactic structures it can occur in, we can dis-
cern at least three different levels of representation of the relation between
a predicate and its argument(s): (i) a lexical semantic representation, (ii) a
lexical syntactic representation, (iii) a syntactic structure representation. The
lexical semantic representation of a predicate, often called lexical concep-
tual structure (LCS), is the ‘deep’ semantic description, which is probably
unique for any particular predicate, or a class of predicates. LCS decom-
poses the meaning of a verb into structures containing variables and meta-
predicates (like CAUSE, BE, etc.). Such a semantic description is mapped
onto the lexical syntactic representation, which is often called predicate
argument structure or argument structure (AS). AS represents how many
arguments a verb requires and to which syntactic argument positions these
are linked, for instance by making a distinction between external and inter-
nal theta roles (Williams 1981). On this view, the number of arguments a
predicate has depends on its meaning. Finally the syntactic representation
will articulate argument structure in the extended projection of the predi-
cate. For further discussion of this approach see Alexiadou, Anagnostopou-
lou & Everaert (2004).
 The theoretical framework 19

2.4.2. Decomposing the VP

2.4.2.1. Ditransitive verbs and binary branching

The alternative ‘decompositional’ syntactic approaches to the licensing of

arguments are inspired by Larson’s (1988) proposals to decompose V.
In a nutshell and simplifying a lot here, Larson’s proposal aimed at rec-
onciling the binary branching X-bar format for structure in (9) with the
observation that ditransitive verbs have three arguments. Assuming that one
of the arguments of the verb could become the subject, the question arose
how to deal with the two remaining arguments, which somehow would
have to both be internal to the VP. The question is how one VP could con-
tain three arguments. According to the X-bar format, there is one specifier
position. If we assume that one argument (the subject) can be associated
with the specifier position, then the internal structure of a verb with three
arguments would have to be realized by postulating two complements. At
first sight one might think of something like (15a):

(15) a. VP

Spec V’

V NP NP

However, (15a) does not respect the binary branching structure since from
(V’) there are three downward nodes.
In order to overcome this problem, Larson (1988) proposed decomposing
the node V and creating layers internal to the projection VP to show internal
structural relations between what seem to be two complements. One pro-
posal would be to replace (15a) by (15b) in which the indirect object and
the direct object form a constituent, here labelled ?P:

(15) b. VP

Spec V’
we
V ?P

Topsy ?’

? the milk
20 Part I – Introduction

This representation respects the binary branching format. The question arises
what label ? corresponds to. One option would be to take into account the
interpretation of the verb give: ‘give’ can be compared to ‘cause to get’: if
we give Topsy some milk then we bring it about (‘we cause’) that Topsy
will get some milk. Many verbs can be said to contain such a ‘causative’
component. It has been proposed that the causative component of a lexical
verb be represented by a special symbol, ‘v’ (‘little v’). The causative com-
ponent of the lexical verb, ‘little v’, is associated with the Agent role: in
(14c) the Agent of the action of giving is we. Thus the Agent is represented
as the specifier of vP, the projection of causative v.
Between indirect object and direct object there is a possessive relation,
brought about by the Agent. The relation between the indirect object and
the direct object could be represented by means of the symbol V, and give
would thus be represented as decomposed into ‘cause’ and ‘get’. CAUSE
and GET in (15c) do not stand for verbs that are realized lexically. Rather
they represent the semantic primitives that build up the interpretation of the
verb give.

(15) c. vP

Spec v’
we
v VP
CAUSE
Topsy V’

GET the milk

Observe that in (15c) the indirect object Topsy and the direct object the milk
form one constituent, VP, which excludes the Agent. 7
There are several other proposals in the literature, but because we will
not be dealing in detail with VP syntax we will not go into them. See Baker
(1997) and Emonds and Whitney (2006) for recent discussion and evalua-
tion of some proposals.

———–——————————
7
See Part IV Chapter 2, section 3.2. for an implementation of this structure to
encode possession in the nominal projection.
 The theoretical framework 21

2.4.2.2. Extending the proposal

Larson’s proposal that V may decompose into different shells (vP and VP
in (15c), for instance) has been extremely influential. The layered structure
of the VP has been generalized also for the cases in which a verb has only
one or two arguments. Hale & Keyser (1993) and Borer (2005) suggest that
the syntactic structure gives rise to a template which in turn determines the
interpretation of arguments. Essentially, what we could call lexical heads
are decomposed and their internal structure encodes the different semantic
relations between the various arguments. This view has been adopted in the
Minimalist program, leaving the status of the theta-criterion rather unclear.
Below we provide a sketch of the motivation for the decompositional ap-
proach. For more details the reader is referred to the literature.
Hale & Keyser (1993) generalize Larson’s VP-shell analysis to mono-
transitive verbs such as drink and propose that the thematic role of Agent,
the entity that initiates the action, is always associated with a separate (‘caus-
ative’) head v (‘little v’). The internal argument of a monotransitive verb
occupies the complement of the lower VP-shell and the external argument
is generated in the specifier position of a higher vP shell. In this view, each
thematic role is uniquely related to a head, i.e. the internal theta role is re-
lated to the lower V-head and the external theta role to the higher v-head. 8

(16) a. vP

DP
Topsy v’
VP
v

Spec V’

V DP

drink the milk

———–——————————
8
As the reader can observe, the final step in this development would be to decom-
pose the representation of (15c) even further and to also analyse ditransitive
verbs in terms of structures involving a separate head for each argument (cf. e.g.
Collins 1997: 53ff.; or Marantz 1993: 115ff.). We will not go into this issue here
as it will not affect the discussion at this point.
22 Part I – Introduction

Note that with respect to the functional/lexical divide, the status of ‘v’, the
head which is related to the external argument, is not completely clear. For
instance, Chomsky (1995) proposes that v is somehow both lexical and
functional. Other labels have also been proposed for the head related to the
external theta role, such as Voice (Kratzer 1994), Act(ive) (Holmberg &
Platzack 1995) or Tr(ansitivity) (Collins 1997).
The decomposition of V is also extended to one-argument verbs, and is
used to draw the distinction between unergative verbs such as sleep or tele-
phone and ergative verbs such as arrive, come. Unergative verbs are treated
as concealed transitives in this system in that they have a non-overt (cog-
nate) object, see (16b), while unaccusative verbs either lack vP altogether,
or contain one with no projected specifier (Chomsky 1995: 315):

(16) b. vP

Topsy
v’
VP
v

Spec V’

V DP

sleep

2.4.3. Nouns and arguments

Having established that the argument structure of V determines the pres-

ence of a number of other components of a clause, Part IV of this book ad-
dresses the question whether the same applies for the relation between N
and its projection.
As illustrated in our examples in (3b) above, one might wish to say that
nouns too are associated with arguments. In the earlier example the genitive
DP Caesar’s seems to refer to the Agent of destruction, in the same way
that Caesar is the Agent of destroy in (3a). Since the NP/DP is a projection
of N, a lexical head, the question that arises is that of the licensing and in-
heritance of argument structure in the nominal domain, which we will ad-
 The theoretical framework 23

dress in Part IV of this book. As we will show in that chapter, proposals

with respect to the argument structure of the nominal head are similar to
those that have been put forward with respect to the verbal projection in
that again, both semantically based proposals and structurally based pro-
posals have been put forward.

2.5. Functional projections

2.5.1. Evidence for functional projections

In general, three types of evidence are advanced for postulating functional

categories/heads: semantic, morphological, and syntactic/distributional. In
this section we will show how this evidence has been applied for postu-
lating a head position in the clausal domain and we will further discuss, for
each type of evidence, how an analogical reasoning could lead us to postu-
late a functional head in the DP. It should be emphasized that the three types
of evidence cannot always be separated as they are here (for ease of expo-
sition). More often than not, morphological, semantic and distributional
evidence will converge to corroborate postulation of a functional category.

(i) Semantic arguments

A first type of evidence for postulating functional projections is semantic.
The line of reasoning is roughly as follows. Lexical categories may be taken
to express certain concepts, but in the context of clauses, these ‘lexical’
concepts are associated with additional notions. The idea is then that these
additional notions are encoded in functional heads that are associated with
the lexical head in question.
For instance, and simplifying a lot here, consider the sentence. It can be
said that its semantic core is the verb and that a verb phrase expresses some
action or state. However, to describe the meaning of the sentence as a
whole, we need to take into account that a sentence adds a temporal dimen-
sion to the action/state expressed by the verb. The temporal reference asso-
ciated with a clause is to some extent independent of the verb in that one
may choose one such temporal expression among the various available
ones (say past tense vs. future tense) in a given language. Verbs are not
tensed ‘as such’. The observation that sentences are associated with a tem-
poral reference, and that this is not an inherent property of the predicate
(verb or adjective, for instance), may then lead us to postulate a specialized
24 Part I – Introduction

head to encode temporal reference. The head that encodes temporal refer-
ence can be labeled Tense (T); it selects a projection of V as its complement
and it projects a TP.
In the nominal system we can apply the same reasoning. Constituents
headed by nouns denote entities (persons, things) but they also contain in-
formation concerning reference (see section 2.3.). Since such information is
not an inherent part of the noun, it is proposed that there is a specialized
head D to encode the referential status of the nominal projection. D selects
NP as its complement and projects DP. From the early days of the DP-
hypothesis, D has been linked with encoding reference. It has also often
been observed that projections headed by nouns may function either as ar-
guments or as predicates, in the latter case the constituent is not referential.
In a number of languages, an NP used as an argument will obligatorily have
to be accompanied by a determiner, while a NP without the determiner may
be used as a predicate. Hence, a functional head D has also been postulated
to encode argument status.
Nominal projections may refer to one or more entities. This difference
concerns number, and again number is not intrinsically part of the N: in-
formally put, we choose the number of the noun depending on the intended
interpretation. The fact that a projection of a noun (or, taking into account
the functional structure, a DP) can be interpreted as referring to one (singu-
lar) or to any number (plural) of entities was taken as evidence that a spe-
cialized projection for encoding Number, namely NumP, should be postu-
lated. As we will see in some detail in the first chapter of Part III, for
instance, Bouchard (2002), attributes the referring capacity of noun phrases
to the properties of the semantic category of Number.

(ii) Morphological evidence

Another type of evidence for postulating functional categories is morpho-
logical. In many languages when lexical heads are inserted into a sentence
they do not come ‘bare’, that is as mere stems. Rather, they are associated
with inflectional morphology. Because morphology is variable (for instance
the verb may be associated with a choice of tenses) it is not taken to be an
intrinsic part of the lexical head as such, but rather it can be argued that the
inflectional morphemes constitute functional heads in the extended projec-
tion of the lexical head.
Let us see how the morphosyntactic argument works. The observation
that verbs can be associated with inflectional morphemes related to mood,
agreement, tense, aspect and voice is invoked as evidence for postulating
 The theoretical framework 25

the relevant functional heads, such as, for instance, Agr, T, Asp, Voice.
Again Ns often inflect for number, which would be taken as evidence for
postulating a functional head Num, thus supporting the NumP hypothesis.
In a similar vein, many researchers have further postulated a Gender
Phrase, based on the fact that at least in some languages nouns are marked
for Gender (or Word Class, cf. Picallo 1991; Bernstein 1993). Although an
obvious difference between verbs and nouns might seem to be the presence
of tense morphology in the former and its absence in the latter, there are
languages in which nouns may be argued to be morphologically marked for
tense. Thus, at least for these languages the morphological evidence could
be said to support postulating a Tense Phrase as a candidate for a functional
projection in the nominal domain (see for instance Wiltschko 2003 on
Halkomelem Salish, and Matthewson 2005 for a different view). Similarly,
in the same way that aspectual projections are postulated for the clause on
the basis of the aspectual inflection of the verb, some languages seem to
provide morphological evidence for aspectual morphology, hence aspectual
projections within the extended projection of N (Alexiadou & Stavrou
1998a; Alexiadou 2001a for Greek).
It is clear that in many cases the semantic argument and the morphologi-
cal argument will coincide, since a semantic concept will often have a mor-
phological expression, and an inflectional morpheme will usually have some
interpretative effect. They are, however, not identical. One case in point has al-
ready been mentioned: while in English the -s ending on plural nouns may be
directly linked to their interpretation, in that it encodes plurality, it is not clear
that the third person singular ending on English verbs has a semantic reflex.
The morphological argumentation is often further supported by the ob-
servation that a bound inflectional morpheme in one language corresponds
to a free morpheme in another language. Since the latter case would moti-
vate postulating a head position, one might invoke a similar position for
languages in which there is a bound morpheme. For instance, while English
uses a free morpheme (will) to express future time, French uses a bound
morpheme, the so called future tense. Thus the fact that one needs to postu-
late a position to host will in English could be used in support of postulat-
ing a similar head in French. However, this kind of reasoning presupposes
that one assumes a universal hierarchy of projections (cf. Cinque 1999).
The universal hierarchy argument could be used in support of postu-
lating a head num within the noun phrase since there are languages such as
Gungbe, described by Aboh (1998), in which number is expressed by a
separate free morpheme:
26 Part I – Introduction

(17) távò xóxó dàxó éhè ló lε

table old big this the PLURAL
‘these big old tables’

Observe that the word order internally to the DP in this language is almost
a perfect mirror of that found in English. We will return to this example in
section 2.6.3.3.

(iii) Distributional/syntactic evidence

As discussed above, the distribution of the lexical head within the constitu-
ent which it heads may also be interpreted as evidence for functional pro-
jections. Two types of argumentation are relevant here. These were formu-
lated by Taraldsen (1990). The first type of argument essentially relates to
the distribution of heads and is based on three widely accepted axioms of
the Government and Binding model: (i) a head can only move to a head
position; (ii) every head X0 is the head of maximal projection Xn; and (iii) a
moved constituent must c-command its trace (Taraldsen 1990: 85–86). The
second type of argument relates more to the distribution of XPs and is
based on the premise that every Xn dominates at most one specifier (see
Taraldsen 1990 for details). Let us consider some illustrations.
Consider the first type of argumentation. The distribution of the lexical
verb with respect to adverbial adjuncts and to markers of sentential nega-
tion shows that the verb cannot always be assumed to remain in its base
position. This is illustrated by the contrast between English (18a) and
French (18b)9:

(18) a. Nelson always eats biscuits.

b. Nelson mange toujours des croquettes. (French)
Nelson eats always biscuits

In (18a) the verb eats is adjacent to its direct object biscuits. We might as-
sume that it occupies its base position in the VP. In the French example
(18b), the verb mange is separated from its object des croquettes (‘bis-
cuits’) by an adverbial adjunct, toujours (‘always’). This suggests that in
(18b) V is not inside VP but has moved leftward. If V moves then we must
conclude that there is a landing site available, i.e. we must postulate a func-
tional head. In English, there is no evidence of this kind, because lexical
———–——————————
9
See Emonds (1978) for a first discussion.
 The theoretical framework 27

verbs fail to occupy displaced positions, but auxiliaries seem to be able to

occupy different positions, suggesting that they move. Thus in (18c) the
non-finite auxiliary have occupies a lower position than its finite counter-
part in (18d).

(18) c. Nelson will already have eaten the biscuits.

d. Nelson has already eaten the biscuits.

By analogy, evidence that N may occupy more than one position in the
nominal constituent could lead us to assume N-movement and hence to
postulate specific head positions as landing sites for N. We have already
briefly discussed one example of this type in section 1, see the data in (4).
The second line of argumentation concerns the distribution of maximal
projections and is invoked when two constituents in an extended projection
seem to have specifier properties. Such evidence will lead to postulating
two specifiers, hence two heads. In other words, in addition to a lexical
head, which can provide one specifier slot, at least one functional head is
required in order to provide the second specifier.10 For instance, it has been
observed that in some languages subjects may occupy different positions in
the clause. In (19) we illustrate the case of Dutch:

(19) a. Dat er morgen drie studenten vertrekken.

that there tomorrow three students leave
‘That there are three students leaving tomorrow.’
b. Dat drie studenten morgen vertrekken
that three students tomorrow leave.
‘That three students are leaving tomorrow.’

In (19a) the subject DP drie studenten (‘three students’) is adjacent to the

lexical verb vertrekken (‘leave’); in (19b) it is separated from the verb by
the adjunct morgen (‘tomorrow’). This might lead us to conclude that the
maximal projection drie studenten has undergone leftward movement. If
the movement of the subject in (19b) can be argued to target a specifier
position, then we need to postulate at least one functional head whose
specifier can host the moved DP. The functional head whose specifier is the
———–——————————
10
The argumentation is based on the assumption that each projection has just one
specifier. See section 2.5.3.2., however, for alternatives which would invalidate
this line of argumentation.
28 Part I – Introduction

canonical VP-external subject position has been identified as AgrS (but see
section 2.5.3.1) or as T. A similar point could be made on the basis of the
Icelandic examples (19c) and (19d):

(19) c. Hann las ekki baekur. (Icelandic)

he reads not books
‘He doesn’t read any books.’
d. Hann las baekurnar ekki. (Icelandic)
he read the books not
‘He doesn’t read the books’

In (19c) the indefinite object DP baekur (‘books’) follows the marker of

sentential negation ekki, in (19d) the definite object baekurnar (‘the books’)
precedes it. The leftward movement of definite object has sometimes been
referred to as ‘object shift’ (Holmberg 1986). If object shift in (19d) can be
argued to target a specifier position, then we need to postulate an additional
functional head whose specifier can host the moved DP. The functional
head whose specifier hosts a moved object has sometimes been identified
as AgrO (Belletti 1990; Chomsky 1991, 1995) (but see 2.5.3.1).
Again, if we observe that DP-internally, maximal projections may oc-
cupy different positions this can motivate postulating specifier positions,
and by implication it provides indirect evidence for functional projections.
As we will also discuss in Chapter 1 of Part III, the position of sentential
adverbials has also been interpreted as evidence for postulating functional
projections in the clause. It has been proposed that adverbial modifiers are
the specifiers of specialized projections. For instance in (20a) the adverbs
frequently and viciously have been argued to be specifiers of functional
projections.

(20) a. Mary frequently viciously criticized John.

(20b) is a nominalization related to (20b): the adverbials in (20a) corre-

spond to adjectives (frequent, vicious) in (20b):

(20) b. Mary’s frequent vicious criticism of John

If adjectival modifiers in the DP are seen as the analogies of adverbial

modifiers in the clause, then again the functional projections postulated for
hosting adverbial adjuncts in the clause could be replicated in the extended
projection of the noun where they would host adjectives.
 The theoretical framework 29

Using the type of argumentation sketched above, research on the struc-

ture of the clausal domain has provided us with a very rich inventory of
functional projections.
Initially, standard generative approaches to clause structure propose that
in the build-up of the clause three distinct layers can be distinguished. (i)
The VP layer is projected around the lexical verb. This layer is the semantic
core of the clause: it contains the predicate and its arguments. (ii) The IP
layer is projected around the inflectional head (I), which encodes modal,
temporal and aspectual properties of the clause. (iii) The CP layer is the
interface between the propositional content of the clause and the context: it
is projected on the basis of the position C, which hosts, among other things,
subordinating conjunctions such as that or if. 11 The CP layer is often re-
ferred to as the ‘left periphery’.
It is assumed that the subject originates VP internally (see section 2.4.2).
In English the subject moves to the specifier of IP, represented as SpecIP.
Originally the requirement that the subject move to the specifier of IP (i.e.
SpecIP) was referred to as the ‘Extended Projection Principle’. Nowadays
in the Minimalist tradition the label EPP is used more widely to refer to the
fact that a particular head requires a specifier. Such a head is then said to
have an EPP feature. We return to the movement of the subject in section
2.5.2.2.1.
(21a) is a schematic representation. For ease of exposition we do not
decompose the transitive VP into vP and VP. We represent nominal projec-
tions as ‘DP’.

———–——————————
11
In main clauses C is either non-overt (ia) or it may be filled by the auxiliary in
contexts of subject-auxiliary inversion (ib). In (ib) the auxiliary will has moved
from its position IP to the position C. It leaves a coindexed trace in its original
position.
(i) a. [CP [IP I will [VP talk to John]]]
b. [CP Willi [IP you ti [VP talk to John]]]
As mentioned before, in the representations, the symbol t stands for the ‘trace’
of the moved constituent, with which it is coindexed. For instance, ti is the trace
of the fronted auxiliary will in (ib). (ic) uses the strikethrough notation
(i) c. [CP Will [IP you will [VP talk to John]]]
30 Part I – Introduction

(21) a. CP

Spec C’

C IP

DP I’

I VP

DP V’

V DP

That John will John buy your book

Later work in the wake of Pollock (1989) suggests that the clause structure
is more richly articulated than this. For instance, it has been argued that IP
should be decomposed into the components T (tense) and Agr (agreement).
We refer the reader to section 2.5.1. for a brief summary of the argumenta-
tion. See also Pollock (1989, 1997). For a critical discussion see also Iatridou
(1990).
Further comparative research has revealed the need for postulating addi-
tional functional nodes in the domain between V and C, e.g. Mood, Aspect
and Voice. On the basis of this, we end up with a rich clause structure in
which IP is argued to decompose into at least the following projections:

(21) b. MoodP > AgrP > NegP > TP > AspP vP/VoiceP VP

It has also been argued that CP should be decomposed into different func-
tional projections. In particular, on the basis of a range of theoretical and
empirical considerations of the same nature as those discussed above, Rizzi
(1997) proposes that the head C (cf. (21a)) be decomposed into a number of
separate projections. In addition to a Force head, associated with encoding
illocutionary force, and a Fin head, which characterizes the morphological
properties of the complement clause, the CP domain may also contain a
unique Focus projection, FocP, whose specifier hosts the focalized con-
stituent and whose head hosts an abstract Focus-feature, and a recursive
 The theoretical framework 31

Topic Projection, whose specifier hosts a topicalized constituent and whose

head hosts a Top feature.

(21) c. ForceP > TopP* > FocP > TopP* > FinP

Given the discussion above, clauses are interpreted as extended projections

of V, i.e. projections of V augmented with a range of functional projections
(see Grimshaw 1991 for the notion of extended projection). Obviously, once
we assume that there is a wide range of functional projections dominating
VP, the question can be raised whether there are also functional projections
that dominate NP within the extended projection of N, and whether one can
identify the same type of functional projections in the nominal domain. Part
II of this book mainly deals with this question.

2.5.2. Functional projections, movement and agreement

2.5.2.1. Features and agreement

Functional projections are projections of functional heads. In the clausal

domain, a functional head, say T, will select an extended projection of V as
its complement. T, for instance, selects AspP. A functional head can attract
a lower head, for instance, T may attract V. As each projection contains a
specifier position, these specifiers will provide us with additional positions
which are the landing site for movement (see our earlier reference to
Taraldsen 1990). For instance, the specifier of TP is available for move-
ment. With respect to the clause, two types of movement have been distin-
guished in the literature: (i) head movement and (ii) movement of maximal
projections. The status of head movement is unclear in current versions of
the Minimalist Program. We do not dwell on this very much here. The
reader is referred to (Chomsky 2001; Lechner 2005) for extensive discus-
sion.
The Minimalist Program attributes an important role to features in the
derivation of the sentence. Features basically drive the concatenations of
elements that will build up the sentence. Heads (both lexical and func-
tional) may be associated with features. For instance, as we have seen al-
ready, both verbs and nouns may be associated with agreement features
such as number. Some features are said to be interpretable or valued, others
are not interpretable or unvalued. As discussed above (see the discussion of
32 Part I – Introduction

(10a, b) in section 2.3.) the feature Number is interpretable (or valued) on

nouns, whereas it is uninterpretable (or unvalued) on verbs. Uninterpretable
features are not tolerated by the system and must be eliminated by check-
ing. Alternatively, in a system that uses feature valuation, unvalued features
must be valued.
In principle, the checking of uninterpretable features – or, in the alterna-
tive approach, valuation of unvalued features – can be achieved without
movement. Let us briefly outline how both of these systems work. Say a
particular functional head contains features that are [–interpretable] or
[–valued]. If they remain in the derivation the structure cannot converge.
For [–interpretable] features to disappear, they must be matched with corre-
sponding [+interpretable] features. The very presence of uninterpretable
features renders them active, so that they can search or ‘probe’ in their c-
command domain for matching interpretable features. Once such features
are located on a goal, they are matched with the uninterpretable features of
the probe, matching leads to agreement, agreement will check and eliminate
an uninterpretable feature.
Observe that the presence of an uninterpretable feature on a probe does
not irrevocably lead to movement. Whether or not movement also occurs
depends on some other property of the system. For instance, if a head car-
ries an EPP feature, this feature will trigger movement.
In an alternative formulation features are [+/– valued]. [–Valued] features
must be valued, that is to say they must receive a value. Again, in order for
unvalued features to receive a value they must enter an Agree relation with
a suitable goal, which will contain matching valued features. Again, the
presence of unvalued features renders them active, so that they probe in
their c-command domain for matching features. Once such features are
located on a goal, they are matched with those of the probe; matching leads
to agreement, and hence valuation of the unvalued feature. Again, whether
or not movement also occurs relates to other properties of the system such
as the presence of EPP features. 12

We have already alluded to the contrast between intrinsic features and non
intrinsic or optional features. This point will be relevant when we discuss
the functional projections in the nominal domain. Intrinsic features are

———–——————————
12
Whichever system (valuation or checking) is adopted, it is also clear that move-
ment must be triggered. A constituent will not move without such a trigger. As
mentioned in section 2.1., movement is a last resort operation.
 The theoretical framework 33

those features that are an inherent inseparable part of a lexical item. Non-
intrinsic or optional features are those features that can be varied. That is to
say, their value can be chosen and this choice is made via the Numeration,
the set of items which constitute the building blocks for the derivation. For
instance, as we will discuss in detail in Part II, Chapter 3, whereas Gender
is an intrinsic feature of the nouns, Number is an optional (or non-intrinsic)
feature. Number is a category the values of which (singular/plural or other)
can be chosen, or put differently, Number features are varied. Gender, as a
rule, cannot be chosen: its values form part of the noun itself.

2.5.2.2. Types of movement

In the Government and Binding tradition, two types of XP movement are

postulated: (i) A-movement and (ii) A'-movement. For discussion of the
contrast we refer to standard textbook introductions such as Haegeman
(1994) and Radford (1998). The contrast between the two types of move-
ment has so far been maintained in the Minimalist tradition. We give a brief
overview of how movement operates. Observe, though, that the discussion
below is a simplification and that there are many different implementations
of the fundamental ideas.

2.5.2.2.1. A movement
It is assumed that the clausal subject DP originates in a VP-internal position.
For transitive verbs this is the specifier position of vP. However, it is clear
that the subject DP, Nelson in (18a) for instance, does not remain VP-inter-
nally. If it did, we would expect it to be adjacent to the lexical verb eats. In
order to account for the fact that the subject is separated from the VP domain
and ends up in the canonical subject position, SpecTP (or SpecIP), it is as-
sumed that the subject has to undergo leftward A-movement.
If the subject moves to SpecTP then we can assume that the trigger for
the movement is an uninterpretable feature located on T. What could this
feature be? We have proposed that nominal projections have interpretable
[Number] features, while the Number features associated with verbs are
uninterpretable. Observe that Number inflection on verbs is also a function
of finiteness: in English and in French only tensed verbs can be associated
with Number. Let us assume that the uninterpretable Number feature of the
verb is encoded on Tense. Thus for the derivation to converge we must
eliminate this uninterpretable feature on T. Recall that the presence of unin-
34 Part I – Introduction

terpretable features renders them active, so that they probe for matching
features in their c-command domain. So the uninterpretable number feature
on T will search for a goal with a matching feature in the clause. The sub-
ject DP, in SpecvP, is such a goal. Once the matching interpretable feature
is located on the goal, it is matched with that of the probe, matching leads
to agreement and leads to the elimination of the uninterpretable Number
feature.
Observe that the subject DP does not move to SpecTP because of the
presence of the uninterpretable number feature. Rather it is assumed that T
has a so-called EPP feature, and it is this feature which requires the filling
of the specifier of T. (22) summarizes the derivation:

(22) a. TP

T’

T vP
[uN, T, EPP]
DP v’
[N]
v DP
Nelson eats biscuits

b. TP

Spec T’

Nelson T vP
[N] [uN, T, EPP]
DP v’
[N]
v DP
Nelson eats biscuits

The moved DP Nelson leaves a copy in its original position: this is repre-
sented by strikethrough in (22b).13

———–——————————
13
Recall that in the Minimalist tradition, copies replace the earlier concept of traces
(see section 1).
 The theoretical framework 35

In (23) we illustrate A’ movement. A constituent of the clause has moved

to the left periphery: the leftward movement of interrogative constituents
how important in (23a), and what in (23b), marks the clauses as questions.

(23) a. [CP [DP What] will [IP the cat will eat what]]?
b. [CP [DP How important] will
[IP the movement will become how important]]?

The system elaborated above will also be implemented to account for A’

movement. In particular, for movement of interrogative constituents it could
be assumed, for instance, that their interpretable [WH] feature can check the
uninterpretable [WH] feature on C. Once again, movement as such will be
triggered by an additional EPP feature on C.

2.5.2.3. Features and movement in the nominal projection

If we assume the mechanisms for movement outlined above for the clause
then the question will arise whether they are also applicable in the nominal
projection. We will return to the concept of head movement in Part II,
Chapter 1 and in Part III, Chapter 1. At various points in the book we will
also turn to the issue of DP movement within the nominal projection.

2.5.3. Challenging functional projections

2.5.3.1. AgrP

In section 2.5.1. we saw that morphological evidence has been used to pos-
tulate functional projections. By this reasoning, the fact that verbs are in-
flected for agreement had led to the assumption that the functional domain
of the clause contains an Agreement projection, AgrP (see (21b), Pollock
1989; Chomsky 1991). Initially, support for AgrP was also provided on the
basis of the distribution of finite verbs. For instance, based on the contrast
between the finite verb and the infinitive in French, Pollock (1989) con-
cludes that IP must be split into at least two projections, which he labels TP
and AgrP. The data are provided in (24):
36 Part I – Introduction

(24) a. Jean ne mange pas souvent de chocolat.

Jean NEG eats not often chocolate
‘Jean doesn’t often eat any chocolate.’
b. Ne pas souvent manger de chocolat, c’est triste.
NEG not often eat chocolate, it is sad.
‘Not often eating chocolate is sad.’
c. Ne pas manger souvent de chocolat, c’est triste.
NEG not eat often chocolate, it is sad.
‘Not often eating chocolate is sad.’

We see that in (24a) the finite verb mange (‘eats’) precedes the marker of
sentential negation pas as well as the adverb of frequency souvent (‘often’).
This order can be derived if we assume that the verb moves from its base
position to a higher functional head. In (24b) the infinitive manger (‘eat’) is
adjacent to its object de chocolat (‘chocolate’) and follows the adverb sou-
vent. Arguably it occupies a position in the VP. But in (24c) the infinitive is
found between pas and souvent: this suggests that it is not VP internal, nei-
ther does it occupy the functional head position which it occupies in (24a).
We conclude that there must be another landing site for V, between the
negation marker and the adverb. In other words, IP decomposes in at least
two projections. Pollock (1989) proposes that TP dominates AgrP. Based
on morphological evidence, however, Belletti (1990) proposes that AgrP
dominates TP (see also Pollock 1997 for a refutation).
However, consider what it would mean to assume a projection AgrP in
terms of the checking theory we have outlined above. Assuming that AgrP
dominates TP, we would assume that the subject DP ends up in SpecAgrP,
that ‘verbal’ agreement features on Agr, such as Number, are [–interpret-
able] (or unvalued) and that the agreement features on the noun (Number,
say) are [+interpretable] (or valued). The [–interpretable] features on Agr
will be a probe searching for a matching interpretable feature in the c-
command domain: this search will locate such features on the subject DP in
SpecvP and by agreement the uninterpretable features on Agr will be
checked and deleted. As a result, though, Agr, which by hypothesis only
contains uninterpretable agreement features, would really have no features
left any more.
In early versions of Minimalism (Chomsky 1991), uninterpretable fea-
tures, such as agreement features associated with the verb, were taken to be
able to project their own functional category. This view has subsequently
been called into question (Chomsky 1995: Chapter 3) and such features are
 The theoretical framework 37

now often taken to be licit only when associated with heads that also have
interpretable features. So, for instance, the uninterpretable agreement fea-
tures associated with the verb are located on Tense, which itself also has
the interpretable Tense feature. 14

2.5.3.2. Multiple specifiers

We have also seen that the distribution of maximal projections can be the
basis for postulating functional heads. For instance, we may observe that
there is a need for two specifier positions in a particular domain. Assuming
that a lexical head can have only one specifier, then, if there is a second
specifier position, we are led to assume that there will be a second func-
tional head. However, this argumentation can also be challenged. In par-
ticular the restriction that each head has one specifier is not universally
accepted and it has been proposed that a head might have more than one
specifier. For instance, Koizumi (1995: 141) proposes that the CP domain
contains one functional projection PolP, ‘Polarity Phrase’. Pol selects IP as
its complement; the head Pol can host a number of different features. Each
feature requires checking and if each feature is associated with the EPP
property, then this leads to multiple movement and to multiple specifiers.
The checking features of Pol are hierarchically ordered: the focus-feature or
the wh-feature is checked in the inner specifier and the topic feature is
checked in the outer specifier. In (25a) the complementizer che (‘that’) is
followed first by a topicalized constituent a Gianni (‘to Gianni’), which is
followed by a focused constituent, il tuo libro (‘your book’) and followed
by an adjunct of time domani (‘tomorrow’). It could be argued that a head
Pol takes IP as its complement and that this head hosts the relevant features
(FOCUS, TOPIC etc) to attract the constituents in the left periphery. Simi-
larly, in French (25b) the topicalized constituent ce livre-là (‘that book’)
precedes the focused interrogative constituent quand (‘when’). Again Pol
could be argued to have a TOPIC feature and a FOCUS feature. Thus in both
(25a) and (25b) Pol would have multiple specifiers.

———–——————————
14
The question whether clausal agreement projections should be admitted has not
been given a final answer. For arguments in favor of agreement projections see
also Belletti (2001), Guasti and Rizzi (2002), Neidle and MacLaughlin (2002),
Pollock (2006: 644, note 25).
38 Part I – Introduction

(25) a. Dicono che a Gianni IL TUO LIBRO domani gli dovremmo dare
(Italian)
they say that to Gianni YOUR BOOK tomorrow him we-should give
‘They say that tomorrow YOUR BOOK we should give to Gianni.’
b. Ce livre-là, quand l’as-tu acheté ? (French)
This book there, when it have you bought
‘This book, when did you buy it?’

