The Layered DP: Form and Meaning of French Indefinites (Linguistik Aktuell Linguistics Today, Volume 124)

The Layered DP

Linguistik Aktuell/Linguistics Today (LA) Linguistik Aktuell/Linguistics Today (LA) provides a platform for original monograph studies into synchronic and diachronic linguistics. Studies in LA confront empirical and theoretical problems as these are currently discussed in syntax, semantics, morphology, phonology, and systematic pragmatics with the aim to establish robust empirical generalizations within a universalistic perspective.

General Editors Werner Abraham

University of Vienna / Rijksuniversiteit Groningen

Elly van Gelderen

Arizona State University

Advisory Editorial Board Cedric Boeckx

Christer Platzack

Guglielmo Cinque

Ian Roberts

Günther Grewendorf

Lisa deMena Travis

Liliane Haegeman

Sten Vikner

Hubert Haider

C. Jan-Wouter Zwart

Harvard University University of Venice

J.W. Goethe-University, Frankfurt University of Lille, France University of Salzburg

University of Lund

Cambridge University McGill University

University of Aarhus University of Groningen

Volume 124 The Layered DP: Form and meaning of French indefinites by Tabea Ihsane

The Layered DP Form and meaning of French indefinites

Tabea Ihsane University of Geneva

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Ihsane, Tabea.   The layered DP : form and meaning of French indefinites / Tabea Ihsane.        p. cm. (Linguistik Aktuell/Linguistics Today, issn 0166-0829 ; v. 124) Revised version of the author's thesis (Ph. D.--University of Geneva, in 2006) presented under the title: The construction of the DP domain. Includes bibliographical references and index. 1.  French language--Article.  I. Title. PC2251.I37   2008 445--dc22 isbn 978 90 272 5507 5 (Hb; alk. paper)

2008005165

© 2008 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents

Acknowledgements

chapter 1. Introduction 1. Setting the scene  1 2. The puzzle  2 2.1 On the interpretation of indefinites  2 2.2 On the structure of indefinites  3 2.3 What moves when and where  4 2.4 The French du/des ‘of.the’ construction  5 3. The proposals  6 4. Theoretical framework  8 4.1 Generative Grammar  8 4.2 The cartographic approach  8 4.3 Relativized Minimality  11 4.4 The structure of nominal phrases  13 4.4.1 From NP to DP  14 4.4.1.1 The co-occurrence of determiners  14 4.4.1.2 The DP-CP parallel  14 4.4.1.3 Movement  16 4.4.2 The nominal architecture  17 4.4.2.1 The inflectional domain  18 4.4.2.2 The left periphery  23 5. The Data  28 5.1 Two-way ambiguities  29 5.1.1 Milsark  29 5.1.2 Fodor and Sag  32 5.1.3 Enç  38 5.1.4 Diesing  41 5.1.5 Summary and comparison  43 5.2 Three-way ambiguities  45 5.2.1 Dobrovie-Sorin and Beyssade  45 5.2.2 Baunaz  47 5.2.3 Summary  50

ix

1

vi

The Layered DP: Form and meaning of French indefinites

chapter 2. un-NPs ‘a-NPs’ 1. Introduction  51 2. Types of un-NPs  53 2.1 Referential noun phrases  53 2.1.1 Definition and characteristics  53 2.1.2 Referential un-NPs  56 2.2 Quantificational noun phrases  61 2.2.1 Definition and characteristics  61 2.2.2 Quantificational un-NPs  62 2.2.3 Presupposition of existence  64 2.3 A third type of noun phrases  68 2.3.1 Interpretation and characteristics  68 2.3.2 Property-denoting noun phrases  71 2.3.3 The distribution of P-un-NPs  78 2.4 Conclusion  83 3. The structure of un-NPs  84 3.1 The structure of arguments  84 3.1.1 Arguments as DPs  84 3.1.2 Previous analysis  87 3.2 The lowest projection of the left periphery  91 3.2.1 The structure  91 3.2.2 Some consequences  94 3.3 The Quantifier Phrase  95 3.3.1 The structure  95 3.3.2 Scope islands  99 3.3.3 Scope interactions  100 3.4 The S-Reference phrase  102 3.4.1 The structure  102 3.4.2 Scope islands  106 3.4.3 Intermediate scope  107 3.4.4 Quantifier Raising without circularity  110 3.4.5 Semantic paraphrases  113 3.4.6 The limits of a Choice Function analysis  115 3.5 Comparison with two-way analyses  116 4. Conclusion  119 chapter 3. Du/des-NPs 1. Introduction  123 1.1 The analysis  123

51

123



Table of contents vii

The ambiguity of du/des-constructions  125 Structures in a pre-DP framework  131 1.3.1 The partitive constituent  132 1.3.2 The quantitative constituent  134 Types of du/des-NPs  137 2.1 S-referential des-NPs  137 2.2 Quantificational du/des-NPs  140 2.3 Property du/des-NPs  143 2.4 Conclusion  146 The structure  148 3.1 The left periphery  148 3.1.1 Property du/des-NPs  148 3.1.2 Quantificational du/des-NPs  149 3.1.3 S-referential des-NPs  152 3.2 The inflectional domain  154 3.2.1 The interpretation  154 3.2.2 The structure  162 3.3 More on partitives  168 3.4 Notes on the-NPs  182 Conclusion  184

1.2 1.3

2.

3.

4.

chapter 4. Bare nouns 1. Introduction  189 2. Types of bare nouns  191 2.1 Germanic and Romance  191 2.2 French  194 3. The structure  197 3.1 The left periphery  197 3.2 The inflectional domain  200 4. Coordinated bare nouns  204 4.1 The interpretation  205 4.1.1 Singulars  205 4.1.2 Plurals  208 4.2 The structure  210 4.2.1 The left periphery  210 4.2.2 The inflectional domain  213 5. Further issues  215 5.1 The generic reading  215 5.2 A note on incorporation  220 6. Conclusion  222

189

viii The Layered DP: Form and meaning of French indefinites

chapter 5. Conclusion 1. The results  225 1.1 The analysis and the proposals  225 1.2 Prediction for language acquisition  228 2. Further issues and future research  229 2.1 The unbounded reading  229 2.2 Languages without determiners  236 Bibliography Index

225

241 255

Acknowledgements This book is the revised version of my doctoral dissertation, submitted to the Department of Linguistics, University of Geneva, in 2006. In contrast to traditional acknowledgements, my first words are for those who are closest to me, in particular my husband who has always been supportive, my children who do not resent their mom for being so busy, my parents and family-in-law who gave me the opportunity to rest when I visited them or by taking care of the kids. On the linguistics side, the first person I would like to thank is Liliane Haegeman, who taught me most of what I know and whose enthusiasm for linguistics is contagious. This book has benefited greatly from stimulating discussions with other linguists and from written comments, pointers and suggestions on (parts of) previous versions. Special thanks go to Hagit Borer, Giuliana Giusti, Genoveva Piskás and Ur Shlonsky for their detailed and helpful remarks and critics and for their encouragements. I am also indebted to Enoch Aboh, Artemis Alexiadou, Lena Baunaz, Donka Farkas, Claire Forel, Eric Haeberli, Liliane Haegeman, Monica Jakots, Christopher Laenzlinger, Jacques Moeschler, Cecilia Poletto, Luigi Rizzi, Michal Starke, Eszter Varga, Eric Wehrli and Roberto Zamparelli. Parts of the analysis developed here have been presented in 2006 at ‘Edges in Syntax’ in Nicosia and at ‘Interface Legibility at the Edge’ in Bucharest. Many thanks to the audiences of these conferences for their questions and comments, in particular to Carmen Dobrovie-Sorin, Nino Grillo, James Higginbotham, EllenPetra Kester, Tom Roeper and Petra Sleeman. My deepest thanks also go to Werner Abraham for offering to publish the book in the Linguistik Aktuell/Linguistics Today series. I am very grateful for his and for Elly van Gelderen’s productive comments. Finally, I would like to express my gratitude to various non linguists for their friendship. Thanks Annick, Catherine, Christian, Floriane, Gaëlle, Hayley, Lada, Lavinia, Leila, Mariane, Mirka, Mirto, Rosanna, Ruth, Sonia, Trudi, Valérie, Véronique and Youssef ! These past two years have been particularly difficult for me and my family for personal and professional reasons but it has allowed me to experience God’s presence and faithfulness in a very special way. This is why, last but not least, my final words of thanks go to Him. Geneva, December 2007

chapter 1

Introduction 1.  Setting the scene At the origin of this book is what I thought to be a trivial question: ‘What is the structure of a simple, ‘basic’ noun phrase like an apple and its French counterpart une pomme?’ This was the beginning of a long journey which led me not only to syntactic considerations but also to semantic, pragmatic and philosophical issues. Extending the question to the plural forms apples and des pommes complicated matters even further. What I thought to be a simple structure revealed itself to be the tip of an iceberg of problems, ranging from purely structural considerations like the position of the determiner and the status of de ‘of ’ in the plural and mass constructions in French, to interface issues, i.e., the relation between interpretation and syntactic structure, and to semantic notions like presupposition and assertion of existence, mass and count interpretations. Other questions popped up as I proceeded, for example the difference between predicative and argumental noun phrases, the representation of partitivity, the role of semantic types (referential, quantificational and potentially property-denoting noun phrases), the structure of bare nouns, and the role of context in the interpretation of indefinites. In a framework where noun phrases have an extended structure with several discrete functional projections all these problems gained in complexity. In order to investigate such questions, I decided to concentrate on a selection of indefinites, namely indefinite noun phrases with the indefinite article and their plural forms. In other words, this book concentrates on un-NPs in French, assuming that aNPs in English are analogous, and on du/des-NPs which raise additional issues. It also examines a subset of bare nouns in Romance and Germanic, that is, determinerless nominals interpreted as du/des-NPs. Only argumental noun phrases are considered, in opposition to predicative ones, and more specifically subjects and objects. The outcome of this exciting expedition, which took me to unfamiliar places and was a constant and fascinating discovery, is a study on the syntax-semantics interface of indefinites. The formal aspects of semantics have however been kept to the minimum. The challenges met on the road and the purposes of the journey, i.e., the puzzles to be addressed and the aims of the book, are mentioned in the next section. Section 3 briefly presents the proposals developed in each chapter of the book. Sections 4 and 5



The Layered DP: Form and meaning of French indefinites

introduce the reader to the literature which is crucial to the analysis proposed here: the former reports the main points of the theoretical framework adopted and the latter summarises the major studies dealing with the interpretation of indefinites. Enjoy the trip!

2.  The puzzle 2.1  On the interpretation of indefinites One aim of this book is to provide an analysis of argumental un-NPs, du/des-NPs and to a minor extent of bare nouns, which accounts for three of their characteristics, namely their (lack of) existentiality, their (lack of) scope properties and their (lack of) referential/quantificational interpretation. That indefinites may be existential, in the sense that they assert existence, is often taken for granted in the literature. However, contexts in which they do not entail existence are often neglected. Yet such environments concern a representative set of data, including questions and intensional contexts with modals, temporal expressions, verbs like want and seek, words that express propositional attitudes (like know and believe, fear and hate) and contexts of direct and indirect speech. The fact that the indefinites mentioned above may fail to entail existence in these contexts is problematic for accounts which consider that indefinites must be bound by Existential Closure (except when they presuppose existence) (Diesing 1992). A first task thus consists in identifying the indefinites which are always existential, the ones which may be existential or not and the ones which cannot be existential. This means that what is responsible for the existential reading of these indefinites needs to be identified. In other words, the question is whether this interpretation is due to Existential Closure or to different mechanisms. Related to this matter is the distribution of existential indefinites in French, as some of them seem to be restricted to certain environments. Another issue addressed in this book is the potential scope properties of indefinites. Although this topic is widely discussed in the literature, many problems remain (see Section 5 for example). One question which needs to be dealt with is how an indefinite with ‘narrow scope’ can have different interpretations. An illustration would be an example like Every girl talked to a man, where a man may co-vary with every girl, i.e., that for every girl there is a man in particular to which she talked, or where a man does not pick out an entity but simply represents a man in a given set of men. Other questions on the program are: Why do only some indefinites have scope properties, i.e., what allows us to differentiate between indefinites which undergo Quantifier Raising (QR) and those which do not (more in Section 2.3)? Among the indefinites which have scope properties, why are only some of them blocked in islands contrary to others? How is it possible for indefinites which move out of island to target two different



Chapter 1.  Introduction

positions at Logical Form (LF), i.e., what triggers these movements? Why is it possible for some indefinites to bypass others at LF but not for others? Another interpretative difference among indefinites is that some of them can be referential contrary to others (see for example Sections 5.1.2 and 5.2.1). This needs to be explained and an important issue is whether this distinction is binary (Section 5.1) or not (Section 5.2). If it is not a classification of indefinites has to be established. In other words, a typology which accounts best for the various characteristics of the indefinites under discussion is required. Many of the above issues have been (partially) addressed in the literature. For example, Van Geenhoven & McNally (2005) deal with the lack of existential entailment of some indefinites, McNally (1995) with the lack of scope properties of some indefinites, Dobrovie-Sorin & Beyssade (2004) with the categorical difference between indefinites. Although most of them touch upon two of the features mentioned at the beginning of the section or sometimes even the three of them, they do not provide a systematic analysis of all three characteristics. This is one aim of this book. To combine (lack of) existentiality, (lack of) scope properties and (lack of) referential/quantificational interpretation is crucial to obtain a clear and fine-grained picture of indefinites and essential as these features are related.

2.2  On the structure of indefinites Another aim of this book is to determine the internal structure of argumental un-NPs, du/des-NPs and bare nouns and to find out to what extent the interpretative characteristics mentioned in the preceding section and the structure of these indefinites are related. This objective concerns two parts of the structure of the indefinites under discussion, namely their left periphery and their inflectional domain (Section 4.4.2). The issues concerning the left periphery are numerous. Some of them are directly related to Longobardi’s (1994, 2005) distinction between arguments and predicates or to Zamparelli’s (2000) division of DP into several layers in a syntax-semantics mapping spirit on which I build. Others are more general. Concerning the argument-predicate distinction, several points have to be clarified. In the syntactic tradition, properties are NPs and arguments DPs, the intuition being that to be arguments nouns somehow need a determiner (Szabolcsi 1987; Abney 1987; Longobardi 1994, 2005). In a cartographic approach like the one adopted here (Section 4.2), this distinction has to be refined. The reason is that, in an extended structure, it is likely that DP represents a domain of projections, namely the left periphery, and not a single layer. What has to be determined is therefore whether there is one projection in particular which turns predicates into arguments or whether any projection of the left periphery may fulfil this function, and also, whether the layer transforming predicates into arguments may be projected in the structure of any kind





The Layered DP: Form and meaning of French indefinites

of predicates or not. Although these points seem to be minor issues, they contribute to the analysis of indefinites proposed in this book and are crucial to determine the internal structure of un-NPs, du/des-NPs and bare nouns. As for Zamparelli’s (2000) proposal to relate nominal expressions with different interpretations to different projections of DP, it is in essence the line of analysis developed in this book. In other words, I assume that noun phrases with different meanings have different internal structures. However, on the basis of the investigations reported here, I question and reject the binary weak-strong classification of noun phrases discussed in Section 5.1.1 and adopted by Zamparelli. For example, I suggest that definite and quantificational noun phrases on the one hand and property-denoting arguments and other arguments on the other hand should belong to different categories. Such proposals raise many issues. One of them is the number of projections required to distinguish between the various types of indefinites. One logical option is to assume that there is a single projection which is the site of interpretation (Giusti 2005, Section 4.4.2.2) and another one is to postulate different layers for the different interpretations. In the spirit of a cartographic approach (Section 4.2) and on evidence coming from minimality effects (Section 4.3), I argue for the second option. The postulation of a nominal left periphery composed of several functional projections is not trivial. Some issues to be addressed are the following: How much structure, if any, is common to all argumental indefinites? Which aspects of the meaning of indefinites are represented in their structure and how are they activated? Do these projections combine or are they in complementary distribution? Are there some layers which are only projected when needed, i.e., facultative projections? Imagine for example that referentiality is encoded in a specific projection. It is legitimate to wonder whether this projection is present in the structure of non-referential noun phrases such as quantifiers, or whether it is simply empty. The possibility of empty projections also concerns the structure of bare nouns. However, as the interpretation focused on here is close to the one of du/des-NPs, it suggests that their structures should be similar despite the absence of determiner. The mechanism triggering the projection of the various layers advocated also has to be determined. Some type of DP-internal movement has probably to be ­assumed (Section 2.3), although this is not obvious for bare nouns. As for the structure of the inflectional domain, it is essentially discussed in relation to du/des-NPs (Section 2.4).

2.3  What moves when and where If some kind of DP-internal movement takes place, for example to legitimate the projection of certain layers as mentioned at the end of the preceding section, the type of movement involved has to be determined. It could be head-movement, most likely of the article, or phrasal movement, i.e., movement of the subpart of the syntactic tree



Chapter 1.  Introduction

which contains the article (Section 4.4.1.3). Assuming that movement has to be motivated, what triggers movement also has to be investigated. These issues are addressed for un-NPs, du/des-NPs and bare nouns. Furthermore, if indefinites can be quantificational, movement at the level of the clause is also relevant as, under a traditional analysis, quantifiers undergo Quantifier Raising, i.e., QR (May 1977, 1985). In a nutshell, the idea is that quantificational noun phrases adjoin to the clause or the VP, essentially for reasons of scope. In the last decade (Beghelli 1995), the unordered adjunction of quantifiers has been called into question. The idea is rather that there are several kinds of quantifiers which move for feature-checking reasons to projections which are hierarchically ordered. Adopting the broad lines of such an analysis implies that the landing site of indefinites undergoing QR has to be identified. If more than one kind of indefinites undergo QR (Section 5.2.2), this means that they could target different clausal projections for different reasons. In other words, what triggers such movements and what results from these movements has to be determined. The portions of structure which move have also to be identified. It could be the whole noun phrase, or one of its subparts (an XP) or simply the determiner (a head).

2.4  The French du/des ‘of.the’ construction Another aim of this book is to provide a thorough analysis of du/des-NPs. In addition to the issues raised in Sections 2.1, 2.2 and 2.3, I am concerned with the structure of their inflectional domain, with how they relate to partitivity and to EN pronominalisation. The first issue is complex as the features of du/des-NPs encoded in their structure have to be identified and contrasted with other features which could be inferred for example from the context or from (a combination of) other features. The discussion inevitably deals with the mass/count distinction, as du-NPs are mass and des-NPs count, and with the question of the quantities associated with du/des-NPs, as these noun phrases may be quantitative or not. Another issue which is addressed here concerns grammatical number. More precisely what is responsible for nominal expressions to be singular or plural has to be determined. To allow comparison, the sections on these matters, addressed in detail in Borer (2005a), also examine un-NPs, le/les-NPs ‘theNPs’ and nominals with beaucoup ‘many/much’ and peu ‘few/little’. The discussion of beaucoup and peu raises questions about ‘bare’ de-NPs, i.e., de-NPs which do not contain any (overt) quantity. One issue concerning these elements is whether they are analogous to the de-NPs in the constructions beaucoup/peu de NP or not and another one whether they contain an empty category or not. A crucial aspect of the discussion of the inflectional domain of du/des-NPs is to determine the role and position of de ‘of ’ and le/les ‘the’ and to find out when/ where these two elements incorporate. More generally this raises the question of the





The Layered DP: Form and meaning of French indefinites

base-position of determiners. In most studies of noun phrases, DP is not split and articles are generated in D, where they stay (Abney 1987). However, as soon as one assumes that DP represents the left periphery and that it consists of several discrete projections (Section 4.4.2.2), it is very unlikely that the base position of articles corresponds to their surface position. This is even more striking in accounts where determiners are generated low in the inflectional domain as suggested by Borer (2005a). As the sections on the inflectional domain also concern un-NPs, the difference between the numeral un and the indefinite article un has to be addressed. The analysis of du/des-NPs allows the comparison with bare nouns, as both kinds of nominals share several interpretative features. One question which arises is to what extent du/des-NPs and bare nouns can be assimilated. In other terms, what characteristics do they share and are there characteristics which distinguish them? Another puzzle I examine in this book, at least partially, is partitivity. Although du/des-NPs are often called ‘partitive’ constructions, they do not necessarily have a partitive meaning. Some of the questions to be tackled are therefore: What is responsible for the partitive reading of du/des-NPs? Are there several ‘sources’ of partitivity? Does this interpretation give rise to a particular structure? In the affirmative, this structure should be able to accommodate partitives with an overt subset like deux des livres ‘two of the books’. Does the de in partitive du/des-NPs differ from the one in non partitive ones? How does partitivity relate to quantification? The discussion of partitivity will also lead us to briefly study EN pronominalisation as in Jean en a trois (de ces livres) ‘John EN has three (of these books)’. One issue is to determine which (portion of) structure EN may replace.

3.  The proposals Chapter 2 has two main objectives. The first one is to establish a typology of argumental un-NPs in French. The main idea I argue for is that un-NPs are three-way ambiguous in that they can be referential (although this notion has to be defined), quantificational or property-denoting. In other words, they can be of three different semantic types, as proposed by Dobrovie-Sorin & Beyssade (2004). Such a classification allows one to distinguish between indefinites which are always existential, indefinites which may be existential depending on the context they appear in and indefinites which are never existential. The second aim is to determine the internal structure of argumental un-NPs and to find out whether different structures may have an impact at the level of the clause, namely with regard to scope properties. On the basis of empirical facts, I show that the left periphery of each type of un-NPs is composed of different projections. This means that the size of the left periphery varies according to the type of indefinite represented.



Chapter 1.  Introduction

Furthermore, I propose that it is the composition of the left periphery which determines whether indefinites undergo QR or not, i.e., whether they have scope properties or not, and whether they can escape islands or not, i.e., whether they can take so-called ‘widest scope’ for example. Finally, the fact that one type of indefinites may target two positions at LF is attributed to their ‘dependent reference’, labelled ‘speaker’s reference’ or ‘S-reference’. Chapter 3 deals with du/des-NPs and begins with some background information. First, I report the interpretations traditionally attributed to du/des-NPs and then the structural analyses corresponding to the two main readings in a pre-DP framework. This information is primarily aimed at readers who are not familiar with the du/des element. After this introduction to du/des-NPs, I turn to the first objective of the chapter which is to determine whether the extendable telescopic left periphery proposed for un-NPs in Chapter 2 extends to du-NPs and des-NPs. Although it seems to be the case for des-NPs, it is not for du-NPs which cannot be S-referential. The second objective is to determine the structure of the inflectional domain of du/des-NPs, in particular the base position of de ‘of ’ and le/les ‘the’. The hierarchy I advocate is inspired by and overlapping with the one proposed by Borer (2005a). However, the study of du/des-NPs leads me to diverge from her account on several points. The analysis developed in this chapter allows me to account for the mass/count distinction as well as for the quantitative/non-quantitative reading of du/des-NPs, and to differentiate the numeral one from the indefinite article un. Furthermore, it allows the distinction, on the one hand, between partitives which overtly express a subset and partitives which overtly express a set, and, on the other hand, between partitivity and quantification. This means that partitive expressions are not necessarily quantificational, whereas quantificational expressions are necessarily partitive. In Chapter 4, I turn to bare nouns. Although the discussion is more succinct than the one of un-NPs and du/des-NPs, I show that the analysis developed in preceding chapters extends to bare nouns, except for bare count nouns excluded from the study. I propose that non-coordinated bare nouns are property-denoting whereas coordinated bare nouns may be property-denoting or quantificational in that they may have a ­covert partitive meaning. Furthermore, I suggest that the composition of the inflectional domain has an impact on the constitution of the left periphery. The idea is that there may be a correlation between quantity and quantification and another one between the count reading and the S-referential interpretation. Finally, a short section is devoted to incorporation phenomena such as ‘semantic incorporation’. What I tentatively suggest is that the objects involved in incorporation processes have a truncated structure. In other words, incorporation is due to a reduced structure.





The Layered DP: Form and meaning of French indefinites

Chapter 5 concludes this book. It summarises the main proposals and raises new questions which are worth investigating in the light of the conclusions reached. In particular, it examines the unbounded reading generally associated with (some) indefinites and mentions determinerless languages.

4.  Theoretical framework 4.1  Generative Grammar The framework of the book is that of Generative Grammar initiated by Chomsky some fifty years ago (Chomsky 1957, 1965, 1981). It assumes Chomsky’s view that there is a Universal Grammar (UG) which is part of the language faculty characteristic of humans and biologically endowed. UG is described as a system of principles which characterizes the class of possible grammars by specifying how particular grammars are organized (what are the components and their relations), how the different rules of these components are constructed, how they interact, and so on  (Chomsky 1979: 180)

UG is constituted of Principles and Parameters. The former are universal innate properties forming the basis of the specific grammars of all possible human languages, whereas the latter account for language-specific properties, for which UG offers a range of choices, thus accounting for the variation among languages.

4.2  The cartographic approach A development in Generative Grammar adopted here is the so-called ‘cartographic approach’ to syntax (Rizzi 1997, 2001a,b; 2004; Cinque 1999, 2002; Belletti 2004 to mention a few references of the last decade), which investigates the make-up of functional categories in the clause, and by extension in nominals. Essentially, it consists in identifying distinct positions in the structure dedicated to different interpretations. At the foundation of the cartographic research is the idea that inflectional morphemes are distributed in the syntax, a view based on the analysis of the English inflectional system (Chomsky 1957). This means that the atoms of syntactic computations can be morphemes, i.e., elements which are not autonomous words. Assuming that these elements head their own projections (Chomsky 1986) suggests that clauses are articulated and formed of a succession of lexical and functional projections. Furthermore, if, as shown by many studies on Case (Vergnaud 1982 and subsequent work), overt morphological richness is a superficial feature of variation which leads to a fundamental assumption of uniformity, clauses should consist of a system of functional heads which is constant in all languages. In other words, even languages in which the



Chapter 1.  Introduction

morphology does not provide direct evidence for the postulation of independent layers should have a richly articulated clausal structure where functional heads can host autonomous function words, affixes or nothing at all. The cartographic approach to syntax has been profitably applied to various domains of the clause such as the inflectional system (Pollock 1989), the left periphery (Rizzi 1997 and subsequent work), the study of particles (Ramchand & Svenonius 2002) and complex predicate constructions across languages (Kratzer 1996; Pylkkänen 1999; Ramchand 2005, 2006c; see also Borer 2005b in a different framework), to mention but a few. A growing number of analyses have also been proposed for nominal structures (see Section 4.4.2). Below I briefly report a pioneer work on the clausal left periphery, namely Rizzi’s (1997) influential paper, in which he proposes that the complementizer system minimally consists of two projections, ForceP and Fin(iteness)P. The former encodes the illocutionary force, determining whether a sentence is declarative, interrogative, relative… The latter encodes finiteness, determining the type of IP selected. Force and Finiteness split when the field of two additional projections is activated, Topic (TopP) and Focus (FocP). Additional layers like Int(errogative) Phrase and Mod(ifier) Phrase are added to the structure in subsequent works to accommodate indirect interrogative particles and adverbs respectively (Rizzi 2001b, 2002). The advocated structure is given below, where the starred projections are recursive:

(1) Force Top* Int Top* Focus Mod* Top* Fin 

IP

(Rizzi 2002 (60))

What is crucial in (1) is that it brings into the light factors that have been ignored for too long, namely the role of information structure and discourse/pragmatics on word order and word order variation. In addition to discourse/pragmatic information, it has been argued at least since Hale & Keyser (1993) that syntactic structures encode information traditionally assumed to be in the lexicon, such as the argument structure/thematic roles of predicates. The idea is that predicates are complex in the sense that event structures are represented in syntax and that each subevent is spelt out by a different head (Kratzer 1996; Pylkkänen 1999; Borer 2005b; Ramchand 2005, 2006c).1 These heads can be recursively combined and are associated to specific thematic elements. Ramchand (2005, 2006a,b,c), to take one of the most recent accounts, proposes that VP should be decomposed into three projections, initiationP (iP), processP (pP) and resultP (rP),

.  ‘Complex predicates’ is taken here in the narrow sense, i.e., abstracting away from mood, tense and aspect features.



 The Layered DP: Form and meaning of French indefinites

whose specifiers contain arguments with the thematic roles Initiator, Undergoer and Resultee respectively: 

(2) [initiationP Initiator i [processP Undergoer p [resultP Resultee XP]]] (Ramchand 2006b, (9))

Initiators are typically subjects of transitive verbs who bring into being a dynamic event (John destroyed the evidence, The key opened the lock, ibid. p.4). Undergoers are internal arguments that have an intimate connection with the run-time of the event (John pushed the cart, ibid. p.3). Resultees are internal arguments which are necessarily holders of result, i.e., the achievers of a final state (John found gold, ibid. p.3). The sequence of projections in (2) accounts for the internal event complexity and suggests that thematic relations are themselves decomposable. In a lexically synthetic language like English the features represented by the different heads are encoded in a single lexical item. This information, according to Ramchand, is available in the lexicon in the form of syntactic labels.2 This means that a verb like run2 in John ran2 the bathwater or John ran2 Mary’s life for example will contain the information ‘v, V’ or ‘iP pP’ (Ramchand 2006a:16). In other words, the lexicon contains a minimal amount of syntactic information in the form of syntactic tags but nothing directly related to the argument structure. The above discussion not only shows that some information traditionally attributed to the lexicon is encoded in the syntax but also that decomposition into functional projections does not only concern words and morphemes but that it should be even more radical in that a single morpheme may spell out more than one functional head. Similar decomposition into features have been proposed for various parts of the clause (including the CP domain (Bhatt & Yoon 1992) and the structure encoding tense information (Löbner 2002, Starke 2005)) but also for the nominal structure (Vangsnes in progress), which is expected if the structure of nominals is parallel to the one of the clause, at least to some extent (Section 4.4.1.2). This is the direction I will take in this book: the study of un-NPs and du/des-NPs in the light of a decomposition approach (to the exclusion of the NP-layer, which I hope to study in future work). As such a trend of research associates polysemous words with different subtrees, it is directly concerned with the relation between structure and interpretation, i.e., by the syntax-semantics interface. The view of the syntax-semantics interface adopted here is thus the one traditionally found in Extended Standard Theory and subsequent work, according to which syntactic structures cannot give rise to ambiguity. In other words, when the semantics gets to interpret a structure, only one reading is possible. In this vein, it predicts that un-NPs and du/des-NPs with different meanings differ structurally. As I argue that these interpretations correspond to different semantic

.  What is available in the lexicon or not has been a matter of debate for many years. Some scholars for example argue that lexical items possess no syntactic information (Marantz 1997; Borer 1998; Borer 2005a,b).



Chapter 1.  Introduction

types (Dobrovie-Sorin & Beyssade 2004, Chapter 2 Section 2), it predicts that each type has its own structure, which further suggests that there is no type-shifting as an independent operation.

4.3  Relativized Minimality An important principle interacting with the cartographic approach discussed in the previous section is the locality principle known as Relativized Minimality (RM) (Rizzi 1990). The basic idea of RM is that core structural relations are local, despite the unbounded length of syntactic representations. To be local, a relation must be satisfied in the minimal environment in which it can be satisfied. This means that in a configuration like (3), a local structural relation between X and Y is impossible if Z has certain characteristics in common with X and Z intervenes between X and Y. In other words, this minimal configuration is relativized to the nature of the structural relation to be established, hence the term Relativized Minimality.

(3) … X … Z … Y… 

(Rizzi 1990: 1)

To accommodate several apparent exceptions to RM (for example some whphrases with special interpretative properties), (3) has been reformulated in terms of structural type (4):

(4) Y is in a Minimal Configuration (MC) with X iff there is no Z such that (i) Z is of the same structural type as X, and (ii) Z intervenes between X and Y.  (Rizzi 2001: 90 (4))

When the different elements involved in (4) are phrasal, i.e., specifiers, the ‘same structural type’ is defined as specifiers licensed by features of the same class. The typology of features advocated by Rizzi (2002) is reported below. For a justification of this classification the reader is referred to the original paper.

(5) a. b. c. d. 

Argumental: person, number, gender, case Quantificational: Wh, Neg, measure, focus… Modifier: evaluative, epistemic, Neg, frequentative, celerative, measure, manner… Topic (Rizzi 2002 (61))

Crucially, this classification of morphosyntactic features has been provided by a cartographic approach. In other words, cartography and locality go hand in hand. This is expressed by Rizzi as follows: On the one hand, the results of the cartographic study provide a sound theoretical and empirical basis for drawing a typology of positions which the



 The Layered DP: Form and meaning of French indefinites

study of locality can build on. On the other hand, selective locality effects can ­provide evidence for differentiating structural positions, thus providing relevant evidence for the cartographic endeavour (2002, Section 1).

As the classes of features in (5) define elements of the ‘same structural type’, we expect RM effects to arise only within the same feature class, but not across classes. As this book deals with arguments, it is most concerned with the features in (5a) defining A(rgument)-positions. Although the features listed in this class are the traditional Phi features, Rizzi observes that “[i]t is entirely conceivable that this class will require finer distinctions” (2002, Section 9). This is, at least to some extent, what I aim at in this book as I will show that argumental un-NPs and du/des-NPs, i.e., elements belonging to the feature class (5a), are sensitive to RM effects only when they belong to the same type. The RM effects of particular interest to my purpose are intervention effects. Starke (2001) for example proposes that the blocking effects in weak islands are cases of RM. More precisely, he observes that intervention effects only occur in constructions involving elements of the same type (6) or when the intervener is interpretatively heavier than the moving element (7):



(6) a. *αi… αj… αi b. *αβi… αβj… αβi  (7) *α … αβ … α 

(Starke 2001: 8(16a, 17b)) (Starke 2001: 8(15a))

In (7), the movement of α is blocked by the α contained in αβ. Conversely, movement of αβ is not blocked by α because it has ‘something more’, i.e., some additional property (8) .3

(8) αβ … α … αβ 

(Starke 2001: 8(16b))

In the extraction contexts studied by Starke, this means that a wh-element, which is quantificational, i.e., part of the feature class (5b), needs an additional property to escape a potential island. According to Starke, this property is what he calls specificity.

.  Starke notes that opinions diverge on what the ‘something more’ is: θ-role, ie. the distinction is between arguments and non-arguments (Huang (1982); Lasnik & Saito (1984); Rizzi (1988)); case or DPhood (Manzini (1992); Rizzi (2000)); ‘referentiality’, as opposed to purely quantificational readings (Cinque (1990); Rizzi (1990)); d-linking (Cinque (1990); Kroch (1989); Comorovski (1996)); individuation (Frampton (1991); Cresti (1995)), richness of internal semantic structure (Szabolcsi & Zwarts (1997)) (2001: 10).



Chapter 1.  Introduction 

Although his analysis focuses on quantificational elements, it may be generalised to the other classes of features as suggested by the abstract representations in (7) and (8).4 The latter in particular will play an important role in my investigation of un-NPs and du/des-NPs (Chapter 2 and 3). The next section concerns the structure of nominals. It is primarily a reminder of basic assumptions.

4.4  The structure of nominal phrases Since Chomsky (1970) much research on the parallelism between the structure of nominal phrases and clauses has taken place in generative syntax. Abney (1987) for example, in his seminal work, applies Chomsky’s (1986) revised X’-theory to the nominal domain. The main issues he addresses are the facts that functional elements project to the phrasal level and that specifiers host phrasal constituents. Building on earlier work (Szabolcsi 1983; Brame 1982), Abney provides empirical and theoretical arguments for a structure in which the determiner projects its own phrase, the Determiner Phrase (DP), which is a functional projection heading the noun phrase. In other words, DP represents the extended projection (in the sense of Grimshaw 1991) of N, the lexical head. (9) represents the structure proposed by Abney for DPs (1987: 25): DP

(9)

D′

Spec

NP

D Spec

N′ N

The ‘DP-hypothesis’ has been adopted and strengthened by much work in the late 1980’s, especially in comparative studies. This section (Section 4.4) is not a state of the art. It simply reminds the reader of some elementary facts, such as the reasons for which NP should be extended to DP in

.  Starke (2001: 6) observes that the term ‘quantifier’, Q, is inadequate as it does not include items with every, some, most … which are traditionally considered as quantificational. In the absence of a better term, he sticks to this term. This observation is in line with the analysis developed here as the quantificational elements studied belong to the ‘Argumental’ feature class (5a) and not to the ‘Quantificational’ class (5b).

 The Layered DP: Form and meaning of French indefinites

the first place and the relevance of other functional material in nominal structures. Needless to say that the evidence presented here is far from exhaustive, and that the theoretical positions I adopt will be refined as I proceed. The reanalysis of NP as DP is supported by at least three striking arguments. The first one is the fact that determiners can co-occur in various languages (Section 4.4.1.1), the second that DP seems to be parallel to the clausal CP (Section 4.4.1.2) and finally the existence of DP-internal movement (Section 4.4.1.3).

4.4.1  From NP to DP 4.4.1.1  The co-occurrence of determiners.  In various languages, determiners like articles, demonstratives or possessives can co-occur as illustrated in (10), suggesting that these elements sit in different positions. (10) a.  b. c. d. e.

a te valamennyi the you(-nom) each il suo libro the his book ‘his book’ bàiatul acesta boy-the this ‘this book’ *the this book *the his book

titk- od Hungarian secret- poss-2sg (Szabolcsi 1994) Italian

Rumanian

As a ‘the’ and te ‘you’ in (10a), il ‘the’ and suo ‘his’ in (10b) and –ul ‘the’ and acesta ‘this’ in (10c) co-occur without leading to ungrammaticality, they cannot occupy the same slot. Although in English articles cannot co-occur with demonstratives or possessives as (10d–e) show, (10a–c) suggest that several positions are available to host determiners, and thus that the structure of nominal phrases should be extended.

4.4.1.2  The DP-CP parallel.  The theoretical arguments in favour of an extended structure of nominal phrases (such as the X’-theory properties extended to nominals) are supported by compelling empirical facts. Below I briefly mention some syntactic and morphological evidence (for a more detailed survey see Bernstein 2001). Support for a nominal structure similar to the clausal one comes from deverbal nouns which share the same argument structure as the verbs they correspond to.5

.  Grimshaw (1990) calls the class of nouns illustrated in (11b) complex event nominals. She does not consider possessives as arguments.



Chapter 1.  Introduction

(11), from Haegeman & Guéron (1999: 412), is adapted from Chomsky’s (1970) wellknown examples. (11) a. The enemy destroyed the city. b. the enemy’s destruction of the city

In (11), the verb destroy and the noun destruction both take an Agent and a Patient. In the a-sentence, the former is realised as the enemy and the latter as the city. This argument structure is preserved in the b-sentence where the Agent of the noun destruction is the enemy’s, and its Patient the city. This suggests that possessors, on a par with clausal subjects, are likely to be generated in the specifier of the lexical projection and to move to the counterpart of Spec,AgrSP to be licensed. Further evidence for the DP-CP parallel is provided by the domain of morphology. Various languages are characterised by matching clausal and nominal agreement affixes. Abney (1987) for instance mentions Yup’ik, a Central Alaskan Eskimo language, Tzutujil, a Mayan language, and Hungarian, a nominative/accusative language.6 Consider the Hungarian example below (cited in Bernstein 2001: 539(3) and (4) from Abney 1987: 44 and Szabolcsi 1983: 90 respectively): (12) a. b. c.

az én- Ø vendég-e-m the i-nom guest-poss-1sg ‘my guest’ a te-Ø vendég-e-d the you-nom guest-poss-2sg ‘your guest’ (a) Mari-Ø vendég-e-Ø (the) Mary-nom guest-poss-3sg ‘Mary’s guest’

(13) Mari-Ø Mary-nom ‘Mary slept.’

Hungarian

alud-t-Ø sleep-poss-3sg

(12) illustrates the nominal agreement pattern and (13) the verbal one. In (12), the possessors bear nominative case exactly like the subject in (13). Similarly the head nouns in (12) agree in number and person with the possessor, whereas in (13) it is the verb which agrees in number and person with the subject. These facts suggest that the parallel between nominal structures and clauses holds. In her 1994 paper, Szabolcsi

.  Yup’ik is discussed by Abney (1987) on page 39 ff., Mayan languages on p. 42 ff. and Hungarian on p. 44 ff. Abney also discusses English gerundive constructions, which support the strong parallel between sentences and noun phrases.



 The Layered DP: Form and meaning of French indefinites

proposes that in the DP-structure, articles can be analysed as functional heads corresponding to the clausal complementizer, a position advocated here.7

4.4.1.3  Movement.  A third argument in favour of the reanalysis of NP into DP is the existence of DP-internal movement. The traditional analysis of word order alternations in Romance and Germanic is that in the former, contrary to the latter, the noun may bypass (some) modifiers (for Romance languages see Valois 1991; Bernstein 1993; Cinque 1994; Longobardi 1994; Giusti 1994, 1997; for Hebrew see Ritter 1988 and Siloni 1991 and for Scandinavian Delsing 1988, 1993; Taraldsen 1990; Santelmann 1993 and Kester 1993).8 (14a) shows that the noun in English follows all the adjectives whereas in Italian it precedes rossa ‘red’. Noun movement may be partial as in (14b) and (15a) or to D as in (15b). (14) a. Eng: the b. It: la `(15) a. b.

Evaluating Size Color beautiful big red ball bella grande palla rossa

un frumos [baˇiat]i român ti a nice boy Rumanian ‘a nice Rumanian boy’ [baˇiat]i –ul frumos t’i (cel român) ti boy-the nice (the Rumanian) ‘the nice (Rumanian) boy’

(Giusti 1997: 97) Rumanian

In (15a), the presence of the indefinite article un ‘a’- a head – prevents the noun baˇiat ‘boy’ from moving further. In (15b) however, the noun moves to D to bind the definite article –ul ‘the’ which is a suffix. N-raising, argues in favour of an extended NP: different positions are necessary to host the noun. In addition to DP-internal head-movement, DP-internal phrasal movement has been postulated. To my knowledge, Szabolcsi (1983) was the first to postulate this kind of movement. Her idea is that Spec,DP, on a par with Spec,CP, can serve as an escape hatch for extraction from the nominal structure. (16) shows that DP-internal phrasal movement is attested in English and that it is analogous to the clausal wh-movement, thus supporting the parallelism between DP and CP. (16) a. [AP How important] is this decision? b. This is [DP a [AP very important] decision]. c. [DP [AP How important] a decision] is this?  (Haegeman & Guéron 1999: 419)

.  The DP-CP parallel is however far from perfect. See for example Giusti 2005, Section 3 for some differences. .  For a different approach where no head movement is postulated in French see Laenzlinger 2005a.



Chapter 1.  Introduction 

(16a) is an illustration of the well-known auxiliary-inversion: the wh-phrase How ­important is preposed and the auxiliary is has moved to C. In (16c), it is a DP, How important a decision, which is preposed. Inside the preposed DP, the AP has moved to the front of DP, illustrating wh-movement taking place inside DP (compare (16b) and (16c)). This A’-movement is similar to movement to Spec,CP in the clause. If these two A’-movements are analogous, the position occupied by a in (16c) is a complementizer position, like the position of is in (16a). Therefore, the same mechanism accounts for the presence of an AP in Spec,CP in (16a), and in Spec,DP in (16c). Both movements are instantiations of A’-movement, reinforcing the analogy between CP and DP.9 This section argues for an extended structure of nominals, where DP is the nominal counterpart to the clausal CP. The next section looks at the internal architecture of such a structure.

4.4.2  The nominal architecture If DP is parallel to the clausal CP as assumed here, it seems natural to propose that the structure of the former is as rich as the one of the latter. It has been proposed for example, that the noun phrase, like the clause, can be divided into three domains, following the cartographic approach to syntactic structures (Cinque 2002; Belletti 2004; Rizzi 2004) (recall Section 4.2) (17).10,11 (17)





a.

[… [

[adj1 … [adj2 …

[n… [.. ] ] ] ] ] ]

b.

[… [

[adv1 … [adv2 …

[v…[.. ] ] ] ] ] ]

left periphery

inflectional domain

NP/VP-shells

.  A second type of XP movement inside DP has gained importance in the literature recently (see for example Aboh 1998; Bhattacharya 1998; Koopman 1999; Laenzlinger 2005a; Scott 1998 to mention only a few). It is the so-called snowballing, which consists of successive movement of bigger chunks. Snowballing proceeds by pied-piping of bigger phrases, i.e., it necessitates an extended structure of nominal phrases. .  The tripartite conception represented in (17) is not new. For example, it parallels the one found in Platzack 2001 and Grohmann 2003 for the clause and Grohmann & Haegeman 2003 and Haegeman 2004 for the nominal structure. .  The domains in (17) do not correspond to the notion of phase introduced by Chomsky (2000) and assumed in subsequent works (2001, 2004, 2005, 2006). The concept of phase has

 The Layered DP: Form and meaning of French indefinites

The NP/VP-shells represent the domain where the arguments of the noun or verb are merged, i.e., the thematic domain (Larson 1988; Chomsky 1995; Grimshaw 1990; Valois 1991). The inflectional domain is the domain where agreement, phi-features and case are checked. It also hosts modifiers, such as adverbs in the clause and adjectives in the noun phrase. Finally, the left periphery subsumes notions like topic and focus. In the nominal structure, it is often associated to features related to the D head such as (in)definiteness, specificity or referentiality (see Rizzi 1997 for the clause and Aboh 2003 for nominals to mention but two references). Assuming a syntax-semantics mapping, the role of information structure and discourse pragmatic will be crucial to determine the structure of indefinites, one of the main issues of this book. Each of the domains in (17) represents several discrete functional projections. The exploration of an expanded INFL(ectional) system, the clausal inflectional domain, started with Pollock’s (1989) seminal article. In the same vein, many papers have grown out of Rizzi’s (1997) paper and subsequent work in which CP, the clausal left periphery, has been replaced by a variety of functional projections. Similarly, the study of nominal phrases has led to the postulation of a proliferation of projections in each of the three domains mentioned above. However, in the next two sections, I will only mention the portions of structure which will play a role in this book. Section 4.4.2.1 deals with the inflectional domain. It concentrates on the layer related to number, on the analysis of modifiers, which include numerals, and on a projection differentiating count from mass nouns. Section 4.4.2.2 focuses on the left periphery and reports different proposals arguing for a split left periphery.

4.4.2.1  The inflectional domain.  One projection widely adopted in the literature is Ritter’s (1991) Number Phrase (NumP). Ritter is the first to propose that the noun’s singular/plural marking corresponds to a functional projection. The idea is that the complement of D is NumP and not np. Her analysis examines two ways of expressing genitive in Modern Hebrew, the construct state and the free state noun phrase. In construct state genitives, the word order consists of a head noun preceding its possessor (18a). On the basis of binding facts, Ritter proposes that this order is derived and that the head noun moves from N to D (19), thus accounting for its incompatibility with the definite article ha ‘the’ (18b). been proposed to refine the Minimalist Program (1993, 1995). The idea is that syntactic derivations take place in incremental chunks, labelled phases. ‘Little’ v and C are phases whereas V and T are not. That D might be a phase has been suggested by Chomsky 2005. Each phase is built from a different lexical sub-array. The complement of the head of the phase, i.e., v, C and possibly D, is transmitted to the interfaces once the phase is completed. This means that operations outside the phase do not have access to this chunk anymore. However, several accounts suggest that phases may be extended (Gallego 2006, Den Dikken to appear).



Chapter 1.  Introduction 

(18) a. b. 

axilat dan et ha-tapuax Hebrew eating Dan acc the-apple ‘Dan’s eating of the apple’ *ha-axila dan et ha-tapuax the-eating Dan acc the-apple (Epstein 2001: 3, adapted from Ritter 1991: 39)

(19)

DP D′

Spec D

NP Subj. Dan

N′ N axilat

Obj. ha-tapuax

In the free genitive constructions involving wel ‘of ’, in contrast, the definite article can precede the head noun (20a). Assuming that ha ‘the’ sits in D, there must be a position between D and N which serves as a landing site for the head noun (20b). (20) a. ha-axila wel dan et ha tapuax ‘Dan’s eating of the apple’ b.

(Ritter 1991: 43(13))

DP Det ha

NUMP NUM

NP Subj ?el dan

N′ N axila

Obj et ha-tapuax

NumP is thus in part motivated to account for the different derived word orders in genitive constructions in Modern Hebrew. On the basis of the order of adjectives relative to the head noun and its arguments, Ritter argues that all Hebrew noun phrases contain a NumP including construct state constructions illustrated in (18a). For the

 The Layered DP: Form and meaning of French indefinites

details of the argument, the reader is referred to the original work. As for Romance languages, Walloon provides independent evidence for a NumP (Bernstein 1991, 1993). Note that the location of NumP in the structure depends on the theoretical positions adopted. If inflectional morphology must be represented structurally and if NumP is the projection hosting plural affixes such as –s in a language like English, then NumP is in the inflectional domain. Alternatively, NumP could be higher in the structure. Aboh (2001) in particular proposes that it is part of the left periphery. He considers that NumP represents the lowest projection of the informational domain (see also Giusti 2006), which is discussed in the next section. In a different framework, NumP might have the function of a counter. This has been proposed by Borer (2005a) who labels the projection corresponding to the traditional NumP, #P. To understand the role of this projection, I will briefly summarise the relevant parts of her analysis. Needless to say that it will be impossible to do justice, in a few paragraphs, to the very detailed and far-reaching theoretical framework she develops and applies to a wide range of typologically different languages. Fundamentally, the significant shift proposed in Borer’s work explains linguistic competence only from the syntactic computation, thus shifting the load from the lexicon of a language. As a consequence, any variation is to be accounted for by differences in the functional structure. In other words, all aspects of the computation emerge from properties of the structure, rather than properties of (substantive) listemes, which are pairings of sound and meaning giving an encyclopaedic knowledge. Such listemes are in opposition to the functional lexicon which includes functional vocabulary like grammatical formatives, affixation and functional structure. As for the structure of nominals, Borer proposes that DPs may be composed by a classifier head (Classifier Phrase), a number head (#/Number Phrase) and a determiner head (Determiner Phrase).12 Crucially, these functional nodes are under-specified (〈e〉). This means that they have open values which are assigned range by a variety of means like free morphs, abstract head features (direct range assignment) and other operators such as independent morphemes like quantifiers and discourse operators (responsible for indirect range assignment). The structure advocated is illustrated below (Borer 2005a:109(27)).

.  Such a structure has been proposed by Peyraube (1998) and Cheng & Sybesma (1999) for Chinese and by Roger-Yun (2002) for Korean. It is also adopted by Guéron (2006).



Chapter 1.  Introduction 

(21)

DP D 〈e〉d

#P

three 〈e〉#

CL cat.〈div〉 〈e〉DIV

L→N cat

The open values 〈e〉d, 〈e〉# and 〈e〉DIV in (21) must be assigned range. The subscripts indicate the kind of range assigner required for each open value. For example, the free standing definite article in English may assign range to 〈e〉d or the abstract head features 〈pl〉 may assign range to 〈e〉DIV. The system does not predict a one-to-one correspondence between these open values and range assigners because it is theoretically possible for an operator to bind more than one variable. However, the reverse is not possible as double marking of one open value by two operators is not allowed as natural languages do not allow a double marking. This means that although a functional head cannot be assigned range twice, a particular morpheme can assign range to more than one functional node. Below I concentrate on the Classifier Phrase and the Number Phrase as they belong to the part of structure I call inflectional domain, in contrast with the determiner head which has a discourse function and which is part of what I call the left periphery. A fundamental point of Borer’s work is that listemes are mass by default. To be counted, they need to be portioned out, divided into countable units via classifiers or plural/ singular functional elements. Put differently, it means that the definite and indefinite articles in English, which are singular, and the plural –s can assign range to the head of ClP, 〈e〉DIV. This means that the classifier head has the role of a divider. Crucially, Borer argues that plural is not a function from singulars to pluralities. This means that, in a language like English, plurality is a divisional function on mass, as the classifier system of Chinese is, and not a quantity specification or a number specification. Such a proposal leads to a unified analysis in which plural morphology and classifiers compete for the same position. Plural is thus a stuff divider, much like classifiers are to be treated as dividers. Once the head of ClP receives a value, it can be the domain of a restrictor which assigns some specific quantity to the mass divisions created by the plural marking.

 The Layered DP: Form and meaning of French indefinites

This function of counting is performed by #P. Singulars are special in that the dividing function and the counting function are semantically identical. In English, for example, in order to have a singular reading, a determiner which realizes both the Classifier and Number function must be present. For example, the singular interpretation is derived in English by the indefinite singular article a, which is the realisation of a set of head features associated with both Cl and #. The identity of function between division and quantity for singulars forces the same element to assign range to both Cl and #. Thus singular is always interpreted as ‘one’. In addition to a Number Phrase and a Classifier Phrase, the inflectional domain contains projections hosting noun modifiers such as adjectives, numerals and demonstratives (Hawkins 1983; Greenberg 1966). A note on noun modifiers is relevant to the topic of this book in that the indefinite article in many languages, including French, and the numeral 1 are identical (French un, German ein, Turkish bir). When un is a numeral it should be considered as an element of the Nral class discussed below (the difference between the indefinite article and the numeral 1 is addressed in Chapter 3 Section 3.2.2). In Chapter 3 Section 3.3, I will also argue that the partitive PP is a kind of modifier. In line with the cartographic approach discussed in Section 4.2, I consider that noun modifiers are generated as specifiers of functional projections and that they head their own projection (Cinque 1994), on a par with clausal adverbs (Cinque 2002). Consider (22). (22)

DP

D′

D

… FP DemP

Spec

F′ Dem′ F

Dem

FP … NralP

…

Nral′

Spec Nral

…



Chapter 1.  Introduction 

Although it has been proposed in the literature that these functional projections are agreement projections (Giusti 1994), I remain vague with respect to their categorial label and use FP for Functional Projection. Concerning the order of noun modifiers, Hawkins (1983) concludes on the basis of the study of over 350 typologically different languages and Greenberg’s (1966) Universals that they can occur in two major patterns (see also Cinque 2005): (23) a. b.

3 modifiers on the left/0 on the right: Dem – Nral – Adj – N 0 modifier on the left/3 on the right: N – Adj – Nral – Dem

Two additional sequences are less common: (24) a. b.

2 modifiers on the left/1 on the right: Dem - Nral - N - Adj 1 modifier on the left/2 on the right: Dem - N - Adj - Nral or Nral - N - Adj - Dem

Greenberg and Hawkins also observe that when demonstratives, numerals and adjectives precede the noun, they always occur in that order and that when they follow the noun “the most frequent order is the mirror-image of the order for preceding modifiers” (Hawkins 1983: 121). Assuming with Kayne (1994) that languages are universally of the type specifierhead-complement, the above observations lead to the conclusion that there is only one base order, DEM-NRAL-ADJ-N (23a). The three other sequences, (23b) and the two less frequent patterns (24), can be accounted for by head or phrasal movement to the left.13 The universal word order postulated for noun modifiers (23a) is repeated as (25). (25) Universal Base Order Demonstrative > Numeral > Adjective > Noun  (Hawkins 1983; Kayne 1994; Cinque 1994, 1996; Aboh 1998…)

(25) implies that sequences that differ from the Universal Base Order adopted must be derived by movement. As this issue is not directly related to the topic of this book, it will not be further discussed. The next section deals with the nominal left periphery.

4.4.2.2  The left periphery.  If DP is to the noun phrase what CP is to the clause, at least to some extent, we expect it to be split into discrete functional projections representing the left periphery of nominals. In other words, the former DP can be replaced by several projections encoding information external to the nominal

.  In his Antisymmetry program, Kayne (1994) tries to reduce word variations to the application of the operation Move α, independently required by the system.

 The Layered DP: Form and meaning of French indefinites

per se.14 Typically, the nominal left periphery is concerned with concepts like topicalisation, focalisation, specificity and referentiality. At the time of writing, there is no consensus in the literature on the structure of the nominal left periphery. The three analyses I will present are Aboh’s (2001), Laenzlinger’s (2005a) and Giusti’s (2005). Aboh (2001) discusses Gungbe, a Gbe language, where the determiner obligatorily follows the noun. The determiners he examines are the markers of specificity, lf¢/2é, and the markers of plurality, lε (26a–b). Specificity is taken in the sense that the entity has been previously established in the discourse (Pesetsky 1987, Enç 1991, Campbell 1996) (see Sections 5.1.3 and 5.1.5 for a more detailed definition of the term specificity). The markers of specificity and of plurality realise the features [± specific] and [± plural] respectively and can co-occur in the nominal structure, in which case the former necessarily precedes the latter (26c–d). (26) a. b. c. d.

távò lf¢ table Det ‘the [specific] table’ távò lε table Num ‘tables’ távò xÓxÓ lf¢ lε table old Det Num ‘the [specific] old tables’ távò xÓxÓ 2é lε table old Det Num ‘some [specific] old tables’

Gungbe

(Aboh 2001: 2–3(2a,7a–c))

Aboh proposes that the markers of specificity and of plurality represent two distinct projections of the nominal left periphery, DP and NumP respectively. The former is the highest projection of the D-system whereas the latter closes it. DP thus represents the boundary between the noun phrase and the discourse and NumP the boundary between the left periphery and the inflectional system of the noun phrase. The structure advocated is reported in (27), where ΣP represents the extended projection of N, i.e., the nominal inflectional system: (27) a. [DP [D lf¢/2é [NumP [Num lε [ΣP modifiers > Noun]]]]]  (adapted from Aboh 2001: 3(9))

.  An alternative is to consider DP as the highest projection of the nominal phrase, and to assume that it can be dominated by additional material, such as a Quantifier Phrase (Shlonsky 1991; Cardinaletti & Giusti 1992; Cardinaletti & Giusti 2002, etc). Such projections are thus ‘outside’ the nominal structure proper.



Chapter 1.  Introduction 

b.

DP D′

Spec D [±specific]

NumP

Spec

Num′

Num [±plural]

∑P



(Aboh 2001: 4(121))

Aboh argues that ΣP moves to Spec,DP through Spec,NumP to check the features of specificity and plurality, an analysis which also accounts for the word order in Gungbe. In his analysis, Laenzlinger (2005a) proposes that the highest projection of the DP-system encodes deixis and that the lowest one expresses determination. Furthermore, he suggests that the determiner head-moves from the lower D to the higher one and that NP targets the specifier of the lowest DP, thus accounting for the post-nominal position of the adjective: (28) a. la voiture rouge the car red b.

French

DPdeixis (=external) D1 la

DPdetermination (=internal) D2 + 15 D2

FPadj rouge





NP

voiture

(Laenzlinger 2005b: 5(10a, 11a))

.  The symbol + is another notation for the traditional ‘bar’ in the X’ -schema.

 The Layered DP: Form and meaning of French indefinites

Laenzlinger argues that in French the left periphery hosts pre-nominal adjectives. According to him, adjectives can only be pre-nominal in three situations, namely, when they have a weak/short form (29a), when they are focused or have a strong subjective reading (29b), or when they are quantificational (29c). (29) a. b. c.

de gros_avions  big airplanes de superbes créatures wonderful creatures de nombreux accidents numerous accidents

French

(Laenzlinger 2005a: 667(41))

The occurrence of such adjectives in the left periphery causes the splitting of the two DP projections attested in (28b). The idea is that weak adjectives move to a projection labelled WeakP, focussed/subjective ones to a SubjP and quantificational ones to a QuantP. In addition, the left periphery contains a functional projection attracting NP for agreement feature checking. It is labelled FPAgr(NP) and might be related to Number and Gender (2005a:665). The resulting structure is the following: (30) DPdeixis > QuantP > SubjP > WeakP > FPAgr(NP) > DPdetermination

(30) is the structure of the nominal left periphery proposed by Laenzlinger. As the analysis is mainly concerned with adjectives, which are out of the scope of this book, I will not discuss it any further. Another possibility is to hypothesise that the nominal left periphery is parallel to the clausal one. This line of analysis is pursued in Ihsane & Puskás (2001) and Giusti (2005) for example. The hierarchy postulated in the former is reported in (31) and the one proposed in the latter in (32). (31) DP > TopP > FocP > DefP (32) DP > TopP* > FocP > TopP* > dP

(Giusti 2005: (29b))

In (31), DefP stands for Definite Phrase, FocP for Focus Phrase and TopP for Topic Phrase. The label DP, for Determiner Phrase, is provisionally maintained although this projection might correspond to a Case Phrase (KP) (see for example Giusti 1995) or a projection related to referentiality. DP can serve as an escape hatch for different constituents such as possessives in Hungarian and corresponds to the clausal ForceP. DefP, which determines whether a noun phrase is definite or not, corresponds to the clausal Fin(iteness)P, which determines whether a sentence is finite or not (recall Section 4.2). In a nutshell, this means that DefP hosts the definite article in languages that attest one, FocP elements marked by contrastive stress and TopP elements that are specific, in the sense that they have been pre-established in the discourse (such an analysis will



Chapter 1.  Introduction 

however be challenged in the next chapters).16 What about Giusti’s structure, based on Rumanian? In (32), DP represents the highest functional head of the DP area and small d the lowest one. The former corresponds to the realisation of Case features assigned to the noun phrase from the outside (Giusti 1993 and subsequent work). The latter, also referred to as defP, is parallel to the clausal FinP (see Ihsane & Puskás 2001 for a similar proposal). Giusti’s approach however radically differs from other split-DP analyses in that only the topmost projection is the site of interpretation. The possible meanings of a noun phrase are due to covert or overt operators in the specifer position of the higest projection, DP, also labelled KP: (33)

KP (=DP = FPmax) K′

Spec OP



K [Case]

…



(Giusti 2005: (32))

Potential covert operators in Spec;KP are the [+ definite] and the [− definite] OPs. According to Giusti, these operators determine the interpretation of the article sitting in K. In other words, the article does not realise semantic features like definiteness or referentiality but is a morphological realisation of Case (Giusti 1993, 1995). An example of an overt operator would be the demonstrative whose presence in Spec;KP, at least in Rumanian, prevents the head K from hosting a determiner. Another aspect of the analysis of ‘simple’ definite and indefinite noun phrases is that they involve only one projection of the left periphery, namely DP/dP (or KP/ defP). In effect, as these nominals do not activate the Topic/Focus field, DP/KP and dP/defP do not split. As for TopP and FocP in (32), they are used differently from TopP and FocP in other split DP analyses, including (31) (see also Giusti 2006).17 Contrary to previous works on split DP, which correlate TopP and FocP with Topic/Focus interpretation

.  The Focus Phrase can also host stressed possessives. Such an idea is alluded to in Ihsane (2003a). .  Giusti 2006 is a refined proposal of Giusti 2005 in which topic and focus features in noun phrases are substituted with a Topic position combined with a Contrast feature, in the spirit of Molnár 2002. As a result, the structure of the split DP is: DP > KonP > dP (2006, (36)), without any FocP.

 The Layered DP: Form and meaning of French indefinites

assigned to the noun phrase in the clause, Giusti proposes that the Topic/Focus distinction inside the noun phrase is independent of the informational structure of the clause.18 This interpretative domain is created by DP-internal movement resulting in a marked order of the nominal constituents. FocP may host postnominal demonstratives (cf. also Ihsane 2003b) and vague numerals and TopP some marked adjectives. Giusti further suggests that functional features may bundle together. For example, the topic and focus features may be merged in the same head as Case and Definiteness respectively. In this case, DP splits into a K/TopP and a Foc/defP. A recurrent feature of split DP analyses is that they generate the indefinite (and definite) article in the left periphery. However, if indefinite articles are derived from numerals as generally assumed for languages like French, the relation between the indefinite article and the numeral should be acknowledged. One possibility would be to assume that un is generated lower than NumP on a par with other numerals (Section 4.4.2.1) and that it ‘becomes’ an article when it moves to D.19 If, in contrast the numeral un is generated below numerals in the same projection as the indefinite article un as suggested by Borer (2005a), what distinguishes these two elements has to be made explicit (see Chapter 3 Section 3.2.2 for a proposal). In the next section I turn to the interpretation of noun phrases and more precisely the interpretation of indefinites, which are at the centre of this book.

5.  The Data The syntax and interpretation of noun phrases are tightly intertwined. Section 4.4 dealt with their structure in general terms, in the sense that I did not examine one kind of noun phrase in particular. In this section, I focus on indefinites and more specifically on their interpretative aspects. As (in)definiteness has been a central issue for linguists and philosophers at least since Russell’s (1905) treatment of definite descriptions and a much debated topic in semantic theories of noun phrases (Carlson 1977b; Barwise & Cooper 1981; Heim 1982; Partee 1986; Ludlow & Neale 1991…), it is impossible to provide a comprehensive overview of the literature. What I will do is review a selection of papers on the various interpretations attributed to indefinites and report the

.  Giusti mentions the following papers (2005, fn. 3): Bernstein (2001) for English and Romance, Ihsane & Puskás (2001) for French and Hungarian, Ihsane (2003b) for Arabic, Aboh (2004) for Gungbe, and Svenonius (2004) on the more theoretical notion of “edge”. .  As numerals are generally considered as phrasal constituents and articles as heads, this would imply that the head of the numeral XP extracts to move to the left periphery. Such a transformation from XP to X is standardly assumed for clitics.



Chapter 1.  Introduction 

argumentation behind the different analyses.20 I will also compare the terminologies used by the authors. Section 5.1 reports several two-way ambiguities and Section 5.2 two three-way ambiguities. As each author of the papers discussed in the former section use their own labels to classify indefinites, it is not only confusing but difficult to compare the typologies. To anticipate, I provide the reader with a table of the different terminologies and classifications of determiners used in each paper (34) (Table 1). In Section 5.1.5 I attempt a classification by sentence type. (34) Table 1. Author

Labels

Determiners

Milsark

Strong (quantificational)

Universal quantifiers: the, each, all, every, both Non-universal quantifier: most Few, some/sm, many, several Numerals Every, all, each, most, few, no Some, several, many, a, numerals

Strong or Weak (cardinal) Fodor & Sag Enç Diesing

Quantificational Quantificational or Referential Specific Specific or Non-specific Presuppositional Presuppositional or Non-presuppositional

Definite NPs, quantifiers Relational specifics: a certain N Several, many, few, numerals Milsark’s strong determiners Milsark’s weak determiners

Let us turn to Milsark’s (1974, 1977) well-known work.

5.1  Two-way ambiguities 5.1.1  Milsark Milsark (1974, 1977) uses the diagnostic of the so-called Definiteness-effect illustrated in (35) to classify noun phrases in general. (35) a. There is a wolf at the door. b. * There is the wolf at the door.  (Milsark 1977: 5, quoted in Abbott 2004 (23))21

.  The ambiguity due to the indefinite/numeral reading of un ‘a/one’ in French will be addressed in Chapter 3 Section 3.2.2. .  Abbott (2004, fn. 9) notes that this example is infelicitous rather than ungrammatical. The asterisk was customary at the time Milsark wrote his paper.

 The Layered DP: Form and meaning of French indefinites

The above examples involving there be are known as existential sentences.22 In the early days of transformational grammar, the ungrammaticality of the b-sentence was attributed to an unidentified problem with definite noun phrases in such contexts. However, to classify NPs, Milsark abandons the traditional terms ‘indefinite’ and ‘definite’ because they cannot be extended to all determiners. He coins the terms weak and strong for noun phrases that are grammatical and ungrammatical, respectively, in existential sentences: (36) a. There are some/several/many/few wolves at the door. b. * There are most/all/those/Betty’s wolves at the door. 

(Abbott 2004 (24))

A list of weak and strong determiners is provided in (37): (37) a. a/an, some, several, many, numerals (one, two, three) b. definite descriptions, demonstratives, possessives, personal pronouns, generics, universal quantifiers (all, every, each), most

To account for the Definiteness-effect illustrated in (35) and (36), Milsark puts forward an analysis in terms of semantic incompatibility between the existential structure and the definite complement. He proposes that strong noun phrases (37b) are quantificational, and that such quantificational elements are incompatible with the existential quantifier expressed by there be.23 That there be is an existential quantifier is based on examples such as (38), where thr represents the weak form of there: (38) a. Thr were three men believed to have been in the garden.  (Milsark 1974: 175 (25a)) b. Three men were believed to have been in the garden.  (ibid, (24a))

According to Milsark (1974: 175ff.), the truth of (38a), involving there be, depends on the existence of three men, contrary to the truth of (38b). The existence of the entity depicted by the complement noun phrase, three men in (38a), is thus asserted by there be. Contrary to strong-quantificational noun phrases, weak noun phrases are compatible

.  There is another type of existential sentences, involving verbs introducing a new entity into the discourse. Lyons gives the following examples: appear, occur, emerge, arise, ensue, arrive, come, follow, and remain (1999: 237). .  Strong determiners and there, which denotes an unrestricted existential quantifier, are of type 〈et,t〉. Hence a type mismatch: there-sentences are uninterpretable when they contain strong determiners. In their predicative use, weak noun phrases are of type 〈e,t〉, and can therefore occur in there-contexts. To put it differently, the existential quantifier there be requires a variable to bind. Strong noun phrases cannot provide such a variable, being themselves quantificational. Weak noun phrases, on the other hand can.



Chapter 1.  Introduction 

with the existential quantifier because in there be contexts they are predicative. What they involve is cardinality and not quantification. The analysis is slightly complicated by the fact that weak determiners have two distinct uses. As weak determiners are not limited to existential constructions such as there be contexts, they must have a second meaning besides the predicative one. Thus, certain cardinality determiners (some/sm, many for instance) can also be used quantificationally, in which case they express a vague proportion of a given set and not a quantity: (39) About a dozen students and three or four lecturers turned up for the discussion. Some students wanted to move to a smaller room. (Lyons 1999: 242)

In (39), some students is a subset of the students mentioned in the preceding sentence. In other words, we are dealing with a partitive relationship between two sets and not with the cardinality of a single set. Consequently some students in (39) is strong. This means that some weak determiners can have a strong and a weak reading. Put differently, these weak noun phrases are ambiguous, which is the point in which I am interested in here.24 What Milsark attributes to the strength (strong vs. weak) of a noun phrase is identified as presuppositionality (or properness) by De Jong & Verkuyl (1985). The appeal of such an approach is that it accounts nicely for the Definiteness-effect in existential sentences. Assuming that such contexts typically assert existence, the ungrammaticality of a strong noun phrase or of a weak one on its strong reading is attributed to their presupposition of existence which conflicts with the assertion of existence specific to the there be construction. The distinction between assertion and presupposition of existence is in fact Frege’s position (1892), later rejected by Russell (1905) (see also Chapter 2 Section 2.2.3). To summarise, Milsark classifies determiners into three groups with respect to the cardinal-quantificational distinction (cf. Table 2):

.  Another characteristic of weak determiners is that, in their weak reading, they are ungrammatical with predications illustrating permanent properties – individual level properties in the sense of Carlson (1977b) (i). In their strong reading however, weak determiners are fine with individual level predicates (ii):



(i) *Sm salesmen are intelligent. (ii) Some (of the) salesmen (but not others) are intelligent.

(Abbott 2004 (27))

To account for the grammaticality of weak determiners like three, many, several with individual level predicates like intelligent (iii), Milsark claims that the cardinal determiners are used quantificationally: (iii) Many salesmen are intelligent.

 The Layered DP: Form and meaning of French indefinites

(40) Table 2 Determiners Only quantificational Ambiguous quantificational-cardinal Only cardinal

- universal quantifiers: the, each, all, every, both - non-universal quantifier: most - f ew, some/sm, many, several perhaps numerals (cf. fn. 24) -a

That the indefinite article a is always a cardinal determiner is not incontestable though. Consider (41): (41) I didn’t buy the house, because a window was broken.

(Lyons 1999: 264)

As a house normally has more than one window, the indefinite in (41) has a partitive reading similar to the one in (39), suggesting that a(n)-phrases can also be quantificational, and therefore that the indefinite article should be classified as ambiguous. As a consequence, there would only be two types of determiners, the ones that are always quantificational and the ones which are ambiguous between a quantificational and a cardinal reading (Lyons 1999: 266). If the case of a can be easily solved, there is a more general problem with Milsark’s classification. His tests for strength, illustrated in (42a,b), predict that proper names and pronouns are strong and therefore that they should be considered as quantificational, which is counterintuitive (Enç 1991: 12, Abbott 2004: 15). Although proper names are intuitively definite, they are traditionally treated as logical constants. (42) a. *There is John/he at the door. b. John/he is intelligent. c. *Every cat is intelligent.

In (42a–b) John and he are banned from the existential sentence but perfectly fine with an individual level predicate. As the latter context does not admit quantificational noun phrases (42c), it suggests that what Milsark calls strong determiners cannot be characterised in terms of quantification.

5.1.2  Fodor and Sag Fodor & Sag (1982) argue that indefinite noun phrases are semantically ambiguous, even in contexts which do not exhibit any scope ambiguity.25 Consider (43):26

.  For simplicity, Fodor & Sag (1982) only discuss a, the indefinite singular article. But their analysis extends to numerical determiners, some, several and many. Every, all, each, most, few and no on the other hand are considered as always quantificational (p. 476). .  Most examples in this section come from Fodor & Sag (1982), in which case the number they correspond to in the original paper is indicated in parenthesis.



Chapter 1.  Introduction 

(43) A student in the syntax class cheated on the final exam. (1)

The above example contains no modal, no negation, no opaque verb and no other quantifier which could interact with the indefinite noun phrase a student. Nevertheless, Fodor and Sag claim that such examples are ambiguous between a referential and a quantificational reading. The former treats a student on a par with proper names or demonstrative phrases, whereas the latter assimilates it to expressions such as few students or each student. As a consequence, sentences like (44) have three interpretations, one on which a student is referential, one on which it has narrow scope (every > a) and one on which it takes wide scope (a > every). (44) a. Every professor met a student in the syntax class. (2) b. Jones believes that a student in the syntax class was cheating. (3)

Fodor & Sag (1982) note that the distinction referential-quantificational is not new, citing Chastain (1975) and Wilson (1978). What Fodor and Sag intend to do in their analysis is to show that there is a genuine semantic ambiguity in such examples, in opposition to a pragmatic distinction. That (43) has two uses in discourse is a familiar observation. These two uses are determined by the speaker’s intention. On the one hand, the intended assertion might be that some specific student in the syntax class, who is not identified by the speaker, cheated on the final exam. On the other hand, the intended assertion might simply be that the set of students in the syntax class who cheated on the final exam is not empty. According to Fodor and Sag, the ambiguity they mention is not associated with distinct syntactic structures. It is a lexical ambiguity, analogous to the ambiguity of bank in a sentence like John is waiting near the bank, and not to a structural ambiguity as in three triangles and squares (1982: 477, sentential examples in italics ours). To support their analysis, Fodor and Sag resort to multi-clause constructions involving islands and VP Deletion, because such contexts are problematic for a theory which considers that any ambiguities in indefinite noun phrases are merely scope ambiguities characteristic of all quantificational phrases. For such a theory, Fodor and Sag use the acronym QIO, for Quantifier-Interpretation-Only theory. A first argument in favour of Fodor and Sag’s analysis comes from some differences between indefinites and the familiar examples of quantificational phrases, and from so-called scope islands. Indefinites have scope properties that are unexpected for a quantificational phrase. Contrary to quantifiers involving each, every or few, indefinites favour wider scope readings. In general, in a clause containing two quantificational noun phrases, it is the first quantifier in the syntactic structure which takes wide scope over the second one (unless the latter is less embedded that the former).27

.  As observed by Fodor & Sag (1982: 486) only for each is the general trend outlined in the text reversed. Each has a predilection for wide scope.

