Demonstratives and Definite Articles As Nominal Auxiliaries (Linguistik Aktuell Linguistics Today)

Demonstratives and Definite Articles as Nominal Auxiliaries

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

Terje Lohndal

University of Maryland

Volume 140 Demonstratives and Definite Articles as Nominal Auxiliaries by Dorian Roehrs

Demonstratives and Definite Articles as Nominal Auxiliaries Dorian Roehrs University of North Texas

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 Roehrs, Dorian. Demonstratives and definite articles as nominal auxiliaries / Dorian Roehrs. p. cm. (Linguistik Aktuell/Linguistics Today, issn 0166-0829 ; v. 140) Includes bibliographical references and index. 1. Grammar, Comparative and general--Noun phrase. 2. Grammar, Comparative and general--Demonstratives. 3. Definiteness (Linguistics) I. Title. P271.R64    2009 415'.5--dc22 isbn 978 90 272 5523 5 (hb; alk. paper) isbn 978 90 272 8977 3 (eb)

2008051209

© 2009 – 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

I dedicate this book to my brother Reimo and my loving parents Rosen und Siggin. Danke, Ihr Lieben, Ihr seid einfach toll

Table of contents

Preface chapter 1 Introduction 1. Introduction  1 2. Basic proposal  6 3. Assumptions and their motivation  12 The structure of the DP  12 3.1 Concord and movement inside the DP  22 3.2 4. Overview of the chapters  25 Chapter 2  25 4.1 4.2 Chapter 3  27 4.3 Chapter 4  28 chapter 2 The syntactic distribution of determiners Introduction  31 1. 2. Word order possibilities of demonstratives and articles in Early Scandinavian  33 Proto-Scandinavian  34 2.1 2.2 Common Scandinavian  35 Old Norse  36 2.3 2.4 Schematic summary  37 3. Formalizing the development of the suffixed article  38 Demonstratives are phrases  38 3.1 Three options to derive the order N-dem  40 3.2 Reanalysis from demonstrative to article  44 3.3 3.4 Some immediate consequences  47 4. Unifying the different determiner systems  48 Demonstratives in different positions  49 4.1 4.2 “Split” demonstratives  51

xi

1

31

 Table of contents

5.

6. 7.

“Split” articles in the Scandinavian languages  52 The basic data  52 5.1 The semantic and syntactic relations of the two determiners  54 5.2 Semantic differences overtly reflected  54 5.2.1 The syntactic relation  60 5.2.2 “Split” determiners  65 5.2.3 Induced agreement domains  67 5.3 Assumptions  67 5.3.1 5.3.1.1 The suffixal determiner  67 5.3.1.2 The free-standing determiner  71 Modified DP  74 5.3.2 Unmodified DP  76 5.3.3 “Violations” of the double definiteness effect  76 5.3.4 Summary  77 5.4 An alternative proposal: Julien (2002 [2005])  78 Conclusion  80

chapter 3 The syntax and semantics of non-/restrictive modifiers 83 Introduction  83 1. 2. The syntax and semantics of restrictive modifiers  85 Derivation of the restrictive reading  87 2.1 2.2 Syntactic consequences: Scandinavian relative clauses  90 3. The syntax and semantics of non-restrictive modifiers  97 Derivation of the non-restrictive reading  102 3.1 Syntactic consequences: more on Scandinavian relative clauses  106 3.2 4. Extension to Romance adjectives  107 Some further issues  111 5. Summary of the syntax of modifiers  111 5.1 Some differences and idiosyncracies  112 5.2 Some restrictions on the “free” interpretation of the determiner  113 5.3 6. Conclusion  116 chapter 4 The strong and weak alternation in German Introduction  117 1. 2. The strong and weak inflections: The traditional endings  120 3. The problem: Determiners with different endings  123 The definite article  123 3.1 The indefinite article  125 3.2

117



4. 5. 6. 7. 8. 9.

10.

11.

Table of contents 

The proposal: Weak and strong null endings more generally  130 Null endings as no endings: The new inflections  133 Distribution of inflections: The traditional picture  134 The new picture  139 The strong and weak alternation as a reflex of Impoverishment  145 Basic derivations  151 The stereotypical cases  151 9.1 9.2 The genitive masculine and neuter  157 Special cases  160 10.1 Split-NPs  160 10.2 Indefinite pronoun constructions  162 10.3 Lexically ambiguous elements  163 10.3.1 Determiner-like pronouns  166 10.3.2 Determiner-like adjectives  167 10.3.3 Determiner-like intensifiers  169 10.3.3.1 Alle  169 10.3.3.2 Ein  170 10.4 Summary and feature realization  172 Conclusion  175

chapter 5 Conclusion

177

References

179

Language index

193

Subject index

195

Preface

Looking back, this book has taken a long time to materialize. It all started when I first taught German as a foreign language at the University of Connecticut in the fall of 1997. At the time, I was looking for a simple way to explain the adjective endings in German to the students. In the summer of 2001, I made the first attempt at explaining this distribution more formally. This idea resulted in a working paper (2002) and eventually in my dissertation (2006). The basic ideas have not changed much since then but the actual technical solutions vary a lot. This book presents the latest version of these ideas and supercedes previous work. All original chapters of the dissertation have undergone numerous, in part, substantial changes. It is my hope that the perspective presented here will be, at the very least, of interest to the reader. Over the years, many people have helped me form my ideas about this and other linguistic topics. In particular, I would like to mention my teachers at the University of Connecticut (Željko Bošković and Howard Lasnik) and Indiana University (Rex Sprouse, Kari Gade, Steven Franks, and Yoshihisa Kitagawa), who have supported me in countless ways. Furthermore, I would like to thank an anonymous reviewer and Elly van Gelderen for numerous and substantial comments on an earlier draft of this book and Werner Abraham for inviting me to submit the manuscript to the Linguistik Aktuell Series. Finally, I would like to gratefully acknowledge the receipt of a Faculty Research Grant (G 33669) from the University of North Texas. On a different note, I would also like to take the opportunity to thank Cristina and Lawrence for providing much needed non-linguistic distraction. Thank you guys for being such good friends. Denton, October 2008 Dorian Roehrs

chapter 1

Introduction 1.  Introduction One of the leading ideas of generative grammar is that all languages are the same at some sufficiently abstract level of representation. Assuming that Universal Grammar makes the same syntactic representation available for all languages, this strong, and thus interesting, claim is challenged every time there seems to be evidence to the contrary; that is, when languages appear to differ with regard to their structure. However, to the extent that the idea of a uniform abstract representation is correct, these differences must be taken to hold on the surface only. In other words, this variation is only apparent, not real in any “deep” sense. For the past twenty years or so, generative syntacticians have sought to account for these differences by appealing to the notion of parameters, morphosyntactic choice-points where individual “languages” may differ in superficial ways. This work is written in the generative tradition and attempts to contribute to this type of investigation, in general, and its cartographic endeavor, in particular. From a cross-linguistic perspective, the apparently different determiner systems in the world’s languages pose a particularly challenging problem. To illustrate, consider the following quotation taken from work on language typology: “…[I]t is mistaken to treat demonstratives which often function like definite articles as articles. After all, the evidence here suggests that they do not exhibit the word order properties that “true” definite articles do. Nevertheless, there is evidence that even when such definite articles are identical to demonstratives, they are still subject to different word order “pressures”… (1) a.

yule mtu the man ‘the man’

b. mtu yule man that ‘that man’

(Swahili)



Demonstratives and definite articles as nominal auxiliaries

…This suggests not only that the position of definite articles cannot be explained (at least not in all cases) in terms of the position of the demonstrative from which they arose, but also that words with the meaning of definite articles are subject to word order “pressures” that demonstratives are not subject to.” (Dryer 1989: 90–1; emphasis added)

Making the well-documented assumption that linguistic reanalysis occurs in a local domain (for the development of complementizers out of demonstratives, see van Gelderen 2004a), the different syntactic distribution of historically related determiner-like elements is unexpected. I will call this phenomenon “the Panchronic Alternation”: (2) Panchronic Alternation Diachronically related elements may occur in different synchronic positions.

If it is correct that reanalysis occurs in a local domain, then lexical items that are related in the relevant sense must be in the same domain at some point in the history of the language. In order to explain the different syntactic distribution of determiner-like elements over time, that is, their change in position, one can exploit the well-documented displacement property of languages. In other words, the differing distribution can be reduced to some general operation, that is, movement or lack thereof of the corresponding elements. With this in mind, consider the Panachronic Alternation in Swahili in light of earlier X′-theoretic assumptions. Assume the following generalized phrase structure schemata XP -> (Spec) X′, X′ -> X (Compl) where X = N. The linear order of the elements is assumed to be a matter of language-specific parameterization (cf. Chomsky 1986: 2–3): (3) a. NP -> (Det) N′ b. N′ -> N (Compl) c. NP Det

N′ N

(Compl)

In order to derive the different distribution of the article vs. the demonstrative in (1), let us first observe that it is very unlikely for the demonstrative to be generated in the complement position of the noun, as in (3b). As we will discuss below, this is the typical position where the complement of the noun merges. With this option excluded, one can proceed in two ways:  (i) the phrase structure schema in (3a) could be “broken up” into two subschemata, or (ii) one could keep (3a) as is and formulate a transformational rule. Considering each in turn, I will show that neither approach can account for the Swahili facts in (1) without problems. The conclusion will be that the noun phrase must contain more structure.



Introduction

As a first option, one could reformulate (3a) as two subschemata. One could assume that due to a parameter, the article is a Specifier on the left (NP1) and the demonstrative is a Specifier on the right (NP2): (4) a. NP1 -> Art N′ b. NP2 -> N′ Dem c. NP 1

Art yule

N′ N mtu



(Compl) NP2

d. N′ N mtu

Dem yule (Compl)

Although the schemata in (4a–b) describe the facts in (1) correctly, this comes at a high  cost. For instance, note that these two statements can only be languagespecific as the language Ute has the opposite distribution (Dryer 1989: 90). More generally, breaking up the category “determiner” by a parameter, the article and the demonstrative are no longer part of a natural class. As a consequence, we lose some important generalizations with regard to their morphological and semantic properties, that is, some commonalities with regard to their inflectional patterns and referential characteristics. In these aspects, determiners are quite different from nouns and adjectives. Moreover, the two historically related elements are not in a local relationship in any obvious way. That is, according to the representations in (4c–d), the historical relationship of the two elements becomes accidental and it is not clear how a smooth transition from the demonstrative to the article can be accomplished. As a second option, one could leave the phrase structure schemata in (3) unchanged and formulate a transformational rule. Leaving all elements in situ in the noun phrase with the article, (5a), the demonstrative raises across the noun in (5b):

(5) a.

NP Art yule

N′ N mtu

(Compl)





Demonstratives and definite articles as nominal auxiliaries

b.

NP Ni mtu

NP Dem yule

N′ ti

(Compl)

This provides the correct surface orders. Thus, employing the same schemata in (3), the historical change from a demonstrative to an article can proceed in a local domain (i.e., in Spec,NP). However, note that the transformation in (5b) adjoins a head to a maximal projection. Assuming structure-preserving operations, this is not allowed (the same would hold if one were to move N′ to adjoin to NP). I conclude that this is not a viable option either. Apparently, more structure is needed. Below, I shall outline a more promising syntactic representation. In the 1980’s, a number of linguists noticed that the traditional structural representation of noun phrases as NPs suffered a number of shortcomings (e.g., Bowers 1987; Brame 1981, 1982; Cowper 1987; Fukui & Speas 1986; Hellan 1986; Horrocks & Stavrou 1987; Hudson 1984: 90–2, 1987: 121–4; Lamontagne & Travis 1987; Lyons, J. 1977: 392, 464; Ritter 1988; Stowell 1989; Szabolcsi 1981, 1983–84, 1987; various manuscripts cited in Abney 1987: 77, and work on different languages in Delsing 1993b: 72). As an alternative, it was proposed that the determiner D is a functional head which projects its own phrase, a DP, and takes NP as its complement:

DP

(6) D the

NP N linguists

Abney (1987) was the first to discuss the DP-Hypothesis in greater detail. With regard to Swahili, this more complex structure allows for a smooth historical development without violating structure-preserving principles. In particular, leaving all elements in situ in the noun phrase with the article, as in (7a), the NP containing the head noun is moved across the demonstrative to Spec,DP, as in (7b):

(7) a.

DP Spec

D′ D yule

NP N mtu



Introduction DP

b. NPi N mtu

D′ D yule

ti

This brings about the correct word orders and structure-preserving principles are adhered to. Furthermore, the historical change can occur in a local domain (i.e., in D). Finally, assuming movement, the DP-Hypothesis has provided a straightforward account for the different distribution of the historically related determiners (the Panchronic Alternation). However, as we will see below, there is evidence that the demonstrative is not a head in D but a phrase presumably in Spec,DP. If so, we are left with the conclusion that (7b) is not sufficient either.1 In a different context, already Abney (1987) suggested that the noun phrase was more complex in that there were intermediate phrases such as a Quantifier Phrase (QP) and an Adjective Phrase (AP) (p. 339):

(8) a.

exceedingly many very beautiful women

b.

DP D

QP

exceedingly

Q′ AP

Q many very

A beautiful

A′ NP N women

Before we return to the Swahili data, note that, although the basic make-up of the DP in (6) has remained more or less unchanged since that time, the nominal middle field in (8) has undergone substantial revisions. For instance, in an influential paper, Ritter (1991) proposed a Number Phrase (NumP). Following that,

1.  I assume that adjunction of NP to DP is impossible (see Chapter 2, Section 3.2).





Demonstratives and definite articles as nominal auxiliaries

other intermediate functional phrases have been identified, although there is little consensus on the kind, number, or sequence of those phrases. Rather than surveying the literature (see Bernstein 2001b; Coene & D’hulst 2003; Longobardi 2001b; for typological overviews, see Plank 2003; Rijkhoff 2002), I proceed to my own goal: in the spirit of “charting out” the noun phrase, I will provide arguments for yet another phrase in the nominal Mittelfeld. I will argue that determiners move from a phrase I call Article Phrase (ArtP) to the DP. Assuming movement operations and that referential noun phrases have an overtly licensed DP (cf. Longobardi 1994), my proposal accounts for both Swahili patterns in (1) using one and the same syntactic representation. Base-generating the article in Art and the demonstrative in Spec,ArtP, I propose that the article moves in (1a) and the noun in (1b). Compare the corresponding simplified derivations in (9) (for a more detailed discussion in a Scandinavian context, see Chapter 2): (9) a. [DP yulei [ArtP Spec ti [NP mtu]]] b. [DP mtuk [ArtP yule tk [NP tk]]]

The ultimate goal of this kind of enterprise is to offer a uniform account of the different types of determiners and determiner systems across the world’s languages (for some typological discussion, see Schroeder 2006). However, this is too ambitious an objective to achieve here. Rather, the more immediate goal for this work is to shed some light on the distribution of determiners in some languages, primarily the Germanic ones. If on the right track, this discussion will also add evidence to the claim that the CP and DP are, in many ways, parallel. The remainder of the introduction is organized as follows: after illustrating the basic proposal in more detail, Section 3 motivates the assumed structure of the DP and briefly discusses issues such as concord and potential Relativized Minimality violations. In Section 4, I provide a survey of the individual chapters.

2.  Basic proposal Movement inside the noun phrase is not a novel claim. Perhaps the most wellknown proposal of movement involves (partial) N-raising (e.g., Ritter 1988; Taraldsen 1990; Cinque 1994; Longobardi 1994). Besides that, other types of movement inside the DP have been proposed: for instance, movement of possessors (e.g., Cardinaletti 1998; Alexiadou 2004), different kinds of (short) movement of quantifiers (e.g., Barker 1995:  p. 124; Zamparelli 2000:  264–5), movement of adjectives (e.g., Giorgi & Longobardi 1991: 123; Zamparelli 1993; Laenzlinger 2005), and (short) movement of determiners (e.g., Valois 1991b: 87, 138;



Introduction

Bernstein 1993: 128, 1997, 2001a; Bennis, Corver & den Dikken 1998: 97; Batllori & Roca 2000; Laenzlinger 2005). Relevant for present purposes, I focus on some cases of the last type. Some analyses involving short movement of a determiner base-generate the determiner in a phrase below the DP-level but above adjectives and subsequently move it to the DP. Brugè (1996, 2002) extensively argues that this kind of movement is longer; that is, that the demonstrative is base-generated in a phrase below all adjectives but above the head noun and its complements and that it is moved from there to the DP (for similar proposals, see Campbell 1996; Carstens 1991; Giusti 1997, 2001, 2002; Grohmann & Panagiotidis 2004; Panagiotidis 2000; Vangsnes 1999:  119–20; Wood 2007). Importantly, Delsing (1988:  p. 70) proposes that, besides movement of the demonstrative, articles may also undergo long movement (cf. also Vangsnes 1996: 17, 1999: 133–4; Alexiadou 2003: 15). Thus, building on the work by these authors, I intend to unify and generalize these proposals to the effect that demonstratives as well as (in-)definite articles undergo this long movement (for a preliminary version of this claim, see Roehrs 2002). Disregarding the details of the nominal Mittelfeld, the simplest form of my proposal can be represented as follows:2 (10)

DP D deti

XP adjective

ArtP Art deti

NP

2.  There are also cases where the determiner has apparently moved out of the noun phrase: (i) a.

am Ende at.the end ‘at the end’

(German)

b.

Vimo-lo neno. saw.we-the child ‘We saw the child.’

(Galician)

Whatever the correct analyses of these cases may be (cf. van Riemsdijk 1998: p. 651 for (ia) and Uriagereka 1995: 81 fn. 5 for (ib)), these explanations do not rely on movement of the determiner from a position below the adjective and thus will not be discussed here.





Demonstratives and definite articles as nominal auxiliaries

Before moving on, it is necessary to establish some terminology. I will refer to demonstratives and (in-)definite articles more generally as “determiners”. Furthermore, considering that unlike determiners, quantifiers are a more diverse group and can themselves be modified (e.g., Svenonius 1992: 102, 106), I take them to be similar to adjectives and thus not part of the group of determiners (cf. Ritter 1991 for Hebrew and Scabolczi 1987, 1994: 271 fn. 12 for Hungarian).3 Finally, I will refer to projections of the head noun as “noun phrases” or “nominal phrases” if their inner make-up is not important for the relevant discussion, and as “DPs” and “NPs” if it is. Proposing that determiners move from a low position, as in (10), we need to clarify their categorial nature. The focus of Abney’s work was to argue for a parallel structure between noun phrases and clauses.4 In particular, not only was the structural representation argued to be parallel in the two domains but also lexical items were taken to be counterparts of one another. Concretely, Abney (1987: 265) took determiners to be the equivalents of modal verbs, merging them in D.  Furthermore, he interpreted pre-nominal descriptive adjectives as the nominal equivalent of auxiliaries in the verb phrase (p. 267). However, there is evidence that Abney’s parallelism is not correct. As a consequence, I will propose below that determiners are similar to auxiliaries in the clause and that adjectives are the equivalent of adverbs. In short, I propose that determiners are – what I will call – “nominal auxiliaries”.5 Starting with the first parallelism, Harbert (2007:  p. 288) notes that English modals are somewhat special in the Germanic languages: for instance, they have defective agreement and do not seem to undergo movement from a lower position. In fact, they can never appear in a lower position, as shown in (11b–c): (11)

a. b. c. d.

She can(*s) go home. *I have (not) could go home. *I (do) not could go home. I could not go home.

3.  As we will see below, I am not claiming here that there are no syntactic or semantic differences between the different determiners, that is, between articles and demonstratives (see, e.g., Giusti 1997, 2002). Furthermore, Chapter 4 shows that quantifiers may have certain features of determiners. 4.  I will not take a stance here as to whether or not the noun phrase is similar to IP (e.g., Abney 1987; Mohammed 1988; Saito & Murasugi 1999), or CP (e.g., Szabolcsi 1981, 1987, 1994; Valois 1991b: p. 30; Gavruseva 2000), or both. The latter position could depend on the language (Horrocks & Stavrou 1987 propose that the noun phrase is similar to IP in English but CP in Greek). 5.  Just like “clausal auxiliary”, “nominal auxiliary” is a cover term for semantically different elements; for instance, there are not only substantive and expletive elements in the clausal domain, but, as we will see, also in the nominal one.



Introduction

Discussing the distribution of different kinds of verbs with regard to various classes of adverbs, Pollock (1989) proposes to split up the traditional inflectional phrase into a Tense Phrase and an Agreement Phrase. In order to account for the special properties of English modals (defective agreement, no movement), he suggests, essentially following work of Chomsky’s (1955 and subsequent work), that modals are base-generated in Tense (pp. 398–9). As they do not move through an Agr position by assumption, this lack of movement accounts for their defective agreement. Importantly, in contrast to English, modals in other languages do show agreement and undergo movement (note that German has a morphological restriction called infinitivus pro participio that, under certain conditions, spells out the past participle of the modal in (12b) as a second infinitive): (12) a.

Du kann-st nach Hause gehen. you can(2pers.sg) to home go ‘You can go home.’

(German)

b. Ich habe nach Hause gehen können. I have to home go could ‘I was able to go home.’ c.

Ich konntei nach Hause gehen ti. I could to home go ‘I could go home.’

Returning to determiners, we will see below that they exhibit agreement. In that respect, determiners are more similar to German, than to English, modals. As for movement, if the parallelism between the clause and the noun phrase also holds in this regard, then we are in fact led to propose that, similar to movement of auxiliaries in the clause (e.g., Emonds 1978; Pollock 1989, and Chomsky 1991), determiners also move in the noun phrase. This is one of the main points of this work. Taking negation as marking the verb phrase boundary, compare (13) through (15), where the auxiliaries for tense, aspect and passive appear in different positions, to (16), which provides examples for the general structure in (10):6 (13) a. Peter would never have done this. b. Peter hasi never ti done this. (14) a. Peter will not be working. b. Peter isi not ti working.

6.  Note that, even if one conceptualizes these auxiliaries as the spell-out of certain features, these features would presumably also undergo movement, later to be spelled out in different positions.



 Demonstratives and definite articles as nominal auxiliaries

(15) a. The task will not be completed. b. The task isi not ti completed. (16) a. thisi nice ti girl b. thei nice ti girl c. ai nice ti girl

If these general considerations are correct, then determiners are similar to modals and auxiliaries in the other, non-English languages. English modals, being exceptional, should not serve as a clausal counterpart of the determiner.7 I propose then that the categorial status of determiners is similar to that of auxiliaries in general (cf. Hudson 1984: 91), subsuming modals under the category of auxiliaries here. Although a movement analysis for determiners is admittedly less obvious from a West Germanic perspective, I will show below that a large number of facts follows from this assumption.8 The parallelism between verbal and nominal auxiliaries can be extended. Ritter (1995: 424) discusses Hebrew, where pronouns can function as copulas in the present tense (see also Panagiotidis 2002: 139; den Dikken 2006: 218–9): (17) a.

dan hu xaxam. Dan he smart(masc) ‘Dan is smart.’

(Hebrew)

b. sara hi xaxama Sara she smart(fem) ‘Sara is smart.’ c.

sara ve dan hem xaxamim Sara and Dan they smart(pl) ‘Sara and Dan are smart.’

Ritter proposes that third person pronouns consist of a NumP with specifications for gender and number and possibly a DP with features for person and definiteness. Although the analysis of the pronouns in (17) is perhaps not entirely straightforward, it seems that, as complex elements containing at least gender and number, they cannot be the spell-out of Tense (but perhaps the spell-out

7.  Interestingly, Matushansky (2006: 92, 104) argues that modal verbs in English do move to Tense from a position below negation. 8.  This does not exclude the possibility that certain cases might be analyzed by basegenerating determiners in different positions in the DP and establishing some construal relationship between them. However, there is a trend in current theorizing to reinterpret construal relations as instances of movement, which is what I simply claim here for all relevant cases.



Introduction

after a series of head movements). Whatever the exact account may be, for my purposes, it is enough to point out that these pronouns are, once again, different from modals in English: they are presumably not just merged in Tense and show agreement with the subject. In all these regards, the pronouns are closer in their behavior to auxiliaries, and, considering that pronouns are determiners (Postal 1966; Roehrs 2005a), I take this as another indication that (pronominal) determiners are auxiliary-like. Turning to the second categorial parallelism, adjectives are an open lexical class and are thus dissimilar from auxiliaries (pace Abney). Furthermore, Baker (2003: p. 230) provides evidence that they are similar to adverbs (also Valois 1991b; Cinque 1994; Crisma 1996). In fact, he argues that adverbs ending in -ly are adjectives. To appreciate the same behavior, note that, while adverbs may modify both a verb and an adjective in the clause, related adjectives can modify the corresponding noun in the noun phrase. Consider the (a)-examples vs. the (b)-examples: (18) a. Italy brutally invaded Albania. b. Italy’s brutal invasion of Albania (19) a. Chris is extremely shy. b. Chris’s extreme shyness

Similarly, degree words may modify both adverbs and adjectives:9 (20) a.

Die Musik wird zu laut gespielt. the music is too loudly played ‘The music is played too loudly.’

(German)

b. die zu laute Musik the too loud music ‘the too loud music’ c.

Die Musik ist zu laut. the music is too loud ‘the music is too loud’

To sum up these introductory remarks, I proposed that determiners are similar to auxiliaries in the clause and adjectives are the equivalents of adverbs.10 In the next section, we consider the structure of the DP in more detail. 9.  There are some restrictions in English on noun phrases containing degree words (see, e.g., Felber & Roehrs 2004). 10.  It is well-known that clausal auxiliaries derive from main verbs. Note now that rather than deriving from nouns, the definite article historically derives from the demonstrative and the indefinite article from the numeral for ‘one’. If this is so, then we also expect that clausal auxiliaries may, in certain ways, have a different relation to main verbs than nominal



 Demonstratives and definite articles as nominal auxiliaries

3.  Assumptions and their motivation While my analysis will be cast in terms of Chomsky’s (1995, 2000, 2001, and subsequent work) Minimalist Program, I intend the discussion to be general enough so that it is also of interest to syntacticians working in other frameworks. Most relevant for present purposes, I discuss and motivate the structure of the DP in this section, followed by some remarks on concord within the DP and on potential Relativized Minimality violations. 3.1  The structure of the DP As in the clause, there is some disagreement on the kinds, number, and sequence of the phrases in the nominal Mittelfeld. Following Julien (2002: 267, cf. also Julien 2005: 11; Anderssen 2005: 94), I will motivate the following hierarchical representation that, moving top-down, consists of a Determiner Phrase (DP), a Cardinal Number Phrase (CardP), an Agreement Phrase (AgrP = Julien’s αP), the proposed Article Phrase (ArtP [Julien’s 2005 nP]), a Number Phrase (NumP), a light noun Phrase (nP) and NP:11, 12 (21) [DP D [CardP Card [AgrP Agr [ArtP Art [NumP Num [nP n [NP N]]]]]]]

auxiliaries to nouns (see the discussion of the Head Movement Constraint below). Perhaps unsurprisingly, the proposed parallelism does not extend to all phenomena. 11.  As a main point of reference, I am using Julien (2002), rather than Julien (2005). This has two reasons: (i) whereas Julien (2005) generates the article in nP, I reserve the name of this phrase for the Larsonian nP-shell (also Anderssen 2005: 95); (ii) Julien’s (2002) article allows the reader more immediate access to her main ideas. 12.  Although this structure is sufficient for current purposes, it is presumably even more complex (see also later footnotes). For instance, there is another projection on top of the DP to host floating quantifiers (which is labeled Intensifier Phrase in Chapter 4), (ia), and there must be at least another phrase between the DP and CardP that can host possessive pronouns as in Italian (ib) (the datum is from Lyons 1999: 293):  (i) a.

alle diese Bilder all these pictures ‘all these pictures’

b.

i miei tre bei quadri the my three beautiful pictures ‘my three beautiful pictures’

(German)

(Italian)

Since possessives will not play a central role in the following discussion, I leave them aside here for the most part (but see Roehrs 2005b).



Introduction 

I consider the individual phrases in more detail. In keeping with traditional terminology, the DP is a projection of D, here assumed to be a null head. This null head is proposed to have three general properties that will become relevant in the discussion below. Semantically, D has a feature for definiteness and it underspecified for referentiality. While the former is a feature that will be valued by a determiner, the latter is a property that is brought about by movement of an element to D (cf. Longobardi 1994; also 1996, 2001a) or by movement of a phrase to Spec,DP (Julien 2002, 2005; Vangsnes 1996:  7). Syntactically, I assume that D has (abstract) case features that trigger movement of the DP to its surface position. Morphologically, certain null D’s need to be supported by another element. We will return to these properties in more detail in the following chapters. CardP hosts numerals or quantifiers in its Specifier position and as such it specifies the number or amount denoted by the nominal. Unlike adjectives, these elements may move to Spec,DP to bring about a strong quantificational reading (cf. Zamparelli 2000). Besides this difference, I provide two arguments that indicate that numerals or quantifiers in CardP have an independent status and, although similar to adjectives in some respects, should not be collapsed with them. First, unlike in literary Icelandic, (22a), the numeral can be “stranded” in common Icelandic. Consider (22b) as derived in (22c) (Vangsnes 1999: 146): (22) a.

hinar þrjár frægu bækur mínar the three famous books my ‘my three famous books’

b. frægu bækurnar mínar þrjár famous books-the my three ‘my three famous books’ c.

(literary Icelandic)

(common Icelandic)

[DP [AgrP frægu bækurnar mínar]i D [CardP þrjár ti]]

As far as I am aware, it is not possible to strand adjectives in Modern Icelandic (but see below for Old Norse). A second difference between numerals/quantifiers and adjectives comes from the discussion of so-called split NPs in German, where a “lower” part of a DP has apparently split off and moved to Spec,CP: (23) [Hemden] hat er immer nur [diese] getragen. shirts has he always only these worn ‘He always wore only these shirts.’

(German)

As discussed in Roehrs (2007a), basically all adjectives can be “stranded” below or split off and moved. Interestingly, Bhatt (1990: 251) points out that, while numerals

 Demonstratives and definite articles as nominal auxiliaries

and quantifiers can be “stranded” in split-NPs, (24a), they cannot be part of the splitoff itself, (24b) (my examples):13 (24) a. [Hemden] hat er immer nur [diese [drei / wenigen]] getragen. shirts has he always only these three / few worn ‘He always wore only these three/few shirts.’ b. *[Drei / Wenige [Hemden]] hat er immer nur [diese] getragen.    three / few   shirts has he always only   these worn ‘He always wore only these three/few shirts.’

As in the case of Modern Icelandic above, we once again find a difference between numerals and quantifiers, on the one hand, and adjectives, on the other. These two differences can be straightforwardly captured by assuming that the relevant elements are in different types of phrases (CardP vs. AgrP) and stipulating that only certain phrases can undergo the relevant operations.14 As far as the syntactic status of adjectives is concerned, I follow Cinque (1994) among others in that AgrP hosts an adjective phrase in its Specifier position.15

13.  Strong stress on the numeral or the quantifier improves the example. 14.  As already noted in footnote 12, the structure of the noun phrase must be more complex:  although a focused adjective cannot precede a quantifier, this is possible with a numeral (cf. Julien 2005: 13): (i) a.

die (*EHRLICHEN) vielen Studenten the       honest many students ‘the many honest students’

b.

die (EHRLICHEN) drei Studenten the    honest three students ‘the honest three students’

(German)

Since I will not focus on differences between quantifiers and numerals in what follows, I will not discuss the contrast in (i), although I assume that there is a focus position between the DP and CardP (cf. Giusti 1996). 15.  Note here that care must be taken not to infer the structural status of adjectives from their ability to have an “effect” on the determiner. For instance, while the determiner in Swiss German exhibits a long form when it (immediately) precedes an adjective and a short form when not (Leu 2001:  55, also Meyer 1967), the opposite state of affairs holds for French (cf. Radford 1993: 99). Importantly, no such difference seems to hold for Rumanian, where the determiner may suffix to both adjectives and nouns (Giusti 1994: 242): (i) a.

di alt autoban vs. the old highway ‘the old highway’

d autoban the highway ‘the highway’

(Swiss German)

b.

d’excellentes oranges vs. some excellent oranges ‘some excellent oranges’

des oranges (excellentes) some oranges  excellent ‘some excellent oranges’

(French)



Introduction 

Taking AgrP to be recurring (in the sense of Cinque 1999; Scott 2002), each phrase contains one adjective in its Specifier. Discussing two other alternatives, I provide two arguments for the structural analysis involving adjectives in Specifiers (see also Kester 1996b: 30–50).16 First, Svenonius (1993b: 445–6) observes that scopal facts show that degree elements such as barely take scope over the first adjective hot but not the second one black (see also Julien 2005: 7–8): (25) some barely hot black coffee

To be precise, while the coffee in (25) may be completely black, it cannot be extremely hot. In order to account for this restriction, Svenonius (1992, 1993b) argues that adjectives cannot be heads in the extended projection of the noun, as in (26), as barely would c-command both adjectives and thus take scope over both (this line of reasoning applies to both the NP being in the complement position of the adjective as in Abney 1987: 326–7 and Barbiers 1992 or in the Specifier position on the right as in Bhatt 1990 & Delsing 1991, 1993b: 81): (26)

QP Q barely

AP A hot

AP A black

NP coffee

Before we proceed, note that there is one potential argument for NPs being in the complement position of the adjective. As is well-known, complements of adjectives cannot follow but only precede their head in the noun phrase in Germanic, (27a). In order to explain this asymmetry, one could suggest that the complement is in Spec,AP in (26) and NP is c.

bǎtrîn-ul om vs. old-the man ‘the old man’

om-ul bǎtrîn man-the old ‘the old man’

(Rumanian)

I take the differences in (ia–b) to be due to language-specific mechanisms. 16.  Besides the three options to be discussed, there are actually two more proposals that I am aware of: one derives pre-nominal adjectives from (reduced) relative clauses (e.g., Kayne 1994: Section 8.4) and the other derives relative clauses from (sentential) pre-nominal adjectives (Fanselow 1986). Although critiquing Kayne’s approach with regard to some languages (pp. 312–3), Alexiadou & Wilder (1998) adopt it for Greek. Following Alexiadou & Wilder’s (1998) critique of Kayne for the languages at hand and considering the fact that Fanselow’s proposal has not been entertained for the languages under consideration in more recent work, I will not discuss these two proposals here.

 Demonstratives and definite articles as nominal auxiliaries

in the complement position of the adjective. However, other languages do allow a following complement, (27b) (the examples are taken from Julien 2005: 6 and Alexiadou & Wilder 1998: 311, also Pereltsvaig 2007): (27) a.

den [af sin datter stolte] mor the   of her daughter proud mother ‘the mother proud of her daughter’

(Danish)

b. i [periphani ja to jo tis] mitera the   proud for the son her mother ‘the mother proud of her son’

(Greek)

Furthermore, note that deverbal adjectives may take several complements in German that can appear in different orders. Compare (28a–b). This suggests that there are at least two phrasal positions associated with the adjective.17 Similarly, if degree modifiers are in a Specifier position, then the intervening degree element in (28c) also suggests that there are several Specifiers: (28) a.

ein [den meisten an Kenntnissen überlegener] Student a   to.the most in knowledge superior student ‘a student superior to most (others) with regard to knowledge’

b. ein [an Kenntnissen den meisten überlegener] Student a   in knowledge to.the most superior student ‘a student superior to most (others) with regard to knowledge’ c.

ein [auf die Kinder sehr stolzer] Vater a   of the children very proud father ‘a father very proud of his children’

In short, although initially plausible, the structure in (26) is not adequate to accommodate all the linguistic material related to adjectives. Furthermore, as pointed out by Giusti (2002: 84 fn. 15), an adjective as a lexical head in the extended projection of the noun goes against Grimshaw’s (1991) original proposal of extended projections. There is also evidence that the degree word itself is not a head in the extended projection of the noun. For instance, Haider (1992: 320) notes that degree elements such as sehr ‘very’ cannot be stranded in split-NPs: (29) a. [Kostbare Vasen] besitzt er nur [drei (*sehr)]. precious vases owns he only   three   very ‘He owns only three very precious vases.’ b. [Sehr kostbare Vasen] besitzt er nur [drei]. very precious vases owns he only   three ‘He owns only three very precious vases.’

17.  This point was inspired by a discussion with Michael Putnam.

(German)



Introduction 

Although one could give an account of the ungrammaticality in (29a) by stipulating that the degree word cannot be stranded, arriving at the opposite state of affairs as in (24), the facts follow directly from the assumption that sehr is in the Specifier of AP and syntactic constituency. Specifically, Corver (1991, 1997) and Zamparelli (1993, 2000: Chap. 7) argue that degree words are part of the extended projection of the adjective. If so, the adjective and noun could not be moved to the exclusion of sehr. Returning to Svenonius’ proposal, he suggests that the adjective phrase is adjoined to NP (also Valois 1991a) with barely in the Specifier of the AP, as in (30).18 Consider the relevant parts of the tree representations: NP

(30) AP QP barely

A hot

NP AP

NP coffee

A black

There is also evidence against (30). Going back to work of Delsing’s (1993b: 143), Julien (2002: 269–70) observes that indefinite determiners can follow pre-nominal adjectives in some Scandinavian dialects, (31). Importantly, these articles cannot be interpreted as adjectival agreement as each adjective has its own (strong) ending marked separately by -t. She suggests that the articles are not part of the AP but are the realization of the head Agr (her α). (31) ?eit stor-t eit styg-t eit hus a big(strong) an ugly(strong) a house ‘a big ugly house’

(Norwegian)

Collapsing both the scopal facts from above and the recurring determiners, we arrive at the following structure where the adjective projects its own phrase inside the Specifier position of Agr. Degree words such as barely are assumed to be in

18.  Actually, Svenonius (1993b:  446) suggests that barely is adjoined to AP. However, as Željko Bošković (p.c.) points out this would allow barely to c-command into the NP and black would be c-commanded after all.

 Demonstratives and definite articles as nominal auxiliaries

Spec,AP and the article moves through Agr, with the option of deleting some of its copies, as in the following English example: (32)

…

AgrP

AP QP barely

A hot

Agr′ AgrP

Agr a

Agr′

AP black Agr a



XP ... coffee

I will basically employ this structure throughout the discussion.19 Turning to ArtP, the proposed phrase, it has often been observed that there is a special relationship between adjectives and determiners. To set the stage, note that semantically vacuous determiners can be optional in unmodified predicate noun phrases or with proper names:20 (33) a.

Sie ist (eine) Lehrerin. she is    a teacher ‘She is a teacher.’

(German)

b. (Der) Hans ist gestern hingefallen. the Hans is yesterday fallen down ‘Hans fell down yesterday.’

19.  Below, I concentrate on descriptive adjectives. Note that Bernstein (1993) discusses adjectives in the Romance languages, arguing that their structural analysis may vary with regard to the type of the adjective:  adjectives of the type mere are heads in the extended projection of the noun (p. 61; also p. 71), pre-nominal adjectives are adjoined to NumP, and post-nominal adjectives are adjoined to NP (p. 55) (for ordering restrictions on adjectives, see Sproat & Shih 1991; Scott 2002; Julien 2005: 9; Svenonius 2007). I will, for the most part, stay agnostic about these other types of adjectives. 20.  English, for instance, is different here: predicates of this type take an obligatory determiner (Stowell 1991) and proper names do not allow a determiner.



Introduction 

However, when modified, the determiner becomes obligatory (the adjective arm ‘poor’ in (34b) is to be interpreted with its “non-restrictive” meaning, cf. Gallmann 1997: 75): (34) a.

Sie ist *(eine) gute Lehrerin. She is   a good teacher ‘She is a teacher.’

(German)

b. {Der arme Hans / *Armer Hans} ist gestern hingefallen. the pitiable Hans / pitiable Hans is yesterday fallen down ‘The pitiable Hans fell down yesterday.’

In other words, while the presence of the adjective is parasitic on that of the determiner, the determiner itself can appear independently. Bearing in mind that these determiners are semantically vacuous, their obligatory presence with an adjective (or, more generally, modifier, see Chapter 3) should be captured in a syntactic way. I suggest that this unidirectional relationship can be straightforwardly expressed in the current system by assuming that Agr selects either another AgrP or ArtP. For the short version of more arguments for ArtP, see the survey of the chapters in Section 4 below. As for NumP, taking Baker’s (1985, 1988) Mirror Principle extended in the sense of Pollock (1989) and Chomsky (1991) as a guide for the investigation, the Scandinavian languages provide some initial evidence that the number morpheme is lower than the suffixal determiner (Julien 2002: 268–9): (35) de röda bil-ar-na the red car-pl-the ‘the red cars’

(Swedish)

As the plural morpheme -ar is closer to the stem bil than the suffixed article -na, the phrase containing the determiner must be higher than that of the plural morpheme.21 With this in mind, and considering that these elements are below the free-standing determiner and the adjective, ArtP, NumP, and NP are assumed to be below the adjective and in the hierarchical relation ArtP > NumP > NP. Besides this first indication, there is other evidence that there is a phrase between ArtP and the head noun.

21.  In view of the discussion on multiple agreement in the DP by Spencer (1992), this can only be suggestive of the hierarchical order of the phrases.

 Demonstratives and definite articles as nominal auxiliaries

In order to account for the word order facts and Binding relations between the possessor and the complement of the head noun in (36), I propose that the head noun raises out of the thematic domain of the noun phrase (cf. Haider 1993: 23, 30; Julien 2005: 143):22 (36) die riesige Wutk des Mannesi tk auf sichi the enormous rage of.the man against himself ‘the enormous rage of the man against himself ’

(German)

Observing that the noun does not incorporate into the determiner in Art (*Wut-die ‘rage-the’) but does raise across the possessor, which is in Spec,nP (see momentarily), there must be another head position between ArtP and nP. In the discussion surrounding (35), I suggested this position to be Num. Finally, I assume that arguments of the noun are merged as part of a Larsonian nP-shell that consists of nP and NP (see Valois 1991b; Ticio 2003, among many others). I illustrate the structure that we have arrived at by fleshing out (21) with an example from Norwegian in (37) below, taken from Julien (2005: 1). Note that, unlike Julien, I interpret the noun forming suffix -ing to be under n:

22.  There is evidence that the possessor in (36) is higher than the complement. First, one the one hand, we know from extraction facts that possessors block the extraction of complements (Giorgi & Longobardi 1991:  68; Ticio 2003:  20–3); on the other, there are interpretative restrictions such that the possessor/agent must precede the theme (Harbert 2007: 150). Here are some examples from German that illustrate the interaction of agent and theme: (i) a.

Cäsars Eroberungen Caesar’s conquests ‘Caesar’s conquest’

b.

die Eroberungen Cäsars the conquests of Caesar

c.

Cäsars Eroberung Galliens Caesar’s conquest of Gaul ‘Caesar’s conquest of Gaul’

(German)

d. *Galliens Eroberung Cäsars Gaul’s conquest of Caesar The ungrammaticality in (id) follows from a c-commanding agent and an economy condition with regard to shortest movement. Second, there are arguments that binding relations cannot be established via feature movement in LF (see Lasnik & Uriagereka 2005: 170–1); that is, the surface order in (36), where the possessor is higher than the complement, also provides the relevant input for binding.



Introduction 

(37)

DP CardP

D QP

AgrP ArtP

AP Art

NumP nP

Num

n′

Poss n

NP PP

N

deii to gamle ti teiknk+ingj-a-nei mine tj the two old draw+ing-s-the my ‘my two old drawings of the town’

tk

P

DP

av of

byen town-the

Some remarks are in order here. I assume that this structure is universal, with surface differences derived by language-specific operations.23 Relevant here, the head noun raises via n to Num and the determiner raises from Art through Agr and Card to D (leaving out the heads Card and Agr for reasons of space, I only indicate the base position and the final landing site). In Chapter 2, I propose that similar to demonstratives (Brugè 1996, 2002), articles may be split up. While one part of the article (dei) moves to the DP, the “stranded” one in Art (-ne) undergoes Merger with the noun in Num in the Scandinavian languages. Finally, considering that noun phrases can be predicative or argumental, I assume that, with the

23.  For the claim that the noun phrase is parameterized with regard to the inner make-up, which in turn depends on the presence of the DP-level, see Bošković 2005; for a reply, see Pereltsvaig 2007 (also Franks & Pereltsvaig 2004).

 Demonstratives and definite articles as nominal auxiliaries

exception of NP, all other phrases are, at least in principle, “optional” projections. Before surveying the individual chapters, I briefly discuss two other issues. 3.2  Concord and movement inside the DP Different elements in the noun phrase are specified for different features, marked by “+” below. In particular, the head noun is specified for gender and the determiner for person and definiteness. Number and case are different in that they are not inherent features with number “originating” inside the DP (i.e., with Num) and case outside the DP (i.e., with the case assigner but it is valued on D). Finally, adjectives are typically not specified for any features and may exhibit overt reflexes of only some categories. For instance, while all features (except person) seem to be relevant for adjectives in the Scandinavian languages, definiteness is not shown in German. Consider the specifications of the individual elements in their base order (top down): (38)

Adjective Determiner Number Noun

[ Person/Definiteness; Number; Gender] [+Person/Definiteness; Number; Gender] [ Person/Definiteness; + Number; Gender] [ Person/Definiteness; Number; + Gender]

Although lower in the DP structure, we have arrived at the same hierarchical ordering of features as in Ritter (1991). Agreement between the clause and the noun phrase as well as concord inside the noun phrase are assumed to be brought about by long-distance agreement. In particular, I assume that the clausal case assigner (i.e., the verb, preposition, functional head) values D for case as in (39a). Turning to concord, I assume that elements inside the DP have to value their features. Specifically, I assume that elements value case and definiteness with D and gender with N.  Finally, in order to explain some cross-linguistic variation, I propose in Chapter 4 that number is valued in different phrases, in NumP in Scandinavian but in CardP in German. Importantly, basically following Bobaljik & Wurmbrand (2005), I propose in Chapter 2 that heads such as Card or Agr induce a separate agreement domain with regard to ArtP, marked by vertical line in (39b): (39) a.

Agreement Between the Clause and the Noun Phrase Case assigner –Case–> [DP D …]

b. Concord Inside the Noun Phrase [DP D Card Agr | Art Num N]

I assume there that long-distance agreement is only possible if the relevant elements are in the same agreement domain of the noun phrase; for instance, in order to value definiteness, the determiner and D must be in the same domain. If Card



Introduction 

or Agr are present, then these elements are not and they have to move into the other domain. With lowering movement banned, this explains the displacement of the determiners. Finally, we look at the different kinds of movements inside the DP with regard to potential Relativized Minimality violations. I concentrate on head movement of the noun and movement of the possessor. As argued above, the Germanic languages have partial N-raising, that is, raising of the head noun to Num. The Head Movement Constraint (Travis 1984) has been widely employed to explain the syntactic distribution of clausal auxiliaries and main verbs. If determiners are auxiliaries, then we could claim that nominal auxiliaries and head nouns are also subject to this kind of constraint. This suggestion is especially tempting since there seems to be evidence that shows that, if a head noun is higher than the base-position of the determiner, it must have moved there as part of a structurally larger element. With a larger element not subject to the HMC, movement across the determiner is allowed, which in turn could indirectly indicate the workings of the HMC. Consider some Romance languages in this regard. Although the traditional account of the high position of the head noun in these languages involves N-raising (Cinque 1994), numerous authors, including Cinque later himself, have noted that not only the head noun may raise but more linguistic material. For instance, Sánchez (1996: 213) concludes that an account other than N-raising must be invoked for examples such as (40a) and Cinque (2003), as cited in Julien (2005: 312), reaches the same conclusion for (40b): (40) a.

una [ladrona de joyas finas] torpe a   thief of jewelry expensive clumsy ‘a clumsy thief of expensive jewelry’

b. la [commedia musicale] americana the   comedy musical American ‘the American musical comedy’

(Spanish)

(Italian)

Other authors have reached the same conclusion (see Alexiadou 2001; Bouchard 1998, 2002; Lamarche 1991; Laenzlinger 2005; Svenonius 1993b:  451–2; Ticio 2003; Valois 1991b: 163–5; and Kester 1996b: 35; for a similar point for Hebrew and Arabic, see Shlonsky 2004 and, more generally, Cinque 2005). Thus, the compatibility of different types of movement  – head movement of the noun below ArtP and movement of a larger nominal element across ArtP – could imply the workings of the HMC. However, I doubt that the HMC restricts the interaction between the movement of the noun and that of the determiner. For the HMC to apply at all, it must be cast in minimalist terms, that is, the determiner and noun must have some feature in common. As the relevant elements have unrelated histories (see footnote 10), the identification of the shared

 Demonstratives and definite articles as nominal auxiliaries

feature is far from straightforward.24 If so, the HMC would not be applicable here to begin with.25 In Chapter 3, I propose that the nominal in Romance adjoins to a certain phrasal projection. Turning to possessor movement, Cardinaletti (1998: 19) argues that the possessive pronoun in (41a) is in its base-position and, when deficient, it must move to the pre-nominal position in (41b): (41) a.

la bella casa sua the nice house his/her ‘his/her nice house’

(Italian)

b. la sua bella casa the his/her nice house ‘his/her nice house’

The relevant movements for (41a–b) are schematized in (42a–b). In particular, I proposed above that the determiner is base-generated in ArtP and moves to the DP. Furthermore, I suggested that the nominal containing the head noun adjoins to a phrasal projection, marked here as XP. Finally, with the possessive pronoun originating in Spec,nP, let us suppose that it moves to Spec,PossP: (42) a. [DP lai [PossP [AgrP bella [XP casak [ArtP lai [nP sua [NPcasak]]]]]]] b. [DP lai [PossP suaj [AgrP bella [XP casak [ArtP lai [nP suaj [NPcasak]]]]]]]

Assuming that the possessive is in a Specifier position, the question arises why the possessive may move across another Specifier, the adjective, in (42b). This is another potential Relativized Minimality (RM) violation. We know from discussions of argument structure inside the DP and extraction facts (e.g., footnote 22, also Valois 1991b; Mallén 1991; Ticio 2003) that there are restrictions on movement. In other words, we cannot simply stipulate that RM does not hold in the noun phrase. In order to allow for these crossing movements, another solution must be found. Discussing the different blocking effects of quantificational adverbs and modifiers in the A′-system, Rizzi (2001: 104) argues that restrictions on movement

24.  Note that phi-features are probably not relevant here as, for instance, the Romance languages exhibit clear cases where a reordering of the noun and another nominal element (the adjective) has taken place. 25.  This means that the present proposal, which base-generates the determiner in a low position, is, in principle, compatible with (high) N-raising. As most of the discussion focuses on Germanic, this question will not play an important part. As a consequence, I wish to remain agnostic about this issue here and I will use the neutral term ‘nominal raising’ for the displacement of the Romance type.