In this way the system ensures that more than one maximal projection is
associated with the CP domain without associating each moved constituent
with a separate head. Rather than having an array of functional projections
in the CP domain, as proposed by Rizzi (1997), and summarized in (21c)
above, there is just one single head with multiple specifiers. The hierarchi-
cal organization of feature checking in the C-domain mimics the hierarchy
of the functional projections postulated above.

2.6. Deriving variations in linear order

2.6.1. Cross-linguistic variation in linear order

So far we have mainly used data from English and French in which typi-
cally the head precedes the complement. For instance, a verb precedes the
direct object. However, it is well known that languages vary with respect to
the relative positions of heads and their complements. We have already
discussed the difference in word order between languages and to account
for that we have used head movement (see (18)). Observe that apart from
differing in V-movement, English and French are similar in the unmarked
positions of subject (Nelson, object (biscuits, des croquettes), and the fre-
quency adverb (toujours, always). The unmarked order is always subject
>adverb>object. The unmarked order is also that in which the verb (and the
VP) follows the auxiliary:

(26) a. Nelson has always eaten biscuits.

b. Nelson a toujours mangé des croquettes. (French)
Nelson has always eaten biscuits

However, other languages display other orders. For instance, Dutch embed-
ded clauses display the order object-verb, and the auxiliary may also follow
the verb:
 The theoretical framework 39

(26) c. Dat Nelson altijd koekjes gegeten heeft

that Nelson always biscuits eaten has
‘That Nelson has always eaten biscuits.’

One approach to such cross-linguistic variation has been to propose that

there is parametric variation in the directionality of the projection schema
and that the structure of Dutch is to some extent the mirror image of English
and French. More precisely, it has been proposed that the X-bar schema as
elaborated in section 2.2. (see (9)) only specifies hierarchical relations, and
that it does not provide information concerning linearity. Thus the schema
in (9), should in fact be read as allowing both specifier-head order and
head-specifier order, and as allowing both head-complement order and
complement-head order. It is proposed that the ordering variation is a mat-
ter of parameter setting. Thus it could be said that in English specifiers con-
sistently precede their heads, complements consistently follow them and
that in Dutch there is some variation: while C, D and P, for instance, pre-
cede their complements, I and V follow them.15

(27) a. English; Dutch CP, DP,PP: b. Dutch IP and VP:

XP XP

Spec X’ Spec X’

X complement complement X

Implementing this variation on the structure of the clause, for instance, we

could then end up with the structure in (27c) for Dutch. (27c) is very
sketchy. In particular we leave aside all articulation in the TP domain, we
leave aside VP-shells, we insert the auxiliary in T and we adjoin the adverb
to VP.

———–——————————
15
Needless to say, the fact that one has to stipulate which projections are head
initial and which are head final is not attractive.
40 Part I – Introduction

( 27) c. CP

C TP

DP T’

VP T

Adv VP

DP V’

DP V

dat Nelson altijd Nelson koekjes gegeten heeft

Consider now the Malagasy examples (28a), taken from Pearson (1998: 2),
his (8), and (28b) taken from Rackowski & Travis (2000: 120), their (6):

(28) a. Nijinja vary tsara ny mpamboly.

past-cut rice well det farmer
‘The farmer harvested rice well.’
b. Manasa lamba tsara Rakoto.
wash clothes well Rakoto
‘Rakoto washes clothes well.’

What is striking about these examples is that the verbs (nijinja (‘cut’), ma-
nassa (‘wash’)) are sentence-initial and that we find the order object-adverb-
subject. We might wish to derive the Malagasy examples by V-movement.
However, simple verb movement is not sufficient since this will not give us
the right ordering of object-adverb-subject. In fact, with respect to the rela-
tive order of the non-verbal constituents in the clause, Malagasy presents
the mirror image of English and French. One might once again propose that
there is parametric variation in the directionality of the projection schema
and that the structure of Malagasy is the perfect mirror image of English
and French. Thus while English (and French) have the ordering in (29a),
Malagasy would have (29b):
 The theoretical framework 41

(29) a. XP b. XP

Spec X’ X’ specifier

X complement complement X

Implementing this variation on the structure of IP, for instance, we end up

with the structure in (30a) for a consistently right-branching language, and
with that in (30b) for a consistently left branching language. We simplify
the structure by ignoring the split IP and VP-shells.

(30) a. TP

DP T’

T VP

DP V’

V DP
b. TP

T’ DP

VP T

V’ DP

V DP

If we also assume that the Malagasy subject DP moves rightward, from

SpecvP into SpecTP then example (30a) would fit the right-branching
structure in (30b).

2.6.2. Antisymmetry and linear order

As an alternative to the directionality parameter to account for linearization

differences between languages, Kayne (1994) proposes the universal base
42 Part I – Introduction

hypothesis, the proposal that the system builds up identical structures across
languages and that the universal schema is the X-bar format presented in
section 2.2. (9).16 All variation in linear order is derived by movement. Let
us first briefly summarize the essence of his proposal.

2.6.2.1. Antisymmetry

Kayne proposes that linear ordering is fully determined by structural hier-

archy. His Linear Correspondence Axiom (LCA) states that only antisym-
metric relations are admitted between nodes in a structure, hence the label
‘antisymmetry’. This means that if a node α c-commands node β then β
must not c-command α.17 Mutual c-command between two nodes is sym-
metric, violating antisymmetry. Structural c-command maps into a left-right
linear ordering. Hence, specifier head and head complement are the only
possible base orders, and all variation in which, say, a head precedes a
specifier and a complement precedes a head are derived by movement.
Moreover, since a moved element targets a c-commanding position, all
movement is to the left. Thus neither the base structure in (30b) nor the
required rightward movement of the subject would be admitted under his
view (for discussion see Beerman et al 1997). The derivation of the English
pattern is not problematic, nor is that of French, in which we continue to
assume that V moves to an inflectional head.

2.6.2.2. Deriving OV-orders

Assuming the X-bar framework as in (9) above for Dutch has the advantage
that we no longer need to stipulate which projections are head initial (CP,
DP, PP) and which are head final (IP, VP): all projections are head initial.
However, how would we derive the order of Dutch embedded clauses in
which the object precedes the verb? There have been a number of proposals
in the literature, and for reasons of space we cannot elaborate them all.
Here we will just look at the derivation of the OV order.
One proposal is that the OV order in Dutch (and German) is derived by
the movement of the object to the right. One implementation of this idea is
to propose that the object DP moves to the outer specifier of vP. Thus in
———–——————————
16
Kayne does not allow for adjunction or for multiple specifiers.
17
For the concept ‘c-command’ see section 2.2.
 The theoretical framework 43

(31a), the direct object koekjes (‘biscuits’) originates to the right of V and
moves leftward, as schematically presented in (31b). Observe that the sen-
tence-final position of the verb suggests that it remains VP-internal (see
Zwart 1993, 1996, 1997 for detailed proposals):

(31) a. Dat Nelson altijd koekjes eet.

that Nelson always biscuits eats
b. [CP dat [TP Nelson [vP altijd [vP koekjes [vP Nelson eet [koekjes]]]]]]

In independent work, Hinterhölzl (2000), Pearson (1998, 2000), Koopman

and Szabolcsi (2001) and Haegeman (2000, 2001) have elaborated an alter-
native proposal to derive the OV order in Dutch and German. The accounts
involve a double movement. Rather than assuming that SOV orders reflect
a low V-position with movement of the complement to a leftward position,
they propose that the OV order is derived by

(i) movement of the finite verb to a functional head in the I domain

(ii) ‘remnant’ movement of the (extended) projection of V to a specifier
position

The second step of the derivation is called ‘remnant movement’ because the
movement affects a ‘remnant’, i.e. it affects a projection from which a con-
stituent (here the head V) has been moved first. Below is a schematic repre-
sentation. The structure is simplified for expository reasons. In (31c) V eet
moves to F, a functional head in the IP domain. In (31d) the remnant pro-
jection moves to the specifier position of the inflectional projection headed
by F. Continuing to assume for expository reasons that adverbials may ad-
join to vP, we label the remnant projection vP. Furthermore, the subject DP
Nelson will have to move to a higher position. This is shown in (31e).

(31) c. dat [FP [F eet]

[vP altijd [vP Nelson [V eet] koekjes]]]
d. dat [FP [vP altijd [vP Nelson [V eet ] koekjes]] [F eet]
[vP altijd [vP Nelson [eet] koekjes]]]
e. dat [Nelson [FP [vP altijd [vP Nelson [V eet koekjes]] [F eet]
[vP altijd [vP Nelson [eet] koekjes]]]]

How can we derive the word order pattern in a language like Malagasy, il-
lustrated in (28)? Recall that in this language the line-up of the constituents
44 Part I – Introduction

of the clause is the mirror image of that found in English. We will not pro-
vide a precise or detailed analysis, here, but we will simply show the spirit
of an analysis in terms of Kayne’s antisymmetry.
We assume the X-bar format in (9a), with all projections head-initial, and
with specifiers to the left of the X’ constituent. If for Malagasy we adopt
the English type of derivation with the subject moving to the highest A-
position, the specifier of TP, and the verb remaining in V, then we would
end up with the order in (32a), clearly not what we want.

(32) a. *[TP Ny mpamboly [vP tsara [vP ny mpamboly nijinja vary]]]

the farmer well cut rice

If in addition to moving the subject to the higher specifier position, we also

move the verb to the highest inflectional head, as happens in French (see
example (18b)), we get (32b), also not the desired order.

(32) b. *[TP Ny mpamboly [T nijinja] [vP tsara [vP ny mpamboly nijinja vary]]]

As a third alternative, we might propose that V moves while the subject DP

remains VP-internal, but then we get the order in (32c).

(32) c. *[TP [T Nijinja] [vP tsara [vP ny mpamboly nijinja vary]]]

Again, though the verb is now indeed initial, this derivation does not pro-
duce the desired order, notably, the subject ny mpamboly now incorrectly
precedes the object vary, and the object incorrectly follows the adjunct
tsara. One option would be to propose that the object vary first moves to a
position to the left of the adjunct tsara.

(32) d. [TP [I Nijinja] [ vary [vP tsara [vP ny mpamboly nijinja vary]]]]

This movement of the object might at first sight be argued to instantiate

object shift as also found in Icelandic and illustrated in (19c,d), repeated
here in (33).
 The theoretical framework 45

(33) a. Hann las ekki baekur. (Icelandic)

he reads not books
‘He doesn’t read any books.’
b. Hann las baekurnar ekki. (Icelandic)
he read the books not
‘He doesn’t read the books’

However, there is a problem with the proposal that the Malagasy object
undergoes object shift. In the Malagasy example, object shift of vary (‘rice’)
would have to move an indefinite object past the manner adverb tsara
(‘well’). In languages exhibiting object shift, such as Icelandic, it is typi-
cally the definite object which undergoes object shift, as shown by the con-
trast in (33): in (33a) the indefinite object baekur (‘books’) remains to the
right of the sentential negator ekki.
Indeed, the proposed movement of the indefinite object in Malagasy
(32d) becomes even more questionable when we compare this example with
(34), in which the object ny vary (‘the rice’) is definite. As can be seen, the
definite object ny vary occupies a position to the left of the subject and to
the right of the manner adverb tsara.

(34) a. Nijinja tsara ny vary my mpamboly. (Pearson 1998: 3)

cut well the rice the farmer

In our derivation, this would have to mean that the definite object has
moved to a position to the immediate left of the subject and to the right of
the manner adverb:

(34) b. [TP [I Nijinja] [vP tsara [vP ny vary [vP ny mpamboly nijinja ny vary]]]]

While this derivation does produce the desired order, it goes against most
assumptions as to the motives for object shift. As shown by the contrast in
(19c,d) repeated here in (33), in general, in languages with object shift,
indefinite objects occupy a lower position (33a) than definite objects (33b).
This is usually related to their interpretation, definite objects expressing some
‘given’ entity (Diesing 1996, 1997). We will not elaborate the details of the
analyses of this phenomenon here. Again then, we should say that somehow
object shift in Malagasy is the mirror image of object shift in Icelandic.
If we continue to assume with Kayne (1994) that only left branching is
possible, further examination of additional examples raises more problems.
46 Part I – Introduction

Suppose we take a sentence with two adverbials in the TP domain between

subject and verb. In French the unmarked order of such adverbials is that
the frequency adverbial toujours (‘always’) precedes the manner adverb
bien (‘well’) and this pattern is generally the unmarked case. For extensive
discussion of adverbial order see Alexiadou (1997) and Cinque (1999).

(35) a. Le paysan coupe toujours bien le riz.

the farmer cuts always well the rice

The Malagasy example (35b) seems to present the mirror image of French
(35a).

(35) b. Nijinja vary tsara foana ny mpamboly.

cut rice well always the farmer (Pearson 1998: 27)

In Malagasy, the frequency adverbial foana (‘always’) follows the manner

adverbial tsara (‘well’). It is hard to see how this order can also be derived.
Suppose, following proposals by Alexiadou (1997) and Cinque (1999), that
adverbials are not simply vP adjoined, as we have been implying so far, but
that they are associated with specific functional projections. Let us say that
the adverbial of frequency is associated with an aspectual projection and that
the manner adverbial is associated with a manner projection (or, possibly,
with VoiceP/vP). Using the mechanism which successfully derived (32d)
we would still end up with the reverse order, as shown in (35c).

(35) c. *[TP [T Nijinja] [vary [AspP foana

[vP tsara [vP ny mpamboly nijinja vary]]]]]

To derive the desired pattern and assuming that adjuncts display a universal
hierarchy which is reflected by their position as specifiers of functional
heads, we would now have to propose that tsara, the manner adverb, also
moves leftward, to a position lower than the moved indefinite object.

(35) d. *[TP [I Nijinja] [vary [ tsara [AspP foana [vP tsara

[vP ny mpamboly nijinja vary]]]]]]

Indeed, from the consideration of additional empirical data it turns out that
we have somehow always to reorder all clausal constituents in Malagasy.
 The theoretical framework 47

For instance, in the double object construction associated with ditransitive

verbs such as give, the indirect object DP usually precedes the direct object
DP as shown by the examples in (36). As shown by (37), Malagasy again
exhibits the opposite order:

(36) a. English John gave Nelson biscuits.

b. Dutch Jan gaf Nelson koekjes.
John gave Nelson biscuits

(37) Nanolotra ny dite ny vahiny ny zazavavy. (Pearson 1998: 2, his (2a))

PAST-offer the tea the guest the girl

Again, to derive (37) we will have to assume that the direct object DP is
obliged to move past the indirect object DP, a pattern which is again most
unusual.
In the various proposals above we have applied the two types of move-
ment; (i) head movement affecting V and (ii) XP movement affecting a con-
stituent of VP or of the clause, such as an object DP, or an adjunct or a sub-
ject. A combination of such movements was also often used. Though we
were able to derive the correct linear orders, each derivation presented us
with an exceptional situation. Notably, definite objects have to remain lower
than indefinite ones, adverbs reorder with respect to each other, and direct
objects must move higher than indirect objects. This type of derivation does
have the advantage of preserving the universal base hypothesis, but it is
unsatisfactory because we require a whole range of unexpected additional
movements.
Pearson (1998, 2000), Rackowski & Travis (2000) and Travis (2006)
propose an alternative derivation for the Malagasy data. We will present the
spirit of their analyses here. Observe that the presentation below does not
correspond to the exact analyses cited. What we want to do is to merely
illustrate in broad lines the alternative proposal as introduced in the papers
referred to. For the detailed and accurate implementation elaborated by the
authors we refer to their own papers. What we need to achieve is that the
verb is in initial position and that all constituents end up in the reverse pat-
tern. Leaving out details which would complicate the picture somewhat, the
essence of Pearson’s proposal is that the initial V-position is not derived by
head movement of V, but rather that the pattern in Malagasy is derived by
the movement of maximal projections, that is VP and extended projections
of VP.
48 Part I – Introduction

Let us start from a schematic structure as (38a).

(38) a. [vP subject [VP verb object]]

Suppose that Malagasy lacks V-to-I movement. To get V to end up in the

leftmost position, we move (extended) projections of V. First VP, compris-
ing V and the object, moves to the left, past the subject DP, to the specifier
of a functional projection FP1. This leads to a linear order – verb-object-
subject as in (38b).

(38) b. [FP1 [VP verb object ] [vP subject [VP verb object ]]]

Then we add the projection whose specifier hosts the manner adverb
‘MannerP’.

(38) c. [ManP adverb [FP1 [VP verb object ] [vP subject [VP verb object ]]]]

The projection FP1, whose specifier hosts the moved VP, moves past the
manner adverb to a specifier position of a higher functional projection FP2.
This movement will also take along vP with the subject. The output is
(38d).

(38) d. [FP2 [FP1 [VP verb object ] [vP subject [VP VP ]]] [ManP adverb [FP1FP1 ]]]

To avoid overly complex representations we will not represent the lexical

elements contained in the copy: thus, for instance, while in (38b) we do
signal the copy of VP by verb object, in (38d) we simply signal the relevant
copy by VP, and we represent the copy of the moved FP1 by FP1.
We will assume that the subject moves to the specifier of an inflectional
projection, represented here as TP. This results in the order subject-verb-
object-adverb (38e).

(38) e. [TP subject [FP2 [FP1 [VP verb object ] [vP subject [VP VP ]]]
[ManP adverb [FP1 FP1 ]]]]

Finally the projection FP2, whose specifier hosts the moved extended VP
(FP1), moves again as a whole past the subject to a higher functional pro-
jection (38f).
 The theoretical framework 49

(38) f [FP3 [FP2 [FP1 [VP verb object ] [vP subject [VP VP ]]]
[ManP adverb [FP1 FP1 ]]] [TP subject] [FP2 FP2]]

This derivation gives the desired output verb-object-manner adverb-subject.

In the derivation proposed there is no instance of V-movement. Instead, we
repeatedly apply what we could call a snowballing XP movement18, where
one projection, say VP in (38a), moves into the specifier of a higher projec-
tion, FP1 in (38b). Next, the entire projection whose specifier has just been
targeted by XP-movement, i.e. FP2, itself moves on, etc. As a result of this
snowballing movement operation we end up with a mirror image pattern,
where English has Su-adverb-V-O and Malagasy has V-O-Adv-Su.
Obviously, the analysis outlined above is presented in a schematic way
and it raises many questions, which we will not address here. It succeeds in
deriving a word-order pattern that is the mirror image of that found in Eng-
lish. The central question is what motivates the movement of the extended
VP in the derivation. What would be, for instance, the trigger for the
movement of VP into [SpecFP1], of FP1 into [SpecFP2] and of FP2 into
[SpecFP3]?

2.6.3.3. Snowballing in the nominal projection?

With respect to word order patterns in the nominal domain, too, we find
languages displaying a pattern that is the mirror image of the English pat-
tern. Aboh (1998) gives example (39a) from Gungbe. Again we find the
reverse ordering of that found in English (39b):

(39) a. távò xóxó dàxó éhè ló lε

table old big this the PLURAL
‘these big old tables’
b. these big old tables

Aboh proposes that the Gungbe order in (39a) is derived by an application

of snowballing movement in the DP. We will come back to this issue and
to other proposals for snowballing movement in Part III, Chapter 1.

———–——————————
18
The term is due to Chris Collins.
50 Part I – Introduction

3. Summary

In this chapter we have touched upon several issues that pertain to the syn-
tax of clauses, that is extended projections of verbs. In particular we have
discussed the contrast between functional head/projection and lexical pro-
jection, the motivation for postulating functional projection (and for postu-
lating movement), concepts such as head movement, XP movement, snow-
balling movement, and feature valuation as a trigger for movement. We
have also briefly hinted at the fact that these issues are equally relevant for
the syntax of extended projections of nouns. In the remainder of this book
we will return to these issues in more detail and offer precise implementa-
tions and more fully elaborated analyses.
Part II

The functional make up of the noun phrase

Part II is concerned with the categories that form part of the extended projection of
the noun. Linguistic research over the last years has shown that the sentence can be
partitioned into three domains: the discourse domain (built around C), which is the
syntactic encoding of discourse-related information, the agreement domain (con-
taining various verb-related functional projections, such as agreement projections,
e.g. IP and/or TP), and the thematic domain which is built around VP-shells (Larson
1988) and encodes thematic information.
Another, by now well-established, hypothesis is that the structure of the nominal
functional domain can be conceived of as paralleling the structure of the clausal
domain, in that specific types of information are typically encoded in specific areas
of the extended projection of the noun and that the structuring of this information is
like that in the clause. It is proposed that the nominal domain consists of a determi-
nation area, which can encode discourse-relevant information, a functional ‘middle’
morphosyntactic area, in which, among other things, the agreement of the various
modifiers with the head noun is encoded, and a still lower area in which thematic
relations are established/licensed, the so-called Theta domain which is built around
NP in the narrow sense, that is, the projection of the lexical head. In this part of the
book, we will take up each of the first two areas in turn; we will return to the Theta
Domain in Part IV.
In the first chapter of Part II we consider the functions of the determination
area, which is headed by the determiner, D, as well as the way(s) it parallels par-
ticular areas in the clause, notably the CP area, and to some extent also the IP area.
We will take a closer look at the semantic import of D, its status across languages,
and we will examine the particular morphemes that are associated with it, articles
and demonstratives being the most prominent among them. In addition, we will ex-
amine to what extent the domain headed by D can encode discourse-related con-
cepts like Topic and Focus, thus bringing it in line with the C-related area in the
clause. In the literature the category D has been associated with a number of di-
verse properties. Among others, it has been considered as (a) the locus of the se-
mantic-pragmatic notion of (in)definiteness, (b) the natural host of the definite arti-
cle, (c) the part of the extended projection of N which is responsible for turning a
predicate, i.e. a noun, into a syntactic argument by anchoring it to the extralinguis-
tic context, (d) a case assigner. We will survey the main arguments for each of
these properties (a)–(d). The picture that will emerge from the various discussions
is in fact one that fails to distinguish the concept D as a syntactic position or as a
syntactic category, from the morpheme for the (definite) article which most fre-
52 Part II – The functional make up of the noun phrase

quently realizes it in many languages. The (definite) article itself has a privileged
position among determiners as the occupant par excellence of the position D, as
there undoubtedly exists a strong link between D and the article. This is the main
reason why in the vast literature on ‘DP’ it is often so difficult to tell whether (a)–
(d) are taken to be inherent properties of D as a position or a category, or of the ar-
ticle itself. In this chapter we will also discuss the structures that have been pro-
posed as an articulation of the relation between D and the article. We will further
compare the definite article with demonstratives in order to be able to assess the
proposals made with respect to the syntactic representation of the latter and we will
evaluate the claim made in the literature (Giusti 1997, 2002; Campbell 1996) that it
is the D(P) category that is semantically interpreted.
The examination of properties (a)–(d) will inevitably take us to an important is-
sue and one that has been occupying the literature for quite a while now: the issue
of articleless noun phrases as well as of articleless languages. We consider this issue
in Chapter 2. More precisely, the question that will emerge is the following: if it is
the definite article as such (rather than the syntactic category D) that is invested
with the various properties listed in (a)–(d), how do we account for the interpreta-
tion of ‘bare’ nouns or ‘bare’ noun phrases in languages that have articles, or,
equally, and perhaps more importantly, in languages that lack an article altogether,
such as, e.g., Latin and Russian? Our survey of the proposals that have been put
forward to explain how the nominal phrase receives its interpretation in the ab-
sence of the article will bring us in line with those who believe that the article has a
purely grammatical function to fulfil, while it is the properties of D that impart
such semantic notions as referentiality and definiteness to the interpretation of the
whole constituent. Crucially, we will arrive at the conclusion that semantic rules
apply to the DP area to interpret the nominal phrase, not simply to the overt material
under it.
The properties of the intermediate morphosyntactic or agreement area dominated
by DP and dominating the thematic domain, NP, will be the topic of Chapter 3. We
will review the evidence that has been adduced in support of inflectional projections
like NumberP, GenderP, WordMarkerP, but also of other functional categories that
are more ‘verb-like’, such as Tense, Aspect and Voice. These categories will be
further dealt with in more detail with particular reference to adjectives in Part III.
Chapter 1

The emergence and the structure of DP.

Articles and demonstratives

1. Introduction

In the Introduction to the book we mentioned in a preliminary way that DP

is the extended projection of N, in the sense of Grimshaw (1991). D is a
functional head; it is regularly occupied by the definite article. D selects as
its complement a (possibly extended, see below) projection of N. DP and
CP are similar in that both are projections that link their complements to
the discourse context or to the non-linguistic context. This parallelism has
been argued for since the eighties by, among others, Szabolcsi (1983, 1987,
1994), Fukui & Speas (1986), Abney (1987), Horrocks & Stavrou (1987),
Loebel (1989), Stowell (1989, 1991), Olsen (1991), and Longobardi (1994).1
Adopting the hypothesis that nominal projections are NPs dominated by DP,
we will from now on use the label DP to refer to the full nominal projection.
We will systematically reserve the label NP for the lexical projection con-
tained within DP.
In the Introduction we have already introduced a number of problems that
the older, and original, X’ theory was faced with in relation to the syntactic
representation of the article. Recall that in Jackendoff’s (1977) X’-theory,
the projection of the NP was not dominated by any functional projection.

(1) a. N’’’

Spec N’’

Spec N’ Compl

N Compl
———–——————————
1
For a comprehensive survey of the literature on determination and nominal pro-
jection, with special reference to English, French and German, see also Kolde
(1996).
54 Part II – The functional make up of the noun phrase

The NP was a single endocentric projection, i.e. it had a unique head, N,

and the other components of the NP occupied either the complement posi-
tion(s) or the specifier position(s). Hence, the article was represented as be-
ing an integral part of the projection of N: specifically it occupied the high-
est specifier position of the NP. But the concept ‘specifier’ and its relation
to the projection was quite different from what it is nowadays. In particular,
while currently it is often assumed that there is one specifier to a projection,
in Jackendoff’s system, there was more than one specifier to a projection
and specifiers of NP were distributed over two bar levels, in particular they
were sisters of N’’’ and sisters of N’’ (Jackendoff 1977: 104). In the earlier
representations the term specifier was used essentially as a cover term to
designate a constituent that appeared to the left of the head (at any bar level
in principle), so it was the position of an element vis-à-vis the head that de-
termined whether it was a specifier or not. There was unclarity as regards
the question whether a specifier could be a phrase or just a word.2 The fol-
lowing extract from Jackendoff (1977: 37) shows the uncertainty surround-
ing the nature of ‘specifier’:

Chomsky considers specifier to represent a syntactic category, but comple-

ment is simply an abbreviatory term for some concatenation of ordinary
syntactic categories. However, there is to my knowledge no evidence that
either complements or specifiers function as constituents – they do not move
or delete as units, and unlike normal constituents, no part can be designated
as a head.

Although Jackendoff drew attention to the difficulty in correlating semantic

regularities with syntactic positions (1977: 103), he did distinguish (intui-
tively) among two major semantic roles that can be performed by the occu-
pants of the specifier system of the noun: he draws a distinction between the
so-called demonstratives and the quantifiers (Jackendoff 1977: 104). These
semantic roles were realized by the two different fillers: demonstratives are
determiner-like constituents, such as definite (and possibly indefinite) arti-
cles and demonstratives in the narrow sense, but also interrogative elements
such as which and what, and quantifiers groups what are traditionally re-
ferred to as existential quantifiers and numerals. In the current chapter we
will be concerned with articles and with demonstratives.
———–——————————
2
For instance, the quantificational phrase a bunch in a bunch of people – i.e. the
Pseudopartitive Construction that we are going to discuss in detail in Part III,
Chapter 2, according to Jackendoff occupied the SpecN’ position.
 Chapter 1 – The emergence and the structure of DP 55

Jackendoff identifies the N’’’ specifier as follows: “The N’’’ specifier is the
position where genitive NPs occur in complementary distribution with de-
monstratives.” (1977: 104). However, we already pointed out in the Intro-
duction that prenominal possessive (or genitive) phrases are not on a par
with articles and demonstratives. For one thing, the latter form closed
classes, while the former are an open class. In Part IV we will discuss pos-
sessive DPs. At this point we also draw attention to the fact that, though ar-
ticles and demonstratives belong to closed-class elements and are usually
taken to be both functional elements, articles and demonstratives do not
form a completely homogeneous class. In section 4 of this Chapter we will
discuss the differences between these two classes of functional elements
and we will propose ways to implement these differences in terms of the in-
ternal structure of the DP. Before that, however, we will go over the role(s)
that have been attributed to the article. Importantly, right away we will see
that the concept of definiteness has been seen as connected intimately with
the definite article and as a consequence with the interpretation of the entire
DP.

2. The role(s) of the article

In the (re)formulation of the nominal projection in terms of the DP hypothe-

sis, the contribution of the definite article to the nominal projection has
played a substantial role. It has long been known that the article is not an
‘optional’ element of the noun phrase, as envisaged in Jackendoff’s original
proposal, in which it was one among several possible specifiers among the
class of demonstratives (Jackendoff 1977: 104). The article somehow stands
out as the determiner par excellence of the noun phrase. The exact role of
the category ‘article’ and its import to the interpretation of the whole pro-
jection it participates in is an issue that is much discussed in relation to the
‘DP Hypothesis’. The discussion concerns both syntax and semantics. All
of the following views are represented in the relevant literature:

The article is:

(i) a grammatical category which in some languages grammaticalizes
the semantico-pragmatic notion of definiteness (Lyons 1999);
(ii) a subordinator assigning argumental status to its NP complement
(Abney 1987; Stowell 1989, 1991; Szabolcsi 1994: 181);
(iii) the ‘natural’ bearer of referentiality (Loebel 1989; Longobardi
1994, 1996);
56 Part II – The functional make up of the noun phrase

(iv) a purely grammatical morpheme whose primary role is to assign

case to its complement NP (Giusti 1993, 1997, 2002).
If we assume that argumenthood and referentiality are closely related, (ii)
and (iii) can be collapsed. (ii) and (iv) can also be related if it is assumed
that an NP is turned to an argument if it is assigned case, and vice versa, if it
is assumed that when a nominal phrase is assigned case then it can function
as argument. (i) makes a different claim: here the function of the article ap-
pears to be primarily semantic/pragmatic. (iv) devoids the article of any
semantic content limiting it to just a grammatical formative. Under this
view, the functions summarized in (i), (ii) and (iii) are assumed to be asso-
ciated with the category D, which, according to most current approaches, is
the natural host of the article. The question then arises whether the func-
tions listed in (i), (ii) and (iii) are related directly to the definite article as
such or whether they are related to the category/structural slot D, and only
indirectly to the article, the natural filler of D. In what follows we will try
to sketch a tentative answer to this question, but we will unfortunately not
be able to provide anything like a definitive statement.

2.1. “Definiteness”
Since the relationship between (in)definiteness and the presence/absence of
the (definite) article has always been taken as obvious, some notes are in
order at this point to illustrate how this ‘link’ is established. So in this sub-
section we will take a quick look at what is usually meant by the term
definiteness, how it is syntactically reflected on the article, and what the
feature [+/-Def], which is standardly connected with D, amounts to. Al-
though a detailed account of definiteness falls outside the immediate goals
of this book, a brief survey of the essential points at issue will hopefully
help the reader to understand the syntax of D.
Accounts of definiteness abound, in the linguistics literature and in the
pragmatics-philosophy literature; the reader is referred to Christophersen
(1939); Krámský (1972); J. Lyons (1977); Loebner (1985); Chesterman (1991)
and C. Lyons (1999), and the references cited there for full details concern-
ing the many approaches to the concept of definiteness.
Let us start by considering the following examples:
(2) a. The cat ran quickly after the mouse.
b. Someone left the cat on my doorstep this morning.
c. The cat was chosen by his wife.
 Chapter 1 – The emergence and the structure of DP 57

Clearly, the underlined DPs in the examples in (2) are interpreted differently
from their indefinite counterparts in (2’).3 The DP the cat is definite, while
a cat is indefinite. Put in a simplistic way, by using the indefinite DPs in
(2’) a referent is introduced into the universe of discourse for the first time;
in (2), the use of the definite DPs implies that reference is being made to an
entity which is accessible because it has already been “introduced”. So, in a
sense we could say that the sentences in (2’) are prior to those in (2).
(2’) a. A cat immediately chased the mouse.
b. Someone left a cat on my doorstep this morning.
c. A Persian cat was chosen.
A this point, we can quote Simon Dik, who illustrates this state-of-affairs in
a very clear manner: “The construction of referents is typically achieved
through indefinite specific terms, as in Yesterday in the park I saw a black
cat; the retrieval of referents is typically guided by definite terms, as in Yes-
terday in the park I saw the/that black cat again.” (Dik 1987: 3).
A number of questions arise at this point: what does the concept of (in)-
definiteness amount to? Is it primarily a semantic notion or is it a gram-
matical notion? What is the relationship between the morpheme realizing
the (in)definite article and the semantic notion of (in)definiteness? Crucially,
is the semantic concept (in)definiteness a universal property? If it is, is it
expressed uniformly across languages?
In relation to the above questions two views have been advanced inde-
pendently, though no doubt they are ultimately interrelated. In both, defi-
niteness is a semantic entity which can be represented as a feature [DEF].
(a) For some, the feature [DEF], which plays an important role in the syntax
of DP, invariably represents a particular semantic-pragmatic4 concept,
———–——————————
3
In the discussion we will not deal with the indefinite article, an equally important
issue, as this article does not directly bear on the formation of the DP-hypothesis.
4
We use the opposition semantic-pragmatic in the current context rather loosely.
The concepts that will be discussed below are mainly pragmatically grounded in
that they rely on coordinates of utterance (speaker, hearer, familiarity, context,
etc.), but above all on identifiability. On the other hand, there is also a logical-
philosophical, or formal semantic (Lyons 1999: 258), account of definiteness,
which goes back to Frege (1892) and Russell (1905), and, much more recently
Strawson (1952), who responded to both his predecessors, and in particular to
claims made concerning the sense-reference distinction. According to this logico-
philosophical tradition, the key concepts in the notion definiteness are the notion
of inclusiveness and the related concept of uniqueness. The classical example
58 Part II – The functional make up of the noun phrase

namely ‘definiteness’. This concept, however, is not uniformly realized

across languages, since many languages lack a definite article. (b) In an-
other view, the feature [DEF] is a purely grammatical feature whose relation
to one or more semantic concepts varies from language to language (Lyons
1999).
In order to better appreciate these two positions, we will first try to clar-
ify certain pragmatic notions primarily involved in definiteness. The quin-
tessential property of definiteness is taken to be identifiability (subsuming
familiarity), a concept which is tightly interwoven with inclusiveness.
Identifiability implies that the speaker signals that the hearer is able to
locate a referent for a particular DP. Concretely, this means that in (2) the
referent of the DPs the cat is identifiable. Along with identifiability, an-
other related interpretative component of definite DPs in (2) is familiarity.
The entity ‘the cat’ referred to by the definite DPs, the cat, in (2), is as-
sumed by the speaker to be familiar to the hearer; to put this differently:
what the DP refers to, is part of the knowledge shared by the speaker and
the hearer. If the hearer happens not to know or be able to locate what the
speaker is talking about, the natural response would be ‘which cat (do you
mean)?’
In the examples in (2’) the use of the indefinite DP, a cat, implies that
no such familiarity can be presupposed. Here, the speaker may be the only
one to know what the DP refers to, that is to identify the entity in question.
Familiarity5 and, more generally, identifiability, is a property of such uses
of the definite article as the situational use in (3-4), general knowledge use
in (5), anaphoric use in (6) and associative use in (7):

———–————————————————————————————
much discussed in this tradition is the sentence The King of France is bald,
which contains the singular definite expression The King of France. This sen-
tence encapsulates the proposition ‘There is only one King of France’ – along
two additional propositions, the existential presupposition (‘There is a king of
France’), and the main assertion of the sentence in question (viz. that the par-
ticular individual is bald). It is the uniqueness component of definiteness, as en-
coded in such definite descriptions, that is replaced in many current accounts by
the concept of familiarity or identifiability. See Lyons (1999, ch. 7) for a concise
informative summary of the logical origins of definiteness.
5
Familiarity is usually opposed to novelty which involves primarily indefinite
noun phrases. See Lyons (1999), section 7.1 in particular, where he gives a con-
cise account of Jespersen’s (1924) degrees of familiarity and where he also pro-
vides other relevant references.
 Chapter 1 – The emergence and the structure of DP 59

(3) Sit in the hand basin just before your human intends to shave.