 The Layered DP: Form and meaning of French indefinites

This means that in (45a) the preferred reading is the one where every has wide scope over many, while in (45b), it is the one where many has wider scope. (45) a. Every actor in our company admires many producers. (39) every > many b. Many producers admire every actor in our company. (40) many > every

Contrary to the every-quantifier in (45b), but like each-phrases in (46), a(n)-phrases tend to take wide scope (47). In (47), an actor can very naturally have scope over many producers. (46) Many producers admire each actor in our company. (41) each > many (47) Many producers admire an actor in our company. (42) a > many

Although (46) and (47) suggest that a is similar to each, i.e., a quantifier favouring wide scope, the analogy breaks down when negation comes into play. Consider the following examples: (48)

John does not respect a friend of mine from Texas. (44)

(49)

Scarcely anyone respects every/each friend of mine from Texas. (47)

In (48), a friend of mine is interpreted outside the scope of the negation not, whereas in (49) the every/each-phrase is more naturally interpreted within the scope of the negative element scarcely anyone. This is unexpected if a(n)-phrases were of the same kind as each-phrases.28 To strengthen the above observations, Fodor and Sag turn to scope islands. Scope islands are closely correlated with syntactic extraction islands. In other words, complements of a noun, relative clauses and unextraposed sentential subjects are considered as scope islands. In (50), for example, each cannot take scope over the highest clause, despite its wide scope preference. (50) a. John overheard the rumor that each of my students had been called before the dean. (56) b. John thinks that for each of my students to be called before the dean would be preposterous.

Contrary to each, whose scope is constrained to islands, a can have scope on the matrix predicate as in (51), which can clearly be about a specific student of the speaker.

.  The scope properties of indefinite noun phrases discussed by Fodor and Sag vary according to the descriptive richness of the noun phrase (p.486 ff.). For more details the reader is referred to the paper.



Chapter 1.  Introduction 

(51) a. John overheard the rumor that a student of mine had been called before the dean. (58) b. John thinks that for a student I know to be called before the dean would be preposterous. (59)

If a were a usual quantifier, the difference with each mentioned above is unexpected. Consider (52), which is identical to (51), except for the subjects of the matrix clause which have been replaced by a quantifier phrase. It shows that the a(n)-phrase can take maximally wide scope over any logical operators or quantifiers outside the island. (52) a. Each teacher overheard the rumor that a student of mine had been called before the dean. (69) b. Each teacher thinks that for a student I know to be called before the dean would be preposterous. (70)

In (52), a can take scope over each teacher. If it were a quantifier, even with special characteristics allowing it to be unconstrained by scope islands, examples such as (52) should have three readings: one where the scope of the indefinite is restricted to the island, an ‘intermediate reading’, where the indefinite has scope over the matrix verb but not over each, and finally a maximally wide scope where the indefinite has scope over the whole sentence, including each. However, this is not what we get as (52) does not have the intermediate reading. In other words, (52a) cannot mean that for each teacher there is some different but nevertheless specific student such that the rumor that that student had been called before the dean has been overheard by that teacher. This reading is represented in (53b), the maximally wide scope in (53c) and the narrow scope in (53a). (53) a. (each teacher: x) [x overheard the rumor that [(a student of mine: y) [y had been called before the dean]]] b. (each teacher: x) [(a student of mine: y) [x overheard the rumor that [y had been called before the dean]]] c. (a student of mine: y) [(each teacher: x) [x overheard the rumor that [y had been called before the dean]]] (Fodor & Sag 1982: 495(71))

According to Fodor & Sag (1982), the lack of intermediate reading of the indefinite in (52) shows that it is not a quantifier when it takes widest scope.29 If it were, it would not

.  Note that this implies that the presence of the partitive PP does not necessarily lead to a quantificational reading (otherwise a student of mine would be quantificational in all three sentences in (52a–c)). This is exactly the position I defend in Chapter 3. More precisely, I argue that what makes a noun phrase quantificational is the projection of QP, independently of the presence/absence of a partitive PP.

 The Layered DP: Form and meaning of French indefinites

only require some special property allowing it to scope out of the relative clause, but it would also have intermediate scope in addition to the wide scope reading represented in (53c). The indefinite has to have either maximal scope over the matrix sentence or scope only over the embedded clause (in which case it is fully non-specific in Fodor and Sag’s terms), which is unaccounted for by a QIO theory. However, if a(n)-phrases can be referential entities as proposed by Fodor and Sag, they are not scoped elements and therefore they do not participate in scope relations, rather like demonstratives or proper names. According to the authors, this conclusion is supported by a decisive argument, VP Deletion. As discussed by Sag (1976a, 1976b) and Williams (1977), VP Deletion is impossible if its antecedent contains a quantificational phrase whose scope is wider than the VP. For (54) to be acceptable, for instance, everyone must have narrow scope. If it takes scope over someone, the sentence is infelicitous. (54) Someone loves everyone. Chris knows that someone does. (75)

VPs containing an indefinite noun phrase however can be deleted even if its scope is wider than the VP: (55) Sandy thinks that every student in our class plays chess better than a guy I beat this morning. Chris



(i) does too

(ii) thinks that every student does too. (79)

Both (55)(i) and (ii) can be understood as attributing to Sandy a belief about a specific person who was beaten at chess by the speaker. This means that the VP can be deleted despite the fact that the indefinite takes scope over the matrix predicate. Fodor and Sag consider such facts as evidence for the referential interpretation of indefinites. A QIO theory cannot account for the acceptability of (55). In short, on the basis of the examples discussed in this section, Fodor and Sag conclude that the referential-quantificational ambiguity of indefinites must be acknowledged in simple sentences like (43). The reason is that the ambiguity is not a matter of relative scope of two quantifiers in a sentence. If it were, the ambiguity should disappear when one of the quantificational elements is removed. As it is not, the ambiguity remains as for lexically ambiguous terms. The classification emerging from Fodor and Sag’s analysis is summarised in (56) (Table 3): (56) Table 3.

Determiners Quantificational Ambiguous quantificational-referential

- every, all, each, most, few, no - some, several, many, a, numerals



Chapter 1.  Introduction 

Fodor and Sag’s analysis is however not without any problems and has been severely criticised (Ludlow & Neale 1991; Abusch 1994…; see also Farkas 1981). Consider the following observations, based on Ludlow & Neale (1991) and mentioned in Diesing (1992, Chapter 3). One argument in favour of the referential reading of indefinites advocated by Fodor and Sag is the absence of intermediate reading in sentences like (52a) repeated below as (57): (57) Each teacher overheard the rumor that a student of mine had been called before the dean. (Fodor & Sag 1982 (69))

According to Fodor and Sag, a student of mine can have widest or narrowest scope, but not intermediate scope. On a quantificational analysis, this is unexpected. That the intermediate scope exists, however, has been shown in the literature (Farkas 1981 and subsequent work, see Chapter 2 Section 3.4.3 for some discussion). As for (57), the absence of such a reading could be due to the semantics of the noun rumor (Ludlow & Neale 1991). As it is indeed difficult to isolate individual instances of the same rumor, it makes difficult to tease apart scope interactions with distinct occurrences of a single rumor. As a demonstration, Ludlow and Neale replace rumor with report, which renders the intermediate scope of the indefinite possible: (58) a. Every student overheard three reports that a Gila monster of mine drowned. (Diesing 1992: 90(54) b. (every student: x) [(a Gila monster of mine: y) [x overheard three reports that [y drowned]]]

The existence of the intermediate reading in examples like (58) invalidates Fodor and Sag’s argumentation. This means that there are two ways of looking at such data. One possibility is to assume that (i) quantificational a-NPs are trapped in islands on a par with other quantificational noun phrases and that the indefinites which can have intermediate and/or wide scope are not quantificational in the traditional sense. The other possibility is that (ii) quantificational a-NPs have some special property, which has to be determined, which allows them to escape islands and to take intermediate and/or wide scope. In the next chapters, I will argue for the first option. Another strategy to bring out the intermediate interpretation is adopted by Kratzer (1986). According to her, the missing reading can be forced pragmatically in certain contexts. In (59) for example, the indefinite can clearly have intermediate scope: (59) Each writer overheard the rumor that she didn’t write a book she wrote.  (Diesing 1992: 90(55))

A book she wrote is within the scope of each writer, but has scope over the rumor… In short, the intermediate reading of Fodor and Sag’s referential indefinites seems in fact

 The Layered DP: Form and meaning of French indefinites

to be possible, contrary to what they claim (see Chapter 2 Section 3.4 for a potential account for such examples).30

5.1.3  Enç In her paper, Enç (1991) discusses the semantics of specificity. She acknowledges that the term specificity has been used in different ways in the literature. Among the definitions she mentions, is the traditional idea that a noun phrase is specific “when the speaker has an individual in mind as its referent” (1991: 1), citing Hellan (1981) and Ioup (1977), and the most wide-spread view which is in terms of scope, i.e., that a noun phrase is specific if it takes wide scope over an operator, citing Fodor & Sag (1982). On the basis of examples such as (60), Enç adopts a position which analyses specificity as independent of scope relations. (60) a. John talked to a logician about this problem. (8a) b. John talked to a certain logician about this problem. (8b)

Scope relations cannot be responsible for the difference of meaning between (60a) and (60b) as these sentences do not contain any operator (as it was the case for (43)). To gain insight into the semantics of specificity, Enç discusses data from Turkish, in which noun phrases are never ambiguous with respect to specificity. Specific and non-specific indefinites in object positions are disambiguated by means of case marking. Specific NPs have overt accusative case contrary to non-specific ones which are not marked for case. This difference is illustrated in the following examples: (61) Ali bir kitab-i aldi. (14) Ali one book-acc bought



‘A book is such that Ali bought it.’

(62) Ali bir kitap aldi. (15) ‘Ali bought some book or other.’

In (61), the object, which bears accusative case, is necessarily interpreted as specific. In (62), on the contrary, the object, which does not bear accusative, cannot be specific. Enç also suggests that specific objects in Turkish, have a covert partitive reading (64a), an analysis which she extends to English (65). In other words, two girls in (64a) is synonymous to two of the girls. .  Another problem for Fodor and Sag’s analysis is that, in some contexts, universally quantified NPs can escape scope islands introduced by propositional attitude verbs, contrary to Fodor and Sag’s claim. In (i), for instance, every Gila monster in New Mexico can take wide scope over thinks but narrow scope with respect to a man in Arizona: (i) A man in Arizona thinks that every Gila monster in New Mexico won the lottery.  (Diesing 1992: 148 fn. 40)



Chapter 1.  Introduction 

(63) Odam-a birkaç çocuk girdi. (16) my.room-dat several child entered ‘Several children entered my room.’ (64) a. b.

Iki kiz-i taniyordum. (17) two girl-acc I.knew ‘I knew two girls.’ Iki köz taniyordum. (18) two girl I.knew ‘I knew two girls.’

(65) a. Several children entered the museum. (23) b. I saw two boys at the movies. (24)

If (63) is the first sentence of a discourse, it can be followed by (64a) or (64b). The difference is that in (64a), contrary to (64b), the two girls represent a subset of the girls introduced in (63). The two interpretations correlate with the difference in case marking: in (64a) the object bears accusative contrary to the one in (64b). In English, the indefinite noun phrase two boys in (65b) can have a non-partitive reading or a covert partitive reading. In other words, the two boys mentioned in (65b) can be included in the referent of several children in (65a) or not. If they are, it means that the two boys might have gone to the movies after their visit to the museum and the reading is specific. If they are not, it means that we are dealing with two boys who did not go to the museum and the reading is non-specific. In fact, what Enç proposes is that specificity places a constraint on the structure of the domain of discourse. (64a) says something about two girls who have previously been introduced into the domain of discourse, whereas (64b) makes first mention of the two girls. Such an approach to specificity implies that specificity is closely related to definiteness. What differentiates specifics from definites is the way they are linked to a referent previously established in the discourse. This linking is weaker and looser for specifics than for definites. For the former, it is a relation of inclusion whereas for the latter it is a relation of identity (of referent). As identity entails inclusion, all definites are necessarily specific. How does this analysis apply to indefinites? Indefinites introduce new referents into the discourse. They are ambiguous because the referent they introduce can be included in a referent which has already been introduced (inclusion) or not. In the first case, they are specific, like (64a), and in the second one they are non-specific, like (64b). The indefinites which are affected by the ambiguity specific-non-specific are listed in (66): (66) a.

bir, iki, yedi, otuz, …31

‘one, two, seven, thirty, …’ (35)

.  Note that the indefinite article in Turkish does not appear in the list of ambiguous elements provided by Enç in (66). The text (1991: 4–5) however suggests that it can be specific or not, on

 The Layered DP: Form and meaning of French indefinites

b. birkaç c. birçok d. az

‘several, a few’ ‘many’ ‘few’

When non-specific, the indefinites in (66) are allowed in existential sentences, like Milsark’s weak determiners in their weak reading. They contrast with Turkish noun phrases containing her ‘every’, bu ‘this’, o ‘that’, pronouns and proper names which are always specific, and which correspond to Milsark’s strong class. To explain the ungrammaticality of specific noun phrases in existential sentences, Enç assumes following Keenan (1987) that they presuppose existence, which is incompatible with the assertion of existence of existential sentences (see also Section 5.1.1). On the basis of such observations, and assuming that specificity is partially constrained by the lexicon – it is not truth conditional – and that it is independent from the definiteness of the noun phrase, Enç proposes that what is known as the Definiteness effect should be labelled Specificity effect (1991: 16). As mentioned above, Enç assumes that specificity involves linking objects to the domain of discourse. One way of achieving this linking is the inclusion relation. So what has been said so far in this section concerns what Enç labels ‘partitive specifics’. Another way of linking objects to the domain of discourse is by relating them to familiar objects. This mainly concerns noun phrases of the form a certain N, labelled ‘relational specifics’: (67) a. Every true Englishman admires a certain woman – his mother. (63) b. John wants to own a certain piano which used to belong to a famous pianist. (4a) c. The teacher gave each child a certain task to work on during the afternoon. (59)

In (67a), the use of a certain woman is felicitous because the relation between Englishmen and the woman they admire is made explicit by the constituent his mother. In (67b), a certain piano is fine because the relation between John and the piano is expressed by the relative clause. In (67c), on the other hand, the NP containing certain is licensed by the agent of the assignment, the teacher. The two linking processes correlate with different distributional properties between the two types of specific noun phrases. Partitive specifics (inclusion) presuppose existence, contrary to relational ones (1991: 21). Consequently, only the latter are grammatical in existential sentences. This means that not only non-specific noun phrases are fine in existential sentences but also relational specifics. The main ideas reported above are summarised in Table 4: a par with the English a. This is not surprising as the Turkish indefinite article is homophonous with the numeral 1, which is included in the list (66).



Chapter 1.  Introduction 

(68) Table 4.

Determiners

Specific Ambiguous specific-non-specific

- definite NPs, quantifiers - familiar indefinites: a certain N - several, many, few, numerals

In the next section, I turn to Diesing’s (1992) influential work. We will see that the classification of determiners is also close to Milsark’s.

5.1.4  Diesing In a sense, Diesing (1992) generalises Milsark’s (1974, 1977) analysis, which focuses on the special case of existential constructions. She proposes that, at LF, there is a silent existential quantifier at the edge of every VP, in opposition to the existential quantifier postulated by Milsark in the there be-construction. This results from Diesing’s attempt to establish the role of syntactic representations in the formation of the logical representations of quantified sentences of the sort developed by Heim (1982). The tripartite representation introduced by Lewis (1975) and proposed by Heim for a sentence like (69a) is reported in (69b): (69) a. Every llama ate a banana. b. Everyx [x is a llama]





quantifier

(∃y) y is a banana ∧ x ate y

restrictive clause

nuclear scope (Diesing 1992: 7(10))

The restrictive clause specifies the set over which the quantifier quantifies. This means that in (69) the quantifier every quantifies over every thing which is a llama. The logical representation thus consists of an operator, a restrictive clause and the nuclear scope, hence the term tripartite. Diesing proposes that there is a procedure, the Mapping Hypothesis (or tree-splitting algorithm), by which the tripartite logical representations mentioned above can be derived from the syntactic representation of a sentence. This process splits the syntactic tree into two parts, which are mapped into the major components of the logical representation, the restrictive clause and the nuclear scope (70a). (70b) is the Mapping Hypothesis expressed in words. (70) a. Mapping Hypothesis (tree splitting)

(Diesing 1992: 9(13))

 The Layered DP: Form and meaning of French indefinites IP

restrictive clause I′

Spec I

VP Spec



V′ V

nuclear scope XP

b. Material from VP is mapped into the nuclear scope. Material from IP is mapped into a restrictive clause.

(ibid, (14))

In (70a), the part of the tree dominating VP represents the restrictive clause and the VP the nuclear scope. As for the quantifier, which constitutes the third part in the logical representation, Diesing suggests that it could be excluded from the restrictive clause and the nuclear scope by means of IP-adjunction, probably at the syntactic level LF. If the VP represents the nuclear scope of the logical representation, it predicts that indefinites which have no quantificational force of their own must be within VP after tree-splitting. The reason is that in this way they can be bound by existential closure – the silent operator mentioned at the beginning of the section. Consequently, indefinites cannot be treated uniformly, which contrasts with Heim’s (1982) analysis in which all indefinites are represented as variables.32 Indefinites which are outside the nuclear scope must have quantificational force of their own and undergo Quantifier Raising (QR), i.e., they are able to form restrictive clause structure. Diesing’s classification of indefinites is based on the contrast between cardinal and presuppositional determiners observed by Milsark (Section 5.1.1). Recall that for the latter all strong determiners are presuppositional, contrary to the weak ones which are ambiguous between a presuppositional and a predicative reading. Diesing thus considers indefinites as ambiguous and proposes that their interpretations correlate with two different syntactic positions. Evidence for this claim comes from German examples like (71) : (71) a.

… weil ja doch zwei Cellisten in diesem Hotel abgestiegen sind. since ‘indeed’ two cellists in this hotel have-taken-rooms

.  The Mapping Hypothesis is given further support by an asymmetry in the semantic and syntactic properties of the subjects of the stage- and individual-level predicates of Carlson (1977b). Cf. footnote 24.



Chapter 1.  Introduction 

b. … weil zwei Cellisten ja doch in diesem Hotel abgestiegen sind. since two cellists ‘indeed’ in this hotel have-taken-rooms  (Diesing 1992: 78(42))

In (71), zwei Cellisten ‘two cellists’ can sit in two different positions, either following the particles ja doch ‘indeed’ or preceding them. These particles are used by Diesing to diagnose the position of the subject: when the subject appears to the right of these particles, it is inside VP, in Spec,VP, and when it appears to their left, it sits in Spec,IP. Depending on the position the subject sits in, it is interpreted differently. When it is in Spec,VP the cardinal reading obtains, whereas when it occupies Spec,IP, the presuppositional reading is more salient, hence the association Spec,IP and presuppositionality. To summarise, Diesing proposes that noun phrases outside of VP are always presuppositional and that all presuppositional noun phrases are outside of VP at LF. As a consequence, the noun phrases external to VP have strong determiners. To put it differently, noun phrases which denote properties are interpreted in situ unlike quantificational ones which have to move out of VP to be interpretable.33

5.1.5  Summary and comparison What is striking in the papers examined in the preceding sections is the differences in terminology, which may or may not correspond to differences in data. To ease the comparison between the categories proposed in the papers discussed, let us try to classify indefinites by author, with sentence type (see Chapter 2 Section 3.5 for a comparison with the typology proposed in this book).34 The tables below (Tables 5, 6 and 7) only concern noun phrases with the indefinite article, except for Fodor and Sag’s quantificational indefinites.35 The contexts listed must take or strongly prefer the type of indefinites given to their left, except in Table 7 where existential sentences may host non-specifics and relational specifics. (72) Table 5.

Milsark 1. Weak weak 2. 3. Weak strong

Context Existential sentences * Individual level predicates Otherwise

.  This movement is overt in Turkish and German which attest scrambling. .  As the different authors do not always discuss the same types of sentences, the comparison is difficult and partial. As Diesing’s classification is very close to Milsark’s it is discarded in the comparison. .  In Table 5, ‘weak weak’ stands for weak indefinites with a weak reading and ‘weak strong’ for weak indefinites with a strong reading.

 The Layered DP: Form and meaning of French indefinites

(73) Table 6.

Fodor & Sag



1. Referential 2. 3. 4. 5. 6. 7. Quantificational 8. 9. 10.

Context ‘Widest’ scope Descriptive richness (including in existential sentences) (76) Topicalisation and Left Dislocation (77) Existential sentences about a particular entity (78) Relative clauses, especially non-restrictive (79) With certain and particular (80) Existential sentences (81) Intensional contexts Narrow scope Numerals

(74) Table 7.

Enç

Context

1. Non-specifics Existential sentences 2. Not introduced in the discourse 3. Intensional contexts 4. Partitive specifics Introduced in the discourse 5. Relational specifics With familiar objects Co-variation with ∀ (82a) Widest scope (82b) 6. Existential sentences

The first observation is that Milsark concentrates on one type of contexts, namely existential sentences. If weak indefinites with a weak reading are restricted to this environment, it implies that everywhere else weak indefinites have a strong reading. This is problematic however for examples discussed by Fodor and Sag and Enç where an indefinite may have two interpretations in a sentence which does not contain any operator. Recall (75) from Section 5.1.2: (75) A student in the syntax class cheated on the final exam.

As this sentence is not existential, Milsark’s analysis predicts that the indefinite has a strong reading. In other words, it cannot be ambiguous, contrary to facts. Concerning Fodor and Sag’s analysis, it is important to note that they consider the ambiguity of indefinites as lexical, i.e., not associated with distinct syntactic structures. Therefore the contexts given in Table 6 illustrate (strong) preferences for one of the two interpretations they advocate. In other words, the other reading should not be excluded. The contexts in Table 6 which are not discussed in Section 5.1.2, namely 2–6, are illustrated below. An example of an existential context asserting the non-emptiness of a set is also provided although it does not contain an indefinite article (81).



Chapter 1.  Introduction 

(76) a. A student that Betty used to know in Arkansas cheated on the exam. (4) b. There’s a man that Kim used to go to school with in the late sixties in Arkansas smoking behind the woodshed. (19) (77) A Frenchman that I met in Tokyo, I went and had dinner with (him) in New York last week. (13) (78) There’s a girl in our syntax class who cheated on the exam. (20) (79) A student in the syntax class, who has a Ph.D. in astrophysics, cheated on the exam. (non-restrictive). (22) (80) a. I accused a certain student of cheating. (27) b. A (one) particular claim in this paper is false. (28) (81)

There was someone smoking behind the woodshed. (18)

Examples like (80) suggest that Fodor and Sag’s referential indefinites are analogous to or at least include some of Enç’s ‘relational specifics’ (compare also (78) and the use of ‘relational specifics’ in existential sentences). More evidence for this similarity comes from intensional contexts like (82b) in which the indefinite takes widest scope. Enç’s relational specifics however also comprise indefinites which co-vary with a universal quantifier (82a). For Fodor and Sag the latter are quantificational; they take narrow scope. (82) a. Every true Englishman admires a certain woman – his mother. (63) b. John wants to own a certain piano which used to belong to a famous pianist. (4a)

One question which arises, and which will be addressed in Chapter 2, is whether the terms quantificational and presuppositional are equivalent as Milsark’s and Enç’s analyses suggest. Recall that in Section 5.1.1 we saw that for Milsark strong means quantificational, and that De Jong & Verkuyl (1985) propose that strong noun phrases are presuppositional. For Enç too noun phrases are banned from existential sentences because they are specific and presuppose existence (1991: 14) (Section 5.1.3). However, as mentioned at the end of Section 5.1.1, this is problematic for definites like proper names and pronouns. On these considerations, I leave the two-way ambiguities and turn to works in which indefinites are attributed three interpretations.

5.2  Three-way ambiguities 5.2.1  Dobrovie-Sorin and Beyssade Dobrovie-Sorin & Beyssade (2004) propose that indefinites can be analysed as individuals, i.e., referential entities, as quantifiers or as properties.36 In other words, they .  To avoid the hypothesis that indefinites exhibit a lexical ambiguity, Dobrovie-Sorin and Beyssade do not exclude that they change type (Partee 1986).

 The Layered DP: Form and meaning of French indefinites

are classified according to their semantic type, two of which are already familiar to us. Recall Fodor and Sag who describe the ambiguity of indefinites in terms of referentiality and quantification (Section 5.1.2). The main proposal of Dobrovie-Sorin and Beyssade is that indefinite noun phrases can be properties in argumental positions (and not just in predicative contexts). That argumental indefinites can be properties has been proposed by Higginbotham (1985) but this approach has only been developed recently in the literature (McNally 1995; Van Geenhoven 1996; Dobrovie-Sorin 1996). Most theoreticians stick to the traditional view that syntactic positions of arguments can only be filled with typical arguments, i.e., referential or quantificational noun phrases. Let us examine Dobrovie-Sorin and Beyssade’s analysis of existential sentences there be, as this construction has also been addressed by the authors mentioned in Section 5.1. To account for the so-called Definiteness-effect, Dobrovie-Sorin and Beyssade propose that it is the semantic type of the noun phrase which matters. In other words, to be grammatical in there be sentences, the noun phrase has to be a non-individual.37 According to them, such an analysis does not only account for the difference of grammaticality between (83) and (84), but also for the difference between (84a–b) and (84c). (83) a. b. 

Il there Il there

y Y y Y

avait un homme devant la porte. French was a man in.front.of the door a un couteau sur la table. is a knife on the table (Dobrovie-Sorin & Beyssade 2004: 61(1))

(84) a. b. c. 

? Il there ? Il there *Il there

y Y y Y y Y

a Fred dehors. French is Fred outside avait la table dans le jardin. was the table in the garden avait chaque table dans le jardin. was each table in the garden (Dobrovie-Sorin & Beyssade 2004: 61(2))

(83) is perfectly fine contrary to (84). The linguistic tradition attributes the difference to (in)definiteness. As the arguments in the former are indefinite, the sentences are grammatical, unlike the sentences of the latter whose arguments are definite (84a–b) and quantificational (84c). Dobrovie-Sorin and Beyssade account for the difference in terms of semantic types: (83) is grammatical because the arguments are properties, contrary to the ones in (84).38 .  This means that not only indefinites, which are properties, but also kinds, sorts or types can occur in there be contexts (2004: 62 fn. 3). .  They also analyse bare nouns as property-denoting expressions.



Chapter 1.  Introduction 

(84) also shows that definite and quantificational noun phrases in there be sentences lead to different judgements of grammaticality: the former are not completely impossible, the latter are. The ungrammaticality of quantificational noun phrases is unsurprising as they cannot be properties. However, if this is on the right track, examples like (85), where the argument is quantified but perfectly licit in a there be construction, are unexpected. (85) Il

y avait toutes sortes de médecins à la

there Y were all 

kinds of doctors

réunion.

French

at the meeting

(Dobrovie-Sorin & Beyssade 2004: 62(3a))

An analysis in terms of semantic types allows Dobrovie-Sorin and Beyssade to account for the difference between (84c) and (85). They propose that the quantifier does not quantify over the domain of individuals but over sets of individuals, in other words over properties (2004: 71 ff.). As for the grammaticality of definite noun phrases in there be contexts (86), Dobrovie-Sorin and Beyssade propose that these sentences are in fact not existential sentences but constructions that depict an event. In other words, the speaker does not assert existence but presents a new event. (86) a. b.



Il there Il there

y Y y Y

a is a is

Jean qui part à Paris demain. Jean who leaves to Paris tomorrow la fille du voisin parmi les figurants. the daughter of.the neighbour among the walkers.on (Dobrovie-Sorin & Beyssade 2004: 62(3b–c))

In (86), the definite noun phrases are grammatical in the there be constructions because they depict new events, i.e., they are arguments of an event and not of the existential there be. In other words, Dobrovie-Sorin and Beyssade argue that there are different there be constructions and that all are not existential (Hoeksema 1996). An analysis is terms of definiteness cannot account for (85) and (86). If Dobrovie-Sorin and Beyssade’s approach is on the right track, there are three types of indefinites: referential, quantificational and property-denoting indefinites. In the next chapter we will see that this is the most interesting and promising way of classifying indefinites. Another study, restricted to French indefinites with the indefinite article, is presented in the next section.

5.2.2  Baunaz A three-way analysis of indefinite noun phrases is proposed by Baunaz (2005a). However, her study only addresses un-NPs ‘a-NPs’ and not indefinites in general. Following Starke (2001), she assumes that there are two kinds of noun phrases which are presuppositional, in the sense presupposed, i.e., already introduced in the discourse

 The Layered DP: Form and meaning of French indefinites

(Starke p.c.). The first type of presupposition corresponds to Enç’s (1991) ‘partitive specifics’. In other words, such un-NPs are subsets of a previously mentioned set. The second type of presupposition does not involve any subset/set relation but refers to familiar individuals, i.e., it corresponds to Enç’s ‘relational specifics’. Note that such noun phrases do not presuppose existence (cf. Section 5.1.3) as will be defined more precisely in Chapter 2. The first type of presupposition is called ‘range-based’ and the second one ‘specificity-based’. They are illustrated in (87) and (88) respectively (Baunaz’ (4) and (5)). (87) Context: Xander shuffles the cards, lays them fan-shaped and presents them to his friends. Each player has to pick a different card, Xander will try to guess which ones. He then says: Xander : - Prenez tous une carte. pick all a card ‘Pick one card each.’ (88) Speaker A: Did you watch Friends yesterday? Is Monica still going out with that insecure guy? Speaker B: Tu devrais quand même savoir que depuis trois saisons, you should all the same know that since three seasons Monica sort avec un homme (i.e., Chandler) ! Monica goes.out with a man ‘You should know that for 3 seasons, Monica [has been] dating a guy.’

In (87), une carte ‘a card’ is part of the set of cards shuffled by Xander, in other words it represents a subset of those cards. In (88), on the contrary, un homme ‘a man’ is not part of a set previously introduced in the discourse but designates a familiar person. The third type of un-NPs discussed by Baunaz is not presuppositional, i.e., it introduces a new entity into the discourse. It is illustrated in (89) (her (6)): (89) a. b.

Ron a un ennemi. ‘Ron has an enemy.’ Nous allons tous essayer de savoir qui c’est. we go all to.try of to.know who it is ‘We are all trying to figure out who it is.’

(89) asserts the existence of Ron’s enemy. It introduces a novel individual. According to Baunaz’ description, this type of un-NP corresponds to Heim’s (1982) indefinites as variables in the sense that it is not quantificational per se but bound by an operator outside the noun phrase. To support the idea that there are two kinds of presuppositional un-NPs, Baunaz examines contexts in which they interact with other quantificational elements, such as tous les étudiants ‘all the students’ and chacun des étudiants ‘each of.the students’,



Chapter 1.  Introduction 

with the features [+ universal, + range] and [+ universal, + specific] respectively (for the reasons for such a differentiation see the original paper). As (90) and (91) show, the scope interactions differ. Range un-NPs are marked with capital letters and specific ones with the diacritic ∨ (see below). (90) a. b.

Tous les étudiants ont lu UN livre. (∀ > UN) ; *(UN > ∀) Tous les étudiants ont lu ∨ un livre. *(∀ > ∨ un) ; (∨ un > ∀) ‘All the students have read a book.’ (17a–b)

(91) a. b.

Chacun des étudiants a lu UN livre. (∀ > UN) ; *(UN > ∀) *Chacun des étudiants a lu ∨ un livre. ‘Each of the students has read a book.’ (18a–b)

Assuming that quantificational noun phrases undergo QR, (90a) suggests that range un-NPs, which must take narrow scope, are blocked, contrary to specific un-NPs which take wide scope over the quantificational noun phrase tous les étudiants ‘all the students’ (90b). In (91), chacun des étudiants ‘each of.the students’ takes scope over the range un-NP, but is ungrammatical with a specific un-NP. That range un-NPs and specific un-NPs interact with universal quantifiers shows that they are quantificational too. In addition, that their scope properties differ suggests that we are dealing with two different kinds of quantificational un-NPs. To corroborate her three-way distinction, Baunaz notes that each type of un-NP corresponds to a different intonation. Range-based presuppositional un-NPs are characterised by a downfall intonation, marked with capital letters (92), whereas specificity based presuppositional un-NPs have a slight fall-rise intonation, marked with the diacritic ∨ (93). (92) Monica aime UN HOMME. ‘Monica loves a man (from a set of men).’ (93) Monica aime ∨ un homme. ‘Monica loves a specific man.’

= downfall (capitals) (Baunaz 2005a (7)) = fall-rise (∨) (Baunaz 2005a (8))

The third type of un-NP has a neutral intonation. (94) Monica aime un homme. ‘Monica loves any man.’

= neutral (Baunaz 2005a (9))

In short, on the basis of intervention effects, Baunaz proposes that un-NPs have two quantificational readings and a non-quantificational one. One difference between the first two is whether they have a covert partitive reading or not.

 The Layered DP: Form and meaning of French indefinites

5.2.3  Summary This section shows that the interpretation of indefinites is even more complex than what was suggested in Section 5.1, as there are very likely more than two interpretations. One line of analysis is Dobrovie-Sorin and Beyssade’s, who consider that indefinites can be of any of the semantic types: referential, quantificational and property-denoting. Another possibility is to refine the two-way ambiguities discussed in Section 5.1 as Baunaz does. Range un-NPs correspond to Enç’s partitive specifics and specific un-NPs are analogous to Enç’s relational specifics.

chapter 2

un-NPs ‘a-NPs’ 1.  Introduction Chapter 1 presents various papers on the interpretation of indefinites, among which Milsark (1974, 1977), Fodor & Sag (1982), Enç (1991) and Diesing (1992) who argue that indefinite noun phrases have two interpretations and Dobrovie-Sorin & Beyssade (2004) and Baunaz (2005a) who propose that there are three kinds of indefinites. Depending on the analyses, the terms referential, quantificational, presuppositional, (non)specific, variable or property have been used to describe indefinite noun phrases. However, these terms have not been clearly defined so far and their use may vary from one author to the next. The first aim of this chapter is thus to define such notions in order to determine the terminology to be adopted in the rest of this book and more importantly to establish a typology of indefinite noun phrases in French. The indefinite noun phrases I am primarily concerned with in this book are unNPs ‘a-NPs’ and du/des-NPs ‘of.the-NPs’. The former are discussed in this chapter and the latter in Chapter 3. Chapter 4 examines bare nouns to a minor extent. The study of un-NPs concentrates on argumental noun phrases involving the indefinite article un/une ‘a’. This means that predicative examples such as Jean est un professeur ‘John is a professor’ will be disregarded. Existential sentences involving there be however will be mentioned in various sections of the chapter. One reason is that the indefinite noun phrases occurring in such constructions are arguments, at least on the syntactic level. Generic sentences like Un chat boit du lait ‘A cat drinks milk’ will be ignored in this chapter but briefly mentioned in Chapter 4. The study of the interpretative properties of un-NPs leads to a three-way classification. The typology I propose in this chapter is analogous to the one proposed by Dobrovie-Sorin & Beyssade (2004) mentioned in Chapter 1, Section 5.2.1 although the terminology varies slightly. The three types I argue for are un-NPs encoding speaker’s reference (S-referential un-NPs), quantificational un-NPs and un-NPs interpreted as properties. In other words, un-NPs belong to different semantic types. That noun phrases in syntactic argument positions cannot belong to a single semantic type is in line with the work of many researchers since the early 1980s (see for example Partee 1986) and contrasts with a theory analysing all noun phrases as generalized quantifiers (Montague 1974; Barwise & Cooper 1981). Strong evidence for the latter

 The Layered DP: Form and meaning of French indefinites

conclusion comes from landmark works such as Kamp’s (1981) and Heim’s (1982). These authors convincingly show that many noun phrases lack quantificational force and that therefore they should not be considered as quantifiers. The second aim of this chapter is to determine the internal structure of the three types of un-NPs identified and to find out to what extent their structure is related to their (lack of) existentiality, (lack of) scope properties and (lack of) referentiality. Assuming with Zamparelli (2000) that different layers of DP encode different interpretations, we expect each type of un-NP to have a different structure. More generally, I show that the role of the internal structure of the indefinites under study is much more important than assumed so far to account for the characteristics mentioned above, namely existentiality, scope and (non)-reference. This suggests that their position in the sentence plays a more minor role in their interpretation than often assumed in the literature (Diesing 1992 for example). This means that some semantic phenomena like scope properties for instance can be accounted for by means of a syntactic analysis, at least to some extent. Needless to say that I do not imply that attributing different internal structures to un-NPs accounts for all the issues arising from the study of indefinites and their scope properties, which have been discussed in the literature for decades (see Reinhart 1997 for an excellent overview). As this chapter focuses on argumental un-NPs, the structure of syntactic arguments in general has to be considered. Building on analyses such as Longobardi’s (1994, 2005) which argue that arguments must involve the DP layer, I propose that their structure must minimally be constituted of one layer of the left periphery. More precisely I argue that it is the lowest layer which turns predicative noun phrases into arguments. The presence of this projection allows a noun phrase to function as an argument but this argument will be semantically undetermined compared to un-NPs with a left periphery involving more than one layer; it will be interpreted as an N or other. A fundamental point of the analysis is therefore that the composition of the left periphery of each type of un-NP differs. The idea is that the interpretation of un-NPs results from the piling up of different functional layers in the nominal left periphery and that each layer of the left periphery corresponds to a different semantic type. This analysis is supported by empirical facts involving intervention effects. For the structure of S-referential un-NPs, I further propose that the highest projection of the left periphery must combine with the QP layer characteristic of quantificational un-NPs. The combination of these projections accounts not only .  For an analysis assuming that no semantic types play any role in the syntactic derivation, see Guéron 2006. .  In Chapter 1, I assumed that DP is parallel to CP. However, if the different structures of the nominal left periphery correspond to different semantic types, this is an important difference with the clausal left periphery.



Chapter 2.  un-NPs ‘a-NPs’ 

for their S-referential interpretation but also for their scope properties, i.e., the fact that they can move out of scope islands and have widest and/or intermediate scope. In other words, the internal structure of un-NPs has an impact on their behaviour at the clausal level and therefore on the interpretation of the clause. Another issue to be tackled is existentiality, in the narrow sense of assertion of existence, as this property is often associated with indefinites. What the investigations presented in this book show is that the existential reading of indefinites may be due to different factors and not only to a mechanism such as Existential Closure. The chapter is organised as follows. The typology of un-NPs advocated here is presented in Section 2. In Section 2.1, I argue that referential un-NPs exist, as proposed by Fodor & Sag (1982). The idea is that reference is not direct reference but rather dependent reference often analogous to ‘speaker’s reference’ in Kripke’s terms (1977). Section 2.2 discusses quantificational un-NPs, which correspond to Enç’s (1991) ‘partitive specifics’. Such indefinites represent a proportion of a set already introduced in the discourse, i.e., presupposed. Finally, Section 2.3 discusses a third type of un-NPs, which I label Property un-NPs. These noun phrases are undetermined in the sense that they mean an N or other. The distribution of such un-NPs shows that they are ­ungrammatical in negative contexts in which de-NPs occur. Section 3 deals with the structure of un-NPs. In Section 3.1.1, I report Longobardi’s (1994, 2005) analysis which shows that arguments are DPs and refine this proposal according to the framework adopted here. This new definition of syntactic arguments will play a crucial role in the rest of the book. This section also presents some of the main points of Zamparelli’s (2000) dissertation, as he proposes to split DP into several projections. Section 3.2 deals with the structure of un-NPs functioning as properties, Section 3.3 with quantificational un-NPs and Section 3.4 with S-referential un-NPs. The idea is that the composition of the left periphery of each type of un-NPs varies with their interpretation. Crucially, the structures proposed account for several characteristics attributed to each type of un-NPs, in particular their scope (or absence of scope) properties. Finally, Section 4 consists of some concluding remarks.

2.  Types of un-NPs 2.1  Referential noun phrases 2.1.1  Definition and characteristics Traditionally, referential expressions are expressions which have a reference, hence designate an individual or entity in some domain of interpretation. Referential noun phrases can be split into two categories, the ones whose referent is rigid, that is, it remains the same, independently of the context, situation or world, and the ones which need some context to be interpreted. The former typically consist

 The Layered DP: Form and meaning of French indefinites

of proper names like John or Mary and definite descriptions like the book or Mary’s cat and the latter of demonstratives like this book or that car, pronouns like I and you and traces. Due to their interpretational properties, pronouns (except for bound variables) and traces are also considered as free variables (Heim & Kratzer 1998: 116), term used by Heim 1982 for indefinites. Free variables denote individuals, “but only relative to a choice of an assignment of a value”, i.e., to an individual (ibid, p. 92, see also pp. 116, 131). In other words, their denotation varies with an assignment, implying that they differ from syntactic variables such as wh-traces. The latter traditionally count as null R-expressions, which are full noun phrases with independent reference. Two criteria are commonly used to classify noun phrases as referential: the Law of Contradiction and the Law of Excluded Middle (Heim & Kratzer 1998: 133–34, Dobrovie-Sorin & Beyssade 2004: 10 ff.). The former is illustrated below.

(1) a. Mount Rainier is on this side of the border, and Mount Rainier is on the other side of the border. (Heim & Kratzer 1998: 133(4)) b. Le fils de la voisine a plus de trente ans et the son of the neighbour-fem has more than thirty years and le fils de la voisine n’ a pas plus de trente ans. the son of the neighbour-fem neg has not more than thirty years ‘The son of the neighbour is over thirty and the son of the neighbour is not over thirty.’  (Dobrovie-Sorin & Beyssade 2004: 12(7a))

The sentences in (1) are necessarily contradictory as they consist of a conjunction of two VPs with disjoint extensions, be on this side of the border and be on the other side of the border in (1a), and avoir plus de trente ans ‘be over thirty’ and ne pas avoir plus de trente ans ‘not be over thirty’ in (1b). In other words, the sentences in (1) are always false. However, such examples are only contradictory if the subjects are referential, i.e., if they denote an individual. In contrast, consider (2):

(2) a. More than two mountains are on this side of the border, and more than two mountains are on the other side of the border.  (Heim & Kratzer 1998: 134(5))

.  Pronouns can be deictically referring, anaphoric or bound. ‘Deictically referring’ means that the element refers to some individual that is somehow salient in the context of utterance, maybe due to some extra linguistic context, ‘anaphoric’ means that the pronoun picks up its reference from another phrase in the surrounding text and ‘bound’ means that the element is in the scope of a quantifier. For example, herj in Maryi likes herj cat and she in She is so clumsy! are deictically referring, heri in Maryi likes heri mother is anaphoric, and hei in Every mani believes that hei is a genius is a bound variable.

Chapter 2.  un-NPs ‘a-NPs’ 



b. Deux enfants sont venus et deux enfants ne sont pas venus. two children are come and two children neg are not come  (Dobrovie-Sorin & Beyssade 2004: 12(7c))

Contrary to the sentences in (1), the ones in (2) are not contradictory. This is because there could be more than two mountains on each side of the border and because two children might have come even though two (different) children might not have come. In other words, the subjects in (2) do not denote entities. Let us turn to the Law of Excluded Middle. The idea is to coordinate by or two sentences whose subjects are identical and whose VPs, when united, exhaust everything there is. Such sentences are tautologies, i.e., they are always true. Consider (3): 

(3) a.

I am over 30 years old, or I am under 40 years old. (Heim & Kratzer 1998: 134(6)) b. Ou bien le fils de la voisine a plus de trente ans, or (well) the son of the neighbour-fem has more than thirty years, ou bien le fils de la voisine n’ a pas plus or (well) the son of the neighbour-fem neg has not more de trente ans. than thirty years (Dobrovie-Sorin & Beyssade 2004: 12(7b))

As somebody necessarily is either over 30 or under 40, (3a) exhausts all the possibilities there are and cannot be false. The same reasoning applies to (3b). However, such sentences are only tautologies when their subjects denote individuals as in (3). Compare with (4).

(4) a. Every woman in this room is over 30 years old, or every woman in this room is under 40 years old. (Heim & Kratzer 1998: 134(7)) b. Chaque fils de la voisine a plus de trente ans, each son of the neighbour-fem has more than thirty years, ou chaque fils de la voisine n’ a pas plus or each son of the neighbour-fem neg has not more de trente ans. than thirty years (Dobrovie-Sorin & Beyssade 2004: 12(7d))

The examples in (4) can be false, i.e., they are not tautological. As tautologies are yielded only with referential subjects, it implies that the subjects in (4) do not have that property. Noun phrases with every and chaque ‘each’ are prototypical quantificational noun phrases (See Section 2.2). In the next section, I turn to un-NPs in order to determine whether they (can) have the properties of referential noun phrases mentioned above.

 The Layered DP: Form and meaning of French indefinites

2.1.2  Referential un-NPs The question whether indefinite (and definite) noun phrases can involve reference is an ongoing debate in the semantics literature. The linguistic and philological tradition commonly treats indefinites as if they were referring expressions. As observed by Heim (1991), when uttering (5), for example, a cat is used by the speaker as if it referred to the cat which is present.

(5) A cat is present.

(Heim 1991 (110))

Such an analysis is problematic however if the speaker is lying or if there is no cat, i.e., if no cat exists, or when a cat is embedded under an operator, such as negation and quantifiers.

(6) a. Hagit won’t bring a cat. b. There is a cat sitting in every violin case.

(Heim 1991 (113–4))

In (6), it is impossible to ask to which individual/entity/cat a cat is referring, implying that in such contexts a cat is not referential, i.e., it does not denote an individual or an entity. Such examples, which will be further addressed in Section 2.3, do however not show that un-NPs cannot be referential. What about the Law of Contradiction and the Law of Excluded Middle mentioned in the preceding section? According to Heim & Kratzer (1998: 134), noun phrases with numerals, no and – crucially – the indefinite article a, pattern like the ones in (2). In other words, as (7a) is not a contradiction, a mountain should not be considered as referential (See Section 2.2 for further discussion).

(7) a. A mountain is on this side of the border and a mountain is on the other side of the border. b. Une montagne se situe de ce côté de la frontière a mountain se lies of this side of the border et une montagne se situe de l’autre côté de la frontière. and a mountain se lies of the other side of the border

As (7b), the French counterpart of (7a), is not a contradiction either, it suggests that in such contexts une montagne ‘a mountain’ can be non-referential. The Law of Excluded Middle leads to the same conclusion. Consider the following example:

(8) a. A woman in this room is over 30 years old, or a woman in this room is under 40 years old. b. Une femme dans cette pièce a plus de 30 ans, a woman in this room has more than 30 years

.  The examples cited as Heim 1991 are in fact from a translation of hers of her paper (ms.).

Chapter 2.  un-NPs ‘a-NPs’ 





ou une femme dans cette pièce a moins de 40 ans. or a woman in this room has less than 40 years

As this example can be false, it suggests that a woman and une femme ‘a woman’ are not (necessarily) referential. The discussion so far shows that un/a-NPs can be non-referential. The question is whether this excludes a referential reading. In other words, could a mountain and a woman refer to an individual/entity, even if examples (7) and (8) would not sound natural? Could a mountain/une montagne and a mountain/une montagne in (7), and a woman/une femme and a woman/une femme in (8) have the same referent, leading to a contradiction or to a tautology respectively? Nothing excludes a/un-NPs from the premise underlying the Law of Contradiction and the Law of Excluded Middle, namely that the subjects of contradictory and tautological sentences denote individuals/entities. The view that indefinites can be referential is Fodor & Sag’s (1982) (Chapter 1, Section 5.1.2) but it is also the position adopted by Donnellan (1978) and Wilson (1978) to cite but two. To show that indefinites have a referential reading, one must invalidate the counter-thesis that the alleged referential indefinites with a are in fact non-referentials with widest scope, as Heim (1991) observes. This is very difficult as an indefinite with a referential meaning and an indefinite which is not referential but which has widest scope are not easily distinguishable. The arguments put forward by Fodor and Sag, namely scope islands and the absence of intermediate reading, succeed in supporting the existence of a specific/referential reading of indefinite noun phrases with a, as acknowledged by Heim (1991). The examples she discusses are reported in (9) to remind us of the crucial points. They involve if-clauses, and not relative clauses with a rumor as the ones reported in Chapter 1 Section 5.1.2 and criticised in Section 5.1.5.

(9) a. If I discussed every objection, I would never get done. b. If I had paid attention back then to a soft-spoken young man passing out leaflets in the street, things would have taken a very different course.  (Heim 1991(138–9))

As (9a) involves two scope taking constituents, every objection and if, we expect the sentence to have at least two different interpretations, depending on their respective scope. However, the scope of every objection seems to be confined to the if-clause. The sentence means that “Indiscriminate attention to all objections would prevent me from getting done” and not “for each objection…, if I were to discuss it, I wouldn’t get

.  In opposition to Kripke 1977, Neale 1990 and Ludlow & Neale 1991 who consider that indefinites (and definites) are never semantically referring. According to them, only ‘singular terms’ such as proper names, demonstratives and personal pronouns are referential.

 The Layered DP: Form and meaning of French indefinites

done… I can’t discuss any objection at all if I want to finish” (Heim 1991: 32). In other words, there is no objection whose discussion alone would take so much time that it would prevent the speaker from finishing. Such examples suggest that the universal quantifier cannot have wider scope than the clause in which it is embedded (but see Farkas 1981). As (9b) essentially has the same structure as (9a), we expect the scope of the a-NP, if it is quantificational, to be limited to the if-clause. However, even if this reading is possible, it is not the most natural one, which is that there was a soft-spoken young man passing out leaflets in the street such that it would have had dramatic consequences if I had paid attention to that man [which] doesn’t preclude that there also were, or could have been, other softspoken young leafletting men who it wouldn’t have made any difference at all whether I paid attention to or not (p. 33).

One possibility to account for this reading, where a soft-spoken young man passing out leaflets in the street is not trapped in the if-clause, is to assume that it is not due to scope properties but to the fact that this un-NP is referential. Such an analysis is supported by even more sophisticated examples such as (10): (10) Everyone would be grateful today if he had paid attention back then to a soft-spoken young man passing out leaflets on the street. (Heim 1991(145))

The a-NP can have widest scope, which is the most natural reading, or narrowest scope, a secondary interpretation, “according to which everybody would be grateful today if he had listened to any soft-spoken young leafletter” (p. 33). The intermediate reading, “that for each person there is a soft-spoken young leafletter who that person

.  The expected readings are reported below for the readers with some semantic background (Heim 1991 (140–41)): (i) ∀x [F(x) → G(x)] ⇒ p (ii) ∀x [F(x) → [G(x) ⇒ p]] where the counterfactual if-then is symbolized by ⇒ and the predicates and sentences abbreviated as follows: F = objection, G(x) = I discuss x, p = I never get done. .  The expected readings are reported below (Heim 1991 (142–43)): (i) ∃x [F(x) & G(x)] ⇒ p (ii) ∃x [F(x) & [G(x) ⇒ p]] where F = soft-spoken young man passing out leaflets in the street, G(x) = we paid attention to x, p = things took a very different course.

Chapter 2.  un-NPs ‘a-NPs’ 



would be grateful to have listened to” (p. 33), is impossible. This is unexpected if the a-NP is a quantificational element, but not if it is referential, as referential noun phrases do not have scope properties. And this seems to be a solid argument for the existence of referential a-NPs. The above discussion also shows that referential un-NPs are associated with wide scope, although this term does not seem appropriate as it should only concern quantificational elements (in opposition to referential ones) (but see Section 3.4 for further discussion). Referential noun phrases, as seen in the preceding section, do not necessarily have a rigid reference. Referential un-NPs, on a par with pronouns and traces are context dependent, as observed by Heim (1991) for referential English a-NPs. This means that their referent may vary with the utterance situation. For example, the interpretation of the proposition in (11) depends on which student the speaker intends to refer to/has in mind. Referential a-NPs are represented by Heim as ar-NPs following Fodor and Sag. (11) Ar student of mine got his Ph.D.

(Heim 1991(135))

Referential is thus used here in the sense of ‘speaker’s reference’, in Kripke’s (1977) terms (see also Farkas 1997). “The speaker’s referent is that individual which the speaker “wishes to talk about” (or “has in mind”) on the occasion of the utterance” (Heim 1982: 15). This means that speaker’s reference and specificity, in the sense ‘individuals in the speaker’s mind’ or ‘familiar to the speaker’, are equivalent. In this book, I will stick to Kripke’s term and abandon the label specific, in the sense ‘an individual in the speaker’s mind’ (the term specific in the sense of Enç’s ‘partitive specifics’ will also be abandoned; see Section 2.2.3). One reason is that there seem to be as many definitions of specificity as authors discussing the topic. The label ‘specificity-based presuppositional un-NPs’ mentioned in Chapter 1 Section 5.2.2 will not be used either. The un-NP in the example mentioned, repeated below as (12), is an ar-NP. In the terminology adopted here, it is a un-NP with speaker’s reference. (12) Speaker A: Did you watch Friends yesterday? Is Monica still going out with that insecure guy?

.  The three expected readings are reported below (Heim 1991 (146–8)): (i) ∀x [∃y [F(y) & G(x,y)] ⇒ H (x)] (ii) ∀x ∃y [F(y) & [G(x,y) ⇒ H (x)]] (iii) ∃y [F(y) & ∀x [G(x,y) ⇒ H (x)]] where G(x,y) = x paid attention to y, H(x) = x is grateful today.

(i) is the reading where the a-NP has narrowest scope, (ii) is the unattested intermediate reading and (iii) the referential reading.

 The Layered DP: Form and meaning of French indefinites

Speaker B: Tu devrais quand même savoir que depuis trois saisons, you should all the same know that since three seasons Monica sort avec un homme (i.e., Chandler) ! Monica goes.out with a man ‘You should know that for 3 seasons, Monica [has been] dating a guy.’

If Speaker B has Chandler in mind when uttering un homme ‘a man’ in (12), then un homme ‘a man’ is S-referential, i.e., it has speaker’s reference. The mentioning of the speaker is important. Implicitly, it excludes the hearer. Consider (13): (13) a. I bought a car. b. Pass me a book.

(Lyons 1999: 165)

In (13), neither a car nor a book is identifiable by the hearer. However, the former refers to something familiar to the speaker, contrary to the latter. In other words, a car is S-referential, contrary to a book (See Section 2.3 for further discussion of such noun phrases). If a car for example were also familiar to the hearer, the definite article would be in order. Another property which is characteristic of un-NPs encoding speaker’s reference is that they are existential. Consider the following example: (14) John loves a woman (it is Jane).

Although (14) has several readings, let us concentrate on the one which states that John loves a particular woman, whose identity seems to be known by the speaker but not by the hearer. If no such woman exists, the sentence is false. This is supported by the paraphrase below: (15) There is an x, x is a woman, and John loves x.

What (15) does, is, by the means of the semantic formula there is an x…, assert that x exists (see Section 3.4.5 for further discussion). In other words, S-referential un-NPs introduce an entity in the discourse and assert their existence. This property will be contrasted with presupposition of existence in Section 2.2.3. The position adopted here is thus that some un-NPs are referential. They are existential and their interpretation is determined by the context of utterance on a par with pronouns and deictic elements like demonstratives (recall (11)). I use the term S-referential for referential un-NPs (and des-NPs in Chapter 3) as they usually depend on the speaker (but see section 3.4.3 for a looser definition). .  As usual, the choice of an appropriate label is a tricky thing. As observed by P. Sleeman (p.c.), the term ‘speaker’s reference’ is too restrictive as S-referential un-NPs may also co-vary with a universal quantifier as in Every man loves a woman – his mother, where the speaker is not involved in the identification of a woman (see Section 3.4.3 (the fact that the speaker is not necessarily involved is also observed by Enç 1991: 19)).



Chapter 2.  un-NPs ‘a-NPs’ 

The next Section deals with quantificational noun phrases and argues that un-NPs may be of that kind, i.e., intrinsically quantificational.

2.2  Quantificational noun phrases In Section 2.1.1, two laws were given to distinguish between noun phrases which are referential and noun phrases which are not, namely the Law of Contradiction and the Law of Excluded Middle. What it means for a noun phrase not to be referential has not been clearly specified however. In this section, we will see that such phrases may be quantificational (see also Section 2.3 which postulates another type of non-referential un-NPs).

2.2.1  Definition and characteristics The difference between quantificational and referential expressions is that the former do not pick out a referent, contrary to the latter. For example, everyone in (16) does not refer to an individual with the property that John saw him. Rather it means something along the lines ‘for every element x, if x is human, it is the case that John saw x’, where x is a variable bound by the quantifier. In other words, the interpretation of the complement of see is not fixed. It varies with the quantifier. Every (and all) are universal quantifiers and someone an existential quantifier (17), symbolised by ∀ and ∃, respectively. (16) John saw everyone.

One property of quantifiers, also called operators, is that they may lead to ambiguity, as is well known for examples like (17): (17) Everyone saw someone.

The two meanings traditionally attributed to this sentence are given below: (18) a. For every x, there is some y such that it is the case that x saw y. b. There is some y, such that for every x, it is the case that x saw y.  (Haegeman 1994: 490(5))

In (18a), each person might have seen somebody different, which means that the number of persons seen depends on the number of persons that see, i.e., on the universal quantifier everyone. In other words, the number of ys depends on the number of xs. In contrast, in (18b) there is one person that was seen by everyone, i.e., the number of persons seen is independent of the universal quantifier. The ambiguity of (17) is due to the different scope of the quantifiers: in (18a), the universal quantifier everyone appears to the left of the existential quantifier someone and we say that it has wide scope over the existential quantifier. In (18b), it is the other way round, it is the universal quantifier which is in the scope of the existential quantifier. Quantifiers take scope over a certain domain and can affect the meaning of certain elements in that domain. In the syntax, this is accounted for by an operation called Quantifier Raising (QR) which adjoins the

 The Layered DP: Form and meaning of French indefinites

quantifier to IP (see Section 3.4.3 for a revised version). qr takes place at the level of representation which encodes logico-semantic properties, LF, for Logical Form: (19) [IP everyonei [IP John saw xi]]

In (19), the logical representation of (16), the operator everyone occupies a scope position, which is assumed to be left-peripheral. The scope of the quantifier is the domain to its right, and more precisely the domain it c-commands. In (19), xi represents the empty category left by the movement of everyone. It is coindexed with the operator. Another way of representing quantificational expression is the tripartite notation mentioned in Chapter 1 Section 5.1.4 and repeated below: (20) a. Every llama ate a banana. b. Everyx [x is a llama]



(∃y) y is a banana ∧ x ate y



quantifier restrictive clause

nuclear scope

(Diesing 1992: 7(10))

In (20) every quantifies over the restriction and takes scope over the nuclear scope. The restriction is the sister node of the quantifier. This relation was described in terms of proportion in Chapter 1 Section 5.1.1.10 In the next section I turn to un-NPs. The aim is to determine whether these noun phrases may be quantificational.

2.2.2  Quantificational un-NPs One reading of Milsark’s cardinality determiners (some/sm, many for instance) is quantificational, in the sense of expressing a proportion of a given set (Chapter 1, Section 5.1.1). In other words, they have a covert partitive meaning (cf. Pesetsky’s term D(iscourse)-linked 1987). One example showing that this is true of the indefinite ­article is repeated below: (21) I didn’t buy the house, because a window was broken.

(Lyons 1999: 264)

In (21), a window represents one of several windows of the house in question. That a/un-NPs can be quantificational is also supported by the tests presented in Section 2.1, namely the Law of Contradiction and the Law of Excluded Middle. That (7) and (8), repeated as (22) and (23), are not necessarily contradictory or true, respectively, shows that the a/un-NPs functioning as subjects in these examples do not (necessarily) denote individuals. This is precisely one characteristic of quantificational noun phrases, i.e., they do not pick out an entity.

.  Note that proportionality and partitivity are distinct notions. Proportional determiners are non intersective and non symmetric contrary to partitives which are intersective and symmetric. Furthermore, proportional noun phrases are quantificational contrary to partitives which may not be (Chapter 3, Section 3.1.2). For further details, the reader is referred to DobrovieSorin & Beyssade 2004: 27ff.

Chapter 2.  un-NPs ‘a-NPs’ 



(22) a. A mountain is on this side of the border and a mountain is on the other side of the border. b. Une montagne se situe de ce côté de la frontière a mountain se lies of this side of the border et une montagne se situe de l’autre côté de la frontière. and a mountain se lies of the other side of the border (23) a. A woman in this room is over 30 years old, or a woman in this room is under 40 years old. b. Une femme dans cette pièce a plus de 30 ans, a woman in this room has more than 30 years ou une femme dans cette pièce a moins de 40 ans. or a woman in this room has less than 40 years

Finally, the scope interactions characteristic of quantificational noun phrases as in (17) – Everyone saw someone – represent further evidence for the quantificational reading of a/un-NPs. Consider (24) which is similar to (17) except that someone has been replaced by the a-NP a woman. (24)

Everyone saw a woman.

The above sentence is ambiguous and has the two readings below, parallel to (18): (25) a. For every x, there is a y such that it is the case that x saw y. b. There is a y, such that for every x, it is the case that x saw y.

The ambiguity illustrated above disappears when the a-NP is replaced by a proper name as in Everyone saw Mary, suggesting that a woman in (24) should be treated on a par with someone rather than with Mary. Another context in which referential noun phrases and a-NPs can have different interpretations is topicalisation as in (26b) where a question is moved to the left periphery of the sentence. This movement affects the truth-conditional meaning of such sentences (Heim & Kratzer 1998: 135 ff.). (26) a. Almost everybody answered a question. b. A question, almost everybody answered.

(26a) can be true when no two people answer the same question, whereas (26b) requires that one question is answered by almost everybody. In other words, the former can be true and the latter false. This is not the case if the noun phrases in (26) are replaced by referential noun phrases: (27) a. John answered question #7. b. Question #7, John answered. (adapted from Heim & Kratzer 1998: 135(8))

 The Layered DP: Form and meaning of French indefinites

Even if the meaning of (27a) and (27b) differ subtly, there is no state of affairs in which one can be true and the other false, in contrast to the pair in (26). If the noun phrases in (26) were treated like those in (27), the truth-conditional effects would be unexpected. This suggests that a question in (26) is closer to at least one question in (28) than to question #7 in (27). (28) a. Almost everybody answered at least one question. b. At least one question, almost everybody answered.  (Heim & Kratzer 1998: 136(10))

All these points show that the quantificational reading of a-NPs is well established. In the next section, I deal with a question of terminology. In Chapter 1, we saw that Milsark (1974, 1977) and Enç (1991) use the terms quantificational and presuppositional, in the sense of presupposition of existence, indistinctively. What I would like to determine is whether these notions are synonymous. Before discussing indefinites let us go back to Russell’s work.

2.2.3  Presupposition of existence Presupposition of existence conflicts with assertion of existence, typical of there be sentences (Chapter 1 Section 5.1.1). The distinction between assertion and presupposition of existence is Frege’s position (1892). It was rejected by Russell (1905), who limited his theory of descriptions to singular definite expressions. His by-now wellknown example with the King of France is reported below: (29) The King of France is bald.

According to Russell this sentence represents the conjunction of the following three propositions: (30) a. There is a King of France. b. There is only one King of France. c. This individual is bald.

(Lyons 1999: 255)

The first proposition in (30) is known as the existential clause and the second as the uniqueness clause. Both are assumed by most logical work on definite noun phrases and are characteristic of descriptions like the King of France. What is not so generally agreed on is the relation between these two clauses and the third one, which asserts that the individual in question is bald. Russell claims that all three propositions are



Chapter 2.  un-NPs ‘a-NPs’ 

asserted, i.e., they are of equal status.11 Consequently, they are logical entailments, and the falsity of any of them leads to the falsity of the whole conjunction. However, this runs counter to the strong intuition that the first two propositions are felt to be background assumptions in contrast to the last one which is asserted, which is Frege’s theory. In fact, on the basis of existential sentences, Frege held that to be referential, expressions presuppose a reference to something, and that such presuppositions must be satisfied to determine the truth-value of a sentence, i.e., whether a sentence is true or false. Adopting such a position would mean that all the singular referential expressions mentioned so far presuppose existence. It would also imply that presupposition is a concept much closer to reference than to quantification, a position radically different from Milsark’s and Enç’s who seem to conflate it with quantification (recall Chapter 1, Section 5.1.1 and 5.1.3). Another possibility is to use the term presuppositional for noun phrases which carry presuppositions about their restrictors, i.e., the sister of the determiner, labelled restrictive clause in (20b). The presupposition may be about the cardinality of the restrictor or about the non-emptiness of the set to which the restrictor belongs (Heim & Kratzer 1998: 153–172). In the first case, the cardinality of the restrictor is known. For example, in the book, book is the restrictor and the presupposes that there is a unique (relevant) book. In other words, the book presupposes that the cardinality of the restrictor is 1. In the same vein, both and neither presuppose that there are exactly two relevant entities. Other examples include the two books, the three books… the n books and partitives such as three of the four books… as the cardinality of the set of books is known. In the second case, the cardinality of the set to which the restrictor belongs to is not known. Rather, what is known is that the set to which the restrictor belongs is not empty. This version of presupposition, adopted here, thus subsumes the first one, i.e., examples in which the cardinality of the set of the restrictor is known. It also includes possessives and partitives in general, as in John’s books (= the books of John) and two of the books respectively, as such examples presuppose that the set of books is not empty. The most radical version of this definition of presuppositionality extends to all quantificational noun phrases, a position adopted by Diesing (1992).12 This means that even in examples like (31), every is considered as presuppositional. (31) a. Every unicorn has exactly one horn. b. Every unicorn is a unicorn.  (cited from Reinhart 1995 in Heim & Kratzer 1998: 164(9))

.  Russell’s account was criticized by Strawson (1950, 1952) who proposed to return to Frege’s position, though in a more elaborate version. .  McCawley 1972 also argues for such an analysis. See also Diesing & Jelinek 1996.

 The Layered DP: Form and meaning of French indefinites

(31) is judged true without hesitation by informants (Reinhart 1995). However, as unicorns do not exist - and all the informants were aware of that, the question whether (31) is true or false should not arise and (31) should give rise to a presupposition failure. This means that, as the presupposition of existence of unicorns is not fulfilled, (31) should be judged as neither true nor false, contrary to facts. To accommodate such data, it is possible to consider the set of mythologically possible unicorns and not of actual unicorns. I leave the technical details of the analysis aside here.13 To maintain such a strict version of presuppositionality, Diesing adopts a position where weak determiners may or may not be quantificational. Such a type ambiguity implies that weak determiners are presuppositional only when quantificational. At first sight, such an analysis does not account for the distribution of the different readings, namely that strong determiners receive a presuppositional reading independently of the position they occupy, i.e., whether they are subjects or not, whereas weak ones may remain neutral in this regard. Consider (32): (32) If you find every mistake(s), I’ll give you a fine reward. most many a no three… (quoted in Heim & Kratzer 1998: 172(22),  from Zucchi 1994 citing Lumsden 1988)

As mentioned above, in (32), the weak determiners many, a, no … remain neutral as to the existence of mistakes, whereas the strong ones, every and most, presuppose that the speaker assumes that there are mistakes. To account for this difference in distribution Diesing invokes her Mapping Hypothesis according to which the interpretation of indefinites correlates with two different syntactic positions: indefinites which are outside VP must have quantificational force of their own, and thus be presuppositional, hence the association Spec,IP and presuppositionality (Chapter 1, Section 5.1.4). Diesing’s position is not uncontroversial though and there is no consensus in the literature on the issue of presupposition.14 However, I will stick to the definition of presupposition given here, namely that the set to which the restrictor belongs is not empty, and not pursue the debate further. The above discussion throws light on the notion presupposition and shows that it is not synonymous with quantification as defined in Section 2.2.1 of this chapter.

.  The idea is that in examples like (31) quantificational noun phrases are under the scope of a modal operator. The quantification would thus be over possible individuals. .  Heim & Kratzer (1998: 172) mention that De Hoop 1992 links the type ambiguity of weak noun phrases to case, thus doing away with Diesing’s Mapping Hypothesis, and that others, such as Reinhart 1995 and Büring 1996, reject a type ambiguity analysis.



Chapter 2.  un-NPs ‘a-NPs’ 

Although the radical definition of presupposition implies that all quantificational noun phrases are presuppositional, it does not imply that all presuppositional noun phrases are quantificational as the book, the two books… mentioned above show. This concept of presupposition also differs from Milsark’s strong noun phrases which include proper names and pronouns, which are clearly not quantificational as observed in Chapter 1 Section 5.1.5, and not presuppositional in the sense discussed here. Another difference with Milsark’s analysis is that noun phrases with the indefinite article a can be quantificational as seen in this section and thus presuppositional, in contrast to Milsark who considers a as unambiguously cardinal, i.e., non-quantificational and thus non-presuppositional. Consider the following example which supports the position adopted here (adapted from Baunaz 2005a (4)): (33) Context: Xander shuffles the cards, lays them fan-shaped and presents them to his friends. Each player has to pick a different card, Xander will try to guess which ones. He then says: Xander: a. Prenez tous une carte. pick all a card ‘Pick one card each.’ b. Avez-vous pris une carte? have you taken a card ‘Have you picked a card ?’ c. Vous prendrez une carte quand je vous le dirai. you will.take a card when I you it will.say ‘You will pick a card when I tell you.’

In (33), the set of cards has been introduced in the context, i.e., it is presupposed. The card which will be picked is one of this set, i.e., it is a subset. This means that une carte ‘a card’ in (33) is a covert partitive. It is quantificational in that it does not pick an entity. It is also presuppositional in that it does not assert existence but presuppose it. This is because the set of cards has been introduced in the context, i.e., we know that the set of cards is not empty. In contrast, if somebody said to Xander J’ai pris une carte, c’est au tour de X de jouer ‘I have taken a card, it is X’s turn to play’, in the sense ‘one of the cards’, and no card existed, Xander would probably reply Quelle carte? ‘which card?’. Such a context is a case of presupposition failure.15 The remarks concerning Milsark’s analysis also apply to Enç’s as she adopts a position similar to Milsark’s in that she attributes the ungrammaticality of noun phrases in there be constructions to their presuppositionality. The indefinite noun phrases which

.  Without a context aforementioned, une carte ‘a card’ in (33) would not necessarily be existential. The lack of existence entailment in imperative, interrogative and future tense structures is a characteristic of the type of un-NPs discussed in the next section.

 The Layered DP: Form and meaning of French indefinites

are presuppositional are labelled ‘partitive specifics’ by Enç because they have a covert partitive reading. However, it seems to me that the distinctions between S-referential and quantificational noun phrases on the one hand, and between noun phrases which presuppose existence and the ones which assert existence on the other hand, are sufficient to distinguish between different types of noun phrases. Consequently, I will ignore Enç’s label and the terms specific and specificity all together. This extends to the term ‘range-based presupposition’ (Chapter 1 Section 5.2.2), which corresponds to Enç’s partitive specifics. Note furthermore that the different interpretations of a/un-NPs cannot be captured by an analysis in terms of specificity only. In effect, using a single feature, such as [± specific], would only allow us to distinguish between two kinds of indefinites, [+ specific] indefinites and [– specific] indefinites. This does not correspond to the facts as (34), which is three-way ambiguous, shows. (34) Everyone failed to stop at a red light.

(Heim 1991(131))

The above example can mean that there is a red light at which nobody stopped, or that for each person there is a red light at which he/she did not stop, or that nobody stopped at any red light. In the first reading the indefinite noun phrase a red light takes widest scope, in the second one it has intermediate scope and in the last one narrowest scope relative to the universal quantifier and the negative verb, everyone and fail respectively. As these three readings cannot be due to a single feature, a combination of features will be necessary to account for the facts. A full analysis of examples like (34) will therefore only be possible once the typology of un-NPs is completed (Section 2.3) and after discussion of un-NPs with intermediate scope (Section 3.4.3). The first objective will be reached in the next section, which deals with a third type of un-NPs, which will be analysed as Property noun phrases.

2.3  A third type of noun phrases 2.3.1  Interpretation and characteristics The traditional view considers that only referential and quantificational noun phrases can fill syntactic positions of arguments. These two types of noun phrases have been discussed in the preceding sections, although I have suggested that referential un-NPs in fact encode speaker’s reference. The study of un-NPs however shows that these two types cannot be the only ones. In effects, un-NPs may have additional properties which cannot be attributed to these types. Consider the following example: (35) Jean veut acheter un livre. Jean wants to.buy a book ‘Jean wants to buy a book.’

In (35), un livre ‘a book’ can be S-referential and denote a book in particular, for example Hamlet (Section 2.1). Or, in a situation involving several books, it might represent

Chapter 2.  un-NPs ‘a-NPs’ 



one of the books introduced in the context, in which case it is quantificational (Section 2.2). The third reading I would like to capture is an interpretation where un livre ‘a book’ means a book or other, in the sense that it is not a particular book in the speaker’s mind, nor a book of a set of familiar books. It is an undetermined book in that what counts is that it is a book and not a magazine or a pencil.16 In other words, un livre ‘a book’ in (35) can mean any book or a book or other. A similar interpretation can be attributed to a cat in (36), repeated from Section 2.1.2, and to a red light in (37), repeated from Section 2.2.3: (36) a. Hagit won’t bring a cat. b. There is a cat sitting in every violin case. (37) Everyone failed to stop at a red light.

(Heim 1991 (113–4)) (Heim 1991 (131))

(36) was mentioned to show that un/a-NPs are not always referential, i.e., there is no individual/entity which can be pointed at in (36), and (37) to show that a two-way distinction for un-NPs is not appropriate. (36a) can even be felicitously uttered if no cat exists. The failure of existential quantification is an important characteristic of the type of un-NPs discussed here. It shows that not all indefinites assert existence. This was also observed for quantificational un-NPs, which presuppose existence (see Section 2.2.3 on the difference between assertion and presupposition of existence). That the third type of un-NPs may be non-existential holds in various contexts. The most striking examples illustrating this characteristic are questions (38)–(39) and intensional contexts (40):17

.  Note however that this undetermined reading is not contrastive, in which case un livre ‘a book’ would be emphasised. .  Compare (38) and (39) with their affirmative versions below, where the un-NPs, although of the third type discussed here, assert existence, i.e., are existential.

(i)

a. b.

Tu as déjà acheté un livre. you have already bought a book ‘You have already bought a book.’ Jean a rencontré un Bernois. Jean has met a Bernese ‘Jean met a Bernese.’

A key element might be the fact that (i) denotes an event which is bound by an existential operator (Davidson 1967, Giannakidou 1997, Kratzer 1989, Parsons 1990, among others). As the event variable is existentially bound, the event is interpreted as actually having occurred (Clarke 2001), which implies the existence of the nominal. Another possibility is that the semantics of extensional verbs like acheter ‘buy’ – in opposition to intensional verbs – supplies an existential operator (this option is further discussed in Section 2.3.2).