Introduction 

cannot be accounted for by a purely geometrical approach, that is, Specifiers blocking the movement of other Specifiers (as is sometimes claimed in the literature dealing with movements inside the DP). Rather, minimality is determined within classes of features, but not across them. For present purposes, I assume then that possessives and adjectives do not share the same relevant class(es) of features.

4.  Overview of the chapters In the last two sections, I provided the main hypothesis of this work and motivated the most relevant background assumptions with regard to the structure of the DP. Assuming that determiners are base-generated in a low ArtP and move to DP, a number of phenomena can be accounted for in a straightforward way. In fact, we will see evidence that this movement proceeds in a successive-cyclic fashion. Concentrating on the Germanic languages, especially the Scandinavian languages and German, I now turn to a brief overview of the individual chapters. My goal here is to illustrate the kind of problems this approach is meant to address and to provide some idea as to what theoretical conclusions will be reached. The remainder of this book is organized as follows: Chapter 2 is concerned with the syntactic distribution of determiners in the Scandinavian languages. In Chapter 3, I deal with the semantic effects of different determiner copies with regard to the non-/restrictive readings of modifiers in the Scandinavian languages. Chapter 4 discusses some morpho-syntactic consequences for German; in particular, it provides an account of the weak/strong alternation of adjective endings. In Chapter 5, the main results are summarized. 4.1  Chapter 2 This chapter focuses on the syntactic distribution of determiners. As is well-known, the Scandinavian languages have an interesting morpho-syntactic asymmetry between homogenous simple DPs, (43), and diverse modified DPs, (44): (43) a.

mann-en man-the ‘the man’

(Swedish, Norwegian)

b. mand-en man-the ‘the man’

(Danish)

c.

maður-inn man-the ‘the man’

(Icelandic)

 Demonstratives and definite articles as nominal auxiliaries

(44) a.

den gamle mann-en the old man-the ‘the old man’

b. den gamle mand the old man ‘the old man’ c.

gamli maður-inn old man-the ‘the old man’

(Swedish, Norwegian)

(Danish)

(Icelandic)

Bearing in mind that both simple and modified DPs can basically be used in the same semantic contexts (e.g., referential uses), the variation in (44) and its lack in (43) is surprising. The apparent difference between (43) and (44) is the absence vs. presence of the adjective. I will assume that the determiner values the feature [def] on D. In a simple DP, this valuing is assumed to take place long-distance. In other words, the determiner in ArtP does not have to move to the DP. As for a modified DP, I propose that adjectives, or more generally modifiers, induce a separate agreement domain with regard to ArtP and, consequently, long-distance agreement cannot take place (cf. Bobaljik & Wurmbrand 2005). With a modifier present, the determiner must move to DP to value features on D.  I propose that languages differ in the way they circumvent the “blocking” effect of the modifier: illustrating that both determiners in (44a) have different semantic import, I will suggest that the article is “split up” in these cases; one part moves to DP, the “stranded” part undergoes Merger with the partially raised head noun. Unlike Swedish and Norwegian, Danish and Icelandic do not allow splitting of their articles. The difference between the latter two languages is that Danish moves its article to DP overtly and Icelandic covertly. Finalizing the derivations for the simple DPs in (43) and Icelandic in (44c), the phrasal node containing the article and the noun (ArtP) in the former case and the phrase containing the adjective, article and noun (AgrP) in the latter case move to Spec,DP to license the DP with regard to referentiality. Finally, the article in Art undergoes Merger with the noun. To sum up: what is new in this discussion of the Scandinavian noun phrase is the interpretation of modifiers as inducing a separate agreement domain (which triggers movement), that articles can be split up, and that the suffixal determiner is brought about by Merger. The first two points are crucially related to the hypothesis that determiners move from ArtP to DP. Additional support for this claim is provided by the discussion of the rise of the suffixal determiner in Early Scandinavian, where we revisit the Panchronic Alternation, and by



Introduction 

cross-linguistic evidence for a lower position of determiners and the possibility of splitting them up. 4.2  Chapter 3 This chapter continues the syntactic discussion of Chapter 2 but concentrates on certain interpretative effects that take the syntactic distribution of determiners into account. In particular, in contrast to the cross-linguistic variation of DPs modified by adjectives, there is another interesting morpho-syntactic asymmetry. This asymmetry is within one and the same language, that is, between homogenous DPs modified by adjectives, (45), and heterogeneous DPs modified by restrictive relative clauses, (46). This inner-language “variation” with relative clauses seems to be parallel to the inter-language diversity with adjectives in (44), where (46a) presents the “Swedish” pattern, (46b) the “Danish” one, and (46c) the “Icelandic” form: (45) den gamle mann-en the old man-the ‘the old man’

(Swedish)

(46) Har du sett _____ som ska köra vårt plan? have you seen who will drive our plane

(Swedish)

a.

den pilot-en the pilot-the

b. den pilot the pilot c.

pilot-en pilot-the

‘Have you seen the pilot that will drive our plane?’

Assuming that a separate agreement domain is also induced by the presence of a relative clause, this morphological asymmetry is proposed to follow from the uniform Specifier position on the left for adjectives and the different analyses for relative clauses both with regard to various adjunction sites on the right and an alternative, head-raising account. Based on these syntactic remarks, I will turn to the main point of this chapter. The discussion of the syntactic variation with adjectives in (44) and relative clauses in (46) allows for an account of the different interpretations of modifiers with regard to non-/restrictiveness at little extra cost. Interpreting determiners as scope-bearing elements, the restrictive interpretation is derived by interpreting

 Demonstratives and definite articles as nominal auxiliaries

the determiner in its derived position in the DP and the non-restrictive one by interpreting it in its base-position in ArtP: DP

(47)

XP

D determiner

relative clause

XP adjective Art determiner

ArtP NP noun

The discussion of (47) provides a simple solution to the “Partee-Chomsky debate”.26 As for appositive modifiers, I suggest that they come in two types, a “syntactic” and a “non-integrated” one. Discussing the former in more detail, I propose that semantic atoms (types ·eÒ and ·tÒ) can be sent off individually in a model of multiple semantic spell-out. Arguing that appositive modifiers are propositions, multiple spell-out explains how these modifiers are interpreted as conjunctions to their hosting proposition although, syntactically, they are part of a DP. To reiterate the main new ideas: rather than moving the modifier to derive the different interpretations, the determiner itself moves and is interpreted in different positions. Non-restrictive modifiers are propositions interpreted as conjunctions to their hosting proposition in a model of multiple semantic spell-out. The discussion is extended to adjective interpretations in the Romance languages. 4.3  Chapter 4 This chapter discusses some morphological reflexes of the proposal in German. Although admittedly more indirect, I argue that the weak/strong alternation of adjective endings in German provides evidence that determiners move from ArtP to the DP in a successive-cyclic fashion. To this end, the Principle of Monoinflection in German will be discussed at some length. For current purposes, let me interpret this well-known morphological principle as follows:

26.  Briefly, while Partee (1976) claimed that restrictive modifiers combine first with nouns before their intersection combines with determiners, Chomsky (1975) pointed out that this does not always seem to be the case.



Introduction 

(48) Principle of Monoinflection The first category within a noun phrase carries the strong and the second one the weak ending (where category stands for instances of determiners or adjectives).

To illustrate, the two adjectives in (49a) belong to the same first category and both modifiers have a strong ending. Importantly, when a determiner precedes them, (49b), this new category carries the strong ending and all the following adjectives have a weak inflection:27 (49) a.

heißer schwarzer Kaffee hot(nom.strong) black(nom.strong) coffee(masc) ‘hot black coffee’

b. der heiße schwarze Kaffee the(nom.strong) hot(weak) black(weak) coffee(masc) ‘the hot black coffee’

In order to explain this distribution, I basically follow Sauerland (1996) in assuming that this alternation is brought about by a mechanism called Impoverishment. In particular, I propose that determiners are marked for triggering Impoverishment. Base-generating them in ArtP and moving them in a successive-cyclic fashion to the DP, I suggest that the underlyingly strong endings of the adjectives are “impoverished” in their feature specifications and are eventually spelled out as weak inflections. This process can be schematically represented for (49b) as follows: (50)

DP de-ri

AgrP Agr′

heiße-(r) de-ri

AgrP

schwarze-(r)

Agr′

de-ri

ArtP de-ri

NP Kaffee

27.  The following abbreviations will be used throughout:  nom = nominative case, acc = accusative, dat = dative, gen = genitive; masc = masculine, neut = neuter, fem = feminine, and pl = plural.

 Demonstratives and definite articles as nominal auxiliaries

Besides providing a morpho-syntactic argument for the main proposal of this book, we will also arrive at some interesting language-specific conclusions:  (i) determiners may move to the DP at different times during the derivation and (ii) singular structurally case-marked elements are special in German. At the end of this chapter, the discussion is extended to combinations of pronouns and nouns (and some other constructions). Assuming that pronouns are determiners, I propose that, like certain determiners discussed in the first part of the chapter, they may also move to the DP at different times. This explains the strong ending on the adjective in (51a) and the weak inflection in (51b): (51) a. {kein / ich / du} armer Idiot {no / I / you} poor(strong) idiot(masc) ‘no/I/you poor idiot’ b. {keine / wir / ihr} armen Idioten {no / we / you} poor(weak) idiots ‘no/we/you poor idiots’

More generally, if we assume that like clausal auxiliaries, determiners are basegenerated in a low position and move to the left, then these syntactic, semantic, and morpho-syntactic phenomena, although apparently unrelated, find a uniform account. I believe that this renders the main proposal, at the very least, interesting and, hopefully, worthy of further investigation.

chapter 2

The syntactic distribution of determiners 1.  Introduction In this chapter, I discuss the syntactic distribution of determiners in the Germanic and some other languages. I show that determiners may occur in different positions. This proposal is straightforward for “heavy” elements such as demonstratives. Briefly illustrating here with Spanish, Bernstein (2001a) argues that the higher demonstrative in (1a) is focally unmarked whereas the lower demonstrative in (1b) has a (non-constrastive) focus interpretation: (1) a.

este libro viejo this book old ‘this old book’

(Spanish)

b. el libro viejo este the book old this ‘this old book’

Below, I propose in detail that the demonstrative in (1a) is in DP and that the one in (1b) is in ArtP. These positions are argued to be related by movement. Furthermore, if we assume that (strong) pronouns are similar to demonstratives (cf. Postal 1966), then these determiner-like elements also provide evidence for two positions. As discussed in Roehrs (2005a:  footnote 15), Italian pronouns occur in a high position when their hosting noun phrase is an argument of a verb and in a low position in vocatives:1 (2) a.

Noi fortunati linguisti possiamo leggere molti libri. we lucky linguists can read many books ‘Us lucky linguists can read many books.’

(Italian)

1.  Vocatives have a different syntax from arguments. For instance, as argued by Cardinaletti (1994), vocatives do not necessarily project a DP-level (cf. Povero lui! ‘Poor you!’). Assuming that there is no DP in the Italian example in (2b), the pronoun must be pronounced in a lower position. Note also that pronouns can be in lower positions even inside arguments, as pointed out for English by Schütze (2001: 215 fn. 14):

(i)

Lucky {us/*we} linguists have to explain our profession to everyone.

Since vocatives deserve a separate treatment, I will not discuss them in much detail here.

 Demonstratives and definite articles as nominal auxiliaries

b. Fortunati noi linguisti! lucky we linguists ‘us lucky linguists’

In contrast to demonstratives and (strong) pronouns, articles usually occur in the left periphery of the noun phrase. In view of the fact that these elements are unstressed, that is, “light”, I assume that they are clitics that, similar to clitics in the clause, have to appear in a higher position for independent reasons. Consequently, it is harder to find evidence for a lower position from these light elements. However, if articles are suffixes, then, due to categorial restrictions, these determiners may appear in lower positions. Of particular interest here are the Scandinavian languages, where suffixal determiners usually attach to nouns (but see Julien 2005: 63–4 for some exceptions in this regard). This low position can be illustrated with Modern Icelandic as well as with Norwegian and Swedish. The latter two languages have both a suffixed and a free-standing determiner, which I will refer to as the “Double Definiteness effect”:2 (3) a.

gamli maður-inn old man-the ‘the old man’

b. den gamle mann-en the old man-the ‘the old man’ c.

den gamle mann-en the old man-the ‘the old man’

(Modern Icelandic)

(Norwegian)

(Swedish)

I will show below that similar to the Spanish demonstratives above, the two determiners in (3b–c) have different semantic import. Like the demonstratives (Brugè 1996, 2002), then, I propose that articles also move to the DP. In fact, I will argue that both demonstratives and articles can be split into two parts, where one part moves to the DP and the other remains in situ. This can be illustrated in the following schematic way:

2.  With regard to Double Definiteness, Alexiadou (2003) discusses Greek, Hebrew, and Swedish. She concludes that this phenomenon has different properties in each of these languages. In view of this conclusion, the language-specific details of the following discussion are meant to be applied to the Scandinavian languages only.





The syntactic distribution of determiners 

(4)

DP detαi

AgrP

adjective

ArtP

detαi+detβ

NP

Since this determiner split just alluded to only occurs if something intervenes between ArtP and DP, I propose that the head of the phrase hosting these “interveners” (e.g., Agr) induces a different agreement domain with regard to its complement (cf. Bobaljik & Wurmbrand 2005). This agreement domain “blocks” long-distance agreement between ArtP and DP but not movement of the determiner. If this interpretation of the facts is tenable, then this discussion provides an argument for the main proposal of this work. The chapter is organized as follows: after motivating the lower position of ArtP by illustrating different word orders in Early Scandinavian (Section 2), I formalize the diachronic reanalysis of the demonstrative as an article in Section 3. In order to account for the variation and change, the determiner is proposed to move from ArtP to DP. Section 4 provides more cross-linguistic evidence for a lower ArtP and movement of determiners. Before the conclusion, I provide my own account of the Scandinavian DP and compare it to Julien’s (2002 [2005]) analysis.3

2.  W  ord order possibilities of demonstratives and articles in Early Scandinavian Recall from Chapter 1 that historically related determiners may have a different syntactic distribution: (5) Panchronic Alternation Diachronically related elements may occur in different synchronic positions.

To illustrate this, I briefly discussed the distribution of articles and demonstratives in Swahili. I now turn to a more detailed discussion of a similar phenomenon in the Scandinavian languages.

3.  As already mentioned in Chapter 1, I will usually refer to Julien’s original work.

 Demonstratives and definite articles as nominal auxiliaries

In this section, I formalize in X′-theoretic terms the change from the Early Scandinavian demonstrative hinn to the article:  pre-nominal (hinn) in literary and post-nominal (-inn) in common Modern Icelandic. (6) a.

hinn gamli maður the old man ‘the old man’

b.

gamli maður-inn old man-the ‘the old man’

(literary Modern Icelandic)

(common Modern Icelandic)

Considering that these articles have the same historical source but different contemporary distributions, this presents another case of a Panchronic Alternation. In order to explain this, I propose that, while the pre-nominal, free-standing article moves to D overtly, the post-nominal, suffixal one raises covertly. As we will see below, this means that unlike literary, archaic Icelandic, common Icelandic has lost a (generalized) EPP feature.4 First, I consider three stages of Early Scandinavian, illustrating different word orders of the demonstratives and articles with regard to the head noun. 2.1   Proto-Scandinavian Proto-Scandinavian is attested in runic inscriptions in the Elder Futhark (2nd–8th centuries). In these inscriptions, there are two demonstratives: sá and hinn. The demonstrative may precede the noun, regardless of whether it is a form of sá (7a) or hinn (7b):5 (7) a.

þat azina this stone-slab ‘this stone slab’

(By, RäF 71)

4.  To be clear, the following discussion will concentrate on how this historical change occurred structurally and derivationally but not on why and how it spread (for discussion, see among others Abraham 1997, 2007; Diessel 1999: chap. 6; Leiss 1994; Lyons 1999: chap. 9; Oubouzar 1997; Philippi 1997; Vincent 1997). Note that the next two sections are based on Roehrs & Sapp (2006). For some other issues not discussed here, see their original paper. 5.  Determiners and head nouns agree with regard to gender, case, and number, which I do not mark here. The examples in Sections 2.1 and 2.2 are taken from RäF (Krause & Jankuhn 1966) and Södermanlands/Upplands Runinskrifter. I follow the convention of transcribing the runic inscriptions with bold, lower-case letters. For clarity, I also use bold print for indicating the pronounced elements in the derivations.



The syntactic distribution of determiners 

b. a hitt lant to this land ‘to this land’

(Eggja; RäF 101)

However, the demonstrative may also follow the noun, both with sá (8a) and hinn (8b). (8) a. runaz þaiaz runes these ‘these runes’

(Istaby, RäF 98)

b. hali hino stone this ‘this stone’

(Strøm, RäF 50)

Neither of these orders is dominant: the entire corpus consists of four examples of the sequence demonstrative-noun (dem-N) and three of the one noun-demonstrative (N-dem). There are no examples that contain both a demonstrative and an adjective. 2.2  Common Scandinavian After the 9th century, the North Germanic runic inscriptions are in a different alphabet (the Younger Futhark), and there are many more inscriptions. At this stage, there are three types of demonstratives: the old sá, which is now the distal demonstrative ‘that’; sási/þessi, a strengthened form of sá with the local meaning ‘this’; and hinn. At this period, the order N-dem has reached 98% for sási/þessi (cf. Perridon 1996: 252), whereas the distribution of sá is more diverse. Most relevant for current concerns, (h)inn frequently occurs with referential elements such as the proper name-like element kristr ‘Christ’. While the free-standing, postnominal (h)inn in (9a) presumably formed the basis for the suffixed determiner, the datum in (9b) is of particular interest as this determiner seems to be well on the way to becoming an article: (9) a.

kristr hin helgi Christ the holy ‘Christ, the holy’

(Uppland 391)

b. in heilhi kristr the holy Christ ‘the holy Christ’

(Södermanland 125)

Specifically, this combination of determiner and referring nominal shows that in some cases, (h)inn has lost its deictic force as a demonstrative and may now have properties of an expletive; that is, the determiner in (9b) seems to have a purely syntactic function.

 Demonstratives and definite articles as nominal auxiliaries

Importantly, at the beginning of this period, the determiner never occurs in a simple DP, but is found only when an adjective is present. Compare (10a) and (10b), where the determiner in (10b) is suffixed to the head noun: (10) a.

kunar … lit kiara mirki fr sial … Gunnar … let make monument for soul … ‘Gunnar … had (the) monument made for (the) soul …’

(Uppland 312)

b. tati iok … mirki-t mikla eftiR faþur sin (Södermanland 41) Tate cut … monument-the big after father his ‘Tate carved the big monument in memory of his father’

In the 11th century (cf. Noreen 1970: 316), we find the first attestation of the article hinn with no adjective (the example is taken from Wessén 1970: 30): (11)

kuþ hialbi ant-ini god help soul-the ‘God help the soul.’

(Uppland 669)

Crucially, the article is in its suffixal form, suggesting it originated in phrases like (10b), where adjectives were present. 2.3  Old Norse Old Norse (13th–15th centuries) is directly descended from Common Scandinavian.6 It has reversed the relative order of the noun and the demonstrative: sá and þessi usually appear as dem-N. As for (h)inn, it has split into two distinct functions. The first is the article, no longer having deictic force. When an adjective is present, the article occurs pre-adjectivally, (12a–c); when there is no adjective, the article appears post-nominally, (12d). The second function of (h)inn is that of a demonstrative, where it can appear in a position not available to the article, that is, directly before the noun, as in (12e). Furthermore, it can occur before the adjective, as in (12c). In other words, (h)inn in (12c) is ambiguous between an article and a demonstrative: (12) a.

maðr-inn gamli man-the old ‘the old man’

(Old Norse)

6.  Throughout the discussion, I am using the term ‘Old Norse’, which is a neutral cover for Old Icelandic and Old Norwegian.



The syntactic distribution of determiners 

b. maðr (h)inn gamli man the old ‘the old man’ c.

(h)inn gamli maðr the/that old man ‘the/that old man’

d. maðr-inn man-the ‘the man’ e.

(h)inn maðr that man ‘that man’

While the presence of ‘h’ can presumably be taken as an indication that the determiner was a free-standing element, its lack in (12b–c) does not necessarily indicate that it is a clitic or an affix. 2.4  Schematic summary The demonstrative hinn gradually developed into a definite article, as shown in table 1. Only a demonstrative in Proto-Scandinavian, the first clear instances of (h)inn as an article occur in modified DPs in Common Scandinavian. In Old Norse, we find the article suffixed to nouns not only in modified noun phrases but also in simple DPs (for bracketed hinn in row two, see footnote 15). Table 1.  Kinds of Demonstratives and Articles in early North Germanic. Language

Demonstrative

Article

Proto-Scandinavian Common Scandinavian Old Norse

sá, hinn sá, þessi [hinn] sá, þessi, (h)inn

– (h)inn (before adjectives only) (h)inn (before adjectives) and -inn (clitic)

Table 2 summarizes the possible positions of the demonstratives and definite article in the various stages for the unmodified DP. After an equal distribution in Proto-Scandinavian, Common Scandinavian shows a clear preference for N-dem. In the latter language, we begin to find the first clear instances of articles (in modified DPs; cf. table 1). With the further advanced development of the definite article in Old Norse, a division of labor seems to be developing in the simple DP between the pre-nominal position used by the demonstrative (reversing the Common Scandinavian preference) and the post-nominal position used by the article.

 Demonstratives and definite articles as nominal auxiliaries

Table 2.  Positions of Demonstratives and Articles in Early North Germanic simple DPs. Language

dem-N

N-dem

Proto-Scandinavian Common Scandinavian Old Norse sá, þessi (h)inn

+ few + + (demonstrative)

+ + few + (clitic article)

To be clear, then, while simple DPs in Old Norse can only have a suffixal determiner, modified noun phrases exhibit a suffixal or a preceding free-standing determiner. Abstracting away from the position of the adjective, this is basically the state of affairs in Modern Icelandic.

3.  Formalizing the development of the suffixed article In the last section, we saw that the determiner system in Early North Germanic underwent some great changes. I now turn to some of these changes and make them more formal. Again following Roehrs & Sapp (2006), I first provide some evidence that demonstratives are phrases (see also Section 4.2. below) and, as such, they are assumed to be in Specifier positions. Next, considering the order N-dem, I discuss three potential analyses, concluding that the demonstrative is base-generated in a lower Specifier position (Spec,ArtP) and that the head noun moves across that Specifier to the DP. Then, I suggest that this resulting order forms the basis of the reanalysis of the demonstrative in the Specifier position as an article in a head position. This reanalysis occurs in a local domain, namely in one and the same phrasal projection: (13) …

XP

dem



X′ X

ZP

Finally, I briefly discuss some advantages and consequences of this proposal. 3.1  Demonstratives are phrases Among many others, Boucher (2003), Brugè (1996, 2002), Campbell (1996), Carstens (1991), van Gelderen (2004a: p. 73, 2007), Giusti (1997, 2002), Panagiotidis (2000),



The syntactic distribution of determiners 

Philippi (1997), Roehrs & Putnam (2008), Valois (1991b:  138–140), and Vincent (1997) argue that demonstratives are phrases and are thus in Specifier positions. In what follows, I report one of these arguments and add one of my own. In Italian, there is evidence from extraction of preposition phrases that demonstratives are different from definite articles. Consider first an example where all elements are in situ, (14a). Now, while the extraction of the PP is fine with definite articles, (14b), the extraction is ungrammatical with demonstratives, (14c) (the data are from Giusti 1996: 107): (14) a.

Conosco il presidente di questa associazione. I-know the president of this association ‘I know the president of this association.’

(Italian)

b. Di che associazione conosci il presidente? of which association you-know the president ‘Of which association do you know the president?’ c. *Di che associazione conosci questo presidente? of which association you-know this president ‘Of which association do you know this president?’

In order to account for this difference, I follow Giusti in that the definite determiner is in D and the demonstrative is in Spec,DP. With the “escape hatch” filled in the latter case, extraction is impossible. Consider a second argument. Demonstratives and (certain) possessives differ in that the former “trigger” a weak ending on the adjective whereas the latter co-occur with a strong adjective. Compare (15a) to (15b): (15) a.

dieses große Glück this great(weak) happiness ‘this great happiness’

(German)

b. mein großes Glück my great(strong) happiness ‘my great happiness’

When the demonstrative precedes the possessive, the adjective has the inflection, triggered by the possessive, (16a), and the possessive itself cannot have a weak ending either, (16b). In other words, unlike in (15a), the higher demonstrative does not have a morphological impact on the DP: (16) a.

dieses mein großes Glück this my great(strong) happiness ‘this my great happiness’

b. diese meine(*n) Freunde these my(strong/*weak) friends ‘these my friends’

 Demonstratives and definite articles as nominal auxiliaries

I suggest that this demonstrative is not part of the DP proper. In Chapter 4, I propose that there is an Intensifier Phrase (IntP) on top of the DP. Assuming that the weak inflection is brought about by Impoverishment, I propose there that only determiners in head and adjoined positions can trigger this morphological process. In view of the fact that the demonstratives in (16a–b) do not trigger Impoverishment, I suggest that they are in the Specifier of IntP. Relevant for current purposes, this is a phrasal position and provides a second argument for the bigger structural status of the demonstrative. Note here that the distribution in (16a–b) is a “marked” option as there are certain restrictions on it; for instance, the demonstrative usually precedes the possessive but not vice versa. This follows from the assumption that IntP is only available for intensifying elements.7 Finally, although the demonstrative is not part of the DP proper, there is evidence that it is part of the noun phrase as a whole. Besides the general fact that this construction can be followed by a finite verb (V2 effect), demonstratives can also be “sandwiched” between quantifiers and the DP proper (cf. also Vater 1991: 28–9; Duden 1995: 286; Bhatt 1990: 217): (17) a.

Doch diese meine Freunde haben mir nicht geholfen. yet these my friends have me not helped ‘Yet, these my friends have not helped me.’

b. alle diese meine Freunde all these my friends ‘all these my friends’

In Chapter 4, I propose that this alle ‘all’ is also an intensifier, located in an even higher position in the left periphery of the DP. In this section, I have provided two arguments that demonstratives are phrasal. Before I turn to the details of the change from the demonstrative to the article, let us consider the structure of the assumed source of the reanalysis, the order N-dem. 3.2  Three options to derive the order N-dem In the last section, I argued that demonstratives are in phrasal positions. In Section 2, I illustrated that demonstratives can appear both before and after the head noun

7.  Note also that Wood (2007: p. 345) argues that the sequence “demonstrative+possessor” is special in that there is a strong preference to use the proximal demonstrative. The latter seems to function as some type of focus marker. She analyses these constructions in Modern English as involving apposition.



The syntactic distribution of determiners 

in Proto-Scandinavian. This then raises the question of how the order N-dem is to be analyzed. The relevant data are repeated here for convenience: (18) a.

runaz þaiaz runes these ‘these runes’

b. hali hino stone this ‘this stone’

In order to derive the sequence N-dem, we could assume that either the demonstrative is in a different base-position, that the head noun (as part of an XP) moves across the demonstrative, or a combination of these two assumptions. In what follows, I discuss these three options, concluding that the base position of the demonstrative is not in Spec,DP but lower in the structure and that the noun (as part of an XP) moves across it.8 As a first option, one could assume that the demonstrative is in Spec,DP and that this Specifier position is on the right: (19)

DP D′ (D)

NP N

Spec dem (D)

However, following Kayne (1994), I assume that Specifiers on the right are universally disallowed, thus concluding that this option is not available. Likewise, I assume that adjunction to the right of this sort, that is, of the demonstrative to the right of the noun phrase is universally disallowed. As a second option, one could again propose that the demonstrative is in Spec,DP but that this Specifier position is now on the left, as in most standard accounts. In

8.  Actually, the sequence N-dem is structurally ambiguous in that all elements can be assumed to be either in one noun phrase, as discussed in the main text below, and in two juxtaposed noun phrases, as in (i):

(i)

[noun] [dem]

I will concentrate on the former case as it provides the basis for reanalysis of the demonstrative and suffixation of the article in a lower position (see footnote 15). Furthermore, assuming such a low position of the determiner, the transition from Old Norse to the two dialects of Modern Icelandic proceeds straightforwardly.

 Demonstratives and definite articles as nominal auxiliaries

order to derive the sequence N-dem, one could suggest that the head noun, contained in another phrase (XP), moves across Spec,DP to a higher position: (20)

?P XPi N

DP Spec dem

D′ D

... ti ...

This option also presents a number of problems. Consider two interpretations of ?P, the first involving movement due to Scrambling, the second movement due to feature valuing. First, if ?P in (20) equals DP, then XP has moved to adjoin to DP, presumably due to Scrambling. In Section 3.1, I observed that a phrasal position to the left of DP is a very “marked” option and is typically only possible with demonstratives (diese meine Freunde ‘these my friends’). Furthermore, the adjunction in question would involve movement to an adjoined position of an argument, banned by Chomsky (1986).9 I conclude then that this option is not viable. In the second scenario, ?P in (20) equals YP. I assume that YP stands for some phrase level in the left periphery of a split-DP (see Haegeman 2004: 236–9, extending ideas of Rizzi’s 1997 split-CP). In this case, movement to the peripheral Spec,YP is assumed to be driven by feature valuing. As far as I have been able to establish, the noun in the N-dem order is neither focused nor topicalized. If so, it is not clear what feature the noun (as part of XP) would value in the peripheral Spec,YP. If movement to this position is only triggered by feature valuing, I conclude that the noun cannot have moved to Spec,YP either.10 As a final option, I propose that the demonstrative is not base-generated in Spec,DP but rather in a lower Specifier position. I assume that determiners are

9.  Chomsky’s work has been interpreted in different ways: while Marcel den Dikken (p.c.) observes in a different context that Chomsky’s system left room for adjunction to DP if brought about by base-generation, Željko Bošković (p.c.) points out to me that Chomsky’s intention was to rule out all adjunction to arguments (for some evidence, see Bošković 2004: 691 fn. 12). 10.  Furthermore, there is some indication that (some of) the modern Scandinavian languages do not have split DPs. For instance, as discussed by Grohmann & Haegeman (2003), Norwegian, unlike West Flemish, does not allow noun phrase-internal left dislocation or possessor-related Quantifier Float. If this also holds for the older varieties, then there would not be a split DP to begin with.



The syntactic distribution of determiners 

merged in an Article Phrase (ArtP). In particular, I suggest that demonstratives are merged in Spec,ArtP (cf. Brugè 1996, 2002; Campbell 1996; Giusti 1997, 2002; Grohmann & Panagiotidis 2004; Panagiotidis 2000; Vangsnes 1999: 119–20). In order to derive the sequence N-dem, I propose that the demonstrative remains in situ and that the head noun, as part of a phrase, moves to Spec,DP:11 (21)

DP XPi N

D′ D

ArtP Spec dem

Art′ Art

... ti ...

If this is correct, then the order N-dem establishes the fact that nouns, as part of a phrase, may move to Spec,DP. Importantly, when an adjective is added, the adjective only follows the demonstrative. Assuming that the adjective is located between the DP and ArtP (i.e., in Spec,AgrP), the order N-dem-adjective can only

11.  There is evidence that the noun in Old Norse raises as a phrase. Lohndal (2007: 292–3) reports that while (ia) and (ib) are possible, (ic) is not mentioned in any of the major descriptive works on Old Norse (Nygaard 1906; Faarlund 2004). Let us assume that (ic) did not exist (unlike Lohndal, who later in his paper assumes that it did exist): (i) a. þau in stóru skip those the big ships ‘those big ships’

(Old Norse)

b. sá ọrninn that eagle-the ‘the eagle’ c.  ??sá hestrinn gamli that horse-the old ‘that old horse’ In agreement with others, I claim that (ia) shows the demonstrative in Spec,DP and the determiner in D. In order to explain the difference between (ib) and (ic), I suggest that the determiner is in Art in (ib) but, with the adjective present, it must be in D in (ic) (for details, see Section 5). If we assume now that the noun moves as part of a phrase (and not by N-raising), then the absence of (ic) is explained as there is no phrasal position between Spec,DP and D. In contrast, in (ib) the phrase containing the noun could be adjoined to ArtP. If so, and crucially assuming that the determiner can be in Art, then we have an argument for phrasal movement of the noun.

 Demonstratives and definite articles as nominal auxiliaries

involve two juxtaposed DPs (as only the noun or the pre-adjectival demonstrative can be in Spec,DP). Consider the schematic representation in (22a). However, if the phrasal demonstrative is reanalyzed as a determiner head, then both the noun (in Spec,DP) and the determiner (in D) can occur in one and the same noun phrase, as illustrated in (22b): (22) a. [noun] [DP dem [D′ D [adjective …]]] b. [DP nouni [D′ det [adjective … ti]]]

Employing movement of both the noun and the determiner, I assume that the simplified structure in (22b) provides the basis for suffixation of the determiner to the preceding head noun. This is the thrust of the proposal in the next section. 3.3  Reanalysis from demonstrative to article Following the traditional literature, I suggest in what follows that the order N-dem-adjective forms the basis for the change from the post-nominal demonstrative to the suffixed definite article. I propose in formal detail that the diachronic reanalysis of the demonstrative to the definite article progressed via several steps. In the course of this discussion, I illustrate the basic derivations for the above data. As discussed in Section 3.2, the Proto-Scandinavian examples motivate the lower position of the demonstrative and movement of part of the noun phrase to Spec,DP. Assuming that referential noun phrases must have overtly licensed DPs (cf. Longobardi 1994), I suggest that Proto-Scandinavian had the option of moving either the demonstrative or the noun (as part of NP) to Spec,DP. The first option is shown in (23b) and the second one in (24b):12 (23) a.

a hitt lant to this land ‘to this land’

b. [DP hitti D [ArtP ti Art [NP lant]]] (24) a.

hali hino stone this ‘this stone’

b. [DP [NP hali]k D [ArtP hino Art tk]] 12.  If NP moves to Spec,DP, we might expect that its complements move along. While I have no evidence for or against this in Proto-Scandinavian, common Modern Icelandic allows its (non-pronominal) complements to vacate their base-position before AgrP moves to Spec,DP (see Vangsnes 2004). If so, then the position of complements to the right does not necessarily indicate N-to-D raising but leaves room for analyzing a preceding noun as movement of a phrase containing the noun.



The syntactic distribution of determiners 

In Common Scandinavian, the determiner is usually obligatory when an adjective occurs.13 Now, considering again the combination of determiner and proper name in (25), the determiner seems to be a semantically vacuous, that is, an expletive element (see also Section 5.2 below). Assuming, then, that it has only a purely syntactic function, we can propose that the determiner has moved to D (rather than Spec,DP).14 If so, the phrasal demonstrative has been reanalyzed as a free-standing article: (25) a.

in heilhi kristr the holy Christ ‘the holy Christ’

b. [DP ini [AgrP heilhi [ArtP ti [NP kristr]]]]

13.  In a few syntactic contexts, the determiner does not occur:  (i)

Guð hialpi salu hans ok Guðs moðiR, helagR God help his soul, and God’s mother, holy



Kristr i himinriki Christ in heaven



‘God help his soul, and God’s mother, holy Christ in heaven.’

(Västergötland 186)

14.  There is independent evidence for the assumption that expletive determiners are in D (and not in Spec,DP). Longobardi (1994: 623) argues that proper names in Italian must undergo N-to-D raising if an expletive determiner as in (ia) is not present. Compare (ib) to (ic): (i) a.

il mio Gianni the my Gianni ‘my Gianni’

(Italian)

b. *mio Gianni my Gianni c.

Gianni mio Gianni my ‘my Gianni’

d.

il Gianni MIO the Gianni my ‘MY Gianni’

The possessive in (id) can only have contrastive reference, marked by capital letters. As discussed by Cardinaletti (1998), the possessive in (id) is in situ, that is, in a low position that the proper name has moved across; the possessives in (ia–b) are in a higher position and (ic) is presumably ambiguous. Most relevant for current purposes, if we assume the expletive determiner to be in D, then the complementary distribution between this determiner and the (raised) proper noun in (ia) vs. (ic) follows straightforwardly. More generally, like with clausal auxiliaries (e.g., copula verbs), there are expletive instances with nominal auxiliaries.

 Demonstratives and definite articles as nominal auxiliaries

In the course of the development from Common Scandinavian to Old Norse, the free article is suffixed to the head noun. As neither hinn N or N hinn sequences are attested in Common Scandinavian, suffixation must, most likely, have resulted from the use of hinn with adjectives, as is traditionally assumed.15 With in a head in D, as suggested for (25), there is now room in Spec,DP for the noun (phrase). If so, then frequent appositives involving proper names were structurally ambiguous between (26b) and (26c): the former illustrates the juxtaposition of two noun phrases but the latter shows just one noun phrase where the head noun, as part of NP, has moved to Spec,DP (eN indicates a null noun). Note that hin is still a freestanding element as it has initial ‘h’: (26) a.

kristr hin helgi Christ the holy ‘Christ, the holy’

b. [DP c.

kristr] [DP hini [AgrP helgi [ArtP ti [NP eN]]]]

[DP [NP kristr]k

hini [AgrP helgi [ArtP ti

tk]]]

I propose that this structural ambiguity paved the way for the actual reanalysis of less-frequent appositives involving common nouns. In particular, once the word stress on and the deictic force of the demonstrative as well as the intonational break between the noun and the demonstrative were lost, the noun inside NP was analyzed to be in Spec,DP and hinn to be in D in one and the same noun phrase. The article could then be suffixed to the noun (“+” indicates suffixation): (27) a.

mirkit mikla monument-the great ‘the great monument’

b. [DP [NP mirki]k +ti [AgrP mikla [ArtP ti tk]]]

15.  However, there are good reasons to believe that Common Scandinavian also had hinn as a demonstrative with unmodified nouns (cf. row two in table 1). The reason I believe that it is not attested has to do with its semantics. Taking Old Norse as a guide where hinn means ‘the other’ or ‘(emphatic) that’ (Zoëga 1910), the use of hinn in inscriptions would presumably be pragmatically odd. Although unattested and usually not assumed to have formed the basis for this reanalysis, the unmodified sequence N-dem could be a second scenario for this change and subsequent suffixation: (i)

a. b.

[DP [NP maðr]i [DP [NP maðr]i

D [ArtP hinn [Art′ Art ti]]] D [ArtP [Art′ inn ti]]]

To the extent that this possibility is correct, it would allow suffixation of inn in the DP-level (after movement of inn and the noun to the DP-level) or in the ArtP-level (after raising of the noun).



The syntactic distribution of determiners 

Finally, although the first suffixed article without an adjective occurs in the 11th century, as shown in (11) above, Old Norse still has free-standing and suffixed forms, here illustrated with the noun phrases for the old man repeated from (12a–c): (28) a. [DP [NP maðr]k+ inni [AgrP gamli [ArtP ti tk]]] b. [DP [NP maðr]k inni [AgrP gamli [ArtP ti tk]]] c. [DP inni [AgrP gamli [ArtP ti [NP maðr]]]]

This variation in Old Norse can be explained by two assumptions: (i) the article is still ambiguous between a free-standing and suffixal element (presumably due to inter-  and/or intra-dialectal variation) and (ii) NP movement to Spec,DP is optional. The first assumption explains the facts in (28a) and (28b) and the second one explains the contrast between (28a–b) and (28c). 3.4  Some immediate consequences The current proposal allows for a straightforward transition into Modern Icelandic (for a discussion of other Scandinavian languages, see Section 5). Discussing some differences between literary and common Modern Icelandic below and in Chapter 3, I assume for now that these two varieties involve different “dialects”. In the former case, (29), optional NP movement is lost: (29) a.

hinn gamli maður the old man ‘the old man’

(literary Modern Icelandic)

b. [DP hinni+D [AgrP gamli [ArtP ti [NP maður]]]]

For the latter case, (30), I suggest that over time, overt movement of the determiner to D is lost (i.e., the relevant EPP feature is gone) and movement of NP is replaced by movement of AgrP to Spec,DP (cf. Julien 2002; Vangsnes 1999, 2004). Suffixation becomes obligatory due to partial N-raising to Art (cf. Taraldsen 1990). Consider these two main steps in the derivation:16 (30) a.

gamli maður-inn old man-the ‘the old man’

b. [DP [AgrP gamli [ArtP c.

(common Modern Icelandic)

-inn [NP maður]]]j D … tj]

[DP [AgrP gamli [ArtP maðurk +inn [NP tk]]]j D … tj]

More generally, considering that the free-standing determiner in literary Modern Icelandic and the suffixal one in common Modern Icelandic derive from the same 16.  In Section 5, I argue that the suffixal determiner actually undergoes Merger with the noun after syntax and, after reconstruction of AgrP in LF, it moves to D covertly.

 Demonstratives and definite articles as nominal auxiliaries

demonstrative, their different synchronic distribution is accounted for by overt vs. covert movement of the determiner to D (cf. footnote 16). As such, the Panchronic Alternation in literary vs. common Modern Icelandic can be explained by adopting a single tree representation and the assumption of displacement of determiners. There are some other consequences of this proposal. It is well-documented that complementizers often evolved from demonstratives. Van Gelderen (2004a:  p. 73) proposes for Germanic that this change consisted of the reanalysis of elements in a Specifier position to elements in a head position of the same phrase, that is, CP. If the analysis above is correct, I can extend van Gelderen’s analysis of the complementizer system to the nominal domain (see also Philippi 1997: 90 for German, Giusti 2001: 167 for Romance, and most recently van Gelderen 2007). In fact, I have presented evidence that, like elements in CP, related “transitive” determiners can be of different structural sizes, as already hinted at in Holmberg (1999: 264), Giusti (1997: 120 fn. 12), and Vincent (1997: 165). In Roehrs (2005a), I make the same claim for pronominal determiners such as us in us linguists.17 I began this section by arguing for the phrasal status of demonstratives. Observing that Proto-Scandinavian exhibits N-dem sequences, I motivated the claim that determiners can be in a lower position. Following van Gelderen (2004a, b, 2007), I proposed that the grammaticalization of the demonstrative to an article proceeded by reanalysis of a phrase to a head. Subsequent suffixation to the head noun occurred either in the DP level and/or in ArtP (cf. footnote 15). Relevant for the main point of this proposal, movement of the article to the DP accounted for the Old Norse pattern and literary Modern Icelandic and loss of this (overt) movement accounted for the change into common Modern Icelandic. This difference in determiner movement provides a straightforward explanation of the Panchronic Alternation with regard to the historically related determiners in literary vs. common Modern Icelandic. I now consider some more general consequences of the analysis. 4.  Unifying the different determiner systems This section makes the claim of a low ArtP more general. Furthermore, it connects the proposal of the rise of the suffixal article offered above to the distribution of determiners in other languages. As such, this discussion represents a first step 17.  One immediate extension for English is that the (unstressed) specific indefinite “demonstrative” this can be analyzed as a head (cf. Diessel 1999:  109, 138; Vangsnes 2001: 258), having derived from the phrasal demonstrative this (similarly for certain uses of that, see Diessel 1999: 106–7). Another extension is the reanalysis of numerals as indefinite determiners (for some general discussion of German, see work of Oubouzar’s).



The syntactic distribution of determiners 

toward unifying the apparently different determiner systems of these (and other) languages. It is not my intention here to be exhaustive or to give complete and detailed analyses of all languages or phenomena. For present purposes, it suffices to provide more evidence for the lower position of ArtP. This discussion will set the stage for a more comprehensive analysis of the Scandinavian DP in Section 5. 4.1  Demonstratives in different positions The claim that demonstratives can be in different positions is not novel (e.g., Bernstein 1993: 149; Dimitrova-Vulchanova & Giusti 1998: 338; Duffield 1996: 331; Giusti 1997:  109–10; Lyons 1999:  119–121; Plank 2003:  344–345). The most comprehensive discussion of these cases that I am aware of is Brugè (1996, 2002). Brugè (1996, 2002) sets out three types of languages: (i) the demonstrative can be both pre- and post-adjectival, (ii) the demonstrative can only be post-adjectival, and (iii) the demonstrative is always pre-adjectival. While the first two types of languages provide straightforward evidence for a lower position of determiners, there is evidence from “stranded” deictic reinforcers that demonstratives in the last type of language also originate in a lower position. I will call the latter constructions “split” demonstratives. In Section 5, I make this discussion more general by developing the notion of a “split” determiner, which includes demonstratives, definite, and indefinite articles. We start by briefly reviewing Brugè’s three types of languages, using some of her examples. In the first main group of languages, the demonstrative can be in pre-adjectival as well as post-adjectival position. Postponing the discussion of the former case (cf. (34) below), the latter is illustrated here with Spanish and Greek ((31b) is taken from Panagiotidis 2000: 721):18 (31) a.

El libro viejo este suyo de sintaxis no me convence. the book old this his/her on syntax not me convince ‘This old book of her on syntax does not convince me.’

(Spanish)

18.  Similar facts hold for Catalan. Furthermore, there are also languages where the demonstrative can be in a lower position but there is no article on the left. For instance, the “marked” option in (ib) in Modern Icelandic seems to be a case in point (Vangsnes 1999: 148 fn. 34): (i) a.

þessi maður this man ‘this man’

b.

maður þessi man this ‘this man’

(Modern Icelandic)

For a list of languages that shows that this is more general, see Rijkhoff (2002: 179–180).

 Demonstratives and definite articles as nominal auxiliaries

b. i nei afti katiki tis polis the new these inhabitants the-gen city-gen ‘these new inhabitants of the city’

(Greek)

Assuming nominal-raising in Spanish (cf. Chapter 1), the position of the demonstrative is below the adjective and above all the complements of the head noun. This is the phrasal level I have identified as ArtP. Besides the base-position and Spec,DP, demonstratives can also occur in intermediate positions. For instance, to the more restricted possibilities in Spanish (Brugè 1996: 13 fn. 15) and Serbo-Croatian (Brugè 1996: 44; Progovac 1998: 174 fn. 6; Trenkic 2004: 1312), I add the apparently more permissive (Ki)Swahili (Carstens 1991: 105–116):19 (32) (hili) shati (hili) langu (hili) zuri (hili) this shirt    this my    this good    this ‘this my good shirt’

(Kiswahili)

In Romanian, the demonstrative can appear in Spec,DP and in an intermediate position, but not in the base-position (for details, see Brugè 1996: 46; Giusti 1997). In the second principal type set out by Brugè, the demonstrative can only appear in its low base-position. Consider examples from Irish and Welsh ((33b) is taken from Roberts 2005: 92): (33) a.

an leabhar nua seo faoi teangedaiocht the book new this on linguistics ‘this new book on linguistics’

b. y pum llyfr newydd hyn gan John the five book new this by John ‘these five new books by John’

(Irish)

(Welsh)

As above, I assume that the lower demonstrative is in ArtP. In the third main type of language, the demonstrative always appears on the left. However, assuming that demonstratives and their deictic reinforcers are generated together (Brugè 1996, 2002; Roehrs & Putnam 2008; cf. also Bernstein 1997, 2001a), evidence for a low ArtP in these languages comes from the “stranded” reinforcer. In other words, demonstratives originate low in the structure and then move to Spec,DP. To set the stage for this discussion, we consider the other case of the first type of language, the discussion of which was postponed earlier on.