(4) Did you see the fat cat just running in?

(5) The sun is burning hot today.

(6) A kitten was sleeping under the tree. A cat then appeared and sat next
to the kitten. She must be his mother.

(7) I met with Artemis outside the opera at 11.00 p.m. The tickets for to-
night’s performance had to be bought by early afternoon.

In all of these examples “the hearer is invited to match the referent of the
definite noun phrase with some real-world entity which he knows to exist
because he can see it, has heard of it, or infers its existence from something
else he has heard” (Lyons 1999: 6). Nevertheless, as is pointed out in the
literature on definiteness, identifiability, let alone familiarity, does not ac-
count for every use of the definite article. As Lyons (1999) points out, as-
sociative/situational uses of the definite article pose problems for identifi-
ability. Consider the following example:

(8) I’ve just come back from a wedding. The bride was wearing red.

In (8) the use of the indefinite article with the N wedding indicates that the
speaker does not presuppose familiarity on the part of the hearer. However,
if the hearer is not familiar with the ‘wedding’ as introduced in the first
sentence, one can hardly imagine that he or she is familiar with the referent
of the definite DP the bride in the second sentence. The hearer very likely
will not be able to identify the referent of the definite noun phrase the bride
in any real sense. He or she may not know who the bride was or in fact any-
thing else about her.
The concept which seems to be at work in the case of examples like (8)
is that of uniqueness. The definite article signals that “there is just one en-
tity satisfying the description used. (…). This description is generally not
absolute, but is to be understood relative to a particular context” (Lyons
1999: 8). Since at every wedding there is by definition a bride, and since
there is normally only one bride, the use of the definite DP the bride is war-
ranted. However, uniqueness itself raises an immediate problem since it
seems to leave unexplained cases involving plural (9a,b) and mass (9c)
nouns:
60 Part II – The functional make up of the noun phrase

(9) a. John was looking for the cats (that lived in his garden).
b. I’ve just come back from a wedding. The bridesmaids were wear-
ing red.
c. The wine you bought needs to be chilled.

If the definite articles in the underlined DPs in (9) can still be thought of as
encoding uniqueness, uniqueness must be taken to concern whole sets and
masses rather than single entities or individuals (Lyons 1999: 11). In addi-
tion, we understand (9a) to mean that John was looking for all the cats that
used to live in his garden, not just some of them; and from (9b) we will
conclude that all the bridesmaids were wearing red. Definiteness here im-
plies reference to the totality of the entities that satisfy the description (Ly-
ons 1999: 11). Such a use of the definite article with the implication of ‘to-
tality’ is considered as a special case of uniqueness. It is often referred to as
inclusiveness, a term which, as Lyons (1999: 11) points out, is due to Haw-
kins (1978). In the case of plural nouns the definite article functions like the
universal quantifier all. The (near) identity of (9a,b) to (9’a,b) shows this
function of the article with a plural noun:

(9’) a. John was looking for all the cats (that live in his garden).
b. I’ve just come back from a wedding. All the bridesmaids were
wearing red.

If there were any cats that used to live in his garden and John was in fact
not looking for them, both (9a,b) and (9’a,b) are equally false. In this case a
response like (9’’a) would be appropriate:

(9’’) a. No, he was already holding one in his hands.

Similarly, if any of the bridesmaids at the wedding were not wearing red,
then both (9b) and (9’b) would be false and a response like (9’’b) would be
appropriate:

(9’’) b. No, one of them was wearing green.

Lyons further suggests that uniqueness can be assimilated to inclusiveness.

“When the noun phrase is singular, inclusiveness turns out to be the same
as uniqueness, because the totality of the objects satisfying the description
is just one.” (Lyons 1999: 13).
Mass nouns lead to the same inclusiveness interpretation. Consider (10a):
 Chapter 1 – The emergence and the structure of DP 61

(10) a. I’ve just come back from a wedding. The wine was awful.

Once again the definiteness of the DP the wine cannot be due to familiarity
on the part of the hearer, as shown by the fact that the DP a wedding is in-
definite, indicating no familiarity is presupposed. Again the DP the wine
seems to imply totality: (10a) is equivalent to (10b):

(10) b. I’ve just come back from a wedding. All the wine was awful.

If some of the wine served at the wedding was actually good, then both
(10a) and (10b) are false and (10c) would be an appropriate continuation:

(10) c. Actually, that’s not true. The dessert wine was very good.

It should also be pointed out here, as is done by Lyons (1999: 158ff), that
the manifestations of definiteness mentioned so far fall into two major
types according to whether the context appealed to is linguistic or non-
linguistic. With respect to the anaphoric use of the definite article (cf. ex-
amples in (2), (2’) and (6)) the context in which the referent is found is lin-
guistic. All the other uses relate to extra-linguistic contexts which crucially
involve general/encyclopedic knowledge or knowledge of the situation re-
lated to the utterance. In this connection it is further interesting to note that
some languages only have an anaphoric (definite) article, while some others
distinguish between an anaphoric definite article and a non-anaphoric defi-
nite article (Lyons 1999: 158–159).
The following question emerges from the preceding discussion: is defi-
niteness a single, unified phenomenon or is it possible that what is com-
monly called definiteness amounts to more than one semantic-pragmatic
category, which in some languages happens to have a unique morphologi-
cal realization? There is no simple answer to this question and any serious
account would inevitably takes us too far afield. For our needs here suffice
it to say that in the examples (3)–(7) above the DPs marked as definite in
English can apparently be used in order to convey a number of distinct se-
mantic-pragmatic notions, such as anaphoricity, familiarity, general knowl-
edge, situation, etc. In this context, definiteness appears to be taken as a su-
perordinate term comprising these distinct semantic/pragmatic concepts
associated with it. Lyons (1999) discusses at length cross-linguistic evidence
that shows that none of these various ingredients of definiteness is com-
pletely reducible to the other, although in many instances one may follow
or be implied by the other. Hence, a theory relying on identifiability will
62 Part II – The functional make up of the noun phrase

make different predictions from a theory relying on inclusiveness, while at

the same time identifiability includes/entails familiarity. As just said, there
exist languages in which these different ‘ingredients’ of definiteness are re-
alized by distinct articles (see Lyons 1999, especially in section 2.2, for
relevant data and discussion).
We can summarize what we have said so far by outlining a proposal by
Lyons (1999), whose work constitutes the most recent detailed cross-lin-
guistic exploration of definiteness. According to Lyons, what is universal is
semantic or pragmatic definiteness, while the grammatical realization of
definiteness is a language specific property (see also Felix 1988 for a similar
view). In particular, Lyons claims that what is realized ‘prototypically’ is
identifiability itself, an interpretational element of all languages, but it is
realized by the definite article (‘grammaticalized’) in only some of them. In
any way, identifiability, as a property of information/discourse structure,
has a role to play in every language. Under the assumption that the definite
article is one realization (‘grammaticalization’ in Lyons’s own terms) of the
semantic and pragmatic concept of definiteness, in languages that do not
grammaticalize definiteness, interpreting a given item as either definite or
indefinite relies on discourse organization, in particular on identifiability
(see the discussion in the next chapter).
With respect to the question whether there is one or several instantia-
tions of definiteness, Lyons’s answer is that despite the range of linguistic
variation as to exactly which components of definiteness may be encoded
by the definite article (or an equivalent form), there is always a large core
of uses of the article that are transparently relatable to identifiability and in-
clusiveness. This approach to definiteness is thus on a par with the standard
approaches to (nearly) all other grammatical categories, such as aspect,
C/case, Tense, etc., as we will have the opportunity to discuss further in
subsequent chapters. With these categories too, a distinction has to be
drawn between abstract, semantic categories and their linguistic or syntac-
tic realization (as concrete morphemes/words).

2.2. The article as a subordinator

As we said above, in Jackendoff’s (1977) approach to the structure of the

NP, articles occupied the specifier position of the maximal projection of N
(i.e. the third bar level). The tree diagram below illustrates Jackendoff’s NP
claim:
 Chapter 1 – The emergence and the structure of DP 63

(1) b. N’’’

 Art
a/the  N’’

 this
some (sing) 
 his  N’

Being specifiers, articles were seen as optional elements of the nominal

projection. However, it doesn’t take long to realize that optionality cannot
be taken for granted, as the ungrammaticality of the bare singular countable
noun in the following examples strongly suggests:

(11) a. *He found cat on doorstep.

vs He found a/the cat on his/the doorstep.
b. *Cat arrived last night.
vs A/the cat arrived last night.
c. *Vrice ghata sto katofli tu.
vs Vrice mia/ti ghata sto katofli tu. (Greek)
found-3SG cat at the doorstep his
found-3SG a/the cat a the doorstep his
‘He/she found a/the cat at his doorstep.’

(11) shows that singular count nouns cannot routinely occupy thematic po-
sitions (e.g. be subjects or objects of verbs)6 unless they are subordinated
under the article. So, in view of these facts the presence of an article ap-
pears to be indispensable with noun phrases that function as arguments of
verbs. But here a question naturally arises: if the article is obligatory for a
noun to assume argumental status, how can ‘bare’ nouns be arguments in
languages that have no (definite) article, such as Latin, Russian, Polish and
so many others? Not less importantly, how can bare NPs ever be arguments
in languages with articles, i.e. how can one explain the fact that the English
sentences below are grammatical despite the fact that no article accompa-
nies the noun that is the object and the subject of the verb?

———–——————————
6
See the next chapter for more on bare singulars and bare nominals.
64 Part II – The functional make up of the noun phrase

(12) a. I am going to get wine and beer for the party.

b. Topsy hates cheese.
c. Ilektra bought flowers for Orestes’s party.
d. Cats are cute creatures.

We observe that in (12a) and (12b) wine, beer and cheese belong to the
subclass of mass nouns; in (12c,d) flowers and cats are plural nouns. Note
in passing that in (12d) the plural cats refers to the ‘genus’ of cats, or to
cats as a species, i.e. it has a generic interpretation.7 We can thus say that
mass nouns and so-called ‘bare plurals’ seem to be able to function as ar-
guments without the presence of the article. We will see further below that
in languages lacking a definite article, bare nouns or noun phrases can in-
deed be arguments. We will address the issue of lack or omission of the ar-
ticle in Chapter 2.
In any case, the ability of nominals to occur ‘bare’ as arguments (of
verbs) is quite restricted; indeed in English it is confined essentially to
cases like those illustrated in (12). Leaving aside this ‘peculiarity’ of cer-
tain types of nominals for the moment, and concentrating on the subordi-
nating role of the article, the core idea behind the ‘DP-Hypothesis’ relies on
the data in (11). Let us describe the function of the article in (11). Cat as
such is a predicate, it denotes what is ‘true of all cats’. The DP the cat re-
fers to an individual, a particular cat; the cat refers to an entity in the world.
Thus, in a sense, the article saturates the predicate ‘cat’: it combines with a
predicate (cat in (11a,b) and turns it into an individual (expressed by the
phrase the cat). Using a more technical language, predicates are of the se-
mantic type <e,t>, individuals are of the type <e>. Articles (and more gen-
erally determiners) are of the semantic type <<e,t>,e>: they are ‘functions’
that take a predicate, which is of the semantic type <e,t>, and yield an indi-
vidual, which is of type <e> (Heim & Kratzer 1998: 52–53). It is along these
lines that one can understand the often used metaphor that articles are what
links language to extra-linguistic reality: articles (like all other determiners
for that matter) anchor linguistic entities to the real world. “A ‘nominal ex-
pression’ is an argument only if it is introduced by a category D.” (Longo-
bardi 1994: 620).

———–——————————
7
Genericity is a complex notion and has given rise to an abundant literature in the
domain of semantics, which we will not go into here. For some discussion see
the papers in Carlson and Pelletier (1995) and also Oosterhof (2006a).
 Chapter 1 – The emergence and the structure of DP 65

For Higginbotham (1985), there is a relation between the article and the
noun that runs in parallel to the relation between a verb and its (thematic)
object: the relation of thematic role saturation or thematic discharge. The
article serves the role of saturating the open place that every common noun
has and by virtue of which it ‘denotes’. This open position is parallel to the
theta grid verbs have. For instance, the noun dog denotes each of the vari-
ous dogs that exist and this is due to the open position this noun possesses
as part of its lexical entry (Higginbotham 1985: 560). The article binds this
open position of the noun, or, putting it differently, it closes off the nominal
structure with respect to the noun’s open position. Higginbotham (1985:
560) calls this process, which is a process of thematic discharge, theta-
binding. He considers it parallel to theta marking of arguments by verbs.
Moreover, there must be a binder to bind the noun’s open position, and, as
Higginbotham points out, there cannot be two. This explains why there is
only one article per nominal projection, i.e.; there is only one ‘binder’
(*every the dog). The biuniqueness of theta binding in the nominal domain
aligns the nominal domain to the sentential domain (see also section 5): in
the verbal domain too each argument can only receive one thematic role
from the verb and the verb can only assign one thematic role to each one of
its arguments.
All this said, some parallelisms suggest themselves here. The anchoring
function of the article in the DP is also seen as similar to that of tense in a
clause: tense anchors the eventuality denoted by the verb (phrase) to the
non-linguistic reality (Abney 1987; Olsen 1991; Loebel 1989). A further
similarity is that between D and C, the complementizer: only DPs and CPs
can function as arguments. It has therefore been argued by Szabolcsi (1983,
1987, 1994), Stowell (1989, 1991) and Horrocks & Stavrou (1987) that the
definite article is to the nominal projection what the complementizer (C) is
to the clause : both turn a non-argumental category into an argument.8 In
sections 3 and 5 below, we will come back to the specifics of Szabolcsi’s
and Horrocks & Stavrou’s proposals concerning the structural representation
of the parallelism(s) holding between the complementizer and the article.

———–——————————
8
This idea of N(P) being embedded under DP goes back to Brame (1982). He
said: “(…) Since DET is the head-selector of DET(N), (…) it would be better to
abbreviate DET(N) as DP, not as NP, and to speak of determiner phrases, rather
than noun phrases.” (Brame 1982: 325).
66 Part II – The functional make up of the noun phrase

2.3. The article and the concept of referentiality

From what we have said so far, it has become clear that referentiality and
argumenthood are related: argumenthood is the syntactic reflex of the con-
cept of referentiality.9
Apart from the situation illustrated in (11), we may ask ourselves if there
is additional evidence that the presence of the article and the subordination
of a noun (or nominal expression) to it brings about a shift of a purely
linguistic entity into something that is ‘closer’ to the external world, i.e., of
a predicative category to an argumental one? The answer to this question is
positive. Consider the following data from Italian (Longobardi 1994: 620),
French and Greek:

(13) a. La mia segretaria e tua collaboratrice sta/*stanno uscendo. (Italian)

the my secretary and your collaborator is/are leaving
‘My secretary and your collaborator is about to leave.’
a’. La mia segretaria e la tua collaboratrice stanno /*sta uscendo.
the my secretary and the your collaborator is/are leaving
‘My secretary and your collaborator are about to leave.’
b. La secrétaire de Jean et collaboratrice de Paul est/*sont à la gare.
(French) (Bouchard 2002: 43)
the secretary of John and collaborator of Paul is/are at the station
‘The secretary of John and collaborator of Paul is at the station.’
b’. La secrétaire de Jean et la collaboratrice de Paul *est/sont à la gare.
the secretary of John and the collaborator of Paul is/are at the station
‘The secretary of John and the collaborator of Paul are at the station.’
———–——————————
9
Though there is no a priori or logical connection between argumenthood and ref-
erentiality, it nevertheless seems to be the case that the two concepts fall together
– and are even often fused-as far as the realization of the article is concerned.
This seems to be an immediate consequence of the DP hypothesis. The relations
can be expressed in a number of ways. Szabolcsi (1994) speaks primarily of NP
subordination (to D) in relation to argumenthood. She says: “Both the comple-
mentizer and the article are subordinators in the sense that they enable the clause
or noun phrase to act as arguments” (1994: 214), whereas Stowell (1991) and
Longobardi (1994) rather explicitly relate argumenthood with referentiality (in
the same spirit cf. also Chomsky 1995: 337).
For a general discussion of the relation between referentiality, definiteness
and argumenthood see also Öztürk (2005).
 Chapter 1 – The emergence and the structure of DP 67

In all the examples in (13) two nominal phrases are coordinated with the
equivalent of English ‘and’ but for the coordinated pairs in the non-primed
examples there is only one article. As a consequence, the coordinated string
refers to a single referent. This is immediately obvious in the Italian example
(13a) and in French (13b) because in both cases a plural form of the finite
verb is excluded. In other words, despite the presence of two descriptive
nouns, segretaria (‘secretary’) and collaboratrice (‘collaborator’) in (13a)
and sécrétaire (‘secretary’) and collaboratrice (‘collaborator’) in (13b),
only one individual is involved/referred to. This is related to the fact that
there is only one article (la).10 On the other hand, the corresponding primed
examples (13a’) and (13b’) contain two articles. In this case there are two
referents and the finite verbs must be plural. From this evidence we con-
clude that it is the article that makes a noun refer to entities (outside the lin-
guistic context). Modern Greek (13c) displays a similar effect: in (13c) the
coordinated DP refers to one referent, in (13c’) there are two referents.
(13c’) with the verb in the singular may in fact be grammatical for inde-
pendent reasons: in Greek, when two singular DPs are coordinated, the verb
may be in the singular too:

(13) c. Irth-e/*-an o antiprosopos tis dikasticis arçis.

came.3SG/PL the delegate of.the court
ce proedhros tis eforefticis epitropis. (Greek)
and chair of the elective committee
‘The representative of the court and chair of the elective committee
has arrived.’

———–——————————
10
Bouchard (2002) explains these facts along the lines of number being realized
on Det in French to the effect that each N in (13b,b’) cannot have a minimal
atomization, hence it cannot have referential capacity by itself. Things are a bit
more complicated in the case of English, as a single determiner may yield an
ambiguous sentence as regards the number of referents involved. Bouchard as-
sumes that Number is realized on N in English. Detailed discussion in this chap-
ter would take us a little further afield than planned. See Part III, Chapter 1, sec-
tion 7, for Bouchard’s approach to the category number, and also Chapter 3 in
this Part for a discussion on variable number realization in different languages.
The interested reader is further referred to Longobardi (1994: 622) and Bouchard
(2002) for interesting detailed discussion.
68 Part II – The functional make up of the noun phrase

c’. Irth-an/*-e o antiprosopos tis dikasticis arçis

came.3PL the friend my and the close collaborator my
ce o proedhros tis eforefticis epitropis.
and the chair of the elective committee
‘The representative of the court and the chair of the elective com-
mittee have arrived.’

On the basis of data like those discussed above, Longobardi (1994: 621)
concludes:
In other words, irrespective of the cardinality of head nouns present, a single
singular determiner is sufficient to impose singular designation on the entire
expression, whereas the sum of two singular determiners automatically im-
poses plural designation.

To put this more simply, there is one referent for each determiner and vice
versa. This one-to-one relation between determiner and referent is apparent-
ly restricted to argumental DPs. In (13d) the predicate contains two coordi-
nated DPs, each with its own article. However, the result of that coordina-
tion does not imply plural reference. In this case the person referred to as
‘Maria’ is both the secretary and collaborator of the speaker.

(13) d. Maria è la mia segretaria e la tua collaboratrice.

(Longobardi 1994: 621)
Maria is the my secretary and the your collaborator
Mary is my secretary and your collaborator.

Examples such as (14) suggest that nouns, alongside verbs, adjectives and
prepositions can function as predicative heads (Stowell 1991) in that they
too can assign theta-roles:

(14) a. The students elected John president (of the union).

b. Gianni è tenente. (Italian)
John is [a] lieutenant
‘John is a lieutenant.’
c. O Petros ine jatros (Greek)
the Peter is doctor
‘Peter is a doctor.’

Leaving theoretical complications concerning theta role assignment aside, we

can say that the noun president in (14a) assigns a theta role (theme) to John.
 Chapter 1 – The emergence and the structure of DP 69

Likewise, tenente (‘lieutenant’) in (14b) and jatros (‘doctor’) in (14c) assign

the theme role to Gianni and Petros respectively. President, tenente and
jatros are predicative heads and are not as a rule accompanied by an article.11
These examples show that singular countable nouns used as predicates do
not necessarily need an article. As arguments, though, singular countable
nouns do require an article. Contrast English (14a) with (15), in which the
omission of the article results in ungrammaticality (and see also (11) above):

(15) *Students elected John president (of the union).

In the light of the data discussed here, linguists have postulated that the ref-
erential capacity of the noun derives form the nature of the (definite) article
(Stowell 1989). Abney (1987: 77) says:

The function of the determiner is to specify the reference of a noun phrase.

The noun provides a predicate, and the determiner picks out a particular
number of that predicate’s extention”. Apparently, the (definite) article en-
joys a privileged position among all those elements that constitute the class
of determiners, as we shall see shortly. However, the key question is how
from this fact, namely the ability of the article to pass referentiality onto the
noun it modifies, the conception of the category DP, as a superordinate
category to NP, emerged.

For the sake of completeness, we add here that indefinite DPs can function
as arguments and as predicates:

(16) a. John is a doctor.

b. They called Mary a thief.

In particular, we refer to Stowell (1989, 1991) and to Higginbotham (1985)

for the discussion of data which show that nominal predicates can, and, in
some cases, even must, be preceded by the indefinite article and for a num-
ber of explanations. The existence of such cases is shown by these authors
not to invalidate the general claim that noun predicates do not need the arti-
cle, rather the presence of the (indefinite) article with such nouns is ac-
counted for on independent principles/grounds.
———–——————————
11
There seems to be variation across languages as to the type of noun phrases that
are permitted as the complements of copulas. In English the counterparts of
(14b–c) are ungrammatical. See the contributions in Zamparelli (2004) for a dis-
cussion of relevant data and explanations.
70 Part II – The functional make up of the noun phrase

As already alluded to at the beginning of section 2.2, the obligatory

character of the determiner together with its semantic impact have led to
promoting it from the status of being a mere specifier within the NP, the
projection of the lexical head N, to being the head of a functional projec-
tion. This projection has been variably labeled Article Phrase (Horrocks &
Stavrou 1987) and Determiner Phrase (Abney 1987 and subsequent litera-
ture). In sections 3 and 5 we will discuss in more detail how this functional
category is articulated. For the moment suffice it to say that the head D
hosts the definite article and that it selects as its complement the lexical NP.
Thus the resulting complex of NP dominated by DP replaces what in Jack-
endoff’s framework was the NP, a category with a unique head, N. DP is
thus considered as the extended projection of N.
The point is put in a simple way by Stowell (1991: 46):

Thus it seems possible to claim that, at least at the level of L[ogical]F[orm],

all NP-type of arguments are referential expressions of one sort or another. It
is plausible to suppose that all of these expressions [i.e., quantifier phrases,
demonstratives, proper names and pronouns – A-H-S] derive their referential
status from their heads, since their distinctive referential properties correlate
with the type of element occurring in the head D position.

To recapitulate the main points covered in sections 2.2 and 2.3 we can fur-
ther cite Longobardi (1994), whose work has been seminal in bringing out
the direct relevance of D to argumenthood/referentiality of nouns:

A “nominal expression” is an argument only if it is introduced by a category D.

DP can be an argument, NP cannot. (Longobardi 1994: 628).

In section 3 below we will attempt an answer to the key question: apart from
its role in creating an argument out of an NP, is there any independent sup-
port for this new DP category? In the next chapter we will go back to Lon-
gobardi’s work centered around the role of D in licensing argumenthood.

2.4. Summary: the functions of the definite article

So far we have presented a relatively simple picture of the structure of the

nominal phrase and the role of the article. The article is conceived of as a
subordinator, on a par with complementizers, which can turn a projection of
a noun into a referential expression, which can then be used as an argument
of a verb (or a preposition). In addition, the article conveys – or grammati-
 Chapter 1 – The emergence and the structure of DP 71

calizes – the semantic-pragmatic concept of definiteness, the latter primar-

ily manifesting uniqueness and identifiability. The article thus can be seen
to perform specific syntactic and semantic functions. Given its central role
in the form and interpretation of the noun phrase, it is analyzed as a head
that projects its own (functional) category DP.
However, at this point the complications of this proposal reveal them-
selves: what is the exact relationship between the article which exists as a
specific morpheme in some, but not all, languages, and the category of
definiteness (with all the interrelated notions it brings about)? Are all the
semantic and syntactic functions we have isolated above (lexical) properties
of the morpheme of the article itself, or are they to be attributed to some-
thing else, something abstract, behind or above the article itself? For in-
stance, could the semantic-syntactic category D be the vehicle of definite-
ness rather than any one of its realizations? In Chapter 2 we will review
some of the responses to these questions. For instance, according to some
researchers, e.g. Giusti (1993, 1997, 2002) the article by itself is nothing
more than a grammatical morpheme responsible for assigning case to the
noun – point (iv) under 2 above. Under such a view, the article is devoid of
any semantic or descriptive content. We turn to this particular view in 2.5.

2.5. The article as a grammatical morpheme

In the introduction we provided the list of diagnostic properties of func-

tional categories, as originally drawn up in Abney (1987: 64f). On the basis
of Abney’s inventory, Giusti (1993, 1997: 102–107, 2002) concludes that
(a) among determiners only articles are functional heads, and (b) in some
languages a definite article is inserted on syntactic grounds regardless of
the referential/semantic properties of the noun phrase.12 Giusti’s (2002)
arguments for these claims are the following:

———–——————————
12
Giusti’s fundamental claim about the ‘grammatical’ character of the definite article
in a sense echoes Krámský’s (1972) claims about the definite article in French
as an indicator of gender and number and in certain cases in German as marking
case. Krámský, however, points out (1972: 28–29) that there is only a very re-
stricted overlap between what the article designates in different languages. He
further says that to give “a precise definition of the article which would be valid
for all languages that posess an article in some or other form, would be, in the
present state of research on this problem, a very difficult if not impossible task”
(1972: 29).
72 Part II – The functional make up of the noun phrase

First, articles, even those that are free morphemes, are phonologically and
morphologically dependent on the lexical category they come with. In some
languages such as Italian, for example, the form of the article is adjusted to
the initial sound of the following word. We observe that in this case the ar-
ticle displays a series of allomorphs:

(17) il ragazzo lo scolaro l’amico (Italian; Giusti 1997: 102)

the boy-MASC the student-MASC the friend-MASC

The dependency of the article on some lexical category is all the more ob-
vious in languages where the article is enclitic. In those languages, the arti-
cle can be suffixed not only on the noun, as illustrated in (18a), (18b), (18c)
and (18d) but it can also be suffixed on the adjective, if that is the initial
constituent of the projection, as shown in the Bulgarian example in (18e).

(18) a. băiatul frumos (Romanian)

boy-the nice
b. djali i mirë (Albanian)
boy-the ARTgood
c. huset mitt (Norwegian)
house-the my
d. hestur-in (Icelandic)
stallion-the
e. goljamoto momče (Bulgarian)
big-the boy

In such patterns, the article seems to play the role of an inflectional ending
spelling out φ-features rather than expressing semantic categories such as
definiteness (see also (25) below).
More revealingly, in Bulgarian the form of the article depends on the
word class to which the noun belongs. This is shown in (19):

(19) a. momce-te
boys-the
b. xora-ta
people-the

Both momce (‘boys’) and xora (‘people’) are masculine plural nouns but
they belong to different word/conjugation classes. In the light of examples
 Chapter 1 – The emergence and the structure of DP 73

such as (18) and (19) Giusti claims that the enclitic article is devoid of de-
scriptive content and is just an inflectional ending of the noun (cf. also ex-
ample (29 below)).
In addition to the observation that the article may be enclitic on other
heads than on nouns, Giusti’s claim that the article is devoid of descriptive
content is based on instances of noun phrases containing more than one
manifestation of the definite article. This phenomenon is referred to as de-
terminer spreading or double definiteness (see Part III, Chapter 1, section 6,
for data and possible analyses) and is illustrated in the following examples
(from Giusti 2002: 61–62):

(20) a. to oreo to vivlio / to vivlio to oreo (Greek)

the good the book / the book the good
the good book
b. djali i mire / i miri djalë13 (Albanian)
boy-the the-good / the good-the boy
the good boy
c. băiatul (cel) frumos / frumosul băiat (Romanian)
boy-the (the) good / good-the boy
the good boy

Similar cases are found among others, in Hebrew and in certain Scandina-
vian languages (Swedish, Norwegian). These examples suggest that the
definite article does not encode definiteness as such: despite multiple mani-
festations of the definite article14 in all of these examples there is just one
referent. In other words, the proposal made in 2.3 above, that for each arti-
cle there is one referent cannot be maintained. This has led some research-
ers to refer to the use of the definite article in determining spreading con-
structions as an expletive use of the definite article, i.e. a use in which the
definite article lacks interpretative substance, in the same way that the defi-
nite article associated with proper names is considered as an expletive (for
proper names see also section 2.3.3. of Chapter 2, for general discussion of
the notion ‘expletive article’ see section 2.3.4. Chapter 2). (For details on
———–——————————
13
The morpheme i in Albanian marks the adjectival class and is not relevant to the
definiteness of the noun phrase. It is a purely morphological entity. See Giusti
(1993) and (2002) for details about the morphological character of the enclitic
articles in Albanian and Romanian.
14
We ignore details having to do with different realizations of the definite article
in some languages, which do not affect Giusti’s argument here.
74 Part II – The functional make up of the noun phrase

the Greek construction, see also Androutsopoulou 1994 and 1996.) Giusti
proposes that in these constructions the ‘definite’ article serves a purely
grammatical role, in particular it encodes agreement between the noun and
its modifier(s) by bearing phi- and case features.
On the other hand, it is clear that Giusti’s claim that the definite article
does not have any semantic import encounters difficulties with respect to
data like those in (13) above, in which the coordination of two DPs each
with their own determiner implied that there were two referents, while the
coordination of two NPs with one determiner implied there was a single
referent. In relation to this issue, Giusti (2002) signals that in Romanian,
the co-occurrence of two definite articles in one DP does not produce a
two-referent interpretation effect. This is because in Romanian the article is
suffixed and cannot be omitted. Consider the following examples from Gi-
usti (2002:62) and compare them with the data in (13) above:

(21) a. Directorul de departament si presidentele de facultate a/au venit aici.

director-the of department and president-the of faculty has arrived here
‘The head of the department and chair of the faculty is here.’
b. *Directorul de departament si presidente de facultate a/au venit aici.
director-the of department and president of faculty has arrived here.

(21a) is ambiguous between the readings in which the coordinated DP has

one or two referents. On the other hand, even when there is just one refer-
ent, example (21b), with one enclitic article, is ungrammatical. This effect
is due to the fact that in Romanian the article is suffixed and cannot be
omitted. One way to solve this complication would be to say that the prop-
erties of free morpheme-articles (such as those in Italian or Greek, for in-
stance) are distinct from those of suffixed articles (such as those in Roma-
nian). But, unfortunately, this solution, as Giusti herself notes, is rather ad
hoc and overshadows the fact that despite their different morphological
status free and suffixed articles share a number of commonalities. In order
to overcome this problem, Giusti puts forward a rather complex explanation
for the Romanian facts, which we will not go into here (see Giusti 2002 for
detailed discussion).
That the presence of an article does not always imply referentiality is
also shown by examples such as (22) in which the (in)definite noun phrase
la/una segretaria di un onorevole is interpreted as non-referential, as
shown by the subjunctive mood in the relative clause:
 Chapter 1 – The emergence and the structure of DP 75

(22) Scommetto che non troverai mai [FP la/una segretaria

bet-PRES-1SG that non find-FUT-2SG never the/a secretary-FEM
[PP di un onorevole che sia disposta a testimoniare contro di lui]].
of a deputee that be-SUBJ-3SG disposed to testify against of him
‘I bet you’ll never find the/a secretary of a deputee who is-SUBJ will-
ing to witness against him.’ (Italian; Giusti 2002: 63)

So once again, referentiality and the presence of the definite article do not
appear to be isomorphic.15
Among the items often listed under the heading ‘determiners’, we also
typically find demonstratives (see below, section 4.1). Giusti distinguishes
between articles and demonstratives and based on a comparison with the
data she has first provided in order to show the genuine functional/gram-
matical behavior of the definite article, she contrasts this behavior to that
displayed by demonstratives. For instance, demonstratives are not morpho-
logically dependent on the head noun, they may even be used independ-
ently.

(23) a. Il ragazzo è partito.

the boy is left
‘The boy has left.’
b. *Il è partito.

(24) a. Questo ragazzo è già partito.

that boy is already left
‘That boy has already left.’
b. Questo è già partito.
that is already gone
‘That one has already left.’