 The Layered DP: Form and meaning of French indefinites

(38) Yes/no question As-tu acheté un livre ? have you bought a book ‘Have you bought a book?’ (39) Wh-(constituent) question Qui a rencontré un Bernois ? who has met a Bernese ‘Who has (ever) met a Bernese?’ (40) Intensional contexts a. Jean veut un chat. Jean wants a cat ‘Jean wants a cat.’ b. Je trouverai un étudiant pour m’ aider. I will.find a student to me to.help ‘I will find a student to help me.’ c. Jean cherche un chat. Jean is.looking.for a cat ‘Jean is looking for a cat.’

Many different elements may create intensional (opaque) contexts: modal (40a) and temporal expressions (40b), verbs like want and seek (40c), but also words that express propositional attitudes (like know and believe, fear and hate) and contexts of direct and indirect speech. Additional examples are given below: (41) a. b. c.

Subjunctives Marie a insisté pour qu’ un étudiant puisse entrer. Marie has insisted for that a student could-subj to.enter ‘Marie insisted that a student may enter.’ Imperatives Va voir un médecin ! go to.see a doctor ‘Go see a doctor!’ Disjunctions Soit un voleur est entré dans la maison, either a thief is entered in the house soit on a oublié d’ éteindre. or one has forgotten of to.switch.off ‘Either a thief entered the house, or we forgot to switch off the light.’

Although none of the above examples asserts the existence of the un-NPs they contain, they are felicitous. This means the un-NPs of this type may be existential or not (see Section 2.3.2 for a more detailed analysis). Note that this characteristic seems to distinguish these un-NPs from S-referential un-NPs which are always existential (Section 2.1.2). Consider the following example:



Chapter 2.  un-NPs ‘a-NPs’ 

(42) Une licorne vient souvent se reposer dans mon jardin. a unicorn comes often refl to.rest in my garden Je l’ai appelée Rousse. I it have called ‘Rousse’ ‘A unicorn often comes to rest in my garden. I have called it Rousse.’

(42) concerns a particular unicorn, i.e., une licorne ‘a unicorn’ is S-referential. If no unicorn exists, the sentence is false. In other terms, S-referential un-NPs seem to assert existence. Another property of the un-NPs under discussion is that they strongly tend to take narrow scope, as illustrated in the following example: (43) Il y avait un visiteur tous les jours. there Y was a visitor each the days ‘Each day there is a visitor, (but not necessarily the same visitor).’  (adapted from Bosveld-de Smet 2004(20))

There is no scope ambiguity in (43). The preferred reading is the one where tous les jours ‘every day’ takes wide scope. As a first diagnostic, this could be due to a lack of quantificational force of the un-NP (see Section 3.2 which supports this hypothesis). That the noun phrases with the characteristics described in this section form a category on their own has been acknowledged in the literature. They have been analysed as property-denoting arguments (or properties). This theoretical approach is discussed in the next section.

2.3.2  Property-denoting noun phrases That argumental indefinites can be property-denoting noun phrases has been proposed by Higginbotham (1985) but this approach has only been explored in the early 1990s (de Hoop 1992; McNally 1992; Ladusaw 1994; Van Geenhoven 1995; McNally 1995; Dobrovie-Sorin 1996).18,19 This line of analysis has been evoked in the outline of various typologies of indefinites (Chapter 1 Section 5.2.1). Henceforth, I will use the label Property-(denoting) noun phrases for argumental noun phrases denoting a property, and Property un-NPs (or P-un-NPs) for such noun phrases attesting an indefinite article. Noun phrases like un médecin ‘a doctor’ in Jean est un médecin ‘John is a doctor’ would be predicative un-NPs. One type of evidence for property-denoting a/un-NPs functioning as arguments comes from verbs like become and consider, which take arguments of type 〈e,t〉 – that is properties. Their arguments can be adjectival or nominal, i.e., APs or NPs (Partee 1986/2002). Consider the following example: .  This paper was published in 2004. .  Longobardi (1991, 1994, 2003) and Chierchia (1998b) reject the hypothesis that propertydenoting noun phrases may function as arguments.

 The Layered DP: Form and meaning of French indefinites

(44) Mary considers John competent in semantics and an authority on unicorns.  (Partee 2002: 360(9))

As conjunction requires that the different constituents be of the same type, and as adjectives are of type 〈e,t〉, the noun phrase an authority in (44) must also be of type 〈e,t〉 (the PPs in semantics and on unicorns do not play any role here). Partee thus uses consider sentences as a diagnostic for noun phrases of type 〈e,t〉, i.e., the Property noun phrases advocated here. (45) Mary considers that an island/two islands/many islands [is/are better than…] *every island/*most islands/*this island…  (Partee 2002: 361, partial (10), italics added)20

As an island/two islands/many islands are grammatical in (45), they are considered as property-denoting noun phrases, contrary to every island/most islands/this island which are out. The proposal that some predicates may take property-type arguments highlights the fact that the study of noun phrases and the one of predicates cannot be dissociated. In effect, such a claim requires new composition rules, i.e., rules which allow Property noun phrases to compose semantically with the predicate selecting them. Several solutions to the composition problem are proposed in the literature (McNally 1995; Van Geenhoven 1998; Farkas & De Swart 2003; Chung & Ladusaw 2004). As the crucial point here is that the composition problem can be solved, the discussion will be limited to a single aspect of one of these works – the idea that predicates may have floating types – as it will play a role in the analysis developed in this book. McNally (1995) is one of the early analyses formalising a compositional semantics for predicates taking property-denoting arguments. She argues that bare nouns in Spanish should be considered as properties even in argumental positions, and not as Heimian indefinites or kinds. To solve the composition problem, she proposes to refine the semantics of predicates in that they should be assigned floating types. In other words, the semantics of predicates selecting referential arguments should differ from the one of predicates selecting properties. The former have been labelled ‘entity predicates’ and the latter ‘existential predicates’ (Van Geenhoven 1995; Dobrovie-Sorin 1997a,b; McNally 1998; Dobrovie-Sorin & Laca 2003), in the sense that they contain an existential quantifier which is able to bind (some of) the variable of their arguments. In other words, the idea that the semantics of predicates selecting

.  Definite noun phrases like the prettiest island/the harbour/Utopia are also grammatical in this example. This shows that definite noun phrases can also be of type 〈e,t〉, i.e., properties. What Partee actually suggests is that definite noun phrases can be of any of the three semantic types, referential, quantificational and predicative.

Chapter 2.  un-NPs ‘a-NPs’ 



Property noun phrases may vary leads to new ways of thinking about the existential force intuitively associated with noun phrases. Consider the following examples: (46) a. λx x is handsome (John) b. λxλy x love y (Mary) (John)

(Dobrovie-Sorin & Beyssade 2004: 43(23’)

(47) a. ∃x∃y x wash y b. ∃x x sleep c. ∃x x is available

(Dobrovie-Sorin & Beyssade 2004: 44(26))

In (46), the predicates are entity predicates. Such predicates do not include an existential quantifier in their lexical representation and cannot apply to Property noun phrases. In (47), in contrast, the predicates are existential predicates. This means that they supply an existential operator to (some of) their variables as the representation given below shows. In (48), the existential predicates are represented as lambda-abstractions over predicate variables: (48) a. λP λQ ∃x∃y [x wash y ∧ Q(x) ∧ P(y)] b. λP ∃x [x sleep ∧ P(x)] c. λP ∃x [x are available ∧ P(x)]  (Dobrovie-Sorin & Beyssade 2004: 44(26’))

This means that monadic predicates such as the ones in (48b-c) differ from the canonical representation - λx P(x) where the predicate takes a variable x of type e (entity) – in two ways. First, they supply an existential operator and second, they select propertydenoting expressions (type 〈e,t〉). In other words, the predicates dormir ‘sleep’ and disponible ‘available’ have their argument position x saturated by an existential operator in the lexicon. However, these predicates remain partially unsaturated as they require a property P to specify their domain of variation. This kind of representation is restricted to existential predicates. As such predicates are saturated in the lexicon, they cannot apply to entity denoting expressions. (48) captures the fact that the existential force of the arguments is contributed by the verb and not by the nominal itself. The formalism has become familiar under the term semantic incorporation (van Geenhoven 1995; McNally & Van Geenhoven 1997; Van Geenhoven & McNally 2005). Note that according to representations like (48) property-denoting subjects which are existential – due to the presence of an existential operator in the predicate – are expected. One such example is given below: (49) [at a party] Un invité a fini ce vin. a guest has finished this wine ‘A guest finished this wine up.’

 The Layered DP: Form and meaning of French indefinites

In (49), the property-denoting subject is existential.21 As it is a property-denoting argument, it is not intrinsically quantificational. That existential force should be associated to predicates – and not to noun phrases was already noted by Carlson (1977a). The latter observes that any existential force associated with the bare plural in stage-level predicates comes from the lexical entailments of the predicate and not from either inherent quantificational force associated with the bare plural or from an existential operator adjoined to the text à la Heim (see below). In effect, in her very influential dissertation, Heim (1982, Chapter II) proposes that indefinites are existential whenever they are in the nuclear scope of a quantifier - the third term of a tripartite construction headed by a quantifier (recall (20) in (Section 2.2.1) - or when they occur in an unembedded sentence.22 The reason is that an existential quantifier ∃ is adjoined either to the nuclear scope of the quantifier or to T(ext), which represents expressions larger than a sentence. This adjunction rule, reported in (50), is known as Existential Closure. (50) a. ‘Adjoin a quantifier ∃ to the nuclear scope of every quantifier.’ b. ‘Adjoin the quantifier ∃ to T.’ (Heim 1982: pp.138–140)

(50) accounts for the existential reading of indefinite noun phrases without assuming that the quantification is due to the noun phrase or the article per se.23 In Chapter III of her dissertation, however, Heim observes that both sub-rules of the Existential Closure can be dispensed with and that indefinites are free variables (1982: 327ff. and 359ff.). In other words, they are not bound by a quantifier and existentiality is associated to felicity and truth conditions. If the existential force is encoded on the predicate and not on the noun phrase, it accounts for the fact that Property noun phrases take narrow scope. This is a characteristic I mentioned in the preceding section for un-NPs (recall (43)) and which is noted by McNally (1995) for Spanish bare plurals. Consider the following examples which contrast indefinites with a determiner and bare plurals (McNally 2004: 120(7–9), (51a-b) and (52a-b) are from Laca 1996: 253,254): (51) a. A la reunión no asistieron algunos profesores. at the meeting not attended some-pl professors ‘The meeting was not attended by some professors.’

(c or d)

.  That existential P-un-NPs are much less frequent in subject position than in object position might be due to independent characteristics of subjects in French, namely that they represent ‘old’ information. .  The existential quantifier binds the indefinite, or more precisely the variable left by the movement of the indefinite. This is due to the ‘NP-Prefixing rule’ which states “Adjoin every non-pronominal NP to S” (Heim 1982: 132). .  Indefinites may also be bound by overt or invisible modal operators (Heim 1982: 168 ff.).

Chapter 2.  un-NPs ‘a-NPs’ 



b. A la reunión no asistieron profesores. c. (∃x: professor(x)) [¬attend(x,m)] d. ¬(∃x: professor(x)) [attend(x,m)] (52) a. b. c. d.

María cree que Sara ha robado unos libros. Maria believes that Sara has stolen a-pl books ‘Maria believes that Sara has stolen some books.’ María cree que Sara ha robado libros. (∃x: book(x)) [believe(m,^stole(s,x))] believe(m,^(∃x: book(x)) [stole(s,x)])

(53) a. b. c. d.

Hoy Juana tiene que leer unos artículos. today Juana has that read.inf a-pl articles ‘Today Juana has to read some articles.’ Hoy Juana tiene que leer artículos. (∃x: article(x)) [ [read(j,x)]] (∃x: article(x)) [read(j,x)]

(d only)

(c or d) (d only)

(c or d)

(d only)

According to McNally, the bare plurals in the b examples above must take narrowest possible scope with respect to any other operators in the same clause. This is illustrated for negation in (51), intensional predicates in (52) and modals in (53). In other words, bare plurals are not ambiguous and their only interpretation is the one given in the d examples where the existential quantifier is in the scope of the other operator. Indefinites with a determiner, in contrast, can have the interpretation in d or the one in c where the existential operator takes scope over the other operator. Furthermore, if Property noun phrases are not intrinsically existential, it comes as no surprise that it is this type of un-NPs which occurs in existential constructions (McNally 1992). Recall the following example from Chapter 1 Section 5.1.1. (54) a. There is a wolf at the door. b. * There is the wolf at the door. (Milsark 1977: 5, quoted in Abbott 2004(23))

As seen in Chapter 1, Milsark assimilates the there is part of existential constructions to an existential operator, thus accounting for the ungrammaticality of quantificational noun phrases, which presuppose existence, in such constructions. The nominals which are licit in existential constructions correspond to the noun phrases qualified as weak by Milsark. In other words, the weak reading involves Property noun phrases (Ladusaw 1994; Dobrovie-Sorin 1997a,b; McNally & Van Geenhoven 1997; Dobrovie-Sorin & Beyssade 2004). This is supported by the inherent narrow scope of the argument of existential predicates: (55) There aren’t many students in the library. (i) ‘It is not the case that many students are in the library.’ (ii) # ‘There are many students such that it is not the case that they are in the library.’ (Van Geenhoven & McNally 2005, 893(24))

 The Layered DP: Form and meaning of French indefinites

In (55), many students only receives the interpretation in (i), i.e., narrow scope with respect to the negation. This is expected if the arguments of existential predicates are property-denoting arguments. The idea that some verbs may have floating types is extended to all verbs by Van Geenhoven & Mc Nally (2005).24 According to them, in principle, verbs can combine with noun phrases in any of their potential types. This means that each verb may have different lexical representations depending on the type of arguments they select (in opposition to an analysis which distinguishes verbs selecting entities from verbs selecting properties). Consider the following example: catch has the following lexical representations: a) and b) depending on the type of arguments it selects: (56) a. catch ⇒ λP λw λx ∃y (catchw (x,y) ∧ Pw (y)) where P is a variable of type 〈s,〈e,t〉〉, w of type s, and x of type e b. catch ⇒ λw λyλx (catchw (x,y))  (Van Geenhoven & Mc Nally 2005, 895(32,33)

(56a) is the lexical representation of catch when it selects a property. An existential operator is incorporated in the semantics of the verb, thus accounting for the existential entailment associated with the object argument y. (56b), in contrast, is the standard lexical representation of catch when it selects two entities. Van Geenhoven and McNally further classify verbs selecting property-denoting complements according to whether or not they supply an existential entailment to this argument. In other words, they aim at accounting for the characteristic of Property un-NPs mentioned in Section 2.3.1, namely that they may be existential or not. This characteristic is illustrated by the following intensional examples: (57) a. Marta is looking for a toy. b. Max wants a book for Christmas.  (Van Geenhoven & McNally 2005: 885(1–2))

Verbs like look for and want are known as absence verbs and they lead to ambiguity as the examples in (57) show. On one reading, a toy means any toy and a book any book. On the other reading, Marta would be happy if she found a particular toy and a particular book is the object of Max’s desire. The former interpretation is also known as the opaque reading as it gives rise to a failure of existential quantification.25 In other

.  Dobrovie-Sorin & Beyssade (2004: 52ff.), in contrast, propose that the essential criterion to determine existential predicates is the possibility to locate their arguments in the space. With predicates denoting events, the assertion of existence of the arguments is entailed because the existence of the event is asserted (Sasse 1987). .  Although the terms opaque and transparent are usually associated with different contexts, I stick to the terminology used by Van Geenhoven & McNally (2005).

Chapter 2.  un-NPs ‘a-NPs’ 



words, the sentences are felicitous even if no toy or no book exists. The second interpretation is known as the transparent reading as, in contrast with the opaque reading, it is associated with existence entailment. In the terms adopted here, the transparent reading corresponds to the existential reading, and the opaque reading to the non-existential reading. Furthermore, I have labelled Property noun phrases nominals like a toy and a book on their any-reading and S-referential noun phrases nominals which refer to a particular object. As seen above, Property noun phrases are transparent/existential when selected by a predicate incorporating an existential operator as in (56a) (and also (48)). To account for the opaque reading of Property noun phrases, Van Geenhoven and McNally propose a semantic decomposition of verbs à la Quine (1960). Quine argues that the complement of verbs of absence like seek, want and look for are interpreted as a proposition and therefore that the verb should be decomposed into a propositional attitude and a relation between individuals, i.e., for example try and find respectively. In other terms, seek means try to find. Indefinite complements of seek are existential (transparent reading) when they take wide scope with respect to try. In contrast, they are not existential (opaque reading) when they take narrow scope with respect to try. The latter reading is illustrated in (58): (58) look for ⇒ λP λw λx (look forw (x,P)) where look forw (x,P) = 1 iff in the world of evaluation w an individual x is trying in w to bring it about that there is an individual y in a world w’ which x (Van Geenhoven & McNally 2005: 896(34)) finds in w’ and which is P in w’

(58) makes explicit that absence verbs (i) denote a relation towards an intensional object, i.e., a non-existential property, and (ii) that they express an attitude towards a proposition. In other words, such verbs are intensional in two respects. What is crucial in such an analysis of opaque verbs is that the individual described by the propertydenoting argument is modally embedded. Applied to (59a), this means that although unicorns may not exist in the world of evaluation, they may exist in a modally embedded world w’, i.e., in the subject’s belief world.26 The lexical representation of look for in (58) thus accounts for the failure of existential entailment of complements of verbs like look for on their opaque reading. In other words, it explains why (59a) may be true even if no unicorn exists, contrary to(59b). (59) a. John is looking for a unicorn. b. John caught a unicorn. (Van Geenhoven & McNally 2005: 896(35–36))

.  Van Geenhoven and McNally acknowledge that a similar account of modal embedding is proposed by Farkas (1997), although she does not make use of property-denoting arguments.

 The Layered DP: Form and meaning of French indefinites

The transparent version of look for is reported below: (60) look for ⇒ λw λx λy (look forw (x,y)) where look forw (x,y) = 1 iff in the world of evaluation w there is an individual x and an individual y such that x is trying in w to bring it about that, in some (Van Geenhoven & McNally 2005: 898(41)) world w’, x finds y in w’

Contrary to (58), the complement of the verb is not modally embedded in (60). This entails its existence in the world of evaluation w. If the opaque and transparent readings of intensional verbs like look for may be distinguished by means of different lexical representations, it seems that an analysis along these lines should be extended to the other contexts allowing opaque un-NPs mentioned in Section 2.3.1, namely questions, subjunctives, imperatives and disjunction. As this is out of the scope of this book, I leave the question and come back to Property un-NPs and more precisely to what seems to be a restriction on their distribution.

2.3.3  The distribution of P-un-NPs The above sections show that the postulation of property-denoting arguments, hence of a third type of un-NPs, is well motivated. One characteristic of P-un-NPs which has not been mentioned so far is that their distribution seems to be restricted. Consider the following examples: (61) Negative sentences * Marie n’ a pas vu un fantôme. Marie ne has not seen a ghost ‘Marie hasn’t seen a ghost.’ (62) Without-clauses * Elle est revenue sans avoir acheté un livre. she is come.back without to.have bought a book ‘She came back without having bought a book.’ (63) Negative subject * Aucun invité n’ a vu un fantôme. no guest ne has seen a ghost ‘No guest has seen a ghost.’

The ungrammaticality of the above examples suggests that P-un-NPs are excluded from negative contexts.27,28 .  Note that the ungrammaticality of Property un-NPs in (61), (62) and (63) does not exclude that other argumental un-NPs, in particular S-referential un-NPs, are fine in such contexts. .  In contrastive examples such as (i), P-un-NPs are fine despite the negation. As such examples might involve different mechanisms, such as focalisation, they will be ignored here.

Chapter 2.  un-NPs ‘a-NPs’ 



To express the ‘Property meaning’ in the above examples, French uses a specific construction, namely de-NP. De-NPs differ from and should not be confused with the socalled ‘partitive’ du/des-NPs ‘of.the-NPs’ which contain the definite article (Chapter 3). Consider the following examples: (64) Negative sentences a. Marie n’ a pas vu de fantôme. Marie ne has not seen de ghost ‘Marie hasn’t seen a ghost.’ b. * Marie a vu de fantôme. Marie has seen de ghost (65) Without-clauses a. Elle est revenue sans avoir acheté de livre. she is returned without to.have bought de book ‘She came back without having bought a book.’ b. * Elle est revenue après avoir acheté de livre. she is returned after to.have bought de book (66) Negative subjects a. Aucun invité n’ a vu de fantôme. no guest ne has seen de ghost ‘No guest has seen a ghost.’ b. * Un invité a vu de fantôme. a guest has seen de ghost (67) Yes/no questions pas acheté de livre? a. N’ as-tu ne have you not bought de book ‘Haven’t you bought a book?’ b. * As-tu acheté de livre ? have you bought de book (68) Wh-(constituent) questions a. Qui n’ a pas rencontré de Bernois? who ne has not met de Bernese ‘Who hasn’t met a Bernese?’ b. * Qui a rencontré de Bernois? who has met de Bernese



(i)

Jean n’ a pas acheté un chat mais un chien. Jean ne has not bought a cat but a dog ‘Jean didn’t buy a cat but a dog.’

 The Layered DP: Form and meaning of French indefinites

(69) Intensional contexts a. Jean ne veut pas de chat. Jean ne wants not de cat ‘Jean doesn’t want a cat.’ b. * Jean veut de chat. Jean wants de cat (70) Intensional contexts a. Je ne trouverai pas d’ étudiant pour m’ aider. I ne will.find not de student for me to.help ‘I won’t find a student to help me.’ b. * Je trouverai d’ étudiant pour m’ aider. I will.find de student for me to.help (71) Intensional contexts a. Jean ne cherche pas de chat. Jean ne is.looking.for not de cat ‘Jean is not looking for a cat.’ b. * Jean cherche de chat. Jean is.looking.for de cat (72) Subjunctives a. Marie a insisté pour que Jean n’ invite Marie has insisted for that Jean ne invite.subj pas d’ étudiant. not de student ‘Marie insisted that Jean didn’t invite any student.’ b. * Marie a insisté pour que Jean invite d’ étudiant. Marie has insisted for that Jean invite.subj de student (73) Imperatives a. Ne va pas voir de médecin ! ne go not to.see de doctor ‘Don’t go see a doctor!’ b. * Va voir de médecin ! go to.see de doctor (74) Disjunctions a. Soit Jean n’a pas vu de voleur, either Jean ne has not seen de thief soit il est un menteur. or he is a liar ‘Either Jean hasn’t seen a thief, or he’s a liar.’ b. * Soit Jean a vu de voleur, soit il est un menteur. either Jean has seen de thief or he is a liar

Chapter 2.  un-NPs ‘a-NPs’ 



At first sight, (64)–(74) suggest that de-NPs represent a kind of Negative Polarity Item (NPI), in the sense that they must be in the scope of a negative operator to be licit (see Kayne 1981). This is illustrated in the above examples where the de-NPs are grammatical in the a examples, which contain a negation, but ungrammatical in the b examples, which do not contain a negation. The analogy with NPIs is supported by the fact that deNPs can be coordinated with qui que ce soit ‘anybody’, the prototypical NPI in French. (75) Je n’ai pas rencontré de professeur ou qui que ce soit (d’autre). ‘I didn’t meet a professor or anybody else.’  (Tovena, Déprez & Jayez 2004: 416(46a))

As only elements of the same type can be coordinated, (75) suggests that de professeur ‘of professor’ is an NPI on a par with qui que ce soit (d’autre) ‘anybody else’. If de-NPs must be in the scope of a negation, it accounts for a further restriction on their distribution, namely the fact that they have to be in object positions (see Kayne 1981 on the distribution of de-NPs). In (64)–(74), all the de-NPs are complements. Adding a negation to examples like (41a, c) (Section 2.3.1) does not make de-NPs licit in subject positions: (76) a. b.

* Marie a insisté pour que d’ étudiant ne Marie has insisted for that de student ne puisse pas entrer. could.subj not to.enter * Soit de voleur n’ est pas entré dans la maison, either de thief ne is not entered in the house soit on a oublié d’ éteindre. or one has forgotten of to.switch.off

As the subject de-NPs in (76) are not in the scope of the negation, they lead to ungrammaticality. However, if the generalisation is that de-NPs have to be in the scope of a negation, it predicts that post-verbal de-NPs subjects should be grammatical. This is borne out as (77) shows: (77) a. b.

Il n’ est pas venu d’ amis. it ne is not come de friends ‘There haven’t come any friends.’ * Il est venu d’ amis. it is come de friends

(Kayne 1981: 126(121b))

In (77), the canonical subject position is occupied by an expletive and the real subject, which is a de-NP, sits in a post-verbal position. However, only (77a) is grammatical supporting the claim that de-NPs have to be in the scope of a negation.

 The Layered DP: Form and meaning of French indefinites

The relation between de-NPs and negation cannot be completely similar to the relation between NPIs like any, anything … and the negation. The latter can notoriously span long distances (78a), contrary to the former (78b).29,30 (78) a. I didn’t know that Mary had met anyone. b.  *Je ne savais pas que Marie avait rencontré de prince. I ne did.know not that Marie had met de prince

On the basis of (78) I conclude that de-NPs are a kind of NPIs leaving for further research the differences with NPIs like anyone. Another characteristic of de-NPs is notable in (64)–(74). What these examples have in common, in addition to the fact that they are negative, is that they are opaque, i.e., they give rise to a failure of existential quantification. In other words, de-NPs are not existential in that they do not assert existence. This is another striking difference with the NPI in (78a). As de-NPs are not existential, the framework discussed in the preceding section implies that the predicates selecting them should have lexical representations analogous to intensional look for in the sense that they should denote a relation towards an intensional object, i.e., a complement which may not exist in the world of evaluation, although it may exist in a modally embedded world w’. (58) is repeated below as (79). (79) look for ⇒ λP λw λx (look forw (x,P)) where look forw (x,P) = 1 iff in the world of evaluation w an individual x is trying in w to bring it about that there is an individual y in a world w’ which x finds in w’ and which is P in w’ (Van Geenhoven & McNally 2005: 896(34))

However, as the ungrammaticality of the a examples in (67)–(74) shows, opaque contexts as such are not sufficient to license de-NPs. They also have to be in the scope of a negation. In other words, de-NPs are property-denoting arguments which occur in the scope of a negation and which lack existence entailment. The lack of existence entailment distinguishes de-NPs from P-un-NPs which may be existential or not. In addition, as de-NPs are licensed in negative contexts and as P-un-NPs are ungrammatical in such contexts, they are in complementary distribution. .  Thanks to Roberto Zamparelli and Lena Baunaz for pointing this out to us. Raising examples with verbs like seem however are fine. This might be due to the subjunctive mood.

(i)

Il ne semble pas que Marie aie rencontré de prince. it ne seems not that Marie has.subj met de prince ‘It doesn’t seem that Marie has met a prince.’

.  Thanks to Greg Ellison for discussing this example. He observes that replacing anyone with any prince is slightly marginal, unless the speaker conveys the extra pragmatic meaning that he doubts that it is true.



Chapter 2.  un-NPs ‘a-NPs’ 

2.4  Conclusion In this section, I take some theoretical and terminological stands, and argue for a three-way classification of argumental un-NPs in French. As the three categories I advocate correspond to the semantic types un-NPs may have, the typology is similar to Dobrovie-Sorin & Beyssade’s (2004) discussed in Chapter 1 Section 5.2.1. First, I acknowledge that un-NPs can be referential as proposed by Fodor & Sag (1982), although I suggest that reference should be understood as speaker’s reference as defined by Kripke (1977). In other terms, the interpretation of such un-NPs is determined by the context of utterance, namely by what/who the speaker ‘has in mind’. This type of un-NPs typically takes wide scope and is existential (but recall footnote 9 in which it is observed that indefinites may also co-vary with universal quantifiers as will be shown in several coming sections). The second type of un-NPs I argue for is quantificational un-NPs. Such indefinites represent a proportion of a set which has been introduced in the discourse, i.e., which is familiar. Quantificational un-NPs do thus not pick out a referent, contrary to S(peaker)-referential un-NPs. In addition, they are presuppositional in the sense that they presuppose existence. Presupposition of existence conflicts with assertion of existence, which characterises for example S-referential un-NPs. The third and most controversial type of un-NPs of the classification is labelled Property un-NPs. This is because such noun phrases are property-denoting arguments. One characteristic of noun phrases of this type is that they may be existential or not. When they assert existence, it is the predicate which supplies the existential operator. As they are not intrinsically existential, they typically take narrow scope, which is another property of theirs. Furthermore, their distribution is constrained in that they are ungrammatical in negative contexts. It is a construction specific to French, namely de-NPs, which occurs in such contexts. As de-NPs have to be in the scope of a negation, they have to be complements or post nominal subjects. Following Kayne (1981) I thus assume that they are (a kind of) Negative Polarity Items (NPI). Finally, we have seen that de-NPs cannot be existential. Admitting un-NPs of different semantic types is in opposition with an approach such as Montague’s, which treats noun phrases uniformly as generalized quantifiers, i.e., of type 〈〈e,t〉, t〉 (see also Barwise & Cooper 1981). It also requires that the semantics of predicates be refined. This is necessary for predicates to be able to select property-denoting arguments. Depending on the lexical representation attributed to a predicate, its property-denoting argument will be existential or not, as mentioned above. This is valid for objects but also for subjects, thus invalidating analyses such as Diesing’s according to which indefinites which are not intrinsically quantificational must be in VP, i.e., in the scope of her Existential Closure, and that indefinites which are outside the nuclear scope must be quantificational.

 The Layered DP: Form and meaning of French indefinites

More generally, the problem of compositional semantics highlights the fact that the study of noun phrases and the one of predicates are closely entangled and that therefore they should not be dissociated. In Section 3 I turn to the internal structure of un-NPs. In essence, the idea is that each type of un-NPs has a different internal structure which may be (partially) responsible for the properties discussed in this section.

3.  The structure of un-NPs 3.1  The structure of arguments 3.1.1  Arguments as DPs Before turning to the internal structure of un-NPs, let us ponder on the structure of arguments in general. A key position I adopt is that to be an argument, a noun phrase needs to have the DP layer projected. This essentially corresponds to Longobardi (1994, 2005), who assumes that this projection may be lexically filled or filled with an empty category.31,32 Part of his analysis is reported below. In Italian the determiner cannot be omitted in argument positions, i.e., as subject (80a), direct object (80b), prepositional object (80c) and inverted subject of either ergative or unergative predicates (80d). In non-argumental positions, however, no determiner is required (81). (80) a. b. c. d. 

* (Un/Il) grande amico di Maria mi ha telefonato. (a/the) great friend of Maria called me up Ho incontrato *(un/il) grande amico di Maria ieri. I met (a/the) great friend of Maria yesterday Ho parlato con *(un/il) grande amico di Maria ieri. I spoke with (a/the) great friend of Maria yesterday Ha telefonato/E’ venuto *(un/il) grande amico di Maria. he called up/came (a/the) great friend of Maria (Longobardi 1994: 612(8))

(81) a. Caro amico, vieni a trovarmi. dear friend, come to visit me

.  See also Szabolcsi 1983 and Stowell 1989. .  According to Longobardi (1994, 2001), a kind of α-raising is required cross-linguistically by the licensing condition on nominals, which states that “A ‘nominal expression’ is an argument only if it is introduced by a category D.” (Longobardi 1994: 620). Aboh (2003) proposes that α-raising may represent XP-movement or X-movement. These movements may be overt or covert.

Chapter 2.  un-NPs ‘a-NPs’ 



b. c. d.

Gianni è tenente. Gianni is lieutenant Diavolo! devil in abito lungo in long dress

(Longobardi 1994: 612–3, (7a), (8a), (9a), (10a))

(81a) illustrates a vocative, (81b) a predicate, (81c) an exclamation and (81d) a modifier probably assimilable to a predicative expression. On the basis of such observations, Longobardi proposes the following generalization: (82) A “nominal expression” is an argument only if it is introduced by a lexically filled D position. (ibid, 613, (11))

(82), however, is too strong as the grammaticality of (83) shows. (83) a. b.

Bevo sempre vino. I always drink wine Mangio patate. I eat/am eating potatoes

(ibid, 613, (12a,b))

In (83), vino ‘wine’ and patate ‘potatoes’ are fine although they are not preceded by any determiner.33 The distribution of such bare nouns is however subject to clear restrictions, as illustrated in (84). (84) a. b.

* Acqua viene giù dalle colline. water comes down from the hills * In questo ufficio marocchini telefonano sempre. in this office Moroccans always call up

(ibid, 616, (14a), (15a))

As bare nouns are grammatical in object positions (83), but not in pre-verbal subject positions (84), it suggests that they involve an empty category which must be lexically governed, on a par with other empty heads.34 Such a proposal accounts for the ungrammaticality of (84) but also for the grammaticality of (85). (85) Amico di Maria sembra essere Gianni. friend of Maria seems  to be Gianni

(ibid, 617, (17))

Whereas (84) is ruled out because the empty category in D cannot be lexically governed, (85), in contrast, is fine because amico di Maria ‘friend of Maria’ is not an argument, hence not a DP, i.e., no empty category in D. Consequently, (82) can be revised as in (86): .  As noted by Longobardi, such bare nouns, mass and plural respectively, appear to be similar to the so-called partitive article in Italian and French (see Chapter 3). .  I stick to the terminology used in the paper, although ‘lexical government’ has disappeared from recent frameworks.

 The Layered DP: Form and meaning of French indefinites

(86) A “nominal expression” is an argument only if it is introduced by a category D.  (ibid, 620, (21))35

The proposal that arguments are DPs is thus, in part, based on the distribution of bare arguments in Italian. Such a claim has further consequences, for example, for the analysis of proper names. If subjects are DPs whose head cannot be empty, as an empty category in this position could not be licensed, it predicts that proper names in examples such as (87) sit in D. (87) Gianni mi ha telefonato. Gianni called me up

(ibid, 622, (24a))

The distribution of proper names suggests that they move to D, i.e., that they are not generated there. Consider (88): (88) a. b. c. d.

Il mio Gianni ha finalmente telefonato. the my Gianni finally called up Il Gianni mio ha finalmente telefonato. the Gianni my finally called up Gianni mio ha finalmente telefonato. Gianni my finally called up * Mio Gianni ha finalmente telefonato. my Gianni finally called up

(ibid, 623, (28))

Proper names can be preceded by the definite article, in which case the possessive (or other adjectives) can either precede (88a) or follow the proper name (88b). If there is no article, the N-initial order (88c) is obligatory, as the ungrammaticality of (88d) shows. To account for examples like (88), Longobardi suggests that in Italian proper names move to D. What is interesting here is that movement to D occurs in contexts where D must be filled. The question that arises at this point is how the analysis mentioned above works in the framework adopted here, where nominal phrases are split into three domains, each of which contains several projections (Chapter 1 Section 4.4.2). If DP is replaced by several layers, I propose to reformulate (86) as (89): (89) A “nominal expression” is an argument only if it is introduced by one of the categories of the DP domain (the left periphery).

The revision of (86) as in (89) will be a building block in the rest of the analysis. The next step consists in determining how much of the left periphery is necessary to form

.  Note that the correlation between structural difference and argumental/predicative meaning is by no means novel. Zamparelli (2000: 124) mentions the following references: Bowers 1988, 1993; Hudson 1989; Mandelbaum 1994; developing Higginbotham’s (1987) ideas.



Chapter 2.  un-NPs ‘a-NPs’ 

an argument. What I will suggest is that the lowest projection is sufficient to form an argument. Intuitively, the difference with the clause would be that usually noun phrases are embedded in a clause, thus benefiting from its top-most layer, which somehow functions as a ‘lid’ closing off a proposition. In Section 3.2, I explore the latter hypothesis, namely the idea that the presence of the lowest projection of the nominal left periphery not only transforms the nominal phrase into an argument, but also that it does not have to be dominated by any other layer. The addition of further projections is only required to account for different interpretations of the noun phrase. Before developing such an analysis, I will report (some of) the main points of Zamparelli’s (2000) dissertation, as he also proposes to divide DP into several layers in a syntax-semantics mapping spirit.

3.1.2  Previous analysis One way to account for the different interpretations of noun phrases like un-NPs (Section 2) is to assume that it is their position in the sentence which determines their interpretation (Diesing 1992; de Hoop 1992; Kratzer 1989). This is the most wide-spread position in the literature. Another possibility is to put the burden on the internal structure of the noun phrases (Zamparelli 2000). This position has, in my opinion, not received enough attention in the literature and will be pursued in this chapter. The idea that there is a strict mapping between syntactic and semantic categories leads Zamparelli (2000) to propose that certain aspects of the interpretation of noun phrases are due to their internal structure. His work, however, is much more ambitious than this one in that it is not limited to indefinite noun phrases but includes all the noun phrases discussed by Milsark (1974) and in that it does not only study the syntax of these noun phrases but also their semantics, which is not addressed here. Like Milsark, Zamparelli examines there be sentences, and proposes to unify them with locative copular constructions. Zamparelli further investigates numerals and their position in a split DP, kind-constructions, some properties of partitives and adjectival phrases. As Zamparelli’s work is extremely multifaceted, leading to many proposals and predictions, it is impossible to do justice to it in a few paragraphs. I will thus report it in outline. Zamparelli builds on previous work on the structure of noun phrases (Abney 1987; Cinque 1990, 1992 and Longobardi 1994) and adopts Milsark’s (1974) classification in terms of strength (see Chapter 1 Section 5.1.1). Strong noun phrases consist of pronouns, proper names, and noun phrases introduced by every, most, both, the, this/ that, focused some/many/few and weak ones of a, some/many/few, unstressed numerals and determinerless noun phrases in a language like English for example. According to Zamparelli, strong and weak determiners sit in different projections in the nominal structure. The former are in the Strong Determiner Phrase (SDP) and the latter in the

 The Layered DP: Form and meaning of French indefinites

Predicative Determiner Phrase (PDP).36 In other words, SDP and PDP constitute the determiner system of a language. PDP also embeds a third functional layer, the Kind Determiner Phrase (KIP), which hosts the NP-proper and any restrictive adjective and which will not concern us directly here.37 The global structure is reported in (90) and the structure for strong, weak and predicative noun phrases in (91a), (91b) and (91c) respectively: (90)

SDP Strong Quantifiers PDP Weak Quantifiers KIP Adjectives

NP Noun complement

(91) a.

SDP SD every, both SOME, he John PD

PDP KIP (adj) N

b.

SDP SD



PDP

PD three, few, some …

KIP (adj) N

.  PDP represents the noun-phrase internal predicate. One of the functions of this layer is to turn nouns into properties (Zamparelli 2000: 16 ff.). .  Zamparelli uses the term ‘kind’ in the sense of Carlson (1977b).

Chapter 2.  un-NPs ‘a-NPs’ 

c.



PDP PD three, few, some …

KIP (adj) N

Zamparelli assigns the referential functions of noun phrases to the SDP and the predicative functions to the PDP. Consequently, only SDPs can function as arguments, whereas PDPs, which are pure property-denoting elements, function like adjectives and cannot appear in argumental positions. As for quantificational noun phrases, they have the same structure as referential ones, except that they undergo Quantifier Raising (QR) at Logical Form (LF). That some noun phrases can have a strong and a weak reading is due to the position of the determiner, which can sit in SDP or PDP as illustrated in (91a-b). These projections can also be empty provided that they are appropriately licensed. For example, weak noun phrases project an empty SDP (91b) whose head must be in the scope of a generic operator (Krifka et al. 1995) or of Existential Closure (Diesing 1992) for the structure to be interpretable. As referential (SDP) and predicative (PDP) noun phrases represent different maximal projections, they are expected to have different distributions, different patterns of extraction… For instance, Quantifier Construal (QC) (Heim 1982) – the operation which moves the quantifier out of a raised NP – can only extract quantifiers from SDP. Other consequences of a structure like (90) are analysed by Zamparelli. Among them, the impossibility to coordinate noun phrases of different type, i.e., SDP, PDP and KIP, the prediction that possessives co-occurring with a numeral as in John’s four wishes lose their predicative reading, the possibility for some Italian wh-quantifiers to appear under a determiner, leading to a meaning shift, and various properties of the Italian ne, for example, that it can only be extracted from under weak quantifiers or that its trace must be governed by a quantifier. For more details, the reader is referred to the original work. A crucial difference between Zamparelli’s work and the one presented here is that I consider the distinction strong/weak inappropriate in that it assimilates definite noun phrases to quantificational ones (see Sections 2.1 and 2.2 which discuss referential and quantificational noun phrases respectively). Although Milsark’s tests for strength, repeated below from Chapter 1, put proper names and pronouns on the side of strong noun phrases, which are quantificational, I consider that the former are referential. Proper names are traditionally treated as logical constants, and not as quantifiers. The reason for Milsark to consider that proper names and pronouns are strong noun phrases is that they are ungrammatical in existential sentences (92a) and grammatical with individual predicates (92b), on a par with quantifiers (93).

 The Layered DP: Form and meaning of French indefinites

(92) a. * There is John/he at the door. b. John/he is intelligent. (93) a. * There is every dog in the garden. b. Every dog is intelligent.

Although John and he pattern like every in existential sentences and with individual predicates, other contexts suggest that they should be distinguished. Recall for example the Law of Contradiction and the Law of Excluded Middle discussed in Sections 2.1 and 2.2. Another context is topicalisation which typically excludes quantificational noun phrases: (94) a. John, I saw (him) on Monday. b. * Every boy, I saw (him) on Monday.

As shown in (94), John is fine in a topic position contrary to every boy. Another wellknown difference is that referential noun phrases, in the traditional sense, do not have scope properties, in contrast with quantificational noun phrases. As for the structures proposed by Zamparelli and reported in (91), there are two ­aspects which seem problematic. The first one is that only SDPs may be arguments. However, as discussed at length in Section 2, argumental un-NPs may have three readings, which cannot be encoded in the same projection. In effect, the syntax-semantics interface adopted here predicts that noun phrases with different interpretations have different structures (Chapter 1 Section 4). Related to this issue, is the fact that if referential and quantificational noun phrases have the same structure, it is unclear how Quantifier Raising may distinguish between these two types of noun phrases. Informally, the question is how QR ‘knows’ which noun phrases it should move. If, in contrast, referential and quantificational nominals represented different categories, hence distinct projections, QR could ‘recognise’ quantificational elements. This is what I will explore in this chapter. As for the similarity of judgement for (92a) and (93a), it could simply be due to the fact that there be sentences assert existence, which is incompatible with referential and quantificational noun phrases which presuppose existence.38 The second issue concerns weak noun phrases (a subset of property-denoting arguments). According to Zamparelli, they are SDPs with an empty head. However, intervention effects as discussed in Chapter 1 Section 4.3 suggest that these noun phrases are ‘smaller’ than the other two types of arguments. This view is developed in the next sections. In short, the structure of noun phrases should not only distinguish between referential and quantificational noun phrases but also between property-denoting arguments and other arguments.

.  Referential here is used in the traditional sense not in the sense S(peaker)-referential (Section 2.1.2).

Chapter 2.  un-NPs ‘a-NPs’ 



As a point of departure, let us focus on the third type of argumental un-NPs discussed in Section 2, namely Property un-NPs, which – I argue – involve the projection representing the boundary between predicative and argumental noun phrases.

3.2  The lowest projection of the left periphery 3.2.1  The structure In Section 3.1.1, it has been suggested that what distinguishes argumental noun phrases from predicative noun phrases is their structure. The former have the DP layer projected contrary to the latter (Longobardi 1994, 2005). This structural difference is a building block of the analysis developed here. However, it has been refined as in (95) to take into account the assumption that the structure of nominal expressions is split into three domains (Chapter 1 Section 2.2.2). (95) A “nominal expression” is an argument only if it is introduced by one of the categories of the DP domain (the left periphery).

A proposal like (95) implies that the left periphery only concerns arguments. In other words, I propose that predicative noun phrases do not have any projection of the left periphery. This means that they are truncated at the level of the highest layer of the inflectional domain, i.e., they are Number Phrases (Chapter 1 Section 4.4.2). In other words, the structure of predicative nominals consists of the NP-shells and the ­inflectional domain and not simply of the NP as often assumed in the literature. That projections involved in agreement phenomena are projected in the structure of ­predicative noun phrases is supported by examples like (96): (96) a. b. c.

Mon fils est un petit coquin. my son is a.masc little.masc rascal.masc ‘My son is a little rascal.’ Ma fille est une petite coquine. my daughter is a.fem little.fem rascal.fem ‘My daughter is a little rascal.’ Mes enfants sont des petits coquins. my children are of.the little.plur rascals.plur ‘My children are little rascals.’

In (96a), the predicative noun phrase un petit coquin ‘a little rascal’ agrees with mon fils ‘my son’, i.e., it is masculine singular. In (96b), une petite coquine ‘a little rascal’ agrees with ma fille ‘my daughter’, which is feminine singular, and in (96c), des petits coquins ‘of. the little rascals’ agrees with mes enfants ‘my children’ i.e., it is plural. In other words, (96) shows that predicative noun phrases comprise the layers responsible for agreement, i.e., the projections of the inflectional domain. It further supports the idea that the distinction argument-predicate should not be stated in terms of presence vs. absence of articles but in terms of structure, i.e., presence vs. absence of some layer(s). If NumP characterises predicates, it corresponds to Zamparelli’s (2000) Predicate Determiner Phrase (PDP).

 The Layered DP: Form and meaning of French indefinites

According to (95), the projection of one layer of the left periphery transforms a predicate, i.e., a NumP, into an argument. What I propose is that such argumental un-NPs are Property un-NPs (P-un-NPs). This means that their structure consists of NumP plus the lowest layer of the left periphery. As this projection is characteristic of property-denoting arguments, I will label it Property Phrase (PropP). In other words, the role of PropP is to transform predicates into arguments. The structure I advocate for P-un-NPs is thus as follows: (97)

PropP Prop′

Spec

NumP

Prop un

Num′

Spec Num ti

… livre

The head of PropP hosts the article un ‘a’, which has moved from Num (or from a lower position (see Chapter 3)). Informally, it is the movement of this element from Num to Prop which activates the projection of PropP (see Section 3.3.1 for a more technical alternative). Arguments for the structural distinction between arguments and non-arguments have been provided in Section 3.1.1. If this line of analysis is on the right track, it suggests that predicative noun phrases (96) and property-denoting arguments (98), should be distinguished, despite their analogous interpretation. (98) Marie cherche un clown. Marie is.looking.for a clown ‘Marie is looking for a clown.’

Such a distinction is supported by the asymmetry among predicates illustrated below. Compare (99) and (100). (99) a. Mary is pretty. b. Mary is in the garden. c. Mary is a woman. (100) a. * John is looking for pretty. (Van Geenhoven & McNally 2005: 897(37)) b. * John is looking for in the garden. c. John is looking for a woman.

Chapter 2.  un-NPs ‘a-NPs’ 



(99) provides examples with adjectival, prepositional and nominal predicates and (100) shows that these predicates lead to different grammatical judgements in intensional contexts typically hosting property-denoting arguments. In other words, only nominal predicates can function as arguments, in contrast with adjectival and prepositional predicates. This means that verbs like look for cannot simply select any predicative element, but that their complements, even with a property interpretation, have to have some ‘argumental’ feature. This is precisely the role of the Property Phrase advocated in (97). In other words, PropP is characteristic of nominal predicates functioning as argument and does not concern other predicates, whatever they may be. The categorical difference between predicative noun phrases (NumPs) and property-denoting arguments (PropPs) resulting from the analysis developed here allows us to distinguish between these two 〈e,t〉-type expressions. In other words, intensional verbs select property-denoting arguments, i.e., PropPs and not NumPs. Other property-denoting arguments are de-NPs found in complementary distribution with P-un-NPs in negative contexts. The following example is repeated from Section 2.3.3. (101) Negative sentences a. Marie n’ a pas vu de fantôme. Marie ne has not seen de ghost ‘Marie hasn’t seen a ghost.’ b. * Marie a vu de fantôme. Marie has seen de ghost

By analogy to un-NPs, the structure of de-NPs should be similar to (97), i.e., a ­Property Phrase. However, as their distribution is restricted to negative contexts, it suggests that contrary to P-un-NPs, de-NPs contain an empty category which needs to be licensed. Kayne (1975) proposes that de-NPs should be analysed as (NP zero elementde-articleless NP) and Kayne (1981) suggests that the zero element is an empty QP which needs an antecedent and which has to be properly governed. The structure emerging from these considerations is given below: (102)

PropP Prop′

Spec ec

NumP

Prop dei

Num′

Spec Num ti

…ti livre

 The Layered DP: Form and meaning of French indefinites

(102) is analogous to (97) except that the specifier of PropP is filled by an empty category and the head with de ‘of ’. De-NPs and their structure are further discussed in Chapter 3 Section 3.2.2 where they are compared to nominal expressions with overt quantities like beaucoup de ‘much/many’ and peu de ‘little/few’. In the next section, I show that the structures advocated above account for various properties of Property noun phrases.

3.2.2  Some consequences Crucially, the structure of P-un-NPs proposed in (97) does not contain any empty category. This predicts that P-un-NPs do not have to be in the scope of an operator or of a potential governor and therefore that their distribution is not restricted to such an environment.39 As (103) shows, P-un-NPs may function as subjects and objects. (104a) illustrates a subject P-un-NP in French. These examples are repeated from Section 2.3.2. (103) a. λP λQ ∃x∃y [x wash y ∧ Q(x) ∧ P(y)] b. λP ∃x [x sleep ∧ P(x)] c. λP ∃x [x are available ∧ P(x)] (Dobrovie-Sorin & Beyssade 2004: 44(26’)) (104) a. [at a party] Un invité a fini ce vin. a guest has finished this wine ‘A guest finished this wine up.’ b. Jean cherche un chat. Jean is.looking.for a cat ‘Jean is looking for a cat.’

The presence of an empty category in P-un-NPs would restrict their distribution. As seen in the preceding section, this is what is observed for de-NPs, but not for P-un-NPs (for more on de-NPs see Chapter 3 Section 3.2.2). Although in French the distribution of property-denoting arguments is not restricted, it seems that in other languages they are (Diesing 1992 for Germanic languages). This suggests that in these languages there is an empty category in their structure, on a par with bare nouns in Italian (Section 3.1.1). If this is correct, this empty category has to be lexically governed in Longobardi’s (1994) terms. .  Note that this also applies to property-denoting du/des-NPs discussed in the next ­ chapter. Consider the following examples which show that these noun phrases are fine in subject ­positions: (i) [at a party] Des invités ont fini ce vin. of.the guests have finished this wine (ii) Du café traînait sur la table. of.the coffee was.lying on the table When du/des-NPs are explicit partitives, i.e., complements of fragmentative verbs, however, I propose that they contain an empty category (Chapter 3 Section 3.3).



Chapter 2.  un-NPs ‘a-NPs’ 

Furthermore, (97) does not imply anything about the existentiality of P-un-NPs. This means that this structure allows P-un-NPs to be existential or not. The existentiality of certain P-un-NPs is due to an existential operator provided by the predicate (Section 2.3.2), i.e., it is independent from the nominal structure. As P-un-NPs are not intrinsically existential, their grammaticality in existential constructions such as There is a Pope is expected. This is because P-un-NPs do not conflict with the there-be part of the sentence considered as an existential operator by Milsark (Chapter 1 Section 5.1.1). In addition, as (97) does not contain any quantificational element, such noun phrases should be interpreted in situ (as observed by Enç 1991 for her non-specifics and by Diesing 1992 for her non-presuppositional indefinites). This means that by themselves they do not have scope properties although the term narrow(est) scope is often used to describe their interpretation. The impression that P-un-NPs take narrow scope might be due to the existential operator supplied by the predicate. However, narrow scope is usually attributed to all property-denoting noun phrases, even when they are not existential. In this case, the term ‘scope’ is clearly misleading as they do not enter any scope relations. Therefore, my position is that to have scope properties, a nominal expression must have some quantificational interpretation. As it is not the case of P-un-NPs, they should not be described in terms of scope relations but simply be attributed an ‘in situ interpretation’. The absence of scope properties of P-un-NPs sharply contrasts with the quantificational expressions discussed in the next section as the latter undergo Quantifier Raising, i.e., move to a scope taking position at LF. Finally, as already mentioned in the preceding section, attributing propertydenoting arguments a structure which differs from the one of predicative noun phrases has important consequences for selection. In effect, it suggests that selection should be formulated in terms of categories and not of semantic types. This means that ­ intensional verbs like look for for example do not select an argument of a particular semantic type, like 〈e,t〉 but an argument of a certain category, like a Property Phrase. This is predicted by (95) which does not allow NumPs to function as ­arguments. The above remarks on P-un-NPs, except for the ones concerning their distribution, extend to de-NPs. In sum, it seems that many of the characteristics of P-un-NPs follow from their internal structure and not from their position in the sentence for example.

3.3  The Quantifier Phrase 3.3.1  The structure In section 3.2.1, I proposed that the projection differentiating arguments from predicates is the Property layer, PropP. As quantificational un-NPs (henceforth Q-un-NPs) are argumental, I assume that the Property Phrase is part of their structure as a building block of their interpretation. In addition, I propose that quantificational phrases

 The Layered DP: Form and meaning of French indefinites

are characterised by a projection dominating PropP, namely a Quantifier Phrase (QP) encoding the feature quantificational. This means that the left periphery of quantificational noun phrases is richer than the one of property-denoting noun phrases: (105)

QP > PropP > NumP…

The idea is that Q-un-NPs differ from P-un-NPs in that they comprise an additional projection and that it is this layer which makes them available for Quantifier Raising (QR). The details of the proposed structure are reported below: (106)

QP Q′

Spec

PropP

Q uni

Prop′

Spec

NumP

Prop ti

Num′

Spec Num ti

… livre

The role of a QP projection is to encode the quantificational feature responsible for its quantificational reading. Crucially, it is the presence of this projection which distinguishes between noun phrases which undergo Quantifier Raising (QR) at Logical Form (LF) and those which do not, like P-un-NPs. Informally, it means that the QR mechanism will look for a structure containing a QP layer and will move it to a higher position at LF. As the QP layer encodes quantification, I assume that it is the category of all quantifiers, in particular of every/each/most …-NPs which I take to be the prototypical quantifiers.40

.  But see Beghelli & Stowell (1997: 102ff.) who observe that “every exhibits some degree of quantificational variability”. This is why, according to the authors, every-NPs are not always distributive contrary to other quantifiers like each-NPs and can or sometimes must be in the scope of a negation.



Chapter 2.  un-NPs ‘a-NPs’ 

Evidence for a QP dominating PropP comes from intervention effects (Chapter 1 Section 4.3). Recall that I assume, following Starke 2001, that only elements which are interpretatively heavier may cross other elements of the same class. The relevant formalisation is repeated as (107): (107) αβ … α … αβ

(Starke 2001: 8(16b))

If interpretative features are encoded in different projections as suggested by a cartographic approach (Chapter 1 Section 4.2), α and β in (107) represent distinct layers. Put differently a nominal expression with more projections should be able to cross another one with fewer projections. In other words, Q-un-NPs, which are QPs with the structure proposed in (106), should be able to cross P-un-NPs, which are PropPs and which are structurally smaller. Consider (108). (108) a. Sm cellists played every suite today. b. Mny cellists played SOME suite today. c. Tw cellists played SOME suite today.

(Diesing 1992: 63 (8))

The indefinites with the deleted vowels represent weak indefinites in their weak/ cardinal reading, i.e., property-denoting indefinites, and the ones in capital letters indefinites in their strong reading, i.e., QPs, like every-NPs. As the QPs every suite and some suite in (108) take scope over sm cellists, mny cellists and tw cellists, it means that they can move over PropPs at LF without giving rise to intervention effects. Replacing sm cellists by a P-un-NPs and every suite by a Q-un-NPs in (108a) for instance yields analogous interpretations: (109) [PropP Un violoncelliste] a joué a cellist has played [QP une suite (= une des suites)] aujourd’hui. a suite (= one of.the suites) today

As the Q-un-NP can cross the P-un-NP at LF, it shows that the former is interpretatively heavier than the latter. In other terms, QPs can pass PropPs because they are structurally heavier, i.e., because they involve one more projection, namely QP above PropP. Had they the same structure as PropPs, they would not be able to cross PropPs at LF. Additional scope interactions between Q-un-NPs and other QPs are examined in Section 3.3.3. As for the details of (106), the idea is that un ‘a’ head-moves to Q from Num (possibly from a lower head, cf. Chapter 3) through Prop. Informally, this means that a layer is only projected when required by the interpretation of the noun phrase and that it is movement, of un ‘a’ in (106), which leads to the projection of an additional layer. Such a movement, from NumP upwards leading to a ‘specific’/strong reading, has

 The Layered DP: Form and meaning of French indefinites

been proposed by Borer (2005a).41 Technically, the fact that QP and PropP combine in the structure of Q-un-NPs can be accounted for by means of the ‘copy and remerge’ mechanism proposed by Shlonsky (2005) for the clausal left periphery in order to account for the fact that selection is local, even when CP is split into different projections. The idea is that CP is represented by a single head, C. This head is equivalent to Rizzi’s (1997) Force and is endowed with several sequenced features 〈T1 … Tn〉. These features correspond to the heads of the clausal left periphery, i.e., Fin, Foc and Top (see Chapter 1 Section 4.4.2.2). They are lexically pre-compiled and are projected in the structure only when needed, by a ‘copy and remerge’ mechanism. Each copy of C can activate a single feature. To access these features, the computational system must respect the order of compilation.42 In an analysis like the one developed here, where the interpretation of noun phrases is built ‘upwards’, the head of the left periphery encoding a sequence of ordered features would be the lowest one (in opposition to the highest, i.e., Force, as proposed by Shlonsky for the clause). In other terms, Prop would have the sequence of features Prop [Prop, Q,…] and it would be copied and remerged with an additional feature expressing quantification to yield Q-un-NPs. In short, Prop would be copied and

.  Note however that in Borer’s framework movement occurs from NumP (#P) to DP. The interpretation of such elements is presuppositional as 〈e〉d is not bound by existential closure. Furthermore, depending on the properties of the determiner, specifically on whether or not it is quantificational, the resulting expression is, or is not, a quantifier (i.e., it is a quantifier if every assigns range to 〈e〉d; it is not a quantifier if a proper name or a definite article (2005a: 145). assigns range to 〈e〉d) In addition, “the resulting expression becomes like a proper name or definite description, taking (descriptively) widest scope” (ibid). What seems problematic to me in this account is that quantifiers typically cannot take widest scope in that they are blocked in islands. Furthermore, as movement from NumP (#P) to DP is also proposed for the specificity marker xit in Hebrew (2005: 150 ff.) (our S-referential reading), it suggests that such noun phrases are ­presuppositional. However, these indefinites do not presuppose existence but assert existence (Section 2.2.3 and Chapter 3 Section 2.1). .  “Nous proposons que CP est constitué d’une tête unique, C, équivalent à Force de Rizzi 1997 et dotée d’un ensemble de traits ordonnés 〈T1 … Tn〉 (voir aussi Laenzlinger 2000). Tandis que Force = C, les traits en question correspondent aux têtes de la périphérie gauche, tel que Fin, foc et Top. Nous admettons que ces traits sont lexicalement précompilés et si besoin est, projetés dans la structure syntaxique. L’accès aux traits par le système computationnel doit suivre l’ordre de compilation. En particulier, la syntaxe exploite le mécanisme de ‘copiage et réinsertion’ (copy and remerge). Dans chaque copie de C, un seul trait est activé” (Shlonsky 2005: 1–2).

Chapter 2.  un-NPs ‘a-NPs’ 



remerged as Q. Informally, it would mean that the property-denoting un ‘a’ is semantically enriched into a quantificational un ‘a’.43 A ‘copy and remerge’ mechanism can account for the fact that PropP is projected when QP is and justify the movement of the article. I will however remain agnostic concerning this mechanism, mainly because it raises questions which are far beyond the scope of this book, namely on what is in the lexicon and on the syntax-lexicon interface.

3.3.2  Scope islands If Q-un-NPs belong to the same category as other quantificational noun phrases, i.e., QPs, it predicts that they should be sensitive to islands. This means that we expect them to be blocked in relative clauses, complements of a noun and unextraposed sentential subjects which represent scope islands, often associated with syntactic extraction islands (Fodor & Sag 1982; Lakoff 1970; Postal 1974; Rodman 1976; May 1977; VanLehn 1978; see also Reinhart 1997; on islands in general see Szabolcsi 1998–2002 and the references therein).44 Scope islands are illustrated in (110) and syntactic islands in (111): (110) a. A doctor will examine the possibility that we give every new patient a tranquilizer. b. A doctor should worry if we sedate every new patient. (Reinhart 1997 (3)) (111) a. * Which patients will a doctor examine the possibility that we give e a tranquilizer? b. * Which patients should a doctor worry if we sedate e?  (Reinhart 1997 (4))

.  In an account of polysemous elements like Ramchand’s (2005, 2006a), it would mean that un ‘a’ might have different categorical tags (Chapter 1 Section 4.2). Property-denoting un would have the label ‘PropP’ and quantificational un the label ‘QP, PropP’, where both layers have to be projected. .  Farkas (1981) shows that universals like every-NPs are not only island-bound but also clause-bound. Consider the following example:

(i)

a. John told a reporter that Peter lives in every French town. b. Where did John tell a reporter that Peter lived? This is the town that John told a reporter that Peter lived in. (Farkas 1981: 60(2–3))

As shown by the grammaticality of the wh-movement in (ib), every French town in (ia) is not in a syntactic island. The sentence should therefore be ambiguous between a reading in which a reporter takes scope over the every-NP and a reading where it is in the scope of the every-NP. However, only the former obtains. In other words, the every-NP cannot have scope over a reporter which is outside its clause.

 The Layered DP: Form and meaning of French indefinites

The sentences in (110) do not allow every new patient to be interpreted with wide scope over a doctor, even when this indefinite is interpreted as property-denoting. In other words, the universal quantifier is trapped in the island on a par with the whelement which patients in (111). Although the reasons for QPs to be blocked in scope islands are poorly understood, we expect Q-un-NPs to be subject to the same restrictions. This seems to be the case as (112), which is parallel to (110b), shows: (112) [At a hospital 12 new patients are waiting to see a doctor (i.e., a doctor or other/ any doctor). One patient faints and a relative of his says to a doctor walking by]: ‘A doctor should worry if a patient faints!’

Although the judgements are delicate, it seems that a patient in (112) cannot take scope over a doctor. This shows that a patient in (112) is a QP parallel to every patient in (110).

3.3.3  Scope interactions If Q-un-NPs are QPs on a par with other quantifiers, it means that they undergo Quantifier Raising (QR) at Logical Form (LF). Consequently, they should interact with other quantificational elements. Let us look at some examples involving universal quantifiers and Q-un-NPs. (113) a. b.

Tous les étudiants ont lu un livre. all the students have read a book Chacun des étudiants a lu un livre. each of the students has read a book

(∀ > unQ); *(unQ > ∀)

(114) a. b.

Un étudiant a acheté tous les journaux. a student has bought all the newspapers Un étudiant a acheté chaque journal. a student has bought each newspaper

(unQ > ∀); *(∀ > unQ)

(∀ > unQ); *(unQ > ∀) (Baunaz 2005b, (15a), (16a))

(unQ > ∀); *(∀ > unQ)

In (113), the universal quantifier in French must have scope over the Q-un-NP (see Baunaz & Cattaneo 2005) whereas in (114) it is the other way round.45 It seems therefore that it is their surface order which determines the scope of the two quantificational noun phrases.46 This suggests that at LF, QPs, i.e., Q-un-NPs and universal quantifiers,

.  The reverse order is possible but would involve an S-referential un-NP (See Section 3.4). .  Observe however that the facts are more complex. It seems that examples like (113a) have several readings. The students could have read the same book, out of a predefined selection of books, or a different book, out of the same selection. For instance, out of a list of books including Hamlet, Moby Dick, Quartet, Sula … John, Mary and Jane could have read Hamlet or John could have read Hamlet, Mary Moby Dick, Jane Quartet … The first situation, where all the students

Chapter 2.  un-NPs ‘a-NPs’ 



cannot cross each other, reminding us of Relativized Minimality (Rizzi 1990). This is exactly what Baunaz & Cattaneo (2005) propose to account for the fact that the Q-un-NPs cannot cross the universal quantifiers in (113). Examples like (114) however are not addressed.47 In the framework adopted here, it means that elements of the same category, namely QPs, cannot cross each other. If QPs cannot cross other QPs and if this applies to Q-un-NPs and every-NPs as seen in (114), the question which arises is how the indefinite can take scope over the every-NP in (115). (115) Every student overheard three reports that a Gila monster of mine drowned.  (Diesing 1992: 90(54))

The reading of (115) which is of interest here is the one where a Gila monster of mine takes scope over every student. If the discussion of (113) and (114) is on the right track, it suggests that the indefinite in (115) is not a QP, because if it were it could not move past the every-NP. This is a very surprising conclusion especially if we assume that all noun phrases with scope properties, including un-NPs which cross universals at LF to take wide scope, are quantificational and that the quantificational feature is encoded in a QP layer. This paradox will be addressed in the next section.

read Hamlet, has two readings depending on whether the type or the token is meant: if the students read a different exemplar of Hamlet each, it is a type reading whereas if they take turn and read the same exemplar of Hamlet, it is a token reading. In terms of scope, it seems that in the latter reading, the Q-un-NP should have scope over the universal, in contrast with (113). If this is on the right track, it is unexpected. The analysis of all the readings of examples like (113a) has to take into consideration the event, whether there is a single one or not, whether it is distributive or not and also the Type-token difference. Note however that the Type-token problem does not arise with a different predicate:

(i)

(about a basket of apples): Tous les enfants ont mangé une pomme. (∀ > unQ); *(unQ > ∀) all the children have eaten an apple

In (i), it is impossible that the students ate the same apple. However they could have eaten an apple of the same kind, say a Golden. In other words, the token reading is excluded contrary to the type reading. .  Note that if the scope relations illustrated in (113) could be accounted for by assuming that every-NPs must for some reason target a position higher than the un-NP at LF, this is not the case in (114). If there is a hierarchy of projections hosting quantificational elements at LF (à la Beghelli 1995, see Section 3.4.3) and if every-NPs are higher than quantificational un-NPs as suggested by (113), the scope relations in (114) remain mysterious because the every-NP should be able to bypass the quantificational un-NP. This is why an analysis in terms of intervention effects or lack thereof is necessary even if quantificational elements are organised hierarchically at LF.

 The Layered DP: Form and meaning of French indefinites

3.4  The S-Reference phrase 3.4.1  The structure What characterises S-referential un-NPs (SR-un-NPs) is the fact that they encode speaker’s reference (Section 2.1). I thus propose that the structure of SR-un-NPs corresponds to the one of Q-un-NPs with an additional layer, related to speaker’s reference and labelled SRefP for Speaker’s Reference Phrase: (116) SRefP > QP > PropP > NumP48 …

The idea that the highest projection contains a feature related to referentiality is not new. Longobardi (1994) proposes that D hosts a (± Referential) feature, which differentiates ‘object-denoting’ (i.e., pronouns and proper names) from ‘kind-denoting’ (i.e., common nouns) and which must be ‘checked’. Zamparelli (2000: 153) also observes that [t]he idea of associating ‘referentiality’ to a section of the noun phrase distinct from the noun itself is once again not a new one. In syntax, it is typically associated with the issue of the position of personal pronouns, going back, at the very least, to a debate between Postal [1966] and Sommerstein [1972] on the nature of determiners and pronouns.

SRefP is the building block which distinguishes SR-un-NPs from Q-un-NPs. It allows the former to pick out a referent, contrary to the latter. Furthermore, I suggest that it is projected above QP. The details of the structure are reported in (117): (117)

SRefP Spec

SRef′ QP

SRef uni Spec

Q′ Q ti

PropP Spec

Prop′ Prop ti

NumP Spec

Num′ Num ti

… livre

.  Recall the structure proposed by Ihsane & Puskás (2001) for the nominal left periphery (Chapter 1 Section 4.4.2.2): DP > TopP > FocP > DefP. Although the arguments in favour of this structure differ from the analysis proposed here, TopP and FocP seem to be analogous to SRefP and QP in (116) respectively in the sense that the latter but not the former is quantificational.

Chapter 2.  un-NPs ‘a-NPs’ 



The postulation of a layer above QP is motivated by intervention effects (Chapter 1 Section 4.3), which only occur with elements of the same type (118a, b) or when the intervener is interpretatively heavier than the moving element (118c) (Starke 2001). When the moving element however is heavier than the intervener, no intervention effects arise (119). (118) a. * αi… αj… αi b. * αβi… αβj… αβi c. * α … αβ … α (119) αβ … α … αβ

(Starke 2001: 8(16a, 17b, 15a)) (Starke 2001: 8(16b))

In (119), the movement of αβ is not blocked by α because it has some additional property. In Section 3.3.1, I have proposed that (119) can be reformulated in terms of structure. Consequently, I have argued that, as Q-un-NP can move past P-un-NPs at LF, they must involve one more projection. The same reasoning applies to SR-un-NPs. In other terms, as they can pass QPs at LF, their structure must be more complex.49 Recall (115) from the preceding section: (120) Every student overheard three reports that a Gila monster of mine drowned.  (Diesing 1992: 90(54))

Although this example has several readings, the one which is relevant here is the one where a Gila monster of mine takes scope over every student. The absence of intervention effects suggests that in such contexts the indefinite is not a QP, otherwise it could not cross the every-NP, but a bigger structure, like (117). Put differently such examples support the proposal that the structure of SR-un-NPs is heavier than the one of Q-unNPs in the sense that there is a projection above QP. Conversely, if this analysis is on the right track, it predicts that every-NPs cannot pass SR-un-NPs at LF. (121) shows that this seems to be the case. (121) A man gained every student’s esteem: the Dean.

Every student in this example cannot take scope over a man. In other words, (121) cannot mean ‘for every student x, there is a man y, such that y gained x’s esteem’, which supports the analysis developed here. In effect, this is an illustration of (118c). (122) is a formalisation of intervention effects or lack thereof for un-NPs. Using syntactic categories, we get (123). (122) a. b. c.

* Q-un-NPs… Q-un-NPs … Q-un-NPs * Q-un-NPs … SR-un-NPs … Q-un-NPs SR-un-NPs… Q-un-NPs… SR-un-NPs where SR-un-NPs contain a QP layer, on a par with Q-un-NPs

.  See also Baunaz & Cattaneo (2005). They observe that what they call ∃QPRange (i.e., Q-un-NPs) cannot cross noun phrases which are specific (i.e., S-referential in the terminology adopted here) because the latter have ‘something more’, although this has not been made explicit.

 The Layered DP: Form and meaning of French indefinites

(123) a. b. c.

* QP… QP… QP * QP… SRefP… QP SRefP… QP… SRefP where SRefPs contain a QP layer

These generalisations show that intervention effects involve features of the left periphery. Whether phi features like person, number, gender or case also play a role has to be determined.50 Recall that the classes in which locality effects take place have been defined as below by Rizzi (2002) (Chapter 1 Section 4.3). (124) a. Argumental: person, number, gender, case b. Quantificational: Wh, Neg, measure, focus… c. Modifier: evaluative, epistemic, Neg, frequentative, celerative, measure, manner… d. Topic (Rizzi 2002, (61))

If interpretative features like quantificational and S-referential play a role in intervention effects among indefinites, (124a) should be modified as (125): (125) Argumentalindef: S-referential, quantificational, (person?, number?, gender?, case?)

As for the position occupied by SR-un-NPs at LF, it is generally the highest one (but see Section 3.4.3). The reason is that, in their most frequent reading, they have scope over any other potential scope-bearing element in their clause. Although I will not address the question of the target site of SR-un-NPs at LF here (see section 3.4.3), some remarks on Quantifier Raising (QR) are in order. The traditional QR mechanism moves quantifiers from their base positions to distinct scope positions in any order at LF (May 1977, 1985). A major criticism however has been that this movement has no specific landing site. This problem has been addressed namely by Beghelli & Stowell (1997), based on previous work of theirs (Beghelli & Stowell 1994 and Beghelli 1995). They show that quantifiers should be classified into different types, with specific syntactic properties and scope positions. What they propose is that the positions occupied by the different types of quantifiers at LF are organised hierarchically. Movement of the quantifiers to their scope position at LF is triggered by a feature-checking mechanism and scope properties are a by-product of these movements. When SR-un-NPs move to the highest projection of the clause at LF, I will talk about ‘widest scope’. This is to avoid the confusion arising from the term ‘wide scope’. In effect, an operator may take wide scope over another one without having widest scope. This is for example the case of intermediate scope (Section 3.4.3).

.  Whether phi features give rise to locality effects or not has to be determined.

Chapter 2.  un-NPs ‘a-NPs’ 



In addition to the intervention effects discussed above, the fact that SR-un-NPs have scope properties supports the presence of a QP layer in their structure. In Section 3.3.1, I suggested that it is this projection which triggers QR. In other words, to have scope properties, a nominal expression must involve a QP layer. As SR-un-NPs clearly enter scope relations – they may have widest scope or intermediate scope (Section 3.4.3) – they must comprise a QP. The question which arises however is whether QP and SRefP are not incompatible as they encode a quantificational and a referential feature respectively. I argue that it is not necessarily the case as the quantificational feature has an interpretative and a syntactic component. The first one implies that quantificational noun phrases do not pick out an entity and the second one that they have scope properties. What I suggest here is that projecting a SRefP above QP affects the first aspect but not the second one. In other words, SRefPs denote an entity, i.e., they are (S)-referential, but have scope properties. In other words, although the presence, in the same structure, of a layer related to quantification and of another one related to referentiality may seem contradictory, I argue that, on the contrary, this is what accounts for the characteristics of SR-un-NPs. If SR-un-NPs undergo QR, it means that this mechanism moves the highest layer of the un-NP along, i.e., the whole SRefP. In other words, SRefP is pied-piped with the QP. Movement of additional material with the QP seems to occur in other contexts (Michal Starke p.c.). Consider the following example of pied-piping in French: (126) a. b.

Pour combien de politiciens as- tu voté ? for how.many of politicians have you voted ‘For how many politicians have you voted?’ * Combien de politiciens as-tu voté pour ? how.many of politicians have you voted for

In (126a), the preposition pour ‘for’ moves with the object and cannot be stranded (126b). Yet, it is the wh-constituent combien ‘how.many’ which has to be moved to form a question. This suggests that the quantificational element embedded in a bigger constituent is active in the sense that it undergoes QR, in which case it carries along the material dominating it. After having presented some arguments supporting the hierarchy of projections advocated in (117), let us briefly consider the movement of un ‘a’, which head moves from Num to SRef, through Prop and Q. It is this movement which triggers the projection of the different layers of the left periphery and leads to an S-referential reading. Alternatively, a mechanism like Shlonsky’s ‘copy-remerge’ mentioned in Section 3.3.1 can be adopted. Adapted to noun phrases it suggests that the lowest projection of the nominal left periphery, Prop, is endowed with several sequenced features. These features are lexically pre-compiled and correspond to the heads of the left periphery. SRefP is thus projected to encode the third feature contained in the sequence of features in Prop: Prop [Prop, Q, SRef…]. This means that for the SRef feature to be

 The Layered DP: Form and meaning of French indefinites

accessed by the computational system, the order of compilation must be respected. In other words, PropP and QP must be projected and QP has to dominate PropP. The copy and remerge mechanism triggers the movement of un ‘a’ from Num to SRef, through Prop and Q.