19.  Carstens argues for N-to-D raising (for another language with a varying lower position of the demonstrative, see Rijkhoff 2002: 326 on West Greenlandic).



The syntactic distribution of determiners 

4.2  “Split” demonstratives Brugè (1996: 23–5) points out that Spanish, a language of the first type, has the option of leaving the demonstrative adjacent to the deictic reinforcer or moving it away: (34) a.

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

(Spanish)

b. Este libro de aquí está mal hecho. this book of here is badly made ‘This here book is poorly made.’

As nothing can intervene between the relevant elements in (34a), she concludes that both parts are generated together in a complex Specifier (p. 27). Considering a more complex noun phrase, we observe that, as expected, this Specifier position is below the adjective and above all the complements of the head noun. (35a) is derived as in (35b): (35) a.

Este libro viejo de aquí de sintaxis está mal hecho. this book old of here of syntax is badly made ‘This here old book on syntax is poorly made.’

b. [DP estei librok [AgrP viejo [ArtP ti de aquí [NP tk de sintaxis]]]]

Assuming with Brugè and Bernstein that the demonstrative and the reinforcer originate together and may be split up, the “stranded” reinforcer indicates the base position of the demonstrative. If the notion of a “split” demonstrative is on the right track, then we have a means to probe into languages that apparently do not have direct evidence for determiners in lower positions. I turn to Brugè’s third type in more detail. Taking German for illustration, the non-proximal demonstrative das can be reinforced by da ‘there’. Recall from Chapter 1 that head nouns in Germanic move out of the theta domain. Importantly, the reinforcer appears to the right of the raised head noun. If das and da originate together and the demonstrative has moved from ArtP, then its reinforcer da overtly shows its originating position (for a more detailed discussion, see Section 5.3 below). I provide my own example for German and an example for Italian from Cardinaletti (1998: 18), where the representations in (b) give the simplified derivations for the examples in (a): (36) a.

das schöne Bild da von Maria that nice picture there of Mary ‘that there nice picture of Mary’

b. [DP dasi schöne [ArtP ti Bildk da [NP tk von Maria]]]

(German)

 Demonstratives and definite articles as nominal auxiliaries

(37) a.

Questo libro qui suo di sintassi non mi convince. this book here his/her of syntax not me convinces ‘this here book of her on syntax does not convince me.’

(Italian)

b. [DP questoi [ librok [ArtP ti qui [NP suo tk di sintassi]]]]

Besides these two languages, Brugè (1996, 2002) also discusses French and Albanian (in Chapter 4, I provide more evidence for a low ArtP in German from the discussion of the strong/weak alternation of adjectives). To sum up this section, I provided evidence from other languages that demonstratives originate in a low ArtP. The evidence came from the overt distribution of the demonstratives themselves or from their “stranded” reinforcers. Proposing that demonstratives can be split up and bearing in mind that they are determiners more generally, one might expect this also to be true for articles. 5.  “Split” articles in the Scandinavian languages The Scandinavian noun phrase has been widely discussed (among many others, Börjars 1998; Börjars & Donohue 2000; Delsing 1988, 1993a, 1993b; Hankamer & Mikkelsen 2002, 2005; Holmberg 1993; Hughes 2007; Julien 2002, 2005; Kester 1993; Santelmann 1993; Sigurðsson 1993, 2006; Vangsnes 1996, 1999, 2004). Concentrating here on definite noun phrases, I proceed as follows: first, I present the basic, wellknown data to lay the foundation for the discussion. Then, I discuss languages with the Double Definiteness effect, illustrating the different semantic contributions of free-standing and suffixal determiners. Maintaining that a syntactic account should take these different semantic contributions into consideration, I propose that, similar to demonstratives, certain languages can split up their articles and move part(s) of them to the DP while leaving the other part(s) in situ. Finally, I briefly compare my analysis with Julien (2002 [2005]), which, to my knowledge, presents the most comprehensive and worked-out proposal for the Scandinavian DPs. 5.1  The basic data The distribution of the definite article in the Scandinavian languages has received a lot of attention. To illustrate the basic patterns, consider the well-known fact that all Scandinavian languages have a homogeneous pattern with definite unmodified DPs, (38). In these cases, the article is suffixed to the noun:20 20.  Exceptions to this are Western Jutlandic (Julien 2002) and perhaps the dialect of Eastern Nyland (footnote 40). For some discussion of singular countable noun phrases without a determiner, see Julien (2002: 272 footnote 5).



(38) a.

The syntactic distribution of determiners 

maðr-inn man-the ‘the man’

b. maður-inn man-the ‘the man’ c.

(Old Norse)

(Modern Icelandic)

mand-en man-the ‘the man’

(Danish)

d. mann-en man-the ‘the man’

(Norwegian)

e.

mann-en man-the ‘the man’

(Swedish)

However, there is considerable syntactic variation with modified DPs. These diverse patterns are illustrated in (39). Concentrating on DPs modified by adjectives (for a discussion of relative clauses, see Chapter 3), note that, while Old Norse and common Modern Icelandic have a suffixed article, literary Icelandic and Danish have a free-standing one. Norwegian and Swedish illustrate – what is traditionally called – the “Double Definiteness effect”: (39) a.

maðr-inn gamli man-the old ‘the old man’

b. gamli maður-inn old man-the ‘the old man’ b′. hinn gamli maður the old man ‘the old man’ c.

den gamle mand the old man ‘the old man’

d. den gamle mann-en the old man-the ‘the old man’ e.

den gamle mann-en the old man-the ‘the old man’

(Old Norse)

(common Modern Icelandic)

(literary Modern Icelandic)

(Danish)

(Norwegian)

(Swedish)

 Demonstratives and definite articles as nominal auxiliaries

Two points should be made here: on the one hand, modified DPs can have the same referential properties as their unmodified, intra-linguistic counterparts (that is, a descriptive adjective does not contribute to the referential status of the DP). On the other hand and simplifying somewhat here (but see below), all the modified noun phrases in (39) can basically have the same interpretation(s) across languages. Again, I conclude that the adjective does not make a semantic contribution in the relevant sense. Following others, I propose that the diversity in (39) is a syntactic phenomenon brought about by the presence of the adjective (the same basically holds for numerals). However, I will not simply provide a surface-oriented analysis of this variation but also take subtle differences in interpretation into consideration. These semantic distinctions have some overt reflexes in the different distribution of the determiners in the languages with the Double Definiteness effect. The following proposal involves two main components: (i) articles are complex elements that can be split into different parts (Section 5.2), and (ii) phrases hosting modifiers induce an independent agreement domain for long-distance agreement but not movement (Section 5.3). As these assumptions and their technical instantiations are new (e.g., vis-à-vis Julien 2002 [2005]), I will devote some space to motivating them in detail. 5.2  The semantic and syntactic relations of the two determiners In what follows I work towards the notion of “split” articles. I argue that in the languages with the Double Definiteness effect, the two overt determiners have different semantic functions. For instance, while the suffixal article seems to bring about specificity interpretations (i.e., it “picks” out a particular item that is known to the speaker, but not to the hearer), the pre-nominal determiner is responsible for uniqueness (cf. Julien’s 2005 inclusiveness and Anderssen’s 2005 person feature) and a deictic reading. As these different functions can also be fulfilled by unmodified DPs (which only have a suffixed determiner), I suggest that all determiners are underlyingly more complex. This inner build-up only becomes evident in modified DPs, where a part of the determiner has split off and moved to the DP. These different parts are then spelled out individually. On a par with “split” demonstratives, I refer to these patterns as “split” articles. 5.2.1  Semantic differences overtly reflected In this subsection, I provide more data from the languages with the Double Definiteness effect, most of which are taken from discussions by Delsing (1993b)



The syntactic distribution of determiners 

and Julien (2002 [2005]).21 After the presentation of the data, I summarize the discussion and draw some conclusions relevant to the main proposal of this work. Svenonius (1993a: 208 footnote 12) observes that some vocatives do not have a pre-nominal determiner but only a suffixal one (cf. Kester 1996b: 146; Delsing 1993b: 39; also Kari Gade, p.c.): (40) dumme idiot-en stupid idiot-the ‘stupid idiot’

(Norwegian)

Second, Delsing (1993b) discusses “definite deictic noun phrases” (p. 123) whose referents are “well-known in the speech situation, by [their] uniqueness in the world or in a smaller speech community” (p. 118): (41) Ta (den) nya bil-en take     the new car-the ‘Take the new car.’

(Swedish)

He observes that the pre-nominal article is normally used when the noun phrase emphasizes contrast. Furthermore, going back to his earlier work (Delsing 1988), he notes that there are a number of adjectives (call them “adjectival determiners”, Börjars 1998:  p. 206) that allow an optional determiner if the noun phrase has deictic reference (p. 119): (42) (det) sista par-et the last pair-the ‘the last pair’

(Swedish)

As a fourth similarity, noun phrases with an individual, specific reading obligatorily take both determiners (Kari Gade, p.c.): (43) Den gamle mammut-*(en) er syk. the old mammoth-the is sick ‘The old mammoth is sick.’

(Norwegian)

In contrast, Julien (2002: 284) observes that in generic noun phrases with an adjective, the suffixal article is optional in Norwegian (cf. Svenonius 1993a: 204 fn. 9). 21.  Note that there are some slight differences between these languages, which I will abstract away from. Among others, for the discussion of variation with noun phrases containing nationality adjectives or proper names, see Delsing (1993b:  118); for other variation, see the next footnote.

 Demonstratives and definite articles as nominal auxiliaries

Apparently, the same holds for kind-referring expressions (data due to Kari Gade, p.c.). Compare (44a) to (44b):22 (44) a.

Den kvite mann-(en) har alltid undertrykt the white man-the has always oppressed



andre kulturar. other cultures



‘The white man has always oppressed other cultures.’

(Norwegian)

b. Den forhistoriske mammut-(en) er utdødd the prehistoric mammoth-the is extinct ‘The prehistoric mammoth is extinct.’

So far, we have seen that vocatives, “definite deictic noun phrases”, noun phrases with “adjectival determiners” and the ones with an individual, specific reading take a suffixal determiner. In that respect, some variation holds in the noun phrases with a generic reading. The presence of the pre-nominal determiner varies from type to type (for relative clauses, see Chapter 3). Let us summarize these complex sets of data in table 3 (pre-nominal determiners in capital letters are only pronounced if the noun phrase has contrastive reference): Table 3.  Summary of the Different Pattern in Languages with the Double Definiteness Effect. Semantics

Swedish, Faroese, Norwegian

vocative deictic

dumme idiot-en (DEN) nya bil-en (det) sista par-et den gamle mammut-en den förhistoriske mammut-%en

familiar and unique “adjectival determiners” individual reading generic reading (cf. fn. 22)

Let us turn to the identification of the different semantic contributions of the pre- and post-nominal determiners in more detail. Besides making the traditional referential distinction between deictic and anaphoric use of determiners (Schroeder 2006), I also discuss specificity, genericity, and vacuity. To be precise, I show that there is a division of labor such that, whereas the pre-nominal determiner brings about deictic, uniqueness, or generic reference, the post-nominal one seems to be specific or expletive in interpretation 22.  This type of data shows the greatest variation among speakers (see also Julien 2005:  p. 35), marked as % in table 3 above. I will propose later that the suffixal determiner in these cases is semantically vacuous. If so, this variation is not surprising (compare in this respect German der Peter ‘(the) Peter’, where the expletive article is obligatory in southern dialects but optional in northern dialects). More generally, despite this apparent speaker variation, perhaps a reflex of a change in progress, I proceed on this empirical basis as it is clear that some elements do make semantic contributions.



The syntactic distribution of determiners 

(below, I will arrange these semantic components in a tree-like fashion). It is worth pointing out that the present discussion differs from those in Anderssen (2005), Julien (2005), and Kester (1996b), who only identify the semantic components deictic/uniqueness and specificity. I start at the top of the table, making the reviewed data relevant for the proposal below. Ad row 1:  Assuming that vocatives do not necessarily project a DP (cf. Longobardi 1994), we easily account for the absence of the pre-nominal determiner here. If so, we also conclude that the suffixal determiner is not sufficient to make a noun phrase argumental. As a consequence, the suffixal determiner of unmodified DPs cannot necessarily be taken to be in D. I propose below that the suffixal article is always in Art. Ad rows 2 and 3: Recall that, under certain conditions, the determiner can be left out if the noun phrase is used deictically. With contrastive emphasis, the determiner reappears. Furthermore, Delsing (1993b:  120) notes that the determiner cannot be left out when its function is anaphoric. Defining deictic in a broad sense, I will treat both uses (situational, linguistic) basically the same unless indicated otherwise. Considering that all these DPs can be arguments, I propose below that, due to feature valuing on D, the pre-nominal determiner is present in all cases but can be elided in situational deixis under Recoverability of Deletion. Ad row 4: DPs with an individual reading have a free-standing and a suffixal article. I suggest that the free-standing article denotes uniqueness and the suffixal article has a specificity interpretation (for an empirical argument, see below). This is consistent with the fact that noun phrases used as vocatives or deictically have a suffixal determiner (i.e., they are specific). Ad row 5:  In contrast, the distribution of generic DPs with adjectives is somewhat different. Languages of the Double Definiteness effect vary in the obligatoriness of the suffixal determiner (footnote 22). Interpreting generics as denoting kinds (cf. Longobardi 2001a), generics denote something similar, but not identical, to uniqueness (intuitively, less semantics is involved here in that the determiner does not “pick” out a unique subset but the entire relevant set). In order to account for the variability of the presence of the suffixal determiner, I suggest that this element is semantically vacuous, that is, it is an expletive. To sum up the discussion thus far, I suggested that there is a division of labor between the pre-nominal determiner, which is unique/deictic or generic, and the suffixal determiner, which is specific or expletive.23 There is more evidence that these determiners differ in their semantic contribution.

23.  If this discussion is on the right track, then, as already noted by many others (e.g., Alexiadou 2003), the Double Definiteness effect is not simply a reflex of agreement in definiteness, as, for instance, Embick & Noyer (2001: 583) and Faarlund (2007) claim.

 Demonstratives and definite articles as nominal auxiliaries

For present purposes, let us follow Enç (1991) in that definiteness is comprised of two parts: uniqueness and specificity. As already discussed by Julien (2002: 283–5), this distinction can be seen in some languages of the Double Definiteness effect such that D provides uniqueness and Art specificity. Evidence for this claim comes from different morpho-syntactic realizations of the noun phrase in certain co-ordinations and when embedded under different types of predicates. Starting with co-ordination, Julien notes that, despite the presence of the suffixal determiner, there is a difference in interpretation in (45): if a pre-nominal determiner is present in both conjuncts, (45a), the professor and the father cannot be the same person. In other words, the conjoined noun phrases refer to two different people. In contrast, if there is only one pre-nominal determiner, (45b), both persons are one and the same ((45a–b) are adapted from Swedish provided by Julien 2002: 283): (45) a.

den unge professor-en og den kjærlige far-en the young professor-the and the loving father-the ‘the you professor and the loving father’

(Norwegian)

b. den unge professor-en og kjærlige far-en the young professor-the and loving father-the ‘the young professor and loving father’

Following Julien, I interpret these data such that the pre-nominal determiner brings about uniqueness (cf. also Longobardi 1994: 620–1). The combination of different noun phrases and predicates bring out different readings. While (46a) has a generic interpretation, (46b) shows individual reference. In particular, with the pre-nominal determiner present in all cases, the suffixal determiner is given as optional in (46a) in Julien (2002: 284) but is dispreferred by my consultant Kari Gade. Crucially, both speakers agree that the suffixal determiner in (46b) must be present (cf. Braunmüller 1999: 178): (46) a.

Den hvite mann-(√/??en) har alltid undertrykt the white man-the has always oppressed



andre kulturer. other cultures



‘The white man has always oppressed other cultures.’

(Norwegian)

b. Den hvite mann-*(en) spiste en is. the white man ate an ice-cream ‘The white man ate an ice-cream.’

In keeping with the discussion above, the suffixal determiner in (46b) seems to have semantic import with regard to specificity (cf. also Kester 1996b:  146–8; Anderssen 2005: 119–35). To conclude, noun phrases in varying co-ordinations



The syntactic distribution of determiners 

and under different predicates provide strong evidence that the two determiners have differing semantic import. Below I suggest that the cross-linguistic, morphosyntactic differences between the DPs with identical interpretation follows from a different syntactic split of the determiner. Before we proceed, there is an important alternative interpretation of these data that needs to be addressed and shown not to be correct. Rather than the lexical items, one could propose that the abstract positions, that is, Art and D, are responsible for the semantic contribution. Recall from Chapter 1 that the adjective is parasitic on ArtP thus indicating the very presence of ArtP. Now, if we were to associate the different semantic contributions with the different syntactic positions (D and Art), then we would expect all languages to have the same interpretations, such that all modified DPs were always unique and specific in reference, which is not the case. I conclude that the different semantic contributions are a matter of the actual overt lexical items. In more detail, starting with the co-ordination data in (45), it is not immediately clear what parts of the noun phrase are conjoined. In other words, it is not clear if there is an empty D in the second conjunct in (45b) or if there is one (overt) D and conjunction is at a lower level. However, we saw above that argumental DPs can occur without a determiner only under very specific conditions. I take this restrictiveness on licensing null D as an indication that there is only one DPlevel in (45b), that is, that the noun phrase is conjoined at a lower level. Assuming across-the-board movement of the determiner from Art to D, we account for the fact that both noun phrases must refer to the same person. In other words, it is the lexical item of the determiner itself that makes the relevant contribution. A similar point can be made for suffixal determiners. With an adjective present in the DPs in (46a–b), Art must be present in both noun phrases. The difference between them is that the Art position in (46a) is (preferably) not filled and the one in (46b) must be. Since the presence of lexical material causes the different grammaticality judgments, I conclude that it is the lexical item itself (rather than the abstract position Art) along with the different predicates that supports the different readings. In this subsection, we have seen that in Scandinavian, unmodified DPs have a homogenous pattern and modified ones are very diverse. Furthermore, I have illustrated that the pre- and post-nominal articles make different semantic contributions, a fact not observable in unmodified DPs. In addition, the languages with the Double Definiteness effect may have slightly different morpho-syntactic realizations depending on the semantic interpretation of the noun phrase. In contrast, many languages have only one determiner in all contexts which fulfills all relevant semantic functions. In other words, the determiner, at first glance a single element, may actually have uniqueness, deictic, generic, specificity, and expletive

 Demonstratives and definite articles as nominal auxiliaries

functions. The question arises, then, in what syntactic relation the semantically different parts of the determiner discussed above stand to one another. In what follows, I will develop the notion of “split” articles in more detail. 5.2.2  The syntactic relation To derive the correlation between the syntactic distribution of the determiners and its corresponding interpretation(s), I propose that determiners are semantically complex elements that consist of “optional” semantic components. Depending on the case at hand, I assume that different semantic components of the determiner can be merged as part of this complex. Each individual component has its own features. Depending on the language, these bundles of features can be spelled out in different ways. With these general statements in mind, I propose that the Double Definiteness effect consists of one complex determiner which has at least two individual components. Due to the presence of pre-nominal modifiers, the determiner has to move to the DP (see Section 5.3). However, rather than moving all the components, Norwegian, Swedish and Faroese move only part of them, “stranding” the rest in situ. In other words, on a par with split demonstratives, I suggest that the Double Definiteness effect involves split articles. We arrive then at the following picture: while Swedish, Faroese, and Norwegian may split their determiners, languages such as Danish and Icelandic do not. I propose that Danish moves all parts of its complex determiner overtly while Icelandic raises all parts covertly. Besides split determiners, there is other evidence that determiners may have different semantic components. For instance, some languages have different overt determiners depending on the semantics (e.g., for Northern Frisian, see Ebert 1971: 71–113; for Western Jutlandic, see Delsing 1993b: 121; for Northern Scandinavian dialects and Icelandic, see Delsing 1993b:  54–5, 1996:  p. 33; for Northern Greek, see Section 5.2.3; more generally, see Schroeder 2006). I interpret these different overt manifestations of the determiner such that different semantic components are present.24 More generally then, different semantic components become apparent when determiners are split, when they are differently spelled out, or a combination of the two, as in Faroese (e.g., tann lítla bók-in ‘the small book’). I assume that these differences are due to different lexical specifications across the languages. Let me try to make this part of the proposal more formal.

24.  If the article were just the phonological realization of morpho-syntactic, but not semantic, features, then different realizations would be unexpected (see also the discussion of the semantically different articles in Icelandic in Chapter 3, Section 5.3).



The syntactic distribution of determiners 

It is not entirely clear how many or what kind of semantic components are relevant for determiners. Besides uniqueness, deixis, genericity, specificity, and vacuity discussed above, there are presumably other semantic components. If so, it is difficult to be precise about the inner structure of the complex article. While this poses a potential problem, there seems to be evidence, however, that these semantic components cannot be freely built up but rather have “internal structure”. Longobardi (1994: esp. 655–9) makes the distinction between substantive and expletive determiners. The latter category divides into articles that are generic or “preproprial” (i.e., used with proper names). Unsurprisingly under current assumptions, these three types of determiners can be differently realized crosslinguistically. Abstracting away from gender, Catalan uses en with proper names but el in generic and definite specific contexts. In contrast, Northern Frisian employs the so-called A-article with proper names and in generic contexts but the D-article in other contexts. In other words, while in Catalan the generic and substantive uses are phonologically neutralized, in Northern Frisian the two expletive uses are phonologically the same. What is interesting to note is that the third type of combination, that is, the phonological neutralization of the substantive and the preproprial article vs. the generic one, does not seem to exist. To the extent that I am aware, this is a more general gap and should be explained. This explanation will provide the basis for the (tentative) discussion of the inner structure of the definite article. Longobardi (1994: 656) states that the generic article is an intermediate one situated between the two other “extremes”. With regard to the semantics, let me interpret this such that the generic article has more semantic components than the preproprial one but fewer than the substantive determiner.25 Remaining vague about the actual semantic components indicated here by Greek letters, assume now that complex determiners are semantically composed in the following way: (47)

substantive generic preproprial α

β

“expletive”

γ

25.  Citing work in progress by Dayal, Longobardi (2005: 32 footnote 31) seems to suggest that definite articles in generic contexts may indeed have some semantics. This is consistent with my interpretation in the main text.

 Demonstratives and definite articles as nominal auxiliaries

Under these assumptions, the preproprial determiner is semantically vacuous (γ = Ø), the generic one is the sum of the semantics of β and γ, and the substantive article is the sum of all three (α + β + γ). If this is correct, then Catalan phonologically neutralizes the semantic components β and higher but Northern Frisian β and lower. The absence of the phonological contrast between substantive and preproprial vs. generic is explained as the semantic components of α and γ do not form a semantic continuous unit to the exclusion of β.26 With this in mind, we return to the definite article in Scandinavian and make some tentative remarks about its internal structure. We saw above that, if split up, the determiner part for uniqueness, deixis, or genericity is in D and the one for specificity or vacuity is in Art. Uniqueness, deixis, and specificity are substantive semantic components. Fleshing out (47), assume that the substantive part of the determiner divides into a uniqueness/deixis component and a specificity one. The structure might look as follows, where I leave α in place and simply add δ: uniqueness/deixis

(48)

specificity generic preproprial δ

α

β

γ

Under these assumptions, we might hypothesize that the Double Definiteness effect is brought about by moving either δ to the DP in noun phrases with uniqueness/deictic interpretations or β in noun phrases with generic readings. In the first case, the semantic components α, β, and γ would be left in Art; in the second case, the component γ.27 This is illustrated in table 4, where the main

26.  This discussion makes the prediction of the existence of an overtly tripartite system. So far, I have not come across a clear case of this yet. 27.  Alternatively, and on an even more tentative note, if determiners are nominal auxiliaries and if clausal auxiliaries are represented in recurring verb phrases, then the structure in (48) might be reinterpreted as recurring ArtPs. This would further extend the parallelisms



The syntactic distribution of determiners 

semantic components of the determiner have been identified by the relevant subscript used in (48). Table 4.  Main Semantic Components in the Double Definiteness Effect. Semantics

Swedish, Faroese, Norwegian

vocative deictic

dumme idiot-enα (DENδ) nya bil-enα (detδ) sista par-etα denδ gamle mammut-enα denβ förhistoriske mammut-%enγ

familiar and unique “adjectival determiners” individual reading generic reading

If it is correct that there are semantically vacuous suffixal elements, then it is less surprising that the suffixal determiner in generic noun phrases modified by adjectives (the last row in table 4) may have such a varying distribution. Needless to say, this analysis raises some interesting issues. For instance, although the Double Definiteness effect was originally due to two different historical sources, a pre- and a post-nominal determiner (see Roehrs & Sapp 2006), the synchronic assumption of movement has some desirable consequences. As already pointed out above, the homogenous unmodified DP can basically be used in all the semantic contexts in which the different modified DP patterns occur (e.g., anaphoric use, specificity). In contrast to modified DPs, the unmodified DP has only a (suffixal) determiner. Its obligatoriness follows from the fact that the determiner is not split and the relevant semantic components that are present must be made visible. As for determiners in modified DPs, they are split by movement due to the presence of the adjectives. As a consequence, the semantic components are separated. As the free-standing and the suffixal determiner are morphologically independent of one another, either semantic component and its overt realization can be absent under certain conditions. Hence, the assumption that determiners are complex and that individual parts can move on their own explains the same interpretations of the homogenous unmodified DPs and the heterogeneous modified noun phrases, on the one hand, and the obligatoriness vs. “optionality” of the relevant overt determiner (parts), on the other. Furthermore, if it is correct that the suffixal part indicates (at least) specificity, then we would expect it to occur in all such contexts. However, the suffixal article only occurs with certain pre-nominal determiners. To be precise, it is not possible

between the nominal and the clausal domain and would allow raising to move words (and not subparts of them).

 Demonstratives and definite articles as nominal auxiliaries

in specific indefinite contexts, (49a), with possessors, (49b), or in pronominal DPs, (49c) ((49b) is taken from Börjars 1998: 15):28 (49) a.

en mus(*-en) a mouse(-spec) ‘a mouse’

(Swedish)

b. min sko(*-en) my shoe(-spec) ‘my shoe’ c.

vi hungriga studenter(*-na) we hungry(weak) students(-spec) ‘we hungry students’

Assuming a movement relation between the two parts, we can suppose that the absence of these patterns is a lexical property of the relevant pre-nominal determiners such that they cannot be split at all.29 Note again that, if double-definiteness were simply agreement, then the lack of it in (49) would be hard to explain. Finally, in Chapter 3, I turn to a semantic argument that the positions Art and D are related by movement of the determiner. Discussing the “Partee-Chomsky debate” (cf. Heim & Kratzer 1998: 83), I suggest that the different adjective interpretations with regard to non-/restrictiveness are derived by different scopes of the determiner with regard to the adjective. This discussion is especially interesting from the Scandinavian point of view, since adjectives can basically have both interpretations, independent of the syntactic distribution of the determiners in the modified DP. Before we turn to the second component of the proposal (induced agreement domains), I show that the claim that determiners may have different functions in different positions is more general.

28.  Another possible case might be the following (although a reviewer states that the absence of the suffix is just a written convention): (i) denna mus(*-en) (standard Swedish) this mouse(-spec) ‘this mouse’ Furthermore, the same reviewer points out that some dialects may have a suffix with possessors (see, e.g., Vangsnes 1999: p. 151). However, importantly for us, not all dialects have a suffix. Finally, as pointed out by Josefsson (1999: 755 fn. 21), vi studenterna is grammatical when both components receive stress (but see also Julien 2005: 129). I interpret this to mean that the elements in this case are combined by adjunction (rather than complementation as in the main text). 29.  In the terms of Distributed Morphology (Halle & Marantz 1993), we could state that the relevant terminal heads undergo obligatory Fusion.



The syntactic distribution of determiners 

5.2.3  “Split” determiners There are other languages where determiner-like elements have different semantic functions in different syntactic positions. Citing earlier work of his and Manolessou’s, Panagiotidis (2000:  723) states that demonstratives in the left periphery have a deictic function and are thus felicitous in the situation described in (50a). In contrast, demonstratives in a lower position, (50b–c), are anaphoric with regard to the linguistic discourse. Consequently, they are infelicitous with situational deixis: (50) (at the butcher’s, pointing to a pork joint) a. Thelo afto to (apaho) butaki. I-want this the     lean joint ‘I want this lean joint’

(Greek)

b.  #Thelo to apaho afto butaki I-want the lean this joint ‘I want this lean joint’ c.  #Thelo to butakik afto tk. I-want the joint this ‘I want this joint’

Note that he argues that the demonstratives in (50b–c) are in the same position (pp. 728–9). Specifically, Panagiotidis argues that the noun in (50c) moves to Num when an adjective is not present in Spec,NumP. So far, we have established that the demonstratives may have different semantic contributions in different positions. Panagiotidis also provides morpho-syntactic evidence from Northern Greek (and South and West Catalan) that the upper determiner in D is not an expletive element. This can be shown with the different distributions of substantial u and expletive i: (51) a.

u skilus aftos (cf. u skilus) the dog this the dog ‘this dog’ ‘the dog’

(Northern Greek)

b. *i skilus aftos (cf. i Yans) the dog this the Yans ‘Yans’

To be clear, then, the datum in (51a) shows two determiners where both have a semantic contribution. Under current assumptions, I claim that this is another type of “split” demonstrative (see Grohmann & Panagiotidis 2004 for a different perspective). A similar point can be made for indefinite noun phrases. With the exception of Icelandic, all Scandinavian languages have an indefinite article. Importantly, they may have some intriguing distributions. Consider some Scandinavian dialects that allow an apparent indefinite determiner to follow every

 Demonstratives and definite articles as nominal auxiliaries

adjective ((52a) is from Delsing 1993b: 143, (52b) is from Julien 2002: 269; see also Anderssen 2005: 65, footnote 20):30 (52) a.

en stor en ful en kar a big an ugly a man ‘a big ugly man’

b. ?eit stort eit stygt eit hus a big an ugly a house ‘a big ugly house’

(Northern Swedish)

(Norwegian)

Delsing (1993b: 143) notes that the post-adjectival article has some special properties. It can co-occur with a non-countable noun, (53a), and has a plural form, (53b):31 (53) a.

Vi ha fått fint e ver. we have got fine a weather ‘We have fine weather.’

b. Dänna var he stor a husa. over-there were there big a-pl houses ‘Over there, there were big houses.’

He observes that the lower determiners have the same properties as non-argumental indefinite determiners in Colloquial Swedish and Norwegian in general (see his Section 2.1). Assuming an analysis of split determiners also for the indefinite article, I suggest that these determiners are split into an argumental and several non-argumental parts. It is presumably these different semantic contributions that explain the possibility that there are several “copies” of the determiner. Finally, these data also illustrate that, due to movement, (part of) the determiner can be “stranded” in intermediate positions. In other words, determiner movement proceeds successive-cyclically (see also Chapters 3 and 4).32

30.  For the discussion of sequences like a such a, see Wood (2002: p. 109); for the discussion of indefinite determiner doubling constructions in certain varieties in German, see Kallulli & Rothmayr (2008). 31.  A reviewer points out that the examples in (53) are not allowed in the dialects that exhibit (52). While more empirical research is certainly needed here, let us take this remark at face value. If so, one could claim that dialects differ in the conditions that regulate the spell-out of the lower copies of the determiner (perhaps, this is a difference due to the singular countable vs. mass/plural opposition in the data above). 32.  There is more cross-linguistic evidence for different positions of indefinite determiners. For instance, although the status of the lower indefinite article in Turkish is not entirely clear (but see Schroeder 2006: p. 592), both the numeral and the lower element occur in different positions (assuming that two (non-coordinated) adjectives do not form a constituent and thus cannot move into Spec,DP). The Turkish data are due Serkan Sener:



The syntactic distribution of determiners 

In the last two subsections, I proposed that articles are semantically complex elements with inner structure. Importantly, while the details may not always be entirely straightforward, there is clear evidence that determiners are semantically complex elements. Certain parts can be split off and moved to the DP. Generalizing this claim to all determiners, different components can be spelled out in different ways in different languages. We now turn to the second main ingredient of the proposal. 5.3  Induced agreement domains What is interesting about the Scandinavian DPs is that, one the one hand, the condition allowing the split is a purely syntactic one (the presence of a modifier) and, on the other hand, this split may have varying morpho-syntactic reflexes corresponding to different interpretations. In order to capture both types of effects, I propose that the separation of determiners is due to the presence of a phrase, hosting the modifier, whose head induces a second, lower agreement domain (cf. Bobaljik & Wurmbrand 2005). This part of the proposal is combined with the assumption, developed above, that articles are complex elements that can be split up in different ways. First, I illustrate and motivate my assumptions. After providing the relevant derivations, I summarize the discussion. 5.3.1  Assumptions With two overt realizations of the determiner, I discuss their nature separately. 5.3.1.1  The suffixal  determiner  We have seen that the noun in the Modern Scandinavian languages always follows the adjective, (54a). Although it is tempting to conclude that the adjective prevents the noun from raising, Old Norse (and Common Scandinavian) provides evidence that this is not the case, (54b). Recall that we proposed for the latter that the NP moves across the adjective to Spec,DP, as in (54c): (54) a. *mannen gamle man-the old

(i) a.

bir büyük yeşil araba one big green car ‘one big green car’

b.

büyük yeşil bir araba big green a car ‘a big green car’

(“Modern Scandinavian”)

(Turkish)

Similarly, as mentioned in Schroeder (2006:  556), Kirmanji has the numeral for ‘one’ as a preposed free morpheme but the indefinite article as a suffix.

 Demonstratives and definite articles as nominal auxiliaries

b. maðrinn gamli man-the old ‘the old man’ c.

(Old Norse)

[DP [NP maðr]k+inni [AgrP gamli [ArtP ti tk]]]

If the present analysis of Old Norse is correct, then adjectives in Specifier positions do not unselectively block the movement of other phrases. This is not unexpected as Relativized Minimality is not purely geometrical, that is, movement of a phrase is not blocked by the simple intervention of any old phrase (cf. Chapter 1). In fact, assuming that numerals are also in Specifier positions, there is evidence from Modern Icelandic that these modifiers do not block such a movement either. Compare literary Icelandic in (55a) with the common form in (55b). The latter is derived by movement of AgrP to Spec,DP as in (55c) (Vangsnes 1999: 145–6):33 (55) a.

hinar þrjár frægu bækur mínar the three famous books my ‘my three famous books’

b. frægu bækurnar mínar þrjár famous books-the my three ‘my three famous books’ c.

(literary Icelandic)

(common Icelandic)

[DP [AgrP frægu bækurnar mínar]i D [CardP þrjár ti]]

I assume that literary and common Modern Icelandic are two different “dialects” (see also Chapter 3). Rather than proposing that adjectives block movement in the Modern Scandinavian languages but not in Old Norse, I will argue for derivations involving different kinds of movements. Before I turn to this discussion, there is an important alternative derivation that needs to be ruled out for the ungrammatical Modern Scandinavian pattern in (54a):  one might suggest that the head noun undergoes N-to-D raising under traditional assumptions, (56a), or that it moves along with the suffixal article to D under my assumptions, (56b):

33.  This movement is not possible in indefinite noun phrases: (i) a.

þrjár frægar bækur three famous books ‘three famous books’

b. *frægar bækur þrjár famous books three This means that AgrP does not always move but only when the DP is definite.



The syntactic distribution of determiners 

(56) a. (*) [DP mannk+en [NumP gamle [NP tk]]] b. (*) [DP { mannk+en}i+D [AgrP gamle [ArtP {mannk+en}i [NP tk]]]]

What both these derivations have in common is long movement of the head, simple in (56a) and complex in (56b). In order to rule out this type of movement, some scholars have argued that, following Abney (1987), adjectives are in head positions (e.g., Santelmann 1993; Delsing 1993b: 81; Kester 1993; Vangsnes 1999) and movement of the (complex) head noun is disallowed by the Head Movement Constraint (Travis 1984). However, besides ruling out the Old Norse pattern under these assumptions, this contradicts what I argued for in Chapter 1, where I concluded that adjectives are in Specifier positions (and not in head or adjoined positions). If so, the HMC cannot be invoked to exclude this kind of derivation (for a more general critique of N-to-D raising, see Giusti 2002: 58–9; Hankamer & Mikkelsen 2002, 2005).34 Keeping NP movement for Old Norse, I propose that head movement to D is not categorically ruled out. However, rather than the noun, only the determiner itself may undergo movement in these cases (note also that it is “closer” to D than the head noun). The reason why the head noun does not move along is that it combines with the determiner after syntax. In other words, in contrast to traditional assumptions, I propose that the head noun does not raise all the way to Art (or higher) but to a lower, intermediate position (Num) and that the determiner in Art undergoes (late) Merger with the noun in Num. Let us flesh this out. Recall from Chapter 1 that word order facts and Binding relations between the possessor and the complement indicate that the head noun moves out of the theta domain (Taraldsen 1990, 1991; cf. Haider 1993: 23, 30): (57) a.

podróż Jankai do swoichi rodziców trip John’s to his parents ‘John’s trip to his parents’

b. l’ aggressività di Giannii contro se stessoi the aggressiveness of Gianni against himself ‘the aggressiveness of Gianni toward himself ’

(Polish)

(Italian)

34.  As already pointed out in Chapter 1, in minimalism, the blocking effect of an adjective could only be instantiated by the assumption that both the lower element and the adjective share the relevant feature to be valued (see Julien’s work for such an argumentation). The Romance languages, which show clear cases of reordering between nouns and adjectives, indicate that this type of analysis is far from straightforward.

 Demonstratives and definite articles as nominal auxiliaries

c.

die Wut des Mannesi auf sichi the rage of.the man against himself ‘the rage of the man against himself ’

(German)

Furthermore, following Brugè (1996, 2002) I argued above that reinforcers are base-generated along with their demonstratives in ArtP. However, while we find the reinforcer on the right of the head noun in German, the determiner can never follow the noun, neither in the DP nor the ArtP-level: (58) a.

das schöne Bild da von Maria that nice picture there of Mary ‘that there nice picture of Mary’

(German)

b. (*Bild-)das schöne Bild(*-das) von Maria picture-the nice picture-the of Mary

In other words, we have evidence from Binding and the position of reinforcers that the head noun partially moves and yet the determiner in German (and other languages) is not suffixal. Note now that this is related to the discussion of Modern Scandinavian in (56b), where the head noun infelicitously moved along with the determiner. In order to explain these two types of cases (56b), (58b), we can make use of an intermediate phrase between the theta-domain of the noun phrase and ArtP. This will allow the head noun to undergo partial raising without moving all the way to ArtP. In Chapter 1, I identified this projection as NumP: (59) … [ArtP Art [NumP Num [nP Poss [NP noun [ complement]]]]]

Illustrating with the example in (60a) below, I propose that the head noun in Germanic moves to Num due to partial N-raising in syntax, illustrated in (60b) by a trace. Furthermore, “stranded” elements in ArtP, that is, (part of) the determiner or the reinforcer, undergo Merger with the head noun after syntax (for recent discussion of Merger, specifically Lowering, see Embick & Noyer 2001). Leaving out some details, Merger is indicated here by strike-through:35 (60) a.

bil-ar-na car-pl-the ‘the cars’

(Swedish)

35.  An alternative would be to assume that N-raising to the suffixal part occurs postsyntactically, which, if local, basically mimics Merger. This would also allow for independent syntactic movement and interpretation of the determiner in overt and covert syntax, which will become important below.



The syntactic distribution of determiners 

b. …

ArtP

-nai -dai

NumP nP

Num bilk Bildk

-ar+nai -Ø+dai

n′

Poss n



NP tk tk

compl

There are a number of advantages of this analysis:  first, Merger applies only to affixes. Assuming that indefinite determiners and demonstratives are not suffixes, this immediately explains their unbound occurrence. Second, Merger of the determiner applies after syntax. As such, it allows the determiner to value features on D in the syntax (see below) before it becomes part of another word. Furthermore, if the determiner independently moves from ArtP to DP in these cases, then a head noun occurs in the DP-level not due to N-raising but only due to phrasal movement, as suggested for Old Norse above. Having discussed the suffixal determiner, we turn to the free-standing one. 5.3.1.2  The free-standing determiner  In this part of the proposal, I employ the notion of “agreement domain” developed by Bobaljik & Wurmbrand (2005) for the clause and extend it here to the noun phrase. I assume with these authors that long-distance agreement is possible without movement. To foreshadow this part of the proposal, assuming that definite determiners have the feature [+def], I propose that the overt determiner in ArtP can value this feature on D from its base position as an instance of long-distance agreement. However, if a modifier is present, then the head of the hosting phrase induces a new agreement domain and the determiner has to move to the DP to value the feature on D. First, I discuss my background assumptions, then I provide the derivations, and after a short summary, I briefly compare this analysis to Julien (2002 [2005]). Bobaljik & Wurmbrand (2005) argue that Agree and A-movement have different (albeit overlapping) locality conditions. Retracing some of their discussion, let us first establish that Agree is a necessary, independent condition; that is, that there is feature valuing without movement (see also Wurmbrand 2006). Of particular interest in this regard are unaccusatives where the indirect object in the dative

 Demonstratives and definite articles as nominal auxiliaries

is higher in the structure than the theme in the nominative. Specifically, given two quantifiers, we note that there is ambiguity in (61a) but not in (61b), where the VP has undergone topicalization: (61) a.

weil mindestens einem Kritiker jeder Film gefallen sollte since at.least one(dat) critic every(nom) film please should ‘since at least one critic should like every movie’  ∃(∀),?∀(∃)

b. ?[VP Jeder Film gefallen] sollte mindestens einem Kritiker. every(nom) Film please should at.least one(dat) critic ‘At least one critic should like very movie.’ ∃(∀),*∀(∃)

The latter example is crucial as it exhibits ‘scope-freezing’; that is, the topicalized VP can reconstruct but it cannot be moved out of or moved into. What ever explains this fact will account for the disambiguity in (61b). If so, then the nominative quantifier remains in the VP in PF and LF and cannot raise to Spec,TP to value nominative case. This in turn means that case is licensed in situ, that is, via Agree. In other words, the scope-freezing facts in (61b) provide the essential argument for valuing abstract case without movement. As a second step, let us look at restructuring infinitives. Unlike the simple clauses above, here lexical verbs such as vergessen ‘forget’ select infinitives with special properties. In the datum in (62a), the embedded theme has moved out of the infinitival clause and must take scope over the matrix verb vergessen, which contains an inherent negation. Importantly, when the matrix clause is passivized, (62b), then the embedded theme occurs in the nominative and agrees in number with passive auxiliary in the matrix clause. The latter construction is called ‘long passive’: (62) a.

weil er alle Fensteri vergessen hat [ti zu schließen] since he all windows(acc) forgotten has to close ‘since he forgot to close all the windows’  *forget(∀),∀(forget)

b. weil alle Fenster zu schließen vergessen wurden since all windows(nom) to close forgotten were ‘since one forgot to close all windows’

In order to explain the obligatory high scope, the case and agreement facts, Bobaljik & Wurmbrand (2005) propose that the theme must move to the matrix clause (and Agree as above is not possible here). In particular, they suggest that lexical verbs such as vergessen induce a new agreement domain (their (21)): (63) Agreement Domain in the Clause The (verbal) complement to a lexical verb delineates an agreement domain.

To be clear, VPs selected by functional elements such as auxiliaries or modals (cf. (61)) do not form an independent agreement domain but VPs selected by lexical



The syntactic distribution of determiners 

verbs do. Now, assuming that restructuring infinitives have no functional structure that could value abstract case on the embedded theme, the object noun phrase has to move out of the lower agreement domain to value case in the matrix clause. To summarize, while the simple clauses above illustrated the independence of Agree from Move, the restructuring cases exemplified the independence of Move from Agree. Most importantly for current concerns, it became clear that certain phrases (VPs) may, under certain conditions, form independent agreement domains. In what follows, I would like to extend Bobaljik & Wurmbrand’s proposal to the DP. In particular, I would like to suggest that ArtP may form an independent agreement domain when it is the complement of a (semi-)lexical head:36 (64) Agreement Domain in the Noun Phrase The ArtP complement to a (semi-)lexical head delineates an agreement domain.

Semi-lexical heads include Card and Agr but, crucially, not functional D.  With the phrase structure of above in mind, we find the following four scenarios, where ArtP in (65a) is not an independent agreement domain but ArtP in (65b–d) is:37 (65)

a. b. c. d.

[DP D [ArtP Art […]]] [DP D [CardP QP Card [ArtP Art […]]]] [DP D [CardP QP Card [AgrP AP Agr [ArtP Art […]]]]] [DP D [AgrP AP Agr [ArtP Art […]]]]

If so, then long-distance Agree is possible in the simple DP in (65a) and movement may or may not take place. In contrast, Agree is not possible in the modified DPs in (65b–d) and movement must take place. Once movement into the higher agreement domain has taken place, overtly or covertly, feature valuing (via Agree) can take place. Note in this regard that Bobaljik & Wurmbrand (2005) suggest that Agree holds at LF. This in turn implies that, if overt movement is involved, we need to assume a generalized EPP feature (as in Chomsky’s latest work). This is what I have tacitly assumed all along. Finally, note that modifiers only indirectly, that is,

36.  Note that the version for the noun phrase is, on the one hand, narrower in its scope, but, on the other, broader: it is narrower in that the complement is specifically ArtP and broader in that the inducer is a (semi-)lexical head. Furthermore, unlike A-movement in the clause, here head and A′-movement are triggered (i.e., movement of the article and demonstrative, see especially Chapter 4). Although not straightforward at this point, there might be a possibility to collapse both generalizations into one, something I will not attempt here. 37.  There is an interesting consequence of this proposal. Considering that Num is a semilexical head, nP might also be a separate agreement domain. This would explain why head nouns must raise to value number, either by partial N-raising as in Germanic or nominal raising as in Romance.

 Demonstratives and definite articles as nominal auxiliaries

via their hosting phrase, induce a separate agreement domain. However, in the discussion to follow, I will often call this the “blocking” effect of modifiers. Returning to the main line of argument, I assume that definite determiners have the feature [+def] and that they must value a [def] feature on D in the course of the derivation. When the DP is simple, that is, when there are no modifiers, then D can value its features by Agree. However, when ArtP and DP are separated by a modifier, then Card or Agr induce an agreement domain with regard to ArtP. With Agree not possible, the determiner moves to the DP, overtly or covertly. Once it is in the higher agreement domain, Agree can occur. Thus, depending on the prevailing locality, there are, then, two ways for D to value its features from the determiner: either from in situ or after movement. Finally, I assume that there is a difference between valuing the feature [def] on D and licensing the DP with regard to referentiality (cf. Longobardi 1994). While the definite determiner can value [def] on D from in situ or after movement, referentiality is brought about by movement of the definite determiner to D (cf. Longobardi 1994) or by movement of a lower phrase to Spec,DP (Julien 2005: p. 14).38 I turn to the individual derivations. 5.3.2  Modified DP The main proposal of this work is that there is a position (ArtP) below adjectives where the overt determiner merges. Extending ideas of Brugè (1996, 2002) and others, I proposed that all determiners (demonstratives, (in-)definite articles) are merged there. Adjectives and a null determiner head are merged on top: (66) [DP D [AgrP adjective [ArtP det [NP noun]]]]

For the derivation to converge, the definite determiner has to value the feature [def] on D. With a modifier and thus its hosting phrase present, ArtP is a separate agreement domain and the determiner moves to the DP under Last Resort. This part of the derivation is shared by all languages under consideration. The languages differ with regard to which part(s) of the determiner is moved to D and when. As discussed above, the languages with the Double Definiteness effect (Swedish, Norwegian, and Faroese) can split up their determiners. To value [def] on D, the relevant part of the determiner moves to D, leaving the other parts in situ. The determiner part in Art undergoes Merger with the head noun and the part in D is licensed 38.  It is not clear why overtness, a pf phenomenon, implies referentiality, an LF phenomenon. To speculate somewhat, this licensing condition of referential DPs could in fact be the nominal equivalent of the (traditional) clausal EPP for subjects. In particular, note in this respect that overt expletive elements can license the DP (e.g., preproprial articles, as in footnote 14).