The demonstrative also makes quite a different contribution to the interpre-

tation of the DP. Compare (22) above with example (25) from Giusti (2002:
64) in which there is a demonstrative to introduce a complex noun phrase
which contains a subjunctive clause.
———–——————————
15
Though, of course, one might try to maintain the isomorphism, and attribute the
apparent lack of a strict relationship between the definite article and referentiality
to independent factors, such as the nature of the predicate, negation and/or the
subjunctive.
76 Part II – The functional make up of the noun phrase

(25) *Scommetto che non troverai mai [FP questa/quella segretaria

bet-pres-1sg that non find-fut-2sg never this/that secretary-FEM
[PP di un onorevole che sia disposta a testimoniare contro di lui]].
of a deputy that be-SUBJ-3SG disposed to testify against of him

As can be seen, the demonstrative cannot replace the determiner: this is be-
cause the subjunctive is incompatible with a referential interpretation of the
DP while, by virtue of the demonstrative, the NP questa/quella segregaria
must be referential. Because of this conflict, the sentence is ungrammatical.
This evidence then supports Giusti’s hypothesis that the definite article does
not invariably imply referentiality. Conversely, the demonstrative does im-
ply referentiality.
Serious doubts on the role of the article as a definiteness/referentiality
marker arise when we further consider the fact that it co-occurs with proper
names as well as the fact that in some languages the definite article co-
occurs with constituents which themselves encode referentiality, such as
demonstratives. Consider the following examples from Greek and Italian
(see the following chapter for more discussion).

(26) a. I Topsy irthe. (Greek)

the Topsy came
‘Topsy came.’
b. Il mio Gianni (Italian)
the my Giani
‘my Gianni’
c. afti i ghata (Greek)
this the cat
‘this cat’

Being ‘rigid designators’ (Kripke 1972), proper names are inherently referen-
tial. Proper names can directly pick out a particular individual in the world.
In the examples (26a) and in (26b), the proper names are accompanied by a
definite article. Since proper names as such are already referential, the defi-
nite article clearly does not itself contribute to the referentiality or definite-
ness of the noun phrase. For this reason, the article which (necessarily) ac-
companies proper names in some languages (such as Greek) is sometimes
referred to as expletive or dummy. We will come back to expletive articles
with proper names in Chapter 2 section 2.3.3.
If, as we have just shown, the demonstrative itself implies definiteness
(see example (25) above), then in (26c) definiteness is conveyed by the de-
 Chapter 1 – The emergence and the structure of DP 77

monstrative afti (‘this’) and hence the definite article in (26c) cannot be re-
sponsible for contributing definiteness.
Further support for the fact that the definite article can be devoid of
semantic (referential) content is suggested by those examples in which it
seems to merely be used as a grammatical device to realize nominal ϕ-
features such as gender, number and case. Consider the German examples
in (27). With the genitive N Kaffees in (27a) and with the dative N Tee in
(27b) the determiner is required, while with the accusative N Kaffee in
(27b) and in (27c) it is not. There is no obvious difference in the referential
status of the nominal projection in these examples. On the basis of these
data, Giusti proposes that the realization of the article is required here sim-
ply by the need to express genitive or dative case.

(27) a. die Zubereitung *(des) Kaffees (German)

the preparation of *(the-GEN) coffee
‘the preparation of coffee’
b. Ich ziehe (*den) Kaffee *(dem) Tee vor.
I draw (*the-ACC) coffee *(the-DAT) tea for
‘I prefer coffee to tea.’
c. Ich trinke gerne (*den) Kaffee.
I drink gladly (*the-ACC) coffee
‘I enjoy drinking coffee.’

In addition, Giusti points out that in several languages the enclitic article
appears with the function of realizing nominal features. In the following
examples from Romanian (Giusti 2002: 64) the bound morpheme -ul which
is suffixed to nouns (or adjectives) in definite noun phrases, is also used as
the ending of indefinite pronouns and quantifiers. In this usage -ul is a fea-
ture marker for gender (here masculine), parallel to Italian -o, and for case
(here nominative):

(28)
Romanian Italian
un(*ul) băiat un(*o) ragazzo ‘a boy’
nici un(*ul) băiat nessun(*o) ragazzo ‘no boy’
am văzut pe un*(ul (ne) ho visto un*( o) ‘I saw one.’
N-am văzut pe niciun*(ul) non (ne) ho visto nessun*( o) ‘I saw no one.’
un*(ul) a spus că un*( o) ha detto che ‘Somebody said that…’
Nici un*(ul) a spus că … un*(o) ha detto che ‘Nobody said that…’
78 Part II – The functional make up of the noun phrase

In Albanian the presence of articles suffixed on adjectives depends only on

the type of adjective stem: in (29a) there is a suffixed article i associated
with the adjective mire (‘good’), in (29b) there is no suffixed article associ-
ated with the adjective besnik (‘faithful’). In (29a) the morpheme i is part of
the adjectival root and does not bear on the definiteness of the noun phrase.

(29) a. një djalë i mire, b. një djalë besnik (Albanian)

a boy the good a boy faithful (Giusti 1997: 104)

As shown by the English gloss of (29a), an indefinite article (nje) and a

definite article (i) co-occur in a single noun phrase, without giving rise to
any conflict in interpretation. This suggests that the definite article in this
example does not contribute ‘definiteness’.
By way of concluding this subsection we cite the following passage from
Giusti (2002: 65):

The definite article is neither sufficient nor necessary to trigger referential

interpretation on the noun phrase. This implies that the article is not the ele-
ment which 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 lan-
guages.

In section 6 we will discuss in some detail Giusti’s (1993) proposal that the
definite article should be seen as an alternative way of expressing case
morphology – under the general assumption that D is the locus of case, and
within a conceptual framework according to which case distinctions in sev-
eral languages approximate the definite-indefinite distinction. See Lyons
(1999), especially Chapter 9, for a survey of the literature on the historical
emergence of definite articles as the result of the loss of case marking on
nouns.
As a final comment on this extract, let us notice that Giusti’s claims (cf.
(iv) above) imply that there should be made a clear point on the distinction
between D as a structural position on the one hand and its realization by the
definite article on the other. Not all researchers adopt such a clear-cut dis-
tinction, with the result of D, as structural position, being rather regularly
confused with its primary occupant, the definite article, as has been pointed
out before.
 Chapter 1 – The emergence and the structure of DP 79

3. The DP hypothesis
3.1. Motivating ‘DP’
In section 2 we examined some of the basic semantic and syntactic proper-
ties that have been associated with the definite article. Though there seems
to be some relation between the presence of the definite article and referen-
tiality of the DP, as well as its capacity to function as argument, we have
shown that this relation is by no means universal or absolute.
Nonetheless, in recent generative work the article is standardly taken to
be located in D, the functional head of the nominal projection. And if the
article itself is not to be taken as responsible for the encoding of referential-
ity (or similar semantic notions), then the conclusion might be that it is D,
the syntactic category realized by the article that is responsible for that part
of the semantic interpretation of noun phrases. Such a conclusion would re-
late one or more semantic properties to a specific syntactic position, D, and
would hence tie in directly with the DP hypothesis. In this section we re-
view some of the evidence for the DP hypothesis.
In the Introduction to the book we mentioned that usually three types of
evidence are used to identify the presence of a functional head: semantic
evidence, morphological evidence and syntactic/distributional evidence.
The evidence we have provided in sections 2.2 and 2.3 in support of the
head D (and the related projection DP) was primarily semantic: given that
the article, as the typical filler of D, does not unequivocally encode
definiteness or related semantic notions, these properties must be derived
from another source; arguably that source is the functional head D itself.
In this section we will be concerned with syntactic/distributional evi-
dence for postulating the functional head D and the functional projection DP.
In Chapter 3 we will provide morphological evidence in order to establish
that there are actually more functional heads within the extended nominal
projection. To anticipate these discussions: the core idea behind the various
types of motivation of the category DP is simple: in order to be able to ac-
commodate facts which were not given due attention some thirty years ago,
we need to postulate more positions, or a richer structure.

3.1.1. Phrasal movement inside the nominal projection

One type of evidence corroborating the need to expand the nominal struc-
ture is the observed parallelism in the behavior of focused constituents on
the level of the sentence and on the level of the nominal domain: in both
80 Part II – The functional make up of the noun phrase

cases such constituents can be fronted (Horrocks & Stavrou 1987: 86). Sen-
tential focusing in Greek is illustrated in the examples in (30b) and (30c), in
which boldface signals contrastive stress:

(30) a. Edhose to vravio tis Afrodhitis.

gave-3SG the prize the-GEN Aphrodite-GEN
‘He gave the prize to Aphrodite.’
b. Tis Afrodhitis edhose to vravio.
the-GEN Aphrodite-GEN gave-3SG the prize
‘To Aphrodite he gave the prize.’
c. To vravio edhose tis Afrodhitis.
the prize gave-3SG the-GEN Aphrodite-GEN
‘The prize he gave to Aphrodite.’

As can be seen, the indirect object tis Afrodhitis (‘Aphrodite’) in (30b) and
the direct object to vravio (‘the book’) in (30c) can be fronted for focalizing
effects. Horrocks & Stavrou (1987) show that focalization in the clause has
the properties of A’-movement. We can assume that the fronted constituent
is moved to the CP domain.
Now consider the word order in the nominal projections in (31), com-
paring the position of the boldfaced constituent in the primed examples
with its position in the non-primed examples:

(31) a. to vivlio afto

the book this
‘this book’
a’. afto to vivlio
b. to vivlio tu Chomsky
the book the-GEN Chomsky
‘Chomsky’s book’
b’. tu Chomsky to vivlio
c. i kritiki tu vivliu
the review the-GEN book-GEN
‘the review of the book’
c’. tu vivliu i kritiki
d. to endhiaferon ja to arthro afto
the interest about the article this
‘the interest in this article’
d’. ja to arthro afto to endhiaferon
 Chapter 1 – The emergence and the structure of DP 81

In (31) the constituent following the head N in the non-primed examples

has been fronted to a pre-N position in the primed examples. The interpreta-
tive effect of such fronting is that of focalizing. Horrocks & Stavrou (1987:
86) say: “All this is obviously reminiscent of the fronting of constituents
that takes place in sentences for the purpose of bringing a particular con-
stituent into prominence.”
In (31) focalization takes place within the nominal projection. If the
fronted constituents in the examples in (30) occupy SpecCP, then in the
light of examples in (31), we can plausibly assume that the noun phrase
must have a similar peripheral position to host focus-moved constituents.
Notice crucially that the focused constituent in the nominal projections in
(31) is found to the left of the article. This means that the moved constitu-
ent targets a position to the left of the position of the article. By analogy to
what happens in the clause, this position can be argued to be a specifier po-
sition. If there is a specifier position to the left of the article, then it is plau-
sible that the article is the relevant head. Thus the article would be in D and
the fronted constituent in SpecDP. This makes SpecDP parallel to SpecCP:
DP is to NP what CP is to VP.
The parallelism between interrogative clauses and interrogative DPs in
Greek strengthens the hypothesis that DP is to NP what CP is to VP. (32a)
is an echo question: the wh-constituent ti (‘what’) does not move to the sen-
tence-initial position. In (32b) the wh-constituent is fronted (all examples
from Horrocks & Stavrou 1987).

(32) a. Ekane ti?

did-3SG what
‘He did what?’
b. Ti ekane?
what did-3SG
‘What did he do?’

The same pattern is again found DP-internally (33):

(33) a. to vivlio tinos?

the book who-GEN
‘whose book?’
b. tinos to vivlio?
who-GEN the book
‘whose book?’
82 Part II – The functional make up of the noun phrase

(34), from Horrocks & Stavrou (1987: 89, their (14)), illustrates the interac-
tion between wh-movement at the clausal level and DP-internal wh-move-
ment. (34a) corresponds most closely to a sentence with minimal move-
ment; in (34b) the interrogative genitive tinos (‘whose’) has been fronted
DP-internally. Following standard practice at the time that Horrocks and
Stavrou wrote their paper, we use the symbol ‘t’ for trace to indicate its
original position. In (34c) and in (34d) wh-movement affects the object DP.
We assume that it transits through the lower SpecCP, in which it leaves a
trace (t), to reach the higher SpecCP. In the former case, wh-movement
does not apply DP internally, and the interrogative genitive tinos (‘whose’)
follows the head noun vivlio (‘book’); in the latter case, the interrogative
pronoun tinos has fronted DP-internally and precedes the definite article to
and the head noun vivlio. In (34e) the interrogative pronoun tinos (‘whose’)
is fronted all by itself to the matrix SpecCP. 16

(34) a. Mu ipes [CP pos dhjavases [DP to vivlio tinos]]?

me-GEN said-2SG that read-2SG the book who-GEN
‘You told me you read whose book?’
b. Mu ipes [CP pos dhjavases [tinos to vivlio t]]?
c. [CP [To vivlio tinos] mu ipes [CP t pos dhjavases t]?
d. [CP [Tinos to vivlio t] mu ipes [CP t pos dhjavases [t ] ?
e. [CP [Tinos ] mu ipes [CP t pos dhjavases [t to vivlio t]] ?

Let us dwell on the derivation of (34e) for a moment. With respect to A’-
movement in the clausal domain, it is generally assumed that a lower
SpecCP serves as an escape hatch for A’-movement to a higher clause, i.e.
SpecCP is the position from which a focalized phrase or a wh-phrase may
be extracted from the clause. This step-by-step derivation follows from
considerations of economy: whatever the precise formulation, it is gener-
ally assumed that movement to a particular position takes place via inter-
mediate landing sites (for formal discussion see Chomsky 1986; Rizzi
1991; Chomsky 1993, 1995; for an informal introduction see Haegeman
and Guéron 1999; Haegeman 2006). In (34c,d) the trace in the intermediate
SpecCP signals this intermediate step in the movement.17 Since we postulate
that DP parallels CP, we assume that in (34e) tinos (‘whose’) has moved
via the intermediate A’-positions, i.e. SpecDP and the lower SpecCP.
———–——————————
16
We use traces here, following Horrocks and Stavrou’s own practice. In Mini-
malist annotation traces are replaced by copies of the moved constituents. In this
book, we sometimes use traces for reasons of clarity.
 Chapter 1 – The emergence and the structure of DP 83

Note that the extraction of tinos from the DP in the Greek example (34e)
may at first appear surprising: ever since the late 1960s it has been assumed
(Ross 1967; Chomsky 1977) that the nominal projection as such tends to
disallow extraction.17 This generalization has become known as the Com-
plex NP Constraint. It is illustrated in the English examples (35a) and (35b).

(35) a. *To whom did you like [the book that your brother gave]?
b. *To whom did you repeat [the story that your brother had given a
book]?

In (35a) to whom is extracted from the NP the book that your brother gave,
which contains a relative clause, and the result is ungrammatical. In (35b)
the wh-phrase whom is extracted from a clausal complement of a nominal
head, story, again leading to ungrammatical results.
However, Greek patterns differently in this respect. Consider the follow-
ing Greek examples:

(36) a. Akuses [ti fimi oti o Petros ce i Maria xorisan]? (Greek)

heard-2SG the rumor that the Peter and the Maria separate-3PL?
‘Did you hear the rumor that Peter and Maria got separated?’
b. Pji akuses [ti fimi oti xorisan]?
who heard-2SG the rumor that separated-3PL?

In (36a) the sequence ti fimi oti o Petros ce i Maria xorisan (‘the rumor that
Peter and Maria got separated’) contains a nominal head fimi (‘rumor’)
which takes a clause as its complement. However, in (36b) pji ‘who’) is ex-
tracted from that clause. This is predicted to be banned by the complex NP
constraint. And yet, though judgments are subtle here, (36b) is grammatical
in Greek.18 This leads Horrocks & Stavrou (1987) to conclude that there
must be a position in the noun phrase which can be used as an escape hatch
for movement. They propose that the specifier position of D is the relevant
escape hatch.19
———–——————————
17
See Corver (1990), Gavruseva (2000), Haegeman (2004), and also the discussion
of Bošković (2005) in Chapter 2, section 3.3.2.
18
See Horrocks & Stavrou for discussion about the subtlety of judgments about
such cases and also for why extraction is altogether excluded out of (restrictive)
relative clauses, which are also complex noun phrases.
19
See Horrocks & Stavrou for extended account of these and related contrasts in
terms of subjacency: they assume that in English IP/DP are bounding nodes,
while in Greek, IP and DP may not be bounding nodes.
84 Part II – The functional make up of the noun phrase

Additional justification for the SpecDP position as an A’-position in the

nominal projection comes from English examples such as (37):

(37) a. [AP How important] is this decision?

b. [DP [AP How important] a decision] is this?
c. This is [DP a [AP very important] decision].

(37a) illustrates a root wh-question formed by A’-movement of the wh-

phrase, how important, and I-to-C movement of the auxiliary. (37b) is also
a root wh-question, in which the preposed constituent is a DP, how impor-
tant a decision. Within the DP how important a decision, the adjectival
wh-phrase how important precedes the indefinite article a.20 The pre-article
position of the adjective important is not the usual position of an adjective,
as shown by (37c). We assume that (37b) illustrates a DP-internal applica-
tion of wh-movement. At first sight, the moved AP in (37b) occupies the
specifier position of DP. This again suggests an analogy between the speci-
fier of DP and the specifier of CP.
All in all then, extraction data suggest that there is a need for a specifier
position to host constituents that undergo movement inside (and out of) the
nominal projection. Typically, such a specifier position will be related to a
functional head. The appropriate type of a specifier becomes available if we
assume there is a functional projection dominating NP; this projection is
the functional projection of the head D, DP. Below we will consider in de-
tail the properties of the specifier position of DP. The hypothesis that NP is
dominated by a functional projection DP is referred to as the ‘DP-hypo-
thesis’.

3.1.2. PRO and the nominal projection (see also section 5.2)

A different type of syntactic evidence for the ‘DP hypothesis’ is discussed

by Abney (1987, section 4 of Chapter 2 in Part II) and in the same spirit
(with minor differences) by Stowell (1991). This evidence involves the dis-
tribution and interpretation of PRO, a pronominal anaphoric non-overt
category which, in the Principles and Parameters model, is taken to be un-
governed. Consider the following data with derived nominals (from Stowell
1991: 43):
———–——————————
20
It must be noted that the phenomenon is restricted to singular only, *how impor-
tant (some) decisions are these? being ungrammatical.
 Chapter 1 – The emergence and the structure of DP 85

(38) a. Johni needs [a PROi good talking to ti].

b. The government condemns [the (PRO-arbi) destruction of private
property (PROi) to make a political statement].

The bracketed NPs in the examples above are the complements of the verbs
need and condemn respectively. The relevant objects receive a theta role
from these verbs. Hence, in terms of the Barriers model (Chomsky 1986),
these constituents would be lexically marked. According to the Barriers
model, a maximal projection can be a barrier if it fails to be lexically
marked, that is, if it is an adjunct/modifier or if it is the complement of a
functional category such as C. Such a barrier may block government. If the
objects of the verbs need and condemn were simple NPs, these NPs would
be lexically marked and hence could not constitute a barrier for outside
government. This means that the constituent PRO would be governed be-
cause there would be no barrier to block government by the verbs. Embed-
ding the lexical NP under the functional category D as its complement en-
sures that it will serve as a barrier.21 So on these grounds, the category DP
can be justified on independent principles or subtheories of the grammar.

3.1.3. Head movement inside the nominal projection

Since the late eighties, the hypothesis has been advanced that in certain
languages the noun moves from its base position in the lexical domain to a
higher functional position within the extended projection of N. The earliest
such accounts include Dobrovie-Sorin (1987), Grosu (1988), Taraldsen
(1991), Ritter (1991), Longobardi (1994), among others. Evidence for such
an operation was provided by strings such as those in (39) in which the
head N appears to precede constituents that it tends to follow in other lan-
guages:

(39) a. gutt-en, hus-et (Norwegian)

boy-the, house-the
b. portretul (Romanian)
portait-the
———–——————————
21
To be precise: both NP and DP will block government of PRO by the higher
verb, because NP is a blocking category; DP is also a barrier because it immedi-
ately dominates a blocking category (Stowell 1991: 43).
86 Part II – The functional make up of the noun phrase

In all the examples in (39) the noun precedes the article, which is suffixed
onto it. Since articles typically occupy a position at the left edge of the
nominal phrase, it was then assumed that the noun moved to a higher posi-
tion attracted by the article. In particular, the assumption was that move-
ment was triggered by to the enclitic nature of the article which made it a
bound morpheme. The position to which the N is moved was identified as a
head position (D) under the independent principle that a head can only move
to a head position. The process of N-movement (which is discussed in more
detail and also evaluated critically in Chapter 3 and also in Chapter 1 of Part
III) is illustrated in (39c):

(39) c. DP

D’

D NP

-et N’

N
hus

Raising of N is also observed in languages without enclitic articles. Cf. the

following noun phrases from Italian:

(40) a. il mio Gianni

‘the my John
b. Gianni mio
John my
‘my John’

In (40a) the article il (‘the’) and the proper name Gianni co-occur, the pos-
sessive adjective mio (‘my’) intervening between the two. However, in (40b)
there is no definite article and the noun Gianni appears to the left of the
possessive adjective mio. As the ungrammaticality of (40c) shows, an initial
N is incompatible with a definite article. These data lead to the hypothesis
that the initial N in (40b) in fact occupies the position of the article in (40a):

(40) c. *mio (il) Gianni

 Chapter 1 – The emergence and the structure of DP 87

The generalization Longobardi (1994: 623) draws from all these and related
data is that in the absence of the article, D still exists but it is phonologi-
cally empty. The proper name then obligatorily moves to the head D in or-
der to fill this empty position and for the whole nominal projection to be in-
terpreted (under the assumption that semantic rules ‘see’ primarily D). In
Chapter 2 we will have more to say about why D must not be empty. In
Chapter 3 we will see that in addition to moving to the functional head D,
there also seems to be evidence for the hypothesis that N moves to func-
tional heads lower than D, which will lead us to postulate additional func-
tional heads.
N-to-D-movement has so far been proposed for two kinds of evidence:
morphological evidence, as is illustrated by the existence of enclitic arti-
cles, and syntactic, as is illustrated by those Italian patterns in which proper
names appear at the left edge of the noun phrase.22 In either case, the result
is a welcome one: the functional head D is justified, in fact it is necessary.
In section 4 below we will examine further how all the observations
made above and hypotheses based on them can be implemented in terms of
DP structure.

3.1.4. DP and the concept of definiteness

Before closing this subsection, it would be useful to turn to a point we

made at the beginning of the chapter, namely the linguistic realization of
the concept of definiteness. Recent research in the generative framework –
both semantic and syntactic – has reached the unanimous conclusion that
grammaticalization of definiteness implicates D (see also next chapter). On
the other hand, it becomes obvious when one goes through the relevant lit-
erature that what is ‘translated’ syntactically through D is reference/referen-
tiality. It thus appears that the other concepts related to definiteness are not

———–——————————
22
But it must be pointed out here already that the hypothesis about the raising of N
to the higher head D, or in fact to intermediate head positions in the extended
projection of N, has been challenged more recently and that in the light of new
theoretical findings and more empirical evidence (Giusti 1994a, 1994b, 2002;
Alexiadou, Stavrou & Haegeman 2001; Bouchard 2002; Shlonsky 2004; Cinque
2005; Laenzlinger 2005, among others), it has eventually been abandoned by
many authors, at least for some languages. This issue is to be further discussed
in Chapter 3 and also in Part III, Chapter 1.
88 Part II – The functional make up of the noun phrase

immediately linked to D/DP, unless it turns out to be the case that referen-
tiality can be argued to subsume all other concepts.
Lyons (1999) claims that D itself is basically the vehicle of semantic
definiteness, which is syntactically encoded as the (grammatical) feature
[+/–DEF]. The relation of [+DEF] to meaning can vary from language to lan-
guage.23

Thus for languages in general there is a range of noun phrase uses which
can in principle be characterized as definite, because they can be described
in terms of identifiability or inclusiveness. These uses represent “semantic
definiteness” but this is not what articles encode. A given language need
not treat the full range of these uses as grammatically definite; so the feature
specification [+DEF] can segment the semantic field at different points in
different languages, its range in a particular language being shown by which
uses require the presence of the definite article or other definiteness marker.
(Lyons 1999: 159, italics A-H-S)

Lyons accordingly assumes that DP is in fact a definiteness phrase. He

writes:

it is reasonable to suggest that only definite determiners are associated with

D and its projection DP. (..) D is definiteness (..) and DP is a definiteness
phrase. So the grammatical category which I have claimed definiteness is
has its representation in syntax in the form of this functional head. This
claim fits in well with the fact that nearly all other proposed functional
heads correspond to grammatical or semantic categories rather than to word
classes. (Lyons 1999: 298–299)

Lyons makes quite explicit a hypothesis that remains implicit in standard

analyses and which is a crucial underpinning of the DP-hypothesis: that DP
is the locus of definiteness. In particular, Lyons claims (1999: 301) that, like
all free form articles, the definite article in English, occupies the specifier
position of DP. If we assume that the free definite article occupies a speci-
fier position and if modifiers also are taken to occupy a specifier position,
———–——————————
23
Since the definite article as such does not encode definiteness, the latter being a
property of D, it can be argued that the definite article is a mere filler for
SpecDP and as such is like the filler for the subject position in (i):
(i) It is likely that Mary won’t be on time.
In this sense the definite article is an expletive, i.e. a meaningless filler for
SpecDP. See also the discussion in Chapter 2, section 2.3.4.
 Chapter 1 – The emergence and the structure of DP 89

the free definite article is similar to a modifier.24 Affixal articles, on the

other hand, would be taken by Lyons to occupy the D position.

3.2. Some challenges for the DP-Hypothesis

Before closing this section, we will briefly return to some of the earlier ar-
guments which have been formulated against the DP-hypothesis.25

3.2.1. N incorporation

On the basis of evidence from incorporating languages (see Baker 1988;

Payne 1993), Payne (1993) casts doubts on the correctness of the hypothe-
sis that the noun phrase is not a single-headed endocentric projection but
that it can be split into a functional domain (DP) and a lexical domain (NP).
In incorporating languages, object noun phrases can incorporate into the
verb, as in the following examples (from Payne 1993: 123, with reference
to Baker 1988: 94):

(41) a. Wisi seuan-in bi-mŭ -ban.

two man-PL I.SG-see- PAST
‘I saw two men.’
b. Wisi bi-seuan- mŭ -ban.
two I.SG-man-see – PAST
‘I saw two men.’

The representation of the resulting incorporation is given in (42) (from

Payne 1993: 123, slightly adapted): 26

———–——————————
24
Modifiers are, for instance, treated as specifiers in recent proposals by Cinque
with respect to the distribution of adverbial and adjectival modifiers (1994,
1999, 2005). Kolliakou (1997) also assumes that the definite article in Greek, as
a marker of definiteness, is an adjunct because it functions like a modifier in that
it restricts the reference of the noun.
25
See also Lyons (1999, paragraph 8.2.3) who points to a couple of weaknesses of
the DP analysis.
26
Payne uses English morphs in the diagrams he reproduces from Baker.
90 Part II – The functional make up of the noun phrase

(42) S

NP VP

V NP

N V QP N

I mani saw two ti

To account for this pattern we may formulate the hypothesis that the noun
raises and adjoins to V. It leaves a trace. The assumption is that traces have
to be either lexically governed or antecedent-governed. In the incorporation
structures, the moved N can antecedent-govern its trace. As Payne points
out, antecedent-government can only hold if there is no intervening head,
such as, for instance, a Quantifier, to block the government relation. But in
a conception of the nominal phrase in which NP is dominated by one or
more functional categories, a problem emerges. The intervening functional
heads would in principle block the movement of the noun since they would
present an obstacle to the requirement of antecedent-government. For one
account for this problem, Payne refers to Baker & Halle (1990). Baker &
Halle (1990) draw a distinction between a functional and a lexical head as
far as their potential for antecedent-government is concerned. Functional
heads (Q, for example) cannot themselves antecedent-govern a lexical trace
and, not being potential antecedent-governors, they do not create a mini-
mality effect, and furthermore the noun is permitted to incorporate into a
verb even with a QP intervening.

3.2.2. D incorporation to N

As a further problem for the DP analysis, Payne cites languages in which

any noun modifier (adjectives, quantifiers, demonstratives, numerals) can
incorporate into the noun itself. The following example is from a dialect of
Koryak, attributed by Payne to Žukova (1980):

(43) a. Yoten-ra -k (Koryak)

this house-LOC
‘in this house’
 Chapter 1 – The emergence and the structure of DP 91

b. Ỵәmәk -ra -k
my house -LOC
‘in my house’

Payne argues that such cases pose a serious problem for an analysis accord-
ing to which the demonstrative (or the possessor) heads a different projec-
tion than the one that N is in. The representation of the structure after in-
corporation would be as follows (Payne 1993: 127):

(44) a. DP

D NP

ti N

D N

үoteni ra -k

(44) requires lowering of the demonstrative onto the noun. From its new
position the demonstrative does not antecedent-govern its trace, in fact the
reverse is the case: the trace governs the demonstrative, an improper situa-
tion altogether. According to Payne, the same problem would arise after
numeral incorporation into the noun.
However, it is obvious that the structure above is not the only conceiv-
able one. We outline just one possible alternative among many. Assuming
that the demonstrative originates in a position lower than D (see below for
discussion) the derivation might not require reference to lowering. It is, for
instance, conceivable for N to move to a higher functional head, say D, and
that from a lower position the demonstrative incorporates into the noun in
that functional head.
92 Part II – The functional make up of the noun phrase

(44) b. DP

D’

D FP

F’

F NP

Dem-N tdem tn tn

Though obviously this analysis would have to be motivated independently,

it is clear that there might be a way of solving the problem raised by
Payne’s example above.

3.2.3. Agreement in the extended projection of N

As a third problem for the DP hypothesis, Payne points to problems related

to the phenomenon of extended agreement between the noun and its modi-
fiers which is found in a number of languages (see, for example the case of
polydefiniteness in Greek discussed in Chapter 1 of Part III). The evidence
that Payne focuses on concerns the scrambling of the nominal projection
around the noun, which according to Payne shows that the noun has a spe-
cial status within the noun phrase.
Payne’s initial assumption is that the base position of nominal modifiers
is prenominal. The modifiers “occur in a fixed order, and a single person,
number and gender marker is postposed to the whole noun phrase” (Payne
1993: 133). Modifiers appearing to the right of the noun must hence have
been postposed. In other words they no longer occupy their base position to
the left of the noun. This reordering has an effect on the morphological prop-
erties of the modifiers. When postposed, each modifier bears its own per-
son, number and case marking. Payne interprets the structures in which the
noun is the first constituent having its own person-gender-number marking
in which it is followed by one or more agreeing modifiers as appositional.
The status of Payne’s counterevidence is not quite clear: it seems to be
dependent on the hypothesis that all ‘traditional’ nominal modifiers such as
adjectives, numerals and quantifiers, must be heads in the extended nominal
 Chapter 1 – The emergence and the structure of DP 93

projection. However, this is not necessarily the case. In fact, as we shall fur-
ther discuss in Chapter 3 and also in Chapter 1 of Part III, rather than being
heads selecting for this NP themselves, at least certain (classes of) adjec-
tives can plausibly be argued to occupy the specifier position of functional
projections dominating the lexical NP. Agreement between the various
modifiers does not entail that they are heads. Agreement can, for instance,
be ensured in terms of specifier head relations between the agreeing modi-
fiers in the specifier positions of functional projections intervening between
NP and DP and the heads of these functional projections. In Chapter 3 we
will see that at least one interpretation of these functional projections is to
interpret them as agreement projections.
The DP-hypothesis, the idea that the nominal projection is selected by a
functional head D, is based on the grounds we saw earlier on in this chapter
and does not obligatorily entail head status for adjectives, or for any other
prenominal modifiers for that matter. The status of such elements should be
considered on independent grounds (see Chapter 1 of Part III for discussion
on this issue).
Payne’s objections to the DP hypothesis seem to be motivated by his as-
sumption that N has a pivotal role in the nominal projection. However, this
hypothesis is not incompatible with the DP hypothesis. Given Grimshaw’s
(1991) hypothesis about Extended Projection, whereby all the projections
within the DP are necessarily of nominal nature, in the same way that all
the projections in the clausal domain are of verbal nature, the pivotal role of
N is maintained. (See also Chapter 2 of Part III for the same idea imple-
mented in terms of the categorial features by Riemsdijk 1998).