3.4.2  Scope islands In Section 3.3.2 we have seen that Q-un-NPs are trapped in islands on a par with other quantificational noun phrases. SR-un-NPs clearly differ in that may take scope out of an island. Recall the following examples discussed by Fodor & Sag (1982) (Chapter 1 Section 5.1.2): (127) a. John overheard the rumor that each of my students had been called before the dean. b. John thinks that for each of my students to be called before the dean would be preposterous. (128) a. John overheard the rumor that a student of mine had been called before the dean. b. John thinks that for a student I know to be called before the dean would be preposterous.

In (127), the each-NPs cannot take scope over the highest clause whereas in (128) the indefinites can. In other terms, the scope of each-NPs is constrained to islands contrary to the one of a-NPs. If they both were usual quantifiers, i.e., QPs, this would be unexpected as seen in Section 3.3.2. The interpretations resulting from the a-NPs taking scope out of the island is that ‘there is an x, x is a student of mine, such that John overheard the rumor that x has been called before the dean’ and ‘there is an x, x is a student I know, such that John thinks that for x to be called before the dean would be preposterous’. That a-NPs are not trapped in islands is further illustrated in (129), which are identical to (128), except for the subjects of the matrix clause which have been replaced by a quantifier phrase. Such examples show that a-NPs can take maximally wide scope over any logical operators or quantifiers outside the island. (129) a. Each teacher overheard the rumor that a student of mine had been called before the dean. (69) b. Each teacher thinks that for a student I know to be called before the dean would be preposterous. (70)

In (129), a student of mine can take scope over each teacher, i.e., widest scope. Crucially, such examples are about a student in particular, as was the case of the a-NPs scoping out of islands in (128). In other words, these a-NPs are S-referential in the sense that they denote an individual known by the speaker. In sum, SR-un-NPs, contrary to Q-un-NPs and quantificational noun phrases in general, can escape scope islands and take widest scope, which represents their typical interpretation.



Chapter 2.  un-NPs ‘a-NPs’ 

Why this should be so is unclear and has to be determined. A few observations are however possible. SR-un-NPs undergo QR because their structure comprises a QP layer. However, this is also the case of Q-un-NPs, which are, in contrast, blocked in islands. What differentiates SR-un-NPs from Q-un-NPs is the presence of a SRefP layer in the former. This suggests that extraction of island might be a matter of category, i.e., SRefP vs. QP. However why the layer SRefP should have an impact on extraction remains mysterious at this stage.51 Descriptively it seems that it is the combination of a layer encoding speaker’s reference (SRefP) and a QP which allows SR-un-NPs to take scope out of islands. A QP by itself is not sufficient, otherwise all quantifiers would move out of islands, contrary to fact (recall (127)). Some kind of reference is not sufficient either as definite noun phrases, including pronouns, do not move out of islands. In the next section, we will see that SR-un-NPs do not necessarily take widest scope. What determines the position SR-un-NPs target at LF is addressed in Section 3.4.4.

3.4.3  Intermediate scope In her seminal paper on quantifier scope and syntactic islands, Farkas (1981) shows that indefinites can not only take widest scope (130), but also intermediate scope (131). Consider the following examples: (130) a. John bought every book that was published by a publishing house in New York. b. Jon dates every girl who knows a diplomat in Washington. c. John gave an A to every student who recited a difficult poem by Pindar.  (Farkas 1981: 62–3(13–14)) (131) a. Each student has to come up with three arguments which show that some condition proposed by Chomsky is wrong. b. Everybody told several stories that involved some member of the Royal Family.

Although the sentences in (130) contain a relative clause, i.e., a syntactic island, they are all ambiguous. The every-NPs can take scope over the a-NP or be in its scope, thus showing that indefinites can escape islands. Similarly, the sentences in (131) have several meanings. For example, (131a) can be about three arguments addressing one and the same condition by Chomsky, which, at the same time, varies with students. This is the intermediate reading. In other words, in (131a), some condition has narrower scope than each student but wider scope than three arguments. Reinhart (1997) however observes that this interpretation might just .  That only S-referential elements can be extracted from islands seems to be supported by Starke’s observation on extraction from Weak Islands (eWI), namely that “eWI can be used only if there are reasons to believe that there exist some entity which the interlocutor has in mind as a referent for the wh-phrase” (2001: 13).

 The Layered DP: Form and meaning of French indefinites

be a matter of vagueness. To control for this, she gives similar examples based, with some variation, on Ruys (1992) (132) and Farkas (1981) (133): (132) a. Most linguists have looked at every analysis that solves some problem. b. [Most linguists]1 [[some problem]3 [every analysis that solves e3]2 [e1 looked at e2]] (133) Each student has to find all arguments in the literature which showed that some condition proposed by Chomsky is wrong.

(132b) represents the intermediate reading syntactically. In this case, different linguists looked at different problems, in opposition to analyses of the same problem. In other words, “for a given problem, the relevant linguists looked at all the analyses that solve this problem” (Reinhart 1997: 345). The above examples show that the intermediate reading of indefinites exists, contrary to what Fodor & Sag (1982) claim. This had already been observed in Chapter 1 Section 5.1.2 and the relevant example is repeated below. (134) a. Every student overheard three reports that a Gila monster of mine drowned. (Diesing 1992: 90(54) b. (a Gila monster of mine: y) [(every student: x) [x overheard three reports that [y drowned]]] c. (every student: x) [(a Gila monster of mine: y) [x overheard three reports that [y drowned]]]

In (134a) a Gila monster of mine may take scope over every student, i.e., widest scope (134b), or below every student, i.e., intermediate scope (134c). As observed for (131) and (132), the latter reading, implies that the indefinite co-varies with the universal. This means that (134b) is about three reports on one and the same Gila monster, which, at the same time, varies with students. In other words, we are dealing with a particular/ certain but different Gila monster for each student. For example, if there were three students (a, b, c) and three Gila monsters (1, 2, 3), (134) could mean that student a overheard three reports about Gila monster 1, that student b overheard three reports about Gila monster 2 and that student c overheard three reports about Gila monster 3. Crucially, co-variation means that the Gila monsters are identified by the student, i.e., they are S-referential as they denote one entity in particular. In sum, the indefinite with intermediate scope in these examples are S-referential, which is not surprising if only SRefPs can escape islands (Section 3.4.2). Co-variation with a universal quantifier and intermediate scope have however to be differentiated. This is because ‘intermediate scope’ suggests there are three operators at play which is not necessarily the case with co-variation with a universal as the following example shows: (135) Every man danced with some women/woman.

(from Borer, p.c.)



Chapter 2.  un-NPs ‘a-NPs’ 

One reading of (135) is parallel to (134c) in that the women, perfectly identifiable, co-vary with the men. This would correspond to the intermediate readings discussed so far. However, as there are only two scope taking constituents in (135), the term ‘intermediate scope’ is inappropriate. Some women/woman takes narrow scope with regard to every man. However, as (135) has another reading where some women/woman does not take scope over the universal quantifier, the term ‘narrow scope’ is problematic as well. This would be a reading where some women/woman is quantificational in the sense that no woman in particular can be picked out in a set of women already introduced in the discourse. In short, as (135) has two different interpretations when the indefinite has ‘narrow scope’, the terminology is misleading. One possibility could be to consider the landing site of these indefinites at LF. In effect, assuming, following Beghelli & Stowell (1997 based on earlier work), that quantifiers should be classified into different types and that the positions occupied by these different quantifiers at LF are organised hierarchically might be an interesting way to distinguish between SR-un-NPs with narrow scope and Q-un-NPs with narrow scope. As, ‘narrow scope’ simply means ‘below a higher operator’, a hierarchy of positions should help us determine how ‘low’ in the structure these indefinites are and assign them a precise position. As the details of the hierarchy proposed by Beghelli and Stowell are not relevant here (and sometimes incompatible with the analysis developed in this book), I will simply assume that the Quantifier Raising (QR) mechanism targets various functional projections labelled FP in (136). (136) FP1 > FP2 > FP3 > FP4

In a hierarchy like (136), FP1 would host SR-un-NPs with widest scope, FP2 universal quantifiers, FP3 SR-un-NPs co-varying with a universal quantifier and FP4 Q-un-NPs. In other words, SR-un-NPs with ‘narrow scope’ and Q-un-NPs with ‘narrow scope’ in examples like (135) occupy different positions, FP3 and FP4 respectively. The impossibility for these elements to sit in the same position is supported by their behaviour in islands. Assuming that SR-un-NPs co-varying with the universal quantifier have to be in the projection immediately below the one hosting the universal quantifier implies that this position cannot be occupied by Q-un-NPs. In effect, indefinites covarying with the universal quantifier may escape islands as seen for example in (134c) contrary to Q-un-NPs. This suggests that in (136), any material responsible for a scope island occurs between FP3 and FP4. In this way, Q-un-NPs are blocked in FP4 whereas SR-un-NPs are able to move out of islands, either to FP1 to take widest scope or to FP3 to be in the scope of a universal quantifier and co-vary with it. Note that the fact that universal quantifiers, which are QPs, may be higher than SR-un-NPs, which are SRefPs, in a hierarchy like (136) is not problematic for Relativized Minimality. This is because the order QP > SRefP in such examples represents

 The Layered DP: Form and meaning of French indefinites

the surface order. In other words, the QP does not bypass the SRefP, which would be impossible as illustrated in (137): (137) Un étudiant a lu tous les livres. a student has read all the books ‘A student has read all the books.’

If un étudiant in (137) is a SR-un-NP, it is necessarily interpreted with ‘widest’ scope in the sense that there is a student in particular, say John, such that he read every book.52 In other words, un étudiant will move to FP1 and tous les livres to FP2. The co-variation reading where, for example, student 1 read the books a and b, student 2 the books b and c, student 3 the books a and c… is impossible. This shows that the universal cannot bypass the SRefP, as expected in the account proposed here. The fact that SR-un-NPs may vary with a universal quantifier or not is due to their dependent reference. Recall that their interpretation is determined by the context of utterance (Section 2.1). More precisely, SR-un-NPs with intermediate scope co-vary with the universal quantifier. SR-un-NPs with widest scope in contrast are similar to pronouns which are deictically referring, in the sense that some extra linguistic context is required to interpret them. The discussion so far suggests (i) that the indefinites which escape islands are SR-un-NPs and (ii) that SR-un-NPs moving out of islands at LF may take either widest scope or ‘intermediate scope’, i.e., move to a position immediately below a universal quantifier (FP3 in (136)). Postulating a type of indefinites with these two characteristics avoids the problems arising with Fodor & Sag’s (1982) account (Chapter 1 Section 5.1.2). Recall that for these authors the intermediate reading of indefinites is unexpected and unaccounted for. The reason is that (i) quantificational un-NPs are blocked in relative clauses and that (ii) their referential indefinites do not have scope properties. In contrast, SR-un-NPs have scope properties and are able to move out of islands.

3.4.4  Quantifier Raising without circularity In the preceding sections, we have seen that SR-un-NPs may target two different positions at LF. In a very tentative hierarchy like (136) repeated below for convenience, they could move to FP1 or FP3. (138) FP1 > FP2 > FP3 > FP4 .  If un étudiant is a P-un-NP, it is interpreted in situ and tous les livres moves to FP2. If it is a Q-un-NP it takes scope over tous les livres (surface order). Note that this is unaccounted for by (136). If the universal is in FP2 the Q-un-NP would have to sit between FP1 and FP2 (it cannot target FP1 which hosts an existential quantifier (see p. 111)). If Q-un-NP sits in FP4, the universal has to be interpreted below. One possibility would be to assume that in some cases quantifiers may be interpreted in their base position (cf. Beghelli & Stowell who have an account along these lines for one of their quantifiers).



Chapter 2.  un-NPs ‘a-NPs’ 

The question tackled in this section is what triggers movement to which projection. Informally, it comes to determining when SR-un-NPs move to FP1 and when they move to FP3. This is a delicate task as it implies avoiding the circularity arising from much work on quantificational expressions which assumes on the one hand that the movement of quantificational expressions is triggered by some interpretative feature of theirs, and on the other hand that the interpretation of quantificational elements results from the target position (thanks to G. Puskás for underlining this problem). Put differently, a quantificational expression x moves to a position y to check a feature z, i.e., the feature z is encoded in x before the movement takes place, and at the same time this quantificational expression x has the feature z because it has moved to y, i.e., the feature z results from the movement. The same problem can be stated in terms of scope. On the one hand, it is assumed that quantificational elements move because they have to take scope on (part of) the sentence, and on the other hand that it is because they have moved that they have scope. Scope is at the same time the trigger of movement and the consequence of movement. In his work, Beghelli (1995) avoids circularity by proposing that (i) the trigger of movement is a scope independent feature-checking mechanism and that (ii) each landing site gives rise to a particular quantificational reading (existential, universal, interrogative…). For example, his Group-Denoting QPs, which comprise indefinite noun phrases headed by a, some and several, will move to a Referentiality Phrase (RefP) or a Share Phrase (ShareP), depending on whether they are referential or not. As the two landing sites contain an existential operator, Group-Denoting QPs will be interpreted as existential in both cases. In other words, the trigger of movement and the interpretative result of movement are different. Scope is a by-product of the mechanism. The result of this analysis is a hierarchy of operators where the highest projection and the one below universal quantifiers, i.e., Beghelli’s RefP and ShareP respectively, host an existential operator (∃). In the analysis developed here, I adopt Beghelli’s proposal concerning existentiality for SR-un-NPs. Recall that SR-un-NPs assert existence (Section 2.1.2). In other words, I assume that the positions targeted by SR-un-NPs (FP1 and FP3 in our terms) contain an existential operator.53 The existential reading of SR-un-NPs .  Another possibility to account for the existentiality of dependent un-NPs, which seems very attractive to me but which I will not develop here, is to extend Heim’s (1982) analysis of indefinites in the restriction of universals to all un-NPs with a dependent reading. Heim assumes that indefinites in the restriction of universals are licensed by the universal quantifier. According to her, examples like (i) represent universal quantification over pairs, illustrated in the formula (ii) and equivalent to the formula (iii) (1982: 141–2). (i) Every man who owns a donkey is rich. (ii) ∀x1 ∀x2 ((man(x1) & donkey (x2) & own (x1, x2)) → rich (x1)) (iii) ∀x1 ((man(x1) & ∃x2 (donkey (x2) & own (x1, x2))) → rich (x1)) This means that “[t]he indefinite does not really have an existential reading in [i]; it has widescope universal reading which happens to be equivalent to a narrow-scope existential reading”.

 The Layered DP: Form and meaning of French indefinites

therefore results from their movement at LF. As for referentiality – more precisely Speaker’s reference, I assume that it is encoded in the structure of SR-un-NPs, namely in the SRefP layer. This means that S-reference is not a feature to be checked. This is supported by the fact that S-referential un-NPs can target two different positions, namely a position above or a position below universal quantifiers. If S-reference had to be checked, it would mean that this could be done in two different positions, which is very unlikely. The question which arises is therefore Why can SR-un-NPs target two positions? In other words, What is the interpretative difference between SR-un-NPs which take scope over universals and SR-un-NPs which are in their scope? According to the definition of S-reference adopted here, S-referential noun phrases are context dependent. Put differently, their referent may vary with the utterance situation (Section 2.1.2) or with a universal quantifier (Section 3.4.3). The former case corresponds to SR-un-NPs with widest scope (Sections 3.4.1 and 3.4.2) and the latter to SR-un-NPs with intermediate/narrow scope (Section 3.4.3). In neither case are they referentially independent, in contrast with definite expressions like John or the-NPs. What I would like to propose here is that the difference in interpretation between SR-un-NPs with widest or intermediate scope is what triggers their movement to different positions, FP1 and FP3 respectively. The idea is that when SR-un-NPs are free, they carry a [+ deictic] feature which triggers movement to the highest projection of the clausal left periphery, i.e., to the edge of the structure to be close to discourse information. In other terms, they have to turn ‘outward’ to the discourse because they depend on extra linguist context. More precisely, these nominals are ‘speaker oriented’ in the sense that they refer to what ‘the speaker has in mind’, hence the term deictic. In contrast, SR-un-NPs which are not deictic have to be in the scope of a universal quantifier to be interpreted. Informally, the question that arises for SR-un-NPs is ‘How can I be interpreted?’ to which there are two possible answers: ‘You need some extra information from the discourse’ or ‘You depend on a universal quantifier’. The former triggers movement to the highest projection in (138), i.e., FP1, and the latter to the projection below the universal quantifier, i.e., FP3. In terms of scope, it means that the former gives rise to widest scope and the latter to intermediate/narrow scope. In sum, according to the analysis developed here, SR-un-NPs, which contain a QP layer, are subject to Quantifier Raising (QR) on a par with other QP elements. What determines their landing site at LF is the way their reference is interpreted. If they are [+ deictic], they take widest scope and move to FP1. If they are not deictic, they have to co-vary with a universal quantifier and move to FP3. In both cases, the movement of SR-un-NPs results in an existential reading. This is because their landing site contains an existential operator. As the trigger and the result of movement are different in such an account, circularity is avoided.



Chapter 2.  un-NPs ‘a-NPs’ 

3.4.5  Semantic paraphrases The aim of this section is to look more closely at the formulae or paraphrases traditionally used to describe scope ambiguities, as they seem to support the analysis developed here. Recall from Section 2.2 the ambiguous example (139) and the formulae used to paraphrase its meanings (140). (139) Everyone saw a woman. (140) a. For every x, there is a y such that it is the case that x saw y. b. There is a y, such that for every x, it is the case that x saw y.

The idea is that (139) is ambiguous because it contains two quantificational expressions, a universal and an existential quantifier, which may take different scopes. In (140a), it is the former which takes scope on the latter and in (140b) it is the other way round, i.e., ∀ > ∃ vs. ∃ > ∀. This however is problematic if we take into account the various remarks and proposals made throughout this book about existential constructions and interpretative properties of un-NPs. In effect, we have seen that presupposition of existence and assertion of existence are incompatible (Section 2.2.3). As existential constructions assert existence and as quantificational noun phrases presuppose existence, it predicts that the latter are ungrammatical in there be contexts (Milsark 1974, 1977). As presuppositional quantificational expressions are QPs in the framework adopted here, it means that the noun phrases following there is in (140) cannot be QPs, including Q-un-NPs. Yet, as (140) shows, these noun phrases seem to have scope properties. This is exactly what I argue for in this section. In other words, the un-NPs occurring in such examples are SR-un-NPs, which can take widest or intermediate scope (in opposition to QPs). Widest scope SRun-NPs are illustrated in (140b) and intermediate/narrow scope SR-un-NPs in (140a). If the there be part of existential sentences represents the existential operator (Milsark 1974, 1977), it means that the noun phrases following it cannot be intrinsically existential. In other words, the source of existentiality is external to the nominal expression. This is not only the case of SR-un-NPs (Sections 3.4.2 and 3.4.3) but also of some P-un-NPs (Section 2.3), predicting that both should be grammatical in paraphrases like (140). However, as P-un-NPs take narrow scope, they should only be possible in the scope of the universal. This is supported by (141) in which the indefinite NPs are slightly modified: (141) a. For every x, there is a particular y such that it is the case that x saw y. b. For every x, there is a y or other (any y) such that it is the case that x saw y.

What (141) shows is that (140a) is ambiguous.54 The indefinite may refer to a particular entity or denote a property, i.e., have an ‘any-flavour’. In our terms, it means that .  Although the truth conditions of the examples in (141) are the same, there is a difference between a and b which should be captured. This is one aim here.

 The Layered DP: Form and meaning of French indefinites

the un-NP can be a SR-un-NP, in which case it is a particular y which co-varies with the universal, or a P-un-NP, in which case it is a y or other. In other words, (139) has (at least) three readings (interpreting a woman as a Q-un-NP would be an additional interpretation) which cannot be accounted for by simply attributing them to different orders of quantifiers. In short, the above discussion leads to important conclusions. First, it shows that there is a y in paraphrases like (140) does not mean that we are dealing with a QP. This is because a QP, which is presuppositional, cannot occur in an existential context. Second, this implies that the role of the there be structure is simply to assert the existence of an entity. In other words, what we know about a noun phrase occurring in a there be context is that it is existential. However, this construction does not allow us to identify the source of existentiality. Consequently, it is impossible, on the basis of a there be structure, to determine what kind of existential un-NP we are dealing with. In effect, according to the analysis proposed here, P-un-NPs and SR-un-NPs may have an existential reading for different reasons. The existential operator may be supplied by the predicate (Section 2.3.2) or by the projections targeted by the un-NPs (Sections 3.4.2 and 3.4.3). The former concerns existential P-un-NPs and the latter SR-un-NPs. Another context which is traditionally considered as ambiguous due to scope interactions is negation, as illustrated in (142): (142) Mary didn’t buy a dog.

The idea is that the indefinite is either in the scope of the negation or has scope over it. The two interpretations are accounted for straightforwardly in the framework developed here. When the indefinite takes scope over the negation it is a SR-un/a-NP, which moves to the highest position of the sentence at LF. When it is the negation which takes scope over the indefinite, the latter is a non-existential P-un/a-NP. In French the latter reading involves a de-NP, the negative counterpart of P-un-NPs as shown in (143a) (Section 2.3.3). To be grammatical in a negative context, the un-NP must be interpreted as a SR-un-NP (143b). In other words, only un-NPs encoding speaker’s reference are licensed in negative contexts in French. (143) a. b.

Jean n’ a pas acheté de livre. Jean ne has not bought de book ‘Jean hasn’t bought any books.’ Jean n’ a pas acheté un livre. John ne has not bought a book ‘Jean hasn’t bought a book.’

In (143a) de livre ‘of books’ is not existential, on a par with a dog in (142) in its ‘narrow scope’ reading. That is, it might very well be that no book or no dog exists. In contrast, un livre ‘a book’ in (143b) and a dog in (142) in its wide scope reading are existential. What this shows is that scope properties are not sufficient to capture the various



Chapter 2.  un-NPs ‘a-NPs’ 

interpretations of un/a-NPs. The intrinsic characteristics of SR-un-NPs (Section 2.1) and P-un-NPs (Section 2.3) play a crucial role.

3.4.6  The limits of a choice function analysis Another advantage of the analysis proposed here is that it does not only account for the scope properties and for the (potential) existential reading of un-NPs. It also accounts for additional interpretative characteristics, namely whether they are S-referential, quantificational or property-denoting. This is, in my opinion, why it is superior to an account like a choice-function analysis, which is gaining importance in the literature (Reinhart 1997). In a choice-function analysis, what accounts for the scope properties of the indefinite noun phrase is the existential quantifier, which can be inserted in different positions. In other words, the function variable is bound by an existential operator which can be arbitrarily far away. A description of choice functions is provided in (144). (144) A function f is a choice function (CH (f)) if it applies to any non-empty set and yields a member of that set. (Reinhart 1997, Section 6.1)

The wide scope reading is illustrated in (145b) and the intermediate scope in (146b). A major difference with a QR account like (145c) is that, in a choice function analysis, the N-restriction remains in situ. (145) a. Every lady read some book. b. ∃f (CH (f) ∧ ∀z (lady (z) → z read f(book)) c. ∃x (book(x) ∧ ∀z (lady (z) → z read x))

(Reinhart 1997: (65))

(146) a. Most linguists have looked at every analysis that solves some problem. b. For most linguists x, ∃f (CH (f) ∧ ∀y (analysis (y) ∧ y solves f(problem) → (x looked at y)). (Reinhart 1997(68a,c)

The choice function in (145b) applies to the set of books. What it means is that ‘a function exists, such that for every z, if z is a lady, then z reads the book selected by this function’. In (146), some problem may take scope over every analysis, and at the same time co-vary with most linguists. As the existential closure is a purely interpretative procedure, it may be introduced arbitrarily far away, including in the scope of another operator. Although such an analysis accounts for the scope properties and for the existential reading of some indefinites, it seems to me that it is not fine grained enough. For example, the fact that un-NPs with wide scope and intermediate scope encode some type of reference, in contrast with other un-NPs, is ignored (Section 3.4.2 and 3.4.3). In addition, no explanation as to why an indefinite takes widest or intermediate scope is provided (Section 3.4.4). Furthermore, the fact that the predicate may be responsible for the existential reading of some indefinites is not acknowledged (Section 2.3.2). That not all indefinites assert existence is not acknowledged either.

 The Layered DP: Form and meaning of French indefinites

Although Reinhart mentions that indefinites may also be standard existential generalized-quantifiers (i.e., Q-un-NPs), she seems to suggest that this assumption is not crucial and that it can be dropped (1997: Section 6.2). In other words, a single interpretative procedure could be sufficient to analyse indefinites. As seen in Section 2, however, empirical considerations show that a three-way distinction is necessary to account for the interpretative and scope properties of at least some indefinites, namely un-NPs.

3.5  Comparison with two-way analyses In the preceding section, we have seen in which way the analysis developed here is superior to a Choice Function account. In this section we will briefly see how it compares with the two-way classifications discussed in Chapter 1 Section 5.1. As the discussion about the scope properties of un-NPs is crucial to establish a list of contexts in which P-un-NPs, Q-un-NPs and SR-un-NPs occur, such a comparison would have been incomplete if done earlier. Table 8 provides a list of contexts in which each type of un-NPs typically occurs. It also indicates to which category/ies proposed by Milsark, Fodor and Sag and Enç the classes advocated here correspond (recall Section 5.1.5 of Chapter 1). The capital letters in parenthesis indicate the authors ((M) stands for Milsark, (FS) for Fodor and Sag and (E) for Enç). (147) Table 8. Type

Context

Parallel

1. P-un-NPs Existential sentences Weak weak (M) Quantificational (FS) (151) Non-specifics (E) 2. Intensional contexts Quantificational (FS) Non-specifics (E) 3. Not introduced in the Non-specifics (E) discourse – i.e., with ‘existential predicates’ Weak strong (M) No co-variation with ∀ 4. Q-un-NPs (Narrow scope) Quantificational (FS) Relational specifics (E) (150) 5. Introduced in the discourse Weak strong (M) (covert partitives) Partitive specifics (E) Referential (FS) 6. SR-un-NPs Widest scope Relational specifics (E) (149) 7. Co-variation with ∀ Relational specifics (E) (148) (Intermediate/narrow scope)  8. Descriptive richness Referential (FS) (Continued)

Chapter 2.  un-NPs ‘a-NPs’ 



Table 8.  (Continued) Type



Context

9. SR-un-NPs Topicalisation 10. Existential sentences about a particular entity 11. Relative clauses, especially non-restrictive

Parallel Referential (FS) (153) Referential (FS) Relational specifics (E) (154) Referential (FS)

A first observation is that Enç’s classification is the closest to the one proposed here. Her two discourse-linked categories, namely partitive specifics (covert partitivity) and relational specifics (familiarity), seem to roughly correspond to Q-un-NPs and SRun-NPs respectively, although the latter are not restricted to a certain N and occur in contexts not discussed by Enç. However, the reasons for ‘relational specifics’ to have scope properties and the difference between intermediate (148) and widest (149) scope, i.e., a reading in which the indefinite co-varies with the universal quantifier and a reading in which it does not, are not addressed in Enç’s account, contrary to the one presented here. (148) Each husband had forgotten a certain date – his wife’s birthday. (E5) (149) John wants to own a certain piano which used to belong to a famous pianist. (E4a)

As relational specifics may also have narrow scope (150), in addition to intermediate and wide scope, it suggests that they simply might have scope properties because they are quantificational. However, quantificational noun phrases, defined in terms of covert partitive reading (Section 2.2), correspond to Enç’s partitive specifics. In other words, relational specifics, which do not have a covert partitive meaning, cannot be quantificational and therefore should not have scope properties. (150) John believes that there are unicorns living in his backyard. He claims that he can distinguish each unicorn from the others, and has even given them names. He believes that a certain unicorn is responsible for destroying his roses, and wants to catch him. (E7)

The problem ‘covert partitive meaning-scope properties’ can be solved if one assumes that some indefinites may have scope properties but no covert partitive reading. This is what I argue for in this section by suggesting that SR-un-NPs encode referentiality and undergo Quantifier Raising because they contain an active QP layer. More generally, it seems that the analysis developed here is more fine-grained than Enç’s as it does not only take into consideration the referential/quantificational issue but also the possible existence entailment of indefinites. In effect, some indefinites always assert existence (SR-un-NPs), others may assert existence (P-un-NPs) and a third group never asserts existence (Q-un-NPs). Furthermore, the contexts in which P-un-NPs are existential or not are clearly defined: they assert existence in existential

 The Layered DP: Form and meaning of French indefinites

sentences and with existential predicates but do not in intensional contexts.55,56 In contrast, Enç simply considers that non-specifics and relational specifics are not presuppositional. In other words, the analysis developed in this book seems superior as it accounts for the referential/quantificational interpretation of un-NPs, for their scope properties and for their (non)-existential reading. The above observation extends to Fodor and Sag’s and Milsark’s classification as none of them discusses all three issues. In addition, as the analysis presented here deals with intermediate scope, it does not run into the problem arising in Fodor and Sag (recall Chapter 1 Section 5.1.2). It also allows us to distinguish between (151) and (154) without assuming that indefinites in there be sentences are quantificational, which is a problematic claim as presupposition and assertion of existence are incompatible, as observed by many scholars, including Milsark and Enç (Chapter 1 Sections 5.1.1 and 5.1.3 respectively. (151) There was someone smoking behind the woodshed. (FS18)

As for Milsark’s analysis, we have seen in Chapter 1 Section 5.1.5 that it cannot capture the ambiguity of an example like (152). (152) A student in the syntax class cheated on the final exam. (FS1)

As Milsark’s weak indefinites with a weak reading are restricted to existential contexts, a student in the above example must have a strong reading, i.e., it cannot be ambiguous. One way to solve the problem could be to treat weak indefinites with a weak ­reading as P-un-NPs. In other words, they would not only occur in existential sentences but also in intensional contexts and with existential predicates. In this way, a student in (152) would be able to have a strong and a weak reading. However, such a refinement would not be sufficient to account for contexts in which indefinites pick out a particular entity, typically topicalisation which cannot host quantificational noun phrases (Section 3.1.2) and existential sentences about a particular individual. The relevant examples are repeated as (153) and (154): (153) A Frenchman that I met in Tokyo, I went and had dinner with (him) in New York last week. (FS13) (154) There’s a girl in our syntax class who cheated on the exam. (FS20)

.  Questions are also a context in which P-un-NPs may be non-existential. However, as the papers under discussion here do not mention them, I left them out of the table. .  Note that the fact that un-NPs may be neither existential nor presuppositional (it is the case of P-un-NPs) is problematic for Diesing’s analysis. According to her, indefinites which are not presuppositional must be bound by Existential Closure, i.e., be existential.



Chapter 2.  un-NPs ‘a-NPs’ 

All the above remarks support the analysis developed here, which classifies un-NPs into three types and which takes into account several of their interpretative features.

4.  Conclusion In this chapter I have addressed two main issues, namely (i) the classification of argumental un-NPs in French and the terminology required to establish such a typology, and (ii) the internal structure of argumental un-NPs and the impact that different structures may have at the level of the clause, namely for scope properties. The typology I argue for corroborates the one proposed by Dobrovie-Sorin & Beyssade (2004). This means that I consider that argumental un-NPs can not only be referential and quantificational but also property-denoting expressions. In other words, un-NPs are classified according to their semantic type. As the interpretation of most referential un-NPs is determined by the context of utterance, namely by what/ who the speaker ‘has in mind’, I have taken this kind of reference to be speaker’s reference, as defined by Kripke (1977). The label adopted for this type of un-NPs is thus S-referential (SR-un-NPs). However, as the interpretation of a minority of SR-un-NPs may depend on a universal quantifier, and not on contextual information, the label ‘S-referential’ should be understood in the more general sense of ‘referentially dependent’. In other words, SR-un-NPs include un-NPs which depend on extra linguistic information to be interpreted and un-NPs which co-vary with a universal quantifier. As the latter interpretation only concerns a minority of cases, the label S-referential has been used for all un-NPs which are referentially dependent. What is crucial here is that this type of reference should be distinguished from the reference of proper names or the-NPs which is rigid. The other two types of un-NPs advocated are Property un-NPs (P-un-NPs) and Quantificational un-NPs (Q-un-NPs). Property-denoting un-NPs are typically interpreted in situ and do not necessarily assert existence, contrary to SR-un-NPs. Their absence of existence entailment is clear in questions and intensional contexts for example. Q-un-NPs in contrast are presuppositional, in the sense that they presuppose existence. De-NPs have also been examined, as they are in complementary distribution with P-un-NPs. More precisely, they are restricted to negative contexts, from which P-un-NPs are banned. As they have to be in the scope of a negative operator, they have been considered as a kind of Negative Polarity Items (Kayne 1981). In addition, I have proposed that de-NPs, on a par with P-un-NPs, are property-denoting arguments and that, in contrast with P-un-NPs, they are always non-existential. Concerning the internal structure of argumental un-NPs, the major claim here is that they exhibit an extendable telescopic left periphery. This lies on two main assumptions. The first one is that the structure of argumental and non-argumental noun

 The Layered DP: Form and meaning of French indefinites

phrases differs (Longobardi 1994, 2005 among others) and the second one that there is a strict syntax-semantics mapping (Zamparelli 2000). The first assumption has led me to refine Longobardi’s distinction between DPs and NPs and to propose that argumental un-NPs minimally involve one layer of the nominal left periphery, the lowest one. In other words, I have argued that it is this layer which turns predicative noun phrases into arguments. As it is characteristic of un-NPs interpreted as properties, I labelled it Property Phrase (PropP). A structural difference between predicative noun phrases and property-denoting arguments is supported by the fact that not all predicates may be arguments. In other words, the projection of PropP allows us to distinguish predicates which can function as arguments, essentially nominal predicates, from predicates which cannot, such as adjectival predicates. This means that the smallest arguments are property-denoting noun phrases (bare nouns are briefly addressed in Chapter 4). Such a structural distinction between property-denoting arguments and predicative noun phrases has led to the conclusion that selection should be reformulated in terms of categories and not of semantic types. For example, intensional predicates may select a PropP and not a particular semantic type like 〈e,t〉. Otherwise, adjectives for instance should be fine as arguments of such predicates, contrary to fact. On the basis of the second assumption, I have proposed that the internal structure of argumental un-NPs varies with their interpretation, i.e., with their type. More precisely, I have argued that the left periphery of P-un-NPs involves one layer, PropP, the one of Q-un-NPs two layers and the one of SR-un-NPs three layers. The projections constituting the left periphery of Q-un-NPs are PropP and a Quantifier Phrase (QP) and the ones forming the left periphery of SR-un-NP are PropP, QP and a ­SRefP encoding speaker’s reference. As mentioned above, I have proposed that PropP transforms predicates into arguments and is characteristic of property-denoting un-NPs. As for QP, I have suggested that it encodes a quantificational feature ­responsible for a reading where no entity can be picked out and that it triggers Quantifier Raising (QR) at Logical Form (LF). Informally, it means that this projection is necessary for a constituent to undergo QR and have scope properties. SRefP encodes S-reference as defined above. In other words, noun phrases become quantificational and S-referential by adding a projection encoding the relevant features. Put differently, what differs in the structure of each type of un-NP is the composition of their left periphery. In other words, the interpretation of un-NPs results from the piling up of different functional layers in the nominal left periphery. Furthermore, I have argued that the structure advocated for each type of un-NPs accounts for some of their interpretative properties. For example, if the projection of QP is necessary for a constituent to undergo QR and have scope properties, it predicts that P-un-NPs, which do not contain such a layer in their structure, are interpreted in situ. In addition, as P-un-NPs do not contain any empty category, their distribution is not restricted, contrary to the one of de-NPs. As the latter are only grammatical in



Chapter 2.  un-NPs ‘a-NPs’ 

negative contexts it suggests that they contain an empty category which has to be licensed by a negative operator. Finally, as the structure of P-un-NPs does not imply anything about existentiality, it is not surprising that they do not entail existence and therefore that they are fine in existential sentences. As for Q-un-NPs, they belong to the same category as other quantifiers like every-NPs, i.e., QP. This correctly predicts that, on a par with the latter, they should be blocked in islands and interact with other quantifiers, namely with regard to intervention effects. The crucial claim about the structure of SR-un-NPs is that it contains both a SRefP and a QP. This accounts for the fact that they pick out an entity, contrary to QPs, and that they have scope properties, on a par with QPs. The idea is that the projection of SRefP above QP cancels the interpretative aspect of QP, namely the impossibility for QPs to have reference, but not its syntactic effect, namely scope properties due to QR. The combination of these two projections also allows SR-un-NPs to escape islands, although why this should be so remains unclear. The projections targeted by SR-un-NPs at LF may be the highest one or the one immediately below the projection hosting distributive universal quantifiers (Beghelli & Stowell 1997). Movement to the former occurs when the reference of SR-un-NPs has to be determined by extra linguistic speaker oriented information, i.e., when they are [+ deictic], whereas movement to the latter occurs when they co-vary with a universal quantifier. In other words, what triggers movement of SR-un-NPs to different positions is their dependent reference. Empirical evidence for structures of different sizes comes from intervention effects (Starke 2001). In a nutshell, intervention effects occur in constructions involving elements of the same type or when the intervener is interpretatively heavier than the moving element. In contrast, when the moving element is interpretatively heavier than the intervener no intervention effects arise. If ‘interpretatively heavier’ is translated in terms of structure as proposed in this chapter for un-NPs, it predicts that QPs can cross PropPs at LF and that SRefPs can cross QRs (and PropPs). Conversely, it predicts that Q-un-NPs cannot move past other QPs or past SR-un-NPs. As all these facts are borne out, it supports the analysis proposed here which attributes left peripheries of different sizes to indefinites of different types. Had these indefinites the same structure, the (lack of) intervention effects would be unexpected and unaccounted for. More generally, the analysis developed here suggests that the internal structure of un-NPs – in opposition to their position in the sentence (Diesing 1992) – not only determines their interpretation but also (part of) their behaviour at the clausal level. This also means that a distinction in terms of scope only is not sufficient to capture the interpretative properties of the three types of un-NPs advocated here. In other words, the analysis developed in this chapter is more fine-grained than a choice-function analysis for example. Although the latter nicely accounts for the scope properties of indefinites, it seems to ignore their difference in semantic types, in particular the difference between S-referential and Property un-NPs. Furthermore, contrary to a choice-function

 The Layered DP: Form and meaning of French indefinites

analysis, I have argued that the existential reading of un-NPs may be due to different sources. In particular, when existential, P-un-NPs owe this interpretative feature to an existential operator supplied by the predicate, i.e., by the clause. SR-un-NPs in contrast are existential because they move to a position hosting an existential operator as mentioned above. In other words, none of the three types of un-NPs is intrinsically existential. Contrary to SR-un-NPs and some P-un-NPs, Q-un-NPs are presuppositional, which accounts for their ungrammaticality in existential contexts. That scope properties are not sufficient to capture the various readings of un-NPs is supported by the paraphrases traditionally used to illustrate scope ambiguities. For example, such paraphrases, namely there is a x, such that for every y…/ for every y, there is an x…, cannot account for the difference between a particular x and an x or other. Rather, it seems that the there is part simply asserts the existence of the expression following it. The source of existentiality however remains undetermined. According to the analysis developed here, the existential operator may be supplied by the predicate selecting a P-un-NP or by the target positions of SR-un-NPs. This does not only account for the different order of operators, that is ∀ > ∃ vs. ∃ > ∀, but also for the difference between a particular entity and an entity with a property reading. In sum, as the analysis developed here combines three fundamental characteristics of indefinites, namely their (lack of) scope properties, their (lack of) existentiality and their (lack of) reference, it is more adequate to account for their interpretation and more fine-grained than any two-way classification examined in Chapter 1. The general idea adopted and argued for in this chapter, namely that arguments can be of different semantic types, is in line with much recent work since the early 1980s (for example Partee 1986) and in opposition to a theory like Montague’s (1974) analysing all noun phrases as generalized quantifiers (see also Barwise & Cooper 1981). The assumption that property-denoting noun phrases can function as arguments requires new composition rules, i.e., rules which allow this type of arguments to compose semantically with the predicate selecting them. This means that the semantics of predicates needs to be refined, and more generally, that the study of nominal expressions and the one of predicates should go hand in hand. On the theoretical side, the analysis developed here supports the idea that it is the syntax which shapes meaning – and not the other way round – and that syntax is not about words or morphemes but about features, which allows more fine-grained distinctions. Such a position predicts that structures become more and more complex to encode interpretative features not considered so far. It also raises many questions about all other noun phrases, for example those with demonstratives, possessives, cardinals…, about wh-elements, and about the structure of their left periphery. In the next chapter, I turn to noun phrases with the so-called ‘partitive article’ in French. If the analysis developed here is on the right track, it should be applicable to these noun phrases.

chapter 3

Du/des-NPs 1.  Introduction 1.1  The analysis In Chapter 2, I examined un-NPs ‘a-NPs’ and argued that they are three-way ambiguous and that the composition of their left periphery varies according to their interpretation. In this chapter, I turn to noun phrases attesting the so-called ‘partitive article’ du ‘of. the.masc’, de la ‘of the.fem’ and des ‘of.the.pl.’ illustrated in (1).

(1) Elle a mangé du gâteau/ de la tarte/ des biscuits she has eaten of.the.masc cake of the.fem tart of.the.pl biscuits ‘She ate (some) cake/tart/biscuits.’

As illustrated in (1), des ‘of.the.pl’ precedes plural nouns whereas, du/de la ‘of.the.sg/ of the.fem’ precedes a singular noun. I will refer to this construction as du/des-NPs ‘of.the-NPs’. The labels du/des-constituent/construction might however also be used. Predicative examples like Sophie et Marie sont des professeures ‘Sophie and Marie are of.the professors’ will be ignored as this study focuses on argumental noun phrases. Generic examples like Des gendarmes peuvent confisquer une voiture ‘of.the military policemen may confiscate a car’ will be disregarded as well. The first aim of the chapter is to determine whether the analysis developed in Chapter 2 extends to du/des-NPs, i.e., (i) whether du/des-NPs can be classified into three groups according to their semantic type and whether this typology holds for both des-NPs and du-NPs and (ii) whether such a typology implies different internal structures. If the analysis of un-NPs proposed in Chapter 2 is applicable to du/des-NPs, it predicts that du/des head moves to the nominal left periphery, as it is the leftmost item

.  De-NPs, where a ‘bare’ NP is preceded by de ‘of ’, will also be addressed in Sections 2.3 and 3.2.2. .  This example is quoted in Bosveld-de Smet (1998: 14(13a)), from Carlier 1989. It seems that du/des-NPs can only be generic with certain modals or when dislocated to the left and pronominalized by ça ‘that’.

 The Layered DP: Form and meaning of French indefinites

of the noun phrase. This movement may trigger the projection of one, two or three layers – Property Phrase (PropP), Quantifier Phrase (qp) and S(peaker)-Reference Phrase (SRefP) – depending on the interpretation of the noun phrase. Propertydenoting, quantificational and S-referential are the interpretative features I will focus on. Whether du/des-NPs give rise to collective, cumulative, iterative, non-iterative or other readings will not be addressed here as it would involve a thorough analysis of the predicates/events involved. This corroborates the observation made in the preceding chapter, namely that the study of noun phrases and the one of predicates are closely intertwined. The second aim of the chapter is to determine the base position and the role of the two elements constituting du/des, namely de ‘of ’ and le/les ‘the’. This means that I will examine the structure of their inflectional domain. Concerning de ‘of ’, there is no consensus in the literature. For some linguists de is an article (Frei 1960; Damourette & Pichon 1952) – i.e., it is part of the nominal structure – and for others it is a preposition (Clédat 1901; Gross 1967; Wagner & Pinchon 1962) – i.e., it is outside the nominal structure. More recently, it has been proposed that de in examples such as (1) has a dual status – preposition or quantificational head (Kupferman 1994). A similar conclusion will be reached in this book although I will propose that de is either a functional element generated low in the inflectional domain (in opposition to a quantificational head) or a preposition. As for the le/les component, various analyses can also be found in the literature. For some scholars it is a generic article (Milner 1978; Gross 1967), for others it is a non-specific article (Kupferman 1979, 1994). What I will propose is that it should be related to grammatical number. The discussion of le/les in du/des will lead me to make a parenthesis on the-NPs, i.e., traditional definite noun phrases. Concerning the meaning of du/des-constituents, there is much more unanimity in the literature. Scholars agree that they have a highly indeterminate meaning as to the things/stuff they refer to and as to the quantity of the things/stuff they refer to (Bosveld-de Smet 1998). The study of du/des-NPs will lead me to examine partitivity in more details. We will see that du/des-NPs represent two kinds of partitives, ‘implicit’ and ‘explicit’ partitives. The former have the covert partitive reading typical of quantificational noun phrases, i.e., they express a subset (Chapter 2 Section 2.2), whereas the latter express the set overtly. More generally, the investigations reported here will suggest that partitivity and quantification should be distinguished. The chapter is organised as follows. The rest of this section provides some background information on du/des-constructions. Section 1.2 reports the traditional ­interpretations attributed to du/des-NPs and Section 1.3 the structural analyses corresponding to the two main readings, namely the partitive and the quantitative interpretation, in a pre-DP framework. We will see that the partitive reading is the least common one, as it is mainly restricted to the complement position of a small number of verbs. Section 2 shows that des-NPs can be property-denoting, quantificational and S-referential whereas du-NPs can

Chapter 3.  Du/des-NPs 



only be property-denoting and quantificational. Section 3 deals with the structure of du/ des-NPs. Their left periphery is addressed in Section 3.1 and their inflectional domain in Section 3.2. In the latter section, du/des-NPs are compared to un-NPs but also to le/lesNPs. De-NPs and en pronominalisation will also be looked at to a minor extent. Section 3.3 examines ‘explicit’ partitives in more details. We will see that they contain a partitive PP, in contrast with all the other du/des-NPs, and that this PP does not necessarily lead to a quantificational reading. In other words, partitive noun phrases are not always quantificational, i.e., QPs. Section 3.4 consists of some observations concerning the-NPs with regard to the analysis developed here. Section 4 is the conclusion to this chapter.

1.2  The ambiguity of du/des-constructions The first observation in order in the study of du/des-constituents is that the label ‘partitive article’ for du/de la/des ‘of.the’ is misleading. The term partitive is used to name the part-of relation. The syntactic expression of this relation can take different forms. In English for example, it can be expressed in compounds such as mountain-top … Most importantly, partitives have the frame det1 (one) of det2+common noun illustrated in (2), where the relevant structures are indicated in italics:

(2) a. b. c. d. e. f.

Some of the senators were acting strange. (Hoeksema 1996: 2) None of us left before 11 p.m. One or two of the first 70 members had died by then. Many of his friends are still at the funeral. Not too many of these problems have been solved. Every single one of a number of proposals was rejected.

In (2), the part-of relation involves a group or a collection and its members. This is not always the case as in (3) which illustrates mass partitives.

(3) a. b. c. d. e. f.

Most of the city is off-limits to foreigners. (Hoeksema 1996: 2) Some of him had stayed behind in his native Rumania. Some of the water was murky. We did not get to see all of her new garden. Half of every donation goes to administrative costs. Rick is not much of a hero.

In these examples, det1 is a mass noun determiner and the part-of relation is expressed either between a quantity of some substance and its subquantities or between an individual and its parts.

.  When an upstairs indefinite determiner co-occurs with a downstairs determiner, the question arises whether the distribution of the whole partitive construction will be determined by the indefiniteness of the former or by the definiteness of the latter.

 The Layered DP: Form and meaning of French indefinites

The partitives in (2) and (3) contrast with bare partitives which are partitives whose prepositional phrase is not preceded by a quantitative or quantificational element. Such constructions, illustrated in (4), are rare in English.

(4) a. Again Tarzan came down into the village and renewed his supply of arrows and ate of the offering of food which the blacks had mad to appease his wrath. (From: E. Rice Burroughs, Tarzan of the Apes)  (cited in Hoeksema 1996: 15(34a)) b. In the breast of his blouse he carried some coarse dark bread; he ate of this between whiles, and sat munching and drinking near Madame Defarge’s counter: (From: Ch. Dickens, A Tale of two Cities)  (cited in Hoeksema, 1996: 16(34b))

In (4), the partitive expressions, involving of, indicate that the object of the verb only partly undergoes the action of eating. In other languages, such noun phrases may be in the partitive case. In French, if the du/des-constituents illustrated in (1) and repeated below for convenience are partitive as their label suggests, they should be treated on a par with of the offering … and of this … in (4), i.e., as bare partitives.

(5) Elle a mangé du gâteau/ de la tarte/ des biscuits she has eaten of.the.masc cake of the.fem tart of.the.pl biscuits ‘She ate (some) cake/tart/biscuits.’

However, conflating du/des-constituents with bare partitives would be simplistic. In effect, not only may du/des-constituents have a non-partitive reading but this reading is also the predominant one (Bosveld-de Smet 1998; Roodenburg 2004a). The observation that du/des-NPs are ambiguous is not trivial. In effect, there is no consensus in the literature on the interpretation of noun phrases with ‘partitive articles’. For some authors, they are unambiguously partitive (Chierchia 1998a), for others they are unambiguously indefinite (Storto 2001) or ambiguous between a partitive and an indefinite reading (Delfitto 1993). To illustrate the ambiguity of du/des-constituents consider the following example (Kupferman 1979: 7):

(6) J’ai mangé du gâteau, I have eaten of.the. cake

a. dont il a aussi mangé d’ailleurs. b. ce qu’ a. of.which he has also eaten besides b. this which

In (6) du gâteau ‘of.the cake’ can be relativised by two different relative constructions, one with dont ‘of.which’ and the other one with ce que ‘this which’. The former is

.  Dont ‘of.which’ can also have a genitive interpretation as in (i):

Chapter 3.  Du/des-NPs 



characteristic of part-of relations, whereas the latter cannot have such a reading. In other words, du gâteau ‘of.the cake’ can be partitive or not. The ambiguity of examples like (6) has also been noticed by Milner (1978: 77) who mentions the following examples:

(7) a. b.

J’ai bu du vin que tu m’ as apporté. I have drunk of.the wine that you to.me have brought J’ai mangé des gâteaux que tu as faits. I have eaten of.the cakes that you have made

According to Milner, (7a) can mean “I had an undetermined quantity of an undetermined wine (and you had brought this wine)” or “I had an undetermined quantity of the determined wine that you had brought” (the question of the undetermined quantity will be addressed in Section 3.2.1). Milner analyses the first interpretation as a quantitative constituent with an appositive relative clause and the second one as a partitive constituent with a restrictive relative clause. He makes a similar proposal for (7b). Crucially, the term quantitative differs from the term quantificational (Chapter 2 Section 2.2) in that it represents quantities and not proportions. The elements Milner considers as quantitative are reported in (8) (1978: 34). Only du and des in (8a) will be considered in this chapter.

(8) a. articles: un ‘a’, du ‘of.the’, des ‘of.the.pl.’ b. numerals: un ‘one’, deux ‘two’, trois ‘three’… c. aucun ‘none, certains ‘some’, plusieurs ‘several’, quelques ‘some’ d. adverbs of quantity: beaucoup ‘a lot’, (un) peu ‘(a) little’, (le) plus ‘(the) more’, (le) moins ‘(the) less’, davantage ‘more’, combien ‘how much’, autant ‘as much’

Following Belletti (1979: 1546), Cardinaletti & Giusti (2002: 10(28) published in 2006), building on their 1992 paper, also observe that quantitative and partitive constructions have different meanings:



(i)

L’article dont je suis l’ auteur a été publié… the article of.which I am the author has been published Some native speakers of French use dont ‘of.which’ very rarely nowadays. They tend to consider it as old-fashioned and replace it with ce que. However, when dont is partitive, ce que cannot express the ‘part-of ’ meaning. Thanks to Claire Forel and Genoveva Puskás for these comments. .  That dont ‘of.which’ is ungrammatical in quantitative constructions is illustrated below: (i) *des livres dont j’ai lu beaucoup/deux of.the books of.which I have read many/two

(Milner 1978: 76)

 The Layered DP: Form and meaning of French indefinites



(9) a. b.

Ho (I).have Ho (I).have prestato. lent

letto cinque libri che mi avevi prestato. read five books that (you) to.me had lent letto cinque dei libri che mi avevi read five of.the books that (you) to.me had

In (9a), the number of books that have been read corresponds to the number of books that have been lent. It is the quantitative reading. In (9b), however, the five books that have been read represent a subset of the books that have been lent, i.e., that there are more than five books that have been lent. It is the partitive reading. The semantic difference between (6a) and (6b) is however more subtle than the one between (9a) and (9b). (6a) means that the two participants had some cake and both have eaten of the same cake; it is clearly partitive, they had a part of the cake in question. To get this interpretation the cake has to be a cake in particular, i.e., S-referential in the terminology adopted in Chapter 2. (6b) however means that the two participants had some cake, but the listener/reader has no idea whether they ate from the same cake or not; they had an undetermined quantity/part of cake. The semantic difference just mentioned leads to an interesting comparison. Both sentences, (6a) and (6b), involve some cake, and as it has been noted in the literature (Milsark 1977; De Hoop 1992; Diesing 1992) some is ambiguous. It has a weak and a strong reading (SOME), as illustrated in (10). (10) a. Some senators were acting strange (There is a set of strange-acting senators). b. SOME senators were acting strange (Part of the senators is acting strange).  (Hoeksema 1996: 2)

Only the stressed SOME (10b) has a partitive reading which can be formally expressed by a partitive phrase, as in some of the senators were acting strange. The weak some, often marked as sm, has a quantitative reading. In the same vein, the second part of (6) can be paraphrased differently, as illustrated in (11): (11) a. part of the cake has been eaten by him (6a) (cf. (10b)) b. there is a quantity of cake that has been eaten by him (6b) (cf. (10a))

(11a) paraphrases the dont ‘of.which’ relative clause, i.e., it represents the part-of relation, whereas (11b) paraphrases the ce que ‘this which’ relative clause, i.e., it is quantitative. Although du/des-constituents may be ambiguous, the partitive reading is constrained as shown in (12) (Kupferman 1979: 7):

.  This is also observed by Kupferman (1979: 10).



Chapter 3.  Du/des-NPs 

(12) Il a vu du gâteau he has seen of.the.masc cake a. * dont nous avons aussi vu d’ailleurs. * of.which we have also seen besides b. ce qu’ nous avons aussi vu d’ailleurs. this which we have also seen besides

(12), contrary to (6), is not ambiguous. In (12a), the partitive relative pronoun dont ‘of.which’ is ungrammatical, suggesting that the verb voir ‘see’ in (12) differs from the verb manger ‘eat’ in (6) in that it cannot be followed by a du/des-NP with a partitive meaning. Put differently, du/des-NPs are not always ambiguous and their interpretation depends on the verb selecting them. Kupferman (1979, 1994, 1998) proposes to classify these verbs into two groups reported in (13). (13) a. Type D/I: manger ‘eat’, boire ‘drink’, prendre ‘take’… b. Type D: voir ‘see’, montrer ‘show’, apporter ‘bring’…

Verbs of Type D/I consist of verbs whose du/des-complement is ambiguous between a partitive and a quantitative reading. Verbs of Type D in contrast consist of verbs whose complement only has the quantitative reading. D and I stand for the type of complement selected, i.e., direct and indirect object respectively (Kupferman 1979: 7). Type D/I corresponds to what Englebert (1992) calls ‘fragmentative’ verbs. As the label indicates, this class is restricted to verbs whose complement can be split, i.e., to roughly twenty verbs. This means that du/des-complements with a partitive reading are restricted to verbs of Type D/I and thus that this interpretation is not the most frequent one (the distribution of partitive du/des-NPs is further discussed in Section 2.2). Another reading of du/des-NPs which has not been mentioned so far is the one which corresponds to bare nouns in English (thanks to H. Borer for bringing this to my attention): (14) a. John bought apples and bread. b. Jean a acheté des pommes et du pain. Jean has bought of.the.pl apples and of.the.masc.sg bread

What differentiates this reading from the quantitative one seems to be the absence of quantity (Borer 2005a). In other words, des pommes corresponds to some apples when it is quantitative and to apples when it is not. Similarly du pain may be interpreted as

.  Milner also mentions some constraints on the ambiguous constructions illustrated in (7). Only a limited number of verbs (Kupferman’s Type D/I and also for example read…) and their complement – in opposition to the subject position – are concerned (1978: 78).

 The Layered DP: Form and meaning of French indefinites

some bread or bread. In the former cases, there is some undefined quantity which is expressed whereas in the latter no quantity is expressed, although this distinction is a very subtle one. Clear contexts may help differentiate these readings as the following examples show (C. Laenzlinger, p.c.): (15) a. b.

J’ ai acheté du sel mais pas beaucoup. I have bought of.the salt but not much ‘I bought (some) salt but not much.’ J’ ai acheté des livres mais pas beaucoup. I have bought of.the books but not many ‘I bought (some) books but not many.’

(16) (In a shop) Ici on vend du tabac. here one sells of.the tobacco ‘Tobacco on sale here.’ (17) Speaker A: Vous vendez des cigarettes? you sell of.the cigarettes ‘Do you sell cigarettes?’ Speaker B:

Non, ici on vend des livres et no here one sells of.the books and des journaux. of.the newspapers ‘No, we sell books and newspapers.’

The examples in (15) are clearly quantitative as the quantity can be contrasted, whereas in (16) and (17) we are rather dealing with a ‘kind/concept’, in a non theoretical sense. As the difference between these two interpretations does not have any impact on the phrasal category of du/des-NPs or on the structure of their left periphery, I will only come back to it in Section 3.2. In the rest of this book, the difference between ‘quantitative’ and ‘non-quantitative’ du/des-NPs illustrated above will thus only be mentioned in the sections specifically dealing with Quantity. Otherwise, the term ‘quantitative’ will be used in opposition to ‘partitive’ as in much literature on the topic (Section 1.3 for example). The take-home message of this section is that du/des-constituents may have several interpretations and that the partitive one is not predominant. This means that the label ‘partitive article’ found in the literature for du/des is inappropriate. In the absence of a better term, I will however stick to it. In the next section, I turn to the structures postulated for du/des-NPs in a pre-DP framework, i.e., before noun phrases were considered as DPs. The relevance of such a section is that it shows that quantitative and partitive du/des-NPs belong to different

Chapter 3.  Du/des-NPs 



categories, thus indirectly supporting the analysis developed in this book, namely, that in a framework with a strict syntax-semantics mapping, semantically different noun phrases are of different ‘sizes’. The ‘layered’ analysis argued for in Chapter 2 will be extended to du/des-NPs in section 3.1.

1.3  Structures in a pre-DP framework Milner (1978) and Kupferman (1979) argue that quantitative and partitive noun phrases have different structures (see also Roodenburg 2004a), contrary to Zribi-Hertz (2003) for example who defends a unitary structure. The first argument supporting the idea that partitive and quantitative du/desconstituents represent different categories comes from coordination. Consider the following examples (Milner 1978: 71ff.): (18) a. b.

J’ ai reçu beaucoup I have entertained many et de mes amis. and of my friends J’ai reçu beaucoup I have entertained many

de mes voisins of my neighbours

de voisins et d’ amis. of neighbours and of friends

(19) * J’ ai reçu beaucoup de mes voisins et d’ amis. I have entertained many of my neighbours and of friends

In (19), de mes voisins ‘of my neighbours’ and d’amis ‘of friends’ can clearly not be coordinated, contrary to de mes voisins ‘of my neighbours’ and de mes amis ‘of my friends’ in (18). As two constituents have to be of the same category to be coordinated, these facts strongly suggest that de mes voisins ‘of my neighbours’, which is partitive, and d’amis ‘of friends’, which is quantitative, belong to different categories. Milner proposes that the former is an N’’ and the latter an N’ (1978: 48). In the same vein, Kupferman (1979) suggests that the former is a PP and the latter an NP, which would correspond to DP, in the sense of topmost functional projection, in a more recent ­(pre-Split-DP) terminology. The arguments for such a differentiation are reported in the next two sections.

.  The label used by Kupferman (1979) is GN for ‘Groupe Nominal’, i.e., noun phrase.

 The Layered DP: Form and meaning of French indefinites

1.3.1  The partitive constituent In his very influential work of the late seventies, Milner (1978) observes that, in addition to coordination, extraction facts (20)–(21) distinguish between partitive and quantitative expressions (1978: 71 ff.):,10 (20) a. b.

C’est de Zola que j’ ai lu deux livres. it is of Zola that I have read two books C’est de Zola que j’ ai lu beaucoup de livres. it is of Zola that I have read a.lot of books

(21) a. * C’est de Zola que j’ ai

lu

deux des

livres.11

it is of Zola that I have read two of.the.pl books

b. * C’est de Zola que j’ ai lu beaucoup des livres. it is of Zola that I have read a.lot of.the.pl books

(20) shows that de Zola ‘of Zola’ can be extracted out of the quantitative phrases deux livres ‘two books’ and beaucoup de livres ‘many books’ whereas it cannot be extracted out of the partitive phrases deux des livres ‘two of the books’ and beaucoup des livres ‘many of the books’ (21). This difference is attributed to the internal structure of these constituents. The idea is that de Zola cannot move out of des livres in (21) because both are Prepositional Phrases (PP) (see Cardinaletti & Giusti 2002: 9). That a PP cannot cross another one has indeed been proposed in the literature (Abeillé et al. 2004 for a recent reference). As deux livres and beaucoup de livres in (20) are not PPs, de Zola can

.  Selkirk (1977) also proposed that simple quantitative expressions and those involving a partitive phrase have independent structures. .  The examples in (20)–(21) do not illustrate bare du/des ‘of.the’ constructions to make sure that the constructions are partitive and not ambiguous between a partitive and a non partitive reading: if deux ‘two’ in (21) were omitted it would be impossible to tell that we are not dealing with a direct object, out of which it would be possible to extract de Zola ‘of Zola’. See (22). .  The ungrammaticality of this example should not be attributed to the definiteness of les livres ‘the books’. In (ic) below extraction out of the definite noun phrase those reviews is fine:



(i) a. Two reviews have been reprinted of Helen’s first symphony. b. * Two of those reviews have been reprinted of Helen’s first symphony. c. Those reviews have been reprinted of Helen’s first symphony.  (Cardinaletti & Giusti 2002: 9(22))

In (ib), however, the extraposition of the PP of Helen’s first symphony leads to ungrammaticality. This is because this constituent would have to cross another PP, the partitive structure of those reviews. When there is no preposition as in two reviews in (ia), of Helen’s first symphony can be extraposed.

Chapter 3.  Du/des-NPs 



be extracted.12,13 This implies that in (21) the du/des-constituents des livres are PPs, headed by the preposition de ‘of ’. If this analysis is on the right track, it also accounts for the fact that (22) cannot have a partitive reading (Luigi Rizzi, p.c.): (22) C’est de Zola que j’ai lu des livres. it is of Zola that I have read of.the books

As de Zola can be extracted without leading to ungrammaticality, des livres in this example is not a PP. If partitive du/des-NPs are PPs, des livres in (22) cannot be partitive. In other words, as des livres is not a PP, it is interpreted as quantitative. This supports the idea that partitive and quantitative noun phrases belong to different categories. If des livres in the partitive construction deux des livres in (21) is a PP, the ­analysis should extend to bare partitives, i.e., to des biscuits in its partitive reading in J’ai mangé des biscuits ‘I have eaten of.the biscuits’. This is supported by the grammaticality of the interrogative locution de quoi ‘of what’ with verbs of Type D/I to which manger ‘eat’ belongs to (23) (see also Section 1.2) (Kupferman 1979: 7): (23) a.

Il m’ a demandé de quoi je mangeais. he me has asked of what I was.eating je buvais. I was.drinking je prenais. I was.taking

b. Il m’ a demandé de quoi * il avait vu. he me has asked of what he had seen * il montrait. he was.showing * il apportait. he was.bringing

As the prepositional locution de quoi ‘of what’ can replace a du/des-construction in its partitive reading (23a), it suggests that the latter is a PP too. More precisely, the argument

.  Note that this implies that de ‘of ’ in beaucoup de livres ‘many of books’ differs from de ‘of ’ in beaucoup des livres ‘many of.the books’: only in the former case can de Zola ‘of Zola’ be extracted (cf. Sections 3.2.2 and 3.3 for a distinction between two de’s: in a nutshell the idea is that in beaucoup de livres, de is a functional head of the inflectional domain whereas in beaucoup des livres it is a preposition). .  Cf. Kupferman (1999: 45–6) for further considerations on the (non)-extraction of partitive de-phrases out of PPs.

 The Layered DP: Form and meaning of French indefinites

is that de in de quoi corresponds to the de present in du/des-constituents and that quoi replaces the rest of the nominal expression, for example, les biscuits ‘the biscuits’. If de in de quoi is a preposition as assumed, it means that it is too in the du/des-constituent replaced by de quoi. If du/des-constituents in their partitive reading are prepositional, they should share some properties with abuser de ‘abuse of ’ and être content de ‘be happy of ’, for example, which take the preposition de. This prediction is borne out as shown in (24) from Kupferman (1979: 5–6): (24) a. b. c.

Il abuse de la bière. he abuses of the beer est content is happy boit drinks La bière dont il abuse/ est content/ boit est … the beer of.which he abuses/ is happy/ drinks is … De quoi abuse-t-il?/ est-il content?/ boit-il ? of what abuses he is he happy drinks he

In (24), abuser de ‘abuse of ’ and être content de ‘be happy of ’ exhibit the same properties as boire ‘drink’, which is of Type D/I, i.e., they are grammatical with the relative pronoun dont ‘of.which’ and with the interrogative locution de quoi ‘of what’. This suggests that if de in abuser de and être content de is a preposition, it is one too in the partitive du/des-constituents. Note however that I do not make any implications as to the interpretation of de in abuser de and être content de. In (24b), for example, dont ‘of.which’ in la bière dont il abuse ‘the beer of.which he abuses’ and in la bière dont il est content ‘the beer of.which he is happy’ does not have a partitive use but a genitive one.14

1.3.2  The quantitative constituent That constituents introduced by ‘partitive articles’ are not always prepositional has been shown in Section 1.2. At the beginning of Section 1.3, it was suggested that they are DPs, in the recent terminology I will adopt in this section.15 Evidence for this .  See Zribi-Hertz (2003) for additional remarks on the restrictions on de quoi ‘of what’ and dont ‘of.which’. .  Kupferman (1994: 196 ff. and following works) analyses de ‘of ’ in quantitative constituents as a quantifier. This is incompatible with the definition of quantification advocated in Chapter 2. In our terms, it is the partitive reading which has the structure of a quantificational constituent (Section 2.2 of this chapter).

Chapter 3.  Du/des-NPs 



conclusion comes from their being grammatical in subject positions. Consider (25) and (26) (adapted from Kupferman 1979: 8). (25) a. b.

La corde traînait par terre. the rope was.lying on the floor De la corde traînait par terre. of the rope was.lying on the floor

(26) a. b.

L’ eau boueuse recouvrait la route. the water muddy was.covering the road De l’ eau boueuse recouvrait la route. of the water muddy was.covering the road

Assuming that subjects have to be DPs and cannot be PPs, de la corde ‘of the rope’ in (25b) and de l’eau ‘of the water’ in (26b) have to be treated on a par with la corde ‘the rope’ and l’eau ‘the water’ respectively, i.e., as DPs.16 Additional data suggest that du/des-constituents in their quantitative reading are DPs. More precisely, negation, en-pronominalisation and existential constructions show that quantitative du/des-NPs are analogous to un-NPs, i.e., to a nominal expression in opposition to a prepositional phrase. The analogy between un-NPs and du/des-NPs corresponds to the ‘traditional conception’ of du/des (Frei 1960). Consider the following examples from Frei 1960, mentioned in Kupferman 1979: 1–2: (27) a. b. c.

Il a du papier. a’. he has of.the.masc. paper Il a un papier. b’. he has a/one paper Il a des papiers. c’. he has of.the.pl. papers

Il n’ a pas de papier. he ne has not of paper Il n’ a pas de papier. he ne has not of paper Il n’ a pas de papiers. he ne has not of papers

In negative contexts (27a’, b’, c’), papier ‘paper’ is preceded by de ‘of ’, independently of the article present in the positive counterpart, i.e., un ‘a’ or du/des ‘of.the’ (see Chapter 2 Section 2.3.3 on de-NPs). With definite noun phrases, in contrast, the definite article remains in negative contexts as shown in (28a’, b’): (28) a.

Il a le papier. a’. he has the paper

Il n’ a pas le papier. he ne has not the paper

.  Kupferman (1994: 200). Note that in English, examples with a prepositional subject (i) are grammatical. Such examples are however highly restricted.

(i)

a. After four would be the best time for me. b. Under the table is a good place to hide.

(Haegeman & Guéron 1999 :119)

 The Layered DP: Form and meaning of French indefinites

b. Il a les papiers. b’. he has the.pl. papers

Il n’ a pas les papiers. he ne has not the papers

The choice of a pronoun also puts du/des-NPs on the side of indefinite noun phrases: (29) a. b. c. d. e.

Il a du papier. a’. he has of.the paper Il a un papier. b’. he has a/one paper Il a des papiers. c’. he has of.the papers Il a le papier. d’. he has the paper Il a les papiers. e’. he has the papers

Il en a. he en has Il en a un.17 he en has a/one Il en a. he en has Il l’ a. he it has Il les a. he them has

In (29a’–c’), the pronoun is en whereas in (29d’–e’) it is a definite pronoun. The former replaces du/des papiers and un papier (for a brief discussion, see Section 3.2.1) and the latter le papier ‘the paper’ and les papiers ‘the papers’. Here again, du/des-NPs behave like un-NPs, in opposition to PPs. Finally, impersonal structures such as (30) support this analogy. (30) a. b. c. d. e. f.

Il est arrivé une lettre. there is arrived a letter Il est arrivé des lettres. there is arrived of.the letters Il est arrivé du courrier. there is arrived of.the mail Il est arrivé * la lettre. there is arrived the letter Il est arrivé * les lettres. there is arrived the letters Il est arrivé * le courrier. there is arrived the mail

Existential constructions (30) only admit indefinite subjects. As des lettres ‘of.the letters’ and du courrier ‘of.the mail’ are grammatical in this context, it suggests that these

.  Un ‘a/one’ in (29b) is ambiguous between an indefinite article and a numeral, which are homonyms in French (Milner 1978: 28).

Chapter 3.  Du/des-NPs 



constituents should be treated on a par with un-NPs and that they are therefore not prepositional. What the above discussion shows is that du/des-NPs in their quantitative reading are analogous to un-NPs in the sense that they are noun phrases and not prepositional phrases. However, as un-NPs come in three different types, namely Property-un-NPs, Quantificational-un-NPs or S(peaker)-Referential-un-NPs (Chapter 2), the question which arises is to which type, if any, quantitative du/des-NPs correspond. This is on the agenda of the next section.

2.  Types of du/des-NPs 2.1  S-referential des-NPs In Chapter 2 reference of indefinites was mainly defined in terms of speaker’s reference (sections 2.1.1 and 3.4.3). This also applies to ‘referential’ du/des-NPs as their referent is often only known by the speaker. Although this reading is not the easiest to get for du/des-NPs, it is attested, at least for des-NPs (Kleiber 1981b; Bosveld-de Smet 1998; Dobrovie-Sorin & Beyssade 2004). The relevant examples may involve specifying relative sentences/adverbials as in (31) or not (32). S(peaker)-referential des-NPs will henceforth be labelled SR-des-NPs. (31) Des amis que j’ ai rencontrés hier of.the friends whom I have met yesterday m’ ont insulté (referential).18 have offended me (Bosveld-de Smet 1998: 33(67a))

.  In examples such as (ia) below, some indeterminacy arises from the fact that we do not know whether friends visited the speaker as a group, in subgroups or individually. Similarly, in (ib), without any pragmatic or contextual clues, the preferred readings are total (one event), collective and cumulative (Bosveld-de Smet 2004). As mentioned in Section 1.1 an analysis of all these readings would require a thorough study of the predicates, which is out of the scope of this book.

(i)

a. b. 

Des amis sont venus me voir, lorsque j’étais malade. of.the friends came to visit me, when I was ill Après l’explosion, de la vapeur s’est échappée de la chaudière. after the explosion, of the steam has escaped from the boiler (ibid (29))

 The Layered DP: Form and meaning of French indefinites

(32) a. b.

Des enfants jouent dans la cour. of.the kids are.playing in the yard Il s’ agit de Pierre, Paul et Jacques. it se is.about of Pierre, Paul and Jacques ‘Some children are playing in the yard, namely Pierre, Paul and Jacques.’ J’ai acheté des livres. I have bought of.the books Il s’ agit de Hamlet, Moby Dick et Sula. it se is.about of Hamlet, Moby Dick and Sula ‘I bought some books, namely Hamlet, Moby Dick and Sula.’

In (31) and (32), the friends and the children are respectively identified by a relative clause or by another identificational means such as an il s’agit de… ‘to be about’ sentence. Additional examples of SR-des-NPs are provided below: (33) a. b. c. d. e. f. g. 

Des verres sont vides. of.the glasses are empty Des fauteuils sont bancals. of.the armchairs are rickety Des verres sont ébréchés. of.the glasses are chipped Des fourchettes sont sales. of.the forks are dirty Des fleurs sont fanées. of.the flowers are faded Des ampoules sont usagées. of.the bulbs are old Des cendriers sont pleins. of.the ashtrays are full (Dobrovie-Sorin & Beyssade 2004: 117(9), from Galmiche 1986)

In (33), the des-NPs can be replaced by quelques ‘several’, which represents evidence for their S-referential reading (in opposition to certains ‘some.pl.’ which would be partitive). Additional support for the S-referential reading comes from the type of predicates illustrated in (33). Examples like (33) are however restricted to particular contexts. For instance, they are only felicitous when the predicate expresses a property which is directly perceivable. On a par with SR-un-NPs, SR-des-NPs may have ‘widest scope’ and are existential.19 Consider the following examples:

.  Recall from Chapter 2 Section 3.4.1 that I adopted the term ‘widest scope’ to avoid the ambiguity arising from ‘wide scope’.



(34) a. b. 