The syntactic distribution of determiners 

by d-support (cf. Santelmann 1993), which I take to be singular in application under Last Resort, similar to do-support in English.39 This is illustrated here for Swedish (abstracting away from partial N-raising): (67) a.

den gamle mannen the old man-the ‘the old man’

(Swedish)

b. [DP deni+D [AgrP gamle [ArtP ti+enk [NP mann+enk]]]]

To derive the different interpretations, I assume that different semantic components of the determiner are merged and split off in language-specific ways (Section 5.2, table 4). In contrast, Danish, literary Icelandic and common Icelandic cannot split up the determiner. Starting with Danish and literary Icelandic, as all the parts of the determiner have moved to D, there are no “stranded” lower parts to undergo Merger. Consider this for Danish (for literary Icelandic, see again Section 3.4 above): (68) a.

den gamle mand the old man ‘the old man’

(Danish)

b. [DP deni+D [AgrP gamle [ArtP ti [NP mand]]]]

As for common Icelandic, AgrP moves to Spec,DP to bring about referentiality (cf. Julien 2002; Vangsnes 1999, 2004) and the determiner in Art undergoes Merger with the partially raised noun (not shown here). Consider these two derivational steps again from Section 3.4, now updated as (69b) and (69c), respectively. Finally, after reconstructing AgrP to its base-position (for more details, see Chapter 3), the determiner values [DEF] on D by covert movement, as in (69d): (69) a.

gamli maðurinn old man-the ‘the old man’

(common Icelandic)

b. [DP [AgrP gamli [ArtP -inn [NP maður]]]j D … tj] c. [DP [AgrP gamli [ArtP -inni [NP maður+inni]]]j D … tj] d. [DP innk-D … [AgrP gamli [ArtP tk [NP maður]]]]

39.  d-support is used here for ease of exposition (see also Embick & Noyer 2001: 581–3) with the understanding that Vocabulary Insertion spells out the feature bundle of the determiner under D as the free-standing element. Now, since the present proposal argues for the splitting of determiners, it is expected that the separated syntactic and semantic features receive independent phonetic realizations.

 Demonstratives and definite articles as nominal auxiliaries

Under the assumptions above, we rule out Doubly-filled DPs in Modern Scandinavian: on the one hand, if the determiner moves to D, not only is the feature [def] valued but the DP as a whole is also licensed with regard to referentiality; on the other hand, if a phrase such as AgrP (or ArtP, see immediately) moves to Spec,DP, the determiner in Art cannot move to D overtly (being itself in Spec,DP). 5.3.3  Unmodified DP In simple DPs, there are no modifiers and, consequently, no second agreement domain is induced. In all these cases, the determiner in Art can value the feature [def] on D without movement. With the features valued by the determiner in situ, ArtP moves to Spec,DP to bring about referentiality, (70b). Finally, the determiner in Art undergoes Merger, (70c). We derive the same, homogenous pattern in all Scandinavian languages, illustrated here with Norwegian:40 (70) a.

mannen man-the ‘the man’

(Norwegian)

b. [DP [ArtP -en [NP mann ]]j D … tj] c. [DP [ArtP -eni [NP mann+eni]]j D … tj]

Finally, I consider the exceptional cases in Swedish, Norwegian, and Faroese, where the pre-nominal article is not present. 5.3.4  “Violations” of the Double Definiteness effect I noted above (cf. (41), (42)) that, besides “adjectival determiners”, “definite deictic noun phrases” also license the deletion of the pre-nominal determiner. What both cases have in common is that, under certain conditions – that is, when the noun phrase is definite either through deixis (due to the situation) or inherently (due to certain adjectives) – the upper determiner in Swedish, Norwegian, and Faroese can be left out. Following Delsing (1993b: 119), I take these two sets of data as related to one another, proposing that the pre-nominal determiner can only be deleted under Recoverability of Deletion.

40.  Wide (2004) reports that the Swedish dialects of Eastern Nyland have a Double Definiteness effect: (i)

te velån the bike-the ‘the bike’

(Eastern Nyland)

Apparently, these are not cases interpreted as demonstratives (see also Julien 2005:  61 footnote 19). To the extent that this is correct, I propose that rather than moving ArtP to Spec,DP for referentiality, here part of the determiner splits up and moves to D to license the DP. The part stranded in Art undergoes Merger.



The syntactic distribution of determiners 

As discussed above, one part of the determiner in Swedish, Norwegian, and Faroese moves to D to value the feature [def] on D. Due to the presence of the “adjectival determiner” with an inherent deictic specification, the upper part of the determiner can then be deleted under Recoverability of Deletion: (71) a. (det) sista paret the last pair-the ‘the last pair’

(Swedish)

b. [DP deti+D [AgrP sista [ArtP park+eti [NP park]]]]

I propose something similar for (72). Here the “deictic” situation licenses the deletion of the free-standing determiner, which itself is deictic in function: (72) a.

Ta (den) nya bilen take the new car-the ‘Take the new car.’

b. [DP deni+D [AgrP nya [ArtP bilk+eni [NP bilk]]]]

This deletion cannot take place when the noun phrase is used in contrastive reference. This follows from our approach of Recoverability of Deletion and is expected under deletion, where elements can, at least “optionally”, appear. (For the deletion of the suffixal determiner under Recoverability of Deletion, see Chapter 3.) Finally, recall that Danish and Icelandic (as well as German and English) do not have split articles. As discussed above, they move all parts to D or leave them all in situ. This (complex) determiner cannot be deleted under the above-mentioned conditions because the entire complex, moved or left in situ, contains the semantic component that is overtly realized as the lower part of the determiner in Swedish, Norwegian, and Faroese. In other words, the deletion of the determiner in these languages would include the semantic contribution of the lower part of the determiner, the deletion of which is not licensed and could thus not be recovered. This explains why these languages do not allow this kind of deletion of the determiner. I briefly summarize the derivations above. 5.4   Summary I proposed that determiners are complex elements, merged in ArtP. Furthermore, (part of) the determiner moves to the DP, covertly or overtly. If the determiner remains (overtly) in Art, we derive all unmodified DPs in a uniform way: ArtP moves to Spec,DP for referentiality. Furthermore, we account for modified noun phrases in common Icelandic, where AgrP moves to Spec,DP. Finally, the determiner in Art undergoes Merger with the noun.

 Demonstratives and definite articles as nominal auxiliaries

Turning to the other modified DPs, I interpreted the presence of modifiers as inducing a separate agreement domain with regard to ArtP. To value the feature [def] on D, determiners must move: either they move to the DP unsplit (Danish and literary Icelandic), part of the determiner moves to the DP and the “stranded” part undergoes Merger (Norwegian, Swedish, and Faroese), or the determiner moves covertly, after reconstruction of AgrP (common Icelandic). This is schematically summarized in table 5. Table 5.  Assumptions in the Derivation of Modified and Unmodified DPs. “Split” article all unmodified DPs Norwegian, Swedish, Faroese Danish, literary Icelandic Icelandic

– + – –

License DP ArtP determiner (split) determiner (unsplit) AgrP

Finally, we turn to a brief comparison between the current analysis and that of Julien (2002 [2005]). 6.  An alternative proposal: Julien (2002 [2005]) Julien (2002 [2005]) offers an alternative account with wide empirical coverage and represents the most comprehensive and detailed analysis that I am aware of (I limit myself to some general points; for the motivation of her proposal, see the originals discussions). She proposes the following structure, where CardP and αP are only projected if lexically filled by a numeral and adjective(s), respectively:41, 42 (73) DP  CardP  αP*  ArtP  NumP  NP

(where * is recursive)

This is basically the same structure that I have been using throughout (see Chapter 1 for arguments). In order to account for the different patterns, she makes the 41.  In Julien (2005), ArtP is reinterpreted as nP, which is assumed to be a strong phase (p. 73, 202). Although the technical instantiation is quite different, both works share many features. While I concentrate on her first publication, many of the points raised here will carry over to Julien (2005). 42.  There are other proposals that argue for a lower position of the suffixal determiner. Both Delsing (1988: 60) and Santelmann (1993: 161) assume that ArtP is located above adjectives but below D. Delsing argues for raising of both the noun and the adjective and Santelmann for lowering of the definite suffix to N. Furthermore, Delsing (1993b: 144) and Kester (1993: 148; cf. also Alexiadou 2003: 15) argue for two DPs in one and the same noun phrase, the former accounts for “double indefiniteness” and the latter for “double definiteness”.



The syntactic distribution of determiners 

following assumptions (I leave out her discussion of Western Jutlandic and Northern Swedish).43 Simplifying somewhat, Julien assumes for unmodified DPs that it is a lexical feature whether Art is overtly realized (Norwegian, Swedish, Faroese, and Icelandic) or D (Danish and literary Icelandic). In all cases, ArtP moves to Spec,DP to license the DP. The determiner in Art is supported by N-raising and the one in D by ArtP, containing the head noun: Table 6.  Julien’s Assumptions for the Unmodified DP. Realize Art

Nw/Sw/Faroese Danish/literary Icelandic Icelandic

License D(P)

Comments

D

Spec,DP

+ –

– +

ArtP ArtP

Move (of ArtP) over Realize (of article in D) Doubly filled DP

+

–

ArtP

(adjective is not present and thus no αP)

As for modified DPs, the presence of modifiers in Specifier positions is assumed to block movement of ArtP to Spec,DP. With Art realized in the same way as in unmodified DPs, Norwegian, Swedish, Faroese, Danish and literary Icelandic license the DP by overtly realizing D. In contrast, Icelandic moves the phrase containing the adjective (αP) to Spec,DP (here, unlike in the other languages, the numeral does not block this movement in Icelandic): Table 7.  Julien’s Assumptions for the Modified DP. Realize Art

Nw/Sw/Faroese + Danish/literary – Icelandic Icelandic +

License D(P)

Comments

D

Spec,DP

+ +

– –

adjectives/numerals block move of ArtP to DP adjectives/numerals block move of ArtP to DP

-

αP

adjectives (but not numerals) block movement of a phrase to Spec,DP

43.  The discussion in the main text extends to these cases fairly straightforwardly: Western Jutlandic does not have Merger (i.e., the definite determiner is not an affix) and Northern Swedish is similar to Icelandic in that the determiner moves to the DP in LF but differs from it in that adjectives and the head noun form a prosodic unit, “fusing” them into one single element (Julien 2002: 291–3). For the cases when numerals are present in Northern Swedish (see Julien 2005: 58), more discussion is needed.

 Demonstratives and definite articles as nominal auxiliaries

Although the present work owes great intellectual debt to Julien (2002 [2005]), there are also some important differences worth taking note of. I focus on two aspects. One of the major differences lies in the treatment of Icelandic. Like Julien, I argued that modifiers have a syntactic effect. Unlike Julien’s blocking, I proposed that modifiers indirectly, that is, via their hosting phrase, induce a different agreement domain. In other words, rather than ruling out phrasal movement due to the presence of an adjective, I suggested that the long-distance agreement relation between ArtP and DP is not possible and the determiner must move to the DP (this analysis will be extended to relative clauses in the next chapter). Importantly, allowing phrasal movement across the adjective straightforwardly explains the Old Norse pattern (maðrinn gamli ‘the old man’) and allows movement of a phrase across a numeral in common Modern Icelandic without further assumptions. The main differences between Old Norse and (common) Modern Icelandic are that (i) movement of NP was, over time, replaced by that of AgrP, such that the adjective now moves along to Spec,DP and (ii) the determiner does not move to D overtly anymore. Both analyses stipulate language-specific operations. While Julien (2002 [2005]) states which determiner position is overtly realized in which language (Art or D), I claim that, if a modifier is present, determiners can be split up by movement in some languages but not others. Besides achieving a homogenous account of the unmodified DP, the current analysis also directly connects the property of splitting up determiners to the corresponding different semantic interpretations of the DPs, overtly manifested in some languages but not in others.44

7.  Conclusion This chapter started out by providing a formal account of the rise of the suffixal article in the Early Scandinavian DP. This discussion was based on the assumption that determiners are base-generated in a lower phrase (ArtP) and then move to DP, either overtly or covertly. This provided a straightforward account for the Panchronic Alternation involving the historically related free-standing and suffixal determiners in literary and common Modern Icelandic. More evidence for

44.  I believe that the connection between the syntax and the semantics is less straightforward in Julien (2002 [2005]). For instance, recall that Julien’s account has different derivations for unmodified DPs across the Scandinavian languages. Now, given these different derivations, we might expect differences in interpretation. To the extent I know, this is not the case.



The syntactic distribution of determiners 

ArtP and determiner movement was provided by highlighting some of Brugè’s (1996, 2002) analysis involving demonstratives and “stranded” reinforcers. Finally, we turned to the discussion of the Scandinavian DP. Documenting that the different determiner parts in languages with a Double Definiteness effect have different semantic import, I proposed that these are cases of “split” articles, where one part is moved to DP and the “stranded” part undergoes Merger. The difference between the individual languages was proposed to follow from a lexical property that determines which determiner can be split up. The differences between the homogenous, unmodified DP and the heterogeneous, modified DP were accounted for by interpreting the modifier as inducing a separate agreement domain that “blocks” long-distance agreement from ArtP to DP but not comparable movement. Finally, the current proposal was compared to that of Julien (2002, [2005]) and some advantages of the present account were highlighted.

chapter 3

The syntax and semantics of non-/restrictive modifiers 1.  Introduction In Chapter 2, I discussed the syntactic distribution of determiners. Adjectives were interpreted as (indirectly) inducing a separate agreement domain with regard to ArtP, which “blocks” long-distance agreement but not movement. It was illustrated there that languages may differ in the way they circumvent this blocking effect. As a consequence, there is morpho-syntactic variation across languages. Furthermore, depending on the way the determiners split and which semantic components are present, the distribution of the determiner may also correlate with different interpretations of the DP as a whole. In this chapter, I turn to interpretative facts inside the DP. In particular, I consider modifiers and their interpretation with regard to (non-)restrictiveness. This discussion is especially interesting from the Scandinavian point of view: to illustrate briefly with restrictive modifiers, the distribution of the determiner in a noun phrase with an adjective is (basically) homogenous within a particular language but diverse across the different languages. Recall the basic patterns from Chapter 2: (1) a.

den gamle mann-en the old man-the ‘the old man’

b. den gamle mand the old man ‘the old man’ c.

gamli maður-inn old man-the ‘the old man’

(Norwegian)

(Danish)

(Icelandic)

d. *gamle mann old man

In contrast, the distribution of the determiner in a noun phrase with a relative clause is heterogeneous within a particular language but basically the same across the different languages. Abstracting away for now from some semantic differences

 Demonstratives and definite articles as nominal auxiliaries

(but see Section 2.2), consider an example from Swedish, where (2a) presents the “Double Definiteness” pattern, (2b) is the “Danish” one, and (2c) shows the “Icelandic” distribution. Note that with both types of modifiers, a determiner has to be present. Compare (1d) and (2d): (2) Har du sett _____ som ska köra vårt plan? have you seen who will drive our plane a.

(Swedish)

den pilot-en the pilot-the

b. den pilot the pilot c.

pilot-en pilot-the

d. *pilot pilot ‘Have you seen the pilot that will drive our plane?’

To reiterate, it is important to point out that, despite the morpho-syntactic variation, all modifiers in (1) and (2) can have a restrictive reading. The goal of this chapter is to account for both the restrictive and non-restrictive readings of the modifiers and the distributional asymmetry of the determiners in noun phrases with adjectival vs. clausal modifiers. To foreshadow the analysis, I follow Heim & Kratzer (1998) in taking the view that the semantics can be “read off ” the syntactic structure. As a consequence, specific syntactic representations will allow only certain interpretations and, conversely, semantic interpretation rules will constrain the syntactic analysis. Concretely, I will be concerned with capturing in more formal terms (i.e., in type-driven semantics) the intuition that determiners are scope-bearing elements with regard to the interpretation of modifiers. In particular, I will propose that restrictive modifiers are in the scope of determiners, (3a), but non-restrictive ones are not, (3b):

(3) a.

Restrictive Interpretation DP XP

D determiner XP

ArtP

adjective Art



relative clause

NP noun



The syntax and semantics of non-/restrictive modifiers 

b. Non-restrictive Interpretation DP XP

D XP adjective



Art determiner

relative clause ArtP NP noun

Arguing against movement of modifiers, I propose that different copies of the moved determiner are interpreted: the restrictive reading is derived by interpreting the determiner in its derived position; the non-restrictive one by interpreting the determiner in its base-position. To the extent that this is correct, this discussion provides a semantic argument for the main proposal of this work. The chapter is organized as follows: Section 2 discusses restrictive and Section 3 non-restrictive modifiers. Although I concentrate on the interpretation of adjectives in Icelandic, the discussion is to be taken more generally. To this end, each section extends the analysis of adjectives, both in the syntactic terms of Chapter 2 and in the semantic framework of this chapter, to the syntactic and semantic account of relative clauses in Scandinavian. In Section 4, I extend the approach to adjectives in Romance. Some further issues are discussed in Section 5 and the main results are summarized in Section 6. Let us begin with the discussion of restrictive modifiers.

2.  The syntax and semantics of restrictive modifiers Heim & Kratzer (1998:  83) discuss a historic debate between Barbara Partee and Noam Chomsky. Put simply, Partee (1976) argues that a restrictive relative clause combines with a noun before their intersection combines with a determiner, (4a). However, Chomsky (1975:  96–101) points out that there is some evidence against this general view such that the determiner apparently forms a unit with the noun, excluding the restrictive modifier, (4b).1 In order to maintain

1.  Actually, Chomsky (1975: 96–101) raises questions about plural DPs in combination with restrictive relative clauses. He argues for the following structure, where the noun and the plural

 Demonstratives and definite articles as nominal auxiliaries

Partee’s view, Heim & Kratzer suggest that (4a) must hold at some other level of representation, (4c): (4) a. Partee: [determiner [noun modifier]] b. Chomsky’s problem: [[determiner noun] modifier] c. Heim & Kratzer: (4a) holds at some other level of representation

As implied by Heim & Kratzer, I will suggest below that this level is LF. To make the discussion concrete, consider adjectival modifiers in Icelandic, which, on the surface, appear to have the problematic structure in (4b) (abstracting away from the order of the elements). Recall from Chapter 2 Section 5.3.2 that literary Icelandic and common Icelandic have different derivations. Concentrating on the latter, (5b), AgrP moves to Spec,DP, as in (5c). This is followed by the determiner in Art undergoing Merger with the (partially raised) noun: (5) a.

hinar þrjár frægu bækur mínar the three famous books my ‘my three famous books’

b. frægu bækurnar mínar þrjár famous books-the my three ‘my three famous books’ c.

(literary Icelandic)

(common Icelandic)

[DP [AgrP frægu bækurnar mínar]i D [CardP þrjár ti]]

Since the possessive pronoun (in Spec,nP) moves along, the overt determiner must be in Art (rather than D). Taking a less complicated example from common Icelandic, (6a), consider the main steps in the derivation in more detail: after base-generating all the elements, (6b), AgrP moves to Spec,DP, (6c), and the determiner in Art undergoes Merger with the partially raised noun, (6d) (the raising is not shown here):

morpheme form a constituent to the exclusion of the restrictive modifier. These elements are dominated by the determiner (p. 100): 

(i)

[NP determiner [N′ [N noun + plural] modifier]]

Without going into any details, the problem is not supposed to arise from the scope of the determiner but the relation between number and the relative clause. Under today’s assumptions, it is not clear, however, if this still presents a problem. Chomsky’s discussion is based on the (then) assumption that number is interpreted at the location where the morpheme appears (i.e., below the relative clause). However, if we assume that number is interpreted otherwise, then the problem does not arise. Avoiding the issue of number, I discuss Icelandic noun phrases with adjectives, essentially following Heim & Kratzer’s discussion of restrictive prepositional phrase in the Scandinavian languages, which present an apparently clear problem with regard to the scope of the determiner, as in (4b).



The syntax and semantics of non-/restrictive modifiers 

(6) a.

gamli maðurinn old man-the ‘the old man’

b. [DP D … [AgrP gamli [ArtP -inn [NP maður ]]]] c.

[DP [AgrP gamli [ArtP -inn [NP maður

]]]j D … tj]

d. [DP [AgrP gamli [ArtP -inni [NP maður+inni]]]j D … tj]

Crucially, none of the stages in (6b–d) provides the correct syntactic input for the semantic interpretation: unlike in (4a), the adjective and the noun do not form a unit to the exclusion of the determiner. 2.1  Derivation of the restrictive reading The main proposal of this work is that determiners move from ArtP to DP. I propose then that, in order to bring about the constellation in (4a), AgrP “reconstructs” to its base-position in LF. This essentially reverses (6c) to the base-generated (6b), which is repeated in (7a) below. Following that, the determiner moves to the DP, (7b):2 (7) a. [DP D … [AgrP gamli [ArtP -inn [NP maður]]]] b. [DP -inni+ D … [AgrP gamli [ArtP -inni [NP maður]]]]

Having assumed that the determiner undergoes Merger after Spell-out, it is free to move to the DP in LF on its own (i.e., without the noun). More generally then, I assume that after movement of the determiner from ArtP to DP, all languages under discussion have a copy of the determiner in ArtP and DP in LF. Consider the syntactic structure in schematic terms (to keep the discussion simple, I will, for the most part, concentrate on cases involving one modifier):

(8) [determiner [adjective [determiner [noun]]]]

The basic proposal of this chapter is that, when the lower determiner is deleted (to be revised in Section 5.3), the upper determiner is interpreted deriving the

2.  Although I will be using the term “reconstruction” throughout, this simply means that, in a copy-and-delete approach, the lower copy becomes relevant. Also, in Section 5.3 below, I claim that the suffixal determiner in Icelandic has to move above the modifier in order to avoid type mismatch. This raises the issue of ‘scope-freezing’ (cf. Chapter 2). Note, however, that the cases in Chapter 2 and here are not identical in that there is only one scope-bearing element in the cases here (i.e., the determiner). Be that as it may, if it turns out that this is indeed a case where no element can move out of AgrP, the reconstructed phrase, then we could still propose that the determiner moves above the modifier inside AgrP (e.g., adjoin to AgrP). With the determiner in the appropriate agreement domain, Agree would occur after the movement (recall from Chapter 2 that Agree holds at LF).

 Demonstratives and definite articles as nominal auxiliaries

restrictive reading; in contrast, when the upper determiner is deleted, the lower one is interpreted bringing about the non-restrictive one: (9) a.

Restrictive Interpretation [determiner [adjective [determiner [noun]]]]

b. Non-restrictive Interpretation [determiner [adjective [determiner [noun]]]]

In what follows, I make this claim more formal, using English examples. Postponing the discussion of non-restrictive modifiers to Section 3, I start with their restrictive counterparts. Consider the following question-answer pair (disregarding pronominalization of John in the answer): (10) Which man did John see? a. John saw the old man. b. John saw the man who is old.

It is clear that the modifiers in (10) contribute to the identification of the referent and are thus part of the denotation of the noun phrases. In other words, these are clear instances of restrictive modifiers. Consider the simplified LF representations of the DPs in (10) in (11) below: DP

(11) a. D thei

AgrP Agr′

AP old Agr thei

ArtP Art thei

b.

DP D thei

AgrP CP [who is old]

AgrP Agr thei

ArtP Art thei

NP man

NP man



The syntax and semantics of non-/restrictive modifiers 

Note that the answer involving the adjective and the one with the relative clause have basically the same meaning. For ease of exposition, I will concentrate on the adjectival modifier. In the framework of Heim & Kratzer (1998), I define the lexical entries of the relevant elements as follows (the entry for the is taken from Heim & Kratzer 1998: 81): (12) [[the]] = λf: f ∈ D〈e,t〉 and there is exactly one x ∈ C such that f(x) = 1. the unique y ∈ C such that f(y) = 1, where C is a contextually salient subset of D. [[John]] = J (John the individual) [[saw]] = λu: u ∈ D〈e〉 . λx: x ∈ D〈e〉 . x saw u. [[old]] = λz: z ∈ D〈e〉 . z is old. [[man]] = λw: w ∈ D〈e〉 . w is a man.

The following calculation illustrates the account of the restrictive modifiers: first, the adjectival predicate old (type 〈e,t〉) is combined with the head noun man (〈e,t〉) by Predicate Modification, (13c–d); then, the resulting intersection (〈e,t〉) combines with the determiner the (〈〈e,t〉, e〉) by Functional Application, (13f–g): (13) [[John saw the old man]] = 1 iff a. [[saw the old man]] ([[John]]) = 1 iff b. [[saw]] ([[the old man]]) (J) = 1 iff c. [[saw]] ([[the]] ([[old man]])) (J) = 1 iff d. [[saw]] ([[the]] (λv: v ∈ D〈e〉 .[[old]] (v) and [[man]] (v) = 1)) (J) = 1 iff e. [λu: u ∈ D〈e〉 . λx: x ∈ D〈e〉 . x saw u] ([λf: f ∈ D〈e,t〉 . the unique y ∈ C such that f(y) = 1] (λv: v ∈ D〈e〉 .[λz: z ∈ D〈e〉 . z is old] (v) and [λw: w ∈ D〈e〉 . w is a man] (v) = 1)) (J) = 1 iff f. [λu: u ∈ D〈e〉 . λx: x ∈ D〈e〉 . x saw u] ([λf: f ∈ D〈e,t〉 . the unique y ∈ C such that f(y) = 1] (λv: v ∈ D〈e〉 .[v is old and v is a man])) (J) = 1 iff g. [λu: u ∈ D〈e〉 . λx: x ∈ D〈e〉 . x saw u] ([the unique y ∈ C such that [λv: v ∈ D〈e〉 .[v is old and v is a man]] (y) = 1]) (J) = 1 iff h. [λu: u ∈ D〈e〉 . λx: x ∈ D〈e〉 . x saw u] ([the unique y ∈ C such that y is old and y is a man]) (J) = 1 iff J(ohn) saw the unique y ∈ C such that y is old and y is a man

To conclude, interpreting the upper determiner derives the restrictive interpretation of the adjective. If the upper determiner were to be deleted and the lower determiner interpreted, a problem would arise:  proceeding bottom-up, the

 Demonstratives and definite articles as nominal auxiliaries

nominal predicate man and the determiner would combine, giving a referential expression (〈e〉). The adjectival predicate old could then take this element as an argument and return an expression of type 〈t〉. However, the verbal predicate saw cannot combine with this outcome and the sentence would be uninterpretable. I consider some syntactic implications. 2.2  Syntactic consequences: Scandinavian relative clauses In Chapter 2, I showed that pre-nominal modifiers (indirectly) induce a separate agreement domain, which leads to different syntactic distributions of the determiner. Assuming that all modifiers are the same in the relevant sense, we expect that the presence of other types of modifiers will also show syntactic effects of this induced agreement domain. This prediction is borne out by restrictive relative clauses in Scandinavian. Recall from the introduction that, in contrast to adjectives, relative clauses allow for more syntactic patterns within a particular language, at the same time making them more homogenous across the languages under discussion. I now turn to an explanation of this asymmetry.3 Consider again the example (2) from Swedish, repeated here as (14) for convenience, where the determiner of the head noun has the “Double Definiteness” distribution in (14a), the “Danish” one in (14b), and the “Icelandic” pattern in (14c). Crucially, a determiner cannot be missing, (14d) (examples from Kester 1996b: 117; for similar data, see Börjars 1998: 142):4 (14) Har du sett _____ som ska köra vårt plan? have you seen who will drive our plane a.

(Swedish)

den pilot-en the pilot-the

b. den pilot the pilot c.

pilot-en pilot-the

d. *pilot pilot ‘Have you seen the pilot that will drive our plane?’

3.  For the discussion of the different relative pronouns across the Germanic languages, see Platzack (2002). 4.  Julien (2005: 88) notes that if an adjective is added, the distribution of the determiner is as in Chapter 2, that is, a free-standing determiner will also occur in (14c). This holds for both the restrictive and non-restrictive relative clauses.



The syntax and semantics of non-/restrictive modifiers 

Julien (2005: 77–82) points out that, depending on the linguistic and situational context, the general distributions in (14a–c) have slightly different interpretations and frequencies in Swedish, Faroese, and Norwegian. In particular, the “Double Definiteness” pattern in (14a) may not only have a unique and specific reading in these three languages but also a non-specific/generic interpretation, (15a). Similar facts hold in Swedish and I assume also in Faroese. In contrast, the “Danish” pattern is only unique and specific in Swedish, (14b), but non-specific/generic in all three languages, (15b) (Julien’s examples (3.9b) and (3.10b)): (15) a.

Den spiller-en som får høyest tall begynner. the player-the who gets highest number begins ‘the player that gets the highest number begins’

(Norwegian)

b. Den spiller som får høyest tall begynner. the player who gets highest number begins ‘the player that gets the highest number begins’

Finally, the “Icelandic” pattern in (14c) is only unique and specific in all three languages. With the exception of the Double Definiteness pattern, Danish also has two forms for restrictive relative clauses: Hankamer & Mikkelsen (2002: 166) point out that (16a) is only restrictive in interpretation. In contrast, (16b) is non-restrictive but can also be restrictive for younger speakers:5 (16) a.

den hest der vandt løbet the horse that won race-the ‘the horse that won the race’

(Danish)

b. hest-en der vandt løbet horse-the that won race-the ‘the horse that won the race’

To the extent that I am aware, both distributions also allow a non-specific/generic interpretation. Under the assumptions of Chapter 2, the absence of the Double Definiteness pattern is expected as Danish does not allow determiner splitting for independent reasons. Thus, once the dis-ability to split determiners is factored in, the distribution of the determiner with restrictive relative clauses within a language is basically parallel to the distribution of the determiner with adjectives

5.  This important asymmetry in the interpretative possibilities in (16a–b) will be captured below by the employment of two different analyses of relative clauses.

 Demonstratives and definite articles as nominal auxiliaries

across the different languages. Let me summarize these complex sets of data and their corresponding interpretations:6 Table 1.  Different Patterns of Restrictive Relative Clauses and their Interpretations. Patterns

Non-specific/generic

Unique and specific

den N-en som den N som N-en som

Sw, Fa, Nw Sw, Fa, Nw, Da Da (younger)

Sw, Fa, Nw Sw, Da Sw, Fa, Nw, Da (younger)

Some remarks are in order here: Ad row 1: Assuming that restrictive relative clauses are modifiers, the Double Definiteness pattern in Swedish, Faroese, and Norwegian is not unexpected. Note again that, since Danish does not split its articles (Chapter 2), the lack of this pattern in the latter language follows. Ad row 2: All languages have the “Danish” pattern with non-specific/generic interpretations. In contrast, only Swedish and Danish have this pattern with unique and specific interpretations. Comparing rows 1 and 2, one can also state that in noun phrases with non-specific/generic interpretation, the suffixal determiner is basically optional in Swedish, Faroese, and Norwegian. In other words, this distribution is like that with adjectives, discussed in Chapter 2, and I assume that the suffixal part is also an expletive element in these cases. Ad row 3:  While only Danish allows the “Icelandic” pattern with a nonspecific/generic interpretation, all languages have this distribution with a unique and specific reading. Crucially, this pattern is not possible when an adjective is present and, consequently, the analyses of these two types of modifiers should not be collapsed.

6.  Recall from Chapter 2 that, with a specific interpretation, a suffixal determiner occurs with noun phrases modified by an adjective. Interestingly, the suffxal determiner can be left out if an additional restrictive relative clause is present (Platzack 1997: 71 footnote 11): (i)

Det röda hus vi just körde förbi ägs av min syster. the red house we just drove by is-owned by my sister ‘the red house we just drove by is owned by my sister’

(Swedish)

This is exceptional and I suggest that, if both types of modifier are present, then the predicative part of the noun phrase (i.e., the adjective, the relative clause, and the noun itself) is semantically “overdetermined” with regard to specificity. This presumably licenses the deletion of the suffixal determiner under Recoverability of Deletion.



The syntax and semantics of non-/restrictive modifiers 

In Chapter 2, I proposed that modifiers (indirectly) induce a separate agreement domain with regard to ArtP. If restrictive relative clauses are modifiers as well, we also expect that they induce a separate agreement domain. In what follows, I will claim just that. However, the greater variety of the distribution of the determiner in noun phrases with relative clauses follows from different adjunction sites of the relative clause, on the one hand, and a second, alternative analysis of relative clauses involving head raising (Vergnaud 1974; Kayne 1994), on the other.7 In particular, while there might be some potential overlap by the analyses, I claim that the “Double Definiteness” and “Icelandic” patterns basically involve the traditional adjunction analysis and that the “Danish” pattern basically involves the alternative head-raising analysis (for general background, see Alexiadou et al. 2000). I start with the former, returning to the latter at the end of this section. First, I consider the syntactic structures of restrictive relative clauses in the languages with the Double Definiteness effect. Assuming that Specifier positions on the right are universally not available (Kayne 1994), I analyze relative clauses to be adjoined on the right. In other words, pace Kayne, I will follow Takano (2003) in assuming “weak antisymmetry”, that is, that all Specifiers are on the left but that adjunction to the right is allowed. In order to determine the adjunction site, recall from Chapter 1 that there is a special relationship between adjectives and determiners:  the former make the latter obligatory. To capture this, I proposed that Agr may select for another AgrP or ArtP. This discussion can now be extended. Note that Delsing (1993b: 32) observes that the article in front of a predicative noun becomes obligatory when the noun appears with a descriptive modifier: (17) Han är *(en) karl som man kan lita på. he is a man that one can trust ‘He is a man one can trust.’

(Swedish)

Similar to the cases in Chapter 1, the indefinite article in (17) is semantically vacuous and its obligatory presence must be syntactically motivated.8 If we assume that the relative clause can only adjoin to AgrP or ArtP, then a determiner

7.  Two analyses of relative clauses in the same language (family) have also been proposed by Hankamer & Mikkelsen (2005: 117) for Swedish and Hulsey & Sauerland (2006) for English. 8.  This is different for non-restrictive modifiers. Here the article is obligatory for a semantic reason, since only specific noun phrases can form the antecedent of appositive relatives (see below).

 Demonstratives and definite articles as nominal auxiliaries

must be present as well. I conclude that the restrictive relative clause cannot be adjoined lower than ArtP (e.g., to NP) or higher than AgrP (e.g., to DP).9 If this is tenable, then the relative clause has a “choice” between adjoining to AgrP and ArtP. In keeping with the discussion of adjectives in Chapter 2, if a relative is adjoined to AgrP, the head Agr will induce a separate agreement domain. As a consequence, the determiner has to move to the DP to value the feature [def] on D. As in the case with adjectives, the determiner will be split up and the lower part undergoes Merger. This derivation brings about the general “Double Definiteness” patterns:10 (18) a. Har du sett den piloten som ska köra vårt plan? have you seen the pilot-the who will drive our plane ‘Have you seen the pilot that will drive our plane?’ b. DP D deni

(Swedish)

AgrP CP [som ska köra vårt plan]

AgrP Agr ti

ArtP Art -eni

YP ... pilot

9.  For two other arguments that restrictive relative clauses cannot adjoin to DP but only lower, see Stroik (1994: 46–8). Actually, there is independent evidence from Icelandic that relative clauses can attach just below DP (the example is taken from Sigurðsson 2006: 209): (i) a.

allar frægu hygmyndirnar þrjár sem all famous ideas-the three that



eru um málfræði are about grammar



‘all the three famous ideas that are about grammar’



[XP allar [DP [AgrP frægu hygmyndirnar ]i D [CardP þrjár ti [CP sem…]]]]

b.

(common Icelandic)

Recalling that Icelandic moves AgrP to Spec,DP and noting that an adjective and a numeral are present, the relative clause could be adjoined to CardP or (the upper segment of) AgrP. 10.  Note that this derivation also explains the “Danish” pattern with a non-specific/generic reading (cf. row two in table 1). In particular, similar to Chapter 2, the “stranded” part of the determiner is assumed to be an expletive and may or may not appear.



The syntax and semantics of non-/restrictive modifiers 

As for the “Icelandic” distribution, I propose that the relative clause is adjoined to ArtP. As there is no semi-lexical head in the structure (e.g., Card or Agr), ArtP does not form a separate agreement domain and the determiner can value features on D without movement: (19) a.

Har du sett pilot*(-en) som ska köra vårt plan? have you seen pilot-the who will drive our plane ‘Have you seen the pilot that will drive our plane?’

(Swedish)

DP

b. D

ArtP CP [som ska köra vårt plan]

ArtP Art -en

YP ... pilot

To finalize the derivation, ArtP moves to Spec,DP to license the DP with regard to referentiality and the determiner in Art undergoes Merger (not shown in (19b)).11 After these syntactic considerations, I turn to the account of the restrictive readings. In the “Double Definiteness” pattern, the modifier is in the scope of the determiner and the restrictive reading follows.12 As for the “Icelandic” pattern,

11.  Noun phrases modified by a prepositional phrase only allow a suffixal determiner, (ia). I assume that the PP is adjoined to ArtP. Interestingly, if a restrictive relative clause is added, the free-standing determiner is possible, (ib). With the above discussion in mind, I suggest that the relative clause is adjoined to AgrP, which induces a separate agreement domain (data from Hankamer & Mikkelsen 2005: 111–2): (i) a.

gris-en med blå pletter pig-the with blue spots ‘the pig with blue spots’

(Danish)

b.

den gris med blå pletter som vi fik af nabo-en the pig with blue spots that we got from neighbor-the ‘the pig with blue spots that we got from the neighbor’

As pointed out by Hankamer & Mikkelsen (2005: 118), the same facts hold for argument PPs and the analysis is similar to that of (i). 12.  The possibility of splitting the determiner raises the question of which part is responsible for bringing about the interpretation with regard to (non-)restrictiveness. As I have no clear

 Demonstratives and definite articles as nominal auxiliaries

the derivation is parallel to that of the restrictive adjectives in Icelandic discussed above: after reconstruction of ArtP and movement of the determiner to the DP in LF, the determiner takes scope over the relative clause and, consequently, the head noun and the relative can combine by Predicate Modification. I turn to the syntax of the “Danish” pattern. Note first that, while the pattern for the non-specific/generic interpretation can be derived by the traditional account (cf. footnote 10), it can also be analyzed involving the alternative, head-raising structure, as proposed by Vergnaud (1974) and Kayne (1994:  87). However, in order to explain that only Swedish has the “Danish” pattern for a unique and specific reading and that the “Danish” pattern has only a restrictive interpretation in Danish, we must employ a different, language-specific analysis.13 Employing the head raising account, I propose that Art takes CP as a complement and the noun, as part of NP, raises to Spec,CP. In order to license the DP, ArtP moves to Spec,DP (not shown here): (20) a.

den hest der vandt løbet the horse that won race-the ‘the horse that won the race’

(Danish)

b. [DP D [ArtP d+en [CP [NP hest]i der ti vandt løbet]]]

Now, unlike with adjectives, the head nominal here is not a head in Num (Chapter 2) but a phrase in Spec,CP. As a consequence, I assume that the determiner in Art cannot undergo Merger with it and undergoes d-support under Last Resort. Finally, if an adjective or numeral is present, only the traditional adjunction analysis is possible. This follows from the assumption that, on the one hand, only NP can move to Spec,CP and, on the other, if the adjective or numeral is merged between DP and ArtP, the “stranded” part in Art cannot undergo Merger in Swedish as the noun is not in a head position (as just discussed). Assuming that d-support can only occur once (here in D), the suffix in Art cannot be supported, which leads to ungrammaticality. Turning to the semantics, the restrictive-only reading of the relative clause follows if movement of IP to Spec,DP at LF is excluded (see Wurmbrand 2006: 193–5). Consequently, both the noun and the restriction are in the scope

evidence bearing on this question, I leave this question for further research but I assume that the relevant part is in the appropriate position, at the latest, in LF. 13.  Note, however, that like Swedish, Faroese and Norwegian may, more restrictively, also have this pattern (Julien 2005:  78) and, if I understand Julien (2005:  97) correctly, Danish may also have a non-restrictive interpretation here. Despite the potential relevance of these empirical observations, I proceed on the basis of the more commonly accepted description.



The syntax and semantics of non-/restrictive modifiers 

of the determiner and the modifier is restrictive. Finally, let us consider again the paradoxical distributions of the determiner in a noun phrase with an adjective visà-vis one with a relative clause (cf. (1), (2)). Being in the Specifier position of AgrP, the adjective, or rather Agr, always induces a separate agreement domain with regard to ArtP. Consequently, there is only one basic pattern per language, though languages vary in the way they resolve the “blocking” effect. This results in a diverse picture across the Scandinavian languages. However, with different adjunction sites available on the right and a second, alternative analysis, more variety is possible with relative clauses within a particular language, basically exhausting all the possibilities with regard to the distribution of the determiner. As a result, relative clauses are similar across the different Scandinavian languages. In this section, I started with the observation that, on the surface, restrictive adjectives in common Icelandic are not in the scope of the determiner. Proposing that the determiner moves to the DP, the relevant semantic constellation holds in LF. Employing independently motivated semantic rules, this derivation presents a semantic argument for the hypothesis that determiners move to the DP, one of the main claims of this work. The discussion was extended to relative clauses, where the asymmetry in the distribution of the determiner followed from the homogeneous Specifier positions for the adjectives and the different adjunction sites and a second type of analysis for relative clauses. In the next section, I turn to non-restrictive modifiers.

3.  The syntax and semantics of non-restrictive modifiers In the last section, I provided an account for restrictive modifiers that involved the interpretation of the determiner after both the predicate head noun and adjective were combined. In other words, the determiner “closes off ” the DP with regard to referentiality and if predicates are in its scope, then they will be part of the extension of the DP. In this sense, the determiner is a scope-bearing element.14 If this is so, we can also imagine “inverse” scope to hold, that is, that the determiner copy below the modifier is interpreted. In fact, assuming base-generation of the determiner in ArtP and movement, we can derive the non-restrictive reading of modifiers at little extra cost by interpreting the low copy. It is well-known that certain relative clauses have a non-restrictive reading (see among many others Bach & Cooper 1978; Borsley 1992; Cinque 1981; de Vries 2006; Emonds 1979; Fabb 1990; Kaisse 1981; Safir 1986; Sag 1997; Sells 1985;

14.  Note that Longobardi (1994: 634) argues that the element in D is an operator.

 Demonstratives and definite articles as nominal auxiliaries

Thompson 1971; Thorne 1972). Adjectives have received less attention in this respect. Recall now that noun phrases with adjectives in Modern Icelandic have two patterns: while the common form involves a weak adjective and a suffixal article, (21a), literary Icelandic also exhibits a weak adjective but has a free-standing article, (21b). Besides these two patterns, there is a third, less often discussed form that involves a strong adjective and a suffixal article, (21c):15 (21) a.

guli bíllinn yellow(weak) car-the ‘the car that is yellow’

b. hinn guli bíll the yellow(weak) car ‘the car, which is yellow’ c.

(common Icelandic)

(literary Icelandic)

gulur bíllinn yellow(strong) car-the ‘the car, which is yellow’

The example in (21c) is not an indefinite DP, which also involves a strong adjective, but a definite one. In contrast to the other Scandinavian languages, these morphosyntactically different forms seem to correlate with differences in interpretation. Citing work by Rögnvaldsson, Delsing (1993b: 132 footnote 25) points out that, while the adjective in (21a) is restrictive in interpretation, (21c) is non-restrictive (see also Thráinsson 1994: 166; Sigurðsson 2006: 200 footnote 3). Furthermore, based on observations by Indriðason, Vangsnes (1999: 131 footnote 24; p.c.) states that the adjective in literary Icelandic in (21b) can only have a non-restrictive reading. In short, while the adjective in the common pattern is restrictive, the other two forms are non-restrictive in interpretation (see Section 5.3 for a lexical account for this asymmetry). In the last section, I derived the restrictive interpretation in (21a) by reconstruction of AgrP and subsequent movement of the article to the DP in LF. I now turn to the account of the non-restrictive readings of (21b–c). In general terms, I propose that Icelandic has two different analyses for non-restrictive adjectives, such that the weak adjective in (21b) is syntactically part of the noun phrase but the strong one in (21c) is not. Arguing against an approach involving movement of the adjective, I propose that the determiner in (21b) is interpreted in its base position in ArtP, resulting in the non-restrictive reading of the now “higher” adjective. This provides an argument for the claim that the article in (21b) must have moved from ArtP to DP, one of the main claims of this work.

15.  For instance, this form is not discussed in Hughes (2007).



The syntax and semantics of non-/restrictive modifiers 

More generally, the overt syntactic distribution of the determiner vs. adjective in (21a–b) is the reverse of what one would expect from the semantic point of view. “Mismatches” of this kind are well-documented in the literature and are usually accounted for by movement or reconstruction of the relevant (quantificational) element (cf. Bobaljik & Wurmbrand 2005). Having interpreted the determiner as a scope-bearing element, we in fact expect the determiner itself to move or reconstruct (rather than a different element). Consider more complex data, where a second adjective has been added to common Icelandic in (22a) and literary Icelandic in (22b) (the data are provided by Þóra Ásgeirsdóttir): (22) a.

ljóti guli bíllinn ugly(weak) yellow(weak) car-the ‘the car that is ugly and yellow’

b. hinn ljóti guli bíll the ugly yellow car ‘the car, which is ugly and yellow’

As is often stated, the free-standing article in literary Icelandic is only possible if an adjective occurs. As argued in Chapter 2, this follows from the assumption that the adjective induces a separate agreement domain, bringing about movement of the determiner to the DP. As in Danish, this movement is overt in literary Icelandic. Now, if adjectives have a “blocking” effect, the non-restrictive adjectives in (22b) must be a syntactic phenomenon as they interact with the determiner. Furthermore, being syntactic, the weak ending of the adjectives is expected in (22b) just as it is in (22a). This is in contrast to the third pattern, which involves a strong adjective and does not allow a preceding determiner, (23b–c): (23) a. ?ljótur gulur bíllinn ugly(strong) yellow(strong) car-the ‘the car, which is ugly and yellow’ b. *hinn ljótur gulur bíll the ugly yellow car c. *þessi ljótur gulur bíll this ugly yellow car

The ungrammaticality in (23b–c) is surprising in view of the fact that Icelandic does allow a free-standing determiner with non-restrictive adjectives, (22b). I suggest that the obligatory absence of the free-standing determiner with the strong adjective is the result of a different analysis from the one in literary Icelandic. Importantly, the claim about two analyses has already been made for non-restrictive relative clauses.

 Demonstratives and definite articles as nominal auxiliaries

Accounting for some differences between non-restrictive relative clauses in Italian introduced by cui vs. il quale,16 Cinque (1981) proposes that there are two analyses of non-restrictive relative clauses: in (24a), the relative clause is part of NP; in (24b), it is not. I will call the former case the “syntactic” and the latter the “non-integrated” type. His structures are slightly updated (CP = Cinque’s S’): (24) a. [NP NP CP] b. NP.., CP, ..

With regard to the syntactic representation of non-integrated relative clauses, Cinque (1981: 286 footnote 26) states that they may “in fact [be] outside of the theory of phrase structure proper” (for related ideas, see Emonds 1979; Fabb 1990; Giorgi & Longobardi 1991: 252 footnote 42; Safir 1986). Finally, assuming parametric variation, English differs from Italian in that it only has non-integrated appositives. As we will see in Section 3.2, Swedish provides evidence for syntactic non-restrictive relative clauses. In the absence of evidence to the contrary, I assume that the two analyses in (24) lead to a similar interpretation. If this is so, syntactic and non-integrated non-restrictive modifiers must be, at least, similar at some level of semantic representation. As the non-integrated type is not part of the syntactic tree representation, it presumably does not enter into any relation with the determiner, which is my main focus here. For present purposes then, I will concentrate on the syntactic type and the third pattern in Icelandic is not further discussed here. As already mentioned above, Partee (1976: 54) proposes that the interpretation of modifiers is often taken to be a matter of scope between the determiner and the modifier (cf. also Crisma 1996: 70; Delorme & Dougherty 1972: 9 footnote 4; de Vries 2006: 234; Hankamer & Mikkelsen 2002: 167; Kayne 1994: 112; Kim 1997; Mallén 2002: 178; Platzack 1997: 77–8; Sproat & Shih 1991: 574; Sybesma & Sio 2006: 13; Ticio 2003: 131).17 Abstracting away from linear order, let me illustrate this again in schematic terms: (25) a. Restrictive Modifier: [determiner [modifier noun]] b. Non-restrictive Modifier: [modifier [determiner noun]]

16.  For instance, both types of relative clauses differ in the way they require the relative pronoun to be present and in the way relative pronouns can pied-pipe certain other material (see also Cinque to appear). 17.  For alternative proposals, see Anderson (2002: 120–4), Bach & Cooper (1978), Longobardi & Giorgi (1991: 246 footnote 21), Radford (1993) and Zamparelli (1993).