4. Determiners, demonstratives and DP

4.1. The interpretation of demonstratives

In this section we turn to demonstratives. According to the earlier proposals,

in which demonstratives and articles are taken to occupy the same position,
they would have the same syntactic status. Concretely, this means that in
earlier accounts articles and demonstratives could be argued to occupy the
specifier of NP; in the more recent accounts both could be argued to occupy
the D position. Whichever account is chosen, if articles and demonstratives
occupy the same slot, they should have the same distribution. This predic-
tion is correct for English. However, comparative evidence shows that de-
monstratives and articles do not always compete for the same position. In
94 Part II – The functional make up of the noun phrase

some languages demonstratives may co-occur with the definite article, while
in other languages demonstratives and articles actually have to co-occur. We
will take up this point below.
The evidence we have discussed so far points to the article as a realization
of a functional head D, which selects the lexical NP and projects DP. As a
result of being embedded under D, the NP can function as an argument (of
a verb, for instance). The basic structure elaborated so far is given in (45):

(45) DP

D’

D NP

N’

Recall that according to earlier versions of X’ theory (Jackendoff 1977),

NP was taken to be a single (endocentric) projection of N. Its leftmost speci-
fier position was taken to be occupied by a number of different elements
such as articles, demonstratives, possessive NPs, possessive adjectives and
interrogative pronouns. The basic reason for grouping of all these elements
was the fact that in English they cannot co-occur; in other words they have
the same distribution:

(46) a. *John’s the book / *the John’s book

b. *this the book / *the this book
c. *the his book / *his the book
d. *John’s this book / *this John’s book

So, associating all the relevant items with one syntactic position leads to
the prediction that these constituents will be in distributional equivalence,
i.e. only one of them will show up. A combination of two or more of them
should be ungrammatical. In the earlier proposals the unique position was
the specifier of NP; following the DP hypothesis and assuming that articles
are hosted by D, we would then have to propose that the elements listed
above occupy D. However, the co-occurrence restrictions identified for
English and which form the basis for a unique position are not universal.
Once we observe that, say, a possessive pronoun and a determiner can co-
occur, as in Italian and in Greek:
 Chapter 1 – The emergence and the structure of DP 95

(46) e. il mia gatta (Italian)

the my cat
f. i ghata mu (Greek)
the cat-my
‘my cat’

We conclude that we need to postulate at least two positions: one to host

the determiner and one to host the possessive.
In addition, even with respect to the English data, the argument is not
straightforward. We have already drawn attention to the fact that genitive
noun phrases (John’s, the student’s, etc.) cannot be treated uniformly with
determiners and demonstrative pronouns: the former belong to open classes
and are definitely phrasal constituents whereas the latter belong to closed
classes, and articles could plausibly be argued to constitute heads.
A salient difference between demonstratives and the definite article is
that demonstratives are a universal category: unlike the definite article, de-
monstratives are found in all languages. They further have a double usage:
they are used either intransitively (English this, Dutch dat, Greek afto),
without an NP complement, or transitively with an NP complement (English
this book, Dutch dat book (‘that book’)), in what is often referred to as the
adjectival use of demonstratives. As will be seen later on, demonstratives
can also be either simple (e.g. French ce livre (‘this book)), or complex,
namely compounded with an adverbial reinforcer (e.g. French ce-ci as in ce
livre-ci (‘this book here’)). This ‘complex’ character of demonstratives has
repercussions on their syntactic account, as we will see.
Demonstratives do have some similarities to definite articles. Like the
definite article, demonstratives form a closed class, they lack substantial de-
scriptive bulk and hence they can be argued to constitute a grammatical/
functional category. Nevertheless demonstratives seem to have slightly more
descriptive content than the definite article. We will review evidence found
in the literature (Giusti 1997, 2002, see in particular section 4.2) that the dis-
tinguishing feature of demonstratives, viz. the deictic feature or [DEM],
constitutes a kind of descriptive content. In what follows we will also see
that the semantic difference between demonstratives and (definite) articles
amounts to demonstratives having a clearly identifiable semantic value: they
are responsible for the deictic interpretation (and thus indirectly of referen-
tiality) of the noun phrase (Giusti 1997, 2002). This property holds irrespec-
tively of the fact that articles historically derive from demonstratives, an
important issue which we will briefly address in the following paragraph.
96 Part II – The functional make up of the noun phrase

4.1.1. A note on the historical connection between demonstratives and the

definite article

In a sense, the definite article and the demonstrative are distributionally

(and functionally) similar: both pertain to the determination area, as said in
the introduction to this part of the book, at some stage of the syntactic deri-
vation both are found within the domain of the head D (see section 4.2). In
languages lacking a separate morpheme for the definite article, demonstra-
tive pronouns are sometimes used with the function of the article. This is,
for instance, the case in Polish and in Latin (see Chapter 2), and even in
Asian languages where classifiers may also play a limited role as articles.
Thus, in Japanese, although a bare NP may denote a definite or indefinite
NP, to stress definiteness a demonstrative is used.
As we will discuss below (section 4.1.2), both definite article and de-
monstrative are definiteness markers (for demonstratives as definite markers
see also the discussion of (25) above). This common property of the defi-
nite article and the demonstrative is not an accident. Indeed it is a reflex of
their diachrony. In general, in those languages in which they do exist, defi-
nite articles have emerged from demonstrative pronouns through a process
of semantic and phonological weakening. For instance, French le, Italian il
and Spanish lo have all developed from the Latin demonstrative ille, through
its phonological reduction from (il)le to le (Greenberg 1978; Lyons 1977;
Giusti 2001 and references therein;27 also Lyons 1999, Chapter 9, among
others).28 The Modern Greek article (realized as o for masculine nouns, i
for feminine nouns and as to for neuter nouns) had a morphologically iden-
tical ancestor in Ancient Greek which was used exclusively as a demonstra-
tive pronoun (Jannaris 1897).29 The same pattern is observed in English.
———–——————————
27
See section 6 below for a different approach by Giusti whereby it is claimed that
the article emerges as a consequence of the weakening/loss of case distinctions
on the noun.
28
We could add here that according to Giusti (2001), this phonological weakening
had as a consequence the reanalysis (in time) of ille from its original position in
the specifier of DP to the head D.
29
As Jannaris (p. 317–318) points out, the final development of the definite article
in Greek becomes apparent: “only in Ancient prose, and particularly in Plato’s
philosophical language, where its presence or absence shows the finest differen-
tiations and distinction between individual and generic notions.” The definite ar-
ticle has retained its usage as a demonstrative in some fixed expressions, for in-
stance, to ce to ‘this and that’.
 Chapter 1 – The emergence and the structure of DP 97

The definite article of the modern language-as well as the modern demon-
stratives themselves-emerged from the Old English demonstrative pronoun
realized as se (masc.), seo (fem.) and Þæt (neut.).30
Notice that the historical approach to the emergence of the definite arti-
cle is independent of the theoretical approach which basically links the
emergence of the definite article to the development of DP structure, given
more general assumptions concerning the development of functional pro-
jections in the child. But as Lyons (1999: 322ff) points out, the theoreti-
cal/acquisitional approach alone cannot explain in a satisfactory way how
definite articles came about in those languages that have them. This is so,
Lyons says, because that would imply first that definiteness markers (the
definite article, in particular) must be determiners (Ds). This, however,
does not necessarily hold, according to Lyons, given the affixal nature of
articles in some languages. Secondly, “it is not clear why an adjectival de-
monstrative in a non-DP language should not weaken to express merely
definiteness while remaining adjectival; and conversely, it is not clear why
a language with determiners (because with DP) should not have only de-
monstrative, cardinal, etc. determiners, without a marker of simple
definiteness” (Lyons 1999: 322–323).31
However interesting this diachronic issue of the evolution of articles
from demonstratives and their respective impact on the expression of
definiteness in the nominal projection may be, we will not pursue it further

———–——————————
30
For reasons of space we offer a simplified account here. The article in English
might well have developed as a consequence of a series of changes concerning
the loss of adjectival inflectional endings. See Spamer (1979) for interesting dis-
cussion.
Similar facts are reported by Vangsness (2004) for certain Scandinavian lan-
guages, where the free morpheme den serves for both the article and the distal
demonstrative. In general, such processes are well-known cases of language
change. In Finnish se is slowly turning into a definite article undergoing a proc-
ess of grammaticalization which eventually will probably result in the definite
article (Laury 1997).
31
Greenberg (1978) hypothesizes a common pattern of ontological development
of the definite article which involves four basic stages: the zero stage with no
definite article as a means of the expression of definiteness, stage 1 whereby the
definite article emerges out of a demonstrative, in stage 2 the use of the definite
article becomes more general and finally in stage 3 it becomes grammaticalized
(a purely grammatical marker). The reader is referred to Greenberg (1978) and
references therein for this interesting issue.
98 Part II – The functional make up of the noun phrase

here. By way of concluding this historical note, we cite an illuminating ex-

tract from Krámský (1972: 34), in which the point where a demonstrative
becomes an article and the subsequent differences between them and their
source is discussed:

And here we come to the conclusion that we can speak about an article only
when the definite article indicates a noun in a GENERAL function (e.g. the
horse is an animal); an individual stands here for a whole class. If the pro-
noun has this meaning it becomes article. Let us mention another difference
between the article and the pronoun: the pronoun is only facultative whereas
the article is obligatory, it is a constant quality of the noun. Moreover, the
article is not mere determination: this results even from the theories which
we have discussed above. The fact that the article adds a definite element
(be it a concretizing, substantivizing, actualizing element, the element of
familiarity, etc.) to the noun is another important factor of the distinction
between articles and demonstratives. In contrast to demonstratives, the arti-
cle is always determination plus something else, some other element, which
modifies the meaning of the word. It seems that the article influences the
noun somehow from the inside, that is to say it influences the noun directly
in its very essence, whereas the demonstrative pronoun merely points from
the outside without substantially affecting the noun. The demonstrative pro-
noun does not insert anything into the noun to which it belongs.

4.1.2. The deictic interpretation of demonstratives

With respect to the interpretation of demonstratives, the important question

that arises concerns their relationship with the definite article: what is the
contribution of demonstratives to the interpretation of the nominal phrase
and how is this similar to/ different from that of the definite article? Both
the definite article and the demonstrative can be said to impart definiteness
and referentiality. One obvious difference between demonstratives and
definite articles is that, though both are definite, only the latter can be used
to refer to a kind term (see Chapter 2 for generic nouns): this dodo in (47b)
and this mobile phone in (48b) cannot refer to kinds.

(47) a. The dodo is extinct.

b. This dodo is extinct.

(48) a. The mobile phone has changed western culture.

b. This mobile phone has changed western culture.
 Chapter 1 – The emergence and the structure of DP 99

Obviously, demonstratives do not have a generic reading. This should be

related to the context dependent nature of demonstratives: their use is im-
mediately connected with the coordinates of the utterance. This is what it
means for the demonstrative to ‘point from the outside’ in the above pas-
sage from Krámský (1972).
Lyons (1999) points out that identifiability is the common denominator
of the definite article and the demonstrative. Just like the definite article,
demonstratives, being inherently definite, serve to identify the referent.
However, demonstratives are directly referential expressions in that they
can directly refer to entities of the linguistic or the extralinguistic (situ-
ational) world. Demonstratives denote entities without describing them. We
can say they lack descriptive content, i.e. while demonstratives have exten-
sion they lack sense (Sinn). Because, besides being definite, demonstratives
are also directly referential, they are considered to be among the deictic
elements of language, viz. those linguistic forms whose use and interpreta-
tion rely crucially on the context in which they are produced.
(49) illustrates the genuinely demonstrative use of the demonstrative that:

(49) That is Melissa’s favorite piece of clothing.

Deictic categories relate the linguistic entities that encode them to the spatio-
temporal, that is the extra-linguistic, context. Quoting Lyons (1977: 637):

By deixis32 is meant the location and identification of persons, objects,

events, processes and activities being talked about, or referred to, in relation
to the spatiotemporal context created and sustained by the act of utterance
and the participation in it, typically, of a single speaker and at least one
addressee.

Other deictic categories are person and tense: being context-related, the ref-
erence of deictic items naturally varies from utterance to utterance or from
context to context. Such context dependent expressions are also referred to
———–——————————
32
It is further useful to give another informative quotation from Lyons (1977: 637)
concerning the terminology employed: «The fact that the Latin-based term ‘de-
monstrative’ has been specialized in linguistic terminology in the sense that the
Greek grammarians gave to ‘deiktikos’, enables us to employ the terms ‘deictic’
and ‘deixis’ in a wider sense. And this is now common practice in linguistics.
(…) Deixis covers not only the characteristic function of the demonstrative pro-
nouns, but also tense and person, and a number of other syntactically relevant
features of the context-of-utterance».
100 Part II – The functional make up of the noun phrase

with the general term indexicals. Pronouns and temporal and locative ad-
verbs such as today, then and there are also indexicals. However, demon-
stratives differ from indexicals such as today, tomorrow, yesterday, I, you,
she, etc. in that they also require an associated demonstration – cf.: “typi-
cally, though not invariably, a (visual) presentation of a local object dis-
criminated by a pointing” (Kaplan 1977: 9).
Let us consider example (50):

(50) Yesterday Topsy gave birth to this lovely kitten here.

All the indexical elements in this example acquire a meaning which de-
pends on the context of their use. To interpret an utterance such as (50), one
needs to know who utters it, when and where, and what is being pointed at
(‘this kitten’) as it is being uttered.
The use of a demonstrative thus involves – to a greater or lesser degree –
a contrast among referents (Hawkins 1978; Lyons 1999)). For instance, in
(50) there is a contrast between ‘this’ kitten and some other kitten (that one,
John’s kitten, etc). While demonstratives and the definite article share the
semantic component of identifiability and referentiality, deicticity (or os-
tensivity) is what sets them apart. The definiteness of demonstratives is not
a matter of inclusiveness (Lyons 1999: 17). In demonstratives, deicticity or
ostentivity is combined with identifiability to give rise to their typical in-
terpretation.
The deictic component of demonstratives helps to locate the referent
with reference to some point in the non-linguistic context (but see immedi-
ately below for the anaphoric use of demonstratives). The deictic feature of
demonstratives is in general interpreted in two ways; either it is encoded in
the opposition [+/– proximal] (or, inversely, [+/– distal]), with the speaker
as the direct anchoring point. This cat denotes a cat that is found closer to
the speaker than does that cat. In this case, the deictic feature functions in a
way parallel to (physical) pointing/gesturing. As Lyons further points out
(1999: 18), the relevant distance may also be temporal (cf. the contrast be-
tween that day and this day/this week). Or, alternatively the deictic feature
is made contingent on the grammatical category of person; it then denotes
association or closeness to the speaker, or a set of individuals that includes
the speaker. For instance this cat can mean ‘the cat I have / I and you have
etc.). In other words, in this case this cat is associated with first person, viz.
with the use of the pronouns I and we. That cat on the other hand is used to
link the referent to a set of individuals that includes the hearer and excludes
 Chapter 1 – The emergence and the structure of DP 101

the speaker or a set which excludes hearer as well as speaker. In other

words, that cat is used to refer to a cat that is either associated with the
hearer (second person) (51a) or with some other entity that excludes the
hearer (third person) (51b).

(51) a. Show me that (?this) letter you have in your pocket.

(Lyons 1999: 18, his (61))
b. Tell her to bring that (?this) drill she has.
(Lyons 1999: 18, his (62))33

The deictic component inherent to demonstratives is taken by some lin-

guists (Giusti 1997) to be a semantic component. In this sense, deixis is the
semantic content that is missing from the definite article. As Giusti claims,
demonstratives, as opposed to articles, have a semantic value, this property
being crucial for “the interpretation of the referential index of the noun
phrase” (1997: 42). Pursuing this point, Giusti (1997) goes a step further,
casting doubt on the status of demonstratives as (completely) functional
elements.
Deixis/ostension is thus considered to be the defining property of demon-
stratives. However, Lyons (1999) makes a more specific claim. He argues
that the main diacritic property of demonstratives, which really sets them
apart from the definite article, is a more abstract feature which he labels
[+/–DEM], and which is to be kept distinct from the deictic feature.34 But
what does [+DEM] amount to? Lyons says that his [+DEM] feature can be
compared to Hawkins’ earlier (1978) ‘matching constraint’. According to
the matching constraint, the hearer is instructed to match the referent of the
DP with some object which is either identifiable/visible in the context (cf.
(49)), or which is known on the basis of previous discourse (Lyons 1999:
20). The contrast between the acceptable definite article and the unaccept-
able demonstrative in (52b) illustrates the point:

(52) a. I got into the car and turned on the engine. (Lyons 1999: 20)
b. *I got into the car and turned on this engine.

———–——————————
33
As Lyons notes (1999: 18–19), this could also be used in these examples if it
were thought as an appropriate word to refer to the speaker.
34
According to Lyons, [+Dem] and [+DEF] are intrinsically connected, so that
marking demonstratives as [+DEF] is even redundant. Demonstratives are nec-
essarily definite.
102 Part II – The functional make up of the noun phrase

In this example the definite article can be used to signal the need for activa-
tion of all-purpose knowledge, namely that cars have engines, and moreo-
ver that they have just one. On the other hand, the use of the demonstrative
signals that the referent must be located in the (non)linguistic context or the
immediate situation. The matching constraint looks close to identifiability
but the type of identifiability intended here is of a more restricted kind than
that implicated by the use of the definite article. The demonstrative signals
that the identity of the referent is directly accessible (cf. ‘direct reference’)
to the hearer (e.g. by pointing), without the need for the hearer to do any of
the inferencing that would be associated with processing definite articles
(Lyons 1999: 21). The context of (52b) is not such as to provide direct ac-
cessibility to the referent ‘the engine’. Hence the use of the demonstrative
is infelicitous. Example (8) above, repeated here as (53a), illustrated a simi-
lar point. Once again the referent of the DP the bride is accessible as a re-
sult of inferencing: there is one bride at a wedding. Again a demonstrative
would not be appropriate (53b):

(53) a. I’ve just come back from a wedding. The bride was wearing red.
b. I’ve just come back from a wedding. *This/that bride was wearing
red.

Lyons’s (1999) decision to replace (spatio-temporal) deixis by the more ab-

stract feature [DEM] is also motivated by the observation that demonstra-
tives are sometimes neutral with respect to spatio-temporal location. This
can be illustrated from a number of languages. English that, for example, is
sometimes neutral with respect to spatio-temporal location (54a), especially
when used as a pronominal in relative clauses (Lyons 1999: 19).

(54) a. She prefers her biscuits to those I make.

(Lyons’s 1999: 19, his ( 63))

Likewise, French demonstrative ce is itself neutral with respect to the cod-

ing of distance/proximity (Lyons 1999: 19). That is why it can be accom-
panied by so-called reinforcers: the bound morphemes -ci and -là are deictic
markers, which are attached to the noun and carry information about dis-
tance. This is shown in (54b). The demonstrative ce itself is neutral between
the proximal this and the distal that.

(54) b. Ce bateau-ci vs ce bateau-là

this boat -here this boat -there
 Chapter 1 – The emergence and the structure of DP 103

Such examples show that spatio-temporal deixis can be dissociated from

the more abstract property [DEM], which Lyons takes to be the constitutive
feature of demonstratives.
Egyptian Arabic is another example cited by Lyons (1999: 19–20) in sup-
port of his claim that, although deixis is a common property of demonstra-
tives, it is not an invariant property. In this language, there is a one-form de-
monstrative system, da (‘this/that’), which lacks deictic contrast altogether.
Not having been lexicalized or grammaticalized, information concerning
the distance of the referent from the speaker is retrieved on the basis of
other contextual means. Observe that Egyptian Arabic has a distinct mor-
pheme for the definite article (?il), so that it could not be argued that da is
an article.
Similar observations can be made for Modern Greek. Modern Greek has
a two-form demonstrative system:35 the relevant forms are afto (‘this’) and
ecino (‘that’). However, afto is often used in a neutral way with respect to
proximity or distance from the speaker, functioning more as a kind of de-
fault demonstrative of the language. Reinforcers can be inserted to make
the proximity/distance clear. This is shown in examples (55a–b), where the
deictic markers edho (‘here’) and eci (‘there’) are added to signal proximity-
distance from the speaker respectively.

(55) a. afto-edho to praghma

this here the thing
b. afto-eci to praghma
this that the thing

The second demonstrative, ecino (‘that’), is different: only the reinforcer

eci (‘there’) can be added to this demonstrative form (55c), edho (‘here’) is
excluded:

(55) c. Fere ecino-eci to trapezi.

bring-2SG-IMPER that-there the table
‘Bring that table there.’
d. *Fere ecino-edho to trapezi.
bring-2SG-IMPER that-here the table
———–——————————
35
It is interesting to note that the Modern Greek article is also related to a demon-
strative usage in pre-homeric Greek, expanded as hode in the classical period,
i.e. a form containing a reinforcer to act as a demonstrative.
104 Part II – The functional make up of the noun phrase

Ecino, then, is unambiguously used to exclude the speaker.

The Finnish demonstrative tämä is used to refer to items close to the
speaker, while the demonstrative tuo is used to refer to items further away.
The language has a third demonstrative se, which is considered to be neu-
tral:

(56) a. tämä
this
b. tuo
that
c. se
‘unmarked’ (Lyons 1999: 113)

To summarize the discussion of the status of demonstratives, we give the

following illustrative extract from Lyons (1999: 21):

A demonstrative signals that the identity of the referent is immediately ac-

cessible to the hearer, without the inferencing often involved in interpreting
simple definites. This may be because the work of referent identification is
being done for the hearer by the speaker, for example by pointing to the ref-
erent. The deictic feature typically expressed on a demonstrative plays a
similar role to pointing, guiding the hearer’s attention to the referent. This
suggests a necessary connection between [+DEM] and [+DEF], the former
implying the latter. I take demonstratives, then, to be necessarily definite.
(Lyons 1999: 21)

4.1.3. The anaphoric use of demonstratives

Before closing this subsection, a brief mention must be made of the most
common non-deictic usage of demonstratives, namely the anaphoric usage.
Anaphoric usage is going to be rather crucial in our syntactic account for
demonstratives (see in particular section 4.2.3). Consider the following ex-
ample:

(57) Every girl brought her favorite piece of clothing to school and wore
that to the party rather than her uniform.

Demonstrative that in (57) refers back to the expression her favorite piece
of clothing, i.e. to an entity referred to already available in the discourse. In
 Chapter 1 – The emergence and the structure of DP 105

this usage, the demonstrative acts like an anaphoric pronoun which is inter-
preted in terms of its connection to an antecedent linguistic expression.
Anaphora is a common non-deictic category involved in demonstrative
systems (Lyons 1999, section 3.1.2). In the anaphoric use of demonstratives
the deictic feature [+/–PROX] tends to be used for anaphoric reference, as
do the person-based systems; first person forms are used as proximal, in an
extended sense, and non-first or third person forms as distal. The Latin de-
monstratives hic (first person) and ille (third person) have regular anaphoric
uses as more (‘the latter’) versus less (‘the former’) recently mentioned.
Latin also has an anaphoric use of the form is, a deictically unmarked form.
But several languages have a special demonstrative for anaphoric usage
(Lyons: 114): such is the case, for example, of the element used in polyde-
finite constructions in Romanian, as well as in Greek (see Chapter 1 of Part
III for discussion of the polydefinite construction).
In their anaphoric use demonstratives can be seen as markers of topic-
hood. Some languages have a special demonstrative for this use. In others it
is the position of the demonstrative relative to the noun that brings about
the anaphoric interpretation. In the next section we will discuss a case from
Modern Greek, in which the demonstrative can serve as an anaphoric pro-
noun when found immediately after the noun or an adjective.
In the light of the observations made in this section, it can be concluded
that demonstratives, despite the fact that they constitute a closed class in all
languages, and even though they lack descriptive content as such, belong to
the (semantic) field of deixis. It can therefore be concluded that demonstra-
tives, unlike the definite article, may be seen as affording at least some se-
mantic content (Giusti 1997) – it essentially consists of the feature [+DEM].
In section 4.2 we examine a number of syntactic differences between the
definite article and demonstratives.

4.2. The syntactic representation of demonstratives

4.2.1. Demonstratives as maximal projections

We have already said repeatedly that in Jackendoff’s model of phrase struc-

ture, demonstratives and articles, along with other elements, belong to the
class of determiners36 and were assigned to the same structural position,
———–——————————
36
In fact Jackendoff distinguishes two major classes, demonstratives and the quan-
tifiers (Jackendoff 1977: 104). Articles and demonstratives belong to the first class.
106 Part II – The functional make up of the noun phrase

namely the leftmost specifier of the NP. As discussed above, this hypothe-
sis was meant to capture the fact that English articles are in equivalent dis-
tribution with demonstratives, so they could be thought of as occupying the
same position:

(58) a. This book is interesting.

b. That book is boring.
c. *This the book is interesting.
d. *The this book is interesting.
e. *That the book is boring.
f. *The that book is boring.

One important remark at this point concerns the cross-linguistic evidence

which strongly suggests that in fact the demonstrative and the determiner do
not occupy the same structural position. One argument for distinguishing the
demonstrative from the article is that in many languages the two elements
can co-occur:

(59) a. ez a haz (Hungarian)

this the house
b. ika n anak (Javanese)
this the baby
c. afto to vivlio (Greek)
this the book
d. omul acesta (Romanian)
man-the this

A further property that sets apart (definite) articles from demonstratives, al-
ready mentioned above, is that demonstratives may stand alone, intransitive-
ly as it were, a property not shared by the definite article, be it a bound or a
free morpheme: 37

(60) a. I like that. vs I like the *(book).

b. Ho visto quello. vs Ho visto il *(ragazzo) (Italian)
I have seen that vs I have seen the (boy)

This property of demonstratives is connected with their autonomy in the

sentence; in contrast to the definite article, they do not depend morphologi-
cally/phonologically on the noun (or an adjective, if there is one).
———–——————————
37
See Giusti (2002) for discussion of some apparent counterexamples.
 Chapter 1 – The emergence and the structure of DP 107

Giusti (1997: 112) discusses the evidence provided by the Italian sen-
tences in (61), which show two interrelated facts. The examples concern the
possibility of extracting material from within the DP, in particular of ex-
tracting a possessor. We see that extraction is possible provided there is no
demonstrative present (Giusti 1997: 111):

(61) a. Di chi hai la foto sulla tua scrivania? (Italian)

of whom have -2SG the picture on-the your desk
‘Whose picture do you have on your desk?’
b. *Di chi hai questa foto sulla tua scrivania?
of whom have-2SG this picture on-the your desk

The first conclusion from these examples is that definiteness as such should
not be taken as the factor blocking extraction. Both the definite article in
(61a) and the demonstrative in (61b) lead to definiteness in the DP. Extrac-
tion from the former is possible while it is blocked from the latter.
The second related conclusion concerns the syntactic status of the de-
monstrative vs. that of the article: Giusti assumes that the contrast in (61) is
due to the different syntactic status of the article and of the demonstrative.
If the article is a head, it does not as such block extraction of a maximal
projection (the di-phrase). On the other hand, Giusti proposes that the de-
monstrative is a maximal projection, it occupies a specifier position and
hence it blocks extraction. Extraction data thus offer some evidence for dif-
ferentiating demonstratives and the definite article in terms of their syntactic
status: the former are maximal projections, the latter are heads.
Further evidence for opposing the phrasal status of demonstratives and
the head nature of articles is provided by the following examples from Ro-
manian, also discussed by Giusti (1997: 107):

(62) a. acest băiat frumos

this boy nice
b. băiatul (acesta) frumos
boy-the (this) nice
c. frumosul băiat
nice-the boy
this nice boy
d. *frumosul acesta băiat
nice-the this boy
this nice boy
108 Part II – The functional make up of the noun phrase

Comparing (62a) with (62b) we see that the demonstrative acest (‘this’) can
be crossed over by the noun bajatul (‘boy-the’). In (62c–d) we observe that
the demonstrative acesta cannot be crossed by an adjective frumosul (‘nice-
the’) (see also Part III, Chapter 1). If the adjective is moved as a phrase (as
argued for by Giusti 1993, and Dimitrova-Vulchanova and Giusti 1998) the
fact that it cannot cross over the demonstrative is further evidence for the
maximal projection status of the demonstrative itself.

4.2.2. Articles in D; demonstratives in SpecDP

We can now combine the above observations with some more syntactic
facts about demonstratives in order to determine their location in relation to
the domain headed by D.
In some languages, English and Greek among them, there is an interpreta-
tive similarity between this and degree modifiers like such, in that both point
to an element known from the discourse context: such a reaction means,
roughly, ‘a reaction of this kind’. This is shown in the following examples:

(63) a. I did not expect this reaction.

‘I did not expect such a reaction.’
b. Dhen perimena afti tin antidhrasi. (Greek)
not expected-1SG this the reaction
c. Dhen perimena tetia antidhrasi.
not expected-1SG such reaction38

Observe also that the English demonstrative that is in fact used as a degree
modifier in an adverbial phrase or in an adjective phrase, and is comparable
to the degree adverb so:

(64) a. I did not expect it to happen [AdvP that quickly].

b. I did not expect it to happen [AdvP so quickly].
c. I did not expect [DP [AP that big] an audience]].
d. I did not expect [DP [AP so big] an audience].

In (64a) such immediately precedes the article in D. As a first approxima-

tion, it can be proposed that such occupies SpecDP. Exploiting the observed
———–——————————
38
Interestingly, traditional grammars of Greek list both aftos (‘that’) and tetios
(‘such’) under demonstrative pronouns.
 Chapter 1 – The emergence and the structure of DP 109

similarity between such, so and the demonstrative, we might formulate the

hypothesis that the demonstrative this in (63a) also occupies the specifier of
DP:

(65) DP

Spec D’

D NP

such a reaction
this
that reaction

This means that the demonstrative appears to be located in SpecDP. Indeed

it is evident from even a superficial inspection of DPs which contain a de-
monstrative in a wide range of languages that the demonstrative often oc-
cupies the leftmost position in the DP. A further question that arises now is
this: is this leftmost position the ‘base’ position or is it a derived position?
Putting the question differently: does the demonstrative start out in the left-
most position of DP or does it originate somewhere lower in the DP or, even,
in the NP? If the latter is the case, then the leftmost position of the de-
monstrative is a derived position to which it must have been moved.

4.2.3. A lower position for demonstratives

An assumption shared by many linguists is that the demonstrative is found

in the position of SpecDP as a result of movement from a lower position
(Giusti 1997, 2002; Brugè 2000, 2002; Brugè & Giusti 1996; Panagiotidis
2000; Grohmann & Panagiotidis 2005; Shlonsky 2004). Support for the
movement analysis comes from a number of languages in which the
demonstrative is indeed found in a lower position. This is illustrated for
Romanian in (66), Spanish in (67) and Greek in (68). As can be seen, in all
three languages the demonstrative may either be the initial constituent of
the DP or it may occur to the right of the head noun.39
———–——————————
39
Note that the difference between Romanian and Spanish on the one hand and
Greek on the other concerns the presence vs. absence of an overt determiner co-
occurring with the demonstrative. The determiner is licit (in fact it is obligatory)
in Greek, but not in Spanish.
110 Part II – The functional make up of the noun phrase

(66) a. acest băiat (frumos) al sau (Romanian; Giusti 2002: 71)

this boy nice of his
b. băiatul acesta (frumos) al sau
boy-the this nice of his

(67) a. este hombre (Spanish)

this man
b. el hombre este
the man this

(68) a. afto to vivlio (Greek)

this the book
b. to vivlio afto
the book this

Brugè (1996) argues that the patterns illustrated above can be accounted for
if one assumes that the demonstrative is generated in a low specifier posi-
tion. The relevant proposal is that the demonstrative is first inserted as the
specifier of a functional category immediately above NP. 40 In addition, it is
assumed that D contains a [+DEF] feature, which needs to be associated
with an overt element (i.e. lexicalized). This requirement may be satisfied
either by the definite article (66b, 67b, 68b) or by the demonstrative (66a,
67a, 68a). Let us consider the examples above.
In Romanian (66a), the uninflected demonstrative acest precedes the N
baiat (‘boy’), which in turn precedes the adjective, frumos. There is no de-
terminer on either the noun or on the adjective. In (66b) the noun baiat is
prefixed to the enclitic definite article -ul and precedes the demonstrative,
which bears agreement inflection (acesta).
If we assume that (66b) is derived by head movement of N, then the
Romanian data support the view that demonstratives are maximal projec-
tions. Head-movement of N in (66b) would cross the demonstrative
(acesta). If the demonstrative acesta were itself to be analyzed as a head we
would have to say that head-movement of the noun baiat can cross a head
in violation of the locality conditions on movement. The assumption that
demonstratives are maximal projections avoids this problem. If we assume
———–——————————
40
Vangsness (2004) also assumes demonstratives are generated at the spec of the
category that hosts the enclitic article in Icelandic. This category immediately
dominates the category that hosts the inflectional morpheme of the noun (the so-
called Word Marker (see Chapter 3)).
 Chapter 1 – The emergence and the structure of DP 111

that the noun in (66b) has moved to D, then the example also suggests that
the position of the demonstrative is lower that the D level.
The Spanish examples in (67a,b) display a similar pattern. The post-
nominal occurrence of the demonstrative in (67b) also offers evidence for a
lower position of the demonstrative. Note that it has been proposed that in
Spanish the noun also raises to an intermediate functional head. In (67b) the
demonstrative occupies a lower position: we can assume it remains in its
base position, the specifier of a functional projection between DP and NP.
The order in (67b) can be derived by noun movement to a head position be-
tween DP and NP. The prenominal occurrence of the demonstrative este
(‘this’) in (67a) can be interpreted as resulting from the demonstrative rais-
ing to SpecDP.
At this point we have derived Romanian (66b) and Spanish (67b) in the
same way: the demonstrative is kept in a low position and the noun moves.
It must be pointed out, however, that although in Romanian (66b) and in
Spanish (67b) the demonstrative immediately follows the noun+article and
the noun respectively, the post-nominal position of the demonstrative in
Romanian can be shown to be different from that in Spanish. Consider
(69):

(69) a. el cuadro redondo este suyo

the painting round this of his
b. tabloul acesta rotund al său
painting this round of his

In both Spanish (69a) and Romanian (69b) the demonstrative follows the
head noun. However, in (69a) the demonstrative este also follows the de-
scriptive adjective redondo (‘round’), while in (69b) the demonstrative
acesta precedes the adjective rotund (‘round’). If the demonstrative occu-
pies its base position in Spanish (69a) then (69b) suggests that it has under-
gone movement in Rumanian.
Giusti (2002: 71–72) takes the position of the demonstrative in the Ro-
manian example (69b) to be derived. She postulates partial movement of
the demonstrative to a position intermediate between DP and the lower base
position of the demonstrative. She assumes that when the enclitic article is
merged in the structure , this creates a further projection. This can be seen
in (69b): in this example there is an enclitic article, -ul, on the noun tabloul
(‘the painting’). The same type of enclitic article appears on the noun
băiatul (‘the boy’) in (66b), while there is no such enclitic article in (66a).
112 Part II – The functional make up of the noun phrase

Merging of the article triggers movement of the noun. Giusti also assumes
that this in turn will necessitate the creation of a new specifier position for
the demonstrative to move to at the level of Logical Form (LF) (see (70)
below for the motivation for LF movement).41 So movement of the demon-
strative to an intermediate position in Romanian is contingent on the bound
nature of the article which triggers N-movement.
But notice now that, in contrast to (66a), the sequence in (66c), in which
the demonstrative precedes the combination noun+ article, is ungrammati-
cal.

(66) c. *acest băiatul (frumos al său)

The format in (66d) summarises the general structure which Giusti postu-
lates to derive the Romanian examples in ((66a-b), (69b), (66c)):42

(66) d. [FPmax [N+art [FP4 dem [N [FP3 AP [N [FP2dem [N [FP1 poss AP

[N]]]]]]]]]]
(Giusti 2002: 72)
where AP=descriptive adjective, possAP =possessive adjective

We will first illustrate how the structure is implemented by providing rep-

resentations for the Romanian examples just given. Then we will explore
some of Giusti’s principles underlying these representations. For (66a),
there is no article on the noun. Giusti assumes that the derivation moves to
FP4, and then moves to FPmax (see (70) for argumentation):

———–——————————
41
The discussion assumes that there are a number of different levels of representa-
tion, D-structure, S-structure and L(ogical) F(orm). These levels were typically
adopted in pre-minimalist approaches, like in the Government and Binding
framework (Chomsky 1981, 1986).
D-structure is the level at which elements are inserted. S-structure results
from various movement operations and is reflected in the overt form of the
sentence. LF is an interpretative level in which non-overt movements may have
taken place to encode semantic relations (scope, for instance). It is assumed that
any movement that can overtly take place before S-structure may also apply
covertly to generate LF-relations.
See also Part I (Introduction) section 2.1. for a discussion of levels of repre-
sentation
42
Following Giusti’s own practice we use copies to indicate moved constituents.
 Chapter 1 – The emergence and the structure of DP 113

(66) a.’ [FPmax acest [FP4 acest [băiat

[FP3 frumos [băiat [FP2acest [băiat [FP1al sau [băiat]]]]]]]]]]

For (66b), the enclitic article -ul is merged with FP4 and heads FPmax. The
clitic in Fmax attracts the noun băiat, which moves cyclically through the
heads of the functional projections (FP2, FP3 FP4). The demonstrative
acesta originates in the low functional projection, FP2, and has moved to
the specifier position of a functional projection FP4.