Chapter 3.  Du/des-NPs 

Chaque dimanche, des amis viennent nous visiter. every Sunday, of.the friends come visiting us Des enfants viennent jouer ici tous les jours. of.the children come playing here all the days (Bosveld-de Smet 1998: 38 (78b), (79b))

The above sentences are ambiguous. They may be about the same friends and the same children or about any friends and any children. The latter corresponds to the property-denoting interpretation whereas the former is the widest scope reading which corresponds to the S-referential interpretation. Although the property reading seems to be preferred, the S-referential one is not impossible. In both cases, the des-NPs are existential (although not intrinsically) in the sense that if there are no friends of us who come and visit us every Sunday and no children who play there every day, the sentences are false. As for du-NPs, the S-referential reading is considered as impossible (Bosveld-de Smet 1998: 33; Dobrovie-Sorin & Beyssade 2004: 48): (35) * De l’ étoffe que j’avais achetée hier traînait par terre.20 of.the material that I had bought yesterday lay about on the floor  (Bosveld-de Smet 1998: 33(68))

An explanation could be that mass contrary to plural noun phrases do not have minimal parts (Landman 1989, 1991). Mass nouns do not have a set of atoms representing minimal entities in their extension (Bunt’s 1985 homogeneity hypothesis) (see Section 3.2 for details on their syntactic structure).21,22 In other terms, as reference means picking out some entity(ies), this is not possible in the absence of entities, i.e., for mass terms.23 If this is on the right track, it accounts for the fact that mass noun phrases show a clear reluctance to take on widest scope (Dobrovie-Sorin & Laca 2003; Bosveldde Smet 1998: 69ff.). This is because in the analysis presented here widest scope is

.  As observed by Bosveld-de Smet (1998: 33), when the relative sentence que j’avais achetée hier ‘that I had bought yesterday’ modifies l’étoffe and not de l’étoffe, the sentence is more easily accepted. .  What mass and plural count nouns share however is their cumulative reference (Link 1983). .  Contra Chierchia 1995 who proposes that all nouns, including mass nouns, have minimal parts in their extension. .  Doetjes argues that the class of mass nouns is not homogeneous. She distinguishes between ‘mass mass nouns’ and ‘count mass nouns’. The former do not have minimal parts and include words like water, whereas the latter do have minimal parts and concern words like furniture (Doetjes 1997: 30ff.).

 The Layered DP: Form and meaning of French indefinites

associated with the S-referential interpretation. As du-NPs cannot be S-referential, they cannot take widest scope either. The use of verbs like aimer ‘like/love’ and détester ‘hate’ and predicates like être intelligent/grand/beau/savoureux… ‘to.be intelligent/tall/handsome/tasty’ … allows us to differentiate between du-NPs and des-NPs. According to Dobrovie-Sorin & Beyssade (2004: 43), such verbs are entity-verbs precisely because they can only take referential arguments (Dobrovie-Sorin 1997a,b). If du-NPs cannot be S-referential, it predicts that they should be ruled out of such contexts. This seems to be confirmed by (36): (36) a. Des enfants étaient grands. of.the kids were tall ‘(Some) kids were tall.’ b. * Du café était savoureux. of.the coffee was tasty

The ungrammaticality of (36b) shows that du café ‘of.the coffee’ is not referential, contrary to des enfants ‘of.the children’ in (36a). As for the assertion of existence attributed to S-referential des-NPs, it can be tested. For instance, if there are not children Pierre, Paul and Jacques, (32a) is false. In other words, the des-NP is existential. The same reasoning applies to the other examples discussed in this section.

2.2  Quantificational du/des-NPs As seen in Chapter 2, quantificational noun phrases cannot be pointed at. They involve a contextually determined set and presuppose existence. These are well-known properties of cardinals (Milner 1978). Consider (37): (37) Deux filles étaient blondes, toutes les autres avaient two girls were blond all the others had les cheveux foncés. the hair dark (Dobrovie-Sorin & Beyssade 2004: 31(40))

That such examples can be quantificational is supported by the tripartite structure reported below (Chapter 2 Section 2.2). This interpretation implies that the two blond girls cannot be pointed at, i.e., it could be any two girls of the contextually determined set of girls. (38) Deux x (fille (x)) [x était blonde]

(Dobrovie-Sorin & Beyssade 2004: 31(41))

As the counterpart of (37) with a des-NP is perfectly fine, it suggests that des filles ‘of. the girls’ in such examples may be quantificational:

Chapter 3.  Du/des-NPs 



(39) Des filles étaient blondes, d’autres avaient les of.the girls were blond others had the cheveux foncés. hair dark

The interpretation of quantificational des-NPs is close to the one of certains NPs ‘some.pl’. However, the quantificational reading of des-NPs is more difficult to obtain. A context favouring such a reading is the use of the anaphoric d’autres ‘of others’ illustrated in (39). This corresponds to the marked use of des, i.e., des filles ‘of. the girls’ is contrasted with others (Attal 1976). Additional examples are reported below: (40) En général, les membres de ce département assistent à toutes les réunions. Mais tous ne sont pas toujours là. ?Ainsi, des professeurs ont séché la réunion d’hier. ‘Generally the members of this department attend all the meetings. But they are not always all present. So, some professors cut yesterday’s meeting’.  (adapted from Laca & Tasmowski 1996 quoted in Roodenburg 2004a: 54) (41) a. b. 

Des licornes sont blanches et d’autres noires. of.the unicorns are white and others black (Kleiber 1981a) Des enfants ont faim, tandis que d’autres ont soif. of.the children are hungry while others are thirsty (Bosveld-de Smet 1998: 15(17a-b))

As the following example shows, the contrast does not have to be overt: (42) Des élèves ne m’ ont pas rendu leurs devoirs. of.the students ne me have not give.back their homework.pl  (Dobrovie-Sorin & Beyssade 2004: 51(41a))

In (42), the des-NP is quantificational if the students who did not hand in their homework cannot be pointed at. In addition, it is clearly partitive. It concerns a subset of a group of students contextually determined, i.e., it has the covert partitive reading typical of quantificational noun phrases. Interestingly, examples (37)–(42) also show that covert partitive des-NPs can function as subjects in contrast with bare partitive du/des-NPs in Section 1.2. In their quantificational reading, des-NPs are expected to be presuppositional. This means that a sentence should be neither true nor false but lead to a presupposition failure if no such des-NP exists. Consider the following example: (43) Paul a rencontré des étudiantes à toi. Paul has met of.the students.fem of yours ‘Paul met some students of yours.’

 The Layered DP: Form and meaning of French indefinites

If I were told (43) and had no female students, I would probably say something like mais je n’ai pas d’étudiantes ‘but I have no female students’ or ce n’est pas possible ‘it is not possible’ but not ce n’est pas vrai ‘it is not true’. The presupposition failure arising shows that the des-NP is presuppositional. Let us now turn to du-NPs. As (44) shows, the contrastive use illustrated above for des-NPs is impossible with du-NPs (Dobrovie-Sorin & Beyssade 2004: 51): (44) * Du café est noir, (de) l’ autre est au lait. of.the coffee is black of the other is with milk

At first sight this might suggest that du-NPs differ from des-NPs in that they may not be quantificational. However, the ungrammaticality of (44) is due to the co-occurrence of the mass noun and d’autres/de l’autre ‘of others/of the other’. In effect, as mass terms do not have minimal parts in their extension, they cannot be used with these contrastive expressions which are only felicitous with nominal expressions encoding minimal entities. That du-NPs may be quantificational is illustrated by the following examples. (45) a. b.

[There is a nice tart on the table, visible to A and B] Tu m’ apporterais de la tarte? you to.me would.bring of the tart ‘Would you bring me some tart?’ [There is a pot of hot coffee on the table, visible to A and B] Tu m’ apporterais du café ? you to.me would.bring of.the coffee ‘Would you bring me some coffee?’

In (45), de la tarte ‘of the tart’ and du café ‘of the coffee’ represent a subset of the tart and of the coffee clearly defined in the context. In other words, they fit the definition of quantification adopted here which implies a covert partitive meaning (Chapter 2 Section 2.2.2). The discussion so far shows that at least two kinds of partitive du/des-NPs should be distinguished, the covert partitives which are quantificational and the partitives restricted to object positions of fragmentative verbs which are not quantificational. I will sometimes use the term ‘implicit’ partitives for the former and ‘explicit’ partitives for the latter. The reason is that des filles in (39) for example represents a subset of a group which is contextually determined but implicit, i.e., not realised. This contrasts with the partitives selected by fragmentative verbs (Section 1.2) and illustrated below: (46) a. b.

Jean a mangé des biscuits. Jean has eaten of.the biscuits ‘Jean ate (some of the) biscuits.’ Jean a bu du café. Jean has drunk of.the coffee ‘Jean drank (some of the) coffee.’

In (45), des biscuits and du café overtly express the set. This difference is encoded in their structure and in Section 3.3 we will see that only explicit partitives contain a PP.

Chapter 3.  Du/des-NPs 



The structure of covert partitives is addressed in Section 3.1.2. The existence of two kinds of partitivity suggests that partitivity and quantification should be differentiated in the sense that not all partitives are quantificational (Dobrovie-Sorin & Beyssade 2004, reach a similar conclusion).

2.3  Property du/des-NPs In Chapter 2, several characteristics have been attributed to property-denoting unNPs. One of them is their lack of scope properties and another one their potential lack of existential entailment. In addition, we have seen that they are banned from negative contexts. As for their interpretation, it is undetermined in the sense that it denotes a property. Consider (47) which means that Mary wants to talk to any psychiatrists and not to some psychiatrists in particular (Roodenburg 2004a: 53). In other words, any person who has the property of being a psychiatrist will do. (47) Marie souhaite parler à des psychiatres. Mary wishes to.talk to of.the psychiatrists (Bosveld-de Smet 2004 (158))

In addition to the property interpretation, des psychiatres ‘of.the psychiatrists’ in (47) necessarily has an opaque reading. Put differently, the truth of the sentence does not entail the existence of any psychiatrist. If it did des psychiatres ‘of.the psychiatrists’ would be interpreted as a SR-des-NP with widest scope. To capture the potential lack of an existence entailment of property-denoting arguments, the semantics of the predicates has to be modified. This is feasible as seen in Chapter 2 Section 2.3.2. When Property du/des-NPs (P-du/des-NPs) are existential, it is the predicate which supplies the existential operator. This implies that property-denoting subjects may be existential on a par with property-denoting objects (in opposition to Diesing’s 1992 analysis): (48) De la glace tachait la nappe. of the ice-cream spotted the tablecloth

(Nolda 2006 (9))

In (48), de la glace ‘of the ice-cream’ is an existential property-denoting subject.24 Additional examples are provided below: (49) a. b.

Des étudiants ont manifesté. of.the students have demonstrated (Attal 1976) Il y a eu une manifestation d’étudiants. there has been a demonstration of students (Bosveld-de Smet 2004 (17))

(50) a. De l’ eau s’est évaporée. of.the water has evapored

.  Du/des-NPs in subject positions are constrained. For example, they cannot have a generic reading (Nolda 2006, fn. 8 citing Galmiche 1986) and are highly dependent on the availability of spatio-temporal boundaries (they are (nearly) impossible with individual-level predicates and subject to spatial and/or temporal anchoring with stage level predicates) (Bosveld-de Smet 2004).

 The Layered DP: Form and meaning of French indefinites

b. Il y a eu une evaporation d’eau. there has been an evaproation of water (ibid (18))

As Bosveld-de Smet (2004) observes, the paraphrases of (49a) and (50a) given in (49b) and (50b) respectively show that these du/des-NPs involve no specific referents and hence that they are P-du/des-NPs (see Attal 1976 Galmiche 1986). The indeterminacy of du/des-NPs has been stated in terms of identificational force or more precisely of lack thereof by Bosveld-de Smet (1998). According to her, this means that du/des-NPs do not give any clue as to the identity of objects/substance involved and that they never involve specific referents. This is why such noun phrases combine perfectly with modifiers denoting physical characteristics as in (51): (51) a. b.

Ici, on fabrique des vases here one manufactures of.the vases de divers styles. of various styles Ce magasin vend du papier de this shop sells of.the paper of de toutes couleurs. of all colors

de diverses tailles, of various sizes,

toutes épaisseurs, all thicknesses, (Bosveld-de Smet 2004(13))

However, contrary to Bosveld-de Smet, I argue that the lack of identificational force only concerns P-du/des-NPs. Consider the following example repeated from Section 1.2. (52) a. b.

J’ai bu du vin que tu m’ as apporté. I have drunk of.the. wine that you to.me have brought25 J’ ai mangé des gâteaux que tu as faits. I have eaten of.the cakes that you have made

(52a) can mean “I had an undetermined quantity of an undetermined wine (and you had brought this wine)” or “I had an undetermined quantity of the determined wine that you had brought”, and a similar distinction can be made for (52b) (Milner 1978: 78). The first reading represents the quantitative one and the second the partitive one. The wine is only undetermined in the former and not in the latter. In other words, explicit partitive du/des-NPs do not seem to lack identificational force. This is because le vin in du vin in (52a) is definite in the partitive reading but not in the quantitative .  The translation of de by of in the quantitative reading of du/des-constituents is inappropriate and misleading. Of would lead to a partitive reading, contrary to fact. In the absence of any better translation I will stick to the one in the text. More generally, note that although the English counterpart of French bare partitives also involves of, it is not the case for the counterpart of quantitative du/des-constituents.

Chapter 3.  Du/des-NPs 



reading (despite of the presence of le, see Section 3.2). However, as the quantity of wine is undetermined in the partitive interpretation (and in the quantitative one), the whole argument is understood as undetermined. In short, although the quantitative and the partitive interpretations seem to have different ‘degrees’ of undeterminacy, they are both property-denoting. What differentiates them, I argue, is there structure (Sections 3.1.1 and 3.3). This means that PropPs have to be classified into a quantitative and a partitive structure. As the above observations show, du/des-NPs can be Property noun phrases on a par with P-un-NPs. This predicts that they should share other characteristics, such as scope properties, or lack thereof (Bosveld-de Smet 2004; Roodenburg 2004a; see also Dobrovie-Sorin & Laca 2003). Consider the following example: (53) a.  b. 

Des voyous ont volé deux bicyclettes. of.the naughty.boys have stolen two bicycles (No more than two bicycles are involved in the robbery) Il y avait de la visite tous les jours. there were of.the people each day (Each day there are visitors, but not necessarily the same visitors) (Bosveld-de Smet 2004(20))

The preferred readings in (53) are the ones where des voyous and de la visite are in the scope of deux bicyclettes and tous les jours respectively. Widest scope of des voyous in (53a) would lead to an S-referential reading. Put differently, when des voyous is property-denoting, it is interpreted in situ. As for de la visite, its grammaticality in a there be sentence shows that it is a non-existential property-denoting noun phrase. The existential operator is represented by the there be part of the construction (Chapter 1 Section 5.1.1). The distribution of du/des-NPs in negative and existential contexts also supports the parallel between P-du/des-NPs and P-un-NPs. Recall (27) repeated as (54): (54) a. b. c.

Il a du papier. a'. he has of.the.masc. paper Il a un papier. b'. he has a/one paper Il a des papiers. c'. he has of.the.pl. papers

Il n’ a pas de papier. he ne has not of paper Il n’ a pas de papier. he ne has not of paper Il n’ a pas de papiers. he ne has not of papers

In (54a',c') the du/des-NPs are replaced by a de-NP, on a par with un papier ‘a paper’ in (54b'). In other words, du/des-NPs are impossible in negative contexts, exactly like P-un-NPs (Chapter 2, Section 2.3.3). Further evidence for this conclusion comes from the following contrasts. (55) Negative sentences a. * Marie n’ a pas mangé de la tarte. Marie ne has not eaten of the tart

 The Layered DP: Form and meaning of French indefinites

b. Marie n’ a pas mangé de tarte. Marie ne has not eaten of tart ‘Marie hasn’t eaten any tart.’ (56)

Without-clauses a. * Elle est revenue sans avoir mangé de la tarte. she is come.back without to.have eaten of the tart b. Elle est revenue sans avoir mangé de tarte. she is come.back without to.have eaten of tart ‘She came back without having eaten any tart.’

In negative contexts, the property reading is obtained with de-NPs. In other words, de-NPs are property denoting arguments restricted to the scope of a negative operator (Chapter 2 Section 2.3.3). Existential constructions also support the analogy between P-du/des-NPs and Pun-NPs (Chapter 2 Section 2.3). The following example is repeated from Section 1.3.2: (57) a. b. c. d. e. f.

Il est arrivé une lettre. there is arrived a letter Il est arrivé des lettres. there is arrived of.the letters Il est arrivé du courrier. there is arrived of.the mail Il est arrivé * la lettre. there is arrived the letter Il est arrivé * les lettres. there is arrived the letters Il est arrivé * le courrier. there is arrived the mail

As existential sentences take property-denoting noun phrases (Chapter 2 Section 2.3.2), the du/des-NPs in (57) must belong to this type. Another context which typically attests property-denoting noun phrases is habitual readings: (58) a. b.

Son mari écrit des romans. (e.g., for a living) her husband writes of.the novels Ce paysan cultive du maïs. (idem) this farmer grows of.the maize (Bosveld-de Smet 2004(4))

In sum, the above discussion shows that to account for the interpretative properties of du/des-NPs a three way classification is necessary. P-du/des-NPs share the same characteristics as P-un-NPs.

2.4  Conclusion In this section, we have seen that argumental des-NPs, on a par with argumental un-NPs, should be classified into three groups, corresponding to their semantic types.

Chapter 3.  Du/des-NPs 



This means that they can be S-referential, quantificational or property-denoting. Du-NPs, in contrast, only have the latter two readings. The different types of du/des-NPs exhibit the same characteristics as their un-NPs counterparts. In other words, SR-des-NPs are existential and can take widest scope (see also Section 3.1.3), Q-du/des-NPs are presuppositional and P-du/des-NPs do not have scope properties, are banned from negative contexts and may lack an existential entailment. I have also distinguished between implicit and explicit partitives. The former concern du/des-NPs which are quantificational in that they represent a subset of a contextually determined covert set. The latter concern du/des-NPs which occur in the complement position of fragmentative verbs and which represent a set. The differences are reported in Table 9 (59): (59) Table 9.

Terminology

Function

Category

Partitive du/des-NPs

1. Implicit (overt subset): Quantificational 2. Explicit (overt set): Not quantificational

subject or object object (with fragmentative verbs)

QP (Section 3.1.2) PropP (Section 3.3)

To complicate the picture, we have also seen that in addition to a partitive reading, property-denoting du/des-NPs can have a quantitative reading. Quantitative here is used in opposition to partitive (as in Section 1.3 but recall Section 1.2). This is represented in Table 10: (60) Table 10.

Function

Interpretation

Property subject Quantitative du/des-NPs du/des-NPs (Section 3.1.1) object du/des-NPs Quantitative Partitive (with fragmentative Vs)

Structure

Relative

No PP in the structure No PP in the structure (Section 3.1.1) Contain a PP (Section 3.3)

ce que (Section 1.2) dont (Section 1.2)

The study of du/des-NPs corroborates the observation made in the preceding chapter, namely that the postulation of property-denoting arguments requires the semantics of predicates to be refined. In other terms, the study of noun phrases and the one of predicates should not be dissociated. The next section focuses on the structure of du/des-NPs. Section 3.1 is about their left periphery, Section 3.2 about their inflectional domain. The latter addresses difficult questions such as the base position and role of de and le/les, the locus of plurality, quantity and atomicity. Section 3.3 concentrates on ‘explicit’ partitives and the structure of partitive noun phrases in general.

 The Layered DP: Form and meaning of French indefinites

3.  The structure 3.1  The left periphery 3.1.1  Property du/des-NPs A first observation is that du/des is not generated in the nominal left periphery but lower. The reason is that du/des-NPs can function as predicates as shown in (61) (in opposition to arguments). As predicates do not involve any projection of the left periphery, du/des must be generated in the inflectional domain. That predicates include projections of the inflectional domain is supported by the agreement features they carry (number and gender). The difference between syntactic arguments and predicates is repeated from Chapter 2 Section 3.1.1 as (62). (61) a. b.

Marie et Sophie sont des professeures/ avocates. Marie and Sophie are of.the professors.fem/lawers.fem ‘Marie and Sophe are professors/lawers.’ C’est du café. it is of.the coffee ‘It’s coffee.’

(62) A “nominal expression” is an argument only if it is introduced by one of the categories of the DP domain (the left periphery).

According to (62), the predicates des professeures and du café in (61) are truncated at the level of the inflectional domain. However, when such du/des-NPs function as arguments the lowest layer of the left periphery is projected, i.e., Property Phrase (PropP). Recall from Chapter 2 Section 3.2.1 that the possibility to have a PropP in their structure distinguishes nominal predicates from adjectival predicates for example. In other words, the latter cannot function as arguments contrary to the former. P-du/ des-NPs are illustrated in (63) and their structure in (64). (63) Marie voudrait acheter des plantes. Marie would.like to.buy of.the plants (opaque reading: any plants) (64)

PropP Prop′

Spec Prop desi

NumP ti plantes

Chapter 3.  Du/des-NPs 



The structure of P-du/des-NPs is analogous to the one of P-un-NPs, except that it is the leftmost element des which moves to Prop.26 It is this movement which triggers the projection of the layer Property Phrase (PropP). As PropPs do not contain any empty category (Chapter 2 Section 3.2.2), P-du/des-NPs are not restricted to the scope of an operator, on a par with P-un-NPs. Examples of P-du/des-NPs in object and subject positions have been provided in Section 2.3. Furthermore, as the structure of P-du/des-NPs does not contain any quantificational element there is no reason for these arguments to undergo Quantifier Raising (QR). This suggests that they are interpreted in situ and therefore that they do not have scope properties, exactly like P-un-NPs. (65), which illustrates this characteristic is repeated from Section 2.3. (65) a. b. 

Des voyous ont volé deux bicyclettes. of.the naughty.boys have stolen two bicycles (No more than two bicycles are involved in the robbery) Il y avait de la visite tous les jours. there were of.the people each day (Each day there are visitors, but not necessarily the same visitors) (Bosveld-de Smet 2004(20))

The intensional context in (63) shows that P-du/des-NPs do not necessarily entail existence. This type of argument is only existential when selected by an ‘existential’ predicate as in (66). (66) Jean a attrapé des lapins. Jean has catched of.the rabbits ‘Jean caught rabbits.’

The preferred reading of this example is the property-denoting one, where John caught any rabbits. If no rabbits existed, this example would be false. This shows that des lapins asserts existence. The source of this interpretation is the existential operator associated with the predicate (Chapter 2 Section 2.3.2).

3.1.2  Quantificational du/des-NPs According to the analysis developed in the preceding chapter, quantificational noun phrases have a two-layer left periphery. The PropP is projected to turn the inflectional domain into an argument and the Quantifier Phrase (QP) to encode quantification. The resulting structure for a quantificational expression like des filles ‘of.the girls’ in (67), repeated from Section 2.2, is given in (68).

.  The incorporation of de + le/les is addressed in Section 3.2.2.

 The Layered DP: Form and meaning of French indefinites

(67) Des filles étaient blondes, d’autres avaient les cheveux foncés. of.the girls were blond others had the hair dark (68)

QP Spec

Q′ Q desi

PropP Prop′

Spec Prop ti

NumP ti filles

In (68), the QP layer accounts for the quantificational reading of des filles, that is, girls who cannot be pointed at. The partitive reading of des filles comes from the fact that des filles represents a subset of a set which is contextually determined. This set is non-overt, i.e., it is not represented in the structure. In other words, (68) only concerns implicit partitives and not the partitive du/des-NPs discussed in Section 1.2, namely the complement of fragmentative verbs whose discussion will be delayed until ­Section 3.3. If the quantificational feature is encoded in a QP layer as assumed here, we expect Q-du/des-NPs to be blocked in islands on a par with every-NPs and to interact with other quantifiers. The example provided in Chapter 2 Section 3.4.2 to show that Q-unNPs are blocked in islands is adapted below to show that Q-du/des-NPs are restricted in the same way: (69) [At a hospital 12 new patients are waiting to see a doctor (i.e., a doctor or other/any doctor). Several patients faint and somebody says to a doctor walking by]: ‘Un médecin devrait s’ inquiéter si des patients a doctor should se to.worry if of.the patients s’ évanouissent!’ se faint

Although the quantificational reading of des-NPs is more difficult to obtain than for un-NPs, it seems that des patients in this example cannot take scope over un médecin. This means that it is trapped in the island, suggesting that it is like everyNPs, a QP.

Chapter 3.  Du/des-NPs 



Let us now examine how Q-du/des-NPs interact with other quantifiers. Consider the following examples: (70) a. b.

Tous les étudiants ont lu des livres. (∀ > desQ); all the students have read of.the books ‘All the students read some books.’ Des étudiants ont acheté tous (desQ > ∀); of.the students have bought all les journaux. the newspapers ‘Some students bought all the newspapers.’

*(desQ > ∀) *(∀ > desQ)

(71) Un étudiant a acheté des journaux. (Paul) (unSRef > desQ); *(desQ > unSRef) a student has bought of.the newspapers ‘A student (in particular) bought some newspapers.’

In (70a), des livres with a quantificational interpretation cannot take scope over tous les étudiants and in (70b) tous les journaux cannot take scope over the quantificational des étudiants. In (71), the quantificational des journaux cannot take scope over un étudiant interpreted S-referentially. In other words, the S-referential un-NP must take ‘widest’ scope in the sense that it is about some student in particular who bought all the newspapers. Put differently the QP tous les journaux cannot move past the S-referential un-NP at LF. The above observations are accounted for if intervention effects occur with elements of the same type (72a,b) and when the intervener is interpretatively heavier than the moving element (72c) (Starke 2001). The by-now familiar representations are repeated below: (72) a. * αi… αj… αi b. * αβi… αβj… αβi c. * α … αβ … α

(Starke 2001: 8(16a, 17b, 15a))

A Q-du/des-NP can however move across a property-denoting argument, which is interpretatively less heavy. An example is provided in (73) and the formalisation repeated in (74). (73) [PropP Un violoncelliste] a joué [QP des suites] aujourd’hui. a cellist has played of.the suites today ‘A cellist played (some of the) suites today.’ (74) αβ … α … αβ

(Starke 2001: 8(16b))

The next section deals with the third type of des-NPs advocated, i.e., S-referential des-NPs.

 The Layered DP: Form and meaning of French indefinites

3.1.3  S-referential des-NPs What characterises S-referential des-NPs is the fact that they typically encode speaker’s reference (Section 2.1). I thus assume that their left periphery comprises three layers, PropP, QP and SRefP, on a par with SR-un-NPs. The structure is illustrated below: (75)

SRefP Spec

SRef′ SRef desi

QP Spec

Q′ Q ti

PropP Prop′

Spec Prop ti

NumP ti … filles

In (75), des ‘of.the’ head moves from the nominal inflectional domain through Prop and Q to SRef. PropP is projected to turn NumP into an argument, QP is responsible for the scope properties of SR-des-NPs and SRefP encodes S-reference. If the QP in the structure of SR-des-NPs triggers Quantifier Raising (QR), it predicts that SR-des-NPs may target two different positions at LF on a par with SR-un-NPs. In other words, they should not only be able to have widest scope (Section 2.1) but also intermediate scope (Chapter 2 Section 3.4.3). This is borne out as the following example shows. (76a) is the French version of (131b) in Chapter 2, modified to host a des-NP. (76b) represents the intermediate reading. (76) a. Tous les étudiants ont raconté plusieurs histoires all the students have told several stories qui impliquaient des membres de la famille royale. that involved of.the members of the family royal ‘All the students told several stories that involved some members of the Royal Family.’ b. tous les étudiants > des membres de la famille royale > plusieurs histoires



Chapter 3.  Du/des-NPs 

Although the judgements are not as clear as for un-NPs, it seems that the intermediate reading is possible. For example, in a world with three students {1,2,3} and three members of the Royal Family {Lady Diana, Prince Charles and Camilla Parker-Bowles}, student 1 could tell stories about Lady Diana and Prince Charles, student 2 about Lady Diana, Prince Charles and Camilla Parker-Bowles and student 3 about Prince Charles and Camilla Parker-Bowles. In other words, the members of the Royal Family vary with the students. Of course, the above illustration is not an exhaustive list of the ­potential combinations of students and members of the Royal Family. As the scope relation between tous les étudiants and des membres de la famille royale in (76) represents the surface order, the universal does not move past the S-referential des-NP with intermediate scope at LF. In contrast, the SR-des-NP can cross the universal to take widest scope, showing that it is interpretatively heavier. In the terminology adopted here, it means that its left periphery contains a SRefP layer above QP. The possibility for SR-des-NPs to take intermediate scope strongly supports the analysis developed in this book. It shows that SR-des-NPs not only encode some kind of reference (co-variation means that the co-varying elements can be identified) but also that they undergo QR on a par with other quantificational expressions. The existentiality of SR-des-NPs, however, is not intrinsic to their structure. It is a by-product of movement at LF. Recall from Chapter 2 Section 3.4.3 that I assume that, at LF, quantificational expressions target different projections which are hierarchically ordered and that the landing sites of S-referential noun phrases contain an existential operator (Beghelli 1995; Beghelli & Stowell 1997). In a very basic hierarchy like (77) (Chapter 2 Section 3.4.3), it means that SR-des-NPs move to FP1 when they take widest scope and to FP3 when they co-vary with a universal, sitting in FP2. Both FP1 and FP3 host an existential operator. (77) FP1 > FP2 > FP3 > FP4

What triggers movement of SR-des-NPs to FP1 or FP3 is their dependent interpretation. When they need some extra linguistic information to be interpreted, i.e., when they are [+ deictic], they target the highest projection, whereas when they depend on a universal quantifier, they move to FP3. A hierarchy like (77) allows us to distinguish SR-des-NPs with ‘narrow scope’ from Q-du/des-NPs with ‘narrow scope’. Consider the following example: (78) Tous les étudiants ont parlé à des professeurs. all the students have talked to of.the professors

When tous les étudiants has scope over des professeurs in (78), the sentence is ambiguous. This is because the professors may co-vary with the students (as the members of the Royal Family with intermediate scope in (76)) or have a covert partitive reading. Put differently des professeurs may be a SRefP or a QP. Two positions are necessary to host

 The Layered DP: Form and meaning of French indefinites

these constituents because only the former can escape islands as seen in (76) vs. (69) in the preceding section. The SRefP co-varying with the universal quantifier moves to FP3 and the Q-du/des-NP to FP4. In the next section, I turn to the structure of the inflectional domain. As du- and des-NPs occur with mass and plural nouns respectively, this section will inevitably touch upon these notions and the way they are encoded in the structure. It will also address the question of number, as du-NPs are grammatically singular whereas des-NPs are plural. As du (de + le) and des (de + les) comprise two distinct parts (Kupferman 1979), the structure of du/des-NPs is bound to be complex in order to host these two components.

3.2  The inflectional domain 3.2.1  The interpretation To establish the role and the position of de ‘of ’ and le/les ‘the’ in du/des-NPs, let us consider the features which characterise the interpretation of these nominals. The discussion in this section and in the next one will concentrate on du/des-NPs to the exclusion of ‘explicit partitives’ as defined in Section 2.2 which are addressed in Section 3.3. It will include un-NPs ‘a-NPs’ and le/les-NPs ‘the-NPs’ to ease comparison. Both the count and the mass readings of le-NPs are considered. As mentioned in Section 1.2, du/des-NPs are also interpreted differently depending on whether they imply a quantity or not. When they are quantitative they correspond to English some-NPs whereas when they are not they correspond to bare nouns. Most of the discussion concerning the inflectional domain is inspired by Borer’s (2005a) work (recall Chapter 1 Section 4.4.2.1). However, the analysis of French proposed here, at least of the small sample of noun phrases examined, will lead me to diverge from some of the conclusions she reaches. Let us begin with the mass/count distinction for which I will use the feature [± count]. Following Borer (2005a) I assume that listemes are not specified for the mass/count distinction in the sense that they are all mass and become count in specific contexts, namely in the presence of some functional elements which include the plural morpheme, the indefinite article, and the count le/les ‘the’ (the next section deals with their structure). However, I depart from Borer’s analysis in that I adopt the standard position on plurality, namely that it is a function from individuals. Informally, it means that ‘count’ is related to individuated elements, i.e., atoms, and not to a division of mass. One reason is that in French it seems that every time a noun phrase is plural there are atoms in its denotation. Intuitively, this is what ‘count noun’ means, that is, something which can be counted (0, 1, 2, 3…), in contrast with something which is measured for example. In order to count, you need atoms. One argument for Borer to reject the view that the plural –s leads to an ‘atom’ reading comes from the interpretation of bare plurals in English (79) and examples with zero or fractions (80).

Chapter 3.  Du/des-NPs 



(79) a. Kim ate apples this afternoon (for an hour) (*in an hour). b. Pat built houses (all summer) (*in two months). c. My kid sister drew circles (all morning) (*in half an hour).  (Borer 2005a :120 (41)) (80) a. zero boys / *boy b. 0,2 apples / *apple

(ibid, 115, (31a,b), (33c))

In English, examples like (79) are consistent with an interpretation whereby no single complete apple has been eaten (but, say, a number of chunks from assorted apples, perhaps no more than one), no whole house was built (by Pat or anybody else), and my kid sister’s drawing gave rise to assorted arcs in incomplete circles, none of which actually qualifies as a circle (ibid, 120–1).

Such a reading is however clearly rejected by native speakers for the French counterparts of (79): (81) a. b. c.

Kim a mangé des pommes cet après-midi. Kim has eaten of.the apples this afternoon Pat a construit des maisons tout l’ été. Pat has built of.the houses all the summer Les enfants de ma sœur ont dessiné des cercles the children of my sister have drawn of.the cercles tout le matin. all the morning

My informants unanimously consider that (81) is about whole apples, whole houses and whole circles, which shows that French and English differ.27 What about the examples in (80)? Although I have no detailed analysis to offer for such examples, a few observations are in order. It seems to me that zero is a numeral on a par with 1, 2, 3… If this is the case, it should be part of the counting process and concern individuals. Intuitively, in sentences like zero boys were playing at the playground we are not thinking of parts of boys. Even if the set is empty at utterance time, it seems that the set concerns individuals, suggesting that the plural –s is used as ‘individuator’. Consider the following examples which support this conclusion (G. Puskás, p.c.): (82) a.

Malheureusement, j’ai trouvé zéro cheval au magasin. unfortunately I have found zero horse at.the shop

.  Note that the French examples in (81) are problematic for an account where the ‘atom’ reading of plurals is due to a counter which assigns range to 〈 e 〉 # as proposed by Borer (2005a:122). This is because the des-NPs which correspond to bare plurals denote individuals although they do not necessarily encode a quantity (Section 1.2).

 The Layered DP: Form and meaning of French indefinites

b. Malheureusement, j’ai trouvé zéro chevaux au magasin. unfortunatelly I have found zero horses at.the shop ‘Unfortunately, I found zero horses at the shop.’ (83) a. b.

Quand je suis rentrée à la maison, il y avait when I am come.home at the house there Y had du cheval sur la table. of.the horse on the table ‘When I came home, there was horse meat on the table.’ Quand je suis rentrée à la maison, il y avait when I am come.home to the house there Y had des chevaux sur la table. of.the horses on the table ‘When I came home, there were horses on the table.’

As in French the plural –s is usually not pronounced, it is more difficult than in English to determine whether the noun is plural or not in examples like (80). This is why (82) and (83) are about horses, as cheval has an irregular plural in French, namely chevaux, and not chevals, which is ungrammatical. In these examples the difference between a mass and a count reading involving individuals is clear. (82a) and (83a) concern horse meat whereas (82b) and (83b) concern individual horses. (82b), for example, could be about a shop which sells animals, which could be alive or toys, but which are individuated. Similarly, in (83b) toy-horses or alive animals could be on the table, although the latter reading is improbable in our world. Such examples suggest that the plural morpheme in French leads to an individuated reading. As for the fraction reading illustrated in (80b), my informants found it more ­difficult to process, probably because it is not an everyday usage. One problem is to imagine 0,2 apples without there being a whole apple first. In other words, although the portion of apple in such examples is smaller than an individuated apple, it seems that it is a fraction of an apple which once existed. If this is correct, it suggests that the notion of atom cannot be dispensed with, even if at utterance time there is no individual apple. And if the plural –s means that atoms are involved, the presence of this morpheme in these examples is not surprising. Such contexts should however be studied in more details. The presence of atoms in plurals can be illustrated with a join semilattice. A join semilattice is a figure which represents sets and which has been used in the literature to represent the extension of mass and plural nouns.28 An example of a join semilattice

.  This set is ordered by the part of-relation, which is a transitive, reflexive and asymmetric relation. For further details, the reader is referred to the relevant literature.



Chapter 3.  Du/des-NPs 

is provided in (84). As du-NPs are mass and des-NPs plural, I assume that such a join semilattice represents their extension. (84) (based on Link 1983) iii. {a, b, c, d} ii. {a,b} {a,c} {a,d} {a,b,c} {a,b,d} {a,c,d} {b,c} {b,d} {b,c,d} {c,d} i. a b c d

The elements a, b, c and d on the bottom line (i) of (84) represent the atoms, i.e., the minimal elements of a set. (84) represents a set with four atoms, which could for ­example be books, socks or bananas. Line (ii) represents pluralities, i.e., sets which contain more than one element, and line (iii) the set of all the books, socks or bananas that exist. What differentiates plural from mass nouns is the presence vs. absence of line (i) in the figure. As mass nouns do not have minimal parts, their extension corresponds to a semilattice without the bottom line (Landman 1989, 1991), contrary to the one of plural noun phrases (recall Section 2.1). Applied to du-NPs and des-NPs, this means that the former do not have line (i) in their extension, contrary to the latter (De Swart 2006 for an account of des-NPs). If this is correct, the fact that des-NPs, contrary to du-NPs, can be S-referential is not surprising (Section 2.1). In effect, a referential reading singles out one or more individuals even if this reference is dependent. Singular count nouns in contrast do not form a join semilattice as their extension is a set of singularities. A question which should be addressed in the discussion of the plural –s is how this morpheme relates (i) to grammatical number and (ii) to the quantities expressed by nominal expressions.29 What I call grammatical number is what determines the agreement of a verb with its subject, i.e., whether the verb is morphologically singular or plural, or of an adjective with the noun it modifies, at least in languages with overt number morphology on verbs and adjectives.30 As the notions of plurality, grammatical number and quantities are closely related, they are much debated in the literature (Heycock & Zamparelli 2003, 2005; Roodenburg 2004a; Borer 2005a; Munn & Schmitt 2005; Farkas 2006). As it would be impossible to do justice to all these works in this short section, I will merely explore one way of combining plurality, grammatical

.  The term quantity here is used in an informal way. For a technical account relating quantity and telicity see Borer 2005a,b where the author carefully reviews Krifka’s (1998) proposals and Kiparsky’s (1996, 1998) slightly different approaches. Her work also allows her to relate the bounded/telic reading to the structure of nominal expressions. This is far beyond the scope of this section (see Chapter 5 Section 2.1). The aim of the discussion here is to determine whether or not a projection related to Quantity is present in the structure of du/des-NPs. .  See Farkas & De Swart (2003) who assume that morphological plurals are semantically plural. Spector (2003) in contrast adopts a different position. He argues that des-NPs are number neutral.

 The Layered DP: Form and meaning of French indefinites

­ umber and quantities for the French data examined here. What this implies for the n structure of these noun phrases is addressed in the next section. The discussion so far has corroborated the idea that in French the plural –s leads to individuated readings. As all noun phrases with this morpheme are grammatically plural, including the ones with zero and fractions (recall (80)), we could assume that grammatical number can be associated with a feature [± plural], where the plural reading obtains when the plural morpheme is present and the singular reading in the absence of such a morpheme (as a kind of default reading). However, it seems to me that such a classification could be refined in that it would imply that singulars comprise two classes of noun phrases, namely mass and count singular expressions. In other words, although the [+ plural] and the [+ count] features somehow conflate, the [– plural] and the [– count] features do not as [– plural] noun phrases may be mass (du pain/lemass pain) or count (un pain/lecount pain). This difference leads me to dissociate grammatical number and count reading and to assume that these features are encoded in different projections (see the next section). The other issue related to plurality and grammatical number mentioned above is the question of the quantities (or absence thereof) associated with a noun phrase. This question is delicate in that it is very difficult to determine whether these quantities are due to encyclopaedic or pragmatic factors or whether they are encoded in the nominal expressions themselves. Furthermore, in some cases the quantity involved is straightforward whereas in other cases it is not. For instance, un-NPs and lecount-NPs clearly imply 1 whereas other examples such as les-NPs are not so clear cut (see Borer 2005a for a very detailed and cross linguistic account which is not restricted to theand a-NPs). Although we know that the latter constructions mean more than one, it is difficult to determine whether this is due to their plurality or to some quantity which is not overtly expressed/morphologically realised. The view I will explore here is that quantity is only encoded in a nominal expression when it cannot be inferred from (a combination of) other features. What this means is that un-NPs and lecount-NPs imply the cardinality 1 because they are count and singular and that les-NPs mean more than one because they are plural. Du-NPs and lemass-NPs do not mean 1 because they are not count, although they are singular. The question which arises is whether this applies to des-NPs. As seen in Section 1.2, these noun phrases may be interpreted as quantitative or as non-quantitative, that is, like someplur-NPs or bare nouns respectively. However, as both readings are plural, they should always mean more than one. This is exactly what I suggest: des-NPs imply more than one entity independently of their quantitative/non-quantitative reading but when they are quantitative, the quantity is somehow more salient and precise. Recall the ­following examples from Section 1.2: (85) J’ai acheté des livres mais pas beaucoup. I have bought of.the books but not many ‘I bought (some) books but not many.’

Chapter 3.  Du/des-NPs 



(86) Speaker A: Vous vendez des cigarettes? you sell of.the cigarettes ‘Do you sell cigarettes?’ Speaker B: Non, ici on vend des livres et no here one sells of.the books and des journaux. of.the newspapers ‘No, we sell books and newspapers.’

In (85) the des-NP is clearly quantitative as the quantity can be contrasted whereas in (86) we are rather dealing with kind/concepts. Despite this difference, it seems to me that both cases are concerned with more than one entity. Although this is obvious in (85), it is less in (86). However, as the job of the sales person is to sell more than one book and more than one newspaper (even if it might happen that his/her shop doesn’t have any book or newspaper in stock anymore), it is possible to associate the more than one reading to the plural forms. As this discussion is rather informal, I leave it for further research. The different properties discussed above are summarised in Table 11 (87). Although I am primarily interested in du/des-NPs (and un-NPs which are discussed in Chapter 2), le/les-NPs are included in the table. (87) Table 11.

A. B. Grammat number Quantity

C. Atoms

D. Count

Not Examples Inferred inferred 1. 2. 3. 4. 5. 6. 7. 8.

un pain a bread le pain count the bread le pain mass the bread du pain quantit. of.the bread du pain ‘bare’ of.the bread des pains quantit. of.the breads des pains ‘bare’ of.the breads les pains the breads

NP

sing.

√ 1

X

√

√ un

pain

sing.

√ 1

X

√

√ le

pain

sing.

X

X

X

X

pain

sing.

X

√ some.sg

X

X

pain

sing.

X

X

X

X

pain

plur.

√ > 1

√ some.plur.

√

√ -s

pain

plur.

√ > 1

X

√

√ -s

pain

plur.

√ > 1

X

√

√ -s

pain

 The Layered DP: Form and meaning of French indefinites

(87) shows that although plural noun phrases are count, singular ones may be mass or count. Columns C and D report the correlation between plural –s and atoms. An individuated reading also arises with the indefinite article and the count le. Finally, only quantitative du/des-NPs encode some quantity which cannot be inferred. Another interpretative property often attributed to du/des-NPs has not been taken into consideration in (87), mainly because it does not seem to be an intrinsic feature of du/des-NPs (see below). It is their unbounded reading from an aspectual point of view (Kupferman 1979; Bosveld-de Smet 1998; Roodenburg 2004a) (see Chapter 5 Section 2.1). This property distinguishes du/des-NPs from un-NPs as (88) shows: (88) a. b.

J’ai mangé une bonne viande. I have eaten a good meat ‘I ate a good piece of meat.’ J’ai mangé de la bonne viande. I have eaten of the good meat ‘I ate good meat.’

Although de la bonne viande in (88b) is indefinite on a par with une bonne viande in (88a), the difference of interpretation between the two sentences is clear. In (88a), viande ‘meat’ is used as a count noun whereas in (88b), it is a mass noun. Another difference is that (88a) is telic whereas (88b) is not. This is illustrated by the restricted use of aspectual modifiers of the type in x hours, for x hours. The use of in- and for- adverbials in (88) leads to different grammaticality judgements as (89) shows. (89) a. J’ai mangé une bonne viande en 10 minutes. I have eaten a good meat in 10 minutes b. ??J’ai mangé de la bonne viande en 10 minutes. I have eaten of the good meat in 10 minutes c. ??J’ai mangé une bonne viande pendant 10 minutes. I have eaten a good meat for 10 minutes d. J’ai mangé de la bonne viande pendant 10 minutes. I have eaten of the good meat for 10 minutes

The sentence in which the du/des-NP co-occurs with a for-adverbial (89d) is better than the one in which it co-occurs with an in-adverbial (89b). Similarly, the sentence in which a un-NP co-occurs with an in-adverbial (89a) is better than the one in which it co-occurs with a for-adverbial (89c). However, the presence of a du/des-NP does not necessarily lead to a non-telic reading as the following examples show: (90) a.

J’ai reconnu des filles dans la rue. I have recognised of.the girls in the street ‘I recognised some girls on the street.’

Chapter 3.  Du/des-NPs 



b. Jean a volé des voitures.31 Jean has stolen of.the cars ‘Jean stole some cars.’

In (90), the sentences are telic. This is supported by the ungrammaticality of for­adverbials in (91): (91) a. * J ai reconnu des filles dans la rue I have recognised of.the girls in the street pendant 10 minutes. for 10 minutes b. * Jean a volé des voitures pendant 10 minutes.32 John has stolen of.the cars for 10 minutes

To determine whether a sentence is telic or not, the kind of predicate it involves has to be taken into consideration (see Borer 2005b).33 As is well-known, achievements (reconnaître ‘recognise’) and accomplishments (voler ‘steal’), are bounded and therefore do not admit for-adverbials (Vendler 1957; Doetjes 1997). As du/des-NPs are not intrinsically unbounded, a feature like [± bounded] is not considered in the discussion here,

.  Thanks to Genoveva Puskás who provided this example (p.c.). .  This example is ungrammatical when there is a single event of stealing. However, an iterative reading would be fine. .  The aspectual properties of the clause do not depend on the (in)definiteness of the DPs it contains nor on its grammatical number (Verkuyl 1993: 71ff.). Consider Verkuyl’s examples (French version mine):

(i) a. Judith ate those three (Judith a mangé ces trois sandwiches. sandwiches.) b. Judith ate three sandwiches. (Judith a mangé trois sandwiches.) c. Judith ate sandwiches. (Judith a mangé des sandwiches.) d. Judith ate that sandwich. (Judith a mangé ce sandwich.) e. Judith ate a slice of bread. (Judith a mangé une tranche de pain.) (i)a,b and (i)d,e are terminative in Verkuyl’s terminology (in opposition to durative). In other words the event is bounded, i.e., completed. Yet, those three sandwiches in (i)a and that sandwich in (i)d are definite whereas three sandwiches in (i)b and a slice of bread in (i)e are indefinite. The difference between singular and plural leads to similar observations: number does not have any influence on the aspectual properties of the event. Although the direct objects in (i)a and (i)c are plural, the former is terminative and the latter durative. (i)a and (i)d on the contrary share the same aspectual properties, even if the direct object of the former is plural and the one of the latter singular. The same analysis applies to the French examples in parenthesis. Note furthermore that the difference mass/count cannot be at the origin of the distinction bound/unbound as desNPs, which are count, are fine in telic and atelic sentences as seen in the text.

 The Layered DP: Form and meaning of French indefinites

and therefore not mentioned in (87). The impact of the features listed in this table on the structure of the nominal expressions concerned is discussed in the next section.

3.2.2  The structure Although my first aim here is to determine the structure of du/des-NPs, this section will also deal with un-NPs (Chapter 2) and with le/les-NPs, to allow comparison. Assuming with Borer (2005a) that all listemes are mass and that it is the structure in which they occur which is responsible for their count reading, I will associate the [+ count] feature to a projection. In other words, this feature is encoded in a projection which is only present in the structure of count nouns. As my account overlaps with Borer’s work, I will not adopt her label to avoid confusion and simply refer to this projection as FP1, as it is the first functional projection above the lexical layer NP. This projection hosts the plural –s the count le ‘the.sg’ and the indefinite article as proposed by Borer. However, I assume that its presence leads to an atomic reading (see the preceding section). Another projection is postulated to differentiate ‘quantitative’ from ‘non-quantitative’ du/des-NPs (Section 1.2). To simplify the discussion I will assume that this projection does not only host quantities like beaucoup ‘much/many’ or peu ‘little’ but also numerals. I will refer to it as FP2. Finally, I propose that the structure of nominal expressions hosts a projection encoding grammatical number, FP3. This projection determines the agreement on verbs and adjectives, at least in languages with overt morphology on these elements. Before turning to the structure of du/des-NPs, which comprises an additional element de ‘of ’, let us look at the structures of un-NPs and le/les-NPs resulting from the above observations. X means that the layer is not projected. (92) Example a. un pain b. le painmass c. le paincount d. les pains

Number FP3 unsing lesing lesing lesplur

Quantity FP2 X X X X

Count FP1 un X le -s

Lexical category NP pain pain pain pain

As mentioned above, un, le and –s are generated in FP1 to account for the count reading of un pain, lecount pain and les pains. This means that un and le move to (are remerged in) higher positions to account for their additional features and that pain moves to FP1 in (92d) to bind the plural morpheme as proposed by Borer (2005a). None of the nominal expressions in (92) have an FP2 in their structure as the quantities associated to them can be inferred: un pain and lecount pain imply the quantity 1 because they are singular and count and les pains imply more than one because they are plural. Appling the same reasoning to the structure of du/des-NPs we obtain (93). (93) only reports the similarities between the nominal expressions in (92) and du/des-NPs to ease comparison and makes abstraction of the element de ‘of ’.

Chapter 3.  Du/des-NPs 



(93) Example a. du painquantit. b. du painnon-quantit. c. des painsquantit. d. des painsnon-quantit.

Number FP3 lesing lesing lesplur lesplur

Quantity FP2 √ X √ X

Count FP1 X X -s -s

Lexical category NP pain pain pain pain

The projection of FP1 in the structure of des-NPs accounts for their count reading and its absence in the structure of du-NPs for their mass reading. To distinguish between du/des-NPs which are quantitative and the ones which are not, I propose that the former comprise a FP2 in their structure contrary to the latter. One question which arises is whether FP2 is empty or not and another one where de is generated. Another issue which should be addressed is where the incorporation of le/les ‘the’ takes place. To help us clarify this matter let us look at nominals involving overt quantities like beaucoup de ‘many/much’, peu de ‘little’… (setting aside their grammatical number which is not relevant here): (94) Example a. beaucoup de pain b. peu de pain c. beaucoup de livres d. peu de livres

Quantity FP2 beaucoup peu beaucoup peu

Count FP1 X X –s –s

Lexical category NP pain pain pain pain

A typical characteristic of beaucoup and peu is that they can occur with mass and plural count nouns. As they represent quantities, I assume that they sit in FP2 and, when they are associated with a plural count noun, that their structure contains a FP1 hosting the plural -s. If this is correct it means that de is generated between beaucoup/peu and –s, i.e., in a projection between FP2 and FP1. If quantitative du/des-NPs are similar to beaucoup/ peu in that they involve a quantity, the de element should be the same in both quantitative expressions. In other words, de in du/des-NPs should also be generated in a projection between FP2 and FP1. As a result, de could move to FP2 and further to FP3 where it merges with le/les.34 This is illustrated in (95): (95) Example a. du painquantit. b. du painnon-quantit. c. des painsquantit. d. des painsnon-quantit.

Number FP3 de + lesing de + lesing de + lesplur de + lesplur

Quant. FP2 deP de de X de de de X de

Count FP1 X X − s − s

Lexical category NP pain pain pain pain

.  That de should be generated below le/les to form the complex de + le, de + les by derivation has been suggested to me by G. Giusti (p.c.).

 The Layered DP: Form and meaning of French indefinites

As shown in (95), FP1 is the lowest functional projection in the structure of des-NPs, whereas in the structure of du-NPs it is the projection hosting de. The le/les component of du/des is generated in FP3, as in the nominal expressions in (92) (except for the count le which is generated in FP1). As de moves to FP3 (through FP2 when this layer is projected) the fusion of de and le/les takes place in FP3. In argumental nominals, the complex du/des element moves further to the left periphery. In predicative du/desNPs, in contrast, it remains in FP3. Put differently, predicative du/des-NPs are FP3s. A question which arises is whether the de element in quantitative du/des-NPs and beaucoup/peu de is similar to the one in de-NPs discussed in Chapter 2 Sections 2.3.3 and 3.2.35 Recall that the restricted distribution of these noun phrases suggests that they contain an empty category which has to be bound by a negative operator. One possibility worth exploring is that this empty category is an empty quantity. The structure of de-NPs would therefore be parallel to the one of beaucoup/peu de:36 (96) Example a. beaucoup de pain b. peu de pain c. [pas] de pain (97) Example a. beaucoup de livres b. peu de livres c. [pas] de livres

Quantity FP2 deP beaucoup de peu de ec de Quantity FP2 deP beaucoup de peu de ec de

Count FP1 X X X

Lexical category NP pain pain pain

Count FP1 − s − s − s

Lexical category NP livre livre livre

If the empty category in (96) and (97) is parallel to beaucoup/peu, it suggests that it is in the specifier position of FP2 (on a par with the trace of wh-elements for example which also have to be bound by an operator). De can thus move through F2 to Prop to form an argument.

.  See Doetjes & Rooryck (2003) for a comparative analysis between quantitative and qualitative constructions involving de. .  One difference however is that beaucoup de cheval ‘much horse’ is about horse meat (mass) whereas pas … de cheval ‘not … a horse’ seems to be about a kind/concept. For the time being, I have nothing to say about this difference. The view that de-NPs in negative contexts and beaucoup/peu de ‘many-much/few-little of ’ have a similar structure differs from the one in Kayne (1975, 1981) in the sense that only the former contain an empty category. According to Kayne (1975, 1981), de-NPs in general should be analysed as (NP zero element-de-articleless NP), where the zero element is an empty QP which needs an antecedent to be properly governed. This element could be bound by beaucoup ‘many/much’, peu ‘few/little’, trop ‘too many/much’ or by a negative operator.

Chapter 3.  Du/des-NPs 



That the quantity element and de occupy different positions is supported by en pronominalisation (G. Giusti, p.c.). Consider the following examples: (98) a. b.

J’ai lu beaucoup/peu de livres. I have read many/few of books ‘I have read many/few books.’ J’en ai lu beaucoup/peu. I en have read many/few

(99) a. b.

J’ai mangé beaucoup/peu de pain. I have eaten much/little of bread ‘I have eaten a lot of/little bread.’ J’en ai mangé beaucoup/peu. I en have eaten much/little

In (98) and (99), en pronominalises de livres and de pain respectively, i.e., deP. In other words, these elements form subtrees and de cannot be the head FP2. Although a thorough discussion of en-constructions is out of the scope of this book, let us see if the above suggestion extends to du/des-NPs. Consider the following example: (100) a. b.

Jean a des livres. Jean has of.the books ‘Jean has books.’ Il en a. / *Il en a des. he en has / he en has of.the

If en in the above example replaced deP, we would expect des to be stranded, contrary to fact.37 This is because in du/des-NPs, du/des sits in a position higher than deP. If, however, en may replace different portions of a structure, (100) is not problematic anymore. Let us consider the idea that there are several en pronouns, which is by no means new.38 One of them replaces the structure below FP2 in nominal expres.  Note that the stranding of des would presuppose that the en pronominalisation takes place after movement of de to FP3 (through FP2 if the du/des-NP is quantitative). If it took place before movement of de, the stranded element would be les. .  For example, there is a genitive en (engenitive) which pronominalises a PP (see the discussion of se souvenir de ‘remember’ in Section 3.3): (i) a. b.

Jean se souvient de tous ses étudiants. Jean REFL remembers of all his students ‘Jean remembers all of his students.’ Jean s’ en souvient. Jean REFL en remembers

In (i)a, the verb se souvenir ‘remember’ takes a PP complement. In (i)b, this complement has been pronominalised by en.

 The Layered DP: Form and meaning of French indefinites

sions with overt quantities (or numerals). Informally, en pronominalises the part of the structure which identifies the quantity which is stranded. This is why I will label it enidentification pronoun. This is the case of (98) and (99) above. In (100), I propose that en replaces the whole nominal expression. If this is correct, the question which arises is whether en can replace any type of du/des-NP. The examples below illustrate desNPs which can be referential, quantificational and property-denoting (Section 2). (101) a. Speaker A : Des enfants jouent dans la cour. of.the kids are.playing in the yard Il s’agit de Pierre, Paul et Jacques. it SE is.about of Pierre, Paul and jacques ‘Some children are playing in the yard, namely Pierre, Paul and Jacques’. b. Speaker B : *Oui, j’en ai vus. yes I en have seen Oui, je les ai vus. yes I them have seen ‘Yes, I’ve seen them.’ (102) a. Speaker A : Des filles étaient blondes, toutes of.the girls were blond all les autres avaient les cheveux foncés. the others had the hair dark b. Speaker B : *Oui, j’ en ai vues. yes I en have seen Oui, je les ai vues. yes I them have seen ‘Yes, I’ve seen them.’ (103) a. b.

Marie cherche des souris. Marie is.looking.for of.the mice ‘Marie is looking for mice.’ Marie en cherche. Marie en is.looking.for

As the ungrammaticality of en in (101b) and (102b) suggests, this pronoun cannot replace des enfants and des filles which are S-referential and quantificational respectively. (103b) in contrast is fine, suggesting that en can pronominalise property-denoting noun phrases. Let us call this pronoun enproperty. This conclusion however has to be refined as P-un-NPs cannot be replaced by en: (104) a. Marie cherche une souris. Marie is.looking.for a mouse ‘Marie is looling for a mouse.’ b. *Marie en cherche. / Marie en cherche une. Marie en is.looking.for / Marie en is.looking.for one

Chapter 3.  Du/des-NPs 



(104b) shows that en cannot replace une souris but that une can be stranded. In the latter case however, une has to be a numeral. That the stranded une is a numeral and not an indefinite article is clear from Italian and English (thanks to G. Giusti for pointing out these facts). In Italian, only the cardinal uno can occur in such contexts (in contrast with the article un) (Cardinaletti & Giusti 2002). Similarly, in English only one is possible. This means that en in the second part of (104b) pronominalises the portion of structure below FP2. In other word, this case can be assimilated to the ones in (98) and (99). The ungrammaticality of the first part of (104b) shows that en cannot replace une souris, even if it is a P-un-NP. Although what distinguishes P-des-NPs from P-un-NPs is not quite clear to me, I hypothesise that it is due to the fact that the latter are singular and imply the quantity 1. That this seems to be on the right track is supported by (105): (105) a. b.

Marie cherche du sel. Marie is.looking.for of.the salt ‘Marie is looking for salt.’ Elle en cherche. she en is.looking.for.

As (105) shows, the property-denoting du-NP can be replaced by en. As du-NPs are mass and therefore cannot imply the quantity 1, it supports the fact that it is the 1 implied by une souris in the first part of (104b) which leads to ungrammaticality. Yet another kind of en pronoun, found in explicit partitives, will be mentioned in the next section (fn. 48). In (92) above we have seen that nominals with the indefinite article comprise a FP1 and a FP3 in their structure (the relevant part of this example is repeated below). The question is how to differentiate between the indefinite article and the numeral 1. A first distinction, I propose, is that nominal expressions with the latter, contrary to the former, contain a FP2 in their structure. Recall that I assume that this projection hosts quantities but also numerals. (106) Example a. unart pain b. unnumeral pain

Number FP3 unsing unsing

Quantity FP2 X un

Count FP1 un un

Lexical category NP pain pain

As proposed by Borer (2005a), I assume that when un is a numeral it differs from other numerals in that it is responsible for the count reading of the nominal containing it (in contrast with other numerals which occur with plural nouns). This is why it is generated in FP1, on a par with the indefinite article. I also assume that the numeral 1, on a par with other numerals (Chapter 1 Section 4.4.2.1), is generated in a specifier position. Put differently, 1 is an XP in Spec,FP1. The above discussion shows that 1 and the indefinite article differ in that the former, contrary to the latter, is generated as an XP on a par with other numerals and in that the noun phrases containing it comprise a FP2 in their structure.

 The Layered DP: Form and meaning of French indefinites

To conclude this section, let us briefly come back to the deP postulated in the structure of du/des-NPs and to the role of de as nothing has been said so far on this issue. Although I have no clear answer as to what the role of de is, I would like to mention two points. First, for some scholars the role of de was to take up the function of the genitive case in Indo-European and genitive in these languages was used to denote a category, in opposition to the object itself. This is interesting as du/des-NPs may be interpreted as kinds/concepts in some contexts, as observed several times in this chapter (see Section 1.2 for example). If the interpretation of nominal expressions is a construction as assumed here, we could hypothesise that there are additional levels in the sense that a piece of structure could be responsible for a category/kind reading (recall Zamparelli 2000). How this would interact with the count reading would be particularly interesting to investigate. Second, for other scholars, de appeared in French because of the weakening of the demonstrative force characteristic of the definite article (on this topic see Giusti 2001). In the framework developed here, this could mean that le/les in du/des differs from le/les in definite noun phrases in that it cannot move higher in the structure to form an argument, especially a referential expression. In other words, le/les in du/des is deficient and is only present in the structure to account for number.39 One role of de would therefore be to move to the left periphery to form an argument.

3.3  More on partitives In Section 2.2, I have made a distinction between implicit and explicit partitive du/ des-NPs. The former express a subset overtly whereas the latter, which concern the complements of fragmentative verbs (Section 1.2), realise the set. Implicit du/des-NPs have been analysed as QPs, i.e., with the typical covert partitive meaning characteristic of quantificational expressions. As for the category of explicit du/des-NPs, I will take them to be property noun phrases, i.e., PropPs. The reason is that, as observed in Section 1.2, their interpretation is undetermined in that it lacks identificational force in Bosveld-de Smet’s (1998) terms, a characteristic of property-denoting arguments discussed in the next section. .  According to Milner (1978: 25ff.) le/les in quantitative du/des-NPs is a kind of generic determiner similar to the one found in les chiens mordent ‘the dogs bite’, corresponding to the English bare plural dogs bite. Gross (1967) also considers this determiner as generic. According to Kupferman (1979, 1994), le/les in his quantitative du/des-NPs is a non-specific determiner, in contrast with le/les in partitive du/des-NPs which is specific. According to Roodenburg (2004a), le/les in (non partitive) du/des is underspecified for number. This means that, intrinsically, it is neither [– plural] nor [+ plural] but it is compatible with both settings. Contreras (1986) and Casalegno (1987) show that mass or plural definites are used in Romance languages where bare singular mass or bare plural count nominals with a ‘kind’ or ‘generic’ reading are used in English.

Chapter 3.  Du/des-NPs 



Concerning the nature of the de component in implicit partitive du/des-NPs, I have proposed in the preceding section (which concerns all du/des-NPs to the exception of explicit partitives) that it is a functional head generated in a projection between the layer related to quantities/numerals and the one responsible for a count reading. The question which arises is whether de in explicit partitives can be treated on a par with the one in implicit partitives. Intuitively, it seems that the former involve a ­prepositional de ‘of ’ on a par with bare partitives in English: (107) Jane ate of/from the cake.

The involvement of a preposition in some partitive du/des-constructions is supported by extraction facts, the use of the relative pronoun dont ‘of.which’ and the locution de quoi ‘of what’ discussed in Section 1.3. The structure of partitive du/des-complements of fragmentative verbs should rather be as in (108), where ∅ represents the undetermined quantity of a cake which is contextually determined. (108) manger [[∅] [PP de [DP le gâteau]]]

In (108), the set is overtly expressed by a PP, hence the label explicit partitives. This PP is contained in the nominal structure, i.e., it does not represent the category of explicit partitives. Before examining the details of such a proposal, some general remarks on the structure of partitive noun phrases are in order, namely on the number of Ns they contain and the position of the PP.40 Let us deal with the first issue. A position widely adopted in the literature is that the structure of partitives is headed by an empty head PRO, i.e., partitives contain two nouns (Jackendoff 1977; Milner 1978), a position adopted here.41 Jackendoff ’s structure is reported below:42 (109)

N′′′ Art′′′

N′′

Art′′

N′

Art′

N

N′′′

Art all

PRO of the men

.  Landmark studies of English partitives are Jackendoff ’s (1977) and Selkirk’s (1977). .  See Martí Girbau 2003 who argues against such a line of research. Kupferman 1999, 2001 and Zribi-Hertz 2003 are among those who do not adopt a structure with an empty element. .  In Jackendoff ’s structure, of does not head a PP because he assumes that of is a case marker inserted at a late stage of the derivation and not a regular preposition. In order to obtain a more regular X-bar structure, Hoeksema suggests that a PP be added to the structure in (109).

 The Layered DP: Form and meaning of French indefinites

An argument in favour of an empty category is that this position is sometimes overtly filled, for example when the first determiner (every, the, and neither in the examples below) requires the presence of a following nominal as in (110): (110) a. every (single) one of my friends b. the only one of them that got away c. neither one of the two women

(Hoeksema 1996: 4)

The idea is that in (110) the pronoun one sits in the N position hosting PRO in (109). In French, this position can also be filled overtly as (111) shows. (111) a. b.

Quatre peintures de celles qui avaient été volées four paintings of those that had been stolen ont été retrouvées. have been found Celui des livres de Zola que j’ ai lu. this.one of.the books of Zola that I have read

(Milner 1978: 84)

Further arguments for an empty element come from (i) the interpretation of partitive noun phrases, (ii) the agreement pattern they give rise to and (iii) en cliticisation in French (see for example Cardinaletti & Giusti 2002: 32ff.).43 In Italian non-overt postverbal subject positions can only have a generic human interpretation, characteristic of arbitrary empty pronouns (cf. Rizzi 1986). The only reading of (112b) is thus that some human beings make the kind of nest in question, which is infelicitous if we except situations in which human beings make nests in their garden to attract birds. (112) a. (Questo tipo di nido), lo fanno molti uccelli migratori. (this kind of nest), CL make many birds migrating b. (Questo tipo di nido), lo fanno molti.  (Cardinaletti & Giusti 2002: 35(119a,b))

The interpretation of (112b) thus suggests that the empty element selected by molti ‘many’ is a pro. Let us turn to agreement. Even when the noun is empty, the numeral agrees with it for number (113a,b) or person (113c). (113) a. b.

Una [e] delle ragazze pensa che … one.fem.sg of.the girls thinks that Uno [e] di noi pensa che … one.masc.sg of us thinks that

.  Cardinaletti & Giusti 2002 argue for an empty DP (in contrast with an empty head). The idea is that the quantifier selects a quantitative DP – which may be overt or not – and a partitive PP. For a short discussion of en pronominalisation see Section 3.2.2.

Chapter 3.  Du/des-NPs 



c.

Molti [e] di noi pensano che … many of us think that

(Cardinaletti & Giusti 2002: 34(115))

In (113a,b) una ‘one.fem.sg’ and uno ‘one.masc.sg’ overtly agree with the empty category, interpreted as ragazza ‘girl’ and ragazzo ‘boy’ respectively and not with the PP which is plural. In (113c), molti ‘many’ is third person contrary to noi ‘us’ in the PP which is second person. Finally, en cliticisation in French represents additional evidence for an empty category in partitive constructions. en is a clitic pronoun which clearly represents a subpart of the constituent in brackets in (114). (114) a. b. c. 

J’ eni J’ eni J’ eni I en

ai pris [un kilo ti de celles-ci]. ai pris [dix ti de celles-ci]. ai pris [beaucoup ti de celles-ci]. have taken a kilo/ten/many of these (Cardinaletti & Giusti 2002: 11(33), brackets and traces added)

In (114), en pronominalises a subpart of the constituent in brackets and is extracted from it. Therefore some structure has to be present in order to be replaced by en (see Section 3.2.2 for more details). Another problem concerns the role/position of the PP in partitive noun phrases. Jackendoff (1977 following Jackendoff 1968) considers that “of the N’’ is part of the N’ complement” on a par with examples such as a group of the men, a gallon of the wine where of the N’’ behaves like a complement of NP (1977: 107) (recall (109)). If the constituent of the men is a complement of the empty noun it means that it is selected by a noun which is non overt. Such a position is also adopted by Abney (1987), who examines partitive PPs in comparative and superlative constructions. He proposes that the partitive PP is the complement of an empty N (1987: 344). Assuming that the partitive phrase in comparative and superlative constructions is parallel to the one occurring with quantities/numerals or in bare partitives (Bresnan 1973; Milner 1978), Cardinaletti and Giusti give different arguments against the selection of the partitive PP by the noun (2002: 16–17). The first one comes from an assumption they attribute to Abney and Jackendoff. This assumption is that only empty Ns can select a partitive PP. According to Cardinaletti and Giusti this is quite ­improbable as (115) below shows: (115) a. the best books of those you lent me b. many books of those you lent me c. *the many books of those you lent me d. *the many of the books you lent me

(Cardinaletti & Giusti 2002: 17(52))

In (115a,b) the head noun books is overt and the presence of the partitive PP is perfectly grammatical. If the PP were selected by an empty noun, this would be ­unexpected.

 The Layered DP: Form and meaning of French indefinites

The ungrammaticality of (115d) would also be unpredicted, as in this example the head noun is non-overt. In other words, if only empty nouns could select partitive PPs, (115a,b) should be out and (115d) should be fine.44 Attributing selectional features to empty elements seems implausible all together. Another argument against the selection of the partitive PP by the noun is that if the N selected the PP, they should form a constituent and hence be grammatical in left-dislocated constructions, contrary to fact: (116) a. *books of those you lent me, I’ve read only the best. b. *books of those you lent me, I’ve read only some.  (Cardinaletti & Giusti 2002: 16(51))

In (116), books of those you lent me cannot be preposed suggesting that it is not a constituent. Such examples contrast with Students of linguistics, I’ve met many, where an N and its complement are moved without leading to ungrammaticality. Another argument against the selection of the partitive PP by the (empty) noun is Baker’s UTAH (Uniformity of Theta Assignment Hypothesis) (1988: 46) according to which the arguments selected by a lexical head are always projected in the structure even if it is non-overtly. If I understand this correctly, it means that the argument of the men selected by the empty man in (109) should always be projected in the structure of the noun man, which sounds unlikely. The above evidence shows that the partitive PP is not an argument of the (empty) noun. What Cardinaletti & Giusti (2002) propose, on the basis of previous work of theirs, is that the partitive PP is an argument of what they call ‘existential quantifiers’, namely vague numerals like many, few and some and cardinals. More precisely, they argue that these quantifiers (which represent quantities or numerals in the terms adopted here) are dyadic elements which do not only select a partitive PP (of those…) but also an indefinite DP, labelled quantitative DP (boys) (117).45 According to Cardinaletti and Giusti, the former represents the set out of which the ‘quantifier’ picks up a subset, its restriction (boys in (117)). The structure they advocate is reported below.46

.  According to Cardinaletti and Giusti, (115) shows that the partitive PP is selected by the quantifier. The idea is that (115c,d) are ungrammatical because in these examples, many is not a quantifier as the presence of the article preceding it shows. It has an adjectival status. Such an analysis suggests that the PP can also be selected by the superlative best (115a). .  When preceded by an article as in the many children, many is analysed as a quantitative adjective generated in the specifier of a projection dominated by DP. .  In their paper, Cardinaletti and Giusti adapt this structure (2002: 36) to make it compatible with Kayne’s (1994) antisymmetry hypothesis and Chomsky’s (1995) bare phrase structure hypothesis. For ease of representation I will stick to the structure reported in the text.

Chapter 3.  Du/des-NPs 



(117)

QP

Spec

Q′

Q′

PP

Q

DP

P

DP

many

boys

of

those …



(Cardinaletti & Giusti 2002: 18(55))

A key assumption is that quantifiers head a projection which is outside DP, as has first been proposed by Sportiche (1988), although his primary interest was not the structure of DP. That quantifiers are heads higher than DP has been independently proposed for Italian by Giusti (1991) and for Hebrew by Shlonsky (1991). One argument to consider quantifiers as heads is that their specifier functions as an escape hatch for the DP they select. Despite the attractiveness of Cardinaletti and Giusti’s analysis and their thorough discussion of the data, including NE and en pronominalisation in Italian and French respectively, I will not adopt it here. One reason is that, extending Cardinaletti and Giusti’s proposal to explicit partitive du/des-NPs as in (118) implies that the element attributed selectional properties by the authors is empty. However, it is unlikely that the partitive PP (and the empty DP) be selected by an empty category. (118) a. b.

Ho letto ∅ dei libri. J’ai lu ∅ des livres. [I] have read of.the books ‘I read some books.’

(Cardinaletti & Giusti 2002: 39(133)

If the quantifier is empty it is implausible that it is able to select arguments, on a par with an empty N, as discussed above (recall the discussion of (115)). More importantly, it seems to me that expressions with a partitive PP are not necessarily quantificational, i.e., QPs, as illustrated in (119) and conversely that the presence of an ‘existential quantifier’ does not imply a partitive reading: (119) a. b.

Je souhaite rencontrer deux des professeurs du département. I would.like to.meet two of.the professors of.the department ‘I would like to meet two of the professors of the department.’ J’ ai vu deux des filles dans la cour. I have seen two of.the girls in the yard

 The Layered DP: Form and meaning of French indefinites

(120) a. b.

Il s’ agit de Marie et Sophie. it se is.about of Marie and Sophie ‘I saw two of the girls in the yard, namely Marie and Sophie.’ Ce matin, j’ai vu beaucoup d’enfants. this morning I have seen many of kids ‘This morning I saw many kids.’ Ce matin, j’ai rencontré trois enfants. this morning I have met three kids ‘This morning I met three kids.’

If somebody arrives at the department and utters (119a), they could very well mean any two professors of the department, which would correspond to a property-denoting reading (PropP). This is supported by the fact that we are dealing with an intensional context (souhaiter ‘wish’), which favors the property reading. Similarly, (119b) may be about two girls in particular. As these girls can be identified, deux des filles is not quantificational, at least not as defined in Chapter 2 Section 2.2. In the terms used here, they are S-referential. In (120), beaucoup d’enfants and trois enfants do not (necessarily) have a partitive meaning. A quantitative reading is perfectly natural, which is unexpected if beaucoup and trois were quantifiers selecting a partitive PP. In effect, if arguments selected by a lexical head are always projected in the structure as suggested by the UTAH mentioned above and assumed by Cardinaletti and Giusti, the partitive PP has to be present in the structure of (120), even if it is phonologically empty, which should lead to a partitive interpretation. It therefore seems to me that, although the examples in (120) express some kind of quantity, they are not quantificational. The analogy with a quantity expression like un kilo supports this conclusion. Recall (114) repeated for convenience: (121) a. b. c. 