The syntax and semantics of non-/restrictive modifiers 

The individual proposals differ in how they bring about the constellation in (25b). Before I turn to my own analysis, I will point out some problems for the most frequent alternative, which involves movement of the adjective across the determiner. To this end, consider noun phrases containing possessives and demonstratives: (26) a. our excellent students b. these excellent students

Note that both noun phrases can have a non-restrictive interpretation of the adjective. Under the assumption that determiners bring about the different interpretations, it is not immediately clear how this works for possessives, where a determiner is apparently not present. However, a possessive can co-occur with an article in some languages, (27a), and I assume that the determiner is responsible for the different interpretations. Although the determiner is presumably always present, I assume that the reason why this pattern is not more frequent has to do with the Doubly filled DP filter, which disallows two overt elements in the DPlevel. Interestingly, when a demonstrative occurs instead of a possessive, (27b), the distribution of the article is slightly different and I assume that this optional instance of the determiner is an expletive element (the data are from Vangsnes 1996: 2, 1999: 157–8): (27) a.

mett te stór húse my the big house-the ‘my big house’

(Lappträsk Swedish)

b. tetta (te) stór húse mett this (the) big house-the my ‘this my big house’

While these data provide evidence for the presence of the determiner, they also show that both the possessive and the demonstrative occupy Spec,DP (see also Roehrs 2005b; Roehrs & Sapp 2006). Now, if this is correct, then the adjective cannot move to Spec,DP, as this position is already filled. Furthermore, the adjective cannot move out of the DP, as it would have to skip the already filled “escape hatch”. In fact, this type of movement would present a violation of the Left Branch Condition (for recent discussion, see Bošković 2005). Importantly, if the adjective in (26) cannot move, it remains in the scope of the determiner and should be restrictive only, in contrast to the facts. Another account is needed. A second argument against the claim that adjectives move across the determiner derives from constructions with two non-coordinated modifiers (for Icelandic, see (22b)). It is clear from the discussion in Chapter 1 that these modifiers do not form a constituent to the exclusion of the lower part of the noun phrase, since they are in different Specifier positions. Consequently, they cannot move to Spec,DP as a

 Demonstratives and definite articles as nominal auxiliaries

unit ((28a) is from Giorgi & Longobardi 1991: 242 footnote 14; (28b) is provided by Elisa Di Domenico, p.c.): (28) a.

l’ ottima nuova segretaria di Maria the excellent new secretary of Mary ‘the secretary of Mary, who is excellent and new’

(Italian)

b. una simpatica piccola ragazza a nice small girl ‘a girl, who is nice and small’

Based on these two arguments, I conclude that adjectives do not move to Spec,DP (for a more detailed discussion of Romance, see Section 4).18 Fortunately, my proposal does not involve any movement of the adjective. In fact, it involves basegeneration of the determiner in ArtP and movement of it to the DP for independent purposes. Continuing in the framework of Heim & Kratzer (1998), I propose in what follows that non-restrictive modifiers are propositions (type 〈t〉) and that the lower determiner copy is interpreted. 3.1  Derivation of the non-restrictive reading Thompson (1971) characterizes the semantics of non-restrictive relative clauses as “an assertion…, a comment injected into the sentence whose truth is being vouched 18.  It is sometimes claimed that non-restrictive adjectives block extraction from the antecedent DP. As the adjectives themselves cannot occupy Spec,DP in LF, as discussed in the text, there must be a different reason for this restriction. Considering the semantics of the antecedents, I assume that this is a semantic issue, since these DPs are specific in interpretation (e.g., Fodor & Sag 1982: 361; Vangsnes 1999: p. 72; Zhang 2004; but cf. also Sells 1985; Authier & Reed 2005). This becomes clear if we make use of Ritter’s (1995: 437 footnote 21) observation that an indefinite specific DP can be co-referential with a personal pronoun but indefinite nonspecific ones can only be co-referential with an impersonal pronoun. Consider in this regard (ia). As the noun phrase is ambiguous, it can be pronominalized by both types of pronouns. However, if übrigens ‘by the way’ is added, (ib), the reading is disambiguated as specific only. In this case, only the personal pronoun may be used: (i) a.

Ich suche [ein rotes Haus]i. Kennst du {esi/einsi}? I am looking  for a red house. Do you know it/one?

b.

Ich suche [ein übrigens rotes Haus]i. Kennst du {esi/#einsi}? I am looking  for an incidentally red house. Do you know    it/one?

(German)

If non-restrictive modifiers, introduced by übrigens, only take specific antecedents, the contrast in (ib) follows. Returning to extraction, although non-restrictive modifiers have a disambiguating function, they themselves are not responsible for the ban on extraction. In other words, specific DPs do not allow extraction for semantic reasons, independently of the presence of the non-restrictive adjective.



The syntax and semantics of non-/restrictive modifiers 

for…independently of the content of the rest of the sentence” (p. 86). I follow this basic semantic intuition, adding that the “presuppositional” assertion of the non-restrictive modifier is associated with its “hosting” proposition at some level of representation. I suggest that this association takes the form of some type of conjunction. As already seen above, restrictive modifiers combine by Predicate Modification, a form of conjoining the relevant properties, (29a). Turning to non-restrictive modifiers, I follow Sproat & Shih (1991: 574), who propose that conjunction holds at a different level, namely between propositions, (29b): (29) a.

This man who is walking is speaking. [this x | (man′(x) & walking ′(x))] is speaking

b. This man, who is walking, is speaking. [this x | man′(x)]i is speaking & hei is walking

The pronoun in the second conjunct in (29b) is co-indexed with its antecedent in the first conjunct. To reiterate, the difference between restrictive and nonrestrictive modifiers is taken to be a matter of the scope of the conjunction (property vs. proposition). Besides this semantic intuition, there is also some syntactic evidence that non-restrictive modifiers are similar to conjunctions. The first argument derives from the behavior of quantifiers. Going back to work of Ross’, Jackendoff (1977: 175–6) points out that every and any can only be part of the “antecedent” of a restrictive relative clause but are not possible with non-restrictive or conjoined clauses: (30) a. Any/Every man who drives a Cadillac is insane. b. *Any/*Every man, who drives a Cadillac, is insane. c. *Any/*Every man is insane, and he drives a Cadillac.

Similar facts hold for Negative Polarity Items such as any ((31a–b) is taken from Fabb 1990: 70): (31) a. Only the tourists who have any imagination go to visit Sicily. b. *Only the tourists, who have any imagination, go to visit Sicily. c. *Only the tourists go to visit Sicily, and they have any imagination.

Thus, non-restrictives are semantically and syntactically similar to conjunctions (for more arguments, see among others Emonds 1979:  Section 2; see also the references in Thompson 1971: 84). In fact, if de Vries (2006) is correct, then this type of relative clause is a specifying conjunct. We are now ready to execute these ideas in more detail. Considering the assumption that non-restrictive modifiers are associated with their hosting proposition by conjunction and the fact that conjunction is only allowed with elements of the same semantic type, the non-restrictive modifier must

 Demonstratives and definite articles as nominal auxiliaries

be a proposition. I propose, then, that non-restrictive adjectives are more complex in that they have a co-occurring null pronoun (cf. Sells 1985: 30). As a consequence, the adjective (〈e,t〉) and the pronoun (〈e〉) can combine by Functional Application to yield a proposition. I will assume here that this element is pro. As already seen above, pro has to be co-indexed with the “hosting” DP: (32) [diese [proi exzellenten] Studenten]i these excellent students ‘these students, who are excellent’

We now turn to the question of how the syntactic type of non-restrictive modifiers gets from its position inside the DP in the syntax to be coordinated with the “hosting” proposition in the semantics. As far as I am aware, there is little discussion on this in the literature and not much relevant evidence. Consequently, the next few paragraphs will be tentative. Importantly, however, the lower position of the determiner will play a crucial role. Consider the following question-answer pair: (33) A: Who did John see, the man or the woman? B: Believe it or not, a. John saw the (incidentally) old man. b. John saw the man, who is old.

Note that the modifiers in this case involve additional information not needed to identify the referent. As such, these modifiers are non-restrictive. As proposed above, I assume that, with non-restrictive modifiers, the upper determiner is deleted and the lower one is interpreted (cf. (11)): (34) a. John saw [thei [old [thei man]]] b. John saw [thei [[thei man] who is old]]

The syntactic representations in (34a) and (34b) are essentially the same. In other words, I will treat all non-restrictive modifiers in the same way, that is, independent of their categorial type (cf. Ticio 2003: 123 footnote 7). Note that the non-restrictive modifier containing old intervenes between the verbal predicate saw and the nominal argument the man in (34). In other words, the verbal functor and its nominal argument cannot combine directly. Assuming that the noun phrase is a phase (cf. Chomsky’s work), we can say that elements of type 〈e〉 and 〈t〉, being semantically complete and atomic, can be sent off individually and are “associated” with the rest of the clause later. Let us flesh this out. The transition of non-restrictive adjectives from the syntactic representation, where they are constructed in, to semantic coordination with the “hosting” proposition minimally involves two different levels of representation (syntax and a “late” level of semantics). I propose that this transition occurs “on-line”; that is,



The syntax and semantics of non-/restrictive modifiers 

there is multiple semantic spell-out. More concretely, assuming that atomic semantic elements can be “sent off ” separately, I propose that expressions of type 〈e〉 are “associated” with predicates and those of type 〈t〉 with the “hosting” proposition at some level of representation: TP〈t〉

(35)

non-restrictive modifier〈t〉 subject DP〈e〉

T′〈e,t〉

 T

VP〈e,t〉 V〈e,〈e,t〉〉



object DP〈e〉

Starting with the most deeply embedded element at the bottom of the syntactic tree (i.e., the object in (34a)), its semantic value is calculated by applying Functional Application to the determiner (〈〈e,t〉, e〉) and the nominal predicate (〈e,t〉). This results in a referential expression (〈e〉), which can be sent off to the semantics. Higher in the object DP, there is the non-restrictive adjective, which consists of a regular adjective (〈e,t〉) and the pronoun pro (〈e〉). These elements combine by Functional Application to give a proposition (〈t〉), which can be sent off as well. Continuing the calculation, the verbal predicate associates with its arguments, the subject and object, bringing about an expression of type 〈t〉. Finally, the nonrestrictive modifier is associated with the main proposition (both are of type 〈t〉) in some kind of conjoined expression. Adopting de Vries’ general system, this coordination structure could, at some point in the derivation, look as follows, where IP stands for the hosting proposition and XP for the non-restrictive modifier (&: = ‘that is’).19

19.  The current proposal seems to be, at least in principle, compatible with the analysis in de Vries (2006). The basic differences are that de Vries coordinates the antecedent and the non-restrictive relative clause at the DP-level (hence the relative is presumably of type 〈e〉) and, for him, there is no detachment after syntax. While de Vries’ account is very successful, I believe the text proposal also has some advantages:  (i) the current account unifies the analyses of non-restrictive modifiers; that is, relative clauses and adjectives are basically treated the same way; (ii) if this is correct, then non-restrictive pre-nominal adjectives pose a problem to de Vries’ account (see especially his remarks on page 264) but not to this proposal as the non-restrictive modifier will follow the antecedent after detachment; (iii) coordination at a higher level also accounts straightforwardly for lack of c-command effects (e.g., ungrammatical bound pronouns).

 Demonstratives and definite articles as nominal auxiliaries

(36)

CoP IP〈t〉

Co′ &:

XP〈t〉

To conclude, interpreting the lower determiner allows for the derivation of the non-restrictive reading. If the lower determiner copy were deleted and the upper one interpreted (as with restrictive modifiers), the semantic calculation would face a problem. Assuming as above that the derivation proceeds bottom-up, the nonrestrictive modifier (〈t〉) intervenes between the nominal predicate (〈e,t〉) and the determiner. As this nominal predicate cannot combine with the non-restrictive modifier, the calculation cannot proceed and the expression is not interpretable. Crucially, in order to guarantee the co-indexation between pro inside the nonrestrictive modifier and the antecedent DP (cf. (29b) and (32a)), I assume that the antecedent of the non-restrictive modifier (i.e., the man) must be calculated before the modifier (i.e., pro old) is sent off. 3.2  Syntactic consequences: more on Scandinavian relative clauses Above I proposed that non-restrictive adjectives in Icelandic have two analyses, a syntactic one and a non-integrated one. I pointed out that this claim was not novel and has already been made for certain kinds of appositive relative clauses in Italian (Cinque 1981). Interestingly, Swedish has two different surface patterns for nonrestrictive relative clauses. To the best of my knowledge, all speakers accept the “Icelandic” pattern in (37a) (data from Kester 1996b: 118; cf. also Börjars 1998: 53; Julien 2005:  79). Platzack (1997:  76, 2000:  279) also discusses structures of the “Double Definiteness” type (37b) (cf. also Santelmann 1993: 158): (37) a.

Jag såg lampan, som Anders förresten köpte i Umeå. I saw lamp-the that Anders, by the way, bought in Umeå ‘I saw the lamp, which, by the way, Anders bought in Umeå’

(Swedish)

b. Det hus-et, som han för övrigt ville riva, är the house-the, that he by the way wanted.to demolish, is

nu till salu. now for sale.



‘the house, which, by the way, he wanted to demolish, is now for sale’

In contrast to restrictive relative clauses, the “Danish” pattern of the antecedent (*den lampa ‘the lamp’) is not possible. This follows from the observation that



The syntax and semantics of non-/restrictive modifiers 

non-restrictive relative clauses have specific antecedents (cf. footnote 18 above) and suffixal determiners in the languages with the Double Definiteness effect can bring about specific interpretations (Chapter 2). In other words, a non-restrictive relative clause has to co-occur with a suffixal determiner on the head noun in Swedish. With the presence of the suffixal determiner due to the semantics, (37a) could be taken to be ambiguous between a non-integrated and a syntactic relative (where the latter is adjoined to ArtP). However, with the determiner split up in (37b), this example, under my assumptions, must be a syntactic non-restrictive relative clause, adjoined to AgrP.20 To sum up, besides the common pattern, noun phrases in Icelandic with adjectives have a literary pattern and a third one. The latter two have adjectives that are non-restrictive in interpretation. Making a distinction between syntactic and non-integrated appositives, I suggested that non-restrictive modifiers are adjoined at a (“late”) propositional level and that the lower copy of the determiner is interpreted in the syntactic type (leaving the analysis of the non-integrated one open). Assuming multiple semantic spell-out, this derived the non-restrictive reading. If the surface position of the free-standing determiner in Icelandic and the position where this determiner is interpreted are related by movement, then this derivation presents another argument for the hypothesis that determiners move to the DP. Finally, the discussion was extended to relative clauses, showing that Swedish has syntactic appositives.

4.  Extension to Romance adjectives As is well-known, the Romance languages allow pre-nominal adjectives under certain conditions (for different types of adjectives, see Bernstein 1993: chap. 2; Bouchard 1998, 2002; Ticio 1993: 113–128). Simplifying somewhat, while French has some obligatorily pre-nominal adjectives (e.g., joli ‘beautiful’), which are lexically specified, French, Italian, and Spanish generally allow pre-nominal adjectives if they are non-restrictive in interpretation. Giorgi & Longobardi (1991: 123) point

20.  A reviewer points out to me that the antecedent in (37b) (i.e., det hus-et) has a deictic flavor and the free-standing determiner should probably be interpreted as a demonstrative (and not an article). If so, then the relative clause does, at least in the dialect of the reviewer, not necessarily induce a separate agreement domain. Consequently, we cannot make the claim of a syntactic non-restrictive relative clause for that dialect.

 Demonstratives and definite articles as nominal auxiliaries

out that, while the post-nominal adjective is ambiguous in its interpretation, the pre-nominal is only appositive (also Martín 1995: 188; Ticio 2003: 122–4):21 (38) a.

una ragazza simpatica a girl nice ‘a girl that is nice’ ‘a girl, who is nice’

(Italian)

b. una simpatica ragazza a nice girl ‘a girl, who is nice’

Above, I argued against an analysis that involves movement of the adjective to Spec,DP or out of the DP altogether. In fact, on my account, adjectives are interpreted in situ. The question, then, arises as to how to capture the ambiguity in (38a) and the lack of it in (38b). Kester (1996b: 60) points out that, independently of the position, all adjectives exhibit the same agreement characteristics. In fact, in the few cases where adjectives do not agree, they can only appear post-nominally (Zamparelli 1993: 156–7, 2000: 262). Thus, I assume that non-restrictive adjectives in the Romance languages are part of syntax (unlike the third construction in Icelandic). Assuming partial nominal-raising in syntax (Chapter 1) followed by reconstruction of the nominal to its base-position in LF (Anderson 2002:  122), the ambiguity in (38a) follows straightforwardly from my proposal: if the interpreted copy of the determiner is higher than the modifier, the interpretation is restrictive; if it is lower, the reading is non-restrictive. Compare the different LF representations in (39b) and (39c): (39) a.

una ragazza simpatica a girl nice ‘a girl that is nice’ ‘a girl, who is nice’

b. [DP una [AgrP simpatica [ArtP una [NP ragazza]]]] c.

[DP una [AgrP simpatica [ArtP una [NP ragazza]]]]

21.  Similar observations have been made for Chinese (Zhang 2004), Japanese (Whitman 1981; Noguchi 1997:  789; Yoshihisa Kitagawa, p.c.), and basically Korean (Kim 1997). It is not clear to me, however, if these non-restrictive modifiers are syntactic or parenthetic in nature. Furthermore, in the main text, I discuss adjective readings in noun phrases with non-theta nouns (e.g., girl). Crisma (1996) points out that with theta-nouns (e.g., reaction), the interpretation of adjectives is similar to that of adverbs in the clause (speaker-, subject-, manner-oriented). Importantly, as Crisma (1996: 66) notes, the pre-nominal adjective is ambiguous but the postnominal one is not. We arrive, then, at opposite judgments vis-à-vis non-theta nouns, discussed in the text. Consequently, these two cases should not be collapsed and should receive a different account (note in passing that this is reminiscent of differences in interpretation of possessives with theta vs. non-theta nouns, see Roehrs 2005b).



The syntax and semantics of non-/restrictive modifiers 

So far, the discussion seems to be straightforward. Now, in order to rule out the restrictive reading in (38b), let us slightly modify the assumptions about the non-restrictive cases. I propose that, like restrictive adjectives, non-restrictive modifiers are also part of syntax (induction of separate agreement domain, same agreement) and I assume that they are also in a Specifier position. Unlike the former, the latter are in – what I will call for clarity’s sake – Spec,FPAGR. Although the distribution of FPAGR is taken to be relatively unconstrained syntactically, which accounts for the different positions of the non-restrictive modifier, we will see that it is subject to restrictions brought about by our semantic assumptions. Returning to the original paradigm, repeating (38a–b) as (40a), the determiner moves in a stepwise fashion to the DP (not shown in (40b)), either through the relevant head positions if it is an article, or by adjunction to phrases if it is a demonstrative (for more details, see Chapter 4 Section 9). Assume that the nominal raises to adjoin to AgrP, the phrase in which garden-variety adjectives are located. With Section 3.1 in mind, let us propose that FPAGR can project on top of any phrase in the noun phrase. This includes ArtP and AgrP and, unlike AgrP, allows FPAGR to occur in several locations:22 (40) a.

una (simpatica) ragazza (simpatica) a    nice girl    nice ‘a girl that is nice’ ‘a girl, who is nice’

b.

(Italian)

DP D unai

FPAGR

simpatica〈t〉

AgrP

ragazzak

AgrP

simpatica〈e,t〉

FPAGR

simpatica〈t〉 Art unai

ArtP NP ragazzak

22.  Unlike Intensifier Phrases (see Chapter 4), FPAGR cannot project on top of the DP. Furthermore, it cannot be on top of NP, which, as we will see momentarily, would lead to type mismatch.

 Demonstratives and definite articles as nominal auxiliaries

Recalling from Section 3 that non-restrictive modifiers are of type 〈t〉, the semantic calculation will only succeed if these modifiers are higher than an atomic semantic element (〈e〉), which consists of a determiner, a noun, and possibly a (restrictive) adjective. If not, this will result in type mismatch and the structure cannot be interpreted (recall again that the nominal syntactically adjoins to AgrP but is interpreted in its base-position). Thus, the relatively free syntactic occurrence of FPAGR combined with our semantic account explains the different distribution of non-restrictive adjectives. In other words, the post-nominal adjective can be ambiguous but the pre-nominal one can not.23 Besides further distributional evidence for this claim, we will see momentarily that French also shows that the determiner moves to the DP in a successive-cyclic fashion. Lexically specified adjectives (e.g., joli ‘beautiful’) as well as pre-nominal non-restrictive adjectives (e.g., lourd ‘heavy’) precede the partially raised nominal in French. Importantly, there is an ordering restriction between both types of adjectives, as shown in (41a–b) (cf. also Zamparelli 1993: 142–3). Besides the general distribution in (41b), the example in (41c) shows an additional postnominal adjective and provides evidence that the nominal has indeed raised: (41) a. ??la jolie lourde valise the beautiful heavy suitcase

(French)

b. la lourde jolie valise the heavy beautiful suitcase ‘the beautiful suitcase, which is heavy’ c.

pauvre petite fille riche poor little girl rich ‘pitiable little rich girl’ (from the movie Titanic)

Let us assume that high pre-nominal adjectives of the joli-type, which are lexically specified, are in a Specifier position of a phrase I call AgrP2 (which is higher than garden-variety AgrP). The markedness of (41a) follows from the current account as syntactic non-restrictive modifiers cannot be below restrictive ones.24

23.  Bernstein (1993), Crisma (1996), and Anderson (2002) report that the post-nominal adjective is only restrictive. To the extent that this is true, there seem to be two dialects, the one discussed in the main text and the one mentioned in this footnote. In order to derive the lack of ambiguity in the latter, I propose that in this dialect, non-restrictive adjectives can only project on top of AgrP (but not ArtP). 24.  Note that the reason why the ungrammaticality of the example in (41a) is not worse has presumably to do with different adjective sequences that are, under certain conditions, possible (see also footnote 26 below).



The syntax and semantics of non-/restrictive modifiers 

To be clear, then, while non-restrictive modifiers have, at least in principle, a freer syntactic distribution than other modifiers, their semantics constrains their occurrence. Note also that (41b) shows a non-restrictive adjective preceding a restrictive one. In order to derive the differences in interpretation between the two adjectives, the determiner must be interpreted below the non-restrictive but above the restrictive modifier. I assume it is located in FAGR: (42) [DP lai+D [FPagr lourde [Fagr′ lai [AgrP2 jolie [ArtP lai [NP valise]]]]]]

If this is so, then we have evidence that the determiner must have moved through an intermediate phrase, that is, in a successive-cyclic fashion. Finally, lacking nominal-raising of the Romance type, this distributional difference of adjectives with various interpretations is not visible in the Germanic languages.

5.  Some further issues In this section, I turn to some final points. 5.1  Summary of the syntax of modifiers In the preceding sections, I proposed that restrictive adjectives are in Spec,AgrP and non-restrictive adjectives are in Spec,FPAGR, which is more freely distributed. Furthermore, I suggested that restrictive relative clauses are right-adjoined to AgrP and ArtP and “syntactic” appositive relative clauses are adjoined to AgrP (Platzack and Santelmann’s examples) and, perhaps, also to ArtP. Table 2.  Structural Status of the Different Modifiers.

restrictive non-restrictive

Adjective

Relative clause

Spec,AgrP Spec,FPAGR

adjunct adjunct

Note first that the restrictive adjective is the least complex element of them all (it does not co-occur with a pronoun of any type). This observation invites the speculation that non-restrictive adjectives might in fact be a reduced version of non-restrictive relative clauses.25 Furthermore, I have suggested for the above cases that non-restrictive adjectives and restrictive relative clauses are part of syntax. Having proposed that they are less constrained in their syntactic distribution,

25.  Fanselow (1986) argues that all pre-nominal adjectives in German are reduced relative clauses. I will not pursue this general line of thought here further.

 Demonstratives and definite articles as nominal auxiliaries

we can relate the different positions of these adjectives in Romance to the diverse syntactic distribution of the determiner with these relative clauses in the Scandinavian languages. Finally, if it is correct that these non-restrictive adjectives and restrictive relative clauses are part of syntax, then due to our type semantics, the former cannot be dominated by restrictive adjectives, as discussed in Section 4, and, conversely, the latter cannot dominate syntactic non-restrictive relative clauses. Put more generally, non-restrictive modifiers must be “higher” than restrictive ones, as already noted by Jackendoff (1977: 171).26 Below, I summarize the different relative clause patterns and their corresponding interpretations. 5.2  Some differences and idiosyncracies Recall that, besides the “Icelandic” and the “Danish” pattern, Swedish, Faroese, and Norwegian also exhibit the “Double Definiteness” form. Furthermore, all non-restrictive clauses lack the “Danish” pattern: Table 3.  Schematic Summary of the Different Relative Clauses.

Sw, Nw, Faroese Danish

Restrictive

Non-restrictive

N-en som … den N som… den N-en som… N-en der … (younger speakers) den N der …

N-en, som … den N-en, som…(Sw: Platzack, Santelmann) N-en, der…

As mentioned above, the lack of the “Double Definiteness” pattern in Danish follows from the assumption that, for independent reasons, this language does not allow determiner splitting. Furthermore, I assume that the lack of the “Danish”

26.  As Leko (1988) extensively discusses, this state of affairs even holds in the noun phrases of free-word order languages like Serbo-Croatian. If this is correct, then “violations” of this state of affairs can only involve the combination of syntactic and non-integrated types of modifiers. A case in point might be the German example in (ia), where the typical adjective ordering can be violated by a non-restrictive, non-integrated adjective. Compare (ia) to (ib): (i) a.

das grüne, übrigens kleine, Auto the green, incidentally small, car ‘the green car, which, incidentally, is small’

(German)

b. ??das grüne kleine Auto the green small car As non-integrated modifiers are not the focus of this chapter, I will not investigate these cases further here.



The syntax and semantics of non-/restrictive modifiers 

pattern with non-restrictive relative clauses (column three in table 3) is due to different reasons: as discussed in Chapter 2, the suffixal ending in Swedish brings about a specific reading. Since the antecedent of a non-restrictive modifier must be specific, the lack of the “Danish” form follows from the general correlation between the distribution of the determiner and its corresponding semantics. As for this lack in Danish, one could assume that its non-restrictive relative clauses are either nonintegrated or, if syntactic, they can only adjoin to ArtP (consequently, they will not intervene and the determiner will not move to the DP). I turn to some other issues. In the last three sections, I argued for the “free” interpretation of the determiner in its base and its derived position, where the former results in the non-restrictive reading and the latter in the restrictive one. While this seems to be fairly elegant, there is an issue with the restrictive “Icelandic” pattern in Danish in that it is only possible for younger speakers. Another issue involves the non-restrictive “Double Definiteness” form in Swedish, which has only been reported by Platzack and Santelmann (as far as I know; see also footnote 20). While these judgments show that the determiner can, in principle, be interpreted in both positions, there seem to be some idiosyncracies related to the syntactic adjunction sites of these relative clauses. 5.3  Some restrictions on the “free” interpretation of the determiner As discussed in Section 2.2, the “Danish” pattern of the restrictive relative clauses in Danish (cf. (20)) only has “surface” scope, which we derived by the head raising analysis and a ban on movement of IP. In contrast, adjectival modifiers in common and literary Icelandic, as reported, only allow “inverse” scope. Repeating (21) here for convenience, the suffixal determiner takes wide scope and the free-standing one narrow scope:27 (43) a.

guli bíllinn yellow(weak) car-the ‘the car that is yellow’

b. hinn guli bíll the yellow(weak) car ‘the car, which is yellow’

(common Icelandic)

(literary Icelandic)

27.  Matushanksy (2006: 104) discusses a similar phenomenon in the clause, where certain modals behave differently. To capture the scopal relations, it is suggested that, after movement of the modal from below negation, can must reconstruct in (ia) but should cannot in (ib): (i)

a. b.

Yolanda can’t leave. Neg > Mod Yolanda shouldn’t leave. Mod > Neg

If determiners are nominal auxiliaries, it is not unexpected that there is such parallelism.

 Demonstratives and definite articles as nominal auxiliaries

I believe that there is an interesting lexical account for this restriction. In the discussion of restrictive modifiers, I proposed to interpret the upper determiner. However, I did not comment much on the nature of the copy in the base position. I basically treated it as if it were deleted, here illustrated by “Ø”: (44)

DP〈e〉 det〈〈e,t〉,e〉

(Functional Application) AgrP〈e,t〉

adjective〈e,t〉

(Predicate Modification)

ArtP〈e,t〉

Ø

NP〈e,t〉 N〈e,t〉

However, if we assume that the copy is not deleted but that something is left behind, we can get a handle on the restriction noted above. Suppose that a trace or copy is left behind, which can combine with the head noun by Predicate Modification. Consider the lower part of the structure in (45). This is followed by combining the adjective with the ArtP node in the usual way. Predicate Abstraction (for details, see Heim & Kratzer 1998) gives an expression of type 〈〈e,t〉,〈e,t〉〉. In order to apply Functional Application in the usual way, the determiner must be of a different type than the traditional 〈〈e,t〉,e〉. In fact, α in (45) must be of type 〈〈〈e,t〉,〈e,t〉〉,e〉:28 DP〈e〉

(45) det〈α〉

(Functional Application) (Predicate Abstraction)

〈〈e,t〉,〈e,t〉〉

1

(Predicate Modification)

AgrP〈e,t〉 adjective〈e,t〉

ArtP〈e,t〉

t1〈e,t〉

(Predicate Modification)

NP〈e,t〉 N〈e,t〉

28.  In the framework of Heim & Kratzer (1998: 184–8), (45) is the semantic representation of the more familiar (but abbreviated) syntactic one [det1 [adjective [t1 [noun]]]], where the index, the binder of the variable (i.e., the trace), is directly on the moved element.



The syntax and semantics of non-/restrictive modifiers 

Let me repeat the traditional definition of the determiner, call it the1 (taken from Heim & Kratzer 1998: 81), and define a new lexical entry for the2 (thanks to Sigrid Beck for help with this definition). Compare (46) to (47) (the relevant differences are underlined): (46) [[the1]] = λf: f ∈ D〈e,t〉 and there is exactly one x ∈ C such that f(x) = 1. the unique y ∈ C such that f(y) = 1, where C is a contextually salient subset of D. (47) [[the2]] = λ f: f ∈ D〈〈e,t〉, 〈e,t〉〉 and there is exactly one x ∈ C such that f(λy.y=y)(x) = 1. the unique y ∈ C such that f(y) = 1, where C is a contextually salient subset of D.

The functions f in both (46) and (47) are the arguments of the1 and the2, respectively. The difference between them is that in the former case f is a property (〈e,t〉) and in the latter it is a function of a higher order (〈〈e,t〉,〈e,t〉〉). As underlined in (47), part of that new function is an added property (〈e,t〉) that maps an individual to an identity statement. As such, this function does not add any crucial semantics to the derivation. With this in mind, assume that the lower copy is not deleted but that something is left behind when the determiner moves from its base-position up. I propose that when the1 moves, it has to reconstruct, closing off the DP at a lower structural level. Under this assumption, the reconstructed determiner directly combines with the head noun by Functional Application. In other words, the determiner must be interpreted in its base position and cannot take scope over an adjective. This is different for the2: as neither the determiner (〈〈〈e,t〉,〈e,t〉〉,e〉) nor the head noun (〈e,t〉) can directly combine with the correspondingly other expression by Predicate Modification or Functional Application, the determiner has to move at LF to resolve this type mismatch. It leaves behind a copy of type 〈e,t〉, which combines with the head noun by Predicate Modification. The adjective, if present, combines with the latter expression by Predicate Modification. This is followed by Predicate Abstraction and Functional Application, as indicated in (45). Returning to the account of the “inverse” scope in Modern Icelandic, I propose that free-standing hinn is of the traditional type the1 and suffixal -inn is of type the2. Although hinn moves to D overtly, it must be interpreted in its base position. This derives the non-restrictive reading of the adjective. Conversely, -inn is in its base-position in syntax but moves to the DP covertly (recall that Merger occurs after syntax). This derives the restrictive reading of the adjective. If this lexical account is correct, then we might expect that there are some languages where the “restrictive” determiner differs morphologically from the “non-restrictive” one. Interestingly, Spencer (1992: 324) points out that, in addition to the word-initial difference, -inn and hinn are not inflectionally identical

 Demonstratives and definite articles as nominal auxiliaries

in the feminine plural nominative:  -nir vs. hinar. Another case might be West Jutlandic. Delsing (1993b: 132 footnote 25) states that this language uses a different determiner in these two cases, de for restrictive and æ for non-restrictive adjectives (cf. Schroeder 2006: 562). 6.  Conclusion Following others, the basic proposal of this chapter is that determiners are scope-bearing elements. If modifiers are in their scope, they are restrictive in interpretation; if not, then they are non-restrictive in reading: DP

(48)

XP

D determiner

relative clause

XP adjective Art determiner

ArtP NP noun

With regard to the main proposal of this work, assuming movement of the determiner from ArtP to DP, the restrictive interpretation is accounted for by interpreting the determiner in its derived position and the non-restrictive interpretation is accounted for by interpreting the determiner in its base-position. As both the derived as well as the base-positions are independently motivated (Chapter 2), this semantic account comes at very little cost and provides a straightforward solution to Chomsky’s objection to Partee’s proposal. To highlight some of the results again, I made a distinction between syntactic and non-integrated appositives. Arguing that non-restrictive modifiers are propositions that are associated with their hosting proposition by conjunction, I suggested that the analysis of the syntactic type involves multiple semantic spell-out. The discussion was extended in an interesting way to different kinds of relative clauses in the Scandinavian languages and to adjective interpretations in Romance. Finally, I provided some evidence from Icelandic that the interpretation of the determiner may not be entirely “free”. If this analysis is accepted, then we have an argument that not all cases of head movement can take place in PF. For the same conclusion, see Longobardi (1994), Zwart (2001), and most recently Matushanksy (2006), Fitzpatrick (2006), and den Dikken (2006).

chapter 4

The strong and weak alternation in German 1. Introduction Chapters 2 and 3 dealt with the syntactic distribution of determiners in the Scandinavian languages and the different interpretations of modifiers derived from it. To briefly review some of this discussion, Chapter 2 provided syntactic evidence for the movement of determiners from the Scandinavian and some other languages. In particular, we discussed cases where movement seems to occur successive-cyclically: (1) a.

en stor en ful en kar a big an ugly a man ‘a big ugly man’

(Northern Swedish)

b. ?eit stort eit stygt eit hus a big an ugly a house ‘a big ugly house’

(Norwegian)

Chapter 3 provided evidence for the movement of determiners from the well-known observation that attributive adjectives can be both restrictive and non-restrictive in interpretation. In order to account for this, I proposed that copies of the determiner may be interpreted in the original and derived positions. Compare the translations and respective derivations in (2b) and (2c): (2) a.

una ragazza simpatica a girl nice

(Italian)

b. ‘a girl, who is nice’ [DP unai+D [FPagr simpatica [ArtP unai [NP ragazza]]]] c.

‘a girl that is nice’ [DP unai+D [AgrP simpatica [ArtP unai [NP ragazza]]]]

In fact, adjacent non-restrictive and restrictive modifiers also provide evidence that the determiner moves in a successive-cyclic fashion: (3) a.

la lourde jolie valise the heavy beautiful suitcase ‘the beautiful suitcase, which is heavy’

b. [DP lai+D [FPagr lourde [Fagr′ lai [AgrP2 jolie [ArtP lai [NP valise]]]]]]

(French)

 Demonstratives and definite articles as nominal auxiliaries

To the syntactic and semantic discussion from Chapters 2 and 3, we now add morpho-syntactic evidence that the determiner moves to the DP. In particular, turning to some language-specific consequences, I raise the question of how much light the main proposal of this work can shed on phenomena related to the German noun phrase. I argue that the proposal that determiners move into the DP also explains the distribution of the strong and weak adjectival inflections in German. Like in the other chapters, this analysis also shows that this movement must occur successive-cyclically. To the extent that this analysis is successful, this agreement phenomenon will provide an overt morphological reflex of the current proposal. To anticipate the discussion, I will propose that determiners reduce the feature content of adjectives in a local fashion (by Percolation and Impoverishment). If so, all underlyingly strong inflections on adjectives, as in (4a), will be spelled out by impoverished, that is, weak endings, as in (4b). This derivation can be schematically represented in (4c): (4) a.

heißer schwarzer Kaffee hot(nom.strong) black(nom.strong) coffee(masc) ‘hot black coffee’

b. der heiße schwarze Kaffee the(nom.strong) hot(weak) black(weak) coffee(masc) ‘the hot black coffee’ c. DP de-ri

AgrP Agr′

heiße-(r) de-ri

AgrP

schwarze-(r)

Agr′

de-ri

ArtP de-ri

NP Kaffee

While the following discussion focuses on the morpho-syntax of German, let us briefly point out that there are also some important differences between the



The strong and weak alternation in German 

Scandinavian languages and German. For instance, while a modifier triggers the presence of a free-standing determiner in the former, here illustrated again by Norwegian, the presence of a modifier has no morpho-syntactic effect on the determiner in German: (5) a.

(den gamle) mann-en the old man-the ‘the old man’

b. ein (alter) Mann the(-) old(strong) man ‘the old man’

(Norwegian)

(German)

In order to capture this cross-linguistic difference, I propose that morphological number is valued in different locations in these languages. Specifically, while the Scandinavian languages and German share that Definiteness and Case are valued in the DP and Gender in the NP, they differ in that number is valued in NumP in the Scandinavian languages but in CardP in German:1 (6) D Card ArtP Num N a. Scand. lang.: Def, Case (+) # Gend b. German: Def, Case # + Gend

Note now that this difference has a simple but important consequence for German. Recall from Chapter 2 that a separate agreement domain may, under certain conditions, be established with regard to ArtP. Unlike in the Scandinavian languages, where the presence of a modifier triggers the presence of a semi-functional head (e.g., Card or Agr), in German CardP is always present for feature valuing of number. Consequently, unlike in the Scandinavian languages, ArtP always forms a separate agreement domain in German, marked by the plus sign under ArtP in (6). Among other facts, this will explain the difference in (5). Returning to the main focus, this chapter identifies effects of determiner movement in German, which are argued to manifest themselves in the strong and weak alternation of adjectival inflections. The chapter is organized as follows: in the first part (Sections 2–5), I discuss the inventory of the nominal inflections with the intention of reducing the relevant five paradigms to two. While the proposal of two basic paradigms is certainly not novel, the particular current assumptions and resultant system are. For instance, reinterpreting the endings in schwa, the number of exponents will be reduced

1.  As it is well-known that there is a close relation between semantic properties and morphological case (e.g., Diesing 1992), I put Definiteness and Case in the same phrase.

 Demonstratives and definite articles as nominal auxiliaries

from five strong and two weak endings to four strong and one weak ending. In the second part of this chapter (Sections 6–10), I discuss the distribution of the endings in more detail. Assuming that determiners move to the DP, I propose to derive the strong and weak alternation by the mechanism of Impoverishment.2

2.  The strong and weak inflections: The traditional endings German is a language with rich nominal inflection, where noun phrases may distinguish case, number, and gender by way of exhibiting different endings. In particular, determiners and adjectives may display morphological inflections that belong to two sets of endings, traditionally called “strong” and “weak”.3 Depending on the structure of the noun phrase, the strong ending may attach either to the determiner or to the adjective but not both. Traditional grammars usually discuss five different paradigms4 where these strong and weak inflections may appear (e.g., Duden 1995: 277–9, 306–7): illustrating with nominals in the nominative and the dative, (7) shows examples from the paradigms of the definite article and the weak adjective, where the article has a strong ending and the adjective a weak inflection: (7) a.

der gute Wein the(nom.strong) good(weak) wine(masc) ‘the good wine’

b. dem guten Wein the(dat.strong) good(weak) wine(masc) ‘the good wine’

2.  The reader who is not interested in all the intricacies of the nominal inflectional system in German may want to skip sections 3, 4, and 10.1–10.3. 3.  The terms “strong” and “weak” go back to Jacob Grimm (1870: 718–756). Other terms found in the literature are “pronominal” vs. “nominal”, “determining” vs. “determined”, and “primary” vs. “secondary”. While I will use the traditional terminology here, it should not be confused with Chomsky’s (1995) strong and weak features. Note that the strong and weak alternation manifests itself differently in the various Germanic languages (see Harbert 2007:  p. 130 for concise discussion). 4.  For the time being, I use the traditional notion of paradigms. In section 10.4, I reinterpret the inflections in these paradigms as underspecified abstract feature bundles, which are later realized by certain overt exponents.



The strong and weak alternation in German 

The data in (8) exemplify cases from the third paradigm, which involves strong adjectives: (8) a.

guter Wein good(nom.strong) wine(masc) ‘good wine’

b. gutem Wein good(dat.strong) wine(masc) ‘good wine’

And, finally, the patterns in (9) provide cases from the paradigms of the indefinite article and the “mixed” adjective, where the adjective has a strong ending in some cases but a weak one in others: (9) a.

ein guter Wein a(-) good(nom.strong) wine(masc) ‘a good wine’

b. einem guten Wein a(dat.strong) good(weak) wine(masc) ‘a good wine’

Let us summarize these five paradigms and their inter-relatedness in schematic terms: (10) a.

Definite Article + Weak Adjective [definite determiner(strong) adjective(weak) noun]

b. Strong Adjective [adjective(strong) noun] c.

Indefinite article + “Mixed” Adjective [indefinite determiner(-/strong) adjective(strong/weak) noun]

These are the basic cases. In the discussion to follow, we will see that each of the general cases in (10) has some specific instantiations that, in some way or other, exhibit some exceptional behavior. Besides explaining the general patterns in (10), this chapter also focuses on these exceptional cases. Let us first take a closer look at the different exponents of these paradigms. It is well-known that determiners and adjectives in German exhibit two basic types of endings. In particular, comparing tables 1 and 2, one can see that the strong adjective has the same endings as the definite determiner dieser ‘this’ (as well as jeder ‘every’, jener ‘that’, and others). As marked in italics in table 2, there

 Demonstratives and definite articles as nominal auxiliaries

is one exception, where the endings in the genitive masculine and neuter are not identical (for detailed discussion, see Section 9.2):5 Table 1.  Inflections on the Demonstrative dies- ‘this’. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

dies-e-r dies-e-n dies-e-m dies-e-s

dies-e-s dies-e-s dies-e-m dies-e-s

dies-e dies-e dies-e-r dies-e-r

dies-e dies-e dies-e-n dies-e-r

Table 2.  Strong Inflections on the Adjective gut- ‘good’. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

gut-e-r gut-e-n gut-e-m gut-e-n

gut-e-s gut-e-s gut-e-m gut-e-n

gut-e gut-e gut-e-r gut-e-r

gut-e gut-e gut-e-n gut-e-r

Looking at tables 1 and 2 in more detail, we can distinguish five different strong endings:  -r, -s, -m, -n, -e. In contrast, the weak set of endings in table 3 below shows only two different endings: -n, -e. In other words, the strong set has three more inflections than the weak one and it is thus better able to disambiguously mark case, number, and gender of the DP. Note also that the two sets of endings in tables 1 and 3 differ in twelve (out of sixteen) cells. This is marked in bold print in table 3: Table 3.  Weak Inflections on the Adjective gut- ‘good’. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

gut-e gut-e-n gut-e-n gut-e-n

gut-e gut-e gut-e-n gut-e-n

gut-e gut-e gut-e-n gut-e-n

gut-e-n gut-e-n gut-e-n gut-e-n

These are the basic sets of endings (for the discussion of the indefinite article and the “mixed” adjective, see section 3.2). To summarize, the above discussion illustrates the following well-known generalization:

5.  I argue in that section that the “strong” adjective in the genitive masculine/neuter has indeed a weak ending (cf. table 3 below).



The strong and weak alternation in German 

(11) (Non-)Identity a. The ending on the definite determiner = the strong ending on the adjective. b. The strong ending on the adjective ≠ the weak ending on the adjective.

In what follows, I will refer to the endings on the definite determiner and the strong adjective simply as ‘strong endings/inflections’ and to those of the weak adjective as ‘weak endings/inflections’. Note that the weak endings form a proper subset of the strong inflections: (12) a.

Strong Endings -r, -s, -m, -n, -e

b. Weak Endings -n, -e

Having collapsed the paradigm of the definite determiner and that of the strong adjective, we are left with four: the strong, weak, and “mixed” adjectival paradigms and the one of the indefinite article.

3.  The problem: Determiners with different endings From the perspective of Section 2, I discuss in this section two, at first glance, unrelated exceptions. However, I will argue that both problems are not only related but also just apparent. 3.1  The definite article The definite article in German differs from the determiners discussed in Section 2 in that the final -e on 〈di-e〉 is not pronounced as schwa, as implied by the segmentation in table 4 below, but 〈die〉 is actually pronounced [di:]. As such, it is different from the strong endings (cf. table 1 above), where this difference is marked in bold print in table 4 below: Table 4.  Inflections on the Definite Article d- ‘the’. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

de-r de-n de-m de-s

da-s da-s de-m de-s

di-e di-e de-r de-r

di-e di-e de-n de-r

Strictly speaking, these four endings of the definite article are exceptional and the traditional segmentation is misleading. While this difference is usually

 Demonstratives and definite articles as nominal auxiliaries

abstracted away from in traditional accounts, I will take it seriously here with the intention to explain it. As a first step, I propose that these instances of the definite article have a strong null ending:6 Table 5.  Inflections on the Definite Article d- ‘the’. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

de-r de-n de-m de-s

da-s da-s de-m de-s

die-Ø die-Ø de-r de-r

die-Ø die-Ø de-n de-r

Interpreted this way, the -e in 〈di-e〉 is now taken to be a spelling convention, indicating long [i:] just as in words like sieben ‘seven’. This discussion can now be extended to personal pronouns of the third person. If Postal (1966) is right in that personal pronouns are determiners (see also Roehrs 2005a, 2006b and references cited therein), then we can update the relevant segmentations as follows:7 Table 6.  Inflections on the Third-person Personal Pronouns. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

e-r ih-n ih-m seiner

e-s e-s ih-m seiner

sie-Ø sie-Ø ih-r ihrer

sie-Ø sie-Ø ih-n+en ihrer

In Section 4, I extend this claim involving null endings to the other definite determiners from Section 2 and adjectives in general.8 The result will be that all these determiners and adjectives will have identical strong inflections without exception. Before we embark on this part of the proposal, let us relate this discussion to the indefinite article.

6.  In section 5, I follow Wiese (1988) in reinterpreting the instances of die in table 5 as elements that have no ending at all. 7.  Note that pronouns in the genitive exhibit “frozen”, archaic forms given in bold print in table 6 and that the dative plural has the additional ending +en (for discussion of the latter, see Lühr 1991). 8.  In particular, I propose in section 4 that all schwas, including the ones in word-final position, are due to an insertion rule.



The strong and weak alternation in German 

3.2  The indefinite article With the above discussion in mind, consider (13). It is interesting to note that – what is sometimes called – ‘ein-words’ (i.e., indefinite determiners, negative articles, the singularity numeral, and possessive pronouns) cannot have an ending.9 This holds in three instances: the nominative masculine and nominative/accusative neuter. Most scholarly work claims that these cases are indeclinable and treats them as exceptions:10 (13) a.

Das ist kein Wein. this is no wine ‘This is no wine.’

b. Das ist kein guter Wein. this is no good(strong) wine ‘This is not a good wine.’

Crucially, though, these accounts do usually not take discontinuous noun phrases, often called ‘split-NPs’, into consideration (for proposals of the latter, see Fanselow 1988; van Riemsdijk 1989; more recently, Roehrs 2007a). Illustrating here with the negative article, there is an interesting divergence from the pattern in (13) when kein is not followed by another overt nominal element. In this case, the ein-word has a strong ending, as shown in (14b): (14) a.

Wein ist das kein guter! wine is this no good(strong) ‘As for wine, this is no good one.’

b. Guter Wein ist das kein*(er)! good(strong) wine is this none(strong) ‘As for good wine, this is not one.’

To the extent that I am aware, there is no independent evidence that the ein-words in (14) are different from those in (13) and the simplest assumption is that these

9.  In Section 8, I return to these different ein-words that are morphologically identical despite their different semantics. I will suggest that they are composite elements consisting of a vacuous ein, which explains the same morphology, and another element, which accounts for the different semantics. 10.  There is actually debate over whether the ein-words in these cases have no ending at all (e.g., Bhatt 1990:  199; Demske 2001:  40; Eisenberg 1999:  233; Olsen 1991:  47 footnote 14) or a (weak) null ending (e.g., Darski 1979:  198; Duden 1995:  307; Roehrs 2006a; Wegener 1995: 98). In Section 7, I propose that ein starts out with a “strong” inflectional feature bundle. However, this feature set cannot be valued during the derivation and, consequently, is not spelled out as an ending.