(66) b.’ [FPmax [băiatul [FP4acesta [băiat [FP3 frumos [băiat [FP2 acesta [băiat
[FP1 al sau [NP băiat]]]]]]]]]]

The ungrammaticality of (66c) shows that, unlike what happens in (66a),

the demonstrative does not move beyond SpecFP4 to land in SpecFPmax.
Giusti (2002) accounts for the ungrammaticality of (66c) on the grounds
that merging of the article is a last resort procedure. If SpecFPmax is occu-
pied by the article, the projection is visible and its specifier need not be
overt, therefore it must not be overt. Hence, the demonstrative cannot move
to SpecFPmax (cf. (71) below).43

(66) c.’ *[FPmax acesta [băiatul [FP4 acesta [băiat [FP3 frumos [băiat [FP2 acesta
[băiat [FP1 al sau [NP băiat]]]]]]]]]]

In order to clarify why demonstratives should move in languages like Ro-

manian and why in some languages they cannot co-occur with the definite
article (Greek), Giusti adopts the following assumptions (for more on this,
see Giusti 2002):
———–——————————
43
Another instance of a last resort process in Romanian concerns the pseudo-
demonstrative cel, which is merged in the structure when the adjective fails to
check the phi-features on D (Campos 2005 for details).
In Romanian there are two words that mean ‘first’: întâi-ul and dintâi. Notice
that only the first one bears definite inflection. The use of the second one re-
quires insertion of cel:
(i) a. inta-ul text b. *dintâi text c. cel dintâi text
first-the text the first text
Since numerals are adjectives merged high in the nominal structure, the noun
does not have the choice of moving past them (i.e. the order is always Num-N).
Thus the last resort Spell-Out of the determiner that Giusti proposes does not
work here and the pseudo-article (the anaphoric-cataphoric demonstrative) plays
the role of the adjective.
114 Part II – The functional make up of the noun phrase

(70) 1. D has a referential feature. The referential feature on D needs to be

associated with an overt element; this is realized either on the head
D itself or on its specifier.
2. “The interpretation of a noun phrase at LF is done in its highest
Specifier position “ (Giusti 2002: 56).
In our examples the highest Specifier position is SpecFPmax in
(66b). SpecFPmax corresponds to DP.
3. Demonstratives, as well as other maximal projections carrying ref-
erential features, must check their referential features in SpecFPmax
/SpecDP at some level of representation (and by LF at the latest).
In (66a) the demonstrative is in the specifier of the highest pro-
jection and can check its features; in (66b) it occupies SpecFP4
but the highest specifier is SpecFPmax. The demonstrative will
therefore have to undergo movement to SpecFPmax at LF to check
its referential features.

Observe that the third condition interacts with the first in that movement of
the demonstrative to SpecDP will satisfy the condition on the overt realiza-
tion of the referential feature on D. Moreover, as we said above, referring
to Lyons (1999), the defining characteristic of demonstratives is the feature
[+DEM]. So we can understand point 3 as saying that what is checked by
raising the demonstrative are not just the referential features of the demon-
strative, but rather the feature [+DEM]. [+DEM] entails definiteness, so when
the demonstrative reaches SpecDP, the whole nominal phrase is interpreted
as definite, as expected.44

———–——————————
44
Brugè (2000) assumes that the demonstrative has a feature [REF] which must be
checked in the DP area. Depending on the strength/weakness of this feature Brugè
predicts the following tripartition of languages: if [REF] is strong the demonstra-
tive will (always) be forced to move to SpecDP (English); if it is weak, the de-
monstrative will remain in situ, i.e. in the lowest specifier position according to
Brugè and Giusti (1996) (Celtic, Hebrew); if it is either strong or weak, the de-
monstrative will either stay in situ or move to SpecDP (Greek, Romanian). This
account captures the cross-linguistic distribution of demonstratives but it also
has a number of shortcomings: first, it implies that in Greek and the languages
that pattern with it, [REF] is both weak and strong. Second, the choice between
the DP position of the demonstrative and the lower one is taken to be free. But
as Panagiotidis (2000: 726), with whom we agree, points out, the different inter-
pretation the two positions receive do not support such a freedom of choice.
 Chapter 1 – The emergence and the structure of DP 115

The question is a more general one: how can the co-occurrence of the de-
monstrative and the (free form) definite article attested in some languages
be best accounted for? One should bear in mind that the realization of
a functional head is seen as a last resort procedure (Giusti 2002, among
others), i.e. a functional head is realized only if it is absolutely necessary. In
addition, Giusti (2002: 70) assumes that a functional projection has to be
licensed.

(71) 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

Clauses (a) and (b) of (71) may operate either disjointly or conjointly de-
pending on the language and on the constituent in the specifier position. In
Greek they are conjoint and the demonstrative and the article can (in fact,
must) co-occur. In English they are disjoint, so that either the article or the
demonstrative can be realized. In the case of disjoint application of (71a)
and (71b) the result is a ‘doubly filled Comp Filter’ effect, whereas when
conjoint, application will result in doubly filled Comp languages (Giusti
2002). In other words, the complementary distribution of a demonstrative
in SpecDP and a determiner in D can be seen as parallel to the complemen-
tary distribution of a wh-phrase in SpecCP and the overt realization of the
complementizer in C.
There have been a number of different implementations of this doubly
filled comp filter effect in the nominal domain. Campbell (1996: 167) pro-
poses a th-criterion (cf. section 2.2. above), whereby «A [+TH] determiner
has a [+TH] specifier and a [+TH] operator specifies a [+TH] determiner».
For Campbell, all demonstratives are specificity/definiteness operators, so
his feature [TH] in all appearances is used as an abbreviation for these se-
mantic categories. Panagiotidis (2000: 724), following Campbell, proposes
the “Demonstrative Criterion” to the same effect.
We can now understand better why (66c) is bad: the demonstrative has
been raised to the highest spec position and the article, being a last resort
procedure, cannot be merged. If the demonstrative is in SpecDP, the head
of this category must remain empty according to clause 2 of (70) in combi-
nation with the disjoint operation of the Principle of Economy of lexical in-
sertion.
116 Part II – The functional make up of the noun phrase

4.2.4. An alternative proposal: head movement and demonstratives

In the discussion above it was assumed that the demonstrative moves from
a low position, the specifier of a functional projection between NP and DP,
to a higher specifier position, viz. SpecDP. The demonstrative moves as a
maximal projection.
Bernstein (1997) elaborates a different implementation of the hypothesis
that the demonstrative moves from a position between DP and NP to the
spec of DP. She assumes that the demonstrative moves as a head, i.e. it
raises and substitutes into the D° position. Her analysis is based on the ob-
servation that in several dialects of French Picard the simple demonstrative
functions as a definite article, a fact which suggests that it has lost its deic-
tic value:

(72) a. chèle école (Boulogne French; Bernstein 1997: 94)

this school
the school
b. che monde
this world
the earth

According to Bernstein, the demonstrative starts out as a phrasal element

and undergoes raising to D, where properties of the head in these dialects
contribute the definite interpretation.
The raising analysis also accounts for another characteristic of demon-
stratives. In English, demonstratives are ambiguous between a deictic inter-
pretation and an indefinite specific interpretation. In the latter case, the de-
monstrative can be paraphrased by the indefinite article:

(73) a. this woman (right here)

b. this woman (from Paris)
= a woman
c. There’s this book (that) you ought to read.
= a book

Bernstein suggests that the deictic interpretation (73b,c) is associated with a

demonstrative that has raised to D°.
Observe that Bernstein’s proposal could also be used to account for the
different forms of the demonstrative in the Romanian example (66) re-
peated here as (74). It could be argued that in (74b) the form acesta, which
 Chapter 1 – The emergence and the structure of DP 117

occupies a postnominal position, is the phrasal variant of the demonstrative,

while the reduced form acest in (74a) results from head movement to D.

(74) a. acest (frumos) băiat (Romanian; Giusti 2000: 71)

b. băiatul acesta (frumos)

However, Bernstein’s head movement hypothesis implies that prenominal

demonstratives and the definite article could have the same distribution,
since they both occupy D. This account faces the problem that in some lan-
guages demonstratives co-occur with determiners, suggesting that the two
do not compete for the same position, as is the case in Greek, as already
said; cf. (75):

(75) afto to vivlio

this the book

An account in terms of a doubly filled DP filter (see above) will more read-
ily account for this type of parametric variation.

4.2.5. Reinforcers as empirical evidence for a lower position

We have already discussed some evidence that though demonstratives oc-

cupy a high, left peripheral position in the DP, they probably originate lower
in the structure. The distribution of reinforcers (see section 4.1.1.) associ-
ated with demonstratives provides some independent evidence for this hy-
pothesis. In the discussion above we illustrated reinforcers in Greek (edho,
eci) and in French (ci, là). Reinforcers are also found in the Germanic lan-
guages: (76) and (77) provide some examples.

(76) a. den här mannen (Swedish)

the here man
b. den där mannen
the there man

(77) a. this here guy (non-standard English;

b. that there guy cf. that guy (over) there)
118 Part II – The functional make up of the noun phrase

In French, demonstratives are marked for proximity vs. lack thereof via the
presence of ci and là (see (54b) above):

(78) a. cette femme-ci

this woman here
b. ce livre-là
that book there

Bernstein (1997: 100) proposes the following structure for DPs containing
demonstratives and reinforcers:

(79) a. [DP [FP cette [F’ ci [NP femme ]]]]

In (79), the demonstrative ce occupies the specifier position of a functional

projection, FP, between DP and NP, the reinforcer ci occupies the head-
position of that functional projection. The N femme (‘woman’) is the head
of NP. The demonstrative moves from SpecFP to D stranding the adverbial
element ci. However, in the examples in (78) the noun (femme (‘woman’),
livre (‘book’) occurs between the demonstrative and the adverbial rein-
forcer. Bernstein argues that in this case the noun moves and left-adjoins to
FP (79b).

(79) b. [DP [FP cette [F’ ci [NP femme ]]]]

Based on the examples above the movement of the noun might be taken to
be head movement. However, the data in (80) show that in fact the relevant
movement is phrasal: in (80) we see that the noun can be modified by ad-
jectives (80a) and that it can be accompanied by its complements (80b):

(80) a. ce livre jaune ci (French)

this yellow book here
b. ce marchand de vin ci
this merchant of wine here

So the phrasal constituent (NP, or more vaguely XP, according to Bernstein)

that contains the noun and its modifiers/complement is moved and adjoined
to a position between the demonstrative and its reinforcer. Bernstein as-
sumes that the trigger of such movement is a strong feature on F, which
 Chapter 1 – The emergence and the structure of DP 119

needs to be checked.45. The resulting structure is as in (80c), where traces

indicate the base positions of the moved constituents.46

(80) c. [DP cei [FP [XP livre jaune j] [FP ti [F’ ci [XP tj]]]]

In the Germanic languages there is no phrasal NP (or XP) movement : the

noun always follows the reinforcer:

(81) a. det här stora huset (Swedish)

this here big house
b. det här ringen av guld
this here ring of gold

Bernstein assumes that the demonstrative raises to DP, as is the case in the
Romance languages:

(81) c. [DP det [FP t [F’ här [NP huset ]]]]

Observe however that in many languages the reinforcer may appear in a

position separate from that of the demonstrative. This is, for instance, the
case in West Flemish, a Dutch dialect, in which the demonstrative dienen
(‘that’) is DP initial and the reinforcer (hier, (‘here’) and doar (‘there’)) is
final.47 The fact that in many languages the demonstrative and its reinforcer
can be separated provides additional evidence for the hypothesis that within
the DP there are at least two positions related to demonstratives.

(82) a. dienen goukden ring hier (West Flemish)

that gold ring here
b. dienen foto van Marie doar
that picture of Mary there
———–——————————
45
Bernstein further assumes that XP movement of the phrase containing the noun
can be extended to Spanish and Catalan, languages which exhibit no reinforcers.
Here too a strong feature on head F triggers raising of the NP (including its
modifiers) to the left of FP, deriving thus the post-nominal position of demon-
stratives available in these languages.
46
As before, rather than using the Minimalist copy notation, we use (coindexed)
traces because they provide for a more legible representation.
47
Observe that WF dienen is compatible with both the proximal reinforcer hier
(‘here’) (82a) and with the distal one, doar (‘there’), (82b).
120 Part II – The functional make up of the noun phrase

So, from within different angles and based on different, but complemen-
tary, types of data linguists have been led to hypothesize (at least) two dif-
ferent positions for demonstratives – a higher one and a lower one. The
question now is: What is the interpretative reflex of these two distinct posi-
tions? Let us therefore return briefly to the interpretation of the demonstra-
tives in the distinct positions, an issue which will lead us to a discussion of
the Greek data illustrated in (68) above.

4.3. Interpreting the positions of the demonstrative in the DP. The case of
Greek.

Bernstein claims that prenominal demonstratives need to raise to SpecDP

due to the fact that they are interpreted deictically. Let us also recall the hy-
pothesis (originally due to Lyons) that demonstratives are intrinsically
[+DEM], [+DEM] being the feature that identifies them.
We said above (section 4.1.2) that demonstratives also have an ana-
phoric use, whereby they pick out referents from the existing discourse. Let
us consider the following Greek data in (83) which can cast some light on
this issue:

(83) Context A: Mary is at the butcher’s pointing to a pork joint that she
wants to buy.
a. Mary: Thelo afto to butaki.
want-1SG this the joint
‘I want this pork joint.’
b. Mary: ??Thelo to butaki afto.48
want-1SG the joint this
Context B: A paragraph from a guide book about a Greek town.
c. I poli eçi pola istorika ktiria pu xronologhunte apo ti vizantini epoçi.
the town has many historical buildings that date back to the Byzantine
period
d. Ta ktiria afta episceptonte kathe xrono ekatondadhes turistes.
the buildings these visit-3SG every year hundreds tourists

———–——————————
48
For some speakers (b) is acceptable with contrastive intonation on the demon-
strative and an accompanying deictic gesture.
 Chapter 1 – The emergence and the structure of DP 121

e. ??? Afta ta ktiria episceptonte kathe rono ekatondadhes

turistes.
these the buildings these visit-3SG every year hundreds
tourists49
‘These buildings are visited every year by hundreds of tourists.’

Manolessou & Panagiotidis (1999), Manolessou (2000), Panagiotidis (2000),

and Grohmann & Panagiotidis (2005) observe that in Greek the pre-article
position of the demonstrative afto entails greater deictic strength, in contrast
with the post-nominal position in which the demonstrative is used as dis-
course anaphoric, namely to refer back to an entity that has been previously
mentioned. Only the pre-article demonstrative can normally be used along
with a pointing gesture: based on this observation it is generally agreed that
only the pre-article demonstrative is genuinely deictic.50 The post-nominal
demonstrative, unlike the prenominal one, cannot be used to contrast two
entities denoted by the noun. To support this, Manolessou (2000: 16–19)
draws on corpus material. She shows that in spoken material, including
plays, the post-nominal use of the demonstrative is statistically very low,
while the pre-article use gets a very high percentage. On the contrary, the
pre-article use is very rare in written language, scientific-theoretical com-
position included, in which the post-nominal use is by far preferred. In par-
ticular, in scientific texts we find 96,47% of demonstratives in post-
nominal position and only 3.53% of demonstratives are prenominal. Con-
versely, in plays only 5.89 % of demonstratives are post-nominal and 94.1%
are prenominal (Manolessou 2000: 17).
Let us further again point to the immediately relevant fact that, although
(83c–d) display the usual first-second mention distinction concerning the
indefinite/definite article use, (83d) involves the demonstrative because the
referent of the DP has already been introduced by the indefinite DP in

———–——————————
49
Notice that the ‘strategy’ discussed in the preceding note is not available for res-
cuing (83e) since the buildings in question are not physically present, and so one
cannot point to them (unlike the butcher’s customer who actually sees the joint).
50
Interestingly, a different proposal with respect to the interpretation of post-nomi-
nal and pre-article demonstratives is made by Tasmowski De Ryck (1990), who
argues that the pre-article demonstrative has a thematic interpretation (i.e. it rep-
resents an entity already known/given) while the post-nominal demonstrative has
a rhematic (i.e. new) interpretation.
122 Part II – The functional make up of the noun phrase

(83c). (83d) would be entirely infelicitous (signaled by #) without the de-

monstrative – cf. (83d’).

(83) d.’ #Ta ktiria episceptonte kathe xrono ekatondadhes turistes.

(Greek)
the buildings visit-3SG every year hundreds tourists

This clearly shows that despite the fact that both the definite article and the
demonstrative are definite and referential, the demonstrative still contrib-
utes something ‘extra’: in the spirit of Lyons (1999:21) we can say that the
demonstrative signals that the reference of the noun involved is immedi-
ately accessible to the hearer/reader. Rather than the speaker pointing to the
referent (remember that (83c–d) represents a written text), the speaker ex-
ploits the postnominal position of the demonstrative to immediately relate
the denotatum to the intra-linguistic context.
Along these lines, consider further some differences between the two
positions or uses of the demonstrative, already shown in (68), repeated here
as (84):

(84) a. afto to vivlio

this the book
b. to vivlio afto
the book this

First, only in the the pre-article position can the demonstrative be stressed
emphatically; emphatic stress is not possible when the demonstrative occurs
post-nominally (but see note 48):

(85) a. afto to vivlio

b. *?to vivlio afto

Second, the post-nominal demonstrative cannot be used independently.

(86) Context: Pjo vivlio aghorases?

Which book did you buy?
a. Afto
this
‘This one.’
b. *to Ø afto
 Chapter 1 – The emergence and the structure of DP 123

To understand a third difference between the post-nominal and the pre-

article position we must at this point introduce some data which illustrate
another position of the demonstrative, namely the ‘post-adjectival’ position.
Consider (87):51

(87) to oreo afto forema

the nice this dress
‘this nice dress’

In (87) the demonstrative appears to be sandwiched between the prenomi-

nal adjective oreo (‘nice’) and the noun forema (‘dress’). It is important to
notice with respect to (87) that it is not the case that the demonstrative sim-
ply occupies its original low position above the noun, because if the adjec-
tive is not present, the result is ungrammatical:

(88) *to afto forema

the nice dress

Taken together (86b) and (88) suggest that the demonstrative found be-
tween the noun and a prenominal adjective is dependent on the adjective, in
the same way that it is dependent on the noun when it follows it (68b/84b),
while the pre-article position allows for an independent use.52
Now, just like post-nominal afto in (84b), post-adjectival afto in (87)
cannot be emphatically stressed:

(89) ??to oreo afto forema

the nice this dress

(85b) and (89) jointly suggest that the demonstrative that appears following
the adjective is also in some lower position. We will come back to this im-
mediately below.
Still another difference between the pre- and the post-nominal position
of the demonstrative in Greek is the fact that the adverbial reinforcers edho
(‘here’) and eci (‘there’) are compatible with the prenominal position of the
———–——————————
51
See Part III, Chapter 1 for discussion of the exclusively prenominal position of
adjectives in Greek.
52
See Stavrou & Horrocks (1989), Manolessou & Panagiotidis (1998), Manolessou
(2000), Panagiotidis (2000), Grohmann & Panagiotidis (2005), among others, for
discussion and different accounts.
124 Part II – The functional make up of the noun phrase

demonstrative but they cannot easily be associated with a demonstrative in

post-nominal or post-adjectival position (see Campos & Stavrou 2004 for
discussion):

(90) a. afto edho to vivlio

this here the book
‘this book here’
b. ??to vivlio afto edho
the book this here
c. *?to oreo afto edho vivlio
the nice this here book

In the light of the data in (84) down to (90) then, and in line with the discus-
sion in Manolessou & Panagiotidis (1999), Manolessou (2000), Panagiotidis
(2000), and Grohmann & Panagiotidis (2005), we conclude that a demon-
strative which appears both post-nominally and post-adjectivally lacks deic-
tic force and is anaphoric.53 In addition, the post-nominal/post-adjectival
demonstrative, being syntactically dependent, behaves much like a weak
pronoun along the lines of Cardinaletti’s (1998) and Cardinaletti & Starke’s
(1999) proposal.54
The crucial question now is how the three positions, the pre-article posi-
tion, the post-nominal position and the post-adjectival position, of the de-
monstrative in Greek are related. In a fairly obvious way, it can be assumed
that the pre-article position is derived by raising of the demonstrative from
its lower position (whether this be SpecXP or SpecNP). This is what has
been assumed for Spanish and for Romanian (see above). The trigger for
such a movement is the need for the demonstrative to check a strong deictic
feature in the DP domain.
Accounting for the post-nominal and the post-adjectival positions is
more intricate. For these two positions, the explanation has often been con-
tingent on N-raising. In order to derive the post-nominal position of the
demonstrative, researchers often appeal to N-movement; the noun rises to
the immediately preceding functional head (see Part III, Chapter 3 and also
Chapter 1), stranding the demonstrative in its original position.

———–——————————
53
See also Panagiotidis (2000).
54
In Cardinaletti’s tripartition of possessive pronouns, the weak or deficient ones
are prenominal whereas the strong ones are post-nominal. The question we turn
to in detail in section 4.4 is how demonstratives fit into this typology.
 Chapter 1 – The emergence and the structure of DP 125

In particular, Manolessou & Panagiotidis (1999), Manolessou (2000)

and Panagiotidis (2000) assume that the nominal head of NP moves to the
head of a functional projection (NumberPhrase) to check its strong mor-
phology (see Bernstein 1993 for an early proposal of N-movement). The
authors also assume that demonstratives are generated in SpecNP. Phrasal
movement of the demonstrative to SpecDP derives the prenominal position
of the demonstrative (84a); the demonstrative is assumed to be attracted
there by a strong [+DEIC] feature on the definite article itself (recall that the
definite article necessarily accompanies the demonstrative in Greek).55
The post-nominal position of the demonstrative (91) is obtained if the
noun moves leaving behind the demonstrative at SpecNP:

(91) to forema afto

the dress this
‘this dress’

Though N-movement may be a tool to derive the right orders in the DP, we
will see in Chapter 3 and in Chapter 1 of Part III that it raises a number of
problems. In the next section we would like to pursue a different account for
the orders demonstrative > noun and noun (or adjective) > demonstrative.56
To account for DPs with a demonstrative after a prenominal adjective,
as in (92), Panagiotidis (2000), assumes that, when present, a prenominal
adjective is able to check the relevant phi-features of NumberP (generated
between DP and NP), thus making the movement of the noun in front of the
demonstrative not necessary – and on economy principles undesired. In this
way (92) is generated.

———–——————————
55
Cf.: “…the locus of deixis, a [DEIC] feature, is always a D.” (Panagiotidis 2000:
736). This universality of the nature of the deixis, Panagiotidis further claims,
does away with any ‘doubly-filled D-filter’, which only has descriptive value but
no explanatory power. [DEIC] can be seen as the same as Lyons’ [DEM] feature.
56
We further refer the reader to Kolliakou (1995, 1997, and references therein) for
a particular implementation in the HPSG framework of the post-nominal and
post-adjectival positions of the demonstrative in Greek. In addition, Stavrou &
Horrocks (1989) elaborate an alternative way to capture these positions, adopting
a Parallel Morphology (Borer 1993) type of model, whereby syntax and mor-
phology interact at all levels of the derivation. Demonstratives are thus treated
as a kind of phrasal affix attached onto the noun or the adjective on the way be-
tween D-structure and S-Structure, creating a morphologically complex N.
126 Part II – The functional make up of the noun phrase

(92) to oreo afto forema

the nice this dress
‘this nice dress’

In more recent work, Grohmann & Panagiotidis (2005) revise the analysis
of (92) along the following lines. The discourse anaphoric demonstrative is
merged (as before) in the Agreement Domain (abbreviated as AD in (93)57).
An empty operator along the lines of that proposed by Campbell (1996) oc-
cupies the specifier of the Discourse Domain (DP)58 and forms a chain with
the demonstrative which remains in situ. This is represented in (93a)
(slightly simplified from Grohmann & Panagiotidis 2005: 12).
The high domain (DP) is related to a strong-deictic interpretation and
the low position (AD), where the demonstrative is merged, is related to a
weak-anaphoric interpretation. Accordingly, afto is expected to have an
anaphoric interpretation in (92). When it moves upwards, towards DP, the
demonstrative crosses the position of the article, a position immediately
above the demonstrative. However, according to Grohmann’s (2003: 26)
Anti-locality Hypothesis, movement must not be too local; it must cross a
minimum distance in order to be well-formed. Grohmann & Panagiotidis
argue that the last step in the movement of the demonstrative to the D do-
main is too local (see Grohmann 2003, and Grohmann & Panagiotidis 2005
for details). Because of the antilocality condition, we need to appeal to a
rescue strategy: Copy Spell Out (Grohmann 2003), by which the violating
copy is spelled out in the form of the definite article, which agrees in all phi-
features with the demonstrative. The relevant derivation is represented in
(93a) (For expository reasons we have simplified Grohmann & Panagiotidis’
(2005: 12) analysis. The interested reader is referred to their work for details.

(93) a. [DP….afta…[ afta → ta [AD nea [ afta [NP phenomena]]]]]

these the new these phenomena

Grohmann & Panagiotidis assume that in the case in which the demonstra-
tive remains in a lower position, the empty operator also moves from the
projection of the determiner (see Grohmann & Panagiotidis 2005: 9) and
thus it likewise makes too local a move, giving rise to Copy Spell Out, as
shown in (93b):

———–——————————
57
Grohmann & Panagiotidis actually use the label ΦΔ.
58
Grohmann & Panagiotidis use the label ΩΔ.
 Chapter 1 – The emergence and the structure of DP 127

(93) b. [DP….OP…[ OP ta [AD nea [ afta [NP phenomena]]]]]

According to this analysis, the article in examples such as (92) is intro-

duced during the derivation by the rescuing strategy Copy Spell Out
(Grohmann & Panagiotidis 2005). The article is not independently merged
in the numeration; it is simply a grammatical formative inserted in the
structure at PF. It is the spelled out Copy of the anti-locally moved demon-
strative, which happens to be homophonous with the definite article. In
other words, it is neither the definite article, nor a generic article, nor a de-
monstrative article as in Panagiotidis (2000). According to Grohmann &
Panagiotidis (2005: 12ff) the article that co-occurs obligatorily with the
demonstrative is not a ‘real’ article but a grammatical formative59 – and this
is true of all the languages in which the article co-occurs with the demon-
strative. This proposal makes explicit claims concerning a more finely ar-
ticulated DP structure overall, an important issue to which we turn in sec-
tion 5.3 below.

4.4. Splitting the DP

The goal of this section is to offer an account of the fact that in some lan-
guages demonstratives may both precede or follow the head noun. In par-
ticular we want to try to derive post-nominal demonstrative positions with-
out having recourse to N-movement. The account is tentative.
Anticipating the discussion in section 5.3. below, let us assume that DP
is not a unitary projection but that it can be analyzed into an articulated array
of projections. This proposal is in line with analogous proposals concerning
the nature of the CP layer (the so-called Split CP hypothesis, see Rizzi’s
(1997), which we return to in section 5.3.) A number of authors (Ihsane &
Puskás 2001; Aboh 2002, 2004a,b; Grohmann & Panagiotidis 2005; Haege-
man 2004; Laenzlinger 2005) analogously propose that DP itself be split
into a number of functional projections. In his work on the structure of CP,
Rizzi proposes that the head C be decomposed into two heads, labelled
‘Force’ and ‘Fin’. Analogously, let us assume that there are two DP layers
internal to the noun phrase as shown in (94a) below.

———–——————————
59
Notice however that a distinction between what is called ‘real’ article and gram-
matical formative is not made clear by the authors.
128 Part II – The functional make up of the noun phrase

(94) a. DP1 (deixis)

FP1

DP2 (determination)

FP2

In addition to Force and Fin, Rizzi proposes that the CP field may also con-
tain TopP and FocP. These are functional projections which are related to
the informational status of fronted constituents. Below we propose that
there is a TopicP between FP1 and DP2.

(94) b. DP1 (deixis)

FP1

TopP

DP2 (determination)

FP2

The relevance of this assumption will become clear presently. In section

5.3. below we briefly come back to the possibility that there may be such
projections in the DP too.
The highest DP layer, DP1, is the locus of that part of interpretation of
the nominal projection that encodes discourse/pragmatic aspects of its inter-
pretation; for example, it may encode concepts such as familiarity, referen-
tiality and deixis. The lower DP, DP2, expresses determination, i.e. defi-
niteness, indefiniteness and so on. For the similarity between the lower DP
and FinP see also Haegeman (2004).60 We continue to assume, in the spirit
of Bernstein (1997), Giusti (2002) and Brugè (2002), that the demonstrative
is generated as the specifier of a lower functional category in DP2.
———–——————————
60
Cf.: “…D is decomposed into separate functional heads. The position that en-
codes (in)definiteness in the D domain and in which the definite article is merge,
is parallel to Fin in the C domain. In the same way that finiteness ‘delimits/an-
chors’ the event in time, (in)definiteness ‘delimits/anchors’ nominal reference in
space.” (Haegeman 2004: 235).
 Chapter 1 – The emergence and the structure of DP 129

As discussed above, Greek prenominal demonstratives are interpreted as

deictic. Post nominal or post-adjectival demonstratives are not interpreted
deictically, rather they are interpreted anaphorically.
In the simple cases, when no reinforcer is present, we take the demon-
strative to originate in SpecFP2 in (94). The assumption is that it has to
move to a higher layer. If it is deictic, it moves to SpecDP1 via SpecDP2
and spec FP1, where it checks its strong deictic feature – Lyons’ feature
[DEM]. This way the pattern with an initial demonstrative, demonstrative
>DP, is generated.
The order DP>demonstrative is derived by moving the demonstrative
only as far as FP161 and by fronting the whole of DP2 to SpecDP1. This en-
sures that the higher DP level is overtly realized and hence DP1 is visible at
LF (see above (70) and (71)).

(95) [DP1[DP2 to praghma] [FP1 afto [DP2 t]]]

Note, however, that this derivation hinges on the prior assumption that the
definite article to originates inside the lower DP (DP2 in (94)). In the ab-
sence of DP2 to DP1 movement, the article itself will move from D2 to D1
to make the projection DP1 visible.
Recall from the discussion and data in sections 4.1.2 and 4.3 that the
post-nominal demonstrative signals that the identity of the referent of the
DP is given in the close context. How can this ‘givenness’ component be
represented in the structure? In his initial elaboration of the CP structure
Rizzi (1997) also postulates that the CP layer contains, among other things,
a TopicPhrase, which hosts topics, i.e. constituents that are accessible in the
context. This proposal has been implemented in the analysis of the DP struc-
ture. Assuming that there is a TopicPhrase between DP1 and DP2 (see sec-
tion 5.3 for more details), we take DP2 to pass through SpecTop on its way
to SpecDP1, checking the feature [+TOP]. This additional move captures
the intuition that when the DP precedes the demonstrative, it has a topical/
anaphoric interpretation. (95) illustrates these assumptions, without includ-
ing the presence of TopP, to which we return in more detail in section 5.3.

———–——————————
61
Perhaps it is a weak pronoun that cannot remain in the low position. Recall from
section 4.3. that the postnominal demonstrative is somehow syntactically de-
pendent (cf. Cardinaletti’s (1998) and Cardinaletti & Starke’s (1999) proposal
concerning the typology of pronouns).
130 Part II – The functional make up of the noun phrase

When a reinforcer is present as the head of FP1, the order Demonstrative

>DP-reinforcer is derived by fronting DP2 to a position preceding the rein-
forcer as shown in (96). On the other hand, to derive the order Dem-
reinforcer-DP we assume D2 is not fronted (97):

(96) [DP1 afto [ DP2 to pragma] [FP1 edho [ t ]]]

(97) [DP1 afto [FP1 edho [DP2 to [NP praghma]]]]

Several proposals deriving the post-nominal position of the demonstrative

rely on N-movement. In view of the discussion in Chapter 1 of Part III, in
which we will cast doubt on the use of the mechanism of N-movement as a
way of deriving the word order of the nominal projection, we have sketched
a way of deriving the post-nominal (anaphoric) and the pre-article (deictic)
position of the demonstrative in Greek without invoking N-movement. In
particular we have sketched a split approach to DP, along the lines of Rizzi
(1997).
Concluding this section, the various word order patterns observed in
various languages involving a demonstrative, a noun and possibly also an
adjective, have been tentatively accounted for by the interplay of a number of
different types of movement such as raising of the demonstrative to SpecDP,
possibly combined with head movement of N to higher head positions.62
We have outlined a possible account to derive the word orders in Modern
Greek. Obviously, more research will be needed for a fully fledged syntactic
account of the various positions and the related interpretations of demon-
stratives in various languages. We hope to have put a base here for such an
account.
In the next section we return to some of the basic parallelisms between
DP and the clause. We return to additional implementations of the split CP
in section 5.3.

5. DP and CP

In this final section of the chapter we return to the general issue of the func-
tional domain of the nominal projection and in particular to the parallelisms
(if any) between the DP layer and the CP layer of the clause. It is assumed
that DP is the extended projection of N. In the original proposals a nominal
———–——————————
62
In Romanian this movement may be as high as D.
 Chapter 1 – The emergence and the structure of DP 131

projection was taken to be selected by a functional head D, occupied by the

determiner. Szabolcsi (1983, 1994), Abney (1987), Horrocks & Stavrou
(1987), Loebel (1989), Stowell (1989), Longobardi (1994) and many others
have emphasised a number of parallelisms between DP and the clause. In
this section we will survey some of the ways in which DP has been seen as
parallel to CP.
In Chapter 3 we will consider further parallelisms between inflectional
categories in the nominal and the clausal domains. Comparing clauses as
extended projections of V and nominal projections as extended projections
of N, the question arises whether we should compare DP to a functional
projection of the IP-type (AGRP, TP), or rather to a functional projection of
the CP-type.