J’ J’ J’ I

eni eni eni en

ai pris [un kilo ti de celles-ci]. ai pris [dix ti de celles-ci]. ai pris [beaucoup ti de celles-ci]. have taken a kilo/ten/many of these (Cardinaletti & Giusti 2002: 11(33), brackets and traces added)

The parallel between (121b,c) and (121a) suggests that dix and beaucoup should be treated on a par with un kilo which is not a quantifier. The necessity for partitive structures to involve a quantity/numeral, be it empty as in bare partitives like explicit du/ des-NPs, also accounts for the fact that a partitive PP is impossible with definites and universal quantifiers as illustrated in (122). Note that this also means that une in (122a) is a numeral and not an article. (122) a. *celle-ci des filles vs. une des filles this of.the girls one of.the girls b. *toutes des filles vs. certaines des filles all of.the girls some of.the girls

Chapter 3.  Du/des-NPs 



Informally, the above discussion suggests that the partitive PP somehow ‘builds on’ a quantity/numeral element. Intuitively, it means that the absence of a quantitative element would prevent the layer hosting the partitive PP to be projected, although the details of such a mechanism would have to be spelt out. Taking these observations into account, I propose that the partitive PP occupies the projection immediately above the quantitative element in the inflectional domain. Put differently, I suggest that the PP is in the specifier of a functional projection on a par with other noun modifiers, including cardinals (Chapter 1 Section 4.4.2.1).47 Such a proposal is compatible with the fact that in quantificational noun phrases the PP is part of the restriction (Doetjes 1997: 160), whereas in noun phrases which are not quantificational, it is similar to a property restricting the domain of the N (Dobrovie-Sorin & Beyssade 2004: 119). The structure of the property-denoting complement in (119a) is given in (123). The projection hosting quantities/numerals is labelled FP2 as in Section 3.2.2 and the one containing the partitive PP FP2, as it is probably lower than the layer related to grammatical number (FP3 in Section 3.2.2). (123)

PropP Spec

Prop′ Prop

… FP2′

PP des prof.

F′ F2′

FP2 deux

F′ F2

… NP ec

If the partitive PP is in the specifier of a functional projection above the numeral, movement to the left periphery will be phrasal (Chapter 1 Section 4.4.1.3) (Kayne 1994). More precisely, FP2 will move to PropP, on a par with the indefinite article in un-NPs (Chapter 2) and with du/des (Section 3.2.2). Movement of the whole chunk,

.  Such a proposal might even play a role in the fact that the PP and the head noun must have the same features, including gender (as observed by Cardinaletti & Giusti 2002).

 The Layered DP: Form and meaning of French indefinites

in opposition to movement of deux, has to be assumed to account for examples where N is overt as in (110) and (111). This analysis extends to the S-referential noun phrase in (119b) except that the chunk which is displaced would target SRefP. These examples crucially show that partitive PPs may occur in the structure of PropPs and SRefPs. Put differently these noun phrases are partitive but not quantificational (in contrast with covert/implicit partitives which are QPs) (see also Dobrovie-Sorin & Beyssade 2004). Let us now concentrate on the structure of what I have called ‘explicit’ partitives, i.e., du/des-NPs in the object position of fragmentative verbs. As seen in Section 2.2, these du/des-NPs are property-denoting, i.e., PropPs. They differ from the examples in (119) in that the quantity is not expressed overtly and in that their distribution is constrained. The structure I advocate for du gâteau/des biscuits in an example like Jean a mangé du gâteau/des biscuits ‘John has eaten of.the cake/biscuits’ with a partitive reading is reported below. (124)

PropP Spec

Prop′ Prop dei

…FP2′

PP des biscuits du gâteau

F’ ti

FP2 Spec

F′ ti

deP Spec

de′ ti

… NP ec

Crucially, I suggest that the de discussed in the preceding section is present in the structure of explicit partitives, that it is generated in the projection labelled deP in the



Chapter 3.  Du/des-NPs 

absence of a better label and that it moves to PropP as illustrated in (124).48 This means that the structure of ‘explicit partitives’ is similar to the one of P-du/des-NPs except that le/les is not present in the structure and that they contain a partitive PP. The latter sits in the specifier of FP2’ as in (123), which comprises an overt quantity/numeral. More generally, it means that all du/des-NPs contain a functional de generated in the inflectional domain. If this is correct is might be at the origin of the amalgam of partitive and non partitive du/des-NPs in the literature of which the following quote is an ­illustration (from Gross 1967, cited in Kupferman 1979: 3). (125) Nous nous intéressons plus particulièrement à certains indéfinis: l’indéfini pluriel: des (de les); les partitifs: de la (de l’) et du (de la ou de l’). Nous considérons que les trois détermineurs ont une nature identique: ils sont formés de la préposition de et d’un article défini générique (Artg): nous les qualifierons tous trois de partitifs. ‘We are more specifically interested in certain indefinites : the plural indefinite: des (de les ‘of the’); the partitives: de la ‘of the.fem’(de l’) and du ‘of.the.masc’ (de la or de l’). We consider that the three determiners are identical in nature: they are formed with the preposition de ‘of ’ and with a generic definite article (Artg): we will qualify the three of them as partitive.’ (translation mine)

As seen in the above discussion, I assume that what all du/des-NPs have in common is a functional de generated in the inflectional domain and targeting a projection of the left periphery when the nominal expression is an argument. What a structure like (124) implies is that there are two successive de’s ‘of ’, i.e., de de + le gâteau. However, at spell out, a single one is pronounced. In effect, according to the Grammar of Port-Royal two successive de’s are impossible because it would give rise to a ‘cacophony’.49 More formally, the disappearance of one de seems to be a

.  If this is the structure of explicit du/des partitives, one issue which should be addressed is what part of it is pronominalised by en in examples like (i): (i) Speaker A: As- tu pris du café? have you taken of.the coffee ‘Have you taken some coffee?’ Speaker B : Oui, j’ en ai pris. yes I en have taken Intuitively, B’s answer seems to mean Oui, j’en ai pris [un peu] ‘yes I en have taken [a little]’. If this is the case en could pronominalise the partitive PP. This enpartitive differs from the en pronouns mentioned in Section 3.2.2. .  « Car de des, et encore plus de de, eût trop choqué l’oreille, et elle eût eu peine à souffrir qu’on eût dit: Il est accusé de des crimes horribles, ou, il est accusé de de grands crimes. » (cited in Frontier 1997: 339, from Arnauld et Lancelot, p.41).

 The Layered DP: Form and meaning of French indefinites

case of haplology, i.e., deletion of one of the offending morphemes.50 That languages exhibit a resistance against accidental repetition of morphemes has been observed by a number of authors (Perlmutter 1971; Stemberger 1981; Menn & Mac Whinney 1983; Yip 1998; Plag 1998 cited in Neeleman & van de Koot 2004). As noted by Neeleman & van de Koot (2004), this resistance is reminiscent of the Obligatory Contour Principle in phonology (Leben 1973; Goldsmith 1979).51 Examples of cases where the repetition of phonologically identical morphemes leads to ungrammaticality are provided below. (126) is a Dutch example and (127) a French one. (126) Context: Which filmstar do you like best? a. die met dat rooie haar that with that red hair ‘the one with the red hair’ b. ??die die dat rooie haar heeft that that that red hair has c. degene die dat rooie haar heeft the.one that that red hair has  (Neeleman & van de Koot 2004 (5a,c,d)) (127) Je préfère que tu restes, plutôt que (*que) tu t’en ailles. I prefer that you stay rather than that you go.away  (Neeleman & van de Koot 2004 (13))

(126) shows that the demonstrative die cannot be followed by a relative clause introduced by the same pronoun. As relative clauses introduced by die may follow the nonhomophonous demonstrative degene as in (126c), it seems unlikely that this is due to semantic or syntactic factors. The Dutch examples above illustrate a strategy of avoidance of repetition. The French example in contrast illustrates deletion (Radford 1977, 1979). In (127), deletion is triggered by the adjacency of the complementizer que ‘that’ and the homophonous comparative. The above examples suggest that it is the second of the two analogous morphemes which is avoided/deleted. However, no arguments are provided to support this proposal. A closer look at the distribution of subject du/des-NPs suggests that, in these cases, it might be the first de ‘of ’ which is not pronounced and more precisely that this element may be present in the structure as an empty category. Consider the following example:

.  Haplology is one strategy among others which avoids the repetition of similar morphemes. Further options would be to associate either morpheme with a different realization, typically based on a subset or a superset of its feature, or to make available so-called ‘portmanteau’ morphemes, which spell out the features of both morphemes. .  Thanks to Eric Haeberli for bringing this literature to my attention.

Chapter 3.  Du/des-NPs 



(128) a. b. 

De la corde / de cette corde traînait par terre. of the rope of this rope was.lying on the floor De l’eau / de cette eau boueuse recouvrait la route. of the water of this water muddy was.covering the road (Kupferman 1979: 8 (23–4))

In (128), the subjects in de la/de l’ cannot be interpreted as explicit partitives (Kupferman 1979). The reason, I argue, is that in their explicit partitive interpretation du/des-NPs contain a null element which has to be lexically governed. More precisely, I suggest that the de ‘of ’ in the PropP of their structure is phonetically null and needs to be licensed, exactly like the null D in bare nouns in subject positions in Italian. Recall that, according to Longobardi (1994), bare nouns in Romance are ruled out from subject positions because the empty determiner they contain is not lexically governed in this position. The relevant example is repeated below from Chapter 2 Section 3.1.1: (129) *Acqua viene giù dalle colline. water comes down from the hills

If an empty category has to be lexically governed, it accounts for the ungrammaticality of (129) and for the impossibility for (128) to host an explicit partitive, including explicit partitive des-NPs. Implicit du/des-NPs, i.e., Q-du/des-NPs with a covert partitive meaning, in contrast, are expected to be fine in subject positions because they involve a single de ‘of ’, hence no empty category. This is illustrated in (130), repeated from Section 2.2. (130) Des filles étaient blondes, d’autres avaient les cheveux foncés. of.the girls were blond others had the hair dark

The idea that in a sequence de de it is the first de which is omitted is not in line with well-known literature. Gross (1967) for example proposes that it is the second de which is deleted (1967: 116). However, the examples he gives differ from the ones examined above. Consider (131): (131) a. *Jean parle de (du sable + de la soupe + des chevaux). John talks about of.the sand + of the soup + of.the horses b. Jean parle de (sable + soupe + chevaux). (Gross 1967: 107)

In (131), the verb selects a PP headed by de, i.e., parler de ‘talk of ’. If it is the second de which is omitted in (131), it concerns the ‘determiner-like’ de and not the preposition. In other words, the element which is deleted in a structure like (124) and in examples like (131) is the same: the ‘determiner-like’ de. This means that we can restate the avoidance of repetition in terms of category instead of sequence. Put differently, instead of speculating on which of the first or second offending element is avoided/deleted, I propose that it is the functional de generated in the inflectional domain of du/des-NPs (not the preposition): in (124) it is the first occurrence whereas in (131) it is the second one.

 The Layered DP: Form and meaning of French indefinites

An advantage of a structure like (124) is that it contains a PP without being a PP. In other words, it accounts for the intuitions represented in (108) that explicit partitive du/des-NPs involve a prepositional de ‘of ’ which differs from the de found in other du/des-NPs. At the same time it accounts for causative constructions of the type faire faire quelque chose à/par quelqu’un ‘make do something to/by someone’ which can only take a DP followed by a PP as (132) illustrates. (132) a. *J’ai fait boire Marie ce lait. I have made drink Mary this milk b. J’ai fait boire ce lait à / par Marie. I have made drink this milk to / by Mary c. J’ai fait boire Marie dans cette tasse. I have made drink Mary in this cup

(Milner 1978: 78)

(132a) shows that the combination of complements DP + DP – Marie + ce lait – leads to ungrammaticality. In (132b) and (132c) however, the DPs ce lait and Marie are followed by a PP, à/par Marie and dans cette tasse respectively, and both sentences are fine. When ce lait is replaced by de ce lait, we observe that the judgements are the same as in (132a,b): (133) a. *J’ai fait boire Marie de ce lait. I have made drink Mary of this milk b. J’ai fait boire de ce lait à / par Marie. I have made drink of this milk to / by Mary

(Milner 1978: 78)

As the causative sentences illustrated in (133) take a combination of complements DP + PP, (133a) should be fine if de ce lait were a PP, contrary to fact. Similarly, as (133b) is grammatical, de ce lait cannot be a PP. Although the meaning of de ce lait is clearly partitive, (133) suggests it is not a PP. This is accounted for by the structure I propose in (124), which contains a PP but belongs to a different category. The presence of the PP in the structure accounts for its partitive reading and the intuitions expressed in (107) and (108). (124) also allows us to distinguish between fragmentative verbs such as boire ‘drink’ and manger ‘eat’ (Section 1.2) and verbs which take a prepositional complement such as se souvenir (de) ‘remember’ and rêver (de) ‘dream’ (Zribi-Hertz 2003). Consider the following example: (134) Jean est venu me voir à l’hôpital et Jean is come me to.see at the hospital and a. il m’ a demandé de quoi je me souvenais. he me has asked of what I REFL remembered b. *il m’ a demandé de quoi je mangeais. he me has asked of what I was.eating  (Zribi-Hertz 2003: 8 (15))



Chapter 3.  Du/des-NPs 

In (134), de quoi ‘of what’, used by Kupferman to distinguish between prepositional and non-prepositional complements of fragmentative verbs, is grammatical with se souvenir ‘remember’ but not with manger ‘eat’. This shows that the complement of these classes of verbs should be distinguished. Zribi-Hertz observes that de quoi ‘of what’ with fragmentative verbs is only felicitous if it refers to noun phrases already introduced in the discourse. In terms of structure, I assume that, in their partitive reading, complements of fragmentative verbs are as in (124), i.e., PropPs containing a PP. In contrast, the complement of verbs of the type se souvenir de ‘remember’ would be PPs. The distinction between these two types of complements is supported by the use of dont ‘of.which’: (135) a. b. 

Quand je suis arrivée chez Marie, elle m’ a when I am arrived at Marie’s she to.me has tout de suite demandé ce dont je souhaitais immediately asked what of.which I wanted {parler /*manger}. to.talk / to.eat Marie avait préparé plusieurs plats : du riz, de la Marie had prepared several dishes of.the rice of the soupe, du goulash. Elle m’ a demandé soupe of.the goulash she to.me has asked ce dont je souhaitais {?manger / reprendre}. what of.which I wanted to.eat / to.take.again (Zribi-Hertz 2003: 10(19))

In (135a), only a verb with a prepositional complement is felicitous. In contrast, the fragmentative manger ‘eat’ is out. In (135b), where the various types of food have been introduced in the context, fragmentative verbs are fine, although some nuances in the grammaticality judgements seem to take place. In (124), this means that the DP embedded in the PP is familiar, which is in accordance with Kupferman’s (1979) analysis. Finally, if the overt de ‘of ’ in explicit partitive du/des-NPs is a preposition, it predicts that it can select any type of DP, in the traditional sense. This argues against a Partitive Constraint à la Jackendoff (1977): (136) Partitive Constraint In an of-N’’’ construction interpreted as a partitive, the N’’’ must have a demonstrative or genitive specifier. (cited in Hoeksema 1996: 6)

To get rid of the disjunction in the Partitive Constraint, Barwise & Cooper (1981) propose that of must be followed by a definite NP.52 The Partitive Constraint has been .  Keenan & Stavi (1986) suggest another structure where all of the in all of the men is analysed as a complex determiner with men as its argument.

 The Layered DP: Form and meaning of French indefinites

widely discussed in the literature and different problematic aspects have been pointed out (Stockwell, Schachter & Partee 1973; Ladusaw 1982; Hoeksema 1984; Abbott 1996; Reed 1996; Wilkinson 1996 etc). Consider for instance the following counterexamples to the Partitive Constraint (from Abbott 1996: 29ff.): (137) a. One of some boys who were playing in the alley got arrested. b. He ate three of some apples he found on the ground.  (Stockwell, Schachter, & Partee 1973: 144) (138) I heard too much of one speech and not enough of the other.  (Selkirk 1977: 315, n 7) (139) a. That book could belong to one of three people. b. This is one of a number of counterexamples to the PC. c. John was one of several students who arrived late. (Ladusaw 1982: 240) (140) a. I’ll be back in three quarters of an hour. b. Why settle for half of a loaf? c. There was most of a birthday cake and all of a large vegetarian pizza sitting on the buffet. d. That sounds like too much of a good thing.

These examples show that the embedded DP of a partitive construction does not have to be definite, which is why I do not adopt the Partitive Constraint given in (136). In sum what this section shows is that explicit du/des-NPs contain a partitive PP in the specifier position of the functional projection immediately dominating the one encoding quantities namely FP2. De moves to Prop, where it is erased as a case of haplology. Partitive constructions with an overt quantity, like deux des filles ‘two of.the girls’, may belong to different categories, including SRefP. This means that all partitives are not quantificational.

3.4  Notes on the-NPs In Section 3.2.2, I have proposed that le/les, in du/des-NPs but also in the-NPs (to the exception of the count le), is generated in the projection encoding grammatical number. This means that the so-called definite article is generated very low in the structure and that it moves to (remerges in) higher positions, depending on the interpretation of the noun phrase it occurs in (Borer 2005a). If le/les has to move to the nominal left periphery to form an argument, the question which arises is whether the framework developed here can be extended to theNPs. In other words, could the left periphery of the-NPs be constituted of several layers responsible for different interpretations? In the affirmative, this would mean that a clear typology of the-NPs should be established in order to relate them to specific structures. Furthermore, assuming that reference is associated to (one of) the highest

Chapter 3.  Du/des-NPs 



projections of the left periphery, it would imply that many the-NPs are not referential (see Kupisch & Koops 2007 for a recent account of ‘non-specific’ definite articles). The reason is that their left periphery could contain a single projection, not related to reference. The differences with the left periphery of the three types of indefinites studied here should also be determined. In effect, it would be surprising that the left periphery of the-NPs contained a Quantifier Phrase (QP) for example. Recall that I have proposed that Quantifier Raising (QR), which is responsible for scope properties, selects noun phrases containing a QP layer in their structure. As the-NPs do not have scope properties, we expect that they do not contain a QP.53 Another issue which should be addressed is whether the reference of referential the-NPs can be assimilated to the one of S-referential indefinites. In effect, the-NPs have rigid reference contrary to S-referential un-NPs and des-NPs whose reference is either context dependent or co-varying with a universal quantifier. In terms of structure, the question is whether referential theNPs comprise an S-reference Phrase (SRefP) or whether their reference is encoded in another projection. This is for the theory. Let us now turn to some examples. The prediction mentioned above, namely that the-NPs should be able to be non referential if their left periphery contained several layers (assuming that only the highest is related to reference), seems to be born out. In effect, many noun phrases with the ‘definite’ article are not referential. This is not only the case of le/les in du/desNPs (Section 3.2.2) but also of generic noun phrases like les chiens ‘the dogs’ in Les chiens ont quatre pattes ‘Dogs have four legs’. Despite the presence of les, les chiens is not definite, in the sense that it does not have to be familiar to the hearer, and it is not referential. In addition to the generic reading, non-referential the-NPs seem to be able to have a property-denoting reading, known as predicative reading (recall fn. 20 in Chapter 2 which concerns Partee’s (2002) observation that the-NPs may be predicative). That the-NPs may have such a reading is also observed by Donnellan’s (1966) well-known work. For example, in some contexts, the murderer might be paraphrased as “whoever murdered …” in which case it does not make reference to a particular individual. Many other examples involve a non referential le. Consider (141): (141) a. b. c.

montrer le chemin, suivre les instructions show the way, follow the instructions le matin, le lundi, l’été the morning, the Monday, the summer l’ amitié, la peur, la fidélité the friendship, the fear, the faithfulness

.  However, see Partee (2002) who assumes that the-NPs may be quantificational.

 The Layered DP: Form and meaning of French indefinites

As mentioned above, the the-NPs in (141) are not referential. This means that they cannot be defined as ‘referring to a familiar and unique entity’ as often assumed in the literature. In addition, the the-NPs in the above example are not familiar. These observations suggest that the analysis of the-NPs should be reconsidered. The fact that the-NPs can have different interpretations suggests that an analysis like the one developed here may apply to them. In other words, it is conceivable that they have an extendable telescopic left periphery, on a par with un-NPs and des-NPs, whose projections have to be identified. Although the left periphery of the-NPs will probably differ from the one of un-NPs and du/des-NPs, the above discussion suggests that the three kinds of nominals may have in common (at least) one projection, the Property Phrase advocated here. Further research is required to determine to what extent the various interpretations of the-NPs can be related to different structures.

4.  Conclusion In this chapter I have addressed two main issues, namely (i) the extension of the analysis developed in Chapter 2 for un-NPs to du/des-NPs in terms of classification and in terms of structure of the left periphery and (ii) the internal structure of the inflectional domain of du/des-NPs taking into consideration interpretative features like [count], [quantity] and [grammatical number]. The analysis presented in this chapter lies on the assumptions made in Chapter 2, namely that (i) for a noun phrase to function as an argument the lowest projection of its left periphery must be projected and that (ii) nominals with different interpretations have different structures. The topic of du/des-NPs has been introduced with some background information. First, I have reported the interpretations traditionally attributed to du/des-NPs and then the structural analyses corresponding to the two main readings of du/des-NPs, namely the quantitative and the partitive reading, in a pre-DP framework. This information has been crucial to distinguish between explicit and implicit partitives and to show that the former contain a Prepositional Phrase in their structure contrary to the latter. This chapter has shown that the conclusions reached in Chapter 2 for un-NPs extend to des-NPs (which have to be distinguished from du-NPs, see below). This concerns their classification into three types, their (lack of) existential reading and their (lack of) scope properties. In other words, I have shown that des-NPs may be S-referential, quantificational or property-denoting and argued that they may belong to different categories. Put differently, it means that they may be S-reference Phrases (SRefPs), Quantification Phrases (QPs) or Property Phrases (PropPs). This implies



Chapter 3.  Du/des-NPs 

that their left periphery may comprise three, two or one layer(s). In the first case, the ­constitution of their left periphery is SRefP > QP > PropP, in the second case QP > PropP and in the last case PropP. Arguments for such a telescopic left periphery come from empirical facts, mainly intervention effects. In addition, as des-NPs may be existential or not, I have assumed that when they are, the source of existentiality may be either the existential quantifier in their landing site at LF or the existential operator supplied by the predicate selecting them. The former option concerns S-referential des-NPs whereas the latter applies to property-denoting des-NPs. Property-denoting des-NPs are not existential in intensional contexts. Finally, I have argued that des-NPs have scope properties when they contain a QP layer, i.e., when they are quantificational and when they are S-referential. In the former case, they are QPs and are blocked in islands, whereas in the latter case they may move out of islands at Logical Form (LF) and take widest scope or ‘intermediate scope’. To account for the fact that S-referential des-NPs target two different positions at LF, I have proposed that they take widest scope when their interpretation is deictic, i.e., depends on extra linguistic information, and that they take intermediate scope, when they are not deictic. In the latter case, they co-vary with a universal quantifier. The strength of this analysis is that it accounts for several interpretative features of these indefinites, namely their (lack of) reference, their (lack of) scope properties and their (lack of) existentiality. Du-NPs, in contrast with des-NPs, have only two interpretations. In other terms, they belong to two semantic types. This means that they may be property-denoting or quantificational. In other words, their left periphery comprises a single layer, PropP or two layers, PropP and QP. The absence of a SRefP accounts for the fact that they cannot be the antecedent of d’autres/les autres ‘others, the others’, contrary to des-NPs. What prevents du-NPs to project the SRefP layer is the lack of atoms in their extension, that is, their mass reading. Finally, when du-NPs are property-denoting, they may be existential or not, on a par with property un-NPs and property des-NPs. This means that they are existential when selected by a predicate supplying an existential operator and that they are not in intensional contexts. When they are quantificational they are presuppositional. As for the discussion of the inflectional domain of du/des-NPs, it extends to other nominals, in particular to un-NPs and le/les-NPs. It also examines, to a minor extent, nominal expressions with overt quantities like beaucoup ‘many/much’ and peu ‘few/little’ and de-NPs, which I suggest contain an empty category representing a quantity. The structure I have postulated is composed of several functional projections related to the features [count], [quantity] and [grammatical number]. Following Borer (2005a), I assume that listemes are mass and that they only become count in the appropriate context. This means that there is a projection responsible for the count

 The Layered DP: Form and meaning of French indefinites

r­ eading. This projection hosts the plural –s, the indefinite article and the count le ‘the. sg’. Contrary to Borer however I have suggested that the projection of this layer leads to an individuated reading, at least in French. I have labelled this projection FP1, as it is the lowest functional projection in the structure of nominals. As for the quantities associated with nominal expressions, I have proposed that they are only encoded in the structure when they cannot be inferred from the other features of the nominal expression. This means that the quantity 1 associated with un ‘a’ and with the count le ‘the.sg’ is due to their being singular and count, and that les-NPs and des-NPs implicitly mean more than one because they are plural. As for du/des-NPs corresponding to some-NPs, I have argued that the quantity they represent cannot be inferred and therefore that a layer related to quantities, FP2, is projected in their structure. This contrasts with du/des-NPs corresponding to bare nouns, which are not quantitative, and which do therefore not contain a FP2. The presence vs. absence of this projection also distinguishes nominals containing the numeral un from the ones containing the indefinite article un. Put differently, I have proposed that only the former comprise a FP2. Another difference between the numeral and the indefinite article un is that the former is generated as an XP contrary to the latter which is a head. Furthermore, I have argued that the inflectional domain of all the nominals examined comprises a projection encoding grammatical number, FP3. The reason for postulating such a projection is that neither the presence nor the absence of FP1 or FP2 allows us to infer whether a nominal is singular or plural. For example, as singular expressions may be count or mass, the [singular] feature cannot be inferred from the presence or absence of FP1 and therefore must be encoded in a separate projection. FP3 hosts the le/les component of du/des but also the mass le and les (only the count le is generated lower, in FP1 as seen above). As for the base position of the de component of du/des, I have hypothesised that it is generated in a functional projection situated between FP1 and FP2, deP, and that it moves higher, for example to the left periphery to form arguments. The idea is that all du/des-NPs, independently of their ‘some-reading’, their ‘bare-reading’ or their partitive reading contain this functional de. Crucially, the account proposed here implies that the fusion of de + le/les in du/des-NPs takes place in FP3. The discussion on quantities led me to consider additional issues. One of them is the structure of de-NPs, which are restricted to negative contexts, from which property-denoting un-NPs and du/des-NPs are banned. What I have proposed is that they contain an empty category in FP2 which has to be bound by a negative operator. This means that their structure is analogous to the one of beaucoup de ‘many/much of ’ and peu de ‘few/little of ’ to the exception that the quantity in FP2 is empty. Another issue I have addressed is en pronominalisation. What I have proposed is that en can replace the structure below FP2 when this projection contains some overt quantity/ numeral or the whole nominal expression, when it is a property-denoting du/des-NP. Property-denoting un-NPs in contrast cannot be replaced by en for reasons which



Chapter 3.  Du/des-NPs 

remain ­unclear. I have also suggested that in explicit du/des constructions en can replace the partitive PP. As en can also be a genitive pronoun (when it replaces a PP in non ­partitive ­structures), it means that there are (at least) four kinds of en pronouns: enidentification, enproperty, enpartitive and engenitive. An extended inflectional domain such as the one proposed in this chapter supports the conclusion I reached in Chapter 2, namely that syntax is not about words or morphemes but about features, which allows more fine-grained distinctions. This predicts that structures become more and more complex to encode interpretative features not considered so far. Another issue dealt with in this chapter is partitivity, which I have split into ‘implicit’ partitivity and ‘explicit’ partitivity. The former expresses a subset of a non-overt set, whereas the latter overtly expresses the set. As implicit partitives have the covert partitive meaning attributed to quantificational noun phrases, they are QPs. As seen in this chapter, du/des-NPs may be implicit or explicit partitives. Explicit du/des-NPs are only attested in the complement position of a restricted set of verbs called fragmentative verbs. As for the structures of partitive du/des-NPs, they nicely follow from the analysis developed here. Implicit du/des-NPs are QPs, with de moving to FP3 to form du/des and du/des moving higher to Q. In contrast, explicit du/des-NPs are PropPs, with de moving to FP3 to form du/des and du/des moving higher to PropP. In addition, what I have proposed is that explicit du/des-NPs contain a partitive PP in a functional projection immediately above FP2, labelled FP2’. This means that this PP is analogous to other modifiers, analysed as XPs in the specifier of a functional projection. ­Furthermore, as the partitive PP is only licit with nominal expressions encoding quantities/numbers, I have suggested that the projection of FP2’ is only possible when FP2 is present in the structure. In other words, FP2’ depends on FP2. Such an analysis means that explicit du/des-NPs involve two successive de’s. The idea defended here is that the highest one, which means the one generated in the inflectional domain, is ­deleted as a case of haplology. Crucially, this accounts for the subject-object asymmetry observed for explicit partitives. In effect, the presence of an empty category in ­explicit du/desNPs, in contrast with implicit du/des-NPs and non partitive du/des-NPs in general which contain a single de, explains their restricted distribution. Finally, this chapter raises the same questions as the preceding one on the structure of noun phrases in general and more specifically on the structure of the left periphery of noun phrases containing other so-called weak quantifiers, demonstratives, possessives and of course definite articles. Only the latter have been alluded to in this chapter and the brief discussion I provided suggested that an analysis along the lines developed here could be extended to this kind of nominals. In effect, not only do argumental the-NPs have different interpretations, namely the typical definite/referential reading and a property reading, but they also comprise an element, the, which on a par with a, is generated in the inflectional domain and head moves to the nominal left

 The Layered DP: Form and meaning of French indefinites

periphery to form an argument (Borer 2005a). In other words, it is plausible that the various readings of the-NPs are due to the movement of the from its base position to different layers of the left periphery. Although the composition of the left periphery of the-NPs has not been addressed, one piece of information which can be inferred from the interpretation of these noun phrases is that they do not contain a QP, otherwise they would have scope properties, contrary to fact.

chapter 4

Bare nouns 1.  Introduction In this chapter, I address the question of bare common nouns, i.e., nominal ­arguments without a determiner. As this book is on French and as this language is traditionally assumed not to have bare nouns (but see Section 2.2), this choice might seem surprising. However, the objective is to see whether the analysis developed here extends to bare nouns interpreted as du/des-NPs in French. This means that a first restriction on the data examined in this chapter is their analogy with the indefinite du/des-NPs studied in Chapter 3. This implies that I concentrate on plural and mass bare nouns, ignoring singular bare count nouns (see for example Dobrovie-Sorin et al. 2006; Smith Stvan 2007). Coordinated bare nouns, which are possible in French, are examined to a minor extent in a separate section. Thus, unless specified otherwise, the term ‘bare nouns’ is used for non-coordinated bare nouns throughout. As bare nouns characterise Germanic and Romance languages in general, the approach is more comparative than in Chapters 2 and 3. As the study of bare nouns has been a major focus of inquiry for almost three decades (since Carlson 1977a,b at least), it will be impossible to address all the issues related to the topic in this chapter. The first question which will be tackled here is the interpretation of bare nouns. Put differently the aim is to determine whether these nominals may have different readings, on a par with du/des-NPs, i.e., whether they may be property-denoting, quantificational and/or S-referential. The ambiguity generally acknowledged mainly concerns Germanic languages whose bare nouns are classified as generic or existential. As genericity lies beyond the scope of this book (but see Section 5.1 for a short parenthesis on the topic), only the so-called existential interpretation will be examined. The second central issue of the chapter is the internal structure of bare nouns, i.e., the composition of their left periphery and of their inflectional domain. The problem of the left periphery is particularly challenging as bare nouns, by definition, do not have any determiner whose movement could trigger the projection of the layer(s) forming this domain, in contrast with du/des-NPs. Yet as bare nouns are arguments, the framework developed in this book predicts that their structure minimally comprises the Property Phrase (Chapter 2 Section 3.1.1). As for the inflectional domain, the

 The Layered DP: Form and meaning of French indefinites

same interrogations as for du/des-NPs arise, namely whether bare nouns have atoms in their extension, whether they encode any quantity, etc… The topic of coordinated bare nouns has been recently investigated in the literature (Heycock & Zamparelli 2003, 2005; Roodenburg 2004b). It is included in this chapter because these works show that (some) coordinated bare nouns may be ambiguous, in contrast with non-coordinated bare nouns. In other words, coordinated bare nouns have been attributed two different interpretations (in addition to their potential generic reading). What has to be established is thus whether these two readings correspond to the classification proposed in Chapter 3 for du/des-NPs. More generally, the study of coordinated bare nouns raises more questions than it solves. One reason is that it would require a thorough analysis of each conjunct, which is out of the scope of this chapter. In effect, it is very difficult to determine which features characterise the conjunction as a whole and which ones should be attributed to each conjunct. Furthermore, the issues concerning the potential scope properties of coordinated bare nouns are not evoked in the literature. Put differently, whether these nominals may have scope properties, at least in some contexts, would have to be investigated and tested. Two more issues raised by the study of bare common nouns are briefly mentioned at the end of the chapter. The aim is not to make any proposals concerning these problems, but rather to formulate a few questions related to the analysis developed in this book and to make a few remarks. The first topic is genericity and the second one incorporation. Both are concerned with contexts involving bare nouns, even though this is not necessarily the case: other kinds of nominals may be found in generic sentences and incorporation constructions.1 The organisation of the chapter is the following. Section 2.1 discusses the interpretation of bare nouns in Germanic and Romance languages and Section 2.2 shows that argumental bare nouns are not totally absent from French, although they generally differ from bare nouns in Germanic and Romance. Section 3 deals with structural considerations. The left periphery of bare nouns is addressed in Section 3.1 and their inflectional domain in Section 3.2. Section 4 examines coordinated bare nouns, focussing first on their interpretation (Section 4.1) and then on their structure (Section 4.2). Genericity and incorporation are mentioned in Section 5 and Section 6 concludes the chapter.

.  The unbounded reading of sentences involving (certain) indefinites and the status of determinerless languages are examined in Chapter 5. The reason is that the discussion will rely on our analysis of un-NPs and du/des-NPs.



Chapter 4.  Bare nouns 

2.  Types of bare nouns 2.1  Germanic and Romance All the Germanic and Romance languages have non-coordinated bare nouns in argument positions (French is mentioned in the next section).2 However, Germanic bare nouns and Romance bare nouns have different characteristics. One of them concerns their semantic interpretation as illustrated in (1) (see among others Laca 1990; Delfitto & Schroten 1991; Longobardi 2001; Dobrovie-Sorin & Laca 2003).

(1) a. John was eating    biscuitsexist . / John hates catsgen. (English) b. Gianni mangiava biscottiexist . / *Gianni odia gattigen. (Italian)  (Roodenburg 2004b: 122(6–7))

In Germanic, bare nouns are ambiguous between an existential and a generic reading (1a) whereas in Romance they can only be existential (1b).3, 4 As genericity has not been considered so far, the discussion will concentrate on the so-called ‘existential’ reading, i.e., the non-generic interpretation (genericity is however mentioned in Section 5.1). On the basis of bare plurals in Spanish, McNally (1995) argues that these nominals should be considered as property-denoting noun phrases even in argumental positions, and not as Heimian indefinites or kinds (see Dobrovie-Sorin & Laca 2003 for Romance in general). As mentioned in Chapter 2 Section 2.3.2, that property-denoting noun phrases can function as arguments is an approach which has only been explored in the early 1990s (Higginbotham 1985; de Hoop 1992; McNally 1992; Ladusaw 1994; Van Geenhoven 1995; McNally 1995; Dobrovie-Sorin 1996). The relevant Spanish examples are repeated below:

(2) a.

A la reunión no asistieron algunos profesores. at the meeting not attended some-pl professors ‘The meeting was not attended by some professors.’

(c or d)

.  For an analysis of bare plurals and bare singulars in Romanian and Spanish, see DobrovieSorin et al. (2006). .  The explanation often given in the literature is that some predicates cannot contribute an existential operator (Chapter 2 Section 2.3.2), a failure which prevents them from providing spatio-temporal anchoring (Carlson 1977a,b; Kleiber 1981a). As a result such predicates take generic arguments like cats in the text. If this is correct, it suggests that odia in Italian, in contrast with hates in English, always contributes an existential operator, thus preventing the generic interpretation of (1b). .  Longobardi (2001) shows that bare nouns in Romance languages can have a generic reading in some restricted contexts. As these cases are clearly limited, they will be ignored here ­(Vogeleer & Tasmowski 2005).

 The Layered DP: Form and meaning of French indefinites

b. A la reunión no asistieron profesores.  c. (∃x: professor(x)) [¬attend(x,m)] d. ¬(∃x: professor(x)) [attend(x,m)]



(3) a. b. c. d.

María cree que Sara ha robado unos libros. Maria believes that Sara has stolen a.pl books ‘Maria believes that Sara has stolen some books.’ María cree que Sara ha robado libros. (∃x: book(x)) [believe(m,^stole(s,x))] believe(m,^(∃x: book(x)) [stole(s,x)])

(4) a. b. c. d.

Hoy Juana tiene que leer unos artículos. Today Juana has that read.inf a.pl articles ‘Today Juana has to read some articles.’ Hoy Juana tiene que leer artículos. (∃x: article(x)) [☐ [read(j,x)]] ☐(∃x: article(x)) [read(j,x)]

(d only)

(c or d)

(d only)

(c or d)

(d only)

According to McNally (1995), the bare plurals in the b examples above are not ambiguous and their only interpretation is the one given in the d examples where they take narrowest possible scope with respect to any other operators in the same clause.5 Although McNally’s paper focuses on bare plurals, Chapter 2 and 3 of this book clearly show that other nominal expressions may function as property-denoting arguments. This is the case of un-NPs and du/des-NPs. As du-NPs are mass and as they may be compared to bare mass nouns (Chapter 3 Section 1.2), I will assume that the analysis of property-denoting arguments does not only concern bare plurals but that it extends to bare mass nouns. Interestingly, (3) and (4) above illustrate another characteristic of bare plurals: the b examples show that they do not necessarily assert existence (Chapter 2 Section 2.3). This observation is very important as it shows that the classification of bare nouns into existential vs. generic nominals is too simple: contrary to what examples like (1) suggest, non-generic bare nouns are not necessarily existential. When they are, however, they owe this reading to an existential operator provided by the predicate (Chapter 2 Section 2.3.2). In effect, existentiality, in the sense of ‘asserting existence’ can have different sources as seen in various sections of this book (see for example Chapter 2 Section 3.4.5). The existential operator may be provided by the predicate when the nominal denotes a property or by the target position of S-referential noun phrases (Chapter 2 Section 3.4) (Glasbey 2006 however shows that not all predicates allow existential readings for their

.  That bare nouns always have narrow scope has already been observed by Kroch 1974. Recall however that the term ‘narrow scope’ is misleading as property-denoting nominals rather seem to have no scope properties (Chapter 2 Section 3.2.2).



Chapter 4.  Bare nouns 

bare plural objects). In other words, bare nouns, on a par with un-NPs and du/des-NPs examined in Chapter 2 and 3 respectively, are not intrinsically existential. That bare nouns can only have narrow scope in English is illustrated in (5b). This is an important distinction between bare plurals and singular indefinite noun phrases.

(5) a. Everyone read a book on caterpillars. b. Everyone read books on caterpillars.  (Dobrovie-Sorin & Beyssade 2004: 38 (10))

(5a), contrary to (5b), is ambiguous in that a book on caterpillars can either take scope on everyone or be in its scope. Books on caterpillars in contrast can only have the latter reading, i.e., narrow scope (Chapter 2 Section 2.3.2). Additional properties differentiate bare nouns from singular indefinite noun phrases (Carlson 1977a,b, see Dobrovie-Sorin & Beyssade 2004, Sections 1.1 and 1.2 for a nice overview). Consider the following examples:

(6) a. John recommended some / certain / three books and Mary recommended others / the others. b. *John recommended books and Mary recommended others.  (ibid, 30 (15a), (16a))



(7) a. Max discovered two rabbits in his yard  in two hours / ??for two hours. b. Max discovered rabbits in his yard   ??in two hours / for two hours.  (ibid, 30 (12a,b))

Bare nouns, in contrast with singular indefinites, cannot function as an antecedent for anaphors of the type others/the others as illustrated in (6). In addition, they have an impact on the aspectual properties of the predicate in such a way that it has an unbounded interpretation (7). That (7b) has an atelic reading is supported by the grammaticality of the time adjunct for two hours (Dowty 1979; Dobrovie-Sorin & Laca 2003; De Swart 2006; Guéron 2006). In two hours, in contrast is fine in bounded/telic contexts (7a) (unboundedness is discussed in more details in Chapter 5 Section 2.1). A last distinction between singular indefinites and bare nouns worth mentioning is that the latter, contrary to the former, cannot have a transparent reading with wish/ believe-type of verbs. Consider the following example:

(8) a. Minnie wishes to talk with a young psychiatrist. b. Minnie wishes to talk with young psychiatrists.

(ibid, 28 (8))

In (8b), young psychiatrists cannot refer to some particular psychiatrists, contrary to a young psychiatrist in (8a). In other words, bare nouns only have an opaque reading and cannot be transparent, contrary to singular indefinites which are ambiguous between the two readings.

 The Layered DP: Form and meaning of French indefinites

In sum, this section describes bare nouns as property-denoting nominals, which may be existential or not, which take narrow scope, which are ungrammatical as antecedents of others/the others, which lead to an unbounded reading and which cannot be interpreted as transparent. The next section deals with French which is traditionally considered as a language without bare nouns (Chierchia 1998b; Longobardi 1994, 2001), although this position has been challenged recently, namely by Roodenburg (2004a, 2004b) (see also Ihsane 2003c).

2.2  French The traditional position in the literature is that French does not have any bare nouns (Chierchia 1998b; Longobardi 1994). For example, the French counterparts of (9), are ungrammatical as shown in (10):

(9) a. b. c. d. 

Students have occupied the building. I saw students in the building. Studenten hebben het gebouw bezet. Ik heb studenten in het gebouw gezien

(10) a. *Etudiants ont occupé l’édifice. b. *J’ai vu étudiants dans l’édifice. 

(English) (Dutch) (Delfitto & Schroten 1991: 155(2)) (French) (Delfitto & Schroten 1991: 156(4))

To save the examples in (10), the bare nouns have to be replaced by des-NPs. In the same vein, the French versions of (5)–(8) involve des-NPs and are ungrammatical with bare nouns: (11) a. Tout le monde a lu everyone has read des livres sur les chenilles. of.the books on the caterpillars * livres sur les chenilles. books on the caterpillars b. Jean a recommandé Jean has recommended des livres et Marie en a recommandé d’autres. of.the books and Marie EN has recommended others * livres et Marie en a recommandé d’autres. books and Marie EN has recommended others c. Max a découvert des lapins / * lapins dans son jardin Max has discovered of.the rabbits / rabbits in his garden ?? en deux heures / pendant deux heures. in two hours / for two hours



Chapter 4.  Bare nouns 

d.

Minnie souhaite parler à Minnie wishes to.talk to des psychiatres /* psychiatres. of.the psychiatrists / psychiatrists

The ungrammaticality of examples like (10) and (11) has lead to the conclusion that French doesn’t have bare nouns. This is however only partially correct as nominal expressions without determiners but functioning as arguments are possible in different contexts in French. One of them is idioms in which bare nouns may function as subjects (12), (14) or objects (13). (12) a. b.

A toute chose, malheur est bon. to all thing bad.luck is good ‘It’s an ill wind that blows nobody any good.’ Pauvreté n’ est pas vice. poverty neg is not vice ‘Poverty is not a crime.’

(13) a. b.

Jamais honteux n’ eut belle amie. never ashamed neg had pretty friend ‘Faint heart never won fair lady.’ Mettre cartes sur table put cards on table ‘To put one’s cards on the table’

(14) a. b.

Mauvaise herbe croît toujours. bad grass grows always ‘Ill weeds grow apace.’ Chien qui aboie ne mord pas. dog that barks neg bites not ‘Its bark is worse than its bite’/ ‘A barking dog never bites.’

Why the determiner may be omitted in the above examples in contrast with (10) and (11) remains unaccounted for. In effect, except for the absence of determiner, nothing seems to distinguish the bare nouns illustrated in (12)–(14) from other nominal expressions in French: they may have number morphology (13b), be modified by agreeing adjectives (13a), (14a) or relative clauses (14b). What might be argued is that the omission of determiners is specific to idioms, which are lexical items learned as chunks part of our vocabulary. Put differently, idioms would represent a register in which the rules determining grammaticality are less strict than the ones of standard French, or at least different in a sense to be determined. Other expressions which are more or less frozen contain bare complements. This is frequent with faire ‘do/make’ (15a) and avoir ‘have’ (15b). (15) a.

faire peur, mal, to.do fear, harm,

envie, pitié envy, pity

 The Layered DP: Form and meaning of French indefinites

b. avoir peur, mal, envie, pitié, faim, soif, congé, tort to.have fear, harm, envy, pity, hunger, thirst, holiday, fault

The fact that the above examples represent rather fixed expressions suggests that these data are out of the scope of this chapter, which concentrates on bare nouns analogous to du/des-NPs (Section 1). However, bare nouns are also possible in expressions which are not frozen. This is the case of existential constructions (16) and coordination (17)–(18). (16) a. b. 

Va au lit de bonne heure go to.the bed early y a école demain. there has school tomorrow ‘Go to bed early, you have school tomorrow.’ A Talensac, y a marché tous les jours. at Talensac there has market all the days ‘At Talensac the market is open every day.’ (Mathieu 2006, (30))

(17) a. b.

Garçons et filles sont venus. boys and girls are come ‘Boys and girls have come.’ Mère et fille se ressemblent. mother and daughter REFL resemble ‘Mother and daughter resemble each other.’

(18) a. b. 

On a servi vin et champagne one has served wine and champagne jusque tard dans la soirée. until late in the evening On a servi moules et crevettes one has served mussels and shrimps jusque tard dans la soirée. until late in the evening (Roodenburg 2004b:122(8–9), translation added)

As the above examples show, coordinated bare nouns can be subjects (17) or objects (18), can contain plural (17a), (18b) or singular nouns (17b), (18a), which can be count (17b) or mass (18a). However, as most examples involving mass and plural bare nouns represent coordination, French will be further discussed in the section dealing with this construction (Section 4). To exclude French from the study of non-coordinated bare nouns is also justified by the fact that the data are limited and often difficult to describe from an interpretative point of view.



Chapter 4.  Bare nouns 

3.  The structure 3.1  The left periphery A key position adopted here about the structure of nominal expressions is that to be arguments they need to have the DP layer projected (Szabolcsi 1983; Stowell 1989; Longobardi 1994, 2005). This distinction between arguments and non-arguments is partly based on bare nouns (Chapter 2 Section 3.1.1). In other words, argumental bare nouns have to be DPs. In the analysis developed in this book, where DP is split into several discrete layers, this means that these nominal expressions minimally involve one projection of the nominal left periphery, as stated in (19) (Chapter 2 Section 3.1.1). (19) A “nominal expression” is an argument only if it is introduced by one of the categories of the DP domain (the left periphery).

More precisely, the idea is that argumental bare nouns have to involve the lowest projection of the left periphery as it turns predicative noun phrases into property-denoting arguments. As this projection is characteristic of noun phrases interpreted as properties, it is labelled Property Phrase (PropP) (Chapter 2 Section 3.2). This implies that the bare nouns discussed in Section 2.1 are Property Phrases, i.e., PropPs. More precisely, bare nouns in Germanic and Romance languages, like biscuits and biscotti in (20), are PropPs. The fact that they are existential (Section 2.1) doesn’t have an impact on their structure (Chapter 2 Section 2.3.2). (20) a. John was eating biscuits. b. Gianni mangiava biscotti.

(French) (Italian)

If biscuits and biscotti in (20) are Property Phrases, one question which arises is whether this projection is empty or not, and if it is not what it hosts. As in the analysis developed here the layers of the nominal left periphery are legitimated by movement, we expect PropP to be filled. However, a distinction between Germanic and Romance languages is in order here. It is based on the different distributions of bare nouns in these languages. Consider the data below: (21) a. b. c. d.  (22) a. b. c. d. 

Students have occupied the building. I saw students in the building. Studenten hebben het gebouw bezet. Ik heb studenten in het gebouw gezien *Estudiantes han ocupado el edificio. Yo he visto estudiantes en el edificio. *Studenti hanno occupato l’edificio. Io ho visto studenti nell’edificio.

(English) (Dutch) (Delfitto & Schroten 1991: 155(2)) (Spanish) (Italian) (Delfitto & Schroten 1991: 155(3))

 The Layered DP: Form and meaning of French indefinites

In Germanic, bare nouns are grammatical both in subject and object positions (21), whereas in Romance, they are excluded from preverbal subject positions (22).6 To account for these data, Longobardi (1994) proposes a parametric difference. The idea is that Romance languages have a null determiner in D whereas Germanic languages have the common noun moving to D at LF. The presence of a null D in Italian accounts for the ungrammaticality of bare nouns in pre-verbal subject positions (22): these bare nouns must be lexically governed to be grammatical, contrary to bare nouns in Germanic which do not contain a empty category (21). This parameter is made explicit in (23): (23)

Bare nouns a. Romance languages: non-overt D subject to lexical government b. Germanic languages: N movement to D at LF

In the framework adopted here, where DP may be split into several discrete projections, the above parameter may be reformulated as in (24):7 (24) Bare nouns (bis) a. Romance languages: non-overt determiner in Prop, subject to lexical government b. Germanic languages: N movement to Prop at LF

The idea is that the lowest projection of the left periphery, i.e., PropP, is filled, either by a non-overt element (ec) or by the noun which moves from its base position, N. The corresponding structures are reported below: (25)

(Romance)

PropP Spec

Prop′ Prop ec

NumP studenti

.  This is a simplification. The judgements depend on the kind of predicate and subject involved. In English, for example, bare nouns are excluded from the subject position of individuallevel predicates (i), and in some dialects as subjects of stage-level predicates (ii) (see Guéron 2006 for more details).

(i) * Men are blond. (ii) a. * Doctors are sad. b. * Children are hungry. .  The interpretative properties attributed by Longobardi to the empty D will be ignored here.



Chapter 4.  Bare nouns 

(26) Spec

Prop′

Prop studentsi



(Germanic)

PropP



NumP ti



As illustrated in (25) and (26) respectively, the ‘bare noun’ studenti in Italian comprises an empty category (ec) in its left periphery, in contrast with students in English. In other words, the latter involves N-movement contrary to the former. Although the content of Prop slightly varies in Romance and Germanic, bare nouns in these languages are interpreted as properties, i.e., they are not ambiguous (excluding their potential generic reading from the discussion; recall Section 2.1). In terms of structure, this implies that bare nouns can only be Property Phrases and that their structure cannot project further. Put differently, they cannot be Quantifier Phrases (QPs) or Speaker’s Reference Phrases (SRefPs) in contrast with un-NPs or du/ des-NPs. This is because in the framework developed here ambiguity is due to different structures, that is, each interpretation is associated to a specific structure. Interestingly, assuming that the categorical status of bare nouns can only be PropP as proposed here accounts for the fact that these nominals don’t have scope properties (Section 2.1). One of McNally’s (1995) examples illustrating this point is repeated as (27): (27) a. b. c. d.

A la reunión no asistieron algunos profesores. at the meeting not attended some.pl professors ‘The meeting was not attended by some professors.’ A la reunión no asistieron profesores. (∃x: professor(x)) [¬attend(x,m)] ¬(∃x: professor(x)) [attend(x,m)]

(c or d)

(d only)

Recall that in (27b), the bare noun profesores ‘professors’ cannot take scope over the negation in contrast with algunos profesores ‘some professors’. This is expected if it is the QP layer which renders a noun phrase visible to Quantifier Raising (QR) (Chapter 2 Section 3.3): as profesores ‘professors’ is a Property Phrase, it does not comprise a QP layer in its structure and therefore cannot undergo QR. If bare nouns do not have a QP layer in their structure, it further predicts that they should not be able to have intermediate scope, a characteristic attributed to Sreferential noun phrases (Chapter 2 Section 3.4.3). This is because the left periphery of SRefPs is composed of a QP layer which is visible to QR. That this prediction is borne out is illustrated in (28): (28) Every professor rewarded every student who read books he recommended. √ narrowest, *widest, ??intermediate (Borer 2005a: 147 (23b))

 The Layered DP: Form and meaning of French indefinites

In the above example, the intermediate reading of the bare noun is very difficult to ­obtain and the widest scope reading completely out, which seems to support the analysis developed here. This contrasts with des-NPs often compared with bare nouns. The example below is repeated from Chapter 3 Section 3.1.3: (29) a. b.

Tous les étudiants ont raconté plusieurs histoires all the students have told several stories qui impliquaient des membres de la famille royale. that involved of.the members of the family royal ‘All the students told several stories that involved some members of the Royal Family.’ tous les étudiants > des membres de la famille royale > plusieurs histoires

In (29), des membres de la famille royale ‘members of the royal family’ can take scope over plusieurs histoires ‘several stories’ and be in the scope of tous les étudiants ‘every student’. In sum, the analysis developed here and the structures proposed for bare nouns account for two of their characteristics, namely their potential existentiality and more importantly their lack of scope properties. Were they QPs or SRefPs, they would have scope properties as both categories comprise a QP layer in their left periphery. The next section deals with the inflectional domain of bare nouns.

3.2  The inflectional domain Recall from Chapter 3 Section 3.2, that the inflectional domain advocated for the nominals examined may comprise the following projections, where FP stands for Functional Projection: (30) Number Quantity Count Lexical category FP3 FP2 FP1 NP

In a nutshell, the projection of FP1 leads to an atomised reading, FP2 is projected when some quantity is encoded in the nominal expression (and not inferred from the context) and FP3 encodes grammatical number. Let us begin our discussion of bare nouns with the lowest functional projection, FP1. A well-known characteristic of bare nouns is that they fail to have individuated reference by themselves (Bosveld-de Smet 1998; Borer 2005a; Dobrovie-Sorin & ­Beyssade 2004; Guéron 2006).8 This is illustrated in (31): .  According to Bosveld-de Smet, individuation can be imposed upon bare nouns (bare ­plurals depend on contextual and pragmatic factors as to the individuation of their reference 1998: 146). In this case they can have wide scope (1998: 152). One example of a bare noun with wide scope is provided below:

(i) There was a bomb alarm this morning in our apartment building. All day policemen searched every apartment. (Bosveld-de Smet 1998: 149(91))



Chapter 4.  Bare nouns 

(31) a. Kim ate apples this afternoon (for an hour) (*in an hour). b. Pat built houses (all summer) (*in two months). c. My kid sister drew circles (all morning) (*in half an hour).  (Borer 2005a: 120 (41))

Concerning the above examples, Borer observes that the interpretation of the bare plurals in [(31)] is consistent with an interpretation whereby no single complete apple has been eaten (but, say, a number of chunks from assorted apples, perhaps no more than one), no whole house was built (by Pat or anybody else), and my kid sister’s drawing gave rise to assorted arcs and incomplete circles, none of which actually qualifies as a circle (2005a: 120–1).

The ungrammaticality of time expressions like in an hour/in two months/in half an hour in (31) is traditionally taken to support the lack of individuated reference of bare nouns (Borer 2005a; Guéron 2006). The idea is that bare nouns lead to an unbounded reading from an aspectual point of view whereas in-adverbials are associated to bounded/telic readings (Chapter 3 Section 3.2.1). The correlation between unbounded reading and lack of individuated reference is however not straightforward. Put differently, it is not obvious that it is the lack of individuated reference which leads to an unbounded reading (recall for example our discussion of des-NPs in Chapter 3 Section 3.2.1: although these noun phrases have individuated reference, they may lead to an unbounded reading or not; see also (34) below). As this issue concerns the interaction between predicates and nominals at the clausal level, it will be postponed until Chapter 5 Section 2.1. That bare nouns fail to have individuated reference is supported by the fact that they cannot function as an antecedent for anaphors of the type others/the others (32) (Section 2.1): (32) a. John recommended some / certain / three books and Mary recommended others / the others. b. *John recommended books and Mary recommended others.

Anaphors like others/the others have to occur with nominals with individuated reference. The ungrammaticality of (32b) shows that books doesn’t have this interpretation. What this means for the extension of bare nouns can be illustrated by means of a join semilattice.

Although the preferred reading of the above example is the one where every apartment has scope over policemen, it seems that the reverse is not impossible. Other authors attribute scope properties to bare nouns, on a par with other indefinite noun phrases (see for example Kratzer 1980; Gillon 1990 and Link 1991).

 The Layered DP: Form and meaning of French indefinites

Recall that join semilattices represent sets and that they are used to represent the extension of mass and plural nouns (Chapter 3 Section 3.2.1): (33) (based on Link 1983) iii. {a, b, c, d} ii. {a,b} {a,c} {a,d} {a,b,c} {a,b,d} {a,c,d} {b,c} {b,d} {b,c,d} {c,d} i. a b c d

The elements a, b, c and d on line (i) represent the atoms, i.e., the minimal elements of the set. Line (ii) represents pluralities, i.e., sets which contain more than one element, and line (iii) the set of all the atoms that exist in line (i). As bare nouns fail to have individuated reference, it suggests that in a join semilattice like (33) the bottom line (i) containing atoms is absent. Put differently, bare nouns don’t have atoms in their extension. The absence of atoms in the extension of bare nouns has an impact on their structure: it means that FP1 in (30) is not projected in their inflectional domain. In effect, in the analysis developed here, individuated reference is associated to the count reading, which is encoded in the functional projection labelled FP1 (Chapter 3 Section 3.2). In other words, if bare nouns do not have atoms in their extension, they cannot be count and therefore do not have FP1 in their structure. This means that bare nouns are closer to du-NPs than to des-NP, at least for this aspect of their interpretation. In effect, one conclusion reached in Chapter 3 is that des-NPs have individuated reference, contrary to du-NPs (Chapter 3 Section 3.2.1). Put differently, bare nouns and du-NPs do not have FP1 in their structure contrary to des-NPs. This is an interesting observation as des-NPs in French are often compared to bare nouns (Chapter 3 Section 1.2). That bare nouns and des-NPs should however be distinguished is supported by (34) and (35), which contrast with (31) and (32b) respectively:9 (34) *J’ ai reconnu des filles dans la rue I have recognised of.the girls in the street pendant 10 minutes. for 10 minutes (35) Des

(Chapter 3, Section 3.2)

filles étaient blondes, d’autres avaient les

of.the girls were blond others had cheveux foncés. hair dark

the (Chapter 3, Section 2.2)

.  Another difference between des-NPs and bare nouns is that the former can take widest scope, contrary to the latter (Section 2.1). In (i), the des-NPs may scope over the universal.

(i)

a. b. 

Chaque dimanche, des amis viennent nous visiter. every Sunday, of.the friends come visiting us Des enfants viennent jouer ici tous les jours. of.the children come playing here all the days (Bosveld-de Smet 1998: 38 (78b), (79b)) (Chapter 3 Section 2.1 (34))



Chapter 4.  Bare nouns 

The ungrammaticality of the for- time adverb in (34) shows that des-NPs do not ­necessarily lead to an atelic interpretation and (35) shows that des-NPs can be the antecedent of d’autres ‘others’. Both characteristics distinguish des-NPs from bare nouns. That bare nouns do not have atoms in their extension, i.e., no FP1 in their structure, might seem obvious for bare mass nouns. However, it is more controversial for bare plurals. The reason is that we have assumed that the plural –s, at least in the French data examined in Chapter 3, leads to a count reading and is generated in FP1. This account however cannot apply to bare plurals in English as these nominals do not have individuated reference despite the presence of the plural –s (see the references provided above). To solve this puzzle I suggest that the –s on bare nouns differs from the ‘traditional’ plural morpheme which leads to a count reading and therefore that it is generated in another projection (see Borer 2005a for a different account). More precisely, I propose that it sits in the projection associated to grammatical number, labelled FP3 in (30). This is illustrated in the example below, where X means that the layer is not projected in the structure: (36)

Example Number FP3 a. books –s b. bread √

Quantity FP2 X X

Count FP1 X X

Lexical category NP book bread

(36a) implies that the noun moves (at least) to FP3 to bind the –s sitting in this projection. From this position, it may reach the left periphery, more precisely PropP, to become an argument (Section 3.1). As bare mass nouns are singular, I assume that the listeme also moves to FP3 before reaching the left periphery to form an argument (36b). As for bare nouns in Romance, the idea is that the listemes move to FP3 to account for their grammatical number and that they remain in this position because Prop contains an empty category (recall the parameter proposed by Longobardi (1994) to distinguish Germanic from Romance; Section 3.1). In other words, argumental bare nouns in Germanic involve N-movement to Prop through FP3, whereas in Romance the noun stops in FP3 because Prop is filled. As for FP2, I assume that it is not projected in the structure of bare nouns. This is because these nominals do not encode any quantity (Chapter 3 Section 3.2.1). The resulting structures are represented below: (37) a. b. c.

Property PropP books bread ec

Number FP3 book + s bread studenti

Quantity FP2 X X X

Count FP1 X X X

Lexical category NP book bread student

Note that the structure proposed for bare mass nouns in (37b) is analogous to the one of non-quantitative du-NPs, except that the components of du, namely de and le, are absent. Consider (38) which takes into consideration du/des-NPs to ease comparison:

 The Layered DP: Form and meaning of French indefinites

(38) Example a. b. c. d. e. f.

du painquantit. du painnon-quantit. des painsquantit. des painsnon-quantit. books bread

Number FP3

Quant. Count Lexical category FP2 deP FP1 NP

de + lesing de + lesing de + lesplur de + lesplur -s √

de de X de de de X de X X

X X –s –s X X

pain pain pain pain book bread

(38) allows us to visualise the fact that bare nouns, on a par with du-NPs but in contrast with des-NPs, do not have atoms in their extension, i.e., they fail to have individuated reference. Another characteristic shared by bare nouns and du-NPs but not by des-NPs is that neither can be S-referential as defined in Chapter 2 Section 2.1.2 (see Chapter 3 Section 2.1 for du/des-NPs and Section 2.1 above for bare nouns). That a correlation exists between the two properties seems unsurprising: individuals/entities can only be identified or referred to if the extension of the nominal contains atoms/individuals. In other words, an S-referential reading is directly dependant on the presence of minimal parts in the extension of the nominal expression. As du-NPs and bare nouns do not have atoms in their extension, they cannot be S-referential. What might be more ­innovative is to formalise this in terms of structure. The above discussion suggests that the SRefP layer can only be projected if FP1 is present in the structure. In other words, to be SRefPs, nominal expressions have to comprise FP1. Although this might seem trivial, it is important to notice that, conversely, the projection of FP1 does not necessarily lead to an S-referential reading (recall our discussion of des-NPs in Chapter 3). More generally, what this suggests is that the constitution of the inflectional domain has an impact on the composition of the left periphery. This section shows that the analysis developed in this book extends perfectly to bare nouns. Whether this is the case of coordinated bare nouns is examined in the next section.

4.  Coordinated bare nouns Coordination has been focussed on in recent literature (Heycock & Zamparelli 2003, 2005; Roodenburg 2004a, 2004b). As the analyses for coordinated bare singulars (N&N) and coordinated bare plurals (Ns&Ns) differ, these noun phrases are at first studied separately. Coordinated bare nouns are considered as a whole throughout, i.e., the internal structure of each conjunct is ignored. French is included in the analysis of Romance languages as this language has coordinated bare nouns (recall (17)–(18) in Section 2.2).



Chapter 4.  Bare nouns 

4.1  The interpretation 4.1.1  Singulars This section concentrates on examples of coordination involving singular conjuncts, i.e., N&N. The aim is to show that these bare nouns differ from non-coordinated bare nouns (Section 2). Heycock & Zamparelli (2003) for example claim that coordinated bare singulars have a ‘definite’ or ‘quasi-definite’ interpretation. Consider the following example: (39) a. At the company meeting, president and vice president gave an optimistic speech. b. ?? At the company meeting, employee and inspector talked about their colleagues’ motivation. (Heycock & Zamparelli 2003: 445 (9))

What Heycock and Zamparelli suggest is that (39a) is grammatical because companies generally only have one president and one vice president. (39b) on the other hand is weird because companies usually have more than one employee and more than one inspector. In other words, a uniqueness condition determines the grammaticality of coordinated bare singulars. It is this uniqueness condition which leads the authors to consider coordinated bare singulars as definite. Another source of definite reading is anaphoricity, illustrated in the following example: (40) A black cat and a brown dog were fighting in the street. [Cat and dog] were equally filthy. (Heycock & Zamparelli 2003: 443 (1))

In (40), the cat and the dog have been introduced in the context. They are therefore anaphoric. If the referents of the conjuncts have been preidentified, they are analogous to definite noun phrases, a conclusion supported by other contexts examined by Heycock and Zamparelli (the reader is referred to the original paper for further details). According to the authors, the ‘definite’ reading is the only interpretation of coordinated bare singulars, in contrast with coordinated bare plurals examined in the next section. If this is on the right track, it suggests that coordinated bare singulars are out of the scope of this book which concerns indefinites and not definites. However, as coordinated bare singulars also differ from definite noun phrases in several aspects, it is worth pursuing the discussion. Consider the following data in which coordinated bare singulars are impossible, in contrast with definite noun phrases. (41) a. *I didn’t see much of the film: man and woman in front of me were very tall. b. *After taking my hand, the doctor examined nail and finger that had become infected. c.    ?*Cat and dog that were fighting all the time had to be separated; the other animals stayed together.

 The Layered DP: Form and meaning of French indefinites

d. ?*Only boys and girls that you met yesterday know Italian. e. ?*There are four tables and a dozen chairs in stock; others can be ordered. Tables and chairs currently available are made of cherry.  (Heycock & Zamparelli 2003: 450 (30)) (42) a. *[Dog that I saw] and [cat that I heard] fight all the time. b.  ??[Knife with the golden blade] and [fork with the silver handle] go on the left. (ibid, 451 (37)) (43) All the pet shop owner had in stock was ten turtles, one dog, two snakes, and one Siamese cat; he had sold all the rest. a. *Only cat and only dog were both small, shaggy animals. b. *Biggest turtle and smaller snake cost the same. (ibid, 451 (38))

In (41a), the two conjuncts are modified together by a PP. In (41b–d), they are modified by a relative with subject or object gap and in (41e) by a reduced relative. Modification by a relative clause or a PP in each conjunct is not possible either (42). Finally, as shown in (43), bare coordinated singulars are ruled out of contexts where the combines with only or with a superlative. When the definite article the is inserted, the examples in (41)–(43) all become grammatical. This suggests that the analogy between coordinated bare singulars and definite noun phrases only partially holds, hence the term ‘quasi-definite’ used by Heycock and Zamparelli. That coordinated bare singulars have only one interpretation as claimed by ­Heycock & Zamparelli (2003) is not uncontroversial. According to Roodenburg (2004b) for example they may be ambiguous between a strong and a weak reading. The examples he provides are reported below. (44) On servait bière et vin jusque tard dans la nuit. ‘They served beer and wine until late in the evening.’  (Roodenburg 2004b:311 (24)) (45) At the company meeting, president and vice president gave an optimistic speech.

According to Roodenburg, (44) represents the weak reading and (45), which is one of Heycock and Zamparelli’s ‘quasi-definite’ examples (39a), the strong one. Interestingly, however, such a classification does not necessarily conflict with Heycock & Zamparelli’s (2003). In effect, if one adopts Milsark’s (1974) definition of weak-strong noun phrases, definite nominals are strong. Recall (46) repeated from Chapter 1 Section 5.1.1 for convenience: (46) a. a/an, some, several, many, numerals (one, two, three) (= weak) b. definite descriptions, demonstratives, possessives, personal pronouns, generics, universal quantifiers (all, every, each), most (= strong)



Chapter 4.  Bare nouns 

However, as the above classification has proven problematic for several reasons (see for example Chapter 2 Sections 2.2 and 3.1.2), it has not been adopted here. The rest of the section is therefore devoted to the analysis developed in this book. The aim is to determine whether the two interpretations attributed above to coordinated bare singulars can be assimilated to the readings characterising du/des-NPs. In the terms adopted here, the weak reading corresponds to the property interpretation (Chapter 2 Section 2.3) and the strong reading to the covert partitive/quantificational interpretation (Chapter 2 Section 2.2). In other words, bière et vin ‘beer and wine’ in (44) is property-denoting and president and vice president in (45) is quantificational. Although it seems unproblematic to attribute a property interpretation to bière et vin, qualifying president and vice president as a covert partitive is less straightforward. However, it could be argued that a ‘company meeting’ is certainly attended by more than two ­people. In other words, president and vice president represent a subset of the people attending the meeting, which represent the set. The question which arises is how this is compatible with the impression that president and vice president in (45) is quasi-definite, as observed by Heycock and Zamparelli. Recall that the authors propose that there is a uniqueness condition at play which distinguishes (45) from (47) (= (39b)). (47) ?? At the company meeting, employee and inspector talked about their colleagues’ motivation.

The puzzle thus consists in accounting for the uniqueness condition of coordinated bare singulars like president and vice president in (45) and for their potential covert partitive reading. These two properties are however not necessarily in conflict if one assumes that the former concerns each conjunct, i.e., [president] on the one hand and [vice president] on the other hand, and that the latter concerns the conjunction as a whole, i.e., [president and vice president]. In other words, the president and the vice president have to be unique and this set of two persons represents a subset of the people attending the company meeting. If this is on the right track, analysing coordinated bare singulars as property-denoting or quantificational as proposed might be possible. As the discussion is highly speculative, it will however not be pursued. Although the uniqueness condition involved by some coordinated bare singulars can be accommodated in our analysis, their potential anaphoricity is more problematic. Recall that anaphoricity is another property of coordinated bare singulars which may lead to a ‘quasi’-definite reading. (40) is repeated below for convenience. (48) A black cat and a brown dog were fighting in the street. [Cat and dog] were equally filthy. (Heycock & Zamparelli 2003: 443 (1))

In (48) the cat and the dog are anaphoric in the sense that they have been introduced in the context. However, in contrast with president and vice president in (45), cat and dog doesn’t represent a subset of a set already mentioned. Rather it represents the set.

 The Layered DP: Form and meaning of French indefinites

In other words, cat and dog doesn’t have a covert partitive reading, i.e., it is not quantificational. Furthermore, cat and dog doesn’t have a property-denoting reading. In effect, this interpretation would imply that the cat and the dog cannot be identified contrary to fact: the referents of the conjuncts have been preidentified. Put differently, it seems that coordinated bare singulars may have a referential reading. Their reference however is not dependent as defined in Chapter 2: it does not depend on a universal quantifier nor on the speaker (the cat and the dog are not ‘in the speaker’s mind’). In other terms, the definition of S-referentiality adopted here doesn’t seem to apply to the reference of cat and dog in (48). This implies that anaphoric coordinated bare singulars fall out of the scope of this chapter which is limited to mass and plural bare nouns analogous to du/des-NPs. This conclusion is supported by the fact that cat and dog in (48) mean le chat et le chien ‘the cat and the dog’ (although the determiners could be omitted). But if the French translation has to be taken into account in the above example, it should also be in (45). This in turn suggests that president and vice president, which is interpreted as le président et le vice président, should also be excluded from our analysis after all. More generally, the conclusion which emerges from the discussion seems to be that coordinated bare count nouns cannot be treated on a par with the bare nouns examined here and hence that they should be studied independently. Although this section is not conclusive, it shows that coordinated bare singulars cannot be treated uniformly. When they involve bare mass nouns however, they seem to fit the analysis developed here in that they are property-denoting. Whether this is also the case of coordinated bare plurals is addressed in the next section.

4.1.2  Plurals In their analysis of Romance and Germanic languages, Heycock & Zamparelli (2003) suggest that coordinated bare singulars and coordinated bare plurals should be distinguished. According to them, coordinated bare singulars have only one reading, namely the ‘quasi-definite’ interpretation reported in the preceding section, whereas coordinated bare plurals may be two-way ambiguous: in addition to the ‘definite’ reading shared with coordinated bare singulars, coordinated bare plurals can be interpreted like existential bare plurals. The former reading is illustrated in (49) and the latter in (50). (49) We had to set the table for the queen. We arranged one crystal gobleth, one silver spooni, two antique gold forksj and two platinum knivesk. [Forksj and (Heycock & Zamparelli 2003: 448 (24)) knivesk] were equally dirty. (50) Giornata di mercato in città. market day in town [Clienti ?? (e curiosi)] gironzolavano per i banchi. customers (and onlookers) walked about the stands  (Heycock & Zamparelli 2003: 447 (22a))



Chapter 4.  Bare nouns 

In Heycock and Zamparelli’s terms, the coordinated bare plural in (49) is anaphoric, i.e., their referents have been introduced in the linguistic context. In (50), it is existential, in the sense that it introduces a new referent in the discourse. Roodenburg (2004b) in contrast proposes that coordinated bare plurals in French are analogous to coordinated bare singulars in that they are ambiguous between a weak and a strong reading (Section 4.1.1). The former is illustrated in (51) and the latter in (52).10 (51) L’inspecteur Williams se rendit dans cette chambre de bonne sans avoir aucune idée de ce qu’il allait y découvrir. Il eut un peu de peine à trouver l’interrupteur. Il n’y avait pratiquement aucun meuble dans la pièce, mise à part une table. Livres et revues jonchaient le plancher. ‘Inspector Williams went to this maid room without having any idea of what he was going to discover there. He had a little trouble to find the switch. There was practically no furniture in the room, except for a table. Books and revues were lying on the floor.’ (Roodenburg 2004a:123(10)) (52) En général, les members de ce département assistent à toutes les réunions. Mais tous ne sont pas toujours là. Ainsi thésards et Aters ont séché la réunion d’hier. ‘In general, members of this department always attend all the meetings. But not everyone is always there. So graduate students and assistant professors stayed away from yesterday’s meeting.’ (Roodenburg 2004b: 309 (23))

In (51), the books and revues are introduced in the context for the first time. They have an existential reading in the sense of assertion of existence. In (52), in contrast, the graduate students and assistant professors represent a subset of the previously introduced set les members de ce département ‘the members of this department’.11 In the framework developed in this book, these two readings correspond to the property­denoting and the quantificational interpretation respectively. In other words, livres et revues in (51) has a property reading whereas thésards et Aters in (52) is quantificational. What about the examples mentioned by Heycock and Zamparelli and reported in (49) and (50)? The existential interpretation of clienti e curiosi ‘customers and onlookers’ in (50) perfectly fits the definition of property-denoting noun phrases adopted here. As for the anaphoric interpretation of forks and knives in (49), it raises the same questions as the anaphoric reading of the coordinated bare singulars mentioned in the

.  G. Giusti (p.c.) observes that (51) is anomalous if the room is a butcher’s or a barber’s shop, which would suggest that books and journals are somehow salient to the objects of a room. .  Note that thésards et Aters in (52) might be translated by les thésards et les Aters or des thésards et des Aters.

 The Layered DP: Form and meaning of French indefinites

preceding section: the kind of reference involved seems to differ from the one associated with the indefinites discussed here. It will therefore not be further addressed. This section suggests that coordinated bare plurals may have a quantificational reading in addition to the property interpretation they share with coordinated bare singulars. When anaphoric, they may also be referential. However, as this kind of reference is not directly related to the topic of this book, it will be ignored in the rest of the discussion. Further research is also required to understand to what extent the meaning of coordinated bare plurals depends on each conjunct or rather on the coordination as a whole. Whether the analysis developed here for the structure of du/desNPs may be extended to coordinated bare nouns is examined in the next section. More precisely, Section 4.2.1 deals with the categorical status of bare nouns and Section 4.2.2 with their inflectional domain.