 Demonstratives and definite articles as nominal auxiliaries

elements are based on the same lexical element (cf. footnote 27). Now, considering the emergence of an ending on kein in (14b), I submit that the cases of inflectionless kein, (13a–b) and (14a), should be taken seriously and not be treated as indeclinable exceptions. This conjecture is of particular interest in view of truly indeclinable elements such as the numeral zehn ‘ten’ that can license split-NPs without any inflection whatsoever. Consider (15a–b). (15) a.

Ich habe zehn Hemden. I have ten shirts ‘I have ten shirts.’

b. Hemden habe ich zehn. shirts have I ten ‘As for shirts, I have ten.’

The question that arises is why kein must have an ending in (14b) in spite of the fact that it does not have an ending when an overt nominal element follows, (13a–b) and (14a), and that inflectionless elements can, at least in principle, license split NPs as just seen in (15b). Let us take the bull by the horns and propose that all instances of ein-words differ  from plurality numerals in that the former are generally declinable.11 In particular, for the three cases where the ein-words do not exhibit an overt ending, 11.  There is other evidence that ein is different from plurality numerals. For instance, in the genitive (but not in the other cases), the numerals zwei ‘two’ and drei ‘three’ are exceptional in that they take an ending. Interestingly, an adjective following ein or one of these plurality numerals has a different type of ending: (i) a.

das Auto EINES netten Freundes the car of.one(strong) nice(weak) friend ‘the car of one nice friend’

b.

das Auto zweier netter Freunde the car of.two(strong) nice(strong) friends ‘the car of two nice friends’

This difference presumably has to do with the change in the morpho-syntax that ein went through in the history of the German language, where ein has changed from adjectival to determiner-like in inflection. Compare the Old High German example in (iia), taken from Demske (2001: 76), to its Modern German equivalent in (iib): (ii) a. b.

mít éin-emo rôt-emo tûoche mit ein-em rot-en Tuch ‘with one red scarf ’

(Old High German)

(Another interesting case in this respect is Zhuang, discussed by Sybesma & Sio 2006, where the indefinite article/singularity numeral is in a different position from the plurality numerals.)



The strong and weak alternation in German 

I propose the presence of a null ending. Now, the reason why a “bare” ein-word cannot license split NPs is that, although declinable, it does not comply with other, independent conditions on the licensing of split NPs. In fact, if an ein-word does adhere to those conditions, the emergence of the overt ending on it as in (14b) is a side effect (for details, see Section 10.1). With this in mind, table 7 presents the relevant endings, including the null inflections, where the three differences between ein-words and the definite determiners are marked in bold print (cf. table 1; for the meaning of shading in the tables, see below):12 Table 7.  Inflections on the (Negated) Indefinite Article kein ‘no’. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

kein-Ø kein-e-n kein-e-m kein-e-s

kein-Ø kein-Ø kein-e-m kein-e-s

kein-e kein-e kein-e-r kein-e-r

kein-e kein-e kein-e-n kein-e-r

This discussion allows us to relate the “exceptional” cases of the definite article from Section 3.1 to the “exceptional” instances of the indefinite article: despite appearances, both elements have an ending, namely a null morpheme. However, these null endings also differ as the definite article has a null strong ending and the indefinite article has a null weak ending (considering the fact that the adjective following the ein-word has a strong ending and that co-occurring determiners and adjectives have different types of endings). Before I make the proposal involving null endings more general, let us quickly turn to the discussion of the endings on the ein-words and those on the “mixed” adjective with the intention of reducing these two paradigms to the basic ones set out in Section 2. Adjectives of the “mixed” paradigm follow ein-words. In particular, in the three instances where the ein-word does not have an (overt) ending (cf. table 7), the adjective takes a strong ending, marked in table 8 below in bold print. In the other thirteen cases, the adjective has the same endings as the weak adjectives.

12.  I illustrate these inflections with kein ‘no’ here. This allows me to provide the endings in the plural. If it is correct that kein, EIN, and mein are composite structures (footnote 27), then this implies that ein does exist in the plural (before it is combined with the negation, numeral, or possessor). Plank (2003: 15–6) and Delsing (1993b: 33–5) discuss some “unbound” nonsingular usages of a and one in a number of languages. Apparently, languages differ with regard to the licensing of these overt non-singular usages.

 Demonstratives and definite articles as nominal auxiliaries

This mixed bag of endings has given this paradigm its name. Compare table 8 to tables 2 and 3: Table 8.  “Mixed” Adjectival Inflections. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

gut-e-r gut-e-n gut-e-n gut-e-n

gut-e-s gut-e-s gut-e-n gut-e-n

gut-e gut-e gut-e-n gut-e-n

gut-e-n gut-e-n gut-e-n gut-e-n

Note that these three exceptions do not form a “natural class”. However, comparing the other three cases in the nominative/accusative singular, I propose to comprise these exceptions into six cells, marked by shading in tables 7 and 8. Let us consider this in more detail. Duden (1995: 280) observes that the mixed adjectival endings and the strong inflections are identical in the singular structural cases (nominative/accusative). Note now that the paradigm of ein-words is exceptional in itself in that it is different from the strong paradigm in three cells. What is surprising here is that the six exceptions in the mixed adjectival paradigm seem to contrast with only three exceptions with the ein-words, although these elements co-occur in one and the same environment. However, consider the inflectional endings in the accusative masculine and the nominative/accusative feminine: (16) a.

mein-en gut-en Wein my good wine(masc) ‘my good wine’

b. mein-e gut-e Limo my good soda(fem) ‘my good wine’ c.

mein-e gut-e Limo my good soda(fem) ‘my good wine’

(acc masc)

(nom fem)

(acc fem)

Notice that in these three cases, both the ending of the ein-word and the ending of the adjective are the same, -e or -en. Now, if we assume – contrary to traditional descriptions  – that the strong ending is actually with the adjective and the weak ending is with the ein-word also in (16), then all singular structurally case-marked ein-words can be collapsed into a natural group. If so, six instances of the ein-words exhibit weak endings and the six corresponding instances of the “mixed” paradigm of the adjective show strong endings. The remaining instances are strong and weak, respectively. In other words, I propose – perhaps somewhat



The strong and weak alternation in German 

counter-intuitively at first glance – that there are not only three exceptions with ein-words but rather six. These instances involve the exact same morphological cases (nominative/accusative) and number (singular) and, in order to explain this mutual mixture of endings in these two paradigms, I provide a detailed morphosyntactic account in the second part of this chapter.13 To briefly anticipate the account, assuming that determiners move from below adjectives to the DP, I propose that determiners bring about the weak endings on adjectives by a mechanism that causes feature reduction (Impoverishment). Now, assuming that indefinite determiners may, under certain conditions (i.e., our six exceptional cases), move to the DP after narrow syntax, these instances cannot trigger this feature reduction and the feature bundles on the adjectives are spelled out as strong inflections. If so, we take an important step toward reducing the number of paradigms from four to two: with the exceptional instances in the paradigms of the ein-words and the mixed adjectives accounted for by “later” movement of certain indefinite determiners to D, only the strong and the weak paradigms remain. Independent evidence for the special character of the singular structurally case-marked elements comes from the weak adjectival paradigm (recall that the “exceptional” ein-words combine with strong adjectives). In particular, Blevins (1995: 146) and Eisenberg (1998: 173–4) observe that the singular weak endings in the structural cases form a contrast vis-à-vis the plural and/or oblique (= inherent) case-marked ones (i.e., the ones in the dative or genitive). While the first type of ending is -e, the second one is -en.14 Eisenberg simplifies the weak adjectival paradigm in table 3 to table 9 such that singular structurally case-marked items represent the “unmarked” instances: Table 9.  Simplified Weak Adjectival Inflections.

(Structural) Inherent

(Singular)

Plural

-e -en

-en -en

13.  Since the discussion of the inventory of inflections is partially dependent on the discussion of the distribution of inflections, there is a certain amount of overlap in the exposition. 14.  Comparing the tables 1 through 8, we see that the endings in the nominative and the accusative are the same, except for the accusative masculine, which is always marked -n. Below I formulate a “low-level” readjustment rule for these cases and I proceed here by assuming that nominative and accusative are always identical.

 Demonstratives and definite articles as nominal auxiliaries

In view of the different properties of these six cases in the ein-word, “mixed”, and weak paradigms, I conclude that German treats singular structurally casemarked elements special. For ease of exposition, I will continue to call these six instances “exceptions”. The discussion of Sections 3.1 and 3.2 has consequences for the number and nature of the strong and weak endings. While we have taken the first step toward reducing the paradigms from four to two, there is one residual problem. Recalling the discussion of Section 2, there is now an item-specific strong null ending (on the definite article and personal pronoun) and an item-specific weak null ending (on the indefinite article). Furthermore, there are now six strong endings vis-à-vis three weak endings, and strictly speaking, these weak endings are not a proper subset of the strong ones anymore: (17) a.

Strong Endings -r, -s, -m, -n, -e, -ØDIE/SIE

b. Weak Endings -n, -e, -ØEIN

Note that we have now introduced item-specific endings, a new and undesirable notion. Second, recall that strong endings not only better distinguish case, number, and gender than weak endings in general, but four endings (-r, -s, -m, and certain cases of -Ø) are disambiguously strong. With the weak endings no longer a proper subset of the strong ones, both types of endings are becoming more alike in nature: now certain cases of -Ø also indicate weak endings disambiguously. While these may only be esthetical flaws, they may also give rise to certain new predictions (e.g., a possible semantic differentiation of the two sets of endings over time). However, as we will see, the constellation in (17) can be simplified again. During that process, we finalize the reduction of the null strong and weak endings as well as that of the “mixed” paradigm bringing them in line with the basic scenario set out in Section 2. 4.  The proposal: Weak and strong null endings more generally Putting aside the monosyllabic definite article and “bare” ein-words, to which we return below, all other pre-nominal adjectives and determiners exhibit a schwa and are thus disyllabic or, more precisely, multi-syllabic (cf. jedweder ‘every’). This holds across the strong and weak distinction (tables 1–3). I will basically follow Wiese (1988) here, who proposes that the schwa in these cases is due to an insertion rule. Let us first assume that suffixation of an inflection adds an abstract segmental position, indicated as X in (18a). In fact, let us interpret this segmental position as



The strong and weak alternation in German 

an abstract suffix that hosts the feature bundle that is eventually spelled out as the overt ending. Second, schwa is inserted according to the rule in (18b):15 (18) Suffixation by X and Schwa-insertion a. nominal stem+__ X b. Ø -> ә / C__ X

I assume that (18a) applies to all nominal stems that can appear in the DP (i.e., determiners, numerals, adjectives, nouns). In contrast, while schwa-insertion in (18b) applies freely to determiners and adjectives provided the stem ends in a consonant, it applies more restrictively to nouns but not at all to numerals (for more detailed discussion, see Wiese 1996: Section 4.3). To be clear, then, the ending schwa is interpreted here to be due to some phonotactic constraint. Reanalyzing schwa in this way, we assume now that a null ending is not only on certain cases of ein-words, the definite article and the personal pronoun (Section 3) but that it is more generally present; that is, it also follows every “bare” schwa. Although we collapsed tables 1 and 2 in Section 2, I provide their updated versions here for clearer exposition: Table 10.  Inflections on the Demonstrative. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

dies-e-r dies-e-n dies-e-m dies-e-s

dies-e-s dies-e-s dies-e-m dies-e-s

dies-e-Ø dies-e-Ø dies-e-r dies-e-r

dies-e-Ø dies-e-Ø dies-e-n dies-e-r

Table 11.  Strong Adjectival Inflections. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

gut-e-r gut-e-n gut-e-m gut-e-n

gut-e-s gut-e-s gut-e-m gut-e-n

gut-e-Ø gut-e-Ø gut-e-r gut-e-r

gut-e-Ø gut-e-Ø gut-e-n gut-e-r

Comparing tables 10 and 11 to 5 and 6, we note that all the strong inflections are now identical. For ease of exposition, I will call the determiners with

15.  We will see below that endings play an important role in the morpho-syntactic analysis. With the inflection located under X, I assume that the suffixation in (18a) occurs early in the syntax. As to (18b), note also that I treat final glides of diphthongs (e.g., neu ‘new’, blau ‘blue’) as consonants.

 Demonstratives and definite articles as nominal auxiliaries

a null ending ‘die-words’. For the weak paradigm, we obtain the following (cf. tables 3 and 9): Table 12.  Weak Adjectival Inflections. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

gut-e-Ø gut-e-n gut-e-n gut-e-n

gut-e-Ø gut-e-Ø gut-e-n gut-e-n

gut-e-Ø gut-e-Ø gut-e-n gut-e-n

gut-e-n gut-e-n gut-e-n gut-e-n

Finally, although Section 3 basically reduced the ein-words and the “mixed” paradigm to the strong and weak endings by way of a (promised) morpho-syntactic account, consider these two updated tables (13 and 14), where the white cells contain the strong endings (cf. table 10 above) and the shaded cells contain the weak endings (cf. table 12 above). Table 13.  Inflections on the (Negated) Indefinite Article. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

kein-Ø kein-e-n kein-e-m kein-e-s

kein-Ø kein-Ø kein-e-m kein-e-s

kein-e-Ø kein-e-Ø kein-e-r kein-e-r

kein-e-Ø kein-e-Ø kein-e-n kein-e-r

Table 14.  “Mixed” Adjectival Inflections. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

gut-e-r gut-e-n gut-e-n gut-e-n

gut-e-s gut-e-s gut-e-n gut-e-n

gut-e-Ø gut-e-Ø gut-e-n gut-e-n

gut-e-n gut-e-n gut-e-n gut-e-n

While I postpone a detailed discussion of the ein-words to Section 7, let me briefly return to the definite article (cf. table 5). In order to explain the different stem vowels, I assume the following three lexical entries: (19) a. da[neut, nom/acc] b. di[fem/pl, nom/acc] c. de (elsewhere case)

Note now that, while rule (18a) applies and the stems in (19) are suffixed by X, rule (18b) does not and schwa is not inserted. Abstracting away from post-vocalic 〈r〉,



The strong and weak alternation in German 

I follow Wiese (1988: 34) in that the stem final vowel will be realized as short if an obstruent follows and as long otherwise: (20) a. short vowel: das, des b. long vowel: den, dem, die

In the next section, we suggest that the assumption of the abstract suffix X allows us to state that certain instances of declinable elements have no endings (rather than null endings).

5.  Null endings as no endings: The new inflections In the last section, we generalized the proposal of the null inflection -Ø replacing -e as a strong and weak ending, where -e is now due to a schwa-insertion rule. We now have a situation where all determiners and adjectives have identical endings within their relevant set of endings. Furthermore, while the same cells are still different in the relevant tables, the weak endings are, once again, a proper subset of the strong ones. In the last section, we also introduced X, an abstract suffix that hosts the feature bundle that is later to be spelled out as the overt inflection. With X a nominal suffix, it is clear that a null nominal ending does not provide much additional morphological information. Suppose then that we interpret the null ending in these cases as a complete lack of ending. We, then, obtain the following final constellation, where there are four exponents for the strong and one for the weak paradigm: (21) a.

Strong Endings -r, -s, -m, -n

b. Weak Endings -n

Note that all strong and weak inflections are consonantal now. Factoring in the morpho-syntactic account of the endings on ein-words and the “mixed” adjectives, only hinted at in Section 3 but fully developed below, the final update of the two sets of endings looks as follows (where the underscore indicates the earlier presence of the null endings; the -n in the accusative masculine, put in parenthesis, is special as pointed out in footnote 14): Table 15.  Strong Paradigm. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

-e-r -e-(n) -e-m -e-s

-e-s -e-s -e-m -e-s

-e_ -e_ -e-r -e-r

-e_ -e_ -e-n -e-r

 Demonstratives and definite articles as nominal auxiliaries

Table 16.  Weak Paradigm. Case\Gender

Masculine

Neuter

Feminine

Plural

Nominative Accusative Dative Genitive

-e_ -e-(n) -e-n -e-n

-e_ -e_ -e-n -e-n

-e_ -e_ -e-n -e-n

-e-n -e-n -e-n -e-n

Recall that all schwas marked by -e in tables 15 and 16 are due to an insertion rule. In Section 4 above, I fleshed this proposal out by way of suffixing X to the relevant stems and stating a schwa-insertion rule that had X in its context of application. Second, having interpreted null endings as a lack of ending, we now have only one weak ending (cf. table 16). I will suggest below that this -n is inserted in all and only the morpho-syntactic contexts where checking (i.e., a certain type of valuing, see below) and Impoverishment have taken place. Third, certain feature combinations have no overt exponent; that is, they are not overtly spelled out. This holds for both the strong and the weak endings. Finally, these paradigms are now more abstract in that they are no longer tied to specific lexical categories but are relevant to determiners or adjectives alike. This paves the way for the discussion in section 10.4, where these exponents are reinterpreted as underspecified abstract feature bundles, only later to be overtly realized. To sum up thus far, I started this chapter with the traditional picture, which involves five paradigms with five strong and two weak endings. I argued that schwa is inserted by a phonological rule, that certain cases involve null, or rather no, endings, and that the three, or rather six, “exceptions” in the paradigms of the ein-words and “mixed” adjectives can be explained by a morpho-syntactic account. As a consequence, we reduced the number of paradigms to two and the strong and weak endings to four and one, respectively. Finally, we proposed that the feature bundle of the inflection is attached to the nominal stem by way of an abstract suffix (X). In the following, second part of this chapter, I take up the discussion of the distribution of endings, in general, and the morpho-syntactic account of the exceptional ein-words, in particular. 6.  Distribution of inflections: The traditional picture In the first part of this chapter, we discussed the inventory of the strong and weak inflections. Starting with this section, we turn to the distribution of these inflections in more detail. As briefly mentioned in the introduction, the strong ending may, depending on the context, attach either to the determiner or to the adjective



The strong and weak alternation in German 

but not both. Consider again some basic examples, where a determiner with a strong ending precedes an adjective with a weak one, (22a), and an unpreceded adjective has a strong ending, (22b): (22) a.

der gute Wein the(nom.strong) good(weak) wine(masc) ‘the good wine’

b. guter Wein good(nom.strong) wine(masc) ‘good wine’

At first glance, it appears that there is only one strong inflection in a noun phrase. This has come to be known as the Principle of Monoinflection and has been widely discussed (e.g., Darski 1979; Eisenberg 1998:  173; Esau 1973; Helbig & Buscha 2001: 274; Wegener 1995: 105, 153). I provide the following formulation: (23) Principle of Monoinflection The first element within a noun phrase carries the strong and the second one the weak ending.

However, we will see that this language-specific principle can be violated in two ways:  on the one hand, there are noun phrases where no strong inflection is present at all and, on the other, there are cases with two or more strong inflections. Integrating the above discussion, I will reach a different characterization of this alternation in German. Note first that this strong/weak alternation can be interpreted in two ways:  (i) the strong ending appears either on the determiner or the adjective but not both, or (ii) the adjective has a strong ending under some conditions but a weak one under others. While the former describes the positional alternation of the strong ending, the latter refers to the inflectional alternation of the adjective. I will usually refer to the latter, that is, the morphological alternation of the adjective. As a first approximation, one can state for (22) that the presence of a determiner brings about a weak ending on the adjective. Before I make this statement more formal, let me point out that the very occurrence of the strong and weak endings is subject to a number of interesting conditions. Contrary to what (23) leads us to believe, the appearance of a strong ending is by no means obligatory. In fact, not only are noun phrases without a strong inflection grammatical, (24a–b), they become ungrammatical when a strong ending is added to an element that does not support such an ending, (24c): (24) a.

Toms Wein Tom’s wine ‘Tom’s wine’

 Demonstratives and definite articles as nominal auxiliaries

b. ein (prima) Wein a (great) wine ‘a (great) wine’ c.

Wein(*er) wine(*strong) ‘wine’

In particular, this lack of inflection concerns certain possessives (*Tom(s)er), some cases of ein-words (*einer), a few adjectives (*prima(n)er), and more generally nouns (*Weiner). Below, I return to a more detailed discussion of ein-words and nouns. Apart from these cases, we will see that the strong ending is either on determiners or adjectives but that the weak ending is on adjectives only: (25) a. determiner b. adjective

strong inflection weak inflection

Turning to the weak inflection in more detail, we notice that the appearance of a weak ending seems to have nothing to do with the presence of a determiner per se. This characterization includes the null determiner assumed for (26c):16 (26) a.

ein guter Wein a good(nom.strong) wine(masc) ‘a good wine’

b. Peters guter Wein Peter’s good(nom.strong) wine(masc) ‘Peter’s good wine’ c.

guter Wein good(nom.strong) wine(masc) ‘good wine’

Rather, it seems to be the presence of a strong ending on the determiner that “triggers” a weak one on the adjective: (27) a.

der gute Wein the(nom.strong) good(weak) wine(masc) ‘the good wine’

b. dieser gute Wein this(nom.strong) good(weak) wine(masc) ‘this good wine’

16.  Evidence for the presence of a null determiner comes from other languages (e.g., du in French) and from the fact that adjectives need to be licensed by the presence of an (overt) determiner in other cases (see Chapter 1, also Section 10.3 below).



c.

The strong and weak alternation in German 

jeder gute Wein every(nom.strong) good(weak) wine(masc) ‘every good wine’

In a sense, we may state that a weak inflection seems to be parasitic on the presence of a strong ending. Crucially, however, unlike the strong endings on the determiners in (27), the same endings on adjectives do not induce a change in ending on other adjectives, again contrary to the Principle of Monoinflection in (23):17 (28) a.

ein guter süßer Wein a good(nom.strong) sweet(nom.strong) wine(masc) ‘a good sweet wine’

b. guter süßer Wein good(nom.strong) sweet(nom.strong) wine(masc) ‘good sweet wine’ c.

von Peters guter süßer Milch from Peter’s good(dat.strong) sweet(dat.strong) milk(fem) ‘from Peter’s good sweet milk’

d. von guter süßer Milch from good(dat.strong) sweet(dat.strong) milk(fem) ‘from good sweet milk’

To sum up thus far, weak endings occur only if two conditions seem to hold simultaneously: first, a strong ending has to be present (cf. the lack of alternation with ein, Peters, and null determiners) and second, an (overt) determiner has to be present (cf. the lack of alternation with strong adjectives). In fact, these conditions must coincide on the determiner. In other words, inflected determiners have – what I will call for now – a “triggering” factor that brings about the inflectional change on the adjective. The generalization involving the occurrence of endings can be stated as a set of entailments where (29b) trivially includes a null determiner: (29) Asymmetry (ultimately not quite correct) a. A weak ending on an adjective → a strong ending on a determiner. b. A strong ending on an adjective → no ending on the determiner.

With determiners syntactically preceding adjectives, we arrive at a distribution that is characterized by a left-to-right asymmetry where determiners with strong 17.  There are two remarks in order here. First, there is one much-discussed exception to this that involves unpreceded co-occurring adjectives in the dative masculine/neuter (for discussion, see, among others, Gallmann 1996; Schlenker 1999; Roehrs to appear and references cited therein). Second, there is a class of adjectives (e.g., folgend ‘following’) that behaves like determiners and, expectedly, (may) trigger a weak ending on a following adjective (see Section 10.3.2).

 Demonstratives and definite articles as nominal auxiliaries

inflections precede weak adjectives and determiners without relevant ending precede strong adjectives. Note also that with determiners located higher than adjectives, the entailments in (29) hold “bottom-up”. To be clear, then, in all the above cases, a strong ending precedes a weak ending but not vice versa. The latter, impossible scenario is illustrated in (30a). Put more generally, a strong ending must come first and a weak ending cannot occur by itself, as shown in (30b): (30) a. *de guter Wein the(weak) good(nom.strong) wine(masc) b. *gute Wein good(weak) wine(masc)

All these ungrammatical distributions fall under the generalization in (29). There seem to be only two (well-known) exceptions to this general pattern. First, if one were to assume that, rather than no ending at all, the ein-word in (31a) had a weak null ending (cf. Section 3.2), then (31a) would be an exception to (30a). Second, if one made the assumption that the strong ending in the masculine/neuter genitive were uniformly -s, then the adjectival ending in (31b) would be weak and (31b) would be an exception to (30b). I mark these problematic interpretations in lower case in the gloss: (31) a.

ein guter Wein a(-/weak) good(strong) wine ‘a good wine’

b. guten Weins good(strong/weak) wine(gen) ‘good wine’

The two cases in (31) are not identical as the second example has an inflection on the noun. From the comparison of these two distributions, many previous analyses have concluded that unlike determiners, inflections can, by themselves, trigger the alternation (i.e., no inflection on certain ein-words results in a strong adjective and an inflection on the noun results in a weak adjective). Various instantiations of this are available in the literature (see, e.g., Section 8). This basically summarizes the more or less traditional picture. Note that besides this apparent qualitative difference, both cases in (31) are also quantitatively exceptional: (31a) only holds with ein-words in the singular structurally case-marked cases and (31b) only holds in the genitive masculine/neuter.18

18.  The inflectional system of nouns in German is very impoverished. Besides the genitive masculine/neuter, there is one more instance where the noun has a case ending (for the dative plural, see Section 9.2).



The strong and weak alternation in German 

Despite the fact that these cases seem to “mask” the general underlying system, I will not simply put them aside here by treating them as true exceptions. In fact, I believe that these instances provide interesting evidence for the assumption that determiners move from below adjectives to the DP. Making certain assumptions, I will conclude that it is the determiners after all, and not the inflections, that are responsible for the alternation. Returning briefly to (31), I will argue that ein-words start out with a full set of (unspecified) features. However, due to the low position at some point in the derivation, these features can not be valued and, consequently, the ending is not spelled out. In the second example, I argue that, due to the presence of an inflected null determiner (i.e., Ø+s), the adjective is indeed weak (note that the inflection on the null determiner is later deleted under Recoverability of Deletion with the inflection on the noun). 7.  The new picture The proposal I would like to make comprises a number of novel components: with the above discussion of the inventory in mind, I will argue that also inflectionless determiners, the die-words, trigger the alternation; conversely, singular structurally case-marked ein-words, although initially with a full set of unspecified features, do not trigger the alternation. In other words, I will propose that determinerhood (and not inflectedness) is the crucial trigger of the alternation. Furthermore, I will assume a fairly complex structure of the noun phrase where the adjective is merged in Spec,AgrP and I will provide a more comprehensive account that goes beyond the basic cases discussed thus far. In particular, I will focus on the fact that the strong ending can appear on both the determiner and adjective but the weak inflection on the adjective only, even in cases where two determiners co-occur (Section 10.3.3). Recall from Section 5 that certain feminine and plural determiners have no ending, where the schwa on some of these elements was suggested to be due to an insertion rule. Crucially, however, despite this lack of ending on the determiners, the adjectives are still weak (rather than strong). To these cases, I now add plural personal pronouns that may also take a weak ending (for detailed discussion, see 10.3.1): (32) a.

die_ guten Weine the  good(weak) wines ‘the good wines’

b. dies-e_ guten Weine these  good(weak) wines ‘these good wines’

 Demonstratives and definite articles as nominal auxiliaries

c.

wir_ guten Studenten we good(weak) students ‘we good students’

In what follows, I refer to all of these elements as ‘die-words’. Let us pause for a moment and juxtapose some of the “exceptional” cases alongside a stereotypical instance. Starting with the latter, consider (33a), where an inflected determiner “triggers” a weak ending on the adjective. The interesting cases are the second and the third example. The datum in (33b) involves a null determiner and an inflection realized on the noun. As mentioned above, it is often claimed that this lower inflection is responsible for the alternation on the adjective. If so, then we seem to have a bipartite and bidirectional triggering factor:  in contrast to (33a), the inflection in (33b) is not only “detached” from the (null) determiner but it also occurs below the adjective (i.e., on the noun). Turning to (33c), note that we cannot simply claim that the inflection alone is the triggering factor as die is without ending but the adjective is still weak:19 (33) a. de-r -> gute Wein b. Ø guten <- Wein-s c. die -> guten Weine

If we take these cases seriously, as I think we should, then the occurrence of the weak endings in (33) appears, at first glance, to be almost paradoxical: in (33b), there is no (overt) determiner but only an inflection and in (33c), there is only a determiner but no (relevant) inflection. Furthermore, if a strong inflection separated from a (null) determiner can trigger the alternation, then combinations of a (null) determiner and a strong adjective could also be expected to bring about a weak ending on another adjective, contrary to fact. In order to rule this possibility out, we must, perhaps contrary to appearances, assume that there is a close connection between determiners, including null determiners, and endings. To begin with, note that, what all these cases have in common is the presence of a determiner. So in order to resolve this undesirable situation, I propose that the triggering factor resides in the stem of the determiner, including the null determiner,

19.  Note that (33b–c) are different with regard to the overtness of the determiner, on the one hand, and the inflection realized on the noun, on the other:  while the genitive -s is homophonous for (overt) determiners and nouns, the (traditional) plural ending -e is not: di-e vs. Student-en ‘students’, Kind-er ‘children’, Auto-s ‘cars’. The same holds if there is no plural ending on the determiner, as discussed in Section 5. Consequently, these cases should not be collapsed such that the plural ending on Wein is similar in function to the genitive ending.



The strong and weak alternation in German 

and that inflections play a different role (see below).20 This in turn means that there is only one condition on the alternation (and not two holding simultaneously). Furthermore, I propose that the bipartite and bidirectional character of the triggering factor is only apparent. To repeat, the bidirectional character of the alternation seemed to be evidenced by the fact that the triggering factor can be either above or below the adjective. With this factor proposed to be located on the stem of the determiner, this contradictory state of affairs can be explained by movement of the determiner from below adjectives to the DP. As to the bipartite character, I propose that determiners and inflections are generated together. However, an inflected noun may, under certain circumstances, license the deletion of the inflection on the null determiner. Consider these assumptions in more detail. Recalling the behavior of die-words from above, we proposed that all determiners, independent of the presence of an ending, can function as the triggering factor of the alternation. If it is true that it is the stem (and not the ending itself) that is the triggering factor, then the question arises how a stem suffixed by an ending can bring about the alternation. It appears that this property of the stem has to be projected onto the morphologically more complex element. Below, this is more formally instantiated by feature union with regard to the relevant elements. For now, let us simply provide the relevant complex structure.

20.  This is not to say that inflections cannot have a direct effect on other inflections. For instance, there is a condition, often referred to as “double or nothing”, where a suffix on the head noun must be licensed by adjectival agreement on a second element and vice versa. Consider the following paradigm (adopted from Müller 2002:  91, going back to Gallmann 1996: 284; also Haider 1992: 330): (i) a.

Orchester ohne dies-en / eigen-en Dirigent-*(en) orchestra without this / own conductor ‘orchestra without this/its own conductor’

b.

Orchester ohne Dirigent-(*en) orchestra without conductor ‘orchestra without conductor’

Müller (2002: 90 footnote 2) points out that not all speakers judge the starred examples as fully ungrammatical. In the course of the discussion, he makes the insightful distinction between primary and secondary inflection. Simplified, while primary inflection is always obligatory and has much sharper judgments, as in basically all the examples in the main text, secondary inflection is subject to some conditions, as in (i). I assume then that primary inflection has a different account from the secondary one. I will not deal with secondary inflection in more detail here.

 Demonstratives and definite articles as nominal auxiliaries

We have assumed all along that there is an abstract suffix, called X, which hosts the feature bundle later to be spelled out as the actual inflection, (34a). This X is suffixed to the stem of a nominal element, as was suggested in (18a) above. As is typical of morphological structures, the right-hand element projects. Consider the concrete example in (34b) and its spell-outs in (34c): (34) a. X | __ b.

c.

X

stem

X

de / ein

[, , ]

der, ein

In order to prevent the proliferation of null elements, I assume that both X and null determiners need to be licensed; that is, they need an overt host. As a consequence, both are incompatible with one another. In other words, null determiners are different in that they are not suffixed by X. I propose the following two lexical entries for null determiners: (35) a.

Ø | [-fem, gen]

b.

Ø (elsewhere case)

If the discussion thus far is tenable, then a second paradoxical situation reveals itself. Returning to the ein-words, we notice that some determiners trigger the alternation, marked by ->, but others do not, indicated by =. Compare (36a) to (36b–c): (36) a. der -> gute Weine b. ein = guter Wein c. Ø = guter Wein

In fact, we saw above that even inflectionless determiners can trigger the alternation. Putting aside the null determiner in the genitive masculine/neuter for a moment (35a), there seem to be some infectionless determiners that trigger the alternation and others that do not: (37) +alternation die (Ø[-fem, gen]) wir

–alternation ein Ø



The strong and weak alternation in German 

Considering (37), it seems at first glance as if indefinite determiners, overt or null, do not trigger the alternation. However, even this characterization is not quite true as oblique case-marked and/or plural ein-words do trigger this change (cf. dative einem guten Wein ‘a(strong) good(weak) wine’). In view of these cases as well as the null determiner in the genitive masculine/neuter (35a), it is clear that inflections do have an effect on the alternation. It seems as if the triggering factor is on the determiner stem but is, in some cases, made “active” by the presence of an ending. Rather than a direct effect, as the traditional accounts would have it, I propose that inflections have an indirect effect in that they may cause movement of the determiner to the DP, thus bringing about the required conditions for the alternation to be triggered. Concretely, I propose that definite determiners always move to the DP in narrow syntax. In contrast, indefinite determiners move to the DP “later” unless they have a certain ending. Specifically, I propose that there are two fundamentally different types of endings, which are subject to different licensing mechanisms:  checking vs. valuing. On the one hand, there are oblique (dative, genitive) and plural inflections, which must be checked, trigger movement of the determiner, and are always spelled out; on the other, there are structural (nominative, accusative) and singular inflections, which may be valued, do not, by themselves, trigger movement of the determiner, and are spelled out only if valued.21 It is well-known that the structural and oblique cases are different. For instance, Bayer et al. (2001) extensively argue that oblique case in German requires morphological case licensing, that is, that oblique case must be rendered visible.22 A representative example, adapted from their discussion (their Section 2.9), comes from topic drop. In particular, while a structurally case-marked topic can be left unpronounced, this is not possible with oblique elements: (38) a.

(Den) hab¢ ich schon gesehen. him(acc) have I already seen ‘I have already seen him.’

21.  Note that although not identical, there are other cases where inflections are not spelled out; for instance, with unpronounced nouns (provided the latter are compatible with X to begin with). Notice also now that we have two mechanisms (checking, valuing) in the derivational system. This seems undesirable. However, with Agree holding at LF (Chapter 2), checking can be taken as a convenient way of referring to movement of an element triggered by a (generalized) EPP-feature and, once the relevant element is in the higher agreement domain, long-distance agreement can occur to value the relevant features. Modifying the system developed in the main text accordingly, this would allow us to maintain the assumption of one mechanism (i.e., valuing). For the sake of clearer exposition, I will keep the dichotomy in the main text. 22.  Bayer et al. (2001) propose that oblique case projects a separate case phrase (KP) on top of the DP.

 Demonstratives and definite articles as nominal auxiliaries

b. *(Dem) würde ich nicht vertrauen. him(dat) would I not trust ‘I would not trust him.’

Similar facts have been noticed for other languages (for some discussion of Russian, see Franks & Pereltsvaig 2004). If the current discussion of adjective endings turns out to be correct, then we provide more evidence that these types of morphological cases are different. We have now laid out all the essential ingredients of our proposal and before we turn to a more detailed, formal discussion of the mechanics, let us anticipate the analysis in more simple terms. The four scenarios in (34b–c) and (35a–b) are argued to explain the following basic data: (39) a.

der gute Wein: the(strong) good(weak) wine ‘the good wine’

b. ein guter Wein: a good(strong) wine ‘a good wine’ c.

guter Wein: good(strong) wine ‘good wine’

d. guten Weins: good(weak) wine-gen ‘good wine’

de+X-r gute Wein

ein+X-(r) guter Wein

Ø guter Wein

Ø-(s) guten Wein-s

With the triggering factor on the stem of the determiner, the low original position of the determiner and its movement to the DP explain the data as follows:  the datum in (39a) shows a definite determiner, which moves to the DP in narrow syntax triggering the alternation in the process. Turning to the indefinite cases, they only move to the DP in narrow syntax if they have a relevant ending. Starting with (39b), ein is merged with a full set of (unspecified) features for a singular structural ending. By assumption, these do not have to be valued and do not trigger movement of the determiner to the DP. As a consequence, ein remains in its baseposition until after narrow syntax and cannot function as a triggering factor (the unvalued feature bundle of the ending is eventually not spelled out).23

23.  In the discussion of split-NPs (Section 10.1), we will see that there are factors that can trigger movement of these indefinite elements to the DP already in narrow syntax. As a consequence, the feature bundle of ein-words in unmodified split-NPs is valued and spelled out as, e.g., einer.



The strong and weak alternation in German 

As to (39c), there is no ending on the determiner and, consequently, no movement to D in narrow syntax. As with ein, the null determiner cannot trigger the alternation. In contrast, the null determiner in (39d) does have an inflection. As this ending is oblique, it has to be checked and triggers movement of the determiner in narrow syntax and, consequently, brings about the alternation. While this is similar to the der-word above, the null determiner cannot support an inflection and the homophonous ending on the noun licenses the deletion of the one on the null determiner. More generally, movement of the determiner and deletion of the genitive ending, licensed by the inflected noun, explain the apparent bidirectional and bipartite character of the alternation. Furthermore, these assumptions allow the derivation to proceed in a bottom-up fashion and capture the alternation, as we will see, as a by-product of movement of the determiner. If so, then we arrive at another argument for the proposal of this work. In the next section, I propose to interpret the “triggering” factor as a lexical feature, [+Impoverishment], present on every determiner in German. Many of the following ideas have been inspired by Schlenker (1999) and Sauerland (1996). 8.  The strong and weak alternation as a reflex of Impoverishment In order to account for the adjectival alternation in German, Schlenker (1999) proposes that the morphological process of Fission applies, where a terminal node is split creating a subsidiary node. The features of the former node that have not been matched (and deleted) by Vocabulary Insertion are copied onto the latter node (cf. Halle & Marantz 1993). In particular, he assumes that sub-trees come with fully specified feature bundles. Lexical elements such as certain affixes are merged as terminal nodes and can, depending on their specifications, reduce the feature composition of that sub-tree. The remaining undeleted features are transmitted onto a lower part of the tree. In the resultant, impoverished environments, only elements of appropriate (e.g., with fewer) specifications can be merged. Interpreting strong endings as fully specified and weak endings as less specified, a top-down derivation guarantees that the merge of an element with a strong ending will reduce the feature bundle of the sub-tree. This, in turn, allows only elements with a weak or no ending to be merged further down in the tree. The triggering factor of this reduction process is the marking [+Fission], which, for Schlenker (1999: 124), is present on some inflections but not others: (40) a. endings on the determiners are [+Fission] b. endings on the adjectives are [–Fission], except for (40c) c. the adjectival ending -em is [+Fission]

 Demonstratives and definite articles as nominal auxiliaries

This is in stark contrast to what I argued for above. Recalling the above discussion of die-words, I proposed that the triggering factor is on the stem of the determiner. In other words, in contrast to Schlenker (1999) (and also to Sauerland 1996: 33–4), I propose that the stem of the determiner is marked with the relevant feature (and not the endings themselves; for cases like mit frischem schwarzen Kaffee ‘with fresh(strong) black(weak) coffee’ and a more general critique of Schlenker’s proposal, see Roehrs to appear).24 In addition to this, I diverge from Schlenker’s system in some other respects. Most importantly, following the discussion in Sauerland (1996), I will employ the mechanism of Impoverishment, which in turn allows me to assume that the derivation proceeds bottom-up (and not top-down). Furthermore, I will try to be more specific about my syntactic structures than Schlenker and Sauerland. Besides Fission, there is a second operation in Distributed Morphology that reduces the feature content of a terminal node but without creating a subsidiary node. Like Fission, Impoverishment also takes more specific vocabulary items out of the competition extending the domain of less specific (=default) elements (see Halle 1997).25 The basic intuition to be developed is that determiner stems are marked for [+I(mpoverishment)] and that, when they move from below the adjective, they trigger feature reduction with regard to the inflection of the adjective. As a consequence, the adjective will be spelled out as weak. Stems of determiners and adjectives have the respective categorial features [+D(eterminer)] and [+A(djective)], which guarantees that, due to selection, the lexical items are merged in their appropriate position in the syntactic tree. Since all determiners trigger Impoverishment, I assume that there is a languagespecific rule, (41a), that assigns the feature [+I] to every element marked [+D], that is, determiners.26 Strong inflections are assumed to be feature bundles that are only compatible with determiners and adjectives (and quantifiers, see Section 10.3, where we also make the categorial features of the relevant vocabulary items more formal). As already briefly discussed above, let us suppose that there are two types of morpho-syntactic features: oblique/plural vs. structural/singular.

24.  Similarly, Esau (1973: 142) proposes that certain elements are inherently marked (in his system with [+CM], which stands for Case Marker). 25.  I owe this idea to an anonymous reviewer and Heidi Harley (p.c.). The latter also directed my attention to Sauerland (1996). I thank Uli Sauerland for helping me find this paper. 26.  With Impoverishment being a more universal process, this language-specific rule guarantees that determiners have this effect only in German. Also, with successive-cyclic movement triggering Impoverishment in a local domain, category features must, on my account, be present in the syntax.



The strong and weak alternation in German 

For concreteness, let us assume that the former are fully specified features, marked here as [+], that must be checked for the derivation to converge; the latter are underspecified features, marked here as [α], that may be valued. If they are, valuation by the relevant probe provides a negative value (41b), making that feature interpretable at the PF interface; if not, the feature will not be overtly spelled out: (41) Language-specific Rule and Valuation a. [+D] -> [+I] b. [α] -> [–]

To be clear, then, there are two ways to license morpho-syntactic features depending on their “degree” of specification: checking and valuing. As briefly mentioned above, there is a division of labor: the determiner stem is specified for definiteness and the inflectional suffix for case and number (and gender, which is not critical here and will, for the most part, be abstracted away from). I assume the following specifications: (42) Definiteness a. +def -> definite b. αdef -> indefinite (43) Structural case a. [αdat, αgen] -> nominative b. [αdat] -> accusative Oblique case c. [+dat] -> dative d. [+dat,+gen] -> genitive (44) Number a. +pl -> plural b. αpl -> singular

Recall that stems of determiners and adjectives combine with their endings by way of X.  For our purposes, we can state that the feature specification on X is the union of the features of the determiner stem and those of the inflection. As a result, X is marked for features triggering Impoverishment and movement. In the introduction, we stated that [def] and Case are checked/valued on D and number, at least in German, on Card. In other words, Card is always present in the structure in German. Now, with the discussion of Chapter 2 in mind, the semilexical head Card, being located higher than ArtP, will always induce a separate agreement domain with regard to ArtP. Consequently, the determiner must move out of ArtP to check the features on D. With this in place, let us revisit the main types of determiners.

 Demonstratives and definite articles as nominal auxiliaries

Starting with the definite determiner in the masculine nominative (der), consider the feature composition in (45a). The article stem de is marked [+def]. This feature must be checked with D and, assuming that Card induces a separate agreement domain for long-distance Agree, the host X moves under Last Resort to the DP in narrow syntax. (I underline the feature that triggers movement of X to the DP.) As a result, Impoverishment can take place (for a step-by-step derivation, see Section 9) and the underspecified features [αdat, αgen, αpl] are valued and, consequently, spelled out as the ending -r, (45b): (45) a.

X[+D, +I, +DEF, αDAT, αGEN, αPL] de[+D, +I, +DEF]

X[αDAT, αGEN, αPL] [αDAT, αGEN, αPL]



b. de-r = > der

Something similar applies to the structurally case-marked feminine and plural determiners as well as the oblique and/or plural-marked ein-words. Exemplifying the former with the singular nominative variant in (46a), die moves to the DP in narrow syntax and all the features are checked or valued. However, since the feature bundle for nominative feminine has no overt exponent, the determiner is realized without an ending (recall that 〈e〉 is taken to be a spelling convention): (46)

a.

X[+D, +I, +DEF, αDAT, αGEN, αPL] di[+D, +I, +DEF]

X[αDAT, αGEN, αPL] [αDAT, αGEN, αPL]



b. di- = > die_

Turning to the oblique case-marked and/or plural ein-words, consider an example in the dative masculine, (47a). Unlike with der and die, here it is the feature [+dat] that triggers the movement of X to the DP. As a result, all the features are checked or valued and after schwa-insertion, we obtain (47b):27 27.  This derivation holds for all ‘ein-words’ (i.e., indefinite determiners, negative articles, the singularity numeral, and possessive pronouns). As pointed out above, while not a homogenous group semantically, these elements behave morphologically the same; in particular, they share the same morphology in the six “exceptional” instances. I believe that not recognizing these commonalities would be missing an important generalization. Let us assume then that ein ‘a’ is a “supporting” element for D in the DP and also with regard to the negative article kein ‘no’,



The strong and weak alternation in German 

(47)

a.

X[+D, +I, αDEF, +DAT, αPL] ein[+D, +I, αDEF]

X[+DAT, αPL] [+DAT, αPL]



b. ein-m = > einem

As for the singular structurally case-marked ein-words, there are no features that could trigger movement to the DP in narrow syntax, (48a), and, consequently, ein remains in situ. This has two consequences: first, the mechanism of Impoverishment cannot be triggered as the determiner is below the adjectives (and only appears

the (usually stressed) numeral EIN ‘one’, and for possessive determiners such as mein ‘my’ (cf. Corver 2008: 83). This is informally represented as follows: (i)

a. kein <= neg= b. EIN <= numeral Ø= c. mein <= possessor head m=

+ (vacuous) ein + (vacuous) ein + (vacuous) ein

Let us make this intuition more formal by discussing the least straightforward case  – the possessive determiners. Syntactically, there are several possibilities to analyze the location of the possessive element: either it is in Spec,DP or in a Possessor Phrase (PossP), just below the DP, as in Italian. Adopting the framework of Distributed Morphology, I will exemplify the latter possibility here assuming that the feature bundle of the possessor (poss) is in the head position of PossP and that vacuous ein-em (indef.art-infl), as in (47a), is in D, (iia). Both elements have moved there from below. Let us further assume that the host X undergoes Lowering onto the possessor à la Embick & Noyer (2001), which creates a tripartite head, (iib). Next, the possessor and ein undergo fusion (to the exclusion of the inflection), (iic). This brings about a bipartite head, consisting of the fused possessor and ein, on the one hand, and the inflection, on the other. Finally, Vocabulary Insertion inserts mein under the fused terminal node and -em under X, (iid): (ii)

a. b. c. d.

[DP [X indef.art [X -infl]]+D [PossP poss= [CardP …. ]]] [Poss poss= [X indef.art [X -infl]]] [Poss poss+indef.art [X -infl]] [mein-em]

Note that spelling out this fused terminal node with one vocabulary item explains cases that are not straightforwardly compositional; for instance, the third-person feminine possessive pronoun ihr-em is assumed to consist of a possessive element, ein, and the inflection. If these composite structures are on the right track, then their identical inflectional behavior follows straightforwardly.

 Demonstratives and definite articles as nominal auxiliaries

in the DP after narrow syntax, hence too “late” for Impoverishment to apply); second, with Card inducing a separate agreement domain with regard to ArtP, the underspecified features of the inflection cannot be valued and, consequently, cannot be spelled out, as illustrated in (48b):28 X[+D, +I, αDEF, αDAT, αGEN, αPL]

(48) a.

ein[+D, +I, αDEF]

X[αDAT, αGEN, αPL] [αDAT, αGEN, αPL]



b. ein- = > ein_

There are, then, two cases without inflections:  on the one hand, although the features in the bundle on die are checked or valued, there is no exponent for it; on the other, although there is an exponent for the feature bundle on ein, these are not valued and, consequently, not spelled out. More generally, there are two steps in the discussion of case here: syntactically, abstract features are checked or valued and, morphologically, features are overtly realized. This is a typical feature of Distributive Morphology. I return to the discussion of null determiners in Section 9.2. Unlike with ArtP, Card does not induce a separate agreement domain with regard to AgrPs and long-distance Agree can check or value all the relevant features on the adjectives in Spec,AgrP: (49)

X[+A, αDAT, αGEN, αPL]

a. gut[+A]

X[αDAT, αGEN, αPL] [αDAT, αGEN, αPL]



b. gut-r = > guter

To summarize the discussion, while adjectives can never trigger Impoverishment, determiners can, depending on whether or not they move to the DP in narrow 28.  There is no overt manifestation of indefiniteness in German and I assume that the unvalued feature [αdef] on ein does not show any reflex. Furthermore, considering the “supporting” character of ein (cf. footnote 27), the indefinite determiner has to be overtly realized and moves to D at some (later) point (note that the same goes for plural definite determiners, which will move from CardP to DP later). This could be instantiated by PF-movement (Boeckx & Stjepanović 2001; Chomsky 2000: 144 footnote 44, 2001: 37–38; Lasnik 2001; Sauerland & Elbourne 2002), which is not subject to the syntactic cycle.