5.1. DP as parallel to CP

There is compelling evidence that DP is a CP-type projection. We review

some of this evidence here. We have already presented arguments in sup-
port of the hypothesis that D is a subordinator turning NP into an argument.
In view of the fact that only DPs and CPs can function as arguments, we
can say that D is like C in that both turn their complements into arguments.
In addition, the interpretative role of D and C is similar: both D and C head
projections that link their complements with the discourse or non-linguistic
context.
In the clause, the CP layer constitutes the interface between the proposi-
tion and the domain of language use. The projections associated with CP
serve to link a proposition (expressed by IP) with the discourse and specify
the nature of this link, whether the proposition is questioned or affirmed,
for instance; in other words CP serves to convert a proposition into a par-
ticular speech act. In the same way, it has been argued that DP links the
content of the nominal projection (a predicative category) to the universe of
discourse indicating whether the referent is already contextually available
(definite D) or is novel in the discourse (indefinite D). “The function of the
determiner is to specify the reference of a noun phrase. The noun provides
a predicate, and the determiner picks out a particular number of that predi-
cate’s extention” (Abney 1987: 77).
I addition to this semantic argument, there is also distributional evidence
making DP parallel to CP. We have already cited evidence from Greek
based on work by Horrocks & Stavrou (1987: 86). The relevant data show
that phrasal movement operates in a parallel way in the clause and the noun
132 Part II – The functional make up of the noun phrase

phrase. In particular, constituents that are focused can be fronted to SpecCP

and SpecDP. Horrocks & Stavrou (1987) show that focalization in the clause
has the properties of A’-movement (see also Introduction).
Moreover, the parallelism between interrogative clauses and interroga-
tive DPs in Greek strengthens the assumption that DP is to NP what CP is
to VP, as shown in examples (32)–(33) repeated below for the reader’s con-
venience as (98)–(100):

(98) a. Ekane ti?

did-3SG what
‘He did what?’
b. Ti ekane?
what did-3SG
‘What did he do?’

(99) a. to vivlio tinos?

the book who-GEN
b. tinos to vivlio?
whose the book
‘whose book?’

(100) a. Mu ipes [CP pos dhjavases [DP to vivlio tinos]]?

me-GEN said-2SG that read-2SG the book who-GEN
‘You told me you read whose book?’
b. Mu ipes [CP pos dhjavases [tinos to vivlio t]]?
c. [CP [Tinos ] mu ipes [CP t pos dhjavases [t to vivlio t]]?
d. [CP [To vivlio tinos]mu ipes [CP t pos dhjavases t]?
e. [CP [Tinos to vivlio t] mu ipes [CP t pos dhjavases [t ]? 63

Horrocks & Stavrou claim that whether DP corresponds to CP or to IP var-

ies cross-linguistically. Thus the Greek DP corresponds to the IP layer in
the English clause but to the CP layer in the Greek clause. This claim is
———–——————————
63
We note that Horrocks & Stavrou’s argument is reinforced if a demonstrative is
included in (100c). If the demonstrative is in [SpecDP], it ought to block extrac-
tion of tinos (‘whose’). Indeed (i) below is ungrammatical. We thank Hector
Campos for bringing this fact to our attention.
(i) *Tinos mou ipes pos dhjavases [afto to vivlio]?
whose you told me that read this book
 Chapter 1 – The emergence and the structure of DP 133

based on the observation that there is an asymmetry between Greek and

English DPs with respect to the structural position of the subject. The Greek
DP lacks a structural (prenominal) subject position, whereas English has
one (see for details Part IV, Chapter 1). Consider the following examples:

(101) a. the enemy’s destruction of the city

b. *tis Marias apoplanisi t apo to Jani (Greek)
the-GEN Mary-GEN seduction by the John

In English there are two positions to which the theta roles of a (deverbal)
noun can be assigned; one seems to correspond to the canonical position of
determiners (the enemy’s, cf. his/the destruction), and the other is the post-
nominal/complement-of-N position regularly occupied by an of-phrase (of
the city).
In Greek by contrast, “there is no genitive NP position distributionally
equivalent to the determiner slot in NPs. (…), ANY pre-head genitive,
including interrogatives, must precede the article and not ‘replace’ it.”
(Horrocks & Stavrou 1987: 93–94). (101b) becomes grammatical if the ar-
ticle is put into place:

(101) b.’ tis Marias i apoplanisi (apo to Jani)

the-GEN Mary-GEN the seduction (by the John)
‘Mary’s seduction by John.’

The genitive tis Marias in (101b’) precedes the article, which suggests that
the genitive phrase is found in specDP, the head D being realized by the ar-
ticle. Notice further that a second genitive noun phrase cannot appear in the
same DP:

(102) a. *tis Marias i apoplanisi tu Jani

the-GEN Mary-GEN the seduction the John-GEN
b. *i apoplanisi tis Marias tu Jani
the seduction the-GEN Mary-GEN the John-GEN

The only way to express the agent role in a case like that in (102) – the role
encoded in the determiner-like possessive DP in English – is via an adjunct
by-phrase (apo to Jani, ‘by John’), as in (101b’).
The syntax of possessors and genitives will be dealt with in detail in
Chapter 1 of Part IV. Here it need only be emphasized that pre-article geni-
tives in Greek are necessarily focused and we assume that they have been
134 Part II – The functional make up of the noun phrase

moved in that position from the complement-of-N position. So the pre-

article genitive in (102) is on a par with the pre-article interrogative in
(31b) and (32e), as well as with the pre-article demonstratives in (68a). In
view of the relevant facts, Horrocks & Stavrou (1987) conclude that the
specifier of the Greek DP is not an argument position (i.e. it cannot encode
the external argument of the noun phrase).64 Because focus is not required
for the English genitive, Horrocks & Stavrou (1987) propose that the pre-
nominal genitive in English is an argument.
We mentioned above that in English there is evidence for the existence
of a specifier position in the DP, parallel to that of the clause (CP). The ex-
amples in (37) were given to illustrate that assumption. Here we will supply
more evidence. Just as the English CP domain may host non-interrogative
constituents introduced by so, the specifier of DP may also host non-
interrogative constituents introduced by so. In (103a) the DP so vivid a pic-
ture has been moved to SpecCP. This DP itself exhibits DP-internal left-
ward movement of the AP so vivid. The landing-site of this movement is to
the immediate left of D, i.e., by assumption, SpecDP. In (103b), such
moves to a DP-internal landing-site to the immediate left of D, SpecDP.
Note that unlike so, the degree modifier such in (103b) fronts independ-
ently of the adjective important, which it modifies:

(103) a. [CP [DP [AP So vivid] a picture] does [IP this program draw of the sit-
uation of these animals that the reader wants to react immediately]].
b. [IP The article had [DP [AP such] an important impact] that the pro-
posal had to be withdrawn.

(104a) and (104b) are additional examples of DP-internal leftward move-

ment of an AP:

(104) a. This is [DP [AP too easy] a conclusion].

b. I did not expect [DP [AP that big] a turnout].

Furthermore, data from Hungarian reinforce the view of DP as being aligned

to CP. As shown by Szabolcsi (1983, 1987) Hungarian offers further con-
firmation for making the D node parallel to C. The Hungarian pre-nominal
———–——————————
64
They draw a number of additional conclusions with respect to SpecDP in Eng-
lish. However, these are framed in an older version of our theoretical model and
would have to be reformulated and updated to be properly evaluated within the
current framework.
 Chapter 1 – The emergence and the structure of DP 135

possessor may occupy two positions. It may follow the determiner, in which
case it has nominative case (105a), or it may precede the determiner, in
which case it has dative case, as shown in (105b).

(105) a. a Mari kalap-ja

the Mari-NOMINATIVE hat-3SG
b. Mari-nak a kalap-ja
Mari-DATIVE the hat-3SG
‘Mari’s hat’

It seems reasonable to propose that Marinak in (105b) occupies the specifier

position of DP. 65 In fact, (105b) is the exact parallel of the Greek pattern in
(101b’). The relevant structure is under (106):

(106) DP

Spec D’

Marinak D NP

a kalapja

Data from topicalization and wh-movement in Hungarian suggest that

SpecDP is also an escape hatch for A’-movement, just as it has been
claimed for Greek by Horrocks & Stavrou. In (107a), the dative possessor
Marinak is topicalized independently of the noun with which it is con-
strued. Topicalization of the nominative possessor is not possible (107b). In
(108), similarly, the interrogative dative possessor has undergone wh-move-
ment independently of the noun with which it is associated:

(107) a. [CP [TopP Marinaki [FocP PETER làtta [IP [DP ti a kalapja]]]]].
Mari-DATIVE Peter saw the hat
‘Peter saw Mary’s hat.’
b. *[CP [TopP Marii [FocP PETER làtta [IP [DP a ti kalapja]]]].

———–——————————
65
But see also Knittel (1998) and Den Dikken (1999) for claims that Hungarian
dative possessors involve a left dislocation configuration.
136 Part II – The functional make up of the noun phrase

(108) [CP [FocP Kineki làtta [IP Kati [DP ti a kalapja]]]?

whose-DATIVE saw Kati the hat
‘Whose hat did Kati see?’

Szabolcsi (1987) proposes (109) to represent the structure of the phrase a

Peter minden kalapja ‘Peter’s every hat’. CN is the nominal counterpart of
CP and IN is the nominal counterpart of I(nfl) in clauses.

(109) CN’’

CN’

CN IN’’

(az) CN’’ IN’

a Peter Art IN

minden N’ IN
[+POSS, 3SG]
N

kalap -ja

(109) suggests a strict parallelism between DP and clausal structure: in the

nominal domain the definite article occupies the position that is occupied
by the complementizer in the clause. In the same way that the clausal sub-
ject occupies the specifier of IP, the possessor DP occupies the specifier of
the nominal inflectional projection (IN). The head of IP is a functional ele-
ment that enters in an agreement relation with the head noun. We may add
at this point that Szabolcsi (1994 in particular), establishes two categories
of determiners; D and Det. D is represented by the article(s) (for instance
az) and is hosted by D in (109), while Det stands for all other determiners,
like minden (‘every’), melyik (‘which’), kevés (‘few’) and others. These de-
terminers are found under the head Art in (109). The reader is referred to
Szabolcsi’s work (1994: 212–213) for details concerning the different se-
mantic functions of these two types of determiners.
In chapter 3 we will explore to what extent we can find evidence for pos-
tulating additional functional projections in the nominal extended projection
which can be paralleled with corresponding projections in the clause.
 Chapter 1 – The emergence and the structure of DP 137

5.2. DP as a VP-like category: DP-shells

Larson (1996) elaborates an interesting and novel approach to the expansion

of DP based on the observation that D, like V, possesses a thematic struc-
ture. We have mentioned repeatedly that the definite article is a subordinator
in that it takes the lexical NP as its complement. Put differently, bare lexical
projections, NPs, cannot act as arguments of verbs or prepositions – they
have to be preceded by a constituent belonging to the general class of de-
terminers. Thus, articles, as well as other determiners, can be thought of as
being ‘transitive’. Crucially, in this line of thought, certain determiners, the
definite article among them, cannot stand by themselves, they necessarily
must take an ‘object’. Consider the following:

(110) a. *the/most
b. the cat/most cats

The notion ‘transitive determiner’ means that a certain determiner can ex-
press a relationship between predicates. Thus, plural some expresses a rela-
tion between the predicate X (=whale) and Y (=mammal) that corresponds
to the non-empty intersection relation as illustrated by (111) (all the exam-
ples are from Larson 1996):

(111) a. SOME (X,Y) iff |Y ∩ X| > 0

b. |{x; x is a whale} ∩ {x: x is a mammal}| > 0

Putting this in more simple terms, the intersection of ‘whales’ and ‘mam-
mals’ is bigger than zero, i.e. a non empty set. (112) gives the same idea for
THE:

(112) THE (X,Y) iff |Y — X| = 0, where |Y| = 1

(112) is thus another way of expressing uniqueness as one semantic com-

ponent of definiteness as we saw in 2.1 above. THE signals the unique Y.
If certain determiners amount to transitive predicates possessing argu-
ment structure, «projection of DPs can be viewed analogously to the projec-
tion of VPs.» (Larson 1996: 147). In a strictly compositional fashion, i.e. in a
way that the syntax of DP reflects one by one its semantic properties, Larson
elaborates an analysis whereby the specifier of DP is a subject position.
The subject of DP is provided by the clause which the DP is a sister of in
138 Part II – The functional make up of the noun phrase

Logical Form.66 For example, for the sentence All whales are mammals,
Larson gives the following structure:

(113) S

DP1 S

XP D’ DP1 VP

D NP t V DP

Pro all whales are mammals

In particular, the DP subject is the pro-form Pro (cf. also Campbell (1996),
who claims that the specifier of DP hosts an operator, which may be empty
or may be realized as the demonstrative depending on the language). Larson
claims that it denotes a variable that ranges over sets – the value of the vari-
able (x) is given by the clause ‘are mammals’. Larson further takes this pro
subject to be always uniformly selected as the highest argument of a DP.
In this line of thought, all whales represents a ‘transitive’ structure. Ac-
cordingly, the English pronoun He entails an intransitive structure like that
in (114):

(114) DP

XP D

Pro he

A further parallelism is drawn by Larson concerning the structure of the DP

containing a relational noun and that of a VP with two arguments. Larson
(1988) argues for the following structure for a ditransitive verb like give:

———–——————————
66
LF is an interpretative level. See note 41.
 Chapter 1 – The emergence and the structure of DP 139

(115) VP

DP V’

John V VP

give DP V’

e V’ DP

V DP that

t Mary

VP expands in shells – each V has as its sister node another VP and so on.
Larson (1996) gives the following structure for DPs, where the category DP
also expands in a shell-like fashion:

(116) DP

Spec D’

D DP

THE DP D’

e D’ NP

D NP mother

t John

(116) is the structure underlying the noun phrase John’s mother. Details
aside, we see that here too the DP creates a shell structure, whereby every D
head has as its sister node another DP, which in turn also contains another
140 Part II – The functional make up of the noun phrase

DP, etc. We observe then that for Larson DP is more strictly paralleled with
the lexical VP in terms of internal structuring.
In Chapter 1 (section 5.2) of Part III we will have more to say about
Larson’s structure (116) on the basis of facts concerning adjective ordering
and distribution DP internally. This will be particularly relevant in relation
to the opposition individual/stage-level adjectives.

5.3. Topic Phrase and Focus Phrase in the DP: some proposals

In sections 5.1 and 5.2 we reviewed ways in which the category DP is

aligned to CP. We said that the parallelism of D with C lies primarily in the
fact that both categories turn a predicate into an argument. They further en-
code functions that can be characterized as discourse-oriented, namely ref-
erentiality, identifiability, deixis and the like.
For the clausal domain, Rizzi (1997) has argued in favor of splitting up
the CP into a number of separate projections. If DP is analogous to CP, then
we might expect that in the nominal domain too, DP will be reanalyzed in
terms of articulated projections. By analogy with Rizzi’s proposal to decom-
pose CP into an articulated structure, there have indeed been a number of
proposals to decompose the functional domain of DP into a more articulated
structure. We have already referred to such proposals in sections 4.3 and
4.4. We review similar proposals here. A number of proposals have been
elaborated and are being elaborated and we will not be able to go over all of
these in detail. Rather we will provide a brief survey of Rizzi’s original
proposals for the split CP and then discuss the kind of data that have been
used to support the articulated structure of the DP.
With respect to CP Rizzi says:

We can think of the complementizer system as the interface between a pro-

positional content (expressed by the IP) and the superordinate structure (a
higher clause or, possible, the articulation of discourse, if we consider a root
clause). As such, we expect the CP system to express at least two kinds of
information, one facing the outside and the other facing the inside.
(Rizzi 1997: 203)

The outside layer of the CP encodes information concerning the type of the
clause (declarative, question, exclamatory, etc.) – this is what Rizzi, follow-
ing Chomsky (1995), calls ‘Force’. FinP relates to the content of (a possi-
bly articulated) IP, which is embedded under C. In section 4.4 above we
 Chapter 1 – The emergence and the structure of DP 141

noted that by analogy with Rizzi’s ForceP and FinP in the clausal domain,
some authors have proposed that DP can be split into a higher DP and a
lower DP.
In addition, as anticipated in section 4.4, Rizzi (1997) proposes that the
C layer includes information that goes beyond the mere encoding the selec-
tional relations with the neighboring systems. The left periphery of the
clause encodes the traditional discourse-informational concepts of topic and
focus. The finer structure that Rizzi proposes (1997: 291) as an articulation
of the ‘upper’ layers of CP is given below:67

(117) ForceP

Force TopP*

Top FocP

Foc TopP*

Top FinP

Fin IP

Topic and Focus are concepts related to the information packaging: they are
the syntactic reflex of the topic-comment and the focus-presupposition (or
background) contrasts (Jackendoff 1972; Chomsky 1972; Lyons 1977).
Aboh (2002, 2004a, 2004b) proposes the following format as a schematic
way of mapping these traditional discourse-linked notions to the structure
of the clause:

(118) a. [CP Topic [IP comment]]

b. [CP Focus [IP presupposition/(shared) background]]

Notice that in the structure proposed by Rizzi, Topic is a recursive category

whereas Focus is not.68 This is compatible with the interpretation of senten-

———–——————————
67
A full motivation of the structure is beyond the scope of this book.
68
This has been refuted by Beninca (2001), where she basically argues that there
is only one [the higher one] recursive topic position. Beninca and Poletto (2004)
exploit and apply this to Medieval Romance.
142 Part II – The functional make up of the noun phrase

tial Focus: the complement of FocP is the part of the informational structure
that is presupposed, whereas its specifier is the focal part. Topic Phrases
undergo free recursion since their complement can in turn be another topic-
comment structure, and so on (see Rizzi 1997, for details). Crucially, a
clause can contain many topics but only one focus. This is illustrated in the
Modern Greek example in (119).

(119) a. O Janis, sta pedhja tu, to spiti tha to afisi

the John, to-the children-his, the house will it leave
xoris epifilaksi. (Greek)
without hesitation
‘John will leave the house to his children unreluctantly.’
b. * O Janis sta pedhja tu tha afisi to spiti (xoris epifilaksi)
the John to-the children-his will leave the house (without
reluctance)

Several proposals seek to establish a parallel articulation of (the left periphery

of) the nominal phrase as comprising a Topic and a Focus category. Aboh
(2002, 2004a,b) presents evidence from Gungbe, showing that the noun
phrase in that language comprises both a TopicP and a FocusP whose
specifiers host topicalized and focalized nominal constituents respectively.
In the spirit of Rizzi (1997), Aboh claims that Top and Foc

project within the DP, the highest projection of the nominal left periphery
that expresses the interface between discourse and the nominal expression,
and NumP, the lowest projection of the system, that is, the juncture between
the nominal left periphery and the nominal inflectional system. As such,
NumP encodes the agreement features and certain referential features (…)
that parallel those of the inflectional domain. (Aboh 2004a: 4)

He further takes D to be the equivalent of Rizzi’s (and Chomsky’s) Force,

because it is a subordinator. Aboh makes the claim that the nominal and the
clausal periphery are strictly parallel and that Topic and Focus project be-
tween D (Force) and Num (Fin).
Ihsane & Puskás (2001) also provide evidence for a Topic Phrase in the
left periphery of noun phrases. The argument the authors bring in favor of
this projection is based on the observed split between definiteness and spe-
cificity. They show that noun phrases introduced by the definite determiner
are definite but that, at the same time, they are not necessarily interpreted as
specific too; specificity should be thus kept apart from definiteness:69
 Chapter 1 – The emergence and the structure of DP 143

(120) a. J’ai pris le train. (French)69

I have taken the train
b. I have taken the train.

The DP le train (‘the train’) in (120a) and the DP the train in (120b) is defi-
nite but it can be interpreted either as a specific or as a non-specific DP.
The latter case applies to the situation in which the DPs in question do not
refer to a specific train; they fail to pick out a particular train (of that time
or that type or color or origin), but they refer to the train used as a means of
transport, as opposed to, for example, the bus, or the taxi or the boat. Con-
trast this interpretation to the specific interpretation of the twelve o’clock
train in (120c).

(120) c. I took the twelve o’clock train.

The authors propose that it is the highest projection in the left periphery
that is associated with encoding specificity. In particular according to them,
the relevant projection is that which contains elements linked to the dis-
course, i.e. Topic Phrase. Definiteness, as distinct from specificity, is en-
coded in a lower DefinitenessP. Thus they have the following hierarchy:

(121) Top >…DefP…

The head of TopP carries the feature [+SPECIFIC] and is the licenser of con-
stituents which give rise to a specific interpretation such as the specific
definite article and demonstratives (cf. also 4.4 above).
Aboh (2002, 2004a,b) produces evidence that in Gungbe specificity is
morphologically marked. Specific noun phrases can be either definite or in-
definite (much as is the case with many other languages, such as English or
Greek or French for instance). In his approach, the specificity marker is
———–——————————
69
Ihsane and Puskás also invoke Giusti’s example (22) above, where, as we saw,
the presence of the definite article does not necessarily induce reference. How-
ever, it is more accurate to say that the contrast Giusti’s example reveals is the
contrast between reference and attribution in Donnelan’s (1967) terms. The
definite article may well designate attribution (which in the particular example
is signaled by the subjunctive on the verb). The relevance of Giusti’s example to
the distinction between definiteness and specificity is thus quite oblique as it is
reference and not attribution that implies specificity is – i.e. a specific noun
phrase can only be referential, not attributive, while a referential noun phrase is
not necessarily specific, as Ihsane & Puskás own example (2001: 120) shows.
144 Part II – The functional make up of the noun phrase

also hosted by the head of TopP. The nominal constituent that is interpreted
as specific is taken to be a predicate that moves from its original position to
the specifier of TopP to check the feature [SPECIFIC] under Top.
Dimitrova-Vulchanova & Giusti (1998) present some empirical evidence
for postulating a DP-internal FocPhrase and a TopPhrase. However, in their
proposal these projections are not considered to be universal. Moreover,
their position is subject to cross-linguistic variation as they can both pre-
cede and follow the DP. As we will see below, for instance, in Albanian a
DP-internal FocP is postulated as the target of a certain type of A’-move-
ment of the AP. This operation is also available for genitive noun phrases.
Since the moved constituent follows the determiner, Dimitrova-Vulchanova
and Giusti propose that the relevant functional projection is situated imme-
diately below DP. In Bulgarian, both Topic movement and Focus move-
ment are found, but here the preposed elements precede the determiner. This
leads the authors to propose that the projection which is targeted dominates
D. Furthermore, in Bulgarian Topic movement is restricted to genitive
phrases. In Romanian, finally, neither extension of DP is found.
In using the term A’-movement to characterize DP-internal movement
processes like those illustrated below, Dimitrova-Vulchanova & Giusti
(1998) intend to characterize a type of operator movement that is motivated
by information structure and which gives rise to a marked word order. We
illustrate some such cases below.
Consider the following Albanian data (from Dimitrova-Vulchanova &
Giusti 1998: 348ff). They contain adjectives modifying an event nominal.
The adjectives appear in a fixed order, the thematic adjective must be lower
than the descriptive adjective. Albanian differs from Italian in that in Alba-
nian the noun precedes both adjectives, while in Italian the noun appears
between the high adjective and the low one (see Chapter 1 of Part III for a
full discussion of adjective orderings):

(122) a. pushtimi i tmerrshëm italian i Shqipërisë (Albanian)

invasion-the terrible Italian of-Albania
b. la terribile invasione italiana dell’Albania (Italian)
the terrible invasion Italian of Albania
‘the terrible Italian invasion of Albania’

(123) a. *pushtimi italian i tmerrshëm i Shqipërisë (Albanian)

invasion-the Italian terrible of-Albania
b. *l’italiana invasione (terribile) dell’Albania (Italian)
the Italian terrible invasion of Albania
 Chapter 1 – The emergence and the structure of DP 145

The prenominal position of the adjective in Albanian is marked. In this po-

sition, either adjective can appear. This is the case both in object-referring
nominals (124) and in event nominals (125):

(124) a. tjetr-a grua e bukur (Albanian)

other-the woman ART-nice
b. *e bukur-a grua tjetër
ART-nice-the woman other
‘the other nice woman’

(125) a. ?i tmerrshëm pushtimi italian i Shqipërisë

b. ??Italian pushtimi i tmerrshëm i Shqipërisë
the Italian invasion terrible the of-Albania
‘the terrible Italian invasion of Albania’
This suggests that the prenominal position is derived by movement of the
AP. Furthermore, if the A/A’-distinction can be applied to the movement of
adjectives, then it can be argued that the adjectival movement observed
here is a kind of A’-movement, because it serves information-structure pur-
poses. Dimitrova-Vulchanova & Giusti assume this movement to be an in-
stance of Focus movement.
If AP-movement is to be assumed to account for these cases, one should
inquire what the target position is. The landing site of the adjective follows
the position of the demonstrative, which the authors identify as SpecDP (cf.
Giusti (1993), Brugè & Giusti (1996), see also the discussion in section 4.2).
A focused AP cannot precede the demonstrative (all the data that follow are
from Dimitrova-Vulchanova & Giusti 1998: 349, unless otherwise stated):

(126) a. kjo (shumë) e bukur(a) grua tjetër

this (very) ART-nice(-the) woman other
b. *e bukur(a) kjo grua
ART-nice(-the) this woman
c. *tjetra/tjetër kjo grua
other(-the) this woman

Observe that not only adjectives may occupy the position to the immediate
right of the demonstratives. Possessives can also be moved there as shown
in (127):

(127) a. ky libër i Benit

this book ART-of-Ben
146 Part II – The functional make up of the noun phrase

b. ky i Benit libër
this ART-of-Ben book
c. *i Benit ky libër
this book of Ben’s
The fact that this derived position can host elements of various categories
strongly supports the proposal that it is a derived position of A’-type. The
structure proposed for Albanian by Dimitrova-Vulchanova & Giusti (1998:
350) is given below:
(128) DP

Spec D’

D FocP

Spec Foc’

Foc AgrP

Spec Agr’

Agr …
NP

Spec N’
N
a. (një) gruai [tjetër] t’i [e bukur] ti
(një) [tjetër]j gruai tj t’i [e bukur] ti
(një) [e bukur]j gruai [tjetër] t’i tj ti
another nice woman
b. [DEF]k gruai-jak [tjetër] t’i [e bukur] ti
[DEF]k [tjetr-ak]j gruai tj t’i [e bukur] ti
[DEF]k [e bukur-ak]j gruai [tjetër] t’i tj ti
the other nice woman
c. kjo [DEF]k gruai(-jak) [tjetër] t’i [e bukur] ti
kjo [DEF]k [tjetër/r-ak]j gruai tj t’i [e bukur] ti
kjo [DEF]k [e bukur(-ak)]j gruai [tjetër] t’i tj ti
this other nice woman
 Chapter 1 – The emergence and the structure of DP 147

Bulgarian also displays A’-movement inside the nominal structure, though it

has different properties. Firstly, DP-internal fronting distinguishes between
possessor DPs and adjectives, in that topicalization applies exclusively to
possessor phrases. Secondly, the landing site of the movement is to the left
of the position where demonstratives are found.

(129) a. tezi novi knigi na Ivan

these new books to Ivan
b. na Ivan tezi mu novi knigi
to Ivan these CL-DAT-3SG new books
‘these new books of Ivan’s’

Dimitrova-Vulchanova & Giusti assume that (129a) underlies (129b) which

is derived via fronting of the na -DP.
Given the relatively free word order in the Bulgarian clause, it is difficult
to establish whether the preposed possessor is actually still inside the DP or
whether it has ‘scrambled’ out of the DP. Dimitrova-Vulchanova & Giusti,
based on a rather complicated interaction of data (1998: 351ff,) conclude
that the possessor forms a constituent with the rest of the DP (see also Giusti
& Stavrou, to appear, for detailed discussion on this).
Of course, the na-DP can be extracted out of its host DP, as an instance
of fronting (130a) or as wh-movement (130b), just as the possessor can
(raise and) be found in front of the definite article in Greek or Hungarian
too, as we saw above:

(130) a. Na Ivan ja procetox knigata *(mu) na studentite.

to Ivan I read book-the *(CL) to students-the
‘A book of Ivan’s, I read to the students.’
b. Na koj izvesten gruzki filisof kupi portet(a) (*mu)?
of which famous Greek philospher did you buy [a]/the portrait?
c. Na IVAN kupix portret(a) (*mu)!
of IVAN I bought the/a portrait!

The authors present the derivation of topicalization in the Bulgarian noun

phrase structure as in (131) (Dimitrova-Vulchanova & Giusti 1998: 354).
Observe that they also postulate a clitic projection (ClP) within the nominal
periphery.
148 Part II – The functional make up of the noun phrase

(131) TopP

Top’

Spec Top CLP

Spec CL’

CL DP

Spec D’

D AgrP

na Ivan tezii mu ti ti … knigi

The position in which the na DP lands is identified as SpecTopP, because

the moved possessor phrase receives a topic interpretation.
If the fronted constituent is not a topic, the na DP in a left peripheral po-
sition is only allowed if it receives contrastive focus. Thus it is claimed that
there is also a Focus position in the extended nominal projection in Bulgari-
an. The head of this projection can be overtly realized by the question clitic
li, as in (132b,c). This type of construction represents the only instance of
DP-internal A’-movement of demonstratives (or APs) in Bulgarian (Dimi-
trova-Vulchanova & Giusti 1998: 355):

(132) a. Na Ivan knigata (*mu)

to Ivan book-the (*CL)
b. Na Ivan li tezi (*mu) knigi
of Ivan Qcl these CL books
(questioning ‘na Ivan’)
c. tazi li kniga/negovata li kniga
this Q-CL book / his-the Q-CL book
(questioning ‘this’/’his’)
 Chapter 1 – The emergence and the structure of DP 149

d. FocP

Spec Foc’

Foc DP

Spec D’

D AgrP

na Ivan li tezii ti... knigi

Note that in the presence of the question clitic li, the possessive clitic can-
not surface at all, which implies essentially that in the construction type in
(132), no TopP is projected. It can be suggested that in focus constructions
the two projections collapse into just one, e.g., FocP (cf. Kiparsky (1995)
for a diachronic proposal viewing CP in Germanic as having collapsed the
Proto-Indo-European [ TopP … [FocP …]] into one projection).
The similarity between (132a) and the cases of possessor fronting in
Greek, as given in (89b’) above repeated here as (133a) and further in
(133b), is striking:

(133) a. tis Marias i apoplanisi

the Mary-GEN the seduction
‘Mary’s seduction’
b. tu Jani to vivlio
the John-GEN the book
‘John’s book’

In the light of these facts from Bulgarian and the accompanying observa-
tions, and aiming at establishing a structure that can account for as many
aspects of a phenomenon as possible, and in as many languages as possible,
we can revise the analysis for Greek as originally proposed by Horrocks &
Stavrou and assume that the structure for Bulgarian in (132d) is relevant
also for Greek. Recall that in Greek the genitive DP encoding the possessor
can be A’-moved to SpecDP. Demonstratives are also standardly taken to
move to SpecDP from a lower position – cf. (68), (84) and (99).
Horrocks & Stavrou argue that both the pre-article genitive and the de-
monstrative bear emphatic stress and are focused. Then one only needs to
150 Part II – The functional make up of the noun phrase

make one more step to also assume that these pre-article constituents in
Greek can be hosted by a projection above DP, namely a FocP. This leads
to the conclusion that FocP is a projection in the extended projection of the
Greek noun too.
Observe that following Dimitrova-Vulchanova and Giusti’s proposal
there is massive parametric variation with respect to the positioning of DP-
internal TopP and FocP. In particular, while sometimes the relevant A’ pro-
jections dominate DP, in other cases the hierarchy is the opposite. Though
as such this may be an option, the proposal ends up being quite different
from Rizzi’s initial split CP proposal according to which parametric varia-
tion was restricted. However, this problem may be only apparent. Recall that
we have proposed that in the same way that Rizzi split C into Force and
Fin, ‘D’ itself can be split into two projections, which we labeled DP1 and
DP2 (see (94)). The lower projection was responsible for definiteness. (See
also Haegeman (2004) for an implementation of this proposal.) Though fur-
ther research is required to substantiate this view, one way of reconciling
the apparent diverging landing sites of A’ movement proposed by Dimi-
trova-Vulchanova and Giusti could be to say that the relevant TopP and
FocP in fact are sandwiched between the higher DP and the lower DP,
much in the same way that Rizzi inserts TopP and FocP between ForceP
(the ‘higher CP’) and FinP (the lower CP).
As has been observed above, the articulation of the DP area also ties in
with the position of possessors in the DP. We return to this issue in Chapter
1 of Part IV. In the same spirit, it has been proposed that, when stressed
with focus intonation the articled adjective in the so-called polydefinite
construction may also be moved as a head to the head Foc of a FocP which
is found above DP (Campos & Stavrou 2004: 163). We turn to these con-
structions in more detail in Chapter 1 of Part III. Another, and related, em-
pirical area which further supports the presence of a FocP, reflecting the
clausal FocP along the lines of Rizzi (1997), is the DP-internal movement
of prenominal adjectives. As we will see in Part III, Chapter 1, DP-internal
adjectives usually present themselves in a canonical order but this canoni-
cal order of adjectives in the DP can be disturbed if an adjective is contras-
tively stressed. The stressed adjective stands out and is apparently displaced
from its original position in the hierarchy. The ‘internal’ FocP can then be
invoked to host the emphatically stressed (prenominal) adjective found in a
position different from that predicted by the universal ordering hierarchy
(see Chapter 1 of Part III for discussion). Consider (134):
 Chapter 1 – The emergence and the structure of DP 151

(134) a. i omorfi palia ksilini karekla (Greek)

the nice old wooden chair
b. i omorfi ksilini palia karekla
the nice wooden old chair

This point is made by Dimitrova-Vulchanova & Giusti (1998) with respect

to Albanian. We will not pursue the issue of the DP-internal Focus position
further here. Suffice it to say that this ‘nominal’ focus category may further
interact in rather intricate ways with sentential focus (see Aboh 2002,
2004a,b4). For a proposal that DP-internal focussing interacts with ellipsis
see Corver and Van Koppen (2006).