4.2  The structure 4.2.1  The left periphery In the preceding section, we have seen that coordinated bare nouns may be interpreted as property-denoting or quantificational, at least when they are plural. Extending the analysis developed in this book to these nominal expressions means that they are Property Phrases (PropPs) in the former case and Quantifier Phrases (QPs) in the latter. More precisely, it suggests that coordinated bare singulars are PropPs, whereas coordinated bare plurals may be PropPs or QPs. Let us examine the property interpretation first: (53) On servait bière et vin jusque tard dans la nuit. ‘They served beer and wine until late in the evening.’  (Roodenburg 2004b: 311 (24)) (54) Giornata di mercato in città. market day in town [Clienti ?? (e curiosi)] gironzolavano per i banchi. customers (and onlookers) walked about the stands  (Heycock & Zamparelli 2003: 447 (22a))

In our framework, nominals interpreted as properties are Property Phrases. This is the case of bière et vin ‘beer and wine’ and clienti e curiosi ‘customers and onlookers’ in the above examples (Section 4.1). An interesting difference between these coordinated bare nouns and non-coordinated bare nouns, however, is that the former are grammatical in the subject position of a language like Italian (54). This contrasts with non-coordinated bare subjects which are not possible in this language, as seen in Section 3.1. The modified version of Longobardi’s parameter



Chapter 4.  Bare nouns 

accounting for the ungrammaticality of non-coordinated bare subjects is repeated as (55) for convenience: (55) Bare nouns (bis) a. Romance languages: non-overt determiner in Prop, subject to lexical government b. Germanic languages: N movement to Prop at LF

The fact that clienti e curiosi ‘customers and onlookers’ may function as subjects suggests that when they are coordinated, bare nouns in Italian are not concerned by the ­parameter proposed by Longobardi (1994) to account for the distributional contrast between bare nouns in Germanic and Romance. More precisely, it suggests that coordinated bare nouns in Italian do not contain an empty category (ec) in their structure, contrary to non-coordinated bare nouns (recall that this ec has to be lexically governed). If this is correct, it raises the question of the legitimating of the Prop layer in the structure of coordinated bare nouns in Italian. What naturally comes to mind is to assume that these nominals involve some movement to PropP at LF, on a par with non-coordinated bare nouns in Germanic (55b). However, as the coordination is a maximal projection, it is likely that the movement taking place is an instantiation of XP-movement, rather than head-movement. In other words, the coordination moves to Spec,PropP as illustrated below (the base position of this constituent will be ignored): (56)

PropP Spec N&Ni Ns&Nsi





Prop′ Prop

NumP ti

(56) represents the (partial) structure proposed for coordinated bare nouns interpreted as properties: they are PropPs and the coordination, represented by N&N when it is singular and by Ns&Ns when it is plural, has moved to Spec,PropP as a maximal projection. This means that these nominals have a left periphery comprising a single projection. Extending the analysis to quantificational coordinated bare plurals seems unproblematic. Recall examples like (57) repeated from Section 4.1.2: (57) En général, les members de ce département assistent à toutes les réunions. Mais tous ne sont pas toujours là. Ainsi thésards et Aters ont séché la réunion d’hier. ‘In general, members of this department always attend all the meetings. But not everyone is always there. So graduate students and assistant professors stayed away from yesterday’s meeting.’ (Roodenburg 2004b: 309 (23))

 The Layered DP: Form and meaning of French indefinites

If the coordination in italics is quantificational, it implies that its categorial status is QP. According to the analysis proposed here, this means that the left periphery of such nominals is composed of two layers, a PropP and a QP. By analogy with coordinated bare nouns interpreted as properties, the coordination may move from its base position to Spec,QP, through Spec,PropP, to legitimate the QP layer: (58)

QP Spec N&Ni Ns&Nsi

Q′ Q

PropP Prop′

Spec ti Prop





NumP ti

(58) is the structure of quantificational coordinated bare nouns which emerges from the analysis proposed in this book. What remains to be determined is whether the predictions associated with it, in particular concerning the scope properties of such bare nouns, are borne out. In effect, according to the analysis developed here nominals with a structure like (58) are expected to have scope properties. This is because Quantifier Raising (QR) is triggered by the presence of the QP layer. As this discussion is very speculative, the issue is left as an open question. The structures in (56) and (58) share two characteristics with the ones proposed by Roodenburg (2004b): first, they are of different ‘sizes’ and second they involve movement of the whole coordination: (59) a. [NumP Num [+ pl] [CoordP [NP N] & [+ pl] [NP N]]] b. [DP D [NumP Num [+ pl] [CoordP [NP N] & [+ qu] [NP N]]]] (Roodenburg 2004b, 312 (26) and (27), adapted from Heycock & Zamparelli 2003)

(59a) corresponds to what Roodenburg calls the weak reading and (59b) to the strong reading. According to the author, all coordinated bare nouns are [+ pl], hence Num[+ pl] in (59). The conjunction, &, may be [+ pl] or [+ qu] depending on the weak/strong reading respectively. The [+ qu] feature on & in (59b) triggers movement of the Coordination Phrase to DP. It is this movement which is responsible for the strong reading. Although the details of (59) are irrelevant here, Roodenburg’s analysis suggests that strong coordinated bare plurals are ‘bigger’ than weak coordinated bare plurals in



Chapter 4.  Bare nouns 

the sense that they involve one more layer. This corroborates the analysis ­developed here in which QPs are ‘bigger’ than PropPs. Furthermore, (59b) suggests that the whole Coordinated Phrase may move. The difference with the analysis proposed here however is that this constituent only moves when coordinated bare nouns have a strong reading, which corresponds to our quantificational interpretation. In the analysis proposed here, in contrast, the coordination moves, whether the construction is propertydenoting or quantificational. The movement of this constituent legitimates PropP and QP and accounts for the distributional difference between coordinated bare nouns and non-coordinated bare nouns in Romance. The next section examines some aspects of the inflectional domain of coordinated bare nouns.

4.2.2  The inflectional domain The question addressed in this section is whether the inflectional domain of coordinated bare nouns is analogous to the one of non-coordinated bare nouns. The structure proposed for the latter is reported below from Section 3.2: (60) Example a. books b. bread

Number FP3 -s √

Quantity FP2 X X

Count FP1 X X

Lexical category NP book bread

The issues are however more complex for coordinated bare nouns as it is difficult to determine which features characterise the coordination as a whole and which ones are specific to each conjunct. Let us first examine the projection encoding number, FP3 in our terms, as this layer is the least controversial. In effect, both coordinated bare plurals and coordinated bare singulars are plural, independently of the number of their individual conjuncts (Roodenburg 2004b; recall (59) above). In terms of structure, this means the FP3 is projected in the inflectional domain of coordinated bare nouns and that it encodes the feature [+ plural]. What about FP1 and FP2? As coordinated bare count nouns like cat and dog have been excluded from the discussion (Section 4.1.1), it seems plausible to assume that the structure of the remaining coordinated bare nouns is analogous to the one of noncoordinated bare nouns. In other words, FP1 and FP2 would not be projected in the inflectional domain of coordinated bare mass nouns and coordinated bare plurals. That FP1 is absent from the structure of coordinated bare nouns is supported by the following examples adapted from (31) and (32b) in Section 3.2: (61) a. Kim ate apples and pears this afternoon (for an hour) (*in an hour). b. Pat built houses and garages (all summer) (*in two months). c. My kid sister drew circles and squares (all morning) (*in half an hour).

 The Layered DP: Form and meaning of French indefinites

(62) *John recommended books and magazines and Mary recommended others.

Recall that such data have been used to show that bare nouns lack individuated reference and hence that their structure does not comprise the FP1 layer. This seems to extend to coordinated bare nouns. As for the FP2, the analogy with non-coordinated bare nouns suggests that it is also absent from the structure of coordinated bare nouns. However, I will explore another possibility here for reasons to be explained shortly. Thinking of the role of coordination I tentatively propose that coordinated bare nouns encode some quantity. In effect, informally, what and (&) does is put together two sets. In other words, we know that the quantity involved is two. Consequently, the structure proposed here for coordinated bare nouns is as follows: (63)

 Example PropP a. N&N b. Ns&Ns

Number FP3 N&N Ns&Ns

Quantity FP2 N&N2 Ns&Ns2

Count Lexical category FP1 NP X N&N X Ns&Ns

If coordinated bare nouns encode the quantity 2, it means that their inflectional domain is composed of FP2, in contrast with the one of non-coordinated bare nouns (recall (60)). This also implies that the inflectional domain of coordinated bare nouns is analogous to the one of quantitative du-NPs whereas the inflectional domain of non-coordinated bare nouns is similar to the one of non-quantitative du-NPs (except for the presence of deP). The structures proposed for du/des-NPs in Chapter 3 are reported below for convenience: (64)

 Example a. du painquantit. b. du painnon-quantit. c. des painsquantit. d. des painsnon-quantit.

Number FP3 de + lesing de + lesing de + lesplur de + lesplur

Quant. FP2 deP de de X de de de X de

Count FP1 X X –s –s

Lexical category NP pain pain pain pain

Another characteristic shared by coordinated bare nouns and du-NPs is that they may have a quantificational reading (see Section 4.1.2 for the former and Chapter 3 Section 2.2 for the latter). Considering that this interpretation involves a set-subset relation, it seems that a correlation between quantificational reading and quantity may exist. Put differently, the idea is that to be able to have a quantificational reading, a noun phrase has to encode some quantity. In terms of structure, this implies that FP2 has to be present in the inflectional domain for the QP layer to be projected. This of course does not mean that all noun phrases with FP2 are quantificational, but rather that to ‘become’



Chapter 4.  Bare nouns 

quantificational some quantity is required. If this is correct, it means that when du/desNPs are quantificational, FP2 is present in their structure. Put differently, the quantitative du/des-NPs in (64a,c) may be quantificational contrary to the non-quantitative du/des-NPs in (64b,d). That there is a correlation quantification-quantity is supported by the interpretation of un-NPs. In effect, when un-NPs are quantificational they have a covert partitive reading, i.e., un livre ‘a/one book’ is interpreted as un 〈des livres〉, where un means 1. If, as proposed here, un-NPs only contain FP2 in their structure when they involve a numeral (Chapter 3 Section 3.2.2), it represents evidence for a correlation quantification-quantity. The structures for the two kinds of un-NPs are repeated below: (65) Example a. unart pain b. unnumeral pain

Number FP3 unsing unsing

Quantity FP2 X un

Count FP1 un un

Lexical category NP pain pain

The existence of a connection between quantity and quantification corroborates the observation in Section 3.2 that the composition of the inflectional domain has an impact on the constitution of the left periphery. Recall that a possible correlation between the projection of FP1 and SRefP was evoked. Although many issues concerning coordinated bare nouns have to be further investigated to reach any firm conclusions, it seems that the general lines of analysis developed in this book can be extended to these nominals, at least to the ones which do not involve count singulars. The next section mentions additional questions related to the study of bare nouns in general. The aim is to draw the reader’s attention to further implications of the work presented in this book.

5.  Further issues 5.1  The generic reading The history of philosophy of language and of linguistics associates different phenomena to genericity. The two basic varieties of genericity are kind-referring nominals, sometimes called generic noun phrases (66), and propositions expressing a general property, also called characterizing sentences (67) (Carlson & Pelletier 1995). (66) a. The potato was first cultivated in South America. b. Potatoes were introduced into Ireland by the end of the 17th century. c. The Irish economy became dependent upon the potato.  (Krifka et al. 1995: 2)

 The layered DP: Form and meaning of French indefinites

(67) a. John smokes a cigar after dinner. b. A potato contains vitamin C, amino acids, protein and thiamine.

(ibid)

A classification into generic nominals and generic sentences would however be simplistic. One reason is that both phenomena, kind referring noun phrases and characterizing sentences, can be combined in a single sentence: (68) a. Potatoes are served whole or mashed as a cooked vegetable. b. The potato is highly digestible. (ibid, 1995: 3)

(68) shows that it is often not sufficient to concentrate on the nominal expression to analyse a sentence, as observed throughout this book. The above examples also show that genericity may involve different types of noun phrases, including the-NPs.12 This is one reason for which genericity has been ignored in this book. Another one is that the generic interpretation of kind-referring nominals is often opposed to indefiniteness (usually associated with existentiality) (see ­Zamparelli 2000 for an analysis of kinds). Put differently, kinds are not concerned by this study which focuses on a subset of indefinites. The aim of this section is thus merely to mention a few issues related to bare nouns, in particular to their internal structure. Recall from Section 2.1 that in Germanic languages bare nouns are ambiguous between an existential and a generic reading, whereas in Romance they can only be existential. The relevant examples are repeated as (69). (69) a. John was eating biscuitsexist . / John hates catsgen . (English) b. Gianni mangiava biscottiexist . / *Gianni odia gattigen . (Italian)  (Roodenburg 2004b: 122(6–7))

In (69a), biscuits is interpreted as existential and cats as generic. The question which arises is whether these bare nouns have different internal structures or not. If the genericity of John hates cats is due to the kind-referring nominal cats, the theory .  This is also true of French: generic contexts may involve nominals with the so-called definite article (i) or with the indefinite article (ii) (see for example Vogeleer & Tasmowski 2005). Note that the les-NP in (ib) corresponds to a bare plural in English.

(i) a. b.



(ii)

La baleine est un mammifère. the whale is a mammal ‘The whale is a mammal.’ Les chemises en cotton se lavent facilement. the shirts in cotton refl wash easily ‘Cotton shirts wash easily.’

Un chien a quatre pattes. a dog has four legs ‘A dog has four legs.’

(French)

(French)



Chapter 4.  Bare nouns 

­ eveloped here predicts that they do. The reason is that, in our framework, structures d are not ambiguous (cf. end of Chapter 1 Section 4.2). In other words, if biscuits and cats differ in interpretation, they cannot be structurally identical. One such proposal is Zamparelli’s (2000) (Chapter 2 Section 3.1.2). The structure he argues for is reported below: (70)

SDP

PDP

Strong Quantifiers

Weak Quantifiers

Adjectives





KIP

NP Noun

complement

According to Zamparelli, the determiner system of a language is constituted of SDP and PDP, which embeds a third functional layer, the Kind Determiner Phrase (KIP). This projection hosts the NP-proper and any potential restrictive adjective. The problem with this analysis is that only SDPs may be arguments, which is incompatible with one of the crucial assumptions adopted here, namely that argumental nominals involve at least one layer of the DP domain. In effect, as kind-referring noun phrases function as arguments, they are expected to comprise, minimally, PropP in their left periphery. The rest of the section will therefore explore such a line of analysis. One way to distinguish between non-generic and generic bare nouns is to assume that the latter, in contrast with the former, host a non-overt determiner which has to be bound by a clausal operator (see for example Carlson & Pelletier 1995; Borer 2005a and the references therein):13 (71) GEN [DP ec [cats]]

.  According to Borer (2005a), the GEN operator assigns range to 〈e〉d, and to 〈e〉# (cf. Chapter 1 Section 4.4.2.1).

 The Layered DP: Form and meaning of French indefinites

In the analysis developed here, where DP is split into several projections, the nonovert determiner would sit in Prop:14 (72)

(generic)

PropP Spec

Prop′ Prop ec



NumP cats 



(72) differs from the structure of bare plurals like biscuits in that the movement of the noun to Prop is blocked by the presence of the empty category (ec). The structure proposed for non-generic bare nouns is repeated below from Section 3.1: PropP

(73) Spec

(existential)

Prop′

Prop biscuitsi



NumP ti





Assuming that kind-referring noun phrases host a non overt determiner in Prop implies that, in Germanic languages, generic and non-generic bare nouns have different structures. However, it also suggests that the former are analogous to Italian bare nouns, restricted to object positions (Section 3.1): (74) Spec

Prop′ Prop ec





(Romance)

PropP

NumP studenti

The difference between (72) and (74) however lies in the kind of empty category involved: in Italian bare nouns, ec has to be bound by a lexical head, whereas in generic nominals like (72) it is bound by a sentential operator. If this is correct, it predicts that generic bare nouns in Germanic are not excluded from subject positions, in contrast

.  Whether generic nominals are PropPs is of course debatable (Chapter 3 Section 3.2.2; see also Zamparelli 2000).



Chapter 4.  Bare nouns 

with Italian bare nouns. This seems to be supported by (66b) above or by examples like (75): (75) Dogs hate cats.

As potatoes and dogs can function as subjects, it supports the idea that the empty category they comprise has to be distinguished from the one characterising Italian bare nouns. Another possibility to account for examples involving kind-referring arguments is to assume that these nominals involve a free variable, i.e., an empty category which is not bound by an operator. Such an account allows Guéron (2006) to deal with genericity without postulating a silent operator in the clause (nor in the nominal structure). The idea is that genericity depends on the tense node. More precisely, Guéron proposes that what distinguishes generic from non-generic sentences is the absence of a positively indexed tense node. This means that the generic reading results from a tense node which has an index 0 (or no index) and that the non-generic, i.e., the episodic, interpretation is due to an indexed tense. The index depends on the tense involved, for example 1 for the present, 2 for the past, etc. The difference between generic and episodic construals is illustrated below: (76) Le chien aboie. ‘The dog is barking/barks.’ (77) a. b.

Ecoute, le chien aboie. [CP C1 [TP le chien T1 [vP aboie]]] ‘Listen, the dog is barking.’ Le chien aboie. [CP C1 [TP le chien T0 [vP aboie]] ‘(*Listen) the dog barks.’

(epis./generic) (Guéron 2006: 220(3)) (episodic)

(generic) (ibid., (4))

According to Guéron, different tense nodes disambiguate sentences like (76) which may typically be episodic or generic. When the construction is episodic, the tense switch is set on and when it is generic the tense switch is set off, hence T1 and T0 in (77a) and (77b) respectively. Episodic and generic sentences involving bare nouns may be distinguished in the same way: (78) a. Dogs are barking. b. Dogs bark. 

(episodic) (generic) (ibid., (11))

The distinction between (78a) and (78b) is that the tense node in the former is indexed, in contrast with the tense node in the latter. An interesting implication of such an account is that “[n]o special grammar of genericity is needed” (2006: 222).

 The Layered DP: Form and meaning of French indefinites

Another aspect of the analysis is however problematic for the framework developed here. In effect, Guéron suggests that the internal structure of bare nouns is the same in episodic and generic constructions (2006: 226) despite their different interpretations.15 Put differently, dogs in (78a) and (78b) are structurally identical although the first one refers to ‘‘real dogs’’ and the second one ‘‘to the mental concept associated with the lexical item DOG’’ (2006: 222). As such a thesis is incompatible with the assumptions adopted here, and in the absence of a better account, I will retain the structures proposed in (73) and (72) respectively. In other terms, the difference between non-­generic and generic bare nouns in Germanic is that the former involve head ­movement to Prop, whereas the latter host a free variable in this position. A binary distinction however is more questionable for a-NPs (and possibly also for the-NPs which are not examined here). Consider the following example: (79) a. A dog is barking. b. A dog barks. 

(episodic) (generic) (ibid., (12))

If, as proposed by Guéron, a dog in (79a) and (79b) represents a real dog and a mental concept respectively, it suggests that this noun phrase may have only two interpretations, which doesn’t correspond to the facts described in Chapter 2. A way to circumvent the problem might be to combine different tense nodes, as suggested by Guéron, and different nominal types/structures, as proposed here. The result would necessarily be more complex but it would also be more fine-grained. Such a combination might be particularly worth exploring as Guéron’s analysis of DP and TP interacts with a third ingredient, namely aspect issues, a topic briefly taken up in Chapter 5 Section 2.1.

5.2  A note on incorporation The notion of incorporation is often associated to bare nouns in non-determinerless languages, where the term bare noun is used in the wide-ranging sense of nominals which are not introduced by a determiner. The general idea is that such nominal expressions suffer from some kind of deficiency (often related to their (lack of) existentiality), which must be compensated by the category governing them, usually the verb. Consequently, these bare nouns will be semantically and/or syntactically incorporated in the verb. The question of incorporation has received considerable attention in the last two decades, as the amount of literature on the topic attests (see for example Sadock 1980; Mithun 1984; Baker 1988; de Hoop 1992; McNally 1995, 1998; Van Geenhoven 1998; Dayal 1999; Massam 2001; Chung & Ladusaw 2003; Farkas & De Swart 2003 and the

.  The overall structure of nominals adopted by Guéron differs from the one proposed here (footnote 12 Chapter 1 Section 4.4.2.1).



Chapter 4.  Bare nouns 

references therein; see also Vogeleer & Tasmowski 2006).16 However, this is not the place to establish a list of issues related to incorporation, even less to review the literature on the subject. This section rather consists in a longish footnote composed of a few thoughts on the structure of incorporated bare nouns. Let us begin with some reminders concerning semantic incorporation. This notion concerns nominals which have fixed scope properties and which are low on individuality (Mathieu 2006 for a recent reference).17 In other words, it affects the bare nouns examined in this chapter (Section 2.1) and more generally our property-denoting arguments (Chapter 2 Section 2.3). It refers to the idea that the ‘narrowest’ scope of some indefinites is ‘semantically incorporated’, i.e., due to the predicate which provides an operator to their interpretation (Chapter 2 Section 2.3.2). Such predicates have been labelled ‘existential predicates’, in opposition to ‘entity predicates’ which can only select referential arguments (Van Geenhoven 1995; Dobrovie-Sorin 1997a,b, McNally 1998; Dobrovie-Sorin & Laca 2003). Incorporated nominals are arguments selected by the former type of predicates. They may have special morpho-syntactic properties such as the presence of a morphologically deficient determiner or absence of determiner. They may also be restricted to specific positions, usually adjacent to the verb. Turning to syntax, I have argued that semantically incorporated nominals have a particular structure, i.e., they are Property Phrases (PropPs) (see Section 3.1 for bare nouns). Compared with S-referential or quantificational indefinites, property-­denoting indefinites are thus structurally smaller. In effect, in the analysis developed here S-referential indefinites have a three-layer left periphery, quantificational indefinites a two-layer left periphery and property-denoting indefinites a one-layer left periphery. What I would like to suggest here is a link between size of structure and incorporation. The idea is that the structure of nominals which are semantically incorporated is truncated. This is the case of Property Phrases, which lack the Quantification Phrase (QP) characteristic of quantificational nominals and the S-reference Phrase (SRefP) typical of S-referential noun phrases. In other words, the topmost layers of their left periphery are missing. Consequently, their relation with the predicate is ‘closer’ than the relation between a predicate and a (S)-referential noun phrase for example. If this is correct, it suggests that it is because their structure is truncated that arguments of this type incorporate to the predicate.

.  The papers in Part I are concerned with various problems related to the notion of incorporation. For example, they address the questions of how incorporation should be defined, what its boundaries are, what the contexts in which it is allowed are etc; alternatives to the incorporation approach are provided in Part II of this volume. .  Farkas & De Swart (2003) argue that narrow scope is not a sufficient condition for semantic incorporation. See also De Swart 2006 and Farkas 2006.

 The Layered DP: Form and meaning of French indefinites

Pushing the reasoning further, one could speculate on the truncation of additional layers of the nominal structure, which would lead to a special relation with the governing head (for example a very restricted distribution). Put differently, the more layers are absent, the tighter the relation with the incorporating element should be. Crucially such a line of analysis predicts that, in restricted cases and under special conditions, argumental nominals may be ‘smaller’ than PropPs. More precisely, it suggests that in certain “constructions in which a verb and one of its arguments form a particularly tight unit” (Farkas & De Swart 2003: 1), more than the left periphery may be truncated. This could for example be the case of ‘pseudo noun incorporation’. In effect, Massam (2001) shows that in some languages like Niuean, nominal incorporation involves a nominal expression which cannot be a DP in a pre-theoretical sense. The incorporated element cannot be a head either, as it may comprise adjectives.18 A ‘pseudo-incorporation’ account referring to Massam is proposed by Dobrovie-Sorin et al. (2006) for bare singulars in Spanish and Rumanian. The authors propose that such nominals are NPs, in contrast with bare plurals, considered as NumPs. In sum, the idea is that the elements involved in incorporation phenomena have a truncated structure. More precisely, incorporation is due to a reduced structure.

6.  Conclusion This chapter shows that the analysis developed in this book extends to (non-­coordinated) bare nouns and to coordinated bare plural and mass nouns. I have ­suggested that bare nouns, which have a single reading, are Property Phrases and that ­coordinated bare nouns may be Property Phrases or Quantifier Phrases when they have a covert partitive reading. Furthermore, I have suggested that there is a correlation between the projection of SRefP and FP1, which is responsible for the count/individuated reading, and between the projection of QP and FP2, which encodes quantities. More precisely, I have hypothesised that nominals can only be SRefPs if their inflectional domain contains FP1 and that they can only be QPs if their inflectional domain contains FP2. These correlations between the structure of the inflectional domain and the composition of the left periphery further imply that only quantitative du/des-NPs can be quantificational (in opposition to non-quantitative du/des-NPs which are analogous to bare nouns) and that when un-NPs are quantificational they involve the numeral un. .  For Baker (1988), incorporation represents a more technical operation. It is defined in syntactic derivational terms and only concerns heads, i.e., category X0. In this sense, incorporation thus involves the adjunction of a lexical head to another lexical head. In the case of nominal incorporation, it concerns unmodified nouns which move from their base position because they fail to be assigned case.



Chapter 4.  Bare nouns 

As bare nouns lack individuated reference and do not involve a quantity, their inflectional domain does not comprise FP1 and FP2 and they cannot be quantificational or S-referential. The plural –s of bare plurals is generated in the layer related to grammatical number, FP3, and the listeme moves to this position to bind it, before moving to the left periphery. As for coordinated bare nouns, I have proposed that the coordination implies the quantity two, i.e., one element/set on each side of &. Finally, I have tentatively suggested that incorporation might correspond to syntactic truncation.

chapter 5

Conclusion 1.  The results 1.1  The analysis and the proposals This book has taken us from the study of argumental un-NPs and du/des-NPs in French to bare nouns in Romance and Germanic. The journey has led us from crucial empirical facts and theoretical assumptions to various hypotheses and interesting conclusions. On the way, we stopped at structural details, strolled through interface considerations and raced through a little formal semantics. The empirical facts this study examines are the various interpretative properties of the above mentioned indefinites, including their scope properties, and their different behaviour at the clausal level. More precisely, we observed that un-NPs and des-NPs may have three readings, that du-NPs and coordinated bare nouns may have two readings and that non coordinated bare nouns have a single reading. The aim here was therefore to capture these differences, at a descriptive and explanatory level. To achieve this goal I adopted a framework where the nominal structure is split into three domains on a par with the clause, namely a left periphery, an inflectional domain and NP-shells, and built on two main assumptions. The first one is that argumental and predicative noun phrases are structurally different in that the former, contrary to the latter, need to have the DP layer projected. The second one is that there is a strict syntax-semantics mapping. On the basis of these assumptions, I made several claims. The first one – not completely new – is that the different interpretations of the indefinites under consideration correspond to different semantic types. As some of these indefinites may have three readings, it implies that they can belong to three different semantic types, which I labelled S-referential for speaker’s reference (type 〈e〉), quantificational (type 〈〈e,t 〉, t 〉) and property-denoting (type 〈e,t〉). Restating the ambiguity of the indefinites examined, it means that un-NPs and des-NPs may be S-referential, quantificational or propertydenoting, that du-NPs and coordinated bare nouns may have the latter two readings and that non coordinated bare nouns are property-denoting. The investigations reported in this book therefore corroborate other accounts considering that property-denoting nominals can function as arguments and that indefinites cannot belong to a single semantic category.

 The Layered DP: Form and meaning of French indefinites

The second – original – claim is that each semantic type has a left periphery of a different size. More precisely, I have proposed that the left periphery of property­denoting arguments involves one layer, called Property Phrase, that the one of quantificational arguments contains two layers, the Property Phrase and a Quantifier Phrase, and that the one of S-referential arguments contains three layers, the Property Phrase, the Quantifier Phrase and an S-reference Phrase. This is in part based on the assumption mentioned above that argumental, unlike predicative, noun phrases must have a DP layer. A third – deriving – claim concerns the structural DP/non-DP distinction between arguments and predicates. In a framework with three domains involving several projections, this distinction has to be refined. What I have proposed is that it is the projection of the lowest layer of the left periphery which turns predicates into arguments. In other words, the structure of predicates is truncated at the level of the inflectional domain. A fourth – more controversial – claim is that the scope properties attributed to some indefinites are mainly due to their internal structure. I have argued for example that S-referential noun phrases may take widest or intermediate scope because the Quantifier Phrase contained in their structure triggers Quantifier Raising. However, they differ from traditional quantifiers in that they encode S-referentiality. As this kind of reference is context dependent, it means that such indefinites are analogous to pronouns, at least to some extent. They may be deictic in the sense that some extra linguistic information is required for them to be interpreted or co-varying with a universal quantifier. The first option gives rise to widest scope and the second one to intermediate scope. A fifth – multifaceted – proposal concerns the structure of the inflectional domain of the indefinites examined, and of noun phrases in general. I have proposed that this domain consists of several discrete projections, each realising a given feature. More precisely, I have postulated a projection encoding a [count] feature, another one encoding [quantity] and a third one encoding [number]. I have also argued that the first projection is responsible for individuated reference and that the second one is only projected when quantity cannot be inferred from other features. As for the components of du/des in French, I have proposed that le/les ‘the’ is generated in the projection encoding number whereas de ‘of ’ is generated in a deP situated between the [count]layer and the [quantity]-layer. As a result, the fusion of de and le/les takes place in the [number]-layer. The study of the inflectional domain has led to additional results. First, it has allowed us to account for the difference between the numeral un and the article un. What I have proposed is that the former is generated as an XP on a par with other numerals, contrary to the latter. This implies that un-NPs involving the numeral 1 comprise a [quantity]-layer in their structure, in contrast with un-NPs involving the article un. Second, I have suggested that de-NPs are parallel to beaucoup de ‘many/



Chapter 5.  Conclusion 

much of ’ and peu de ‘few/little of ’ except that they contain an empty category in the [quantity]-layer. As this element has to be bound by a negative operator, the distribution of de-NPs is restricted. Third, a brief examination of en pronominalisation has shown that en pronouns can be classified into enidentification, enproperty, enpartitive and engenitive. Fourth, the study of the inflectional domain has suggested that there might be some correlations between its structure and the composition of the left periphery. The hypothesis put forward is that the S-reference Phrase can only be projected if the [count]-layer is present in the inflectional domain. In the same vein, the Quantifier Phrase would depend on the [quantity]-layer. A final result concerns explicit ­partitives, i.e., partitives in which the set is overtly expressed. I have proposed that the partitive PP sits in the specifier of a projection immediately above the [quantity]-layer and that the de generated in the deP is not realised overtly as a case of haplology, thus accounting for the restricted distribution of explicit du/des partitives. A consequence of this analysis is that articles in French are generated in the lowest functional projection of the inflectional domain and that they move through the whole functional structure of the nominal expression. It is this movement which is responsible for the projection of the layers forming the nominal structure. This movement is probably triggered by a feature-checking mechanism. This would mean that articles are (lexically) endowed with features which have to be realised in the structure. A sixth – not completely new – claim concerns partitivity. I have suggested that partitives should be classified as explicit or implicit. The former express a set overtly whereas the latter express a subset. As implicit partitives are covert partitives, they are quantificational. This is not the case of explicit partitives which may belong to different semantic types. Put differently I have argued that partitivity and quantification should be distinguished. A seventh – encouraging – claim is that the analysis developed here extends to bare nouns, be they coordinated or not (to the exclusion of bare singulars). What I have proposed is that non-coordinated bare nouns are Property Phrases, whereas ­ coordinated bare nouns (plural or mass) may be Property Phrases or Quantifier Phrases. As for their inflectional domain, the proposal is that it does not comprise the [count]-layer nor the [quantity]-layer when they are non-coordinated but that they may have the latter when they are coordinated. This implies that the plural –s of noncoordinated bare nouns differs from the traditional plural morpheme which leads to an individuated reading. What I have proposed is that the –s of bare nouns is generated in the [number]-layer. An eighth – interesting – suggestion is that incorporation is due to structure ­truncation. For example, Property Phrases, which lack an S-reference Phrase and a Quantifier Phrase, incorporate semantically. More generally, I have hypothesised that it is because the structure of this type of arguments is incomplete that they incorporate semantically.

 The Layered DP: Form and meaning of French indefinites

An important conclusion of the analysis proposed in this book concerns the syntaxsemantics interface. This study shows that syntax drives semantics – and not the other way round. In other words, meaning is shaped by structure. This is striking in the account of S-referential noun phrases – their scope properties are due to a specific aspect of their internal structure – and of incorporation – a truncated structure triggers incorporation. Related to this conclusion, the investigations reported here clearly show that syntax is not about words or morphemes but about features, which allows more subtle distinctions and gives rise to more complex structures. Another significant conclusion concerns the syntax-lexicon interface. If arguments may be of different sizes, it implies that selection is not simply in terms of nominal or prepositional arguments for example. It has to be much more fine-grained to be able to distinguish between Property Phrases, Quantifier Phrase and S-reference Phrases, which are all nominal arguments. Finally, an analysis like the one presented here, which takes into consideration the interpretative properties of the different semantic types examined as well as their structural complexity, proves superior to a choice function account for example. For instance, it allows a distinction between two sources of existentiality: when propertydenoting arguments assert existence, it is because the predicate selecting them provides an existential operator; when they are S-referential, it is their landing site at LF which provides such an operator. In sum, the trip was worth it. It has allowed interesting discoveries and has aroused our curiosity about further destinations (see for example Section 2 below).

1.2  Prediction for language acquisition Attributing nominals an extendable telescopic left periphery makes (at least) one prediction. In effect, if the nominal left periphery is a construction with several layers piled up step-by-step as proposed in this study, it predicts that children will first acquire nominals with the lowest layer of the left periphery. This corresponds to property-denoting arguments, i.e., Property Phrases. This prediction seems to be borne out. Although it is difficult to determine what steps are taken by children, Roeper (2006) for instance argues that less specific interpretations emerge very early and that children “move from less specific to more specific” (59a). For example, he argues that the first stage in the acquisition of nominals reflects the notion of Kind in the sense that the child tries to identify a class and not to name an individual. Roeper further points out that children use expletive definite articles very early. This concerns examples like here’s the plan, what’s the difference, go to the hospital where the ‘definite’ article does clearly not entail a prior referent. Matthewson & Schaeffer (2000) have also observed that children use definite articles ‘incorrectly’ without a prior reference. If this is on the right track, it does not only



Chapter 5.  Conclusion 

support the idea of a left periphery built layer-by-layer but also that such an analysis should be applicable to different kinds of nominal expressions and not only to some indefinites (recall Section 3.4 in Chapter 3).

2.  Further issues and future research In this book I have focussed on a variety of indefinites. However, it would be interesting to see whether the framework developed may account for other ‘weak’ indefinites like bare cardinal plurals (including many) and wh-phrases. Furthermore, if the nominal left periphery may comprise different layers related to the interpretation of the nominal expression, we expect the line of analysis explored here to extend to other nominals, for example with a definite article, with possessives or demonstratives. The parallel established between S-referential noun phrases and pronouns also deserves further attention. Among the issues worth examining in more detail, we find of course genericity and kinds, only briefly mentioned in this work. Finally, as notions like topicality and focus generally associated with the left periphery have been left out of the discussion, future research will have to determine their role in the system advocated here. The general issues mentioned above are not the only topics which should be investigated. Two questions are dealt with in some more detail in the rest of the section. The first one is the unbounded reading often associated with (some) indefinites and the second one determinerless languages.

2.1  The unbounded reading That the interpretation of nominal expressions (mainly in object position) has an impact on the aspectual nature of the sentence as a whole has been clear from the seventies onwards (Verkuyl 1972; Bennett & Partee 1978; Dowty1979; see also Borer 2005a,b, De Swart 2006; Stark, Leiss & Abraham 2007 for more recent references). The aim here is not to do justice to all the literature on the topic, which would obviously be impossible in a few paragraphs, or to contribute to the debate on the interrelation between verbal and nominal characteristics, mentioned towards the end of the section. Rather, it is to highlight some aspects of the analysis developed in this book concerning the interpretation of nominals typically leading to atelicity/unboundedness, namely bare nouns and du/des-NPs, in contrast with singular un-NPs (and definite noun phrases) which generate a telic/bounded reading. Before turning to the interpretation of nominals involved in atelicity, a few introductory remarks are in order. Aspect is a concept which is often difficult to grasp, especially for speakers of Indo-European languages, because it is related to other notions

 The Layered DP: Form and meaning of French indefinites

like tense. Furthermore, there is a distinction between lexical aspect and grammatical aspect which is important. The former is a classification of verbs, whereas the latter is a classification of various forms of a single verb. In other words, lexical aspect, also called situation aspect, is an inherent property of verbs, in contrast with grammatical aspect. It is also known as Aktionsart, especially by Slavic and German linguists. One of the most cited typology of verbs is probably Vendler’s (1957). It classifies verbs into four categories: activities (shop), accomplishments (build a house), achievements (buy) and states (have). Although activities and accomplishments allow the use of continuous tenses (in contrast with achievements and states), they differ from each other by telicity: accomplishments have a natural end point, whereas activites do not. The verbs of the former class are called telic whereas the ones of the latter are called atelic. These two categories may be distinguished by the use of for-phrases, which describe a time duration. For example, for-adverbials are impossible with accomplishments and achievements, but grammatical with states and activities. In other words, they may be used as a test for atelicity/unboundedness. Crucially, however, telicity is not a characteristic of verbs in isolation. It concerns the entire verb phrase, which implies that the potential object of the verb has to be taken into consideration. The properties of these arguments are examined below; grammatical aspect is mentioned at the end of the section. The example mentioned in Chapter 4 Section 2.1 to illustrate the atelic reading of sentences involving bare nouns is repeated below:

(1) a. Max discovered two rabbits in his yard in two hours / ??for two hours. b. Max discovered rabbits in his yard ??in two hours /for two hours.

As mentioned above, for-adverbials present the sentence as an unbounded/atelic event. In-adverbials in contrast describe a bounded/telic event. This well-known test shows that sentences involving bare nouns are aspectually characterized as unbounded (1b). The following examples illustrate des-NPs, un-NPs and les-NPs (from Bosveld-de Smet 1998, quoted in De Swart 2006).

(2) a. Marie a cueilli des fraises pendant des heures. Marie picked indef.pl strawberries for hours b. *Marie a cueilli des fraises en une heure. Marie picked indef.pl strawberries in an hour



(3) a. *Marie a cueilli une fraise / les fraises pendant des heures. Marie picked a strawberry / the strawberries for hours b. Marie a cueilli une fraise / les fraises en une heure. Marie picked one strawberry / the strawberries in an hour

(French)

These data show that only des-NPs lead to an unbounded interpretation: the foradverbial is grammatical in (2a) but not in contexts with un-NPs and les-NPs (3a). Similar examples with du-NPs are provided below:





Chapter 5.  Conclusion 

(4) a. Marie a ramassé du coton pendant des heures. Marie has picked indef.sg cotton for hours ‘Marie picked cotton for hours.’ b. *Marie a ramassé du coton en une heure. Marie has picked indef.sg cotton in an hour

As the grammaticality of pendant des heures ‘for hours’ in (4a) shows, du-NPs occur in atelic environments, on a par with des-NPs and bare nouns. What leads to an atelic reading has been discussed in the literature at least since Verkuyl (1972) (recall also footnote 33 in Chapter 3). Verkuyl classifies noun phrases into [+ sqa] vs. [− sqa]. sqa stands for ‘referring to a specified quantity of A’ and A for the interpretation of the common noun. The positive setting of the feature leads to a telic reading whereas a minus value gives rise to an atelic interpretation. The feature is set negatively, i.e., [− sqa], only when other possible sources of quantizedness are not available. This account thus suggests that there is a correlation between atelicity and absence of quantity and between telicity and quantity. The notion of quantity involved by nominals is often referred to as ‘(lack of) quantificational force’ in the literature. Note however that the term ‘quantificational’ in this expression differs from the one used in this book: quantificational force is not related to quantification as defined in Chapter 2 Section 2.2. The problem with the abovementioned thesis is that definite nominals lead to a telic/ bounded reading although they do not involve any quantity (Borer 2005a for a different proposal). Verkuyl (1993) is aware of the problem but doesn’t have any solution to propose. That the absence of quantity is not sufficient to explain the aspectual contrast between (non-quantitative) des-NPs and les-NPs in French has been observed by Bosveld-de Smet (1998). In effect, although none of these nominals involve a quantity, only des-NPs lead to an atelic reading. Put differently, if the absence of quantity were at the origin of the unbounded reading, les-NPs, on a par with des-NPs, should lead to such an interpretation, contrary to fact (recall (2) and (3)). In the same vein, the lack of individuated reference of a nominal expression cannot be the cause of the atelic reading either (recall that this question was raised in Chapter 4 Section 3.2 for bare nouns). If it were, it would predict that des-NPs, which have individuated reference (Chapter 3 Section 3.2), lead to a telic/bounded reading contrary to fact (at least in a majority of cases; cf. footnote 1). The interpretative properties associated with the noun phrases mentioned in Chapter 3 are repeated below for convenience. The reader is referred to the relevant sections, in particular Section 3.2.1, for any explanations. (5) illustrates the nominals which seem to lead to an atelic reading, whereas (6) mentions the ones characterising a telic reading.

.  The observation in Chapter 3 Section 3.2.1 that some du/des-NPs may be involved in telic contexts is discussed shortly.

 The Layered DP: Form and meaning of French indefinites





(5) Example a. du painquantit. b. du painnon-quantit. c. des painsquantit. d. des painsnon-quantit. e. books f. bread (6) Example a. un pain b. le painmass c. le paincount d. les pains

Number FP3 de + lesing de + lesing de + lesplur de + lesplur - s √

Number FP3 unsing lesing lesing lesplur

Quant. FP2 deP de de X de de de X de X X

Quantity FP2 X X X X

Count FP1 un X le - s

Count FP1 X X - s - s X X

Lexical category pain pain pain pain book bread

Lexical category NP pain pain pain pain

What is striking if we compare the nominals in (5) and (6) is that there is no feature which allows us to distinguish between the two groups. In other words, it is impossible to isolate one factor which would lead to an unbounded interpretation, in contrast with a bounded reading. For example, as mentioned earlier, atomicity can be related to nominals in (5) (des-NPs) or (6) (un/lecount/les-NPs). Similarly, the (lack of) quantity involved does not characterise one group. If atelicity is not due to a single feature, the question is whether it could be due to a combination of features and if so, what the combination is. As the feature [− sqa] is persistently associated with unboundedness in the literature, let us consider the idea that it is one of the ingredients necessary to this interpretation after all. If this is correct, it implies that quantitative du/des-NPs are excluded from atelic constructions. In effect, by definition, quantitative means [+ sqa], a feature leading to a telic reading. This is however a welcome result: recall from Chapter 3 Section 3.2.1 that du/des-NPs are not totally excluded from bounded contexts. This is illustrated below.

(7) J’ ai reconnu des filles dans la rue. I have recognised of.the girls in the street

(French)

That the above sentence may be telic is supported by the ungrammaticality of the for-adverbial:

(8) *J’ ai reconnu des filles dans la rue pendant 10 minutes. I have recognised of.the girls in the street for 10 minutes

Although du/des-NPs are generally involved in atelic environments, examples like (8) show that this is not necessarily the case. Put differently, du/des-NPs are possible in both bounded and unbounded contexts. However, as they may be quantitative or not, the conclusion that the feature [− sqa] plays a role in atelicity may be maintained: when



Chapter 5.  Conclusion 

du/des-NPs do not encode a quantity, they lead to an unbounded reading and when they encode a quantity, they lead to a bounded reading. This suggests that the nominals typically involved in atelic contexts are the following:

(9) Example a. du painnon-quantit. b. des painsnon-quantit. c. books d. bread

Number FP3 de + lesing de + lesplur − s √

Quantity FP2 deP X de X de X X X X

Count FP1 X − s X X

Lexical category NP pain pain book bread

If the above discussion is on the right track, the question which immediately comes to mind is what differentiates the nominals in (9) from the ones in (6), which do not involve a quantity either. A first element of answer is that the former have all been examined in this book, i.e., they are all indefinites. In other words, indefiniteness may play a role in unboundedness, as proposed by De Swart (2006). The combination of features leading to such an interpretation would thus be indefiniteness and lack of quantity. This generalisation however is problematic for un-NPs, which share both of these characteristics with the nominals in (9), but which are involved in telic sentences. The solution to this puzzle might come from the other features of un-NPs. Recall that it has been proposed in Chapter 3 (Section 3.2) that the value ‘one’ of these nominals may be inferred from their being count and singular (see Farkas 2006 who proposes that singulars (including definites) have atomic reference by default). In other words, the features [singular] and [count] (which implies individuated reference) are sufficient to set the value [sqa] of un-NPs positively, and thus to lead to boundedness. In sum, the above discussion suggests that indefiniteness and lack of quantity may be the indispensable features of nominals involved in atelic constructions and that this generalisation holds for non-quantitative du-NPs, non-quantitative des-NPs and bare nouns, be they plural or mass. An analysis very close to the one mentioned above has been proposed by De Swart (2006) for des-NPs. According to her, it is the combination of weak quantificational force and of lack of determined reference which leads sentences involving des-NPs to an unbounded reading. The weak quantificational force of des-NPs is the result of two factors, (1) their being semantically plural (relying on Farkas 2006, contra Spector 2003 and Roodenburg 2004b for example) and (2) the lack of numerical/quantificational value

.  Her account is mentioned in more detail below. .  Conversely, a bounded reading may be due to determined reference or strong quantificational force. In other words, the telic interpretation of sentences with definites and cardinals, for example, is due to the determined reference and the cardinality of these nominals respectively.

 The Layered DP: Form and meaning of French indefinites

contributed by the article (2006: 166). Their lack of determined reference contrasts with definites which have ‘determined reference’, i.e., the choice of their referent is fixed (Farkas 2002). The determined reference of singular definites is due to uniqueness (10a) or anaphoricity (10b) (Farkas 2002) and the one of plural definites to maximal reference within the context (11a,b) or anaphoricity (11c) (De Swart & Farkas 2005): (10) a. The planets turn around the sun. b. A man and a woman came in. The man sat down.  (De Swart 2006: 168 (9)) (11) a. b. c.

The planets revolve around the sun. (Looking at a painting that has two apples in the foreground). The apples are beautiful! Mary bought applesi. The applesi are on the countertop.

(ibid, (10))

As for un-NPs, De Swart (2006) suggests that these nominals lead to a telic interpretation because they have strong quantificational force. The reason is that singulars have atomic reference by default (Farkas 2006). Although this is not the place to compare De Swart’s proposal with the one sketched above on the basis of the interpretative features mentioned in (9) and (6), it would be interesting to see whether De Swart’s account extends to bare nouns (mass and plural) and to du-NPs excluded from the paper. This question is particularly significant as the analysis involves a mapping/oneto-one relation from individuals to events, i.e., a notion which has not been associated to bare nouns and du-NPs in the framework developed here. De Swart’s (2006) paper also deals with des-NPs in habitual sentences. This is where grammatical aspect comes into play. As mentioned at the beginning of the section, grammatical aspect is a classification of different forms of a particular verb. The most familiar distinction is between perfective and imperfective aspects. In a nutshell, the former represents a single event conceived as a unit, whereas the latter refers to a habitual or repeated event, or to an event in the process of unfolding. Put differently, a perfective aspect denotes an event bounded in time, whereas an imperfective aspect denotes an event unbounded in time. A typical contrast discussed in the literature is between episodic and habitual readings: (12) a. b.

Paul répare des bicyclettes. (Hab./Epis.) Paul repairs/is repairing of.the bikes Paul répare une bicyclette. (*Hab./Epis.) Paul repairs a bike ‘Paul is repairing a bike’. (De Swart 2006: 182 (29–30))

.  The paper also discusses different semantic issues which would be out of the scope of this short section.



Chapter 5.  Conclusion 

(12a) may assert the existence of a certain habit or have an episodic interpretation. In (12b), in contrast, the habitual reading is blocked. De Swart offers a semantic account for this contrast building on her conclusions on the interpretation of des-NPs, namely their weak quantificational force and their lack of determined reference, as defined above. In other words, the habitual reading is possible only if the object has these two characteristics. The analysis thus takes into account the predicate and the object. What De Swart proposes (along the lines of Bosveld-de Smet 1998) is that it is the combination of a predicate interpreted as a bijection and the cumulative meaning of the object which is responsible for the habitual reading. As the details of the study would take us too far afield, the reader is referred to the original paper. To conclude on this work, it is worth mentioning again that the issue of du-NPs would be interesting to investigate, especially if the bijection is a one-to-one relation between individuals and events. An example illustrating du-NPs is provided below: (13) a. b.

Paul répare du matériel électronique. Paul repairs/is repairing of.the material electronic Paul fume du cannabis. Paul smokes/is smoking of.the cannabis

(Hab./Epis.) (Hab./Epis.)

(13) suggests that du-NPs, on a par with des-NPs, are possible in both habitual and episodic contexts. If this is correct, a unifying analysis is necessary. The interrelation between verbal and nominal characteristics is also investigated by Guéron (2006). She examines episodic, i.e., perfective, and generic, i.e., imperfective, sentences and establishes a correlation between the type of tense node involved and the interpretation of the subject. More precisely, she proposes that “[t]he DP subject of an unindexed Tense node is itself construed as generic while that of an episodic Tense node is construed as referential” (2006: 219). Recall in effect that one of Guéron’s proposals is that the difference between generic and non-generic sentences lies in the tense node (Chapter 4 Section 5.1). The idea is that the generic reading results from a tense node which has an index 0 (or no index) whereas the non-generic one is due to an indexed tense. If this analysis is on the right track, it suggests that episodicity may be related to referential marking whereas genericity implies non-referentiality. This is illustrated below: (14) a. The dog is barking. b. The dog barks. 

(episodic) (generic) (Guéron 2006: 222 (10))

(15) a. Dogs are barking. b. Dogs bark. 

(episodic) (generic) (ibid, (11))

(16) a. A dog is barking. b. A dog barks. 

(episodic) (generic) (ibid, (12))

 The Layered DP: Form and meaning of French indefinites

As (14a), (15a) and (16a) are episodic, their tense nodes are indexed (T1 for example) and their subjects referential in the sense that they represent a real dog. (14b), (15b) and (16b) in contrast are generic and hence have an unindexed tense node (T0) and a non-referential subject in the sense that it is interpreted as a mental concept. More generally, this suggests that the perfective aspect (episodic) examined should be associated with a referential subject, as defined by Guéron. Note however, that this notion seems to be close to what has been called existentiality in the framework developed here. In other words, dogs in (15a) would be an existential Property Phrase and a dog in (16a) either an existential Property Phrase if the dog cannot be identify or an S-Reference Phrase if the dog is a dog in particular. The fact that a dog in (16a) may be interpreted in two different ways suggests that a binary classification of nominals, as referential and non-referential for example, is not fine grained enough to capture the subtle differences of meaning illustrated by the data (this was already observed in Chapter 4 Section 5.1). Furthermore, (14)–(16) show that different tense nodes, i.e., T1 and T0, cannot be associated with a specific nominal feature: the subjects of the perfective sentences (i.e., episodic examples with T1) for instance may be definite (14a) or indefinite (15a), (16a), singular (14a), (16a) or plural (15a). Although this section shows that there are links across categories and that the aspectual characteristics of a verb interact with the properties of DP as well as with the ones of TP, many issues should be further investigated. As we will see in the next section, one of them is the role of case in languages without determiners.

2.2  Languages without determiners This section offers a glimpse at determinerless languages. It first carries on the discussion of unboundedness presented in Section 2.1 before raising more general issues related to the analysis developed in this book. The aim is to show that the questions addressed in preceding chapters also concern determinerless languages and that the absence of determiners may be the source of additional problems (Abraham, Stark & Leiss 2007). A well-known fact about languages which do not possess overt determiners is that case has an impact on the aspectual nature of the sentence as a whole. Kiparsky (1998) for example proposes that in Finnish unboundedness is associated with partitive case. More precisely, he suggests that partitive case is either assigned to an object which is quantitatively undetermined or to an object which is governed by an

.  Note that the term ‘referential’ here does not correspond to De Swart’s (2006) determined reference nor to the concept dealt with in this book. .  Guéron (2006) also assumes that there is an interaction between aspect and DPs. She concentrates on bare plurals and makes a distinction between constructions which allow such nominals in their subject position and constructions which do not. .  Partitive subjects will be ignored here. The reader is referred to the original paper.



Chapter 5.  Conclusion 

atelic verb. This means that the complement of an unbounded verb must be partitive whether it is itself bounded or not. This is summarised in Table 12: (17) Table 12.  Adapted from Kiparky 1998: 19 (42) (Russian examples ignored)

[+ B] object

[− B] object

[+ B] verb (telic) [− B] verb (atelic)

kirjoitti kirjeet (acc.) ‘wrote the letters’ kirjoitti kirjeitä (part.) ‘was writing the letters’

kirjoitti kirjeitä (part.) ‘wrote letters’ kirjoitti kirjeitä (part.) ‘was writing letters’

The examples with a partitive object are in bold in the above table. They take into account both the predicate and the object. Interestingly, the column on the right corresponds to the cases of unboundedness studied in the preceding section. As the English translation shows, these sentences involve bare nouns, i.e., indefinites which do not involve a quantity. The example on the bottom line of the central column however has not been considered so far. It comprises a bounded object and an atelic verbal predicate. Although such data will not be considered here, they suggest that the partitive case associated with unboundedness cannot be related to a nominal feature like (in)definiteness for example (see Abraham, Stark & Leiss 2007 and the references therein for a different analysis). In effect, the partitive object in was writing the letters is definite whereas in wrote letters and was writing letters it is indefinite. Further evidence for this conclusion comes from the example with a telic verb and a bounded object, namely wrote the letters. Here the definite object has accusative case. That case doesn’t correspond to a nominal feature in particular is further supported by sentences with stative verbs. In effect, the indefinite object of such predicates may be marked with accusative or partitive case depending on the (lack of) quantity it involves. Consider the Finnish example (18) from Kiparsky 1998: 5–6(8c–f): (18) a. b.

Aki-lla on iso-t Aki-adess have-3sg big-pl.acc silmä-t / viikse-t eyes-pl.acc / mustache-pl.acc ‘Aki has big eyes / a mustache.’ Aki-lla on iso-j-a Aki-adess have-3sg big-pl-part silm-i-ä / viiksi-ä eyes-pl-part / mustache-pl.part ‘Aki has big eyes / mustaches in his possession.’

.  To be more precise, it is one sentence which is three-way ambiguous.

 The Layered DP: Form and meaning of French indefinites

c. d.

Vauva-lla on pitkä-t hiukse-t baby-adess have-3sg long-pl.acc hair-pl.acc ‘The/a baby’s hair is long.’ Vauva-lla on pitk-i-ä hiuks-i-a baby-adess have-3SG long-pl-part hair-pl-part ‘The/a baby has (some) long strands of hair’ (on its head, in its hand, etc.).’

The accusative case in (18a) implies inalienable possession. In other words the eyes and the mustache must be Aki’s and there must be two of the former and one of the latter (speaking of ‘normal’ human beings in our world). (18b), in contrast, implies “an indeterminate number of alienably possessed objects, such as glass eyes or anatomical samples in a vat, false mustaches (of which there must now be several), etc.” (Kiparsky 1998: 6), hence the partitive case. Similarly, (18c) refers to the totality of hair belonging to the baby, whereas (18d) refers to some hair which could belong to the baby, or to loose strands of hair in the baby’s possession whose number is indeterminate. The difference of case assigned to the relevant objects depends on the (lack of) quantity involved and more generally, as seen above, on their being bound or not: when the set is clearly delimited, the complement bears accusative case, and when the quantity is indeterminate, it bears partitive case. The contrast in (18) suggests that the complements illustrated are analogous to du/des-NPs in that they may be quantitative or not. If this is correct, the arguments with accusative case should have FP2 in their structure in contrast with the partitive objects (Chapter 3 Section 3.2.2). The above discussion raises many questions on the interrelation between interpretative features like (un)boundedness, (in)definiteness and case. For example, the mechanisms involved in the interaction between (un)boundedness and case have to be determined. One reason is that unboundedness may result from the verbal predicate or from the object (recall (17)). This suggests that a distinction between structural and inherent case might be relevant: an unbounded verb would assign structural partitive case to its object, whereas an unbounded object would have inherent partitive case (see Kiparsky 1998 for further discussion). Even if this is on the right track, it is unclear how a property like quantity (or lack thereof) may have an impact on the case marked on the DP. In addition, if unbounded objects must be indefinite as suggested in the preceding section, whether this feature plays a role in determining the case on the nominal expression has to be determined. More general issues related to determinerless languages should also be addressed (see Abraham, Stark & Leiss 2007 and the references therein). One of them is determination. In effect, although this section only briefly mentions Finnish, it is clear from the glosses that nominals in this language correspond to the-NPs, a-NPs and bare nouns in English. This implies that concepts like determination and (in)definiteness cannot be ignored in the study of such languages. It also means that these notions are not (necessarily) dependent on the presence of a determiner. Put differently, information



Chapter 5.  Conclusion 

concerning the referential status of nominals for example has to be accessible in the absence of the article. The question is whether such interpretative features are encoded in the structure of languages which do not possess overt determiners, i.e., whether a D-system is available or not. Although this issue is debated in the literature, the framework developed in this book suggests the affirmative. In effect, the thesis defended here is that it is the structure which is responsible for the interpretation of nominals. For instance, (S-)referentiality is encoded in a specific layer and indefiniteness is associated with three types of nominals, each represented by a different syntactic category. As such an analysis seems to extend to bare nouns in Germanic and Romance (Chapter 4), it is conceivable that it is also applicable to languages which do not have overt determiners at all. However, to determine whether this is on the right track or not represents a new expedition. This is why our journey stops here.

Bibliography Abbott, Barbara. 1996. Doing without a partitive constraint. In J. Hoeksema (ed.), Partitives, 25–56. Berlin: Mouton de Gruyter. Abbott, Barbara. 2004. Definiteness and indefiniteness. In L.R. Horn & G. Ward (eds), Handbook of pragmatics, 122–149. Oxford: Blackwell. Abeillé, Anne, Olivier Bonami, Danièle Godard & Jesse Tseng. 2004. The syntax of French de-N’ phrases. In S. Müller (ed.), Proceedings of the HPSG04 Conference. Stanford CA: CSLI. Abney, Steven P. 1987. The English noun phrase in its sentential aspect. Ph.D. dissertation, MIT. Aboh, Enoch. 1998. On the syntax of Gungbe noun phrases. Clearinghouse on Languages and Linguistics. ERIC Documentation Reproduction Service n° ED 420 209. Aboh, Enoch. 1999. From the syntax of Gungbe to the grammar of Gbe. Sierre, Suisse: Editions à la carte S.A. Aboh, Enoch. 2001. Morphosyntaxe de la périphérie gauche nominale. Presented at the Conference “La syntaxe de la définitude”, Université de Paris 8, Feb. 8–9. Aboh, Enoch. 2003. The morphosyntax of complement-head sequences. Oxford: OUP. Aboh, Enoch. 2004. Topic and focus within D. Linguistics in the Netherlands 21: 1–12. Abraham, Werner, Elisabeth Stark & Elisabeth Leiss. 2007. Introduction. In E. Stark, E. Leiss & W. Abraham (eds), Nominal determination. Typology, genesis, and historical emergence of DP. Amsterdam: John Benjamins. Abusch, Dorit. 1994. The scope of indefinites. Natural Language Semantics 2: 83–136. Attal, Paul. 1976. A propos de l’indéfini des: Problèmes de représentation sémantique. Le Français Moderne 2: 126–142. Baker, Mark. 1988. Incorporation. Chicago IL: University of Chicago Press. Barwise, Jon & Robin Cooper. 1981. Generalized quantifiers and natural language. Linguistics and Philosophy 4(2): 159–219. Baunaz, Lena. 2005a. The syntax and semantics of wh in-situ and existentials: The case of French. Leiden Working Papers in Linguistics 2(2): 1–27. Baunaz, Lena. 2005b. Un NPs and Wh in situ: An argument for an indefinite analysis. Generative Grammar in Geneva GG@G 4: 1–43. Baunaz, Lena & Andrea Cattaneo. 2005. Personne ‘nobody’: An instance of universal quantification. Ms. University of Geneva. Beghelli, Filippo. 1995. The phrase structure of quantifier scope. Ph.D. dissertation, UCLA. Beghelli, Filippo & Tim Stowell. 1994. The direction of quantifier movement. GLOW Newsletter 32: 56–57. Beghelli, Filippo & Tim Stowell. 1997. Distributivity and negation: The syntax of each and every. In A. Szabolcsi (ed.), Ways of scope taking, 71–108. Dordrecht: Kluwer. Belletti, Adriana. 1979. Sintagmi nominali quantificati e strutture dislocate a sinistra. Annali della Scuola Normale Superiore, Classe di Lettere e Filosofia, Serie III, Vol. IX(4): 1525–1568. Pisa. Belletti, Adriana (ed.). 2004. Structures and beyond. The cartography of syntactic structures. Oxford: OUP.

 The layered DP: Form and meaning of French indefinites Bennett, Michael & Barbara Partee. 1978. Toward the logic of tense and aspect in English. Ms, Indiana University Linguistic Club. Bernstein, Judy. 1991. DPs in French and Walloon: Evidence for parametric variation in nominal head movement. Probus 3: 1–26. Bernstein, Judy. 1993. Topics in the syntax of nominal structure across Romance. Ph.D. dissertation, City University of New York. Bernstein, Judy B. 2001. The DP hypothesis: Identifying clausal properties in the nominal domain. In M. Baltin & C. Collins (eds), The handbook of contemporary syntactic theory, 536–561. Oxford: Blackwell. Bhatt, Rakesh & James Yoon. 1992. On the composition of COMP and parameters of V2. In Proceedings of the West Coast Conference on Formal Linguistics. Bhattacharya, Tanmoy. 1998. DP-internal movement. Working Papers in Linguistics 10. London: University College London. Borer, Hagit. 1998. Deriving passive without theta roles. In S.G. Lapointe, D.K. Brentari & P.M. Farrells (eds), Morphology and its relation to phonology and syntax, 60–99. Stanford CA: CSLI. Borer, Hagit. 2005a. In names only. Oxford: OUP. Borer, Hagit. 2005b. The normal course of events. Oxford: OUP. Bosveld-de Smet, Leonie. 1998. On mass and plural quantification: The case of French des/ du-NPs. ILLC dissertation series/Groningen dissertation series in linguistics, Amsterdam/ Groningen. Bosveld-de Smet, Leonie. 2004. Towards a uniform characterization of noun phrases with des or du. In F. Corblin & H. De Swart (eds), Handbook of French semantics, 41–54. Stanford CA: CSLI. Bowers, John. 1988. Extended X-bar theory, the ECP and the left branching condition. Proceedings of WCCFL-7, Vol. 7, 46–62. Stanford CA: Stanford Linguistics Association. Bowers, John. 1993. The syntax and semantics of nominals. Proceedings of SALT-III, Vol. 3: 1–30. Brame, Michael. 1982. The head-selector theory of lexical specifications and the nonexistence of coarse categories. Linguistic Analysis 10(4): 321–5. Bresnan, Joan. 1973. Syntax of the comparative clause construction in English. Linguistic Inquiry 4: 275–343. Bunt, Harry C. 1985. Mass terms and model-theoric semantics. Cambridge: CUP. Büring, Daniel. 1996. A weak theory of strong readings. In T. Galloway & J. Spence (eds), Proceedings from SALT VI, 17–34. Ithaca NY: Cornell University. Campbell, Richard. 1996. Specificity operators in SpecDP. Studia Linguistica 2: 161–188. Cardinaletti, Anna. 1993. On the internal structure of pronominal DPs. University of Venice Working Papers in Linguistics 3(1). Cardinaletti, Anna & Giuliana Giusti. 1992. Partitive ne and the QP-hypothesis. In E. Fava (ed.), Proceedings of the XVII Meeting of Generative Grammar, 121–141. Turin: Rosenberg and Sellier. Cardinaletti, Anna & Giuliana Giusti. 2002. The syntax of quantified phrases and quantitative clitics. http://www-uilots.let.uu.nl/syncom/uiltjes/case_131_qcon.htm Carlier, Anne. 1989. Généricité du syntagme nominal sujet et modalités. Travaux de Linguistique 19: 33–56. Carlson, Greg N. 1977a. A unified analysis of the English bare plural. Linguistics and Philosophy 1: 413–56.



Bibliography 

Carlson, Greg N. 1977b. Reference to kinds in English. Ph.D. dissertation, University of Massachusetts at Amherst. Carlson, Greg N. & Francis J. Pelletier (eds). 1995. The generic book. Chicago IL: University of Chicago Press. Casalegno, Paolo. 1987. Sulla logica dei plurali. Teoria 2: 125–143. Chastain, Charles. 1975. Reference and context. In K. Gunderson (ed.), Language, mind and knowledge [Minnesota Studies in the Philosophy of Science 7], 194–269. Minneapolis MN: University of Minnesota Press. Cheng, Lisa & Rint Sybesma. 1999. Bare and not so bare nouns and the structure of NP. Linguistic Inquiry 30: 509–542. Chierchia, Gennaro. 1995. Plurality of mass nouns and the notion of Semantic Parameter. Ms, University of Milan and DIPSCO. Chierchia, Gennaro. 1998a. Partitives, reference to kinds and semantic variation. In A. Lawson (ed.), Proceedings of Semantics and Linguistic Theory VII, 73–98. Ithaca NY: CLC Publications. Chierchia, Gennaro. 1998b. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton. Chomsky, Noam. 1965. Aspects of the theory of syntax. Cambridge MA: The MIT Press. Chomsky, Noam. 1970. Remarks on nominalization. In R. Jacobs & P. Rosenbaum (eds), Readings in English transformational grammar, 184–221. Waltham MA: Ginn. Chomsky, Noam. 1979. Language and responsibility (based on conversations with Misou Ronat). New York NY: Pantheon. Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris. Chomsky, Noam. 1986. Barriers. Cambridge MA: The MIT Press. Chomsky, Noam. 1993. A minimalist program for linguistic theory. In K. Hale & S.J. Keyser (eds), The view from building 20. Essays in honor of Sylvain Bromberger, 1–52. Cambridge MA: The MIT Press. Chomsky, Noam. 1995. The minimalist program. Cambridge MA: The MIT Press. Chomsky, Noam. 2000. Minimalist inquiries: The framework. In R. Martin, D. Michaels & J. Uriagereka (eds), Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge MA: The MIT Press. Chomsky, Noam. 2001. Derivation by phase. In M. Kenstowicz (ed.), Ken Hale: A life in language, 1–52. Cambridge MA: The MIT Press. Chomsky, Noam. 2004. Beyond explanatory adequacy. In A. Belletti (ed.), Structures and beyond: The cartography of syntactic structures, Vol. 3, 104–131. Oxford: OUP. Chomsky, Noam. 2005. Three factors in language design. Linguistic Inquiry 36: 1–22. Chomsky, Noam. 2006. On phases. In R. Freidin, C.P. Otero & M.-L. Zubizaretta (eds), Foundational issues in linguistic theory. Cambridge MA: The MIT Press. Chung, Sandra & William A. Ladusaw. 2004. Restriction and saturation. Cambridge MA: The MIT Press. Cinque, Guglielmo. 1990. Types of A’ dependencies. Cambridge MA: The MIT Press. Cinque, Guglielmo. 1992. Evidence for partial N movement in the Romance DP. GLOW newsletter 28: 12–13. Paper presented at GLOW 15. Cinque, Guglielmo. 1994. On the evidence for partial N-movement in the Romance DP. In G. Cinque, J. Koster, J.-Y. Pollock, L. Rizzi & R. Zanuttini (eds), Paths towards Universal Grammar. Studies in honour of Richard Kayne, 85–110. Washington DC: Georgetown University Press.

 The layered DP: Form and meaning of French indefinites Cinque, Guglielmo. 1996. The ‘antisymmetric’ programme: Theoretical and typological implications. Journal of Linguistics 32: 447–464. Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford: OUP. Cinque, Guglielmo. 2005. Deriving Greenberg’s Universal 20 and its exceptions. Linguistic Inquiry 36: 315–332. Cinque, Guglielmo (ed.). 2002. Functional structure in the IP and DP. The cartography of syntactic structures, Vol. 1. Oxford: OUP. Clarke, Melody. 2001. Aspect, negation and adjunct phrases. Poster session, Perspectives on aspect, December, Utrecht. Clédat, Lucien. 1901. La préposition et l’article partitif. Revue de philologie française et de littérature XV: 81–131. Coene, Martine & Yves D’Hulst. 2003. From NP to DP. Vol. 2. Amsterdam: John Benjamins. Comorovski, Ileana. 1996. Interrogative phrases and the syntax-semantics interface. Cambridge MA: The MIT Press. Contreras, Heles. 1986. Spanish bare NPs and the ECP. In I. Bordelois, H. Contreras & K. Zagona (eds), Generative studies in Spanish syntax, 25–49. Dordrecht: Foris. Cresti, Diana. 1995. Extraction and reconstruction. Natural Language Semantics 3: 79–122. Damourette, Jacques & Edouard Pichon. 1952. Des mots à la pensée. Essai de grammaire de la langue française, Tome I, 467–528. Paris: D’Artrey. Davidson, Donald. 1967. The logical form of action sentences. In N. Rescher (ed.), The logic of decision and action. Pittsburgh PA: University of Pittsburgh Press. (Reprinted in Davidson 1980, Essays on action and events, 105–122. Oxford: Clarendon Press). Dayal, Vaneeta. 1999. Bare NPs, reference to kinds, and incorporation. Proceedings of SALT IX. Ithaca NY: Cornell University, Cornell Linguistics Club. De Hoop, Helen. 1992. Case configuration and noun phrase interpretation. Ph.D. dissertation, University of Groningen. De Jong, Franciska & Henk Verkuyl. 1985. Generalized quantifiers: The properness of their strength. In J. van Benthem & A.G.B. ter Meulen (eds), Generalized quantifiers in natural language, 21–43. Dordrecht: Foris. De Swart, Henriëtte. 2006. Aspectual implications of the semantics of plural indefinites. In S. Vogeleer & L. Tasmowski (eds), Non-definiteness and plurality, 161–189. Amsterdam: John Benjamins. De Swart, Henriëtte & Donka Farkas. 2005. Généricité et indéfinitude. Une analyse dans la théorie de l’optimalité. In C. Dobrovie-Sorin (ed.), Noms nus et généricité, 97–126. Paris: Presses universitaires de Vincennes. Delfitto, Denis 1993. A propos du statut lexical de l’article partitif en français: Quelques hypothèses sur l’interaction entre morphologie et forme logique. In A. Hulk et al. (eds), Du lexique à la morphologie: Du côté de chez Zwaan, 103–116. Amsterdam: Rodopi. Delfitto, Denis & Jan Schroten. 1991. Bare plurals and the number affix in DP. Probus 3(2): 155–185. Delsing, Lars-Olof. 1988. The Scandinavian noun phrase. Working Papers in Scandinavian Syntax 42: 57–79. Delsing, Lars-Olof. 1993. The internal structure of noun phrases in the Scandinavian languages. A comparative study. Ph.D. dissertation, Lund University. Den Dikken, Marcel. To appear. Phase extension: A reply. Theoretical Linguistics. Diesing, Molly. 1992. Indefinites. Cambridge MA: The MIT Press.



Bibliography 

Diesing, Molly & Eloise Jelinek. 1996. Distributing arguments. Natural Language Semantics 3(2): 123–176. Dobrovie-Sorin, Carmen. 1996. Syntactic configurations and references: SE/SI in Romance. In K. Zagona (ed.), Grammatical theory and Romance languages, 73–86. Amsterdam: John Benjamins. Dobrovie-Sorin, Carmen. 1997a. Existential arguments and existential predicates. In A. Lawson (ed.), Proceedings of SALT VII. Ithaca NY: CLC Publications. Dobrovie-Sorin, Carmen. 1997b. Classes de prédicats, distribution des indéfinis et la distinction thétique-catégorique. Le gré des langues, 58–97. Paris: L’Harmattan. Dobrovie-Sorin, Carmen & Brenda Laca. 2003. Les noms sans déterminant dans les langues romanes. In D. Godard (ed.), Les langues romanes, 235–281. Paris: CNRS Editions. Dobrovie-Sorin, Carmen & Claire Beyssade. 2004. Définir les indéfinis. Paris: CNRS Editions. Dobrovie-Sorin, Carmen, Tonia Bleam & M. Teresa Espinal. 2006. Bare nouns, number and types of incorporation. In S. Vogeleer & L. Tasmowski (eds), Non-definiteness and plurality, 51–79. Amsterdam: John Benjamins. Doetjes, Jenny. 1997. Quantifiers and selection. On the distribution of quantifying expressions in French, Dutch and English. Ph.D. dissertation, Holland Institute of Generative Linguistics, Leiden University. Doetjes, Jenny & Johan Rooryck. 2003. Generalizing over quantitative and qualitative constructions. In M. Coene & Y. D’hulst (eds), From NP to DP, Vol. 1, 277–295. Amsterdam: John Benjamins. Donnellan, Keith. 1966. Reference and definite descriptions. Philosophical Review 75: 281–304. Donnellan, Keith. 1978. Speaker reference, descriptions and anaphora. In P. Cole (ed.), Syntax and semantics, Vol. 9: Pragmatics, 47–68. New York NY: Academic Press. Dowty, David. 1979. Word meaning and Montague grammar. Dordrecht: Reidel. Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22: 1–25. Englebert, Annick. 1992. Le petit mot DE. Etude de sémantique historique. Genève: Droz. Epstein, Melissa A. 2001. On the singular indefinite article in English. Ms, UCLA. Farkas, Donka. 1981. Quantifier scope and syntactic island. CLS 17: 59–66. Farkas, Donka. 1997. Evaluation indices and scope. In A. Szabolcsi (ed.), Ways of scope taking, 183–215. Dordrecht: Kluwer. Farkas, Donka. 2002. Specificity distinctions. Journal of Semantics 19: 1–31. Farkas, Donka. 2006. The unmarked determiner. In S. Vogeleer & L. Tasmowski (eds), Nondefiniteness and plurality, 81–105. Amsterdam: John Benjamins. Farkas, Donka & Henriëtte de Swart. 2003. The semantics of incorporation. Stanford CA: CSLI. Fodor, Janet Dean & Ivan A. Sag. 1982. Referential and quantificational indefinites. Linguistics and Philosophy 5: 355–398. Frampton, John. 1991. Relativized minimality: A review. The Linguistic Review 8: 1–46. Frege, Gottlob. 1892. Über Sinn and Bedeutung. Zeitschrift für Philosophie und Philosophische Kritik 100: 22–50. Frei, Henri. 1960. Tranches homophones: A propos de l’article partitif. Word 3: 317–322. Frontier, Alain. 1997. La grammaire du français. Paris: Belin. Gallego, Angel J. 2006. Phase sliding. Ms, UAB/UMD. Galmiche, Michel. 1986. Référence indéfinie, événements, propriétés et pertinence. In J. David & G. Kleiber (eds), Déterminants: Syntaxe et sémantique, 41–71. Paris: Klincksieck.