The strong and weak alternation in German 

syntax. Let us look at the process of Impoverishment in more detail by providing a step-by-stem derivation.

9.  Basic derivations In the first subsection, I discuss run-of-the-mill cases of the alternation and in the second subsection, I turn to the genitive masculine and neuter instances. 9.1  The stereotypical cases The basic proposal is that checked or valued feature bundles are spelled out as the strong inflections. However, if these bundles undergo Impoverishment, then these “reduced” feature sets are spelled out as weak endings; that is, weak inflections are the result of Impoverishment applied to potentially strong endings (Sauerland 1996). In general, let us assume the following two basic mechanisms: Percolation and Impoverishment. Put simply, the former “spreads” new features by adjunction whereas the latter eliminates existent features under sisterhood. In more detail, Percolation occurs as a consequence of adjunction whereby the feature composition of a structure is changed by forming the union of the two relevant feature sets. This feature union percolates up or down on the projections of the element adjoined to by the determiner. There are two sub-cases: on the one hand, articles as heads adjoin to X and “spread” their feature [+I] upwards to XP; on the other, demonstratives as phrases adjoin to XP and “spread” their feature downwards to X: (50) a. article XP[+I] X′[+I]

art[+I]

XP[+I]

dem[+I]

X[+I]



b. demonstrative

XP[+I] X′[+I]

X[+I]



X[+I]

In contrast, Impoverishment occurs under sisterhood whereby the triggering sister and the immediate mother node must be of the same type but of different levels of projection (i.e., X-X′ and X¢-XP). Consider (51). In other words, non-maximal projections (X, X¢) trigger Impoverishment and maximal ones (Specifiers and

 Demonstratives and definite articles as nominal auxiliaries

complements) undergo it, marked by strike-through. The result of Impoverishment then spreads by Percolation as in (50) above: XP[+I]

(51) YP[+DAT]

X′[+I] X[+I]

Y′[+DAT] Y[+DAT]

ZP[+DAT] Z′[+DAT] Z[+DAT]

Finally, if a head Z moves to adjoin to a head X (cf. (51)), then the in-situ projections of Z will undergo Impoverishment (i.e., ZP and by Percolation Z¢). In other words, reduction is “recursive” and may affect the non-minimal projections of a moved head. Similar assumptions hold for the movement of the phrasal determiners (cf. (50b)). Following Julien (2002:  267), we motivated and argued for the following hierarchical representation in Chapter 1 that, moving top-down, consists of a DP, a CardP, AgrPs, the proposed ArtP, a NumP, an nP and NP: (52) [DP D [CardP Card [AgrP Agr [ArtP Art [NumP Num [nP n [NP N]]]]]]]

(For the sake of simplicity, I leave out n in what follows and return to Card in Section 10.) Sidestepping a number of issues involving number and mass nouns, let us illustrate the two processes, Percolation and Impoverishment, with the masculine nominative noun phrase der kleine schwarze Wagen ‘the small black car’. The count noun Wagen ‘car’ projects an NP. The null head Ø under Num specifies the semantic singularity of the noun phrase (i.e., it does not have to be checked/valued). Num merges with NP, projecting NumP: (53)

NumP Num

NP

Ø

Wagen

Next determiners and adjectives, suffixed by X, come into play. The host X contains the features formed by the union of the features of the determiner/adjective stem and those of the inflectional feature bundle. In other words, both suffixed



The strong and weak alternation in German 

determiners and suffixed adjectives have the complete set of features at the beginning of the derivation. Depending on the categorial feature on X, they are merged in ArtP ([+D]) or AgrP ([+A]). Continuing the derivation of (53) in (54), X with the marking [+D] is adjoined to Art. The resulting treelet merges with NumP, projecting ArtP (for expository clarity, I will provide the overt exponent of the inflection here and leave out some intermediate projections): (54)

ArtP[+I, +DEF, αDAT, αGEN, αPL] Art[+I, +DEF, αDAT, αGEN, αPL] X[+D, +I, ...] Art[+D, +I, ...]

NumP

Num

NP

Ø

Wagen

d+r[+D, +I, +DEF, αDAT, αGEN, αPL]

Due to adjunction of the definite determiner, the features including [+I] spread across the projections of Art. Next, Art moves to adjoin to – what I will call here – Agr1 by sideward movement (e.g., Bobaljik & Brown 1997). This treelet merges then with ArtP. Agr1 triggers Impoverishment on ArtP and its lower projections: (55)

Agr′1[+I, +DEF, αDAT, αGEN, αPL] Agr1[+I, +DEF, αDAT, αGEN, αPL] Art[+I, ...]i

Agr1[+I, ...]

ArtP[+I, +DEF, αDAT, αGEN, αPL] ti

d+r

NumP NP

Consonant with the discussion above, the adjective is merged in Spec,AgrP1. This, now, is the crucial instance of Impoverishment where Agr′1 is the sister of AP, reducing the latter’s feature content: (56)

AgrP1[+I, +DEF, αDAT, αGEN, αPL] AP[+A, αDAT, αGEN, αPL] schwarz+(r)

Agr′1[+I, +DEF, αDAT, αGEN, αPL]

Agr1[+I, +DEF, αDAT, αGEN, αPL] Art[+I, ...]i d+r

Agr1[+I, ...]

ArtP[+I, +DEF, αDAT, αGEN, αPL] ti

NumP NP

 Demonstratives and definite articles as nominal auxiliaries

Next, merging the second adjective, the derivation basically proceeds as in (56). Agr1 moves to adjoin to Agr2 by sideward movement and Agr2 is merged with AgrP1 projecting AgrP2. The adjective in Spec,AgrP2 undergoes Impoverishment: (57)

AgrP2[+I, +DEF, αDAT, αGEN, αPL]

AP[+A, αDAT, αGEN, αPL] klein+(r)

Agr′2[+I, +DEF, αDAT, αGEN, αPL]

Agr2[+I, +DEF, αDAT, αGEN, αPL]

Agr1[+I, ...]j

AgrP1[+I, +DEF, αDAT, αGEN, αPL]

Agr2[+I, ...] AP[+A, αDAT, αGEN, αPL] Agr′1[+I, +DEF, αDAT, αGEN, αPL]

d+r

schwarz+(r)

tj

ArtP[+I, +DEF, αDAT, αGEN, αPL] ti

NumP NP

Finally, Agr2 moves to D, which merges with AgrP2: (58)

D′[+I, +DEF, αDAT, αGEN, αPL]

D[+I, +DEF, αDAT, αGEN, αPL] Agr2[+I, ...]k d+r

AgrP2[+I, +DEF, αDAT, αGEN, αPL]

D[+I, ...] AP[+A, αDAT, αGEN, αPL] Agr′2[+I, +DEF, αDAT, αGEN, αPL] klein+(r)

tk

AgrP1[+I, +DEF, αDAT, αGEN, αPL] AP[+A, αDAT, αGEN, αPL] Agr′1[+I, +DEF, αDAT, αGEN, αPL]

schwarz+(r)

tj

ArtP

Similar assumptions hold for phrasal demonstratives. They originate in Spec,ArtP (where they do not trigger Percolation or Impoverishment), move to adjoin to AgrP (where they do trigger Percolation and Impoverishment), and land in Spec,DP. I assume that this step-wise displacement involves instances of A′-movement. Finally, depending on the resulting feature composition on the relevant elements,



The strong and weak alternation in German 

these sets will be spelled out by the strong endings in table 15 or, alternatively, by -n in checked but impoverished contexts (see also Section 10.4). Importantly, if Percolation and Impoverishment occur in a local domain, then head movement of the article as well as A¢-movement of the demonstrative provide evidence that movement to DP must be successive-cyclically. Furthermore, this implies that Impoverishment must be iterative in order to capture the non-local effects of the alternation evidenced when several adjectives co-occur.29 To sum up, I have proposed two mechanisms that, in tandem, explain the inflectional alternation in German. As a result of Percolation and Impoverishment, non-maximal projections trigger the alternation and maximal projections undergo it.30 The triggering factor was proposed to be the feature [+I], assigned to determiners by a language-specific rule. The alternation on the adjective was shown to be a side-effect of the movement of the determiner from below the adjectives to the DP. Assuming that inflections are the overt spell-out of feature bundles, it was suggested that weak endings are the spell-out of impoverished feature sets. With this in place, we are ready to turn to some necessary readjustment rules. In order to explain some unsystematic exceptions, we formulate the following informal readjustment rules: first, we saw above that, except for the masculine accusative (cf. tables 15 and 16), the structural cases have the same endings in the strong and weak paradigms. In order to account for the surprising presence of -n, I suggest the rule in (59a). Second, according to the rules in (18), which suffix X and insert schwa, we would expect unvalued ein-X to be always spelled out as eine. Rather than blocking rule (18b) in certain contexts, I account for the lack of schwa

29.  Note that jeder ‘every’ is similar to the determiners discussed in the main text. In order to explain the alternation with this element, we assume that elements like this also move from below adjectives. Note that this is in keeping with the assumptions in Chapter 3. 30.  The reader will have noticed that, the way the system is set up, Impoverishment cannot be triggered by a complement or Specifier. In particular, complements (e.g., AgrP) cannot trigger Impoverishment on heads (e.g., D), and Specifiers (e.g., AP even if marked as [+I]) cannot trigger Impoverishment on intermediate projections (e.g., Agr¢). This is a desired result as it rules out the possibility of weak determiners. All being equal, we could now allow adjectives to be marked for Impoverishment and claim that it is the type of position the element is in that is responsible for the fact that adjectives do not induce Impoverishment. However, I will leave the discussion unchanged, as not all scholars may agree that adjectives are in Specifiers, and if Bernstein (1993) is right in that different types of adjectives may occupy different positions, then adjectives in head positions must also be prevented from triggering Impoverishment.

 Demonstratives and definite articles as nominal auxiliaries

in the three instances of “bare” ein (cf. table 13) by stating a rule that deletes the schwa in the relevant contexts, (59b): (59) a.

masculine accusative: X | [] -> -n

b. unvalued nominative masculine and nominative/accusative neuter ein: ә -> Ø

These morpho-phonological readjustment rules apply in the context of particular morpho-syntactic features. The first rule spells masculine accusative features out as -n; the second undoes schwa-insertion when the feature bundle of certain items has not been valued. More generally, while (59a) allows us to keep the simple generalization that the exponents in the structural cases are the same, (59b) provides a handle on this and some other usages of ein. To our case in (60a), we add (60b–c), where schwa cannot occur. Besides that, there are other instances that may have an optional schwa, as illustrated in (60d), slightly adapted from Grewendorf (1991: 304) and van Riemsdijk (1989: 124): (60) a. b.

Ich habe kein(*e/*s) Geld. I have no money ‘I have no money.’ ein(*e)s, zwei, drei, … one, two, three, … ‘one, two, three, …’

c.

Es ist um ein(*e)s. it is at one ‘It is one o’clock.’

d. Geld habe ich kein(e)s. money have I none ‘As for money, I have none.’

Now, while the rule in (59b) is obligatory in (60a–c), it is optional in (60d).31 With my current understanding of these issues, I believe that all these exceptional cases are best captured by the readjustment rules in (59) applying to certain elements in certain contexts and, consequently, these readjustments are low-level phenomena. Considering that these patterns are rather frequent, I assume that the acquisition of these readjustment rules does not pose a problem for the language-learning child. 31.  In order to capture all these cases, the (context of) application of the schwa-deletion rule will have to be fine-tuned, something I will not do here.



The strong and weak alternation in German 

To sum up, in the last section we saw that “late” movement of determiners has some explanatory potential. It explains why there is no Impoverishment triggered by the exceptional ein-words and the null determiners although they are marked for that feature. Second, this type of movement can explain why some of the einwords have no ending (as they were not valued and hence not spelled out). Finally, the low position of the determiner and movement to the DP allowed us to employ Schlenker’s basic proposal in bottom-up derivations and accounts for the genitive cases, to which we turn next. 9.2  The genitive masculine and neuter As already mentioned above, adjective-noun combinations in the genitive masculine and neuter, as in (61a–b), respectively, are special: on the one hand, a weak inflection is on an adjective in a seemingly “strong” context (i.e., there is no overt determiner); on the other, an (apparent) strong ending on a noun follows a weak inflection on an adjective: (61) a.

guten Wein-s good(gen.weak) wine(gen.masc) ‘(of) good wine’

b. guten Brot-s good(gen.weak) bread(gen.neut) ‘(of) good bread’

The question that arises is how to account for this exceptional inflectional behavior. Methodologically, there are two ways to proceed: (i) we could assume that the adjective inflection in (61) is strong after all. If so, then this ending on the adjective basically follows from the above system but we would admit that there are two different strong endings in the genitive for determiners and adjectives (perhaps instantiated by another readjustment rule). Besides this undesirable assumption, we would also admit that there is no correlation between the case ending on the noun and -en on the adjective; (ii) we could stick with the claim that the adjective in (61) is indeed weak (also Sauerland 1996:  30). If so, we could locate the exceptional behavior in the assumption that there are two lexical entries for null determiners. While this would also allow us to keep the above system unaltered, we could continue claiming that strong and weak endings in the genitive are different (-s vs. -n) and we could better explain the relation between the inflections on the adjective and noun. In what follows, I develop the second perspective in more detail and flesh out the assumptions about the null determiners (cf. (35)). Recall from above that null determiners and X, both being null elements, are not compatible with one another. I state the two lexical entries as follows where

 Demonstratives and definite articles as nominal auxiliaries

(62a) illustrates the assumptions about the mass determiner in the genitive masculine/neuter and (62b) shows the elsewhere case. Note that the line under Ø in (62a) indicates where the relevant feature bundle of the inflection is located. (62) a.

Ø[+D, +I, αDEF, +DAT, +GEN, αFEM]



 | __

b. Ø[+D, +I, αDEF]

Due to the specification for oblique case in (62a), the genitive determiner will move to D in narrow syntax. It triggers Impoverishment and the adjective will be spelled out as weak. In contrast, the determiner in (62b) will stay in situ in narrow syntax and does not trigger Impoverishment. In that sense, it is similar to the einwords discussed above but differs from them in that it was merged without any ending at all. Before we make a suggestion as to what happens with the inflection of the null determiner in (62a), let us consider another potential case like this. We could basically make the same claim about the dative plural, where, similar to the genitive, the noun has a case ending (-n). Unlike the genitive, the ending on the adjective (-n) is ambiguous between a strong and a weak one. To make these cases parallel, suppose the latter, (63a). The relevant lexical entry for the null determiner is given in (63b): (63) a. b.

guten Wein-e-n good(weak) wine-s-(dat) ‘(to) good wines’ Ø[+D, +I, αDEF, +DAT, +PL] | __

Movement and co-occurring Impoverishment are like above. Now, what is important to point out about the case-related inflections on nouns is that not only are they restricted to two instances (genitive masculine/neuter and dative plural), they can also “double” the inflection on a determiner:32

32.  Under certain circumstances, the inflections on the noun can or must be left out: (i) a.

Ich erinnere mich des Barock(s). I remember refl the(gen) baroque ‘I remember the baroque.’

a′. Ich erinnere mich guten Barock*?(s). I remember refl good(gen) baroque ‘I remember good baroque.’



(64) a.

The strong and weak alternation in German 

de-s guten Wein-s the-infl good(weak) wine-infl ‘(of) the good wine’

b. de-n guten Wein-e-n the-infl good(weak) wine-s-infl ‘(to) the good wines’

The fact that definite determiners may occur with a doubled inflection on the noun indicates that the inflection on the noun is, at least in part, independent of the presence of a determiner. With this in mind, consider the analysis of the data in (61), where the inflected null determiner moves to the DP, triggering Impoverishment: (65) [DP [Ø+s]i+D [AgrP guten [ArtP [Ø+s]i [NPWein+s]]]]

Note now that the feature bundles of the inflections on the noun and the determiner are identical and that unlike the overt noun, the null determiner cannot support an inflection. I propose, then, that the features on the noun license the failure to spell out those on the null determiner under Recoverability of Deletion. Notice that this relation between the lower copy of the determiner and the noun is a local one. In conjunction with movement of the determiner triggering Impoverishment “from below”, this captures the apparent bidirectional structure of the alternation.33

a″. während des Sozialismus(*ses) during the(gen) socialism ‘during socialism’ b. *den Pos(e)n the(dat) bottoms ‘(to) the bottoms’ This is different from inflections on determiners and adjectives, which, with the exception of the die-words and ein-words, distribute across the entire paradigm and are obligatory. We may conclude, then, that there are special conditions on the realization of the inflections on nouns, something I will not pursue here. 33.  Although different from the text case, there is another instance, where the presence of a second -s licenses the deletion of the first. In particular, in the nominative and accusative singular, the neuter demonstrative dieses ‘this’ can be replaced by dies. Independent of which form is used, the adjective is weak: (i) dies-es / dies schöne(*s) Haus this / this nice(weak) house ‘this nice house’ Following Gallmann (2004:  152 footnote 3), I assume that dies is underlyingly dies-s. Unlike in the genitive in the main text, I assume that the stem -s optionally licenses the “late”

 Demonstratives and definite articles as nominal auxiliaries

The discussion has revealed two fundamental differences between structural/ singular and oblique/plural endings. Structural/singular endings do not trigger movement to the DP and are not spelled out if they are not valued. In contrast, oblique/plural inflections do trigger movement to the DP and must be spelled out. For the latter, there is only one exception, namely, the feature bundle on the null determiner is not spelled out as, on the one hand, it cannot be supported by a null element and, on the other, it is “recoverable” from the presence of identical features on the noun.

10.  Special cases In the first part of this section, we return to the discussion of split-NPs from Section 3.2 with the intention of explaining the alternation between structurally case-marked ein and einer (or ein and eines, for that matter). In the second subsection, we discuss the inflections in indefinite pronoun constructions such as etwas anderes ‘something different’. In the third sub-section, we account for the differing inflectional behavior of adjectives with personal pronouns, determinerlike adjectives (e.g., folgend ‘following’), and semantic intensifiers such as floating quantifiers (e.g., alle ‘all’) and similar elements. The final sub-section summarizes the account and discusses the interpretation of the syntactic features in the morphology and their overt realization. 10.1   Split-NPs Split-NPs are discontinuous noun phrases where the lower part of a DP is topicalized and the higher part stays in situ (Fanselow 1988; van Riemsdijk 1989). Consider a simple example: (66) a.

Ich habe immer nur [Peters Hemden] getragen. I have always only    Peter’s shirts worn ‘I have always worn only Peter’s shirts.’

b. [Hemden] habe ich immer nur [Peters eN] getragen. shirts have I always only    Peter’s worn ‘As for shirts, I have always worn only Peter’s.’

Let us call the in-situ part, Peters ‘Peter’s’ in (66b), the ‘licenser’ of the split-NP and assume that there is a null noun following it. While I cannot discuss all the deletion of the inflectional -s (actually, deletion of X on my account, which will also block later schwa-insertion).



The strong and weak alternation in German 

intricacies of the construction in detail here (but see Roehrs 2006a: Chapter 4, Part II; 2007a), it seems clear that split-NPs have to be licensed by an overt element in a high position in the DP. To see this, consider first a complex genitive possessor, which can only follow the head noun in Modern Standard German:34 (67) a.

Hemden meines Vaters shirts of.my father ‘shirts of my father’

b. ??meines Vaters Hemden my father’s shirts

Since definite noun phrases can, under certain conditions, be split, as just seen in (66b), let us exchange Peters with meines Vaters. We notice that the result is fairly marked: (68) a.

Ich habe immer nur Hemden meines Vaters getragen. I have always only shirts of.my father worn ‘I have always worn only shirts of my father.’

b. ??Hemden habe ich immer nur meines Vaters getragen. shirts have I always only of.my father worn

There are two analyses for (68b): first, if the possessor is in a high, pre-nominal position, (69a), then the ungrammaticality follows from whatever causes the ungrammaticality in (67b) above. Alternatively, we could assume that the possessor is in a low, post-nominal position, (69b). While this position of the possessor is in principle fine, as already seen in (67a), we need something else to explain the marked status: (69) a. ??Hemden habe ich immer nur [DP meines Vaters eN] shirts have I always only of.my father

getragen. worn

b. ??Hemden habe ich immer nur [DP Ø eN meines Vaters] getragen. shirts have I always only    of.my father worn

Guided by the behavior of simple possessors like Peters, I suggest that this additional condition is that licensers have to be in a high position of the split-NP. In short, conditions on the positioning of complex genitive possessors and conditions on the positioning of overt licensers in split-NPs pose contradictory requirements on the possessor in (68b), explaining the marked status of the example. We assumed above that Card and Agr induce a “blocking” effect for longdistance Agree but not movement. Let us further assume that semi-lexical heads

34.  This is different for possessor proper names, which may occur before or after the possessum in German.

 Demonstratives and definite articles as nominal auxiliaries

also pose a problem for licensers in ArtP and positions below it. If this is on the right track, then we get a handle on the alternation between ein and einer. Recalling the basic paradigm from (13) and (14), repeated in (70) for convenience, we can explain the data as follows: there are no special conditions on non-split NPs, as in (70a), and the derivations proceed as discussed above. In contrast, licensers of split-NPs have to be in a position higher than ArtP. Now, while the adjective in (70b) fulfills that condition and the ein-word can remain in situ in narrow syntax, there is no other overt element in (70c) and, in order to fulfill the licensing condition, the ein-word moves to the DP in narrow syntax under Last Resort. Due to this independent licensing condition, all the features can be valued and, consequently, spelled out as discussed above: (70) a.

Das ist kein (guter) Wein. this is no good(strong) wine ‘This is not a good wine.’

b. Wein ist das kein guter! wine is this no good(strong) ‘As for wine, this is no good one.’ c.

Guter Wein ist das kein*(er)! good(strong) wine is this none(strong) ‘As for good wine, this is not one.’

To sum up, an independent condition on the positioning of the licenser of splitNPs brings about the valuation and thus spell-out of the ending on the ein-word. In other words, the inflection on the ein-word is a side-effect of the high position required for licensers. 10.2  Indefinite pronoun constructions In Roehrs (2008), I argue that indefinite pronoun constructions such as the ones in (71a–b) are binominals consisting of a pronoun nominal and a modifier nominal. The modifier nominal, containing the adjective, is adjoined to an Indefinite Pronoun Restrictor Phrase (IPRP) inside the pronoun nominal. This adjunction is mediated by a Modifier Phrase (ModP; Rubin 1996). Consider (71c). Importantly, it is argued there that both the pronoun and adjective agree in their phi-features and that the adjective has a strong ending (for some morphological variation with jemand ‘somebody’, see Roehrs to appear): (71) a. mit etwas anderem with something different(strong) ‘with something different’



The strong and weak alternation in German 

b. %mit jemand anderem with somebody different(strong) ‘with somebody different’ c.

Indefinite pronoun construction DP

D

IPRP IPRP



et je

ModP

IPR

NP

-was -mand

eN eN

Mod

AgrP [anderem eN] [anderem eN]

The strong inflection on the adjective follows from our discussion. Above, I assumed that adjective stems are suffixed with a complete feature bundle (mediated by X). Now, although marked for Impoverishment, the determiner cannot trigger the alternation on the adjective, as AP is embedded in the modifier nominal (i.e., in AgrP and ModP) and is thus never in the complement or Specifier position of the determiner. Consequently, the feature bundle is spelled out as a strong ending.35 10.3  Lexically ambiguous elements In this section, we extend the discussion to the following, in this regard less frequently discussed, cases. The example in (72a) shows the combination of a pronoun and adjective (see Duden 2007: 39), the datum in (72b) exemplifies a special type of adjective (see Helbig & Buscha 2001: 277), and the data in (72c–d) illustrate combinations of – what looks like – two determiners (see Roehrs 2007b): (72) a.

wir %nette / netten Studenten we   nice(strong/weak) students ‘we nice students’

35.  If there was some kind of null determiner in the modifier nominal, it would not be able to trigger Impoverishment under our assumptions either. Of particular relevance here would be the genitive (cf. Section 9.2). Note, however, that indefinite pronoun constructions in the genitive are fairly awkward (see Roehrs 2008; to appear). One could blame this on the lack of the relevant licensing context. In particular, the lack of an overt noun in ModP (which implies the lack of an ending on that element) can not license the deletion of the inflection on the null determiner.

 Demonstratives and definite articles as nominal auxiliaries

b. folgendes %neues / neue Beispiel following   new(strong/weak) example ‘the following new example’ c.

mit einer %jeder / jeden netten Frau with an(st)  every(st/wk) nice(wk) woman ‘with each nice woman’

d. alle diese(*n) guten Studenten all(st) these(st) good(wk) students ‘all these good students’ dʹ. alle gute*(n) Studenten all(st) good(wk) students ‘all good students’

Note first that what all these cases have in common is that the weak ending on a garden-variety adjective (nett ‘nice’, neu ‘new’, gut ‘good’) is always fine. I interpret this to mean that all these DPs have a determiner, which gets marked for Impoverishment. Interestingly, though, strong inflections can sometimes occur after what looks like a determiner (with the exception of (72d′)). In order to capture these co-existing options, I propose that some elements are not only determiners, thus triggering the alternation, but that they may also be quantifiers or intensifiers, which do not bring about Impoverishment. In what follows, I propose that some elements are lexically ambiguous. Notice first that there are actually three elements that take part in the weak and strong alternation: besides determiners and adjectives, quantifiers may also exhibit these inflections: (73) a. b.

die wenigen netten Leute the(strong) few(weak) nice(weak) people ‘the few nice students’ wenige nette Leute few(strong) nice(strong) people ‘few nice students’

In fact, quantifiers have a hybrid character: in some ways, they are like adjectives, as just seen with the inflectional parallelism in (73) and the fact that they can be preceded by a determiner (cf. Esau 1973: 141–2); in other ways, they behave like determiners. As to the latter, it is well known that both determiners and quantifiers can license singular countable nouns, where the indefinite determiner is in D and the homophonous (but stressed) quantifier is in Spec,CardP:36

36.  I assume that due to the complementary distribution of numerals and quantifiers, the former is a type of the latter. Note also that there is cross-linguistic evidence that the



The strong and weak alternation in German 

(74) a.

ein guter Student a good(strong) student ‘a good student’

b. EIN guter Student one good(strong) student ‘one good student’

Let us summarize these hybrid properties: (75) a. Distribution of Endings (morpho-syntactic): b. Licensing of DP (semantico-syntactic):

D≠Q=A D=Q≠A

With this in place, let us finally be more explicit about the feature compositions of these elements. Above, we concluded that there is one set of strong endings, which can attach to determiners or adjectives. With the discussion of quantifiers in mind, let us assume that these endings can also attach to quantifiers. Assume the following features, where (76a) and (76b) are taken from Chomsky (1981: 48): (76)

Feature Composition of Vocabulary Items: a. N [+N, –V] b. A [+N, +V] c. Q [+N, +V, +Q] d. D [+N, +V, –Q]

As proposed above, the abstract suffix X is attached to nominal stems, thus sensitive to the lexical feature [+N], and strong inflections, or rather their feature

indefinite determiner and the singularity numeral are two different elements. In particular, in dialects of Ladin, a Rhaeto-Romance language, both elements can co-occur (Haiman & Benincà 1992: 152): (i)

da øna na skwadra from one a team ‘from one team’

(Badiot)

Similar facts hold also in the dialect of Gardena. In footnote 27, I proposed that the singularity numeral is a composite consisting of a null element and vacuous ein. Consonant with the discussion thus far, I assume that the null element of the numeral is in Spec,CardP, which is supported by one of the copies of moved ein. The assumption of a null element has also the potential of explaining the diachronic relatedness of these two elements as well as their synchronic differences in pronunciation: one vs. a(n) (English, see Perlmutter 1970), ejn vs. a(n) (Yiddish, see Jacobs 2005: 172, 191), õa vs. a (southern German and Austrian dialects, see Bhatt 1990: 201, citing a personal communication by Hubert Haider; see also Karnowski & Pafel 2004: 175, footnote 8).

 Demonstratives and definite articles as nominal auxiliaries

bundles, are only compatible with the stems of determiners, adjectives, and quantifiers. Hence, these bundles are sensitive to [+N, +V]: (77) Feature Contexts for Suffixes: a. abstract X [+N] b. “strong” feature bundle [+N, +V]

We are now ready to make the discussion more concrete for the three cases set out above. 10.3.1  Determiner-like pronouns Following Postal (1966), I assume that personal pronouns are determiners. As discussed in Roehrs (2005a), personal pronouns are similar to ein-words such that singular structurally case-marked pronouns take a strong adjective and oblique and/or plural pronouns a weak one. For instance, consider the following examples and note that like the other die-words, the pronouns do not have an obvious strong inflection: (78) a.

{kein / ich / du} armer Idiot {no / I / you} poor(strong) idiot ‘no/I/you poor idiot’

b. {keine / wir / ihr  } armen Idioten {no / we / you} poor(weak) idiots ‘no/we/you poor idiots’

I proposed there that like ein-words, personal pronouns may move to the DP at different times, depending on their feature specifications. Interestingly, as pointed out there, plural pronouns may not only take a weak adjective, as is expected under Postal’s conjecture, but they may also be followed by a strong one, which is surprising: (79) a.

wir netten Studenten we nice(weak) students ‘we nice students’

b. %wir nette Studenten we nice(strong) students ‘we nice students’

In that work, pronouns were assumed to be “agreeing” and “non-agreeing” to explain the weak and strong endings in (79), respectively. In what follows, I make this proposal more formal by suggesting that “agreeing” means that the pronoun is



The strong and weak alternation in German 

lexically a determiner whereas “non-agreeing” means that the pronoun is a quantifier. Now, since only determiners receive the feature [+I] by the rule (41a), updated now with the above feature compositions in (80a), the quantifier version of the pronoun does not trigger the alternation. Consonant with the discussion above, assume that pronouns are merged as stems and that they are suffixed with the relevant feature bundle as follows: (80) a. [+N, +V, –Q] -> [+I] b. X[+N, +V, ±Q, αDAT, αGEN, +PL] w[+N, +V, ±Q]

X[αDAT, αGEN, +PL] [αDAT, αGEN, +PL]

To be clear, then, depending on the categorial specification on the stem w with regard to either [–Q] or [+Q], the pronoun triggers Impoverishment or not. Finally, in view of the fact that pronouns in Germanic are portmanteau, rather than segmental, forms and that their paradigms are suppletive (Howe 1996: 70), I assume that the pronoun stem and the inflection in (80b) undergo Fusion, creating one terminal head. Vocabulary Insertion spells out this node as the pronoun. Note that these pronouns are different from the ones of the third person, which, segmentally, are similar to garden-variety determiners. Here the stem and inflection are spelled out separately, that is, Fusion does not take place. Next, I turn to some cases that are not as general as the cases discussed above but, nonetheless, deserve attention from our perspective, especially, in view of an important prediction. 10.3.2  Determiner-like adjectives As already mentioned above, singular countable nouns cannot be licensed by garden-variety adjectives alone, but must be preceded by a determiner (or a quantifier). Compare (81a) to (81b). Interestingly, there is one type of adjective that, on its own, can license these noun phrases, (81c): (81) a. *Neues Beispiel wird das illustrieren. new(strong) example will this illustrate b. *(Ein) Beispiel wird das illustrieren. an example will this illustrate ‘An example will illustrate this.’ c. *(Folgendes) Beispiel wird das illustrieren. following(strong) example will this illustrate ‘The following example will illustrate this.’

 Demonstratives and definite articles as nominal auxiliaries

These adjectives are special in that they are similar to definite expressions.37 Let me refer to these elements, then, as ‘determiner-like adjectives’. Interestingly, as noted by, among others, Gallmann (1996: 298) and Haider (1992: 329), a second, following adjective can be weak or, less commonly, strong: (82) a. Folgendes neue Beispiel wird das illustrieren. following(strong) new(weak) example will this illustrate ‘The following new example will illustrate this.’ b. %Folgendes neues Beispiel wird das illustrieren. following(strong) new(strong) example will this illustrate ‘The following new example will illustrate this.’

Similar to the personal pronouns above, I assume that these elements are lexically ambiguous between determiners and quantifiers. If the former, they trigger the alternation and the adjective will be weak; if the latter, they do not and the adjective will be strong. Note that in each case, the determiner-like adjective itself is not preceded by an (overt) determiner and the strong ending on it follows from the above system.

37.  There is morpho-syntactic evidence for this assumption in Swedish. In particular, although the noun phrase in (i) is semantically definite, there is no suffixal determiner on the noun. As such, these type of adjectives are parallel to pronouns or demonstratives, (ii) ((i) is from Börjars 1998: 208): (i) 

ovannämnda institution(-*en) above-mentioned department(-the) ‘above-mentioned department’

(ii)  a.

(Swedish)

vi hungriga studenter(-*na) we hungry(weak) students ‘we hungry students’

b. denna mus(-*en) this mouse ‘this mouse’ These adjectives, then, behave on a par with clear instances of determiners. Interestingly, if a free determiner is added, (iii), the suffixal determiner also appears ((iii) is from Andersson 1994: 284) (iii)

den ovannämnda kommitté-n the above-mentioned committee-the ‘the above-mentioned committee’

The addition of a determiner is also possible in German, and as is expected, the ending of the adjective is weak: das folgende Beispiel ‘the following(weak) example’. I will not examine the semantic/pragmatic differences between these constructions.



The strong and weak alternation in German 

10.3.3  Determiner-like intensifiers In this subsection we turn to cases where  – what looks like  – two determiners co-occur. These instances are relatively rare. Two cases are of particular interest here: on the one hand, I discuss combinations of alle ‘all’ and definite determiners; on the other, combinations of ein and jeder ‘every’. I will assume that the first element in this combination is a semantic intensifier where alle intensifies the exhaustiveness of the set being considered and ein has an individualizing function (see Roehrs 2007b). In these combinations, we will see that, categorially, this type of alle is only an intensifier thus never triggering the alternation but that ein is ambiguous between an intensifier and a determiner-like element. 10.3.3.1  Alle  With the above discussion in mind, we arrive at an important prediction:  if successive-cyclic movement takes place from below adjectives and Impoverishment and Percolation occur in a local fashion, then elements “higher” than the phrase where the determiner ultimately lands should not undergo Impoverishment and, consequently, should not be weak. This prediction is borne out. Consider the data in (83). Under the current system, alle as well as diese must have moved from below the adjective in (83a–b). The interesting case is (83c), where both of these elements co-occur and they both have a strong ending: (83) a.

alle kleine*(n) Autos all(strong) small(weak) cars

b. diese kleine*(n) Autos these(strong) small(weak) cars c.

alle diese(*n) kleinen Autos all(strong) these(strong) small(weak) cars

To be clear, neither alle nor diese triggers Impoverishment on the respectively other determiner. In order to account for this, I propose that diese moves from ArtP to DP triggering Impoverishment on the adjective klein ‘small’. As to alle in (83c), I suggest that this type of element is an intensifier and I assume that it merges in the Specifier of a higher phrase (cf. Esau 1973: 143), which I will call Intensifier Phrase (IntP). (84) [IntP alle [DP diesei [AgrP kleinen [ArtP ti [NP Autos]]]]]

In this position, it will not trigger Impoverishment and being in a higher phrase it will not undergo Impoverishment. Consequently, the complete feature bundles of alle and diese will be spelled out as strong endings. This part of the proposal also applies to cases with “special” diese ‘this’ and combinations of diese and alle.

 Demonstratives and definite articles as nominal auxiliaries

In Chapter 2, we discussed cases such as the following: (85) a.

dieses mein großes Glück this(strong) my great(strong) happiness ‘this my great happiness’

b. alle diese meine Freunde all(strong) these(strong) my(strong) friends ‘all these my friends’

Following Wood (2007) in that these constructions are special, I suggested there that dieses ‘this’ is also in Spec,IntP. In fact, as (85b) indicates, IndP can be recursive. If so, the strong endings in (85) follow. To sum up, in the stereotypical case, intensifiers are merged in a Spec,IntP where they cannot trigger Impoverishment. Assuming that Impoverishment and Percolation occur locally, these intensifiers cannot undergo Impoverishment either. If this discussion is on the right track, then one could suggest that none of the semantic intensifiers trigger Impoverishment. This is not so as we will see next. 10.3.3.2  Ein  Turning to the second combination, we notice that by themselves, ein and jeder ‘every’ trigger the alternation, (86a–b). Interestingly, when these elements co-occur, ein may trigger the alternation, (86c), but does not have to, (86d): (86) a.

mit einer netten Frau with a(strong) nice(weak) woman

b. mit jeder netten Frau with every(strong) nice(weak) woman c.

mit einer jeden netten Frau with an(strong) every(weak) nice(weak) woman

d. %mit einer jeder netten Frau with a(strong) every(strong) nice(weak) woman

In Roehrs (2007b), I point out that cases like (86c) are more frequent than the ones in (86d). Importantly, ein can also be without an ending when it precedes jeder in the nominative: (87) ein jeder nette Student an every(strong) nice(weak) student ‘each nice student’

On my account, bare ein must be in ArtP in narrow syntax and jeder must move from below the adjective to trigger the alternation on the adjective. In other words,



The strong and weak alternation in German 

both elements raise from their low base-position at some point in the derivation. With two elements raising, I will assume that they are of different structural sizes. While there are other possibilities, let me assume here that jeder is categorially ambiguous between a determiner and a quantifier. As a determiner, it is a head and surfaces in D; as a quantifier, it is a phrasal element surfacing in Spec,CardP. The intensifying element ein is also categorially ambiguous. When it is a determinerlike element, it raises to D; when it is an intensifier, it is phrasal and moves to Spec,IntP. If only heads and phrases can co-occur, then determiner-like ein (X0) can only be combined with quantifier jeder (XP) and intensifier ein (XP) only with determiner jeder (X0). Consider the derivations of the main data. First, I consider the combination of determiner-like ein and quantifier jeder. In (88a), quantifier jeder surfaces in Spec,CardP and does not trigger the alternation.38 As for ein, it moves to D after narrow syntax and does not trigger the alternation either. Both the quantifier and the adjective are spelled out with a strong ending. As to (88b), oblique case-marked ein moves to D in narrow syntax and triggers Impoverishment on both the quantifier and the adjective: (88) a.

ein jeder guter Student an every(strong) nice(strong) student ‘each good student’

b. einem jeden guten Studenten an(strong) every(weak) nice(weak) student ‘(to) each good student’

We turn to the combination of intensifier ein and determiner jeder. In (89a), jeder moves to D in narrow syntax triggering the alternation. The intensifier moves to Spec,IntP after narrow syntax. As for (89b), both elements move to their respective positions in narrow syntax. The adjective undergoes Impoverishment: (89) a.

ein jeder gute Student an every(strong) nice(weak) student ‘each good student’

b. einem jedem guten Studenten an(strong) every(strong) nice(weak) student ‘(to) each good student’

38.  We could also assume that quantifiers are base-generated below adjectives (e.g., in Spec,NumP) and move to Spec,CardP. While this assumption is in the spirit of Chapter 3, note that even under these assumptions, quantifiers cannot trigger Impoverishment.

 Demonstratives and definite articles as nominal auxiliaries

Notice that ein is never in the Specifier or complement position of jeder and vice versa and as such they themselves do not undergo Impoverishment:  both elements move either through head positions like articles or via phrasal adjunction like demonstratives. If this is correct, then we have a straightforward way of ruling out the following example, where all elements are strong, that is, Impoverishment has not occurred: (90) *mit einer jeder guter Frau with an(strong) every(strong) good(strong) woman ‘with each good woman’

Note that intensifier ein and quantifier jeder do not trigger Impoverishment. However, since both elements are phrasal, they cannot be combined under our assumptions. 10.4  Summary and feature realization In the last three sections, we saw that besides the (proto-)typical cases, where determiners trigger Impoverishment but garden-variety adjectives do not, there are elements with varying behavior. I suggested that they are categorially ambiguous between determiners and quantifiers or determiners and intensifiers. As determiners, they trigger the alternation; as quantifiers or intensifiers, they do not. Consider the summary of these elements in table 17: Table 17.  Summary of the Lexical Ambiguities. Intensifier (in Spec,IntP)

Determiner (in DP)

ein alle diese

die, jene wir folgendes jeder ein alle diese

Quantifier (in Spec,CardP) wir folgendes jeder

Besides these lexical ambiguities, we suggested that certain elements may, under certain conditions, move to the DP at different times. In particular, while the determiners der ‘the’, dieser ‘this’, jeder ‘every’, jener ‘those’, and mancher ‘some’ take a weak adjective and always move to the DP in narrow syntax, ein-words, null determiners and personal pronouns take strong and weak endings depending on their specifications for number and case. In other words, they may move



The strong and weak alternation in German 

at different times. Note also that each group contains semantically definite as well as indefinite elements. Let us assume, then, that each of these elements is marked for definiteness syntactically and semantically (cf. Karnowski & Pafel 2004: 177). Consider the following main cases: (91)

Definiteness a. de- b. pronoun c. ein- / Ø d. manch-

[+ synt, + sem] [ synt, + sem] [ synt, sem] [+ synt, sem]

I suggest that it is the syntactic feature for definiteness [+synt] that triggers movement of the element to DP. Finally, let us return to the paradigms in tables 15 and 16 with the intention of stating the endings as realizations of partially underspecified feature bundles. In section 8, we set up a system of abstract syntactic features that were proposed to drive movement. In order to be realized as overt inflections, these features must be interpreted in the morphology. As noted by Müller, Gunkel, & Zifonun (2004), ideally, there is a one-to-one correspondence between syntactic and morphological features. However, they point out that there are some irreducible differences between the two; for instance, the relevance of inflectional classes (e.g., with Russian nouns) for morphology but not syntax (see also Sadler & Spender 2001). Accepting the conclusion that these two types of features are not identical, I will interpret the syntactic features from Section 8 slightly differently for the morphology. Case is fairly straightforward: (92)

Case (morphological) a. [–dat, –gen] b. [–dat, +gen] c. [+dat, -gen] d. [+dat, +gen]

-> nominative -> accusative -> dative -> genitive

Now, with three different genders and one plural that does not differentiate between the three different genders, I collapse gender and number into four features. In particular, I interpret the syntactic feature [+pl] as the neutralization of gender, (93a), and [–pl] as relevance of gender, (93b–d): (93)

Number and gender (morphological) a. [+masc, +fem] -> plural b. [+masc, –fem] -> masculine singular c. [ –masc, +fem] -> feminine singular d. [ –masc, –fem] -> neuter singular

 Demonstratives and definite articles as nominal auxiliaries

With this in place, consider again the strong endings and their relevant syncretisms: Table 18.  Syncretisms of the Strong Adjectival Inflections Feminine

Plural

-m

-r

-n

-s

-r

-r

Masculine

Neuter

Nominative

-r

-s

Accusative

-n

-s

Dative

-m

Genitive

-s

The general syncretisms in table 18, marked by putting the relevant elements in boxes, are disguised by the presence of two instances of -n, put in circles. The first case, the accusative masculine, also holds in the weak paradigm and we stated the readjustment rule (59a). For the second case, the dative plural, let us formulate a fully specified Vocabulary item and assume with Halle & Marantz (1993: 122) that Vocabulary items with the most matching features are inserted, (94a); that is, modulo the slight interface interpretations above, the features of the item have to be not only compatible with those of the relevant syntactic terminal node but also more specific than other Vocabulary items competing for this node. With these cases out of the way, we can make use of underspecification to capture the syncretic forms, (94b–d). Finally, the weak ending is inserted in reduced and checked contexts, (94e). The specifications of the exponents are given in the order of their complexity and insertion: (94) a. b. c. d. e.

[+masc, +fem, +dat, –gen] [–fem, +dat, –gen] [+masc, -fem, –dat] [+fem, +dat] [–fem] [+masc,+fem] or/and [+dat]

-> -n -> -m -> -r -> -r -> -s -> -n

Note that syntactic nodes are always fully specified for features but that Vocabulary items may be underspecified. Recall in this regard that we argued that the nominative/accusative feminine and plural have no overt exponents (cf. the empty cell in table 18). The latter have, minimally, the features [+fem] and [–dat]. Importantly, none of the Vocabulary items in (94) are, minimally, specified for either one or both of these features. Thus, these feature combinations will not be overtly realized. I summarize the main proposals of this chapter.



The strong and weak alternation in German 

11.  Conclusion In this chapter, I turned to some language-specific consequences of the main proposal. I suggested that the strong and weak alternation of the inflection on the adjective is a reflex of determiner movement in German. Specifically, the first part of this chapter illustrated and simplified the inventory of the morphological exponents. I proposed that determiners and adjectives are complex elements where the stem and inflection is combined via the abstract element X. Furthermore, I suggested that all schwa endings are due to an insertion rule. Assuming that singular structurally case-marked determiners move to the DP after syntax, we reduced the number of paradigms from five to two and the number of the exponents from five strong and two weak endings to four strong ones and one weak one. In the second part of this chapter, I turned to the discussion of the distribution of the endings. Having briefly critiqued some traditional assumptions, I proposed that the alternation can be derived by successive-cyclic movement of the determiner from ArtP to DP. Following Sauerland (1996) and Schlenker’s (1999) proposals in certain aspects, I suggested that determiners are marked for the feature [+Impoverishment], which reduces the feature content of other elements (i.e., the adjectives) in a local, successive-cyclic fashion. Reduced feature bundles are spelled out as weak inflections and unaffected sets as strong ones. In the final section, we discussed some apparent variation proposing that some elements are categorially ambiguous between determiners and quantifiers or intensifiers.

chapter 5

Conclusion The main proposal of this book is that determiners (i.e., demonstratives and definite articles) are parallel to auxiliaries in the clause; that is, determiners are auxiliaries in the nominal domain. Extending the parallelism from lexical correspondences to syntactic operations, determiners were argued to be base-generated above the theta domain (ArtP) and to move to a higher position in the extended projection of the noun (DP). Throughout this work, I employed and argued for the following structure of the noun phrase, here illustrating the successive-cyclic movement of a determiner (the) across an adjective (big) in an Agreement Phrase (AgrP) and a numeral (three) in a Cardinal Phrase (CardP). Finally, nouns (car) are argued to move from the Noun Phase (NP) to the Number Phrase (NumP): (1) a. the three big cars b. [DP thei [CardP three thei [AgrP big thei [ArtP thei [NumP cark-s [NP cark]]]]]]

I provided three main arguments for this structure and for this movement. In Chapter 2, I discussed the syntactic distribution of the definite article in the Scandinavian languages from a diachronic and synchronic point of view. On the basis of this distribution, Chapter 3 dealt with the interpretation of the different determiner copies in order to derive the non-/restrictive readings of modifiers in the Scandinavian languages. In Chapter 4, I considered some morpho-syntactic consequences of this proposal for German; in particular, I discussed the strong/weak alternation of adjective endings. Besides giving evidence for (1), these apparently unrelated linguistic phenomena find a uniform account here. I will briefly summarize the main results of each chapter. Chapter 2 provided a formal account of the rise of the suffixal article in the Early Scandinavian noun phrase. The assumption that determiners are basegenerated in a lower phrase and then move to DP (overtly or covertly) offered a straightforward account for the Panchronic Alternation involving the different positions of historically related determiners in Modern Icelandic. After some cross-linguistic discussion, I turned to the explanation of the Scandinavian noun phrase. Assuming that the presence of a modifier induces a separate agreement domain, long-distance agreement between ArtP and DP is blocked. Consequently,

 Demonstratives and definite articles as nominal auxiliaries

determiners have to move to DP to value features on D. The differences between the individual languages were proposed to follow from the different times determiners move and whether or not some semantic components of the determiner can be split off. Chapter 3 proposed that determiners are scope-bearing elements. In general terms, it was suggested that, when modifiers are in their scope, they are restrictive in interpretation and when not, they are non-restrictive. Specifically, assuming movement of the determiner from ArtP to DP, the restrictive interpretation was accounted for by interpreting the determiner in its derived position and the nonrestrictive one by interpreting the determiner in its base-position. This account provided a straightforward solution to the “Partee-Chomsky debate”. Making a distinction between syntactic and non-integrated appositives, I argued that non-restrictive modifiers are propositions that are associated with their hosting proposition by conjunction, in a model of multiple semantic spell-out. Chapter 4 discussed the strong/weak alternation of adjective endings in German. In order to account for the relevant generalization (cf. the Principle of Monoinflection), the mechanism of Impoverishment was employed. Assuming that determiners are complex elements triggering Impoverishment (i.e., feature reduction), the successive-cyclic movement of a determiner reduces the feature bundles of the adjectives, which are eventually spelled-out as weak endings. In order to account for apparent exceptions, I proposed that certain elements are lexically ambiguous between determiners and quantifiers or intensifiers and that singular structurally case-marked elements are special:  ein-words and personal pronouns move from ArtP to DP after syntax. In the course of the discussion, we provided evidence for movement at all levels of derivation. In particular, we discussed two types of endings in German, where oblique and plural features have to be checked and structural and singular features have to be valued. Assuming that Agree holds at LF, the former involves a (generalized) EPP feature, which triggers movement in overt syntax, but the latter does not. Furthermore, in order to resolve semantic type mismatch, the suffixal determiner -inn in Icelandic must move at LF providing evidence for movement at this level. Finally, assuming that ein-words and personal pronouns in German support D and other elements, we find justification for the claim that certain singular structurally case-marked elements move to the DP at PF.