6. The category D and Case

Before closing the chapter a brief mention needs to be made of the connec-
tion between Case and the category D. How are these two functional cate-
gories related?
The relation between case and definiteness is a common theme in the
literature. It is a well-known fact that in some languages case markings in-
teract with definiteness-indefiniteness or specificity-non-specificity or even
the whole-part distinction (Finnish, Hungarian).70 In other languages the di-
rect object is marked by accusative (or another marker) only if it is definite
(Turkish, Modern Hebrew), as will be shown in the data in (135) below.
Such phenomena have been taken as indications of a close relationship be-
tween case and definiteness.
An interesting observation in connection with this issue is found in Giusti
(1995); although verbal inflection has been analyzed as

a syntactic process that takes place due to the presence of functional projec-
tions (cf. Pollock 1989, Beletti 1990, among others), the study of the prop-
erties of nominal inflection has not been related to functional nominal pro-
jections. (Giusti 1995: 77)

Giusti was writing these lines back in 1995. We already saw that the inade-
quacies of the older accounts referred to by Giusti have been eliminated
over the years by the introduction of various functional categories that were
———–——————————
70
See Lyons (1999), section 5.1 in particular for a comprehensive presentation of
these and other related phenomena.
152 Part II – The functional make up of the noun phrase

thought of as relevant in the nominal domain and which parallel the func-
tional architecture of clause. In the following chapter we will turn to the
role and the raison d’ être of certain inflectional categories projected be-
tween D and NP.
In older formulations of the DP-hypothesis, D was thought of as simul-
taneously hosting features of (in)definiteness, gender, number and, crucially,
Case. Such was the approach by Loebel (1989, 1993) and it has also been
maintained by Giusti throughout her work on DP (1993, 1995, 1997, 2002).
However, Loebel (1994), echoing work by Lamontagne & Travis (1986,
1987), proposes that the nominal structure also contains a Kase Phrase,
dominating DP itself. In other words she proposes that DP is split between
a category that bears Case features and a category that hosts the determiner.
Let us go briefly over the argumentation behind this proposal, as first formu-
lated by Lamontagne & Travis (1986, 1987) and developed by Loebel (1994).
The primary evidence for postulating a separate functional head for Case
concerns the alternation between Case marking and zero realization of a
Case feature in some languages. The following examples from Turkish illus-
trate the point:

(135) a. Hasan dün bu pasta-yi ye-di. (Turkish)

Hasan yesterday this cake-ACC eat- PAST (from Loebel 1994: 41)
‘Hasan ate this cake yesterday.’
b. Hasan dün pasta ye-di.
Hasan yesterday cake eat- PAST
‘Hasan ate cake yesterday.’

In (135a) the accusative marking appears on the object noun, whereas in

(135b) it does not. This may well be due to the specificity vs. non-specificity
of the noun phrase involved. When it is interpreted as specific the accusative
marking is present, when non-specific the object appears without the accu-
sative marking. But things are clearly not as simple as that: even when non-
specific, the accusative marking cannot be dropped if the object is not adja-
cent to the verb; thus (135c) is ungrammatical (Loebel 1994: 42):

(135) c. *Hasan pasta dün ye-di.

Hasan cake yesterday eat- PAST

The same effect is reproduced for Japanese (Loebel 1994: 42).

 Chapter 1 – The emergence and the structure of DP 153

Based on such evidence Lamontagne & Travis propose there is a parallel-

ism between CP and a nominal projection involving C/case. 71 Recall from
what we said in the preceding section that DP has been seen as parallel to
CP on general semantic and syntactic grounds. In (136) we notice that
Comp is optional:

(136) a. John believes (that) Mary will win.

b. John believes wholeheartedly *?(that) Mary will win.
(Lamontagne & Travis 1986: 57; Loebel 1994: 42)

Optionality of complementizer depends on nothing intervening between the

embedded and the matrix clause (136a). When something intervenes, the
complementizer is obligatory (136b). Exploiting this observation, Lamon-
tagne & Travis claim that COMP-drop and Case-drop are very similar, as
both involve the presence of an empty element and in both this element is
subject to strict adjacency: adjacency is the key notion uniting COMP-
deletion and the omission of the morpheme realizing Case on nominals.
The parallelism between the two phenomena is captured by Lamontagne &
Travis by the postulation of a category dubbed KP, ‘Kase’ Phrase. Kase
Phrase is aligned to CP in clauses. KP selects DP. 72
Loebel adopts and expands this idea of Lamontagne & Travis. The divi-
sion of labor between D and K is expressed by Loebel (1994) as follows:

I propose applying this argument to the noun phrase and “splitting up” the
category D into two separate functional categories, one with semantic con-
tent, where D itself functions as a feature bearer of referential features, and
one with primarily syntactic function, i.e. K for Case. (Loebel 1994: 51)

K is licensed by the feature [+Case].73 The modified nominal structure is as

follows:

———–——————————
71
Lyons (1999) also assumes a K head distinct from D.
72
Notice that Abney too argued for a functional category KP, however in his case
KP is the category that occurs at the position of SpecDP in nominal phrases like
Caesar’s destruction of the city (1987: 103).
73
Bittner & Hale (1996) also hypothesize a separate KP above DP but only for
‘marked’ cases – nominative, for example, is unmarked, so the DP encoding no-
minative case is caseless and therefore not a KP.
154 Part II – The functional make up of the noun phrase

(137) a. VP

V KP

K DP

D NP

Loebel illustrates her basic claims with the following sentence:

(137) b. Barbara hilft der Freundin. (German)

B. help-3SG the friend-dat
‘Barbara helps the friend.’

Which she analyses as in (137c):

(137) c. VP

KP V
[+Dative]
K DP helf-
[+Dative]
[+AGR]
D NP
-er [+def]
[+AGR] N

d- Freundin

KP is the projection directly selected by a higher case assigning head. As

Loebel underlines, K does not assign case; KP is case. The surface structure
is obtained through head movement of D which adjoins to K, giving rise to
the article form der. Loebel offers an articulated analysis relying on the
structural-inherent Case distinction and Case assignment. She argues that
Case assignment interacts with the referential features [+/– definite] and
[+/– specific] and it further depends upon such nominal features as [+/–
human] and [+/– animate]. We are not going to go into the details of Loebel’s
analysis here, as this would take us too far. The reader is referred to her ar-
ticle for details concerning the ways her approach to a split D/K hypothesis
capture the alternation between partitive and nominative case in Finnish and
 Chapter 1 – The emergence and the structure of DP 155

the partitive (genitive) case in Russian, unaccusative verbs in English, the

personal a-object construction in Spanish and bare NP- adverbs in English.
By contrast, the connection of the category D with case does not lead
Giusti (1995; also 1993) to propose a separate functional category but, on
the contrary, it leads here to a unified analysis of the relevant nominal pro-
jections. As we mentioned earlier on (see the discussion of (27) in 2.5), for
Giusti, the article is a syntactic means of expressing case. As we also said
at the beginning of this section, definiteness is linked by some linguists to
the loss, or reduction, of case markings, as is the case with the Romance
and the Germanic languages (Philippi 1997; also Giusti 1993, 1995. But
see Lyons 1999: 324 and also in pp. 217–219 for doubts about the central-
ity of determiners in the expression of nominal inflectional categories – let
alone case). Giusti postulates that case markings are ‘transferred’ onto the
determiner after they have been lost on the noun;74 in either case the ex-
pression of case and through it the exact argumental function of a given
noun phrase is ensured.
Giusti assumes the projection DP as a saturator of the predicate NP (in
the sense of Longobardi (1994) and others, see section 2.2), and that only
case-marked constituents can function as arguments. Giusti (1993, 1995)
postulates one functional projection FP, which conflates or unifies referen-
tiality and case. FP75 is taken to be the highest nominal projection – accord-
ingly the nominal structure expands as follows:

(138) FP

AgrP
. ..
..
.
NP

While the head of FP is universally instantiated to satisfy the Case filter, its
specific instantiation varies cross-linguistically; F may host the article or a
case morpheme, in a language that marks case on nouns, or even nothing at
all (see following chapter). Giusti’s proposals for the conflation of case and
reference/argumenthood under a single functional head (namely F), is con-
———–——————————
74
Giusti appeals to Renzi (1984), who proposed that in those languages that have
an article, it emerged at a stage in which morphological case was eliminated or
at least weakened.
75
Giusti’s FP corresponds to the ‘traditional’ DP. The label FP is used to accom-
modate those data in which case seems to take over the role of the article.
156 Part II – The functional make up of the noun phrase

tingent on the assumption that articles tend to be in complementary distri-

bution with case morphology.
It should be noted here that viewing the emergence of the article as a
consequence of the reduction or the loss of case distinctions in the case sys-
tem of a language leads to different claims compared with what we said
earlier on (see section 4.1) about the emergence of the article from some
kind of reduction or weakening of demonstratives. To the best of our
knowledge no analysis of the rise of the article has combined the two routes
– via the (weakening of) demonstrative and via (loss of) case markings on
nouns. Quite importantly, there are languages where the article in all of its
basic functions co-exists with a rather rich case system on nouns (Greek,
Icelandic, German and others). The issue of the interrelations between case
and (in)definiteness will not occupy us any more here.

7. Summary of the chapter

In this chapter we have discussed a number of issues that relate to the deter-
mination area of the nominal projection.
In section 2 we concentrated on the function of the article, in particular
of the definite article. We have examined the various properties that have
been attributed to it, both semantic/pragmatic and syntactic.
In section 3 we turned to the DP hypothesis: the idea that in the same
way in which the lexical projection VP is dominated by functional material,
the lexical projection NP is dominated by DP, the projection whose head
hosts the article. We also surveyed some problems for the DP hypothesis.
Section 4 is essentially focused on definite articles and demonstratives,
elements which are commonly classified as determiners and seem to be
naturally related to D and its functional domain. Though definite article and
demonstratives at first sight share a number of semantic and distributional
properties it turns out that there are also arguments for not amalgamating
them completely. In one particular proposal it is argued that while the arti-
cle is a head element that fills D, the demonstrative is a phrasal constituent
that may (but need not) fill SpecDP. 76 We also examine in more detail the
———–——————————
76
See Lyons (1999, especially his Chapter 8) for a different proposal according to
which free form definite articles are specifiers of DP, while affixal articles are
realized under D. This proposal echoes Cinque’s work on modifiers as specifiers
of functional categories, and also relies on the assumption that DP is essentially
a definiteness phrase. See also subsection 3.1.4.
 Chapter 1 – The emergence and the structure of DP 157

various positions that can be occupied by the demonstrative and the ac-
counts that have been proposed to relate these positions.
In section 5 we returned to the more general build-up of the DP and we
examined further analogies between the nominal periphery and the clausal
periphery. Among other things we showed that just as there can be argu-
ments for splitting up the CP layer of the clause into a number of articulated
projections, arguments have been put forward for splitting up DP into a
range of articulated projections.
Finally, in section 6 we briefly mentioned possible correlations between
case marking and DP, formulated by two different recent approaches; one
approach holds that case is encoded on a separate functional head (K) and
the other holding that D (labeled F) subsumes both case and whatever se-
mantic notions are carried by the definite article.
In the next chapter we will address the issue of the omission or lack of
the definite article in certain languages or in certain DPs in languages
where there is a definite article.
Chapter 2

Determinerless Noun Phrases

1. Introduction: D across languages

1.1. Languages without (definite) articles

In the previous chapter we listed a number of roles that are regularly attrib-
uted to the functional element D. We also discussed the fact that according
to many researchers D is like C in that they both turn their complements
into arguments, and in that only DPs and CPs can function as arguments
(Stowell 1989, 1991; Szabolcsi 1994; Longobardi 1994, among many oth-
ers). This is basically attributed to the fact that D imparts referentiality to
its complement and that (syntactic) arguments are entities that have refer-
ence (Higginbotham 1985). One issue which we remained vague about,
however, was the exact relation between D as a functional head, semantic
entities such as definiteness and referentiality, and the items that are stan-
dardly taken to be the natural realizations of D, e.g. the (definite) article,
demonstratives, etc. The natural realizations of D such as the definite article
and demonstratives are usually conceived of as the natural expressions of
the semantic categories of definiteness and referentiality. So, in a sense, the
functional head D mediates between semantic entities (definiteness, refer-
entiality) and their phonetic expression. Our vagueness on this issue re-
flects the general vagueness and uncertainty found in much of the recent
literature with respect to the precise functions of D. In a nutshell the prob-
lem can be stated as follows:

Is reference conveyed to the NP by virtue of the functional

head D as such, or is it conveyed by the lexical realizations
(i.e. the fillers) of D?

In the previous chapter we surveyed some views according to which it is

essentially the structural position D that assigns referentiality to its NP
complement, and it is not the article per se that is to be held responsible for
that.
160 Part II – The functional make up of the noun phrase

In this chapter these views will be further supported by the cross-linguistic

evidence we will discuss and which mainly bears on the ways in which
noun phrases without articles (‘bare nouns’) are licensed and interpreted.
The starting point for our discussion here is the observation that not all lan-
guages have the type of morphemes that could be considered as articles
(definite/indefinite).1 And yet, noun phrases in these languages are clearly
interpreted in the same way that noun phrases in languages like Greek or
English, which do have articles, are interpreted.
For an illustration consider the following examples from Latin, a lan-
guage without a definite article (the relevant nouns are in italics; the nota-
tion ‘:’ marks long vowels):2

(1)
a. Annum age:ns sextum decimum patrem a:mi:sit.
year during 16th father lost.
‘During his 16th year he lost his father.’
(Suetonius, Life of Julius Caesar 1.1)
b. In expugna:tio:ne Mytile:na:rum coro:na: ci:uica:
in storming of Mytilene crown civic
do:na:tus est.
awarded PART MASC was
‘At the storming of Mytilene he was awarded a civic crown.’
(viz. like a medal). (Suetonius, Life of Julius Caesar 2)
c. Uastante regio:ne:s proxima:s Mithrida:te.
devastating regions neighboring Mithridates
‘while Mithridates was devastating the neighboring regions.’
(Suetonius, Life of Julius Caesar 4.2)
———–——————————
1
Krámský (1972) proposes a typology of seven language types based on the for-
mal means that express the category ‘determination’ (see below, subsection 2.2).
Type A consists of languages in which the category ‘determination’ (vs. ‘inde-
termination’) is expressed by independent words, type B consists of languages
in which one member of the category is expressed by an independent word, and
the other by an enclitic or a proclitic. Type C contains languages where both
members of the category are coded in enclitics or proclitics. Type D, which di-
rectly concerns our discussion here, consists of languages in which this category
is inherent in the noun itself. In type E the category ‘determination’ is expressed
by inflection, while in type F it is expressed by prosodic means. Finally, type G
contains languages that have a zero category for determinedness. Here, following
the mainstream literature on articleless nouns/languages, we will be concerned
exclusively with the lack of the definite article.
2
We thank Geoffrey Horrocks for providing us with the Latin examples.
 Chapter 2 – Determinerless Noun Phrases 161

As the English translations show, in all of the above examples, the under-
lined Latin noun corresponds to a noun preceded by a definite article or an
equivalent determiner (e.g. a possessive adjective) in English. It is thus
natural to conclude that the English-type article is just one way to express
definiteness and indefiniteness (Felix 1988; Lyons 19993).
Another language that illustrates this point is Polish. Polish has no defi-
nite article, but the demonstrative pronoun ten (‘this’ MASCULINE ), ta
(‘this’ FEMININE), to (‘this’, NEUT) is regularly used in front of common
nouns in much the same way that the definite article is used in languages
like English (Masiejewska 1996):4

(2) Kiedy kończysz tę swoją pracę?

when finish-2SG this your work?
‘When are you finishing off your work?’

Such a use of a demonstrative pronoun to replace the article is also ob-

served in Latin (3a), as well as in Japanese, another language in which there
is no definite article, as shown in (3b):

(3) a. Illa die:s ueniet mea qua: lu:gubria po:nam.

that day come-FUT-3SG my when mourning put-1SG
‘The day will come when I will put aside my mourning.’
(Ovid, Tristia 4.2.73)
b. Peter-to Mary-wa sono heya-ni odori-konda. (Japanese)
Peter and Mary-TOP that (the) room-into danced
‘Peter and Mary danced into the room.’
———–——————————
3
For instance, as Lyons (1999, section 2.4) points out, other ways for languages
to express definiteness are adpositional marking, agreement with the verb, word
order, pronominal marking. Likewise Krámský (1972) writes:
Languages that do not possess an article of the same kind as English or German does,
can express the same concept by means of other grammatical categories. Some Asian
languages possess an objective case which corresponds with the definite article of let
us say English. The difference between the nominative and the partitive case in Fin-
nish resembles in a way the function of articles. (Krámský 1972: 28–29)
See Chapter 1, section 2.5. for Giusti’s claims concerning the purely grammati-
cal functions of the definite article.
4
Krámský (1972: 188) attributes the same claim about the use of the demonstra-
tive pronoun ten as a definite article in Czech to Josef Zubaty. And the same is
reported for Sorbian (1972: 190).
162 Part II – The functional make up of the noun phrase

Two questions arise at this point: (i) How do noun phrases get interpreted
in languages that do not have articles? (ii) Under the assumption that D
turns the NP into an argument, does the projection DP exist in all languages
regardless of whether a language does or doesn’t have articles? To put it
differently: Do languages like, e.g., Latin, Slavic languages like Polish,
Hindi, Mohawk, which do not have a definite article, have/need a DP layer
the head of which will remain empty (non-lexicalized)? Or is it the case
that these languages have impoverished nominal projections, in particular,
are their nominal projections merely NPs?
The answers that have been proposed to these questions vary. While
many authors adhere to the more or less standard view about the universality
of D as a category relevant to semantic interpretation, this has been chal-
lenged by Lyons (1999), who argues in favor of distinguishing the seman-
tic/pragmatic notion of definiteness from its grammatical exponent, namely
D. In his view, D is only projected in a language if this language encodes
semantic definiteness. If a language fails to encode definiteness, DP is ab-
sent.5 Lyons argues (1999, especially Chapters 8 and 9) that the creation of
DP entails the creation of definiteness marking (see also the preceding
chapter, 4.1.1 on the historical development of determiners). He says:
there can be no definite article in languages lacking DP structure, and, to the
extent that it is obligatory to have some expression of a projection, lan-
guages with DP structure must have a definite article. (Lyons 1999: 323)

On the other hand, Stowell (1989, 1991), Longobardi (1994), Szabolcsi

(1993, 1994), Giusti (1993, 1997, 2002), among others, maintain that the
functional head D is as such responsible for the interpretation of a noun
phrase as definite, generic, etc. This means that even in the absence of an
overt article a nominal projection that is interpreted as, say, definite will
have a D-projection. Concretely, while for Lyons (1999) the underlined
NPs in (1) are mere NPs, for the other authors mentioned they will be DPs
in spite of the absence of a determiner.
In this chapter we will survey some approaches to the issues sketched
above.
———–——————————
5
Lyons writes:
With this framework, it is possible to maintain that DP structure is necessary for a
language to have a definite article, and even that the creation of an article is a neces-
sary concomitant of the emergence of DP structure. This is a desirable position be-
cause the empirical evidence is for a much closer on the part of definite articles than
of other ‘definite determiners’ like demonstratives. (Lyons 1999: 323)
 Chapter 2 – Determinerless Noun Phrases 163

1.2. The distribution of the definite article in languages that have one

The question of the universality of D arises most clearly with respect to

those languages such as Latin and Polish, illustrated above, which lack the
article altogether. However it also arises with respect to languages that do
have articles. This is because in the latter languages determinerless NPs can
also be found in specific circumstances. In fact it is the latter issue that will
be the focus of our discussion. We will also suggest ways of handling de-
terminerless languages, but as will be shown in section 3.3. below, the de-
bate as to how to deal with these remains fairly open.
Let us look at some examples of languages that have determiners but
which also display determinerless nominal constituents. Both Greek and
English have a definite article. However, in both languages there are cases
where a nominal constituent can apparently function as an argument without
the article. This is illustrated by the underlined bare noun phrases in (4).
(4) a. Topsy loves cheese; Peter loves alcohol.
b. Battered cats can usually find a refuge.
c. I Topsy pini ghala. (Greek)
the Topsy drink-3SG milk
‘Topsy drinks milk.’
d. Dhjavazi (pola) vivlia.
reads-3SG (many) books
‘He (she) reads (many) books.’
But in the same languages such bare noun phrases are not generally admit-
ted and may give rise to ungrammaticality. In (5a,b) cat is ungrammatical
in the absence of an article, similarly in (5c) the absence of an article with
student leads to ungrammaticality. (5d,e) show similar effects in Greek:

a. *Peter loves cat.6

(5)
b. *I saw cat.
c. *Student came in.
———–——————————
6
One interesting complication arises here. While singular count nouns cannot
function as arguments they can do so in coordination, at which point the coordi-
nated string has a definite reading. The examples are drawn from Heycock and
Zamparelli (2003: 443) to whom we refer for detailed discussion.
(i) A black cat and a brown dog were fighting in the street.
a. *Cat was filthy.
b. Cat and dog were equally filthy.
164 Part II – The functional make up of the noun phrase

d. *O Petros aghapai ghata. (Greek)

the Peter love-3SG cat
‘*Peter loves cat.’
e. *Mathitis bice mesa.
student came-PAST-3SG in
‘*Student came in.’

One factor that determines the different status of (4)–(5) is that the bare
nominal constituents in (4) are headed by mass nouns (4a,c) or by plural
count nouns (4b,d). The nominal constituents in (5) are headed by singular
count nouns: apparently in both English and Greek a bare singular count
noun can neither appear in object position (5a,b,d) nor in subject position
(5c,e).7 However, Greek bare plurals do not have a free distribution in ar-
gument positions either. In Greek, a bare plural noun such as pedhja (‘chil-
dren’) is not fully acceptable in (preverbal) subject position:

(5) f. ??Pedhja pezun.

children play.3PL
Children play.8

In the light of the differences in grammaticality between the examples in (4)

and in (5) the following questions arise: What licenses bare noun phrases in
languages that have a definite article? How does a bare noun phrase get its
interpretation in the absence of an article?
One important difference between English and Greek is that while in
English the article does not normally occur before proper names and generic
nouns ((4a) and (6a)), in Greek, in contrast, it is required in front of both
noun types (6b,c):

(6) a. Cats are adorable creatures.

———–——————————
7
This pattern is thus different from the marked case in Brazilian Portuguese dis-
cussed by Schmitt & Munn (2003), in which bare singulars are allowed freely,
generally with the interpretational properties of bare plurals. We will come back
to Brazilian Portuguese later in the discussion. For deteminerless count nouns
with definite reading in dialects of Dutch see Oosterhof (2006a) and the refer-
ences cited there.
8
Marinis (2003) and Alexiadou (1996) discuss the conditions under which a bare
plural can be allowed in (preverbal) subject position. The reader is referred to
these works (and the references therein) for details concerning Greek.
 Chapter 2 – Determinerless Noun Phrases 165

b. *(I) Topsy aghapai *(to) tiri.

the Topsy love-3SG the cheese
‘Topsy loves cheese.’
c. *(I) ghates ine aksiolatrefta plasmata.
(the) cats are adorable creatures
‘Cats are adorable creatures.’

Thus a final question to be dealt with is: What underlies crosslinguistic

variation with regard to the licensing conditions of bare nouns (4)–(6)?

The questions that arise with respect to the absence of articles in languages
with determiners are very similar to those raised in the previous sub-section
with regard to languages that do not have a definite article. It will become
clear, as the discussion unfolds, that the understanding of each one of these
questions can cast light on the other. Questions concerning the (non)exis-
tence of articles in certain languages and concerning the interpretation of
determinerless noun phrases in others have given rise to a vast literature
both in the typological tradition (Krámský 1976; Christophersen 1939;
Chesterman 1991; Gil 1987; Lyons 1999, a.o.) and in the generative tradi-
tion (Longobardi 1994; Chierchia 1998, a.o.). As we will see, both tradi-
tions contribute to our better understanding of these issues.

The remainder of the chapter is organized as follows: in section 2 we will

review some earlier views on the issue of the universality of D as a nominal
functional category. These views emerged in direct response to the DP-
hypothesis. In section 3 we turn to two recent widely discussed accounts of
noun phrases that do not instantiate a determiner. In section 4 we summa-
rize the basic points discussed.

2. The functional head D and the configurationality of noun phrases

2.1. Setting the scene

Often the article is the leftmost constituent in the noun phrase and is not
preceded by anything else. Recall from the previous chapter (section 1) that
before the emergence of the DP-hypothesis, the commonly held assumption
was that the article occupied a specifier position within the nominal projec-
tion (NP). In particular, in Jackendoff’s three-level system, the article was
taken to occupy the highest specifier. In that framework, the noun phrase
was seen as an endocentric construction, headed by N. A specifier was pro-
166 Part II – The functional make up of the noun phrase

jected when there was an overt filler available. If an article preceded the
noun, the highest specifier was projected; if no article was present but a
quantifier or quantificational adjective was available (e.g. many books), the
immediately inferior specifier was projected, etc. Crucially for our discus-
sion here, if nothing preceded the noun, only N is projected:

(7) N’ ( )

N0

The idea that the position occupied by the article is a specifier of NP and
may or may not be projected, depending on its overt realization, is obviously
in sharp contrast with basic tenets of the DP-hypothesis, according to which
the nominal projection is no longer seen as a single endocentric construction
but is reinterpreted as the projection of the (functional) category D, which
selects the lexical NP as its complement. In this more recent approach, the
role of the article has in a sense been upgraded: being inserted in D, the
article is now associated with the head of the projection, i.e. it has become
associated with an obligatory constituent of the nominal projection. But of
course, it is this very hypothesis that has given rise to the discussion about
the status of the head D in the absence of an overt article. In particular, it
gives rise to the question whether, if no article shows up, D is (still) re-
quired as a structural position?
In the following sections we shall review some of the answers that have
been given to the question above.

2.2. Gil 1987, Loebel 1989

Gil (1987) claimed that languages split into two broad types according to
two co-varying parameters, the parameter of configurationality and the pa-
rameter of the count-mass distinction. The latter parameter is based directly
on the existence of plurality markers and the use of numeral classifiers in a
language. See also section 3 below.9
———–——————————
9
For Gil, these two parameters are the summary of seven typological correlates.
The details of these correlates need not concern us here, except for a brief men-
tion of his seventh correlate, which concerns the order of adjectives in the noun
phrase and for which Gil predicts that in non-configurational languages the hier-
archical interpretation of stacked adjectives is not available. See Chapter 1 of
Part III for detailed discussion of this issue.
 Chapter 2 – Determinerless Noun Phrases 167

The following table (Gil 1987: 256; Loebel 1993: 184) illustrates the classi-
fication: English and Japanese illustrate two types of languages:

Table 1. The configurationality parameter for noun phrases

Type A (English) Type B (Japanese)

10
a configurational noun phrases non-configurational noun phrases
b count vs mass nouns all nouns as mass

Gil’s typology is based on two interrelated observations. The first concerns

the obligatory marking of (in)definiteness. In languages like Japanese there
is no definite article. In languages like English, there is an article (a book,
the book); and articleless nouns have a very restricted occurrence compared
with the articled ones. As we will see below, in the latter type of languages,
only plurals and mass nouns can appear without an article.11
This leads Gil to propose that in Japanese the noun hon meaning ‘the/a
book’ is a zero level category (N0), there is no branching inside the NP; the
———–——————————
10
The issue of configurationality is an important one, but we cannot go into it here,
as it goes well beyond the immediate aims of our study. Put informally, a typical
property of non-configurational languages is that their constituent order seems
to be much freer than that of configurational languages. The reader is referred to
Lyons (1999: 153–156) for a concise excursus to the various views on (non)con-
figurationality. In terms of the theoretical framework adopted here, Chomsky
(1981) accounts for the distinction between configurational and non-configura-
tional languages like Japanese by proposing that in non-configurational lan-
guages there is no hierarchical structuring of the clause. The verb, the subject, the
direct object, the indirect object and the oblique are all part of a flat structure,
that is to say all of these constituents are dominated by a single node (labelled
‘S’ in the 1981 framework), without any more hierarchical structuring among
them. (Also Hale 1978, 1982, 1983.)
It is worth mentioning here that according to Lyons, languages are not neces-
sarily configurational or non-configurational but may vary in the different parts
of their grammars, being, for instance, configurational in noun phrases and non-
configurational in their clauses (like Hungarian) (Lyons, 1999: 154). According
to him, there are degrees of configurationality, languages showing properties of
each type to a varying degree.
11
As we will see later on (3.1.1), these are the same properties on which Chierchia
(1998) bases his own distinction between languages like English on the one
hand and languages like Chinese or Japanese on the other.
168 Part II – The functional make up of the noun phrase

NP is said to be ‘non-configurational’. This is represented in (8a). The Eng-

lish equivalents the book or a book are of the category N 0+1, i.e. the NP is
internally structured and hence is said to be ‘configurational’ (see note 10).
In particular, in English the NP is a three level endocentric category, ac-
cording to the Jackendovian framework (8b).

(8) a. Japanese b. English

N'''

Art N''

N'

N0 N0

hon the/a book

Similarly, Fukui & Speas (1986), echoing Fukui (1986), argue that in Japa-
nese the nominal phrase is non-configurational. In particular they propose
that it is not a DP but an NP. These linguists argue that Japanese lacks all
of D, COMP and INFL and as a consequence all the relevant constituents
are X’ and not X’’ (Fukui & Speas (1986: 134). See also Introduction, sec-
tion 2.3).
The conception of the nominal projection as a three level endocentric
category as illustrated above would obviously not be compatible with one
version of the DP hypothesis. According to this view, the article is the head
of a nominal functional projection, D and the functional head D is always
present because it is always interpreted at the component of Logical Form
(LF), independently of whether it is lexicalised or not (see Chapter 1).12
This view is captured in Loebel’s (1993) account which incorporates the
DP hypothesis. Loebel’s leading idea is that within the framework of the
DP hypothesis, languages do not differ with regard to configurationality: in
both English and Japanese, for instance, the category DP is present, i.e. NP
is embedded under D. For Loebel (1993), the source of crosslinguistic
asymmetries with respect to the presence or absence of the definite article
is found in a lexical property called ‘determinedness’ or ‘determination’.
Determinedness/determination is a property of the noun itself and not of
———–——————————
12
See Part I (Introduction) section 2.1 for a discussion of levels of representation.
 Chapter 2 – Determinerless Noun Phrases 169

determiners.13 Languages only differ in the lexical make-up of their nouns.

In particular, the inherent lexical attribute ‘determination’ gets different
values (+ or –).
Quoting Loebel:

In particular, it is claimed that all nouns of all languages are (1) either count-
able or non-countable, and are (2) either inherently determined or non-deter-
mined. We claim that countability and ‘determinedness’ resp. ‘determination’
in this sense are both lexical properties which are parametrized, i.e. that these
properties constitute ‘possible parameters’ which have syntactic influence
with regard to inflection (namely pluralization) and the configurationality of
noun phrases (namely the obligatory occurrence of the article).
(Loebel 1993: 183)

‘Determination’ is the superordinate term which comprises definiteness and

indefiniteness: [definite] and [indefinite] are simply realizations of ‘deter-
mination’. ‘Determination’ is an inherent property of all nouns, which are
thus characterized as [+determined] (and then as +/–definite), or as [–deter-
mined]. The noun hon (‘book’) in Japanese is [+determined], whereas the
noun book in English is [–determined]. If a noun is marked [+determined],
it will not be accompanied by an article; if it is marked as [–determined] it
will co-occur with an article. To explain the occurrence of the article, Loe-
bel invokes Emonds’ Invisible Category Principle (Emonds 1985: 227),14, 15
which states, informally, that a functional category which has a particular
———–——————————
13
Loebel’s (1993) account is reminiscent of Krámský’s (1972) account in which
the typology of nominal constituents is based on the term ‘determinedness’.
14
Emonds puts forward this principle to account for manifestations of so called
‘adverbial case NPs’ (see Emonds 1985: 224 for the term) or of NPs that occur
as indirect objects of verbs of ‘giving’. He assumes in particular that such noun
phrases are the complements of empty Ps that bear certain semantic features such
as DIRECTION, LOCATION, DATIVE (Emonds 1985: 224ff). Note that Emonds uses
the term NP rather than DP.
15
Loebel invokes the Invisible Category Principle in combination with the para-
metrized feature ‘determination’ concerning the interpretation of noun phrases
not only across languages but also within one and the same language. The feature
‘determination’ is parametrized both across languages, as we saw, but also within
the same language. The Invisible Category Principle can account for the fact
that English proper nouns – as well as mass nouns – do not have an article, while
singular common nouns must have an article. For Loebel proper nouns in English
are marked as [+determined], just like all nouns in Japanese are.
170 Part II – The functional make up of the noun phrase

feature or a feature complex, may remain empty if this feature (or these
features) is (or are) morphologically transparent on a phrasal sister of the
functional category. We can apply this principle to the functional category
D. If the feature of determination (along with features for gender, case and
number) is positively marked on a noun, as Loebel assumes, then by the
Invisible Category Principle, D may remain empty. This is typically the
case of Japanese: nouns themselves are [+determined] and hence D may
remain empty. Conversely, if nouns are marked as [–determined], as is the
case in English, then D will have to be phonetically realized.
Loebel points out that, being syntactically relevant, the feature [+/–deter-
mined] must be a syntactic feature and not a (purely) semantic feature. She
further claims that, with respect to the projection of D, the structure of the
nominal constituent in English and Japanese is the same. In particular, D is
present both in Japanese which lacks a morpheme for the (definite) article
and in languages like English, in which there is definite article. The reason
for this claim is the assumption (Loebel 1993: 192) that D is the site where
Case is realized and where Case is assigned (for some discussion of case in
relation to D, see Chapter 1 section 6, and see also the discussion of case in
Chapter 3). The DP the book has Case (nominative or accusative, according
to where it occurs): the book corresponds, according to Loebel, not to Japa-
nese hon (‘book’) but to the Japanese DP hon-ga ((‘the/a) book-NOMIN’)
(see Chapter 3). The relevant structures for the DPs in the two languages
are given in (9) (Loebel 1993: 192):

(9) a. DP b. D’

D’ NP D

D NP N’ Nom

Nom N’ N

the book hon- -ga

[+N, –determ.] [+N, +determ.]

The structure in (9b) does not contradict Loebel’s claims about why there is
no definite article in Japanese. In (9b) D is present because it hosts the
morpheme ga which realizes (nominative) case. It is in this respect that in
 Chapter 2 – Determinerless Noun Phrases 171

English and in Japanese the noun phrase is expanded in a parallel fashion.

(See also Chapter 1 section 6 for the alternation of case markers and articles
in D).
Now English proper names such as John also do not co-occur with a de-
terminer. We will return to these in more detail belong but we will briefly
sketch Loebel’s analysis here. Loebel treats the noun John in English as
parallel to the common noun hon in Japanese in that she takes both to be
inherently marked as [+determined]. For Loebel the [+definite] marking on
the noun allows for the D position to remain phonetically empty, the role of
the article hosted by D being taken over by the N itself in these cases:

(9) c. DP

D’

D NP

Nom N’

John
[+N, +determ.]

More recently, Lyons (1999: 155) has cast doubt on the validity of Gil’s
empirical evidence for the correla