 The layered DP: Form and meaning of French indefinites Giannakidou, Anastasia. 1997. The landscape of polarity items [Groningen Dissertations in Linguistics 18]. Groningen: University of Groningen. Gillon, Brendan. 1990. Towards a common semantics for English count and mass nouns. Linguistics and Philosophy 15: 597–639. Giusti, Giuliana. 1993. La sintassi dei determinanti. Padova: Unipress. Giusti, Giuliana. 1994. Enclitic articles and double definiteness: A comparative analysis of nominal. Linguistic Review 11: 231–255. Giusti, Giuliana. 1995. A unified structural representation of abstract case and article. Evidence from Germanic. In H. Haider, S. Olsen & S. Vikner (eds), Studies in comparative Germanic syntax, 77–93. Dordrecht: Kluwer. Giusti, Giuliana. 1997. The categorial status of determiners. In L. Haegeman (ed.), The new comparative syntax, 95–123. Harlow: Addison Wesley Longman. Giusti, Giuliana. 2001. The birth of a functional category: From Latin ILLE to the Romance article and personal pronoun. In G. Cinque & G. Salvi (eds), Current studies in Italian syntax. Esseys offered to Lorenzo Renzi, 157–171. Amsterdam: Elsevier. Giusti, Giuliana. 2005. At the left periphery of the Romanian noun phrase. In M. Coene & L. Tasmowski (eds), On space and time in language. Cluj: Clusium. Giusti, Giuliana. 2006. Parallels in clausal and nominal periphery. In M. Frascarelli (ed.), Phases of interpretation, 151–172. Berlin: Mouton de Gruyter. Glasbey, Sheila. 2006. Bare plurals in object position: Which verbs fail to give existential readings, and why? In S. Vogeleer & L. Tasmowski (eds), Non-definiteness and plurality, 133–157. Amsterdam: John Benjamins. Goldsmith, John. 1979. Autosegmental phonology. New York NY: Garland. Greenberg, Joseph H. 1966. Some universals of grammar with particular reference to the order of meaningful elements. In J.H. Greenberg (ed.), Universals of language, 73–113. Cambridge MA: The MIT Press. Grimshaw, Jane. 1990. Argument structure. Cambridge MA: The MIT Press. Grimshaw, Jane. 1991. Extended projection. Ms, Brandeis University. Grohmann, Kleanthes K. 2003. Prolific domains. On the anti-locality of movement dependencies. Amsterdam: John Benjamins. Grohmann, Kleanthes K. & Liliane Haegeman. 2003. Resuming reflexives. Proceedings of the 19th Scandinavian Conference of Linguistics. Nordlyd 31(1): 46–62. Gross, Maurice. 1967. Sur une règle de cacophonie. Langages 7: 105–119. Guéron, Jacqueline. 2006. Generic sentences and bare plurals. In S. Vogeleer & L. Tasmowski (eds), Non-definiteness and plurality, 219–234. Amsterdam: John Benjamins. Haegeman, Liliane. 1994. Introduction to Government and Binding theory. 2nd ed. Oxford: Blackwell. Haegeman, Liliane. 2004. DP periphery and clausal periphery: Possessor doubling in West Flemish? Relating nominal periphery to clausal periphery. In D. Adger, C. de Cat & G. Tsoulas (eds), Peripheries, 211–239. Dordrecht: Kluwer. Haegeman, Liliane & Jacqueline Guéron. 1999. English grammar. London: Blackwell. Hawkins, John A. 1983. Word order universals. New York NY: Academic Press. Heim, Irene. 1982. The semantics of definite and indefinite NP. Ph.D. dissertation, University of Massachusetts, Amherst (Published in 1989 by Garland, New York NY). Heim, Irene. 1991. Artikel und Definitheit. In A. von Stechow & D. Wunderlich (eds), Semantik. Ein internationales Handbuch der zeitgenössischen Forschung, 487–535. Berlin: Mouton de Gruyter.



Bibliography 

Heim, Irene & Angelika Kratzer. 1998. Semantics in generative grammar. Oxford: Blackwell. Hellan, Lars. 1981. On semantic scope. In F. Heny (ed.), Ambiguities in intensional contexts, 47–81. Dordrecht: Reidel. Heycock, Caroline & Roberto Zamparelli. 2003. Coordinated bare definites. Linguistic Inquiry 34: 443–469. Heycock, Caroline & Roberto Zamparelli. 2005. Friends and colleagues: Plurality, coordination, and the structure of DP. Natural Language Semantics 13: 201–270. Higginbotham, James. 1985. On semantics. Linguistic Inquiry 16: 547–593. Higginbotham, James. 1987. Indefiniteness and predication. In E.J. Reuland & A.G.B. ter Meulen (eds), The representation of indefiniteness, 43–70. Cambridge MA: The MIT Press. Hockett, Charles F. 1948. Potawatomi I, II, III, IV. International Journal of American Linguistics 14: 1–10, 63–73, 139–149, 213–225. Hoeksema, Jacob. 1984. Partitives. Ms, University of Groningen. Hoeksema, Jacob. 1996. Partitives: Studies on the syntax and semantics of partitive and related constructions. Berlin: Mouton de Gruyter. Huang, James C.-T. 1982. Move WH in a language without WH movement. The Linguistic Review 1: 369–416. Hudson, W. 1989. Functional categories and the saturation of noun phrases. Proceedings of NELS-19: 207–222. Ihsane, Tabea. 2003a. A typology of possessive modifiers. In M. Coene & Y. D’Hulst (eds), From NP to DP, Volume II, 23–42. Amsterdam: John Benjamins. Ihsane, Tabea. 2003b. Demonstratives and reinforcers in Arabic, Romance and Germanic. In J. Lecarme (ed.), Research in Afroasiatic Grammar II. Selected papers from the Fifth Conference on Afroasiatic Languages, Paris, 2000, 263–285. Amsterdam: John Benjamins. Ihsane, Tabea. 2003c. Notes on bare arguments in French. Ms, University of Geneva. Ihsane, Tabea & Genoveva Puskás. 2001. Specific is not definite. Generative Grammar in Geneva GG@G, Vol. 2, 39–54. Ioup, Georgette. 1977. Specificity and the interpretation of quantifiers. Linguistics and Philosophy 1: 233–245. Jackendoff, Ray. 1968. Quantifiers in English. Foundations of Language 4(4): 422–442. Jackendoff, Ray. 1977. X’-Syntax. A study of phrase structure [Linguistic Inquiry Monograph 2]. Cambridge MA: The MIT Press. Kamp, Hans. 1981. A theory of truth and semantic representation. In Groenendijk et al. (eds), Formal methods in the study of language, 277–322. Amsterdam: Mathematical Centre. Kayne, Richard. 1975. French syntax. Cambridge MA: The MIT Press. Kayne, Richard. 1981. ECP extensions. Linguistic Inquiry 12: 93–133. Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge MA: The MIT Press. Keenan, Edward L. 1987. A semantic definition of ‘indefinite NP’. In E.J. Reuland & A.G.B. ter Meulen (eds), The representation of indefiniteness, 286–317. Cambridge MA: The MIT Press. Keenan, Edward L. & Jonathan Stavi. 1986. A semantic characterization of natural language determiners. Linguistics and Philosophy 9: 253–326. Kester, Ellen-Petra. 1993. The inflectional properties of Scandinavian adjectives. Studia Linguistica 47: 139–53. Kiparsky, Paul. 1996. Partitive case and aspect. Ms, Stanford University. Kiparsky, Paul. 1998. Partitive case and aspect. In M. Butt & W. Greuder (eds), The projection of arguments: Lexical and compositional factors, 265–308. Stanford CA: CSLI.

 The layered DP: Form and meaning of French indefinites Kleiber, Georges. 1981a. Problèmes de référence: Descriptions définies et noms propres. Paris: Klincksieck. Kleiber, Georges. 1981b. Relatives spécifiantes et relatives non-spécifiantes. Le Français Moderne 49: 216–233. Koopman, Hilda. 1999. The internal and external distribution of pronominal DPs. In D. Johnson & I. Roberts (eds), Beyond Principles and Parameters, 91–132. Dordrecht: Kluwer. Kratzer, Angelika. 1980. On the notional category of modality. In H. Eikmeyer & H. Rieser (eds), Words, worlds and contexts, 38–74. Berlin: Mouton de Gruyter. Kratzer, Angelika. 1986. Conditionals. In A.M. Farley, P. Farley & D.E. McCullough (eds), Papers from the parasession on pragmatics and grammatical theory, 1–15. Chicago IL: Chicago Linguistic Society. Kratzer, Angelika. 1989. Stage and individual level predicates. Papers on quantification. Department of Linguistics, University of Massachusetts at Amherst. (Reprinted in G.N. Carlson & F.J. Pelletier 1995). Kratzer, Angelika. 1996. Severing the external argument from the verb. In J. Rooryck & L. Zaring (eds), Phrase structure and the lexicon, 109–137. Dordrecht: Kluwer. Krifka, Mandfred. 1998. The origins of telicity. In S. Rothstein (ed.), Events and grammar, 197–236. Dordrecht: Kluwer. Krifka, Manfred et al. 1995. Genericity: An introduction. In G.N. Carlson & F.J. Pelletier (eds), The generic book, 1–124. Chicago IL: University of Chicago Press. Kripke, Saul. 1977. Speaker’s reference and semantic reference. In P.A. French, T.E. Uehling & H.K. Wettstein (eds), Contemporary perspectives in the philosophy of language, 6–27. Minneapolis MN: University of Minnesota Press. Kroch, Anthony. 1974. The semantics of scope in English. Ph.D. dissertation, MIT. Kroch, Anthony. 1989. Amount quantification, referentiality, and long wh-movement. Ms, University of Pennsylvania. Kupferman, Lucien. 1979. L’article partitif existe-t-il? Le français moderne 47: 1–16. Kupferman, Lucien. 1994. Du: Un autre indéfini? Faits de Langue 4: 195–203. Kupferman, Lucien. 1998. Des: Pluriel de du? In M. Bilger, K. v.d. Eynde & F. Gadet (eds), Analyse linguistique et approches de l’oral. Recueil d’études offert en hommage à Claire BlancheBenveniste, 229–238, Paris: Peeters. Kupferman, Lucien. 1999. Réflexions sur la partition: Les groupes nominaux partitifs et la relativisation. Langue française 122: 30–51. Kupferman, Lucien. 2001. Quantification et détermination dans les groupes nominaux. In X. Blanco, P.-A. Buvet & Z. Gavrilidou (eds), Détermination et formalisation, 219–234. Amsterdam: John Benjamins. Kupisch, Tanja & Christian Koops. 2007. The definite article in non-specific direct object noun phrases: Comparing French and Italian. In E. Stark, E. Leiss & W. Abraham (eds), Nominal determination. Typology, genesis, and historical emergence of DP, 189–213. Amsterdam: John Benjamins. Laca, Brenda. 1990. Generic objects: Some more pieces of the puzzle. Lingua 81: 25–46. Laca, Brenda. 1996. Acerca de la semántica de los ‘plurales escuetos’ del Español. In I. Bosque (ed.), El sustantivo sin determinación. La ausencia de determinación en la lengua Española, 241–268. Madrid: Visor Libros. Laca, Brenda & Liliane Tasmowski-de Ryck. 1996. Indéfini et quantification. Recherches linguistiques de Vincennes 25: 107–128. Ladusaw, William. 1982. Semantic constraints on the English partitive construction. In D. Flickinger, M. Macken & N. Wiegand (eds), Proceedings of the First West Coast Conference



Bibliography 

on Formal Linguistics, 231–242. Stanford CA: Stanford University. Ladusaw, William. 1994. Thetic and categorical, stage and individual, weak and strong. In M. Harvey & L. Santelmann (eds), Proceedings of Semantics and Linguistic Theory IV, 220–229. Ithaca NY: Cornell University. Laenzlinger, Christopher. 2000. More on adverb syntax. In A. Alexiadou & P. Svenonius (eds), Adverbs and adjunction [Linguistics in Potsdam 6], 103–132. Potsdam: ZAS. Laenzlinger, Christopher. 2005a. French adjective ordering: Perspectives on DP-internal movement types. Lingua 115: 645–689. Laenzlinger, Christopher. 2005b. Some notes on DP-internal movement. Generative Grammar in Geneva GG@G, Vol. 4, 227–260. Lakoff, George. 1970. Repartee. Foundations of Language 7: 389–422. Landman, Fred. 1989. Groups I and Groups II. Linguistics and Philosophy 12, 559–605 and 723–744. Landman, Fred. 1991. Structures for semantics. Dordrecht: Kluwer. Larson, Richard K. 1988. On the double object construction. Linguistic Inquiry 19: 335–392. Lasnik, Howard & Mamoru Saito. 1984. On the nature of proper government. Linguistic Inquiry 15: 235–289. Leben, William. 1973. Suprasegmental phonology. Ph.D. dissertation, MIT. Lewis, David. 1975. Adverbs of quantification. In E.L. Keenan (ed.), Formal semantics of natural language, 3–15. Cambridge: CUP. Link, Godehard. 1983. The logical analysis of plurals and mass terms: A lattice-theoretical approach. In R. Bäurle et al. (eds), Meaning, use, and interpretation of language, 302–323. Berlin: De Gruyter. Link, Godehard. 1991. Plural. In A. von Stechow & D. Wunderlich (eds), Semantik. Ein internationales Handbuch der zeitgenössischen Forschung, 418–440. Berlin: De Gruyter. Löbner, Sebastian. 2002. Is the German perfekt a perfect perfect? In G. Katz, S. Reinhard & P. Reuter (eds), Sinn and Bedeutung VI, Proceedings of the Sixth Annual Meeting of the Gesellschaft für Semantik, 369–391. Osnabrück: University of Osnabrück. Longobardi, Giuseppe. 1991. Proper names and the theory of N-movement in syntax and Logical Form. University of Venice Working Papers in Linguistics 9, Venezia. Longobardi, Giuseppe. 1994. Reference and proper names: A theory of N-Movement in syntax and Logical Form. Linguistic Inquiry 25: 609–665. Longobardi, Giuseppe. 2001. The structure of DPs: Some principles, parameters and problems. In M. Baltin & C. Collins (eds), The handbook of syntactic theory, 562–603. London: Blackwell. Longobardi, Giuseppe. 2003. Methods in parametric linguistics and cognitive history. Linguistic Variation Yearbook 3: 103–140. Longobardi, Giuseppe. 2005. Toward a unified grammar of reference. Zeitschrift für Sprachwissenschaft 24(1): 5–44. Ludlow, Peter & Stephen Neale. 1991. Indefinite descriptions: In defense of Russell. Linguistics and Philosophy 14: 171–202. Lumsden, Michael. 1988. Existential sentences. London: Croom Helm. Lyons, Christopher. 1999. Definiteness. Cambridge: CUP. Mandelbaum, Deborah. 1994. Syntactic conditions on saturation. Ph.D. dissertation, MIT. Manzini, Rita. 1992. Locality. Cambridge MA: The MIT Press. Marantz, Alec. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. In A. Dimitriadis & L. Siegel (eds), Proceedings of the 21st Annual Penn Linguistics Colloquium [University of Pennsylvania Working Papers in Linguistics], 201–225. Philadelphia: University of Pennsylvania.

 The layered DP: Form and meaning of French indefinites Martí Girbau, Núria. 2003. Partitives: One or two nouns? XXIX Incontro di Grammatica Generativa. Urbino, 13–15 February. Massam, Diane. 2001. Pseudo-incorporation in Niuean. Natural Language and Linguistic Theory 19: 153–197. Mathieu, Eric. 2006. Bare nouns and morpho-syntactic reflexes of semantic incorporation: Some new facts. In L. Bateman & Ch. Ussery (eds), Proceedings of NELS 35. Charleston SC: BookSurge Publishing. Matthewson, Lisa & Jeannette Schaeffer. 2000. Grammar and pragmatics in the acquisition of articles systems. In J. Gilkerson, M. Becker & N. Hyams (eds), Language Development and Breakdown 1 [Working Papers in Linguistics No. 5], 1–39. Los Angeles CA: UCLA. May, Robert. 1977. The grammar of quantification. Ph.D. dissertation, CUNY Graduate Center, distributed by Indiana University Linguistics Club. May, Robert. 1985 Logical Form: Its structure and derivation. Cambridge MA: The MIT Press. McCawley, James D. 1972. A program for logic. In D. Davidson & G. Harman (eds), Semantics, 498–544. Dordrecht: Reidel. McNally, Louise. 1992. An interpretation for the English existential construction. Ph.D. dissertation, UCSC. (Published in 1997 by Garland Press, New York NY). McNally, Louise. 1995. Bare plurals in Spanish are interpreted as properties. In G. Morrill & R. Oehrle (eds), Proceedings of the 1995 ESSLLI Conference on Formal Grammar. (Republished in 2004, Catalan Journal of Linguistics 3: 115–133). McNally, Louise. 1998. Existential sentences without existential quantification. Linguistics and Philosophy 31: 353–392. McNally, Louise. 2004. cf. 1995. McNally, Louise & Veerle Van Geenhoven. 1997. Redefining the weak/strong distinction. Paper presented at the Second Paris Syntax and Semantics Colloquium. Menn, Lise & Brian Mac Whinney. 1983. The repeated morph constraint: Toward an explanation. Language 60: 519–541. Milner, Jean-Claude. 1978. De la syntaxe à l’interprétation. Paris: Editions du Seuil. Milsark, Gary. 1974. Existential sentences in English. Ph.D. dissertation, MIT. (Published by the Indiana University Linguistics Club in 1976). Milsark, Gary. 1977. Toward an explanation of certain peculiarities of the existential construction in English. Linguistic analysis 3: 1–29. Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60: 847–894. Molnár, Valeria. 2002. Contrast – from a contrastive perspective. In H. Hasselgård, S. Johansson, B. Behrens, C. Fabricious-Hansen (eds), Information structure in a cross-linguistic perspective, 147–162. Amsterdam: Rodopi. Montague, Richard. 1974. Formal philosophy. Selected papers of Richard Montague. New Haven CT: Yale University Press. Munn Alan & Cristina Schmitt. 2005. Number and indefinites. Lingua 115: 821–855. Neale, Stephen. 1990 Descriptions. Cambridge MA: The MIT Press. Neeleman, Ad & Hans Van de Koot. 2004. Syntactic OCP effects. http://www.phon.ucl.ac.uk/ home/hans/haplology.html. Nolda, Andreas. 2006. De indéfini et de partitif. In F. Corblin, S. Ferrando & L. Kupferman (eds), Actes du colloque Indéfinis et prédication, 155–167. Presses de l’Université Paris-Sorbonne, Paris. Parsons, Terence. 1990. Events in the semantics of English. A study in subatomic semantics. Cambridge MA: The MIT Press.



Bibliography 

Partee, Barbara. 1986. Noun phrase interpretation and type-shifting principles. In J. Groenendijk, D. de Jong & M. Stokhof (eds), Studies in discourse representation theory and the theory of Generalized Quantifiers, 115–143. Dordrecht: Foris. Partee, Barbara H. 2002. Noun phrase interpretation and type-shifting pinciples. In P. Portner & B.H. Partee (eds), Formal semantics. Oxford: Blackwell. (First published in 1986. In J. Groenendijk, D. de Jongh & M. Stokhof (eds), Studies in discourse representation theory and the theory of Generalized Quantifiers, 115–143). Perlmutter, David. 1971. Deep and surface structure constraints in syntax. New York NY: Holt, Rinehart and Winston. Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In E. Reuland & A. ter Meulen (eds), The representation of indefiniteness, 98–129. Cambridge MA: The MIT Press Peyraube, Alain. 1998. On the history of classifiers in archaic and medieval Chinese. Studia Linguistica Seria: Proceedings of the 3rd International Conference on Chinese Linguistics, July 1994, 39–66. City University of Hong Kong. Plag, Ingo. 1998. The polysemy of –ize derivatives. The role of semantics in word formation. In G. Booij & J. Van Marle (eds), Yearbook of Morphology 1997, 219–242. Dordrecht: Foris. Platzack, Christer. 2001. Multiple interfaces. In U. Nikanne & E. van der Zee (eds), Cognitive interfaces. Constraints on linking cognitive information, 21–53. Oxford: OUP. Pollock, Jean-Yves. 1989. Verb movement, Universal Grammar and the structure of IP. Linguistic Inquiry 20: 365–424. Postal, Paul. 1966. On so-called ‘pronouns’ in English. In F. Dinneen (ed.), Monograph Series on Language and Linguistics, Vol. 19, 177–206. Washington DC: Georgetown University Press. Postal, Paul. 1974. On certain ambiguities. Linguistic Inquiry 5: 367–424. Pylkkänen, Liina. 1999. Causation and external arguments. In L. Pylkkänen, A. van Hout & H. Harley (eds), Papers from the UPenn/MIT Roundtable on the Lexicon [MITWPL 35], 161–183. Cambridge MA: MIT. Quine, Willard. 1960. Word and object. Cambridge MA: The MIT Press. Radford, Andrew. 1977. Counter-filtering rules. York papers in linguistics 7: 7–45. Radford, Andrew. 1979. The functional basis of transformations. Transactions of the Philological Society 1–42. Ramchand, Gillian. 2005. Verb meaning and the lexicon. Ms, University of Tromso. Ramchand, Gillian. 2006a. Against a generative lexicon. Lecture given at the University of Geneva. Ramchand, Gillian. 2006b. Predication and the internal structure of events. Lecture given at the University of Geneva. Ramchand, Gillian. 2006c. Verb meaning and the lexicon: A first phase syntax. Ms. University of Tromso. Ramchand, Gillian & Peter Svenonius. 2002. The lexical syntax and lexical semantics of the verb-particle construction. In L. Mikkelsen & C. Potts (eds), Proceedings of WCCFL 21. Somerville MA: Cascadilla. Reed, Ann. 1996. Partitives, existentials and partitive determiners. In J. Hoeksema (ed.), Partitives, 143–178. Berlin: Mouton de Gruyter. Reinhart, Tanya. 1995. Interface strategies. OTS Working Papers in Linguistics. Utrecht: Utrecht Institute of Linguistics, Utrecht. Reinhart, Tanya. 1997. Quantifier scope – How labor is divided between QP and choice functions. Linguistics and Philosophy 20: 335–397.

 The layered DP: Form and meaning of French indefinites Ritter, Elisabeth. 1988. A head-movement approach to construct-state noun phrases. Linguistics 26: 909–929. Ritter, Elisabeth. 1991. Two functional categories in noun phrases: Evidence from Modern Hebrew. In S. Rothstein (ed.), Perspectives on phrase structure [Syntax and Semantics 26], 37–62. New York NY: Academic Press. Rizzi, Luigi. 1986. Null objects in Italian and the theory of pro. Linguistic Inquiry 17: 501–557. Rizzi, Luigi. 1988. On the status of referential indices. Ms, University of Geneva. Rizzi, Luigi. 1990. Relativized minimality [Linguistic Inquiry Monograph 16], Cambridge MA: The MIT Press. Rizzi, Luigi. 1997. The fine structure of the left periphery. In L. Haegeman (ed.), Elements of grammar, 281–337. Dordrecht: Kluwer. Rizzi, Luigi. 2000. Reconstruction, weak island sensitivity, and agreement. Ms, University of Sienna. Rizzi, Luigi. 2001a. Relativized minimality effects. In M. Baltin & C. Collins (eds), The handbook of contemporary syntactic theory, 89–110. Oxford: Blackwell. Rizzi, Luigi. 2001b. On the position interrogative in the left periphery of the clause. In G. Cinque & G. Salvi (eds), Current studies in Italian syntax. Esseys offered to Lorenzo Renzi, 287–296. Amsterdam: Elsevier. Rizzi, Luigi. 2002. Locality and left periphery. In A. Belletti (ed.), Structures and beyond. The cartography of syntactic structures, Vol. 3, 1–27. Oxford: OUP. Rizzi, Luigi (ed.). 2004. The structure of CP and IP. The cartography of syntactic structures. Oxford: OUP. Rodman, Robert. 1976. Scope phenomena, ‘movement transformations’, and Montague grammar. In B.H. Partee (ed.), Montague grammar, 165–176. New York NY: Academic Press. Roeper, Tom. 2006. Watching noun phrases emerge. Seeking compositionality. In V. van Geenhoven (ed.), Acquisition meets semantics, 37–64. Dordrecht: Kluwer. Roger-Yun, Soyoung. 2002. Les expressions nominales classificateurs et les propositions à cas multiples du coréen. Ph.D. dissertation, Université Paris 3. Roodenburg, Jasper. 2004a. Pour une approche scalaire de la déficience nominale: La position du français dans la théorie des noms nus [LOT Dissertation Series 99]. Roodenburg, Jasper. 2004b. French bare arguments are not extinct: The case of coordinated bare nouns. Linguistic Inquiry 35: 301–313. Russell, Bertrand. 1905. On denoting. Mind 14: 479–493. Ruys, Eduard. 1992. The scope of indefinites. Ph.D. dissertation, Utrecht University. Sadock, Jerrold. 1980. Noun incorporation in Greenlandic: A case of syntactic word formation. Language 56: 300–319. Sag, Ivan A. 1976a. A logical theory of verb phrase deletion. In S. Mufwene et al. (eds), Papers from the Twelfth Regional Meeting of the Chicago Linguistic Society. Chicago IL: University of Chicago. Sag, Ivan A. 1976b. Deletion and Logical Form. Ph.D. dissertation, MIT. (Published by Garland, New York NY, 1980). Santelmann, Lynn. 1993. The distribution of double determiners in Swedish: Den support in Do. Studia Linguistica 47: 154–176. Sasse, Hans-Jürgen. 1987. The thetic/categorical distinction revisited. Linguistics 25: 511–580. Scott, Gary-John. 1998 Stacked adjectival modification and the structure of nominal phrases. SOAS Working Papers in Linguistic and Phonetics Vol. 8, 59–89. Selkirk, Elisabeth O. 1977. Some remarks on noun phrase structure. In A. Akmajian, P. Culicover & T. Wasaw (eds), Studies in formal syntax, 285–325. London: Academic Press.



Bibliography 

Shlonsky, Ur. 1991. Quantifiers as functional heads: A study of quantifier float in Hebrew. Lingua 84: 159–180. Shlonsky, Ur. 2004. The form of Semitic nominals. Lingua 114(12): 1465–1526. Shlonsky, Ur. 2005. La sélection sémantique dans un cadre cartographique. Ms, University of Geneva. Siloni, Tal. 1991. Noun raising and the structure of noun phrases. MIT Working Papers in Linguistics 14: 255–170. Smith Stvan, Laurel. 2007. The functional range of bare singular count nouns in English. In E. Stark, E. Leiss & W. Abraham (eds), Nominal determination. Typology, genesis, and historical emergence of DP, 171–187. Amsterdam: John Benjamins. Sommerstein, Alan R. 1972. On the so-called ‘definite article’ in English. Linguistic Inquiry 3: 197–209. Spector, Benjamin. 2003. Plural indefinite DPs as plural polarity items. In J. Quer et al. (eds), Romance Languages and Linguistic Theory 2001, 295–313. Amsterdam: John Benjamins. Stark, Elisabeth, Elisabeth Leiss & Werner Abraham (eds). 2007. Nominal determination. Typology, genesis, and historical emergence of DP. Amsterdam: John Benjamins. Starke, Michal. 2001. Move dissolves into merge: A theory of locality. Ph.D. dissertation, University of Geneva. Starke, Michal. 2005. Lecture on ‘nanosyntax’ given at the University of Geneva. Stemberger, Joseph Paul. 1981. Morphological haplology. Languages 57: 791–817. Stockwell, Robert P., Paul Schachter & Barbara H. Partee. 1973. The major syntactic structures of English. New York NY: Holt, Rinehart and Winston. Storto, Gianluca. 2001. On the status of the partitive determiner in Italian. Ms, UCLA in http://semanticsarchive.net. (Published as Storto 2003). Storto, Gianluca. 2003. On the status of the partitive determiner in Italian. In J. Quer, J. Schroten, M. Scorretti, P. Sleeman & E. Verheugd (eds), Romance languages and linguistic theory 2001: Selected papers from Going Romance 2001, 315–330. Amsterdam: John Benjamins. Stowell, Timothy. 1989. Subjects, specifiers, and X-bar theory. In M. Baltin & A. Kroch (eds), Alternative conceptions of phrase structure, 232–262. Chicago IL: The University of Chicago Press. Strawson, Peter F. 1950. On referring. Mind 59: 320–344. Strawson, Peter F. 1952. Introduction to logical theory. London: Methuen. Svenonius, Peter. 2004. On the edge. In D. Adger, C. de Cat & G. Tsoulas (eds), Peripheries, 261–287. Dordrecht: Kluwer. Szabolcsi, Anna. 1983. The possessor that ran away from home. The Linguistic Review 3: 89–102. Szabolcsi, Anna. 1987. Functional categories in the noun phrase. In I. Kenesei (ed.), Approaches to Hungarian 2, 167–89. Szeged: JATE. Szabolcsi, Anna. 1994. The noun phrase. In F. Kiefer & K.E. Kiss (eds), The syntactic structure of Hungarian [Syntax and Semantics 27], 179–274. New York NY: Academic Press. Szabolcsi, Anna. 1998–2002. Strong and weak islands. In M. Everaert & H. van Riemsdijk (eds), The syntax companion. http://homepages.nyu.edu/~as109/papers.html Szabolcsi, Anna & Frans Zwarts. 1997. Weak islands and an algebraic semantics for taking scope. In A. Szabolcsi (ed.), Ways of scope taking, 217–262. Dordrecht: Kluwer. Szabolcsi, Anna (ed.). 1997. Ways of scope taking. Dordrecht: Kluwer. Taraldsen, Tarald. 1990. D-projections and N-projections in Norwegian. In M. Nespor & J. Mascaró (eds), Grammar in progress, 419–31. Dordrecht: Foris.

 The layered DP: Form and meaning of French indefinites Tovena, Lucia, Viviane Déprez & Jacques Jayez. 2004. Polarity sensitive items. In F. Corblin & H. De Swart (eds), Handbook of French semantics, 403–411. Stanford CA: CSLI. Valois, Daniel. 1991. The internal syntax of DP. Ph.D. dissertation, MIT. Van Geenhoven, Veerle. 1995. Semantic incorporation: A uniform semantics for West Greenlandic noun incorporation and West Germanic bare plural configurations. In A. Dainora, R. Hemphill, B. Luka, B. Need & S. Pargman (eds), Papers from the 31st Meeting of the Chicago Linguistics Society, 171–186. Chicago IL: Chicago Linguistics Society. Van Geenhoven, Veerle. 1996. Semantic incorporation and indefinite descriptions: Semantic and syntactic aspects of noun incorporation in West Greenlandic. Ph.D. dissertation, University of Tübingen. Van Geenhoven, Veerle. 1998. Semantic incorporation and indefinite descriptions. Stanford CA: CSLI. Van Geenhoven, Veerle & Louise McNally. 2005. On the property analysis of opaque complements. Lingua 115: 885–914. Van Lehn, Kurt. 1978. Determining the scope of English quantifiers. Technical report, Artificial Intelligence Laboratory, MIT. Vangsnes, Oystein. In progress. The nanosyntax of wh-determiners: A Scandinavian perspective. Ms, University of Tromso. Vendler, Zeno. 1957. Verbs and times. Philosophical Review 56: 143–160. (Reprinted in Vendler 1967). Vergnaud, Jean-Roger. 1982. Dépendances et niveaux de représentation en syntaxe. Thèse de doctorat d’état, Université de Paris VII. Verkuyl, Henk. 1972. On the compositional nature of the aspects. Dordrecht: Kluwer. Verkuyl, Henk. 1993. A theory of aspectuality. Cambridge: CUP. Vogeleer, Svetlana & Liliane Tasmowski. 2005. Les N, un N et des N en lecture générique. In S. Vogeleer & L. Tasmowski (eds), Déterminants indéfinis en français [Travaux de linguistique 50], 53–78. Vogeleer, Svetlana & Liliane Tasmowski (eds). 2006. Non-definiteness and plurality. Amsterdam: John Benjamins. Wagner, Robert-Léon & Jacqueline Pinchon. 1962. Grammaire du français classique et moderne. Paris: Hachette. Wilkinson, Karina. 1996. Bare plurals, plural pronouns and the partitive constraint. In J. Hoeksema (ed.), Partitives, 209–230. Berlin: Mouton de Gruyter. Williams, Edwin. 1977. Discourse and Logical Form. Linguistic Inquiry 8: 101–140. Wilson, George. 1978. On definite and indefinite descriptions. Philosophical Review LXXXVII: 48–76. Yip, Moira. 1998. Identity avoidance in phonology and morphology. In S. LaPointe, D. Brentari & P. Farrell (eds), Morphology and its relation to phonology and syntax, 216–246. Stanford CA: CSLI. Zamparelli, Roberto. 2000. Layers in the determiner phrase. New York NY: Garland. Zribi-Hertz, Anne. 2003. Pour une analyse unitaire de DE partitive. In F. Corblin, L. Kupferman (eds), Actes du colloque Indéfinis et prédication. Paris: Presses de la Sorbonne. Zucchi, Alessandro. 1994. The ingredients of definiteness and the definiteness effect. Natural Language Semantics 3(1): 33–78.

Index

A A’-movement see movement Abney,  13, 15, 171 Aboh,  20, 24, 25 accomplishment,  161, 230 accusative see case achievement,  161, 230 acquisition,  228 activity,  230 adjective,  22, 23, 25, 26, 88 agreement,  15, 91, 162, 170 Aktionsart,  230 ambiguity,  10, 32, 33, 36, 39, 61, 63, 76, 126, 127 three-way ambiguity,  49, 68, 83, 146 two-way ambiguity,  69, 116, 122, 208 anaphoric,  141, 193, 201, 205, 207, 208, 209, 234 argument,  12, 18, 47, 52, 72, 73, 90, 95, 119, 120, 122, 140, 164, 168, 172, 173, 174, 189, 191, 195, 203, 221, 222, 225, 226, 228 argument structure,  9, 10, 15, 89 argument-predicate distinction,  3, 84, 86, 87, 89, 91, 92, 148, 197, 225 see also kind-referring noun phrase, propertydenoting argument aspect,  229, 230, 234, 236 see also bounded, unbounded assertion of existence,  2, 31, 64, 69, 70, 82, 113, 117, 119, 122, 140, 192, 209, 228 see also existential, existentiality atelic,  193, 203, 229, 230, 231, 232, 233, 236, 237 see also telic, unbounded

atom,  139, 154, 156, 157, 202, 203, 204, 232 see also individuated reference B bare nouns,  72, 85, 94, 129, 154, 158, 179, 186, 189 bare partitives see partitive bare plurals see plurals Barwise & Cooper,  181 base order,  23 Beghelli,  111 Beghelli & Stowell,  104, 109 Belletti,  127 Borer,  20, 21, 28, 98, 154, 162, 167, 201 Bosveld-de Smet,  144, 168, 231, 235 bounded,  161, 193, 201, 229, 230, 231, 232, 233, 234, 237, 238 see also unbounded C cardinal,  29, 31, 32, 42, 43, 167 Cardinaletti & Giusti,  127, 170, 171, 172, 173, 174 cardinality,  31, 65, 158 Carlson,  74 cartographic approach,  8, 9, 11, 17, 97 case,  8, 11, 18, 27, 28, accusative,  38, 39, 237, 238 genitive,  168 inherent,  238 nominative,  15 partitive,  126, 236, 237, 238 Case Phrase,  26 causative,  180 cacophony,  177 characterizing sentence,  215, 216 choice-function,  115, 116, 121, 228 Chomsky,  8, 13, 15, 18

circularity,  111, 112 Classifier Phrase,  20, 21, 22 clausal operator see operator co-variation,  2, 44, 45, 83, 108, 109, 110, 112, 114, 115, 116, 117, 119, 121, 153, 183, 185, 226 complementizer system,  9 composition rules,  72 construct state,  18, 19 contextually determined interpretation,  140, 141, 142, 147, 150 contrastive stress,  42 contrastive use,  142 coordination,  131, 196, 204, 227 copy-remerge mechanism,  98, 105 correlation count-S-reference,  204, 222 quantity-quantification,  214, 215 quantity-telicity,  231 tense-subject,  235 count,  154, 158, 159, 160, 162, 163, 164, 167, 185, 186, 202, 203, 226, 227, 233 see also mass counter,  20 covert partitive see partitive D de ‘of ’,  133, 134, 135, 168, 169, 177, 178, 179, 181, 226 de quoi 133, 134, 180, 181 decomposition approach,  10, 77 definite definite article,  16, 18, 21, 26, 60, 86, 89, 135, 168, 177, 206, 228 definite bare noun,  205, 207 definite noun phrase,  26, 27, 29, 30, 39, 41, 46, 47,

  Index 54, 64, 107, 112, 135, 144, 174, 181, 182, 183, 187, 206, 231, 234, 236, 237 Definite Phrase,  26 Definiteness-effect,  29, 30, 31, 46 deictic,  60, 110, 112, 153, 226 De Jong & Verkuyl,  31, 45 de-NP,  79, 81, 82, 93, 94, 114, 119, 120, 145, 146, 164, 185, 186, 226, 227 deP,  163, 164, 165, 168, 176, 186, 226, 227 dependent interpretation,  153, 157, 208 De Swart,  72, 157, 233, 234, 235 determination,  25, 238 determinerless language,  236, 238 Determiner Phrase,  13, 20, 26, 87, 88, 91, 217 deverbal noun,  14 Diesing,  29, 37, 41, 42, 43, 65, 66, 83 discourse,  21, 24, 33, 39, 40, 44, 60, 116, 117, 209 discourse information,  112 see also deictic discourse operator see operator discourse pragmatic,  9, 18 see also referent distribution of bare nouns,  85, 197, 211, 213 of du/des-NPs,  145 of partitives,  176, 178, 227 of specific noun phrases,  40 of strong determiners,  66 of un-NPs,  53, 78, 81, 82, 83, 93, 94, 119, 227 restricted distribution,  78 divider,  21 Dobrovie-Sorin & Beyssade,  6, 45, 46, 47, 140 Dobrovie-Sorin et al.,  222 domain,  13, 17, 18, 86, 91, 148, 197, 226 Donnellan,  57, 183 DP-hypothesis,  13 E empty category,  62, 84, 85, 86, 93, 94, 120, 121, 164, 170, 171, 173, 178, 179, 185,

186, 187, 198, 199, 211, 218, 219, 227 Enç,  29, 38, 39, 40, 44, 45, 48, 65, 67, 68, 116, 117, 118 EN pronoun,  136, 165, 166, 167, 171, 186, 187, 227 episodic,  219, 234, 235, 236 existence entailment,  77, 82, 117, 119, 143 see also assertion of existence existential,  2, 53, 60, 64, 74, 77, 111, 112, 114, 115, 118, 121, 122, 139, 140, 184, 185, 191, 192, 194, 209, 218, 236 existential bare plurals,  208 existential closure,  42, 53, 74, 83, 89, 115 existential construction/ sentence,  30, 31, 32, 40, 41, 44, 46, 47, 65, 75, 90, 95, 113, 116, 117, 135, 136, 146, 196 see also there be construction existential entailment,  76, 77, 143, 147 existential operator see operator existential predicate see predicate existential quantification see quantification existential quantifier see quantifier intrinsically existential,  75, 83, 95, 113, 193 non-existential,  69, 77, 114, 119, 145 see also presuppositional existentiality,  2, 53, 74, 95, 111, 113, 114, 121, 122, 153, 185, 192, 200, 216, 220, 228, 236 explicit partitive see partitive Extended Standard Theory,  10 extension,  139, 142, 156, 157, 185, 202, 203, 204 disjoint extension,  54 extraction,  12, 16, 34, 89, 99, 107, 132 F false,  54, 55, 57, 60, 63, 64, 65, 66, 71, 139, 140, 141, 149

Farkas,  107, 108, 233, 234 feature class,  12 finiteness,  9 floating type,  72, 76 Focus,  9, 26, 27, 28 Fodor & Sag,  29, 32, 33, 34, 35, 36, 37, 44, 45, 53, 57, 59, 83, 106, 108, 110, 116, 118 Force,  9, 73, 74 formulae,  113 fragmentative verb,  129, 142, 147, 150, 168, 169, 176, 180, 181, 187 free genitive,  19 free variable see variable Frege,  31, 64, 65 functional de,  177, 179, 186 fusion of de and le/les,  164, 186, 226 G Gbe,  24 Generative Grammar,  8 generic,  89, 124, 170, 177, 183, 189, 191, 192, 215, 216, 217, 218, 219, 220, 235, 236 genitive,  18, 19, 134, 181, 187 see also case Germanic,  16, 191, 197, 198, 199, 203, 208, 211, 216, 218, 220, 239 Giusti,  26, 27, 28, 173 government,  85, 93, 94, 179, 198, 211, 236 grammatical number,  5, 157, 158, 162, 175, 185, 186, 200, 203, 223 Greenberg,  23 Gross,  177, 179 Guéron,  219, 220, 235, 236 Gungbe,  24, 25 H habitual,  146, 234, 235 Hale & Keyser,  9 haplology,  178, 182, 187, 227 Hawkins,  23 head movement see movement Heim,  41, 42, 48, 52, 56, 57, 58, 59, 74 Heim & Kratzer,  56 Heycock & Zamparelli,  205, 206, 207, 208, 209 Higginbotham,  46, 71

Index  I identificational force,  144, 168 idiom,  195 Ihsane & Puskás,  26 imperfective,  234, 235 implicit partitives see partitive indefinite article,  44 (in)definiteness,  18, 28, 46, 237, 238 indefiniteness,  216, 233, 239 indeterminacy,  124, 144 individuals,  45, 47, 48, 54, 55, 57, 59, 62, 154, 155, 156, 157, 204, 234 individuated reading,  154, 156, 158, 160, 186, 222, 227 inferred feature,  158, 159, 160, 162, 186, 226, 233 inflectional domain,  7, 17, 18, 20, 21, 22, 91, 124, 148, 149, 152, 154, 175, 177, 179, 185, 186, 187, 200, 202, 204, 213, 214, 215, 222, 223, 225, 226, 227 inherent case see case in situ,  43, 95, 115, 119, 120, 145, 149 intensional context,  2, 45, 70, 75, 76, 77, 78, 82, 93, 95, 118, 119, 120, 149, 174, 185 interface syntax-lexicon interface,  99, 228 syntax-semantics interface,  10, 90, 228 intermediate reading,  35, 36, 37, 53, 57, 58, 68, 104, 107, 108, 109, 110, 112, 113, 115, 117, 152, 153, 185, 199, 200, 226 see also scope intervention effects,  12, 49, 52, 90, 97, 103, 104, 121, 151, 185 island,  12, 33, 34, 35, 37, 53, 57, 99, 100, 106, 107, 108, 109, 110, 121, 150, 154, 185 J Jackendoff,  169, 171, 181 K Kamp,  52 Kayne,  23, 83, 93 Keenan,  40

kind-referring noun phrase,  215, 216, 217, 218, 219 Kiparsky,  236, 237 Kratzer,  37 Kripke,  53, 59, 83, 119 Kupferman,  129, 131, 134, 135, 181 L Laenzlinger,  25, 26 landing site,  19, 104, 109, 111, 112, 185 language faculty,  8 left-dislocated,  172 left periphery,  4, 6, 7, 9, 18, 20, 23, 52, 63, 86, 91, 96, 98, 105, 112, 119, 120, 148, 164, 168, 175, 182, 183, 184, 185, 189, 197, 203, 204, 210, 215, 217, 221, 222, 225, 226, 227, 228, 229 licensing,  82, 86, 89, 93, 114, 121, 179 listeme,  20, 21, 154, 162, 185, 203, 223 locality,  11, 12, 104 see also Relativized Minimality Logical Form,  41, 43, 62, 96, 100, 103, 104, 109, 112, 114, 120, 121, 151, 152, 153, 185, 198, 211, 228 Longobardi,  84, 85, 86, 91, 102, 179, 198, 211 Ludlow & Neale,  37 M Mapping Hypothesis,  41 mass,  21, 139, 142, 154, 156, 157, 158, 159, 160, 162, 163, 167, 186, 196, 203, 213, 233 see also count Massam,  222 Matthewson & Schaeffer,  228 Mc Nally,  72, 74, 75, 191, 192, 199 mental concept,  220, 236 Milner,  127, 131, 132 Milsark,  29, 30, 31, 32, 40, 41, 43, 44, 62, 65, 67, 87, 89, 116, 118, 206 minimal parts,  139, 142, 157, 204 mirror-image,  23 modifier,  16, 18, 22, 23, 85, 144, 160, 175, 187 Montague,  83, 122

movement,  4, 5, 62, 63, 86, 92, 97, 99, 104, 105, 124, 188, 197, 212, 213, 218, 227 A’-movement,  17 DP-internal movement,  16, 28 head movement,  16, 23, 198, 199, 220 phrasal/XP movement,  16, 23, 175, 211 trigger of movement,  106, 111, 112, 153 wh-movement,  16 N narrow scope see scope negation,  34, 56, 75, 81, 82, 83, 114, 135 negative context,  78, 82, 83, 93, 114, 119, 135, 145, 146, 147 Negative Polarity Item,  81, 82, 83 non-specific see specific non-overt determiner,  198, 211, 217, 218 nuclear scope see scope Number Phrase,  18, 20, 22, 91 see also NumP numeral,  22, 23, 28, 29, 30, 32, 36, 41, 56, 127, 155, 162, 167, 170, 172, 174, 175, 177, 186, 215, 222, 226 NumP,  18, 19, 20, 24, 25, 91, 92, 152, 212 see also Number Phrase O Obligatory Contour Principle,  178 opaque reading,  76, 77, 78, 143, 193 opaque context,  70, 82 operator,  20, 21, 27, 35, 38, 41, 44, 48, 56, 61, 62, 75, 94, 104, 106, 108, 109, 111, 149, 219, 227 clausal operator,  217, 218 discourse operator,  20 existential operator,  73, 74, 75, 76, 77, 83, 95, 111, 113, 114, 115, 122, 143, 145, 149, 153, 185, 221, 228 generic operator,  89 see also generic

  Index negative operator,  81, 119, 121, 146, 164, 186, 227 silent operator,  42, 219 see also quantifier P Partee,  183 part-of relation,  125, 127, 128 partitive,  7, 31, 32, 127, 128, 129, 131, 132, 133, 145, 184, 186 bare partitives,  126, 133, 169, 174 covert partitive,  38, 39, 49, 67, 68, 116, 117, 141, 142, 207, 215 explicit partitive,  142, 144, 147, 168, 169, 173, 174, 176, 177, 179, 180, 181, 182, 187, 227 implicit partitive,  142, 147, 150, 169, 176, 179, 187, 227 partitive article,  123, 125, 126, 130, 134 partitive case see case Partitive Constraint,  181, 182 partitive PP,  171, 172, 173, 174, 175, 177, 187 partitive specifics,  40, 48, 68, 116, 117 see also relational specifics, specific, specificity see also distribution partitivity,  143, 187 perfective,  234, 235, 236 phrasal specifier,  11 phrasal category,  130 phrasal movement see movement pied-piping,  105 plural -s,  21, 154, 155, 156, 157, 158, 160, 163, 203, 227 plurality,  21, 24, 25, 154, 157, 158 plurals,  156 bare plurals,  74, 75, 154, 191, 192, 193, 201, 203, 208, 209, 210, 211, 212, 213 Pollock,  18 Port-Royal,  177 pragmatic,  9, 18, 33, 158 predicate,  72, 73, 74, 77, 83, 85, 91, 92, 115, 122, 138, 161, 185, 193, 221, 235, 237, 238

argument-predicate distinction see argument dyadic predicate,  172 entity predicate,  72, 73, 140, 221 existential predicate,  72, 73, 75, 76, 116, 118, 149, 185, 221 individual level predicate,  32 matrix predicate,  34, 36 monadic predicate,  73 predicative expression,  31, 42, 52, 71, 85, 88, 89, 91, 92, 93, 95, 120, 164, 183 Predicative Determiner Phrase,  88 preposition,  105, 133, 134, 169, 177, 179, 181 see also partitive PP presupposition of existence,  31, 40, 45, 48, 64, 65, 66, 67, 68, 83, 113, 140 presupposition failure,  66, 67, 141, 142 see also range-based presupposition presuppositional,  29, 42, 43, 45, 48, 49, 64, 65, 66, 67, 83, 113, 122, 141, 142, 147 presuppositionality,  31, 43, 65, 66, 67 Principles and Parameters,  8 pronoun,  30, 32, 54, 60, 102, 110, 170, 187, 226 relative pronoun,  129, 134, 169 see also EN pronoun proper name,  32, 33, 54, 63, 86, 102 property reading,  71, 72, 73, 74, 75, 76, 77, 83, 92, 93, 113, 119, 139, 143, 145, 146, 168, 174, 203, 207, 209, 210 property-denoting argument,  43, 45, 46, 47, 51, 71, 72, 73, 74, 76, 77, 78, 82, 83, 89, 92, 93, 94, 95, 97, 99, 119, 120, 122, 139, 143, 145, 146, 147, 149, 151, 166, 167, 174, 175, 183, 185, 186, 191, 192, 197, 199, 207, 208, 209, 210, 211, 221, 225, 226, 228

Property Phrase,  92, 93, 95, 120, 124, 148, 149, 184, 189, 197, 199, 210, 221, 222, 226, 227, 228, 236 Q quantification,  32, 65, 66, 74, 96, 98, 105, 142, 143, 149, 215, 227 existential quantification,  69, 76, 82 quantificational force,  42, 66, 74, 233, 234, 235 lack of quantificational force,  52, 71, 231 quantificational reading,  33, 63, 64, 96, 111, 141, 150, 210, 214 non quantificational,  37, 48, 67, 142, 174, 175, 176, 208 quantifier existential quantifier,  30, 61, 72, 172, 173 universal quantifier,  29, 30, 32, 45, 49, 58, 61, 100, 108, 109, 110, 111, 112, 117, 119, 121, 153, 174, 185, 226 interaction with quantifiers,  48, 49, 100, 121, 150, 151 see also operator Quantifier-InterpretationOnly,  33, 36 Quantifier Raising,  5, 42, 61, 89, 90, 96, 100, 104, 105, 107, 109, 112, 115, 117, 120, 121, 149, 152, 153, 183, 199, 212, 226 quantitative,  126-135, 144, 145, 147, 154, 160, 162, 164, 172, 174, 175, 186, 203, 214, 215, 222, 232, 238 quantity,  21, 22, 31, 124, 125, 128, 129, 154, 157, 158, 159, 162, 163, 164, 166, 167, 174, 175, 176, 177, 182, 184, 185, 186, 187, 200, 203, 214, 215, 223, 226, 227, 231, 232, 233, 237, 238 undetermined quantity,  127, 130, 144, 145, 169 Quine,  77 R Ramchand,  9, 10 range,  20, 21, 22,

Index  range-based presupposition,  48, 49, 68 reference,  52, 53, 54, 56, 65, 83, 107, 115, 121, 122, 139, 153, 157, 182, 184, 185, 208, 210, 221, 225, 227, 228 dependent reference,  53, 59, 110, 112, 119, 121, 183, 226 determined reference,  233, 234, 235 individuated reference,  200, 201, 202, 203, 204, 214, 223, 226, 231, 233 see also atom rigid reference,  53, 59, 119, 183 speaker’s reference,  53, 59, 60, 83, 102, 107, 112, 114, 119, 120, 137, 152, 199, 225 see also referential, referentiality referent,  38, 39, 53, 57, 59, 61, 83, 102, 112, 137, 209, 228, 234 referent established in the discourse,  24, 26, 47, 48, 53, 83, 109, 181 referential,  56, 89, 119 not referential,  56, 57, 140, 183, 184 not S-referential,  140, 204 referentiality,  18, 26, 27, 46, 52, 102, 105, 112, 117, 208, 226, 235 referring expressions,  56 Reinhart,  107, 108, 115 relational specifics,  40, 45, 48, 117, 118 see also partitive specifics, specific, specificity Relativized Minimality,  11, 12, 101, 109, 252 see also locality restrictive clause,  41, 42, 65 restrictor,  21, 65, 66 Ritter,  16, 18, 19 Rizzi,  9, 11, 12, 18, 98, 104 Roeper,  228 Romance,  16, 179, 189, 191, 197, 198, 199, 203, 208, 211, 213, 216, 218 Roodenburg,  194, 206, 209, 212 Russell,  28, 64 Ruys,  108

S Sag,  36 scope,  62, 63, 81, 82, 89, 94, 95, 97, 103, 104, 107, 106, 109, 111, 119, 121, 146, 150, 151 narrow scope,  33, 35, 36, 44, 45, 49, 71, 74, 75, 76, 77, 83, 95, 109, 112, 113, 114, 116, 117, 153, 193 nuclear scope,  41, 42, 62, 74 wide scope,  33, 34, 35, 36, 37, 38, 49, 61, 71, 77, 83, 100, 104, 106, 114, 115 widest scope,  35, 44, 45, 57, 58, 68, 104, 105, 106, 107, 108, 109, 110, 112, 138, 139, 140, 147, 152, 153, 185, 200, 226 scope ambiguity,  32, 122 scope island,  99, 106, 109 see also intermediate reading scope properties,  2, 33, 49, 53, 101, 104, 105, 113, 115, 117, 120, 121, 145, 152, 185, 221, 226, 228 lack of scope properties,  59, 95, 147, 149, 199 selection,  73, 76, 83, 93, 95, 98, 120, 171, 172, 173, 181, 221, 228 semantic decomposition,  77 semantic incorporation,  73, 221 semantic type,  46, 47, 50, 51, 52, 83, 95, 119, 120, 121, 122, 225, 226, 227, 228 semilattice,  156, 157, 202 set,  31, 48, 62, 67, 69, 83, 109, 128, 142, 168, 169, 172, 187, 207, 214, 238 context determined set,  140, 147, 150, 209 delimited set,  138 empty set,  155 non-empty set,  33, 41, 44, 65, 66, 115 see also explicit partitives, subset shells,  18, 91, 225 Shlonsky,  98, 105, 173 silent operator see operator speaker oriented,  112, 121 speaker’s reference see reference

specific,  24, 26, 29, 33, 34, 35, 36, 38, 39, 40, 41, 45, 49, 57, 59, 68, 97, 144 non-specific,  29, 38, 39, 40, 41, 44, 118, 124, 168 see also distribution, partitive specifics, relational specifics specificity,  12, 18, 24, 25, 38, 39, 40, 48, 49, 59, 68 Sportiche,  173 Starke,  12, 47, 97 state,  230 strong noun phrase/ determiner,  29, 30, 31, 32, 40, 42, 43, 44, 45, 66, 67, 88, 89, 97, 116, 118, 128, 206, 207, 209, 212, 213 see also weak noun phrase/ determiner structural difference,  91, 92, 120 structural type,  11, 12 subset,  31, 39, 48, 67, 128, 141, 142, 147, 150, 168, 172, 187, 207, 209, 214 see also implicit partitives, set syntactic label,  10 syntactic tag,  10 syntax-lexicon interface see interface syntax-semantics interface see interface Szabolcsi,  16 T tautology,  55, 57 telic,  160, 161, 193, 201, 229, 230, 231, 232, 233, 234, 237 see also atelic tense,  219, 220, 235, 236 thematic domain,  18 thematic role,  9, 10 there be construction,  30, 31, 41, 46, 47, 64, 67, 90, 113, 114, 118, 145 see also existential construction/sentence three-way classification/ ambiguity see ambiguity

  Index Topic,  9, 11, 26, 27, 28 topicalisation,  24, 63, 90, 118 transparent reading,  77, 78, 193, 194 tree-splitting,  41, 42 trigger of movement see movement tripartite notation,  41, 62, 74, 140 true,  55, 62, 63, 64, 65, 66, 77, 141, 142 truncated structure,  91, 148, 221, 222, 226, 228 truncation,  222, 223, 227 two-way classification/ ambiguity see ambiguity Type D verb,  129 Type D/I verb,  129, 133, 134 typology,  6, 11, 51, 83, 119, 123, 182, 230

U unbounded,  160, 161, 193, 194, 201, 229, 230, 231, 232, 233, 234, 237, 238 see also atelic, bounded uniqueness,  64, 205, 234 Universal Grammar,  8 universal quantifier see quantifier Universals,  23 UTAH,  172, 174 V Van Geenhoven & Mc Nally,  76, 77 variable,  21, 42, 54, 61, 72, 73, 74, 76, 115 free variable,  219, 220 Verkuyl,  231 VP Deletion,  36

W weak noun phrase/determiner,  12, 26, 29, 30, 31, 40, 42, 43, 44, 66, 75, 88, 89, 90, 97, 116, 118, 128, 206, 207, 209, 212 see also strong noun phrase/ determiner wide scope see scope widest scope see scope Williams,  36 Wilson,  57 X X’-theory,  13, 14 Z Zamparelli,  4, 52, 87, 89, 90, 91, 102, 217 Zribi-Hertz,  181

Linguistik Aktuell/Linguistics Today A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 132 Biberauer, Theresa (ed.): The Limits of Syntactic Variation. Expected September 2008 131 De Cat, Cécile and Katherine Demuth (eds.): The Bantu-Romance Connection. A comparative investigation of syntactic structure. Expected September 2008 130 Kallulli, Dalina and Liliane Tasmowski (eds.): Clitic Doubling in Balkan Languages. ix, 431 pp. + index. Expected August 2008 129 Sturgeon, Anne: The Left Periphery. The interaction of syntax, pragmatics and prosody in Czech. xi, 139 pp. + index. Expected August 2008 128 Taleghani, Azita H.: Modality, Aspect and Negation in Persian. ix, 179 pp. + index. Expected July 2008 127 Durrleman-Tame, Stephanie: The Syntax of Jamaican Creole. A cartographic perspective. xii, 186 pp. + index. Expected July 2008 126 Schäfer, Florian: The Syntax of (Anti-)Causatives. External arguments in change-of-state contexts. xi, 318 pp. + index. Expected June 2008 125 Rothstein, Björn: The Perfect Time Span. On the present perfect in German, Swedish and English. xi, 170 pp. + index. Expected August 2008 124 Ihsane, Tabea: The Layered DP. Form and meaning of French indefinites. 2008. ix, 260 pp. 123 Stoyanova, Marina: Unique Focus. Languages without multiple wh-questions. 2008. xi, 184 pp. 122 Oosterhof, Albert: The Semantics of Generics in Dutch and Related Languages. 2008. xviii, 286 pp. 121 Tungseth, Mai Ellin: Verbal Prepositions and Argument Structure. Path, place and possession in Norwegian. 2008. ix, 187 pp. 120 Asbury, Anna, Jakub Dotlačil, Berit Gehrke and Rick Nouwen (eds.): Syntax and Semantics of Spatial P. 2008. vi, 416 pp. 119 Fortuny, Jordi: The Emergence of Order in Syntax. 2008. viii, 211 pp. 118 Jäger, Agnes: History of German Negation. 2008. ix, 350 pp. 117 Haugen, Jason D.: Morphology at the Interfaces. Reduplication and Noun Incorporation in Uto-Aztecan. 2008. xv, 257 pp. 116 Endo, Yoshio: Locality and Information Structure. A cartographic approach to Japanese. 2007. x, 235 pp. 115 Putnam, Michael T.: Scrambling and the Survive Principle. 2007. x, 216 pp. 114 Lee-Schoenfeld, Vera: Beyond Coherence. The syntax of opacity in German. 2007. viii, 206 pp. 113 Eythórsson, Thórhallur (ed.): Grammatical Change and Linguistic Theory. The Rosendal papers. 2008. vi, 441 pp. 112 Axel, Katrin: Studies on Old High German Syntax. Left sentence periphery, verb placement and verbsecond. 2007. xii, 364 pp. 111 Eguren, Luis and Olga Fernández Soriano (eds.): Coreference, Modality, and Focus. Studies on the syntax–semantics interface. 2007. xii, 239 pp. 110 Rothstein, Susan (ed.): Theoretical and Crosslinguistic Approaches to the Semantics of Aspect. 2008. viii, 453 pp. 109 Chocano, Gema: Narrow Syntax and Phonological Form. Scrambling in the Germanic languages. 2007. x, 333 pp. 108 Reuland, Eric, Tanmoy Bhattacharya and Giorgos Spathas (eds.): Argument Structure. 2007. xviii, 243 pp. 107 Corver, Norbert and Jairo Nunes (eds.): The Copy Theory of Movement. 2007. vi, 388 pp. 106 Dehé, Nicole and Yordanka Kavalova (eds.): Parentheticals. 2007. xii, 314 pp. 105 Haumann, Dagmar: Adverb Licensing and Clause Structure in English. 2007. ix, 438 pp. 104 Jeong, Youngmi: Applicatives. Structure and interpretation from a minimalist perspective. 2007. vii, 144 pp. 103 Wurff, Wim van der (ed.): Imperative Clauses in Generative Grammar. Studies in honour of Frits Beukema. 2007. viii, 352 pp. 102 Bayer, Josef, Tanmoy Bhattacharya and M.T. Hany Babu (eds.): Linguistic Theory and South Asian Languages. Essays in honour of K. A. Jayaseelan. 2007. x, 282 pp.

101 Karimi, Simin, Vida Samiian and Wendy K. Wilkins (eds.): Phrasal and Clausal Architecture. Syntactic derivation and interpretation. In honor of Joseph E. Emonds. 2007. vi, 424 pp. 100 Schwabe, Kerstin and Susanne Winkler (eds.): On Information Structure, Meaning and Form. Generalizations across languages. 2007. vii, 570 pp. 99 Martínez-Gil, Fernando and Sonia Colina (eds.): Optimality-Theoretic Studies in Spanish Phonology. 2007. viii, 564 pp. 98 Pires, Acrisio: The Minimalist Syntax of Defective Domains. Gerunds and infinitives. 2006. xiv, 188 pp. 97 Hartmann, Jutta M. and László Molnárfi (eds.): Comparative Studies in Germanic Syntax. From Afrikaans to Zurich German. 2006. vi, 332 pp. 96 Lyngfelt, Benjamin and Torgrim Solstad (eds.): Demoting the Agent. Passive, middle and other voice phenomena. 2006. x, 333 pp. 95 Vogeleer, Svetlana and Liliane Tasmowski (eds.): Non-definiteness and Plurality. 2006. vi, 358 pp. 94 Arche, María J.: Individuals in Time. Tense, aspect and the individual/stage distinction. 2006. xiv, 281 pp. 93 Progovac, Ljiljana, Kate Paesani, Eugenia Casielles and Ellen Barton (eds.): The Syntax of Nonsententials. Multidisciplinary perspectives. 2006. x, 372 pp. 92 Boeckx, Cedric (ed.): Agreement Systems. 2006. ix, 346 pp. 91 Boeckx, Cedric (ed.): Minimalist Essays. 2006. xvi, 399 pp. 90 Dalmi, Gréte: The Role of Agreement in Non-Finite Predication. 2005. xvi, 222 pp. 89 Velde, John R. te: Deriving Coordinate Symmetries. A phase-based approach integrating Select, Merge, Copy and Match. 2006. x, 385 pp. 88 Mohr, Sabine: Clausal Architecture and Subject Positions. Impersonal constructions in the Germanic languages. 2005. viii, 207 pp. 87 Julien, Marit: Nominal Phrases from a Scandinavian Perspective. 2005. xvi, 348 pp. 86 Costa, João and Maria Cristina Figueiredo Silva (eds.): Studies on Agreement. 2006. vi, 285 pp. 85 Mikkelsen, Line: Copular Clauses. Specification, predication and equation. 2005. viii, 210 pp. 84 Pafel, Jürgen: Quantifier Scope in German. 2006. xvi, 312 pp. 83 Schweikert, Walter: The Order of Prepositional Phrases in the Structure of the Clause. 2005. xii, 338 pp. 82 Quinn, Heidi: The Distribution of Pronoun Case Forms in English. 2005. xii, 409 pp. 81 FuSS, Eric: The Rise of Agreement. A formal approach to the syntax and grammaticalization of verbal inflection. 2005. xii, 336 pp. 80 Burkhardt, Petra: The Syntax–Discourse Interface. Representing and interpreting dependency. 2005. xii, 259 pp. 79 Schmid, Tanja: Infinitival Syntax. Infinitivus Pro Participio as a repair strategy. 2005. xiv, 251 pp. 78 Dikken, Marcel den and Christina Tortora (eds.): The Function of Function Words and Functional Categories. 2005. vii, 292 pp. 77 Öztürk, Balkız: Case, Referentiality and Phrase Structure. 2005. x, 268 pp. 76 Stavrou, Melita and Arhonto Terzi (eds.): Advances in Greek Generative Syntax. In honor of Dimitra Theophanopoulou-Kontou. 2005. viii, 366 pp. 75 Di Sciullo, Anna Maria (ed.): UG and External Systems. Language, brain and computation. 2005. xviii, 398 pp. 74 Heggie, Lorie and Francisco Ordóñez (eds.): Clitic and Affix Combinations. Theoretical perspectives. 2005. viii, 390 pp. 73 Carnie, Andrew, Heidi Harley and Sheila Ann Dooley (eds.): Verb First. On the syntax of verbinitial languages. 2005. xiv, 434 pp. 72 FuSS, Eric and Carola Trips (eds.): Diachronic Clues to Synchronic Grammar. 2004. viii, 228 pp. 71 Gelderen, Elly van: Grammaticalization as Economy. 2004. xvi, 320 pp. 70 Austin, Jennifer R., Stefan Engelberg and Gisa Rauh (eds.): Adverbials. The interplay between meaning, context, and syntactic structure. 2004. x, 346 pp. 69 Kiss, Katalin É. and Henk van Riemsdijk (eds.): Verb Clusters. A study of Hungarian, German and Dutch. 2004. vi, 514 pp. 68 Breul, Carsten: Focus Structure in Generative Grammar. An integrated syntactic, semantic and intonational approach. 2004. x, 432 pp. 67 Mišeska Tomić, Olga (ed.): Balkan Syntax and Semantics. 2004. xvi, 499 pp.

66 Grohmann, Kleanthes K.: Prolific Domains. On the Anti-Locality of movement dependencies. 2003. xvi, 372 pp. 65 Manninen, Satu Helena: Small Phrase Layers. A study of Finnish Manner Adverbials. 2003. xii, 275 pp. 64 Boeckx, Cedric and Kleanthes K. Grohmann (eds.): Multiple Wh-Fronting. 2003. x, 292 pp. 63 Boeckx, Cedric: Islands and Chains. Resumption as stranding. 2003. xii, 224 pp. 62 Carnie, Andrew, Heidi Harley and MaryAnn Willie (eds.): Formal Approaches to Function in Grammar. In honor of Eloise Jelinek. 2003. xii, 378 pp. 61 Schwabe, Kerstin and Susanne Winkler (eds.): The Interfaces. Deriving and interpreting omitted structures. 2003. vi, 403 pp. 60 Trips, Carola: From OV to VO in Early Middle English. 2002. xiv, 359 pp. 59 Dehé, Nicole: Particle Verbs in English. Syntax, information structure and intonation. 2002. xii, 305 pp. 58 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 2: Morphology, phonology, acquisition. 2003. vi, 309 pp. 57 Di Sciullo, Anna Maria (ed.): Asymmetry in Grammar. Volume 1: Syntax and semantics. 2003. vi, 405 pp. 56 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 2: The expression of possession in noun phrases. 2003. x, 295 pp. 55 Coene, Martine and Yves D’hulst (eds.): From NP to DP. Volume 1: The syntax and semantics of noun phrases. 2003. vi, 362 pp. 54 Baptista, Marlyse: The Syntax of Cape Verdean Creole. The Sotavento varieties. 2003. xxii, 294 pp.  (incl. CD-rom). 53 Zwart, C. Jan-Wouter and Werner Abraham (eds.): Studies in Comparative Germanic Syntax. Proceedings from the 15th Workshop on Comparative Germanic Syntax (Groningen, May 26–27, 2000). 2002. xiv, 407 pp. 52 Simon, Horst J. and Heike Wiese (eds.): Pronouns – Grammar and Representation. 2002. xii, 294 pp. 51 Gerlach, Birgit: Clitics between Syntax and Lexicon. 2002. xii, 282 pp. 50 Steinbach, Markus: Middle Voice. A comparative study in the syntax-semantics interface of German. 2002. xii, 340 pp. 49 Alexiadou, Artemis (ed.): Theoretical Approaches to Universals. 2002. viii, 319 pp. 48 Alexiadou, Artemis, Elena Anagnostopoulou, Sjef Barbiers and Hans-Martin Gärtner (eds.): Dimensions of Movement. From features to remnants. 2002. vi, 345 pp. 47 Barbiers, Sjef, Frits Beukema and Wim van der Wurff (eds.): Modality and its Interaction with the Verbal System. 2002. x, 290 pp. 46 Panagiotidis, E. Phoevos: Pronouns, Clitics and Empty Nouns. ‘Pronominality’ and licensing in syntax. 2002. x, 214 pp. 45 Abraham, Werner and C. Jan-Wouter Zwart (eds.): Issues in Formal German(ic) Typology. 2002. xviii, 336 pp. 44 Taylan, Eser Erguvanlı (ed.): The Verb in Turkish. 2002. xviii, 267 pp. 43 Featherston, Sam: Empty Categories in Sentence Processing. 2001. xvi, 279 pp. 42 Alexiadou, Artemis: Functional Structure in Nominals. Nominalization and ergativity. 2001. x, 233 pp. 41 Zeller, Jochen: Particle Verbs and Local Domains. 2001. xii, 325 pp. 40 Hoeksema, Jack, Hotze Rullmann, Víctor Sánchez-Valencia and Ton van der Wouden (eds.): Perspectives on Negation and Polarity Items. 2001. xii, 368 pp. 39 Gelderen, Elly van: A History of English Reflexive Pronouns. Person, Self, and Interpretability. 2000. xiv, 279 pp. 38 Meinunger, André: Syntactic Aspects of Topic and Comment. 2000. xii, 247 pp. 37 Lutz, Uli, Gereon Müller and Arnim von Stechow (eds.): Wh-Scope Marking. 2000. vi, 483 pp. 36 Gerlach, Birgit and Janet Grijzenhout (eds.): Clitics in Phonology, Morphology and Syntax. 2001. xii, 441 pp. 35 Hróarsdóttir, Thorbjörg: Word Order Change in Icelandic. From OV to VO. 2001. xiv, 385 pp. 34 Reuland, Eric (ed.): Arguments and Case. Explaining Burzio’s Generalization. 2000. xii, 255 pp. 33 Puskás, Genoveva: Word Order in Hungarian. The syntax of Ā-positions. 2000. xvi, 398 pp. 32 Alexiadou, Artemis, Paul Law, André Meinunger and Chris Wilder (eds.): The Syntax of Relative Clauses. 2000. vi, 397 pp.

31 Svenonius, Peter (ed.): The Derivation of VO and OV. 2000. vi, 372 pp. 30 Beukema, Frits and Marcel den Dikken (eds.): Clitic Phenomena in European Languages. 2000. x, 324 pp. 29 Miyamoto, Tadao: The Light Verb Construction in Japanese. The role of the verbal noun. 2000. xiv, 232 pp. 28 Hermans, Ben and Marc van Oostendorp (eds.): The Derivational Residue in Phonological Optimality Theory. 2000. viii, 322 pp. 27 Růžička, Rudolf: Control in Grammar and Pragmatics. A cross-linguistic study. 1999. x, 206 pp. 26 Ackema, Peter: Issues in Morphosyntax. 1999. viii, 310 pp. 25 Felser, Claudia: Verbal Complement Clauses. A minimalist study of direct perception constructions. 1999. xiv, 278 pp. 24 Rebuschi, Georges and Laurice Tuller (eds.): The Grammar of Focus. 1999. vi, 366 pp. 23 Giannakidou, Anastasia: Polarity Sensitivity as (Non)Veridical Dependency. 1998. xvi, 282 pp. 22 Alexiadou, Artemis and Chris Wilder (eds.): Possessors, Predicates and Movement in the Determiner Phrase. 1998. vi, 388 pp. 21 Klein, Henny: Adverbs of Degree in Dutch and Related Languages. 1998. x, 232 pp. 20 Laenzlinger, Christopher: Comparative Studies in Word Order Variation. Adverbs, pronouns, and clause structure in Romance and Germanic. 1998. x, 371 pp. 19 Josefsson, Gunlög: Minimal Words in a Minimal Syntax. Word formation in Swedish. 1998. ix, 199 pp. 18 Alexiadou, Artemis: Adverb Placement. A case study in antisymmetric syntax. 1997. x, 256 pp. 17 Beermann, Dorothee, David LeBlanc and Henk van Riemsdijk (eds.): Rightward Movement. 1997. vi, 410 pp. 16 Liu, Feng-hsi: Scope and Specificity. 1997. viii, 187 pp. 15 Rohrbacher, Bernhard Wolfgang: Morphology-Driven Syntax. A theory of V to I raising and prodrop. 1999. viii, 296 pp. 14 Anagnostopoulou, Elena, Henk van Riemsdijk and Frans Zwarts (eds.): Materials on Left Dislocation. 1997. viii, 349 pp. 13 Alexiadou, Artemis and T. Alan Hall (eds.): Studies on Universal Grammar and Typological Variation. 1997. viii, 252 pp. 12 Abraham, Werner, Samuel David Epstein, Höskuldur Thráinsson and C. Jan-Wouter Zwart (eds.): Minimal Ideas. Syntactic studies in the minimalist framework. 1996. xii, 364 pp. 11 Lutz, Uli and Jürgen Pafel (eds.): On Extraction and Extraposition in German. 1996. xii, 315 pp. 10 Cinque, Guglielmo and Giuliana Giusti (eds.): Advances in Roumanian Linguistics. 1995. xi, 172 pp. 9 Gelderen, Elly van: The Rise of Functional Categories. 1993. x, 224 pp. 8 Fanselow, Gisbert (ed.): The Parametrization of Universal Grammar. 1993. xvii, 232 pp. 7 Åfarlí, Tor A.: The Syntax of Norwegian Passive Constructions. 1992. xii, 177 pp. 6 Bhatt, Christa, Elisabeth Löbel and Claudia Maria Schmidt (eds.): Syntactic Phrase Structure Phenomena in Noun Phrases and Sentences. 1989. ix, 187 pp. 5 Grewendorf, Günther and Wolfgang Sternefeld (eds.): Scrambling and Barriers. 1990. vi, 442 pp. 4 Abraham, Werner and Sjaak De Meij (eds.): Topic, Focus and Configurationality. Papers from the 6th Groningen Grammar Talks, Groningen, 1984. 1986. v, 349 pp. 3 Abraham, Werner (ed.): On the Formal Syntax of the Westgermania. Papers from the 3rd Groningen Grammar Talks (3e Groninger Grammatikgespräche), Groningen, January 1981. 1983. vi, 242 pp. 2 Ehlich, Konrad and Jürgen Rehbein: Augenkommunikation. Methodenreflexion und Beispielanalyse. 1982. viii, 150 pp. With many photographic ills. 1 Klappenbach, Ruth (1911–1977): Studien zur Modernen Deutschen Lexikographie. Auswahl aus den Lexikographischen Arbeiten von Ruth Klappenbach, erweitert um drei Beiträge von Helene MaligeKlappenbach. (Written in German). 1980. xxiii, 313 pp.