References Abney, Steven. 1987. The English noun phrase in its sentential aspect. Ph.D. Dissertation, MIT. Abraham, Werner. 1997. The interdependence of case, aspect and referentiality in the history of German: The case of the verbal genitive. In Ans van Kemenade & Nigel Vincent (Eds), Parameters of morphosyntactic change, 29–61. Cambridge: CUP. Abraham, Werner. 2007. The discourse-functional crystallization of the historically original demonstrative. In Elisabeth Stark, Elisabeth Leiss and Werner Abraham (Eds), Nominal determination. Typology, context constraints, and historical emergence, 21–48. Amsterdam: John Benjamins. Alexiadou, Artemis. 2001. Adjective syntax and noun raising: Word order asymmetries in the DP as the result of adjective distribution. Studia Linguistica 55(3): 217–48. Alexiadou, Artemis. 2003. On double definiteness. In Lars-Olof Delsing, C.  Falk, Gunlög Josefsson & Halldór Ármann Sigurðsson (Eds), Grammar in Focus, Vol 2. Festschrift for Christer Platzack, 9–16. Lund: Dept. of Scandinavian Languages, University of Lund. Alexiadou, Artemis. 2004. On the development of possessive determiners:  Consequences for DP structure. In Eric Fuß & Carola Trips (Eds), Diachronic clues to synchronic grammar, 31–58. Amsterdam: John Benjamins. Alexiadou, Artemis & Chris Wilder. 1998. Adjectival modification and multiple determiners. In Artemis Alexiadou & Chris Wilder (Eds), Possessors, Predicates and Movement in the Determiner Phrase, 303–332. Amsterdam: John Benjamins. Alexiadou, Artemis, Paul Law, André Meinunger & Chris Wilder. 2000. Introduction to The Syntax of Relative Clauses, 1–51. Amsterdam: John Benjamins. Anderson, Bruce. 2002. The fundamental equivalence of native and interlanguage grammars: Evidence from argument licensing and adjective position. Ph.D. Dissertation, Indiana University. Anderssen, Merete. 2005. The acquisition of compositional definiteness in Norwegian. Ph.D. Dissertation, University of Tromsø. Andersson, Erik. 1994. Swedish. In Ekkehard König & Johan van der Auwera (Eds), The Germanic languages, 271–312. London: Routledge. Authier, J.-Marc & Lisa Reed. 2005. The diverse nature of noninterrogative Wh. Linguistic Inquiry 36(4): 635–47. Bach, Emmon & Robin Cooper. 1978. The NP-S analysis of relative clauses and compositional semantics. Linguistics and Philosophy 2: 145–150. Baker, Mark. 1985. The mirror principle and morphosyntactic explanation. Linguistic Inquiry 16(3): 373–415. Baker, Mark. 1988. Incorporation:  A theory of grammatical function changing. Chicago IL:  University of Chicago Press. Baker, Mark C. 2003. Lexical categories. Verbs, nouns, and adjectives [Cambridge Studies in Linguistics 102]. Cambridge: CUP. Baltin, Mark. 2002. Movement to the higher V is remnant movement. Linguistic Inquiry 33(4): 653–659.

 Demonstratives and definite articles as nominal auxiliaries Barbiers, Sjef. 1992. Adjectives as auxiliaries of the noun phrase. In Reineke Bok-Bennema & Roeland van Hout (Eds), Linguistics in the Netherlands 9, 13–24. Amsterdam: John Benjamins. Barker, Chris. 1995. Possessive descriptions. Standford CA: CSLI. Batllori, Montse & Francesc Roca. 2000. The value of definite determiners from Old Spanish to Modern Spanish. In Susan Pintzuk, George Tsoulas & Anthony Warner (Eds), Diachronic syntax, 241–54. Oxford: OUP. Bayer, Josef, Markus Bader & Michael Meng. 2001. Morphological underspecification meets oblique case: Syntactic and processing effects in German. Lingua 111: 465–514. Bennis, Hans, Norbert Corver & Marcel den Dikken. 1998. Predication in nominal phrases. The Journal of Comparative Germanic Linguists 1: 85–117. Bernstein, Judy B. 1993. Topics in the syntax of nominal structure across Romance. Ph.D. Dissertation, The City University of New York. Bernstein, Judy B. 1997. Demonstratives and reinforcers in Romance and Germanic languages. Lingua 102: 87–113. Bernstein, Judy B. 2001a. Focusing the ‘right’ way in Romance determiner phrases. Probus 13: 1–29. Bernstein, Judy B. 2001b. The DP hypothesis:  Identifying clausal properties in the nominal domain. In Mark Baltin & Chris Collins (Eds), The Handbook of Contemporary Syntactic Theory, 536–561. Malden MA: Blackwell. Bhatt, Christa. 1990. Die syntaktische Struktur der Nominalphrase im Deutschen [Studien zur deutschen Grammatik 38]. Tübingen: Gunter Narr. Blevins, James B. 1995. Syncretism and paradigmatic opposition. Linguistics and Philosophy 18: 113–152. Bobaljik, Jonathan David & Samuel Brown. 1997. Interarboreal operations:  Head movement and the extension requirement. Linguistic Inquiry 28(2): 345–356. Bobaljik, Jonathan David & Susi Wurmbrand. 2005. The domain of agreement. Natural Language & Linguistic Theory 23(4): 809–865. Boeckx, Cedric & Sandra Stjepanović. 2001. Head-ing toward PF. Linguistic Inquiry 32(2): 345–355. Börjars, Kersti. 1998. Feature distribution in Swedish noun phrases [Publications of the Philological Society 32]. Boston MA: Blackwell. Börjars, Kersti & Mark Donohue. 2000. Much ado about nothing:  Features and zeroes in Germanic noun phrases. Studia Linguistica 54(3): 309–353. Borsley, Robert D. 1992. More on the difference between English restrictive and non-restrictive relative clauses. Journal of Linguistics 28: 139–148. Bošković, Željko. 2004. Be careful where you float your quantifiers. Natural Language & Linguistic Theory 22(4): 681–742. Bošković, Željko. 2005. On the locality of Left Branch Extraction and the structure of NP. Studia Linguistica 59(1): 1–45. Bouchard, Denis. 1998. The distribution and interpretation of adjectives in French: A consequence of Bare Phrase Structure. Probus 10: 139–183. Bouchard, Denis. 2002. Adjectives, number and interfaces. Why languages vary. Amsterdam: Elsevier. Boucher, Paul. 2003. Determiner phrases in Old and Modern French. In Martine Coene & Yves D’hulst (Eds), From NP to DP, Vol.  1:  The syntax and semantics of noun phrases, 47–69. Amsterdam: John Benjamins.



References 

Bowers, John. 1987. Extended X-bar theory, the ECP, and the Left Branch Condition. In Proceedings of the West Coast Conference on Formal Linguistics 6, 47–62. Stanford CA: Stanford Linguistics Association. Brame, Michael. 1981. The general theory of binding and fusion. Linguistic Analysis 7(3): 277–325. Brame, Michael. 1982. The head-selector theory of lexical specification and the nonexistence of coarse categories. Linguistic Analysis 10(4): 321–325. Braunmüller, Kurt. 1999. Die skandinavischen Sprachen im Überblick, 2nd Edn. [UTB 1635]. Tübingen: Franke. Brugè, Laura. 1996. Demonstrative movement in Spanish: A comparative approach. University of Venice Working Papers in Linguistics 6(1): 1–61. Brugè, Laura. 2002. The positions of demonstratives in the Extended Nominal Projection. In Guglielmo Cinque (Ed.), Functional structure in DP and IP: The cartography of syntactic structures, Vol.1, 15–53. Oxford: OUP. Campbell, Richard. 1996. Specificity operators in Spec,DP. Studia Linguistica 50(2): 161–188. Cardinaletti, Anna. 1994. On the internal structure of pronominal DPs. The Linguistic Review 11: 195–219. Cardinaletti, Anna. 1998. On the deficient/strong opposition in possessive systems. In Artemis Alexiadou & Chris Wilder (Eds), Possessors, predicates and movement in the determiner phrase, 17–53. Amsterdam: John Benjamins. Carstens, Vicky M. 1991. The morphology and syntax of determiner phrases in Kiswahili. Ph.D. Dissertation, University of California, Los Angeles. Chomsky, Noam. 1955 [1975]. The logical structure of linguistic theory. New York NY: Plenum. Chomsky, Noam. 1975. Questions of form and interpretation. Linguistic Analysis 1(1): 75–109. Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris. Chomsky, Noam. 1986. Barriers. Cambridge MA: The MIT Press. Chomsky, Noam. 1991. Some notes on economy of derivation and representation. (Reprinted in Noam Chomsky. 1995. The minimalist program, 129–166). Chomsky, Noam. 1995. Minimalist program. Cambridge MA: The MIT Press. Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Roger Martin, David Micheals & Juan 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 Michael Kenstowicz (Ed.), Ken Hale. A life in language, 1–52. Cambridge MA: The MIT Press. Cinque, Guglielmo. 1981. On the theory of relative clauses and markedness. The Linguistic Review 1: 247–294. Cinque, Guglielmo. 1994. On the evidence for partial N-movement in the Romance DP. In Guglielmo Cinque, Jan Koster, Jean-Yves Pollock, Luigi Rizzi & Raffaella Zanuttini (Eds), Paths towards Universal Grammar. Studies in Honor of Richard S. Kayne, 85–110. Washington DC: Georgetown University Press. Cinque, Guglielmo. 1999. Adverbs and functional projections:  A cross-linguistic perspective. New York: Oxford University Press. Cinque, Guglielmo. 2003. The dual source of adjectives and XP- vs. N-raising in the Romance DP. Talk given at the CASTL Kick-off Conference, University of Tromsø. Cinque, Guglielmo. 2005. Deriving Greenberg’s universal 20 and its exceptions. Linguistic Inquiry 36(2): 315–332.

 Demonstratives and definite articles as nominal auxiliaries Cinque, Guglielmo. To appear. Two types of nonrestrictive relatives. In Proceedings of the Colloque de Syntaxe et Sémantique de Paris 2007. Coene, Martine & Yves D’hulst. 2003. Introduction: The syntax and semantics of noun phrases. Theoretical background. In Martine Coene & Yves D’hulst (Eds), From NP to DP, Vol. 1: The syntax and semantics of noun phrases, 1–33. Amsterdam: John Benjamins. Corver, Norbert. 1991. Evidence for DegP. Proceedings of NELS 21: 33–47. Corver, Norbert. 1997. The internal structure of the Dutch extended adjectival projection. Natural Language & Linguistic Theory 15: 289–368. Corver, Norbert. 2008. Uniformity and diversity in the syntax of evaluative vocatives. The Journal of Comparative Germanic Linguistics 11(1): 43–93. Cowper, Elizabeth A. 1987. Pied piping, feature perculation and the structure of the noun phrase. Canadian Journal of Linguistics 32(4): 321–338. Crisma, Paola. 1996. On the configurational nature of adjectival modification. In Karen Zagona (Ed.), Grammatical theory and Romance linguistics, 59–71. Amsterdam: John Benjamins. Darski, Józef. 1979. Die Adjektivdeklination im Deutschen. Sprachwissenschaft 4: 190–205. Delorme, Evelyne & Ray C. Dougherty. 1972. Appositive NP Constructions: We, the men; we men; I, a man; ETC. Foundations of Language 8: 2–29. Delsing, Lars-Olof. 1988. The Scandinavian noun phrase. Working Papers in Scandinavian Syntax 42: 57–79. Delsing, Lars-Olof. 1991. Quantification in the Swedish noun phrase. Working Papers in Scandinavian Linguistics 47: 89–116. Delsing, Lars-Olof. 1993a. On attributive adjectives in Scandinavian and other languages. Studia Linguistica 47(2), 105–125. Delsing, Lars-Olof. 1993b. The internal structure of noun phrases in the Scandinavian languages. A comparative study. Ph.D. Dissertation, University of Lund. Delsing, Lars-Olof. 1996. Nominalfrassyntax i skandinaviska dialekter. Nordica Bergensia 9: 24–74. Demske, Ulrike. 2001. Merkmale und Relationen:  Diachrone Studien zur Nominalphrase des Deutschen. Berlin: Walter de Gruyter. Diesing, Molly. 1992. Indefinites. Cambridge MA: The MIT Press. Diessel, Holger. 1999. Demonstratives. Form, function, and grammaticalization. Amsterdam:  John Benjamins. Dikken, Marcel den. 2006. Relators and linkers. The syntax of predication, predicate inversion, and copulas. Cambridge MA: The MIT Press. Dimitrova-Vulchanova Mila & Giuliana Giusti. 1998. Fragments of Balkan nominal structure. In Artemis Alexiadou & Chris Wilder (Eds), Possessors, predicates and movement in the determiner phrase, 333–360. Amsterdam: John Benjamins. Dryer, Matthew S. 1989. Article-noun order. Chicago Linguistic Society 25: 83–97. Duden. 1995. Die Grammatik, Vol. 4. Mannheim: Dudenverlag. Duden. 2007. Richtiges und gutes Deutsch. Vol. 9. Mannheim: Dudenverlag. Duffield, Nigel. 1996. On structural invariance and lexical diversity in VSO languages: Arguments from Irish noun phrases. In Robert D. Borsley & Ian Roberts (Eds), The syntax of the Celtic languages. A comparative perspective, 314–40. Cambridge: CUP. Ebert, Karen Heide. 1971. Referenz, Sprechsituation und die bestimmten Artikel in einem Nordfriesischen Dialekt. Bräist/Bredstedt. Eisenberg, Peter. 1998. Grundriß der deutschen Grammatik, Band 1: Das Wort. Stuttgart: Metzler. Eisenberg, Peter. 1999. Grundriß der deutschen Grammatik, Band 2: Der Satz. Stuttgart: Metzler.



References 

Embick, David & Rolf Noyer. 2001. Movement operations after syntax. Linguistic Inquiry 32(4): 555–595. Emonds, Joseph E. 1978. The verbal complex V’-V in French. Linguistic Inquiry 9(2): 151–175. Emonds, Joseph E. 1979. Appositive relatives have no properties. Linguistic Inquiry 10(2): 211–243. Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22(1): 1–15. Esau, Helmut. 1973. Form and function of German adjective endings. Folia Linguistica 6(1/2): 136–45. Faarlund, Jan Terje. 2004. The syntax of Old Norse. Oxford: OUP. Faarlund, Jan Terje. 2007. From clitic to affix: The Norwegian definite article. Working Papers in Scandinavian Syntax 79: 21–46. Fabb, Nigel. 1990. The difference between English restrictive and nonrestrictive relative clauses. Journal of Linguistics 26: 57–78. Fanselow, Gisbert. 1986. On the sentential nature of prenominal adjectives in German. Folia Linguistica 20(3–4): 341–380. Fanselow, Gisbert. 1988. Aufspaltung von NPn und das Problem der “freien” Wortstellung. Linguistische Berichte 114: 91–113. Felber, Sarah & Dorian Roehrs. 2004. *So weird a baffling construction. Snippets 8: 7–9. Fitzpatrick, Justin M. 2006. Deletion through movement. Natural Language & Linguistic Theory 24(2): 399–431. Fodor, Janet Dean & Ivan A. Sag. 1982. Referential and Quantificational Indefinites. Linguistics and Philosophy 5: 335–398. Franks, Steven, & Asya Pereltsvaig. 2004. Functional categories in the nominal domain. In Olga Arnaudova, Wayles Browne, Maria Luisa Rivero & Danijela Stojanovic (Eds), Proceedings of FASL 12. The Ottawa meeting 2003, 109–28. Ann Arbor MI:  University of Michigan Press. Fukui, Naoki & Margaret Speas. 1986. Specifiers and projection. MIT Working Papers in Linguistics 8: 128–72. Gallmann, Peter. 1996. Die Steuerung der Flexion in der DP. Linguistische Berichte 164: 283–314. Gallmann, Peter. 1997. Zur Morphosyntax der Eigennamen im Deutschen. In Elisabeth Löbel & Gisa Rauh (Eds), Lexikalische Kategorien und Merkmale, 73–86. Tübingen: Niemeyer. Gallmann, Peter. 2004. Feature Sharing in DPs. In Gereon Müller, Lutz Gunkel, and Gisela Zifonun (Eds.), Explorations in Nominal Inflection, 121–160. Berlin: Mouton de Gruyter. Gavruseva, Elena. 2000. On the syntax of possessor extraction. Lingua 110: 743–772. Gelderen, Elly van. 2004a. Economy, innovation, and prescriptivism: From Spec to Head and Head to Head. Journal of Comparative Germanic Linguistics 7: 59–98. Gelderen, Elly van. 2004b. Grammaticalization as economy. Amsterdam: John Benjamins. Gelderen, Elly van. 2007. The definiteness cycle in Germanic. Journal of Germanic Linguistics 19(4): 275–308. Giorgi, Alessandra & Giuseppe Longobardi. 1991. The syntax of noun phrases. Configuration, parameters and empty categories. Cambridge: CUP. Giusti, Giuliana. 1994. Enclitic articles and double definiteness:  A comparative analysis of nominal structure in Romance and Germanic. The Linguistic Review 11: 241–255. Giusti, Giuliana. 1996. Is there a FocusP and a TopicP in the noun phrase structure? University of Venice Working Papers in Linguistics 6(2): 105–128. Giusti, Giuliana. 1997. The categorial status of determiners. In Liliane Haegeman (Ed.), The new comparative syntax, 95–123. London: Longman.

 Demonstratives and definite articles as nominal auxiliaries Giusti, Giuliana. 2001. The birth of a functional category: From Latin ILLE to the Romance article and personal pronoun. In Guglielmo Cinque & Giampaolo Salvi (Eds), Current studies in Italian syntax. Essays offered to Lorenzo Renzi, 157–171. Amsterdam: Elsevier. Giusti, Giuliana. 2002. The functional structure of noun phrases. A bare phrase structure approach. In Guglielmo Cinque (Ed.), Functional structure in DP and IP: The cartography of syntactic structures, Vol.1, 54–90. Oxford: OUP. Grewendorf Günther. 1991. Aspekte der deutschen Syntax. Eine Rektions-Bindungs-Analyse, 2nd Edn. Tübingen: Gunter Narr. Grimm, Jacob. 1870. Deutsche Grammatik, Vol 1, 2nd Edn. Berlin. Grimshaw, Jane. 1991. Extended projections. Ms, Brandeis University. Grohmann, Kleanthes K. & Liliane Haegeman. 2003. Resuming reflexives. Proceedings of the 19th Scandinavian Conference on Linguistics. Nordlyd 31(1): 46–62. Grohmann, Kleanthes K. & Phoevos Panagiotidis. 2004. Demonstrative doubling. Ms, University of Cyprus and Cyprus College. Haegeman, Liliane. 2004. DP-periphery and clausal periphery:  Possessor doubling in West Flemish. In David Adger, Cécile de Cat & George Tsoulas (Eds), Peripheries. Syntactic edges and their effects, 211–240. Dordrecht: Kluwer. Haider, Hubert. 1992. Die Struktur der Nominalphrase. Lexikalische und funktionale Strukturen. In Ludger Hoffmann (Ed.), Deutsche Syntax. Ansichten und Aussichten. Institut für deutsche Sprache Jahrbuch 1991, 304–333. Berlin: Walter de Gruyter. Haider, Hubert. 1993. Deutsche Syntax – Generativ. Tübingen: Narr. Haiman, John & Paola Benincà. 1992. The Rhaeto-Romance languages. London: Routledge. Halle, Morris. 1997. Distributed Morphology:  Impoverishment and Fission. MIT Working Papers in Linguistics 30: 425–449. Halle, Morris & Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In Kenneth Hale & Samuel K. Keyser (Eds), The view from building 20. Essays in Honor of Sylvain Bromberger, 111–176. Cambridge MA: The MIT Press. Hankamer, Jorge & Line Mikkelsen. 2002. A morphological analysis of definite nouns in Danish. Journal of Germanic Linguistics 14(2): 137–175. Hankamer, Jorge & Line Mikkelsen. 2005. When movement must be blocked. A reply to Embick and Noyer. Linguistic Inquiry 36(1): 85–125. Harbert, Wayne. 2007. The Germanic languages. Cambridge: CUP. Heim, Irene & Angelika Kratzer. 1998. Semantics in generative grammar. Cambridge MA: Blackwell. Helbig, Gerhard & Joachim Buscha. 2001. Deutsche Grammatik. Ein Handbuch für den Ausländerunterricht. Berlin: Langenscheidt. Hellan, Lars. 1986. The headedness of NPs in Norwegian. In Pieter Muysken & Henk van Riemsdijk (Eds), Features and projections, 89–122. Dordrecht: Foris. Holmberg, Anders. 1993. On the structure of predicate NP. Studia Linguistica 47(2): 126–138. Holmberg, Anders. 1999. Comments on Cardinaletti and Starke:  ‘The typology of structural deficiency’. In Henk van Riemsdijk (Ed.), Clitics in the languages of Europe, 263–266. Berlin: Mouton de Gruyter. Horrocks, Geoffrey & Melita Stavrou. 1987. Bounding theory and Greek syntax: Evidence for wh-movement in NP. Journal of Linguistics 23: 79–108. Howe, Stephen. 1996. The personal pronouns in the Germanic languages: A study of personal pronoun morphology and change in the Germanic languages from the first records to the present day. Berlin: de Gruyter.



References 

Hudson, Richard A. 1984. Word grammar. Oxford: Basil Blackwell. Hudson, Richard A. 1987. Zwicky on heads. Journal of Linguistics 23: 109–132. Hughes, Michael. 2007. Phrasal concord in Icelandic. Journal of Germanic Linguistics 19(3): 171–204. Hulsey, Sarah & Uli Sauerland. 2006. Sorting out relative clauses. Natural Language Semantics 14: 111–137. Jackendoff, Ray S. 1977. X’ syntax: A study of phrase structure. Cambridge MA: The MIT Press. Jacobs, Neil G. 2005. Yiddish. A linguistic introduction. Cambridge: CUP. Josefsson, Gunlög. 1999. On the semantics and syntax of Scandinavian pronouns and object shift. In Henk van Riemsdijk (Ed.), Clitics in the languages of Europe, 731–57. Berlin:  Mouton de Gruyter. Julien, Marit. 2002. Determiners and word order in Scandinavian DPs. Studia Linguistica 56: 264–315. Julien, Marit. 2005. Nominal phrases from a Scandinavian perspective. Amsterdam: John Benjamins. Kallulli, Dalina & Antonia Rothmayr. 2008. The syntax and semantics of indefinite determiner doubling constructions in varieties of German. The Journal of Comparative Germanic Linguistics 11(1): 95–136. Karnowski, Paweł & Jürgen Pafel. 2004. A topological schema for noun phrases in German. In Gereon Müller, Lutz Gunkel & Gisela Zifonun (Eds), Explorations in nominal inflection, 161–188. Berlin: Mouton de Gruyter. Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge MA: The MIT Press. Kaisse, Ellen M. 1981. Appositive relatives and the cliticization of who. Chicago Linguistic Society 17: 108–115. Kester, Ellen-Petra. 1993. The inflectional properties of Scandinavian adjectives. Studia Linguistica 47(2): 139–153. Kester, Ellen-Petra. 1996a. Adjectival inflection and the licensing of empty categories in DP. Journal of Linguistics 32: 57–78. Kester, Ellen-Petra. 1996b. The nature of adjectival inflection. Ph.D. Dissertation, Utrecht University. Kim, Young-Kook. 1997. Agreement phrases in DP. UCL Working Papers in Linguistics 9: 1–24. Krause, Wolfgang & Herbert Jankuhn. 1966. Die Runeninschriften im älteren Futhark. Göttingen: Vandenhoeck and Ruprecht. Lamarche, Jacques. 1991. Problems for N0-movement to Num-P. Probus 3(2): 215–236. Lamontagne, Greg & Lisa Travis. 1987. The syntax of adjacency. WCCFL 6: 173–186. Laenzlinger, Christopher. 2005. French adjective ordering:  Perspectives on DP-internal movement types. Lingua 115(5): 645–689. Lasnik, Howard. 2001. A note on the EPP. Linguistic Inquiry 32(2): 356–62. Lasnik, Howard & Juan Uriagereka with Cedric Boeckx. 2005. A course in minimalist syntax. Foundations and prospects. Malden MA: Blackwell. Leiss, Elisabeth. 1994. Die Entstehung des Artikels im Deutschen. Sprachwissenschaft 19: 307–319. Leko, Nedžad. 1988. X-Bar theory and internal structure of NPs. Lingua 75: 135–169. Leu, Thomas. 2001. A sketchy note on the article-modifier relation. Evidence from Swiss-German which sheds light on Greek determiner spreading. Generative Grammar in Geneva 2: 55–69. Lohndal, Terje. 2007. On the structure and development of nominal phrases in Norwegian. In Elisabeth Stark, Elisabeth Leiss & Werner Abraham (Eds), Nominal determination. Typology, context constraints, and historical emergence, 285–308. Amsterdam: John Benjamins.

 Demonstratives and definite articles as nominal auxiliaries Longobardi, Giuseppe. 1994. Reference and proper names: A theory of N-movement in syntax and logical form. Linguistic Inquiry 25(4): 609–665. Longobardi, Giuseppe. 1996. The syntax of N-raising:  A minimalist theory. OTS Working Papers, Utrecht. Longobardi, Giuseppe. 2001a. How comparative is semantics? A unified parametric theory of bare nouns and proper names. Natural Language Semantics 9: 335–369. Longobardi, Giuseppe. 2001b. The structure of DPs:  Some principles, parameters, and problems. In Mark Baltin & Chris Collins (Eds), The handbook of contemporary syntactic theory, 562–603. Malden MA: Blackwell. Longobardi, Giuseppe. 2005. Toward a unified grammar of reference. Zeitschrift für Sprachwissenschaft 24: 5–44. Lühr, Rosemarie. 1991. Die deutsche Determinansphrase aus historischer Sicht. Zur Flexion von der, die, das als Demonstrativpronomen, Relativpronomen und Artikel. Beiträge zur Geschichte der deutschen Sprache und Literatur 113(2): 195–211. Lyons, Christopher. 1977. The demonstratives of English. Belfast Working Papers in Language and Linguistics 2(5): 1–23. Lyons, Christopher. 1999. Definiteness. Cambridge: CUP. Mallén, Enrique. 1991. Noun phrase structure, clitic-doubling, and extraction. Linguistische Berichte 134: 276–309. Mallén, Enrique. 2002. On the distribution of restrictive vs. nonrestrictive adjectives in Germanic and Romance. In Irgmengard Rauch & Gerald F.  Carr (Eds), New insights in Germanic linguistics III. Berkeley Insights in Linguistics and Semiotics  52:  177–210 New York: Peter Lang. Martín, Juan. 1995. On the syntactic structure of Spanish noun phrases. Ph.D. Dissertation, University of Southern California. Matushanksy, Ora. 2006. Head movement in linguistic theory. Linguistic Inquiry 37(1): 69–109. Meyer, Rudolf. 1967. Zur Morphologie und Sprachgeographie des Artikels im Schweizerdeutschen. Frauenfeld: Huber & Co. Mohammad, M.A. 1988. On the parallelism between IP and DP. WCCFL: 241–254. Müller, Gereon. 2002. Syntactisch determinierter Kasuswegfall in der deutschen NP. Linguistische Berichte 189: 89–114. Müller, Gereon, Lutz Gunkel & Gisela Zifonun. 2004. Introduction in Explorations in Nominal Inflection, 1–20. Berlin: Mouton de Gruyter. Noguchi, Tohru. 1997. Two types of pronouns and variable binding. Language 73(4): 770–797. Noreen, Adolf. 1970. Altnordische Grammatik I. Tübingen: Max Niemeyer. Nygaard, Marius. 1906. Norrøn syntax. Oslo: H. Aschehoug & Co. Olsen, Susan. 1991. Die deutsche Nominalphrase as ‘Determinansphrase’. In Susan Olsen & Gisbert Fanselow (Eds), DET, COMP und INFL. Zur Syntax funktionaler Kategorien und grammatischer Funktionen, 35–56. Tübingen: Niemeyer. Oubouzar, Erika. 1997. Zur Frage der Herausbildung eines bestimmten und eines unbestimmten Artikels im Althochdeutschen. Cahiers d’études germaniques 32: 161–175. Panagiotidis, Phoevos. 2000. Demonstrative determiners and operators:  The case of Greek. Lingua 110: 717–742. Panagiotidis, Phoevos. 2002. Pronouns, clitics and empty nouns. ‘Pronominality’ and licensing in syntax. Amsterdam: John Benjamins. Partee, Barbara H. 1976. Some transformational extensions of Montague Grammar. In Barbara H. Partee (Ed.), Montague grammar, 51–76. New York NY: Academic Press.



References 

Pereltsvaig, Asya. 2007. On the universality of DP:  A view from Russian. Studia Linguistica 61(1): 59–94. Perlmutter, David M. 1970. On the article in English. In Manfred Bierwisch & Karl Erich Heidolph (Eds), Progress in linguistics. A collection of papers, 233–248. The Hague: Mouton. Perridon, Harry. 1996. Noun phrases in Runic Swedish. In Kjartan G. Ottósson et al. (Eds), The Nordic languages and modern linguistics, Vol. 9: Proceedings of the Ninth International Conference of Nordic and General Linguistics, 248–61. Oslo: Novus. Philippi, Julia. 1997. The rise of the article in the Germanic languages. In Ans van Kemenade & Nigel Vincent (Eds), Parameters of morphosyntactic change, 62–93. Cambridge: CUP. Plank, Frans. 2003. Double articulation. In Frans Plank (Ed.), Noun phrase structure in the languages of Europe [Empirical Approaches to Language Typology, EUROTYP 20–7], 337–395. Berlin: Mouton de Gruyter. Platzack, Christer. 1997. A representational account of restrictive and non-restrictive relatives: The case of Swedish. Working Papers in Scandinavian Syntax 59: 65–96. Platzack, Christer. 2000. A complement-of-N0 account of restrictive and non-restrictive relatives: The case of Swedish. In Artemis Alexiadou, Paul Law, André Meinunger & Chris Wilder (Eds), The syntax of relative clauses, 265–308. Amsterdam: John Benjamins. Platzack, Christer. 2002. Relativization in the Germanic languages, with particular emphasis on Scandinavian. In Patricia Poussa (Ed.), Relativisation on the North Sea littoral, 77–96. München: Lincom. Pollock, Jean-Yves. 1989. Verb movement, universal grammar, and the structure of IP. Linguistic Inquiry 20(3): 365–424. Postal, Paul M. 1966. On the so-called pronouns in English. In F. Dinneen (Ed.), Nineteenth monograph on language and linguistics, 177–206. Washington DC: Georgetown University Press. Progovac, Ljiljana. 1998. Determiner phrase in a language without determiners. Journal of Linguistics 34: 165–179. Radford, Andrew. 1993. Head-hunting: On the trail of the nominal Janus. In Greville G. Corbett, Norman M.  Fraser & Scott McGlashan (Eds), Heads in grammatical theory, 73–113. Cambridge: CUP. Riemsdijk, Henk van. 1989. Movement and regeneration. In Paola Benincà (Ed.), Dialect variation and the theory of grammar, 105–136. Dordrecht: Foris. Riemsdijk, Henk van. 1998. Head movement and adjacency. Natural Language & Linguistic Theory 16: 633–678. Rijkhoff, Jan. 2002. The noun phrase. Oxford: OUP. Ritter, Elizabeth. 1988. A head-movement approach to construct-state noun phrases. Linguistics 26: 909–929. Ritter, Elizabeth. 1991. Two functional categories in noun phrases:  Evidence from Modern Hebrew. In Susan D. Rothstein (Ed.), Perspectives on phrase structure: Heads and licensing [Syntax and Semantics 25], 37–62. San Diego CA: Academic Press. Ritter, Elizabeth. 1995. On the syntactic category of pronouns and agreement. Natural Language & Linguistic Theory 13: 405–443. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (Ed.), Elements of grammar, 281–337. Dordrecht: Kluwer. Rizzi, Luigi. 2001. Relativized minimality effects. In Mark Baltin & Chris Collins (Eds), The handbook of contemporary syntactic theory, 89–110. Malden MA: Blackwell. Roberts, Ian G. 2005. Principles and parameters in a VSO language:  A case study in Welsh. Oxford: OUP.

 Demonstratives and definite articles as nominal auxiliaries Roehrs, Dorian. 2002. The boundary between morphology and syntax: Determiners Move into the determiner phrase. IULC Working Papers Online 2: 1–46. Roehrs, Dorian. 2005a. Pronouns are determiners after all. In Marcel den Dikken & Christina Tortora (Eds), The function of function words and functional categories, 251–285. Amsterdam: John Benjamins. Roehrs, Dorian. 2005b. Possessives consists of heads and complements:  Some notes. In Dorian Roehrs, Ock-Hwan Kim & Yoshihisa Kitagawa (Eds), Syntax and beyond [Indiana University Working Papers in Linguistics 5], 117–247. Bloomington IN: IULC. Roehrs, Dorian. 2006a. The morpho-syntax of the Germanic noun phrase: Determiners MOVE into the determiner phrase. Ph.D. Dissertation, Indiana University. Roehrs, Dorian. 2006b. Pronominal noun phrases, number specifications, and null nouns. In Jutta M.  Hartmann & László Molnárfi (Eds), Comparative Studies in Germanic Syntax:  From Afrikaans to Zurich German [Linguistik Aktuell/Linguistic Today 97], 143–180. Amsterdam: John Benjamins. Roehrs, Dorian. 2007a. Split-NPs as an instance of sideward movement. In Tatjana Scheffler, Joshua Tauberer, Aviad Eilam & Laia Mayol (Eds), Proceedings of the 30th Annual Penn Linguistics Colloquium. Penn Working Papers in Linguistics 13(1): 323–336. [also available at: http://ling.auf.net/lingBuzz/000275] Roehrs, Dorian. 2007b. Complex determiners: A case study of German ein jeder. Ms, University of North Texas. Roehrs, Dorian. 2008. Something inner-  and cross-linguistically different. The Journal of Comparative Germanic Linguistics 11(1): 1–42. Roehrs, Dorian. To appear. Inflectional parallelism with German adjectives. Interdisciplinary Journal of Germanic Linguistics and Semiotic Analysis. Roehrs, Dorian & Christopher D.  Sapp. 2006. Different word orders in the early North Germanic DP: The rise of the suffixal article. In Michal Temkin Martínez, Asier Alcázar & Roberto Mayoral Hernández (Eds), Proceedings of the Western Conference on Linguistics 2004, 290–301. Fresno CA: California State University. Roehrs, Dorian & Michael Putnam. 2008. Complex demonstratives. Ms, University of North Texas and Carson-Newman College. Rubin, Edward J. 1996. The transparent syntax and semantics of modifiers. Proceedings of the 15th WCCFL, 429–439. Sadler, Louisa & Andrew Spencer. 2001. Syntax as an exponent of morphological features. In Geert Booij & Jaap van Marle (Eds), Yearbook of Morphology 2000, 71–96. Dordrecht: Kluwer. Safir, Ken. 1986. Relative clauses in a theory of binding and levels. Linguistic Inquiry 17(4): 663–689. Sag, Ivan A. 1997. English relative clause constructions. Journal of Linguistics 33: 431–483. Saito, Mamuro & Keiko Murasugi. 1999. Subject predication within IP and DP. In Kyle Johnson & Ian Roberts (Eds), Beyond principles and parameters. Essays in memory of Osvaldo Jaeggli, 167–188. Dordrecht: Kluwer. Sánchez, Liliana. 1996. Word order, predication and agreement in DPs in Spanish, Southern Quechua and southern Andean bilingual Spanish. In Karen Zagona (Ed.), Grammatical theory and Romance linguistics, 209–218. Amsterdam: John Benjamins. Santelmann, Lynn. 1993. The distribution of double determiners in Swedish: Den support in D0. Studia Linguistica 47(2): 154–176. Sauerland, Uli. 1996. The late insertion of Germanic inflection. Ms, MIT.



References 

Sauerland, Uli & Paul Elbourne. 2002. Total reconstruction, PF movement, and derivational order. Linguistic Inquiry 33(2): 283–319. Schlenker, Philippe. 1999. La flexion de l’adjectif en allemand:  La morphologie de haut en bas. Recherches Linguistiques de Vincennes 28: 115–132. (http://rlv.revues.org/document1216.html). Schroeder, Christoph. 2006. Articles and article systems in some areas of Europe. In Giuliano Bernini & Marcia L. Schwartz (Eds), Pragmatic organization of discourse in the languages of Europe [EALT/EUROTYP], 545–611. Berlin: Mouton de Gruyter. Schütze, Carson T. 2001. On the nature of default case. Syntax 4(3): 205–238. Scott, Gary-John. 2002. Stacked adjectival modification and the structure of nominal phrases. In Guglielmo Cinque (Ed.), Functional structure in DP and IP: The cartography of syntactic structures, Vol.1, 91–120. Oxford: OUP. Sells, Peter. 1985. Restrictive and non-restrictive modification. Report #CSLI-85-28, Stanford University. Shlonsky, Ur. 2004. The form of Semitic noun phrases. Lingua 114: 1465–1526. Sigurðsson, Halldór Ármann. 1993. The structure of the Icelandic NP. Studia Linguistica 47(2): 177–197. Sigurðsson, Halldór Ármann. 2006. The Icelandic noun phrase: Central traits. Arkiv för nordisk filologi 121: 193–236. Spencer, Andrew. 1992. Nominal inflection and the nature of functional categories. Journal of Linguistics 28: 313–341. Sproat, Richard & Chilin Shih. 1991. The cross-linguistic distribution of adjective ordering restrictions. In Carol Georgopoulos & Roberta Ishihara (Eds), Interdisciplinary approaches to language. Essays in honor of S.-Y. Kuroda, 565–593. Dordrecht: Kluwer. Stroik, Thomas S. 1994. Saturation, predication, and the DP hypothesis. Linguistic Analysis 24: 39–61. Stowell, Tim. 1989. Subjects, specifiers, and X-bar theory. In Mark R. Baltin & Anthony S. Kroch (Eds), Alternative conceptions of phrase structure, 232–262. Chicago IL: The University of Chicago Press. Stowell, Tim. 1991. Determiners in NP and DP. In Katherine Leffel & Denis Bouchard (Eds), Views on phrase structure, 37–56. Dordrecht: Kluwer. Svenonius, Peter. 1992. The extended projection of N: Identifying the head of the noun phrase. Working Papers in Scandinavian Syntax 49: 95–121. Svenonius, Peter. 1993a. Selection, adjunction, and concord in the DP. Studia Linguistica 47(2): 198–220. Svenonius, Peter. 1993b. The structural location of the attributive adjective. In Erin Duncan, Donka Farkas & Philip Spaelti (Eds), The proceedings of the Twelfth WCCFL, 439–454. Stanford CA: Stanford Linguistics Association. Svenonius, Peter. 2007. The position of adjectives and other phrasal modifiers in the decomposition of DP. (Available at lingBuzz/000329). Sybesma, Rint & Joanna Sio. 2006. D is for demonstrative. Investigating the position of the demonstrative in Chinese and Zhuang. Ms, University of Leiden. Szabolcsi, Anna. 1981. The possessive construction in Hungarian: A configurational category in a non-configurational language. Acta Linguistica Academiae Scientifiarum Hungaricae 31: 261–289. Szabolcsi, Anna. 1983–84. The possessor that ran away from home. The Linguistic Review 3: 89–102.

 Demonstratives and definite articles as nominal auxiliaries Szabolcsi, Anna. 1987. Functional categories in the noun phrase. In István Kenesei (Ed.), Approaches to Hungarian, Vol. 2: Theories and analyses, 167–189. Szeged: Jate. Szabolcsi, Anna. 1994. The noun phrase. In Ferenc Kiefer & Katalin É. Kiss (Eds), Syntax and semantics 27: The syntactic structure of Hungarian, 179–274. San Diego CA: Academic Press. Takano, Yuji. 2003. How antisymmetric is syntax? Linguistic Inquiry 34(3): 516–526. Taraldsen, Knut Tarald. 1990. D-projections and N-projections in Norwegian. In Joan Mascaró & Marina Nespor (Eds), Grammar in progress. Glow essays for Henk van Riemsdijk, 419–431. Dordrecht: Foris. Taraldsen, Knut Tarald. 1991. Two arguments for functional heads. Lingua 84: 85–108. Thompson, Sandra Annear. 1971. The deep structure of relative clauses. In Charles J. Fillmore & D. Terence Langendoen (Eds), Studies in linguistic semantics, 79–94. New York NY:  Holt, Rinehart and Winston. Thorne, James Peter. 1972. On nonrestrictive relative clauses. Linguistic Inquiry 3: 552–6. Thráinsson, Höskuldur. 1994. Icelandic. In Ekkehard König & Johan van der Auwera (Eds), The Germanic languages, 142–89. London: Routledge. Ticio, María E. 2003. On the structure of DPs. Ph.D. Dissertation, University of Connecticut. Travis, Lisa. 1984. Parameters and effects of word order variation. Ph.D. Dissertation, MIT. Trenkic, Danijela. 2004. Definiteness in Serbian/Croatian/Bosnian and some implications for the general structure of the nominal phrase. Lingua 114: 1401–1427. Uriageraka, Juan. 1995. Aspects of the syntax of clitic placement in Western Romance. Linguistic Inquiry 26: 79–123. Valois, Daniel. 1991a. The internal syntax of DP and adjectival placement in French and English. Proceedings of NELS 21: 367–381. Valois, Daniel. 1991b. The internal syntax of DP. Ph.D. Dissertation, University of California, Los Angeles. Vangsnes, Øystein Alexander. 1996. The role of gender in (mainland) Scandinavian possessive constructions. Working Papers in Scandinavian Syntax 58: 1–28. Vangsnes, Øystein Alexander. 1999. The identification of functional architecture. Ph.D. Dissertation, University of Bergen. Vangsnes, Øystein Alexander. 2001. On noun phrase architecture, referentiality, and article systems. Studia Linguistica 55(3): 249–299. Vangsnes, Øystein Alexander. 2004. Rolling up the Scandinavian noun phrase. Paper presented at the 27th GLOW Colloquium in Thessaloniki, Greece. Vater, Heinz. 1991. Determinatien in der DP. In Susan Olsen & Gisbert Fanselow (Eds), DET, COMP und INFL. Zur Syntax funktionaler Kategorien und grammatischer Funktionen, 15–34. Tübingen: Niemeyer. Vergnaud, Jean-Roger. 1974. French relative clauses. Ph.D. Dissertation, MIT. Vincent, Nigel. 1997. The emergence of the D-system in Romance. In Ans van Kemenade & Nigel Vincent (Eds), Parameters of morphosyntactic change, 149–169. Cambridge: CUP. Vries, Mark de. 2006. The syntax of appositive relativization: On specifying coordination, false free relatives, and promotion. Linguistic Inquiry 37(2): 229–270. Wessén, Elias. 1970. Schwedische Sprachgeschichte, Vol. 3. Berlin: Walter de Gruyter. Wegener, Heide. 1995. Die Nominalflexion des Deutschen  – verstanden als Lerngegenstand. Tübingen: Niemeyer. Whitman, John. 1981. The internal structure of NP in verb final languages. Chicago Linguistic Society 17: 411–418.



References 

Wide, Camilla. 2004. Adnominal uses of den ‘the/it’ with unmodified nouns in the Swedish dialects of Eastern Nyland. Paper presented at the 20th Scandinavian Conference of Linguistics, Helsinki. Wiese, Richard. 1988. The proper treatment of inflection in the German article system. Wiener Linguistische Gazette, Beiheft [Supplement] 7: 32–4. Wiese, Richard. 1996. The Phonology of German. Oxford: OUP. Wood, Johanna L. 2002. Much about such. Studia Linguistica 56(1): 91–115. Wood, Johanna L. 2007. Demonstratives and possessives. In Elisabeth Stark, Elisabeth Leiss, & Werner Abraham (Eds), Nominal determination. Typology, context constraints, and historical emergence, 339–361. Amsterdam: John Benjamins. Wurmbrand, Susi. 2006. Licensing case. Journal of Germanic Linguistics 18(3): 175–235. Zamparelli, Roberto. 1993. Pre-nominal modifiers, degree phrases and the structure of AP. University of Venice Working Papers in Linguistics 3(1): 138–163. Zamparelli, Roberto. 2000. Layers in the determiner phrase. New York NY: Garland. Zhang, Niina. 2004. Representing specificity by the internal order of indefinites. Ms, National Chung Cheng University. Zoëga, Geir T. 1910. A concise dictionary of Old Icelandic. Oxford: Clarendon Press. Zwart, C. Jan-Wouter. 2001. Syntactic and phonological verb movement. Syntax 4(1): 34–62.

Language index

Note that this index does not distinguish between some of the different dialects of certain languages. C Catalan  49, 61 f, 65 Common Scandinavian  35 ff, 45 f, 67 D Danish  16, 53, 60, 75, 79, 83, 91 f, 95 f, 99, 112 f E English  4 f, 8 ff, 18, 31, 40, 48, 77, 93, 100 f, 103, 113, 165 F Faroese  56, 60, 63, 74, 76 ff, 91 f, 96, 112 French  14, 52, 107, 110, 117, 136 Frisian  60 ff

I Icelandic  13 f, 32, 34, 38, 44, 47 ff, 53, 60, 65, 68, 75, 77, 79 f, 83, 86 f, 94, 96 ff, 106 ff, 113, 115 Irish  50 Italian  12, 23 f, 31, 39, 45, 51 f, 69, 100, 102, 106 ff, 117 J Jutlandic  52, 60, 79, 116 L Ladin  165 N Norwegian  17, 20 f, 32, 42, 53, 55 f, 58, 60, 63, 66, 74 ff, 83, 91 f, 96, 112, 117, 119

G Galician  7 German  7, 9, 11 ff, 16, 18 ff, 22, 39, 48, 51 f, 56, 66, 70, 72, 77, 102, 111 f, Chapter  4 Greek  8, 15 f, 32, 49 f, 60, 65

O Old High German  126 Old Icelandic see Old Norse Old Norse  36 ff, 43, 46 ff, 53, 67 ff, 80

H Hebrew  8, 10, 23, 32

P Polish  69

Proto-Scandinavian  34 f, 37 f, 41, 44 R Romanian see Rumanian Rumanian  14 f, 50 S Spanish  23, 31 f, 49 ff, 107 Swahili  1 ff, 50 Swedish  19, 32, 53, 55 f, 58, 60, 63 f, 66, 70, 74 ff, 84, 90 ff, 100 f, 106 f, 112 f, 117, 168 Swiss German  14 T Turkish  66 f U Ute  3 W Welsh  50

Subject index

A Adjective as Adjunct  17 ff Adjective determiner-like  167 ff Adjective as head in the extended projection of the noun  15 ff, 69 Adjective as Specifier  14 ff Agreement see Agreement domain, Concord, Long-distance agreement Agreement domain  22 f, 71 ff, 76, 87, 90, 93 ff, 97, 99, 107, 109, 119, 143, 147 f, 150 Agreement Phrase  12, 17 Article different semantics  56, 61 Article expletive  35, 45, 56 f, 59, 61, 65, 74, 92, 94, 101 Article indefinite  11, 17, 65 ff, 93, 121 ff, 143 f, 148, 150, 156, 164 f Article null  136 f, 140 ff, 145, 157 ff, 163, 172 Article Phrase  12 Auxiliary  8 ff, 23, 45, 62, 72, 113 C Cardinal Phrase  12 f Concord  22 Coordination  58, 103 ff D Definiteness syntactic  173 Deictic reinforcer  49 ff, 70 Demonstrative as a phrase  38 ff Demonstrative specific indefinite  48 Determiner see Article, Demonstrative, Pronoun Distributed Morphology see Fission, Fusion, Impoverishment,

Morphological Lowering, Morphological Merger, Underspecification Double Definiteness  32, 53 f, 56 ff, 62 ff, 74, 76, 78, 92 f Doubly filled DP  76, 79, 101 DP-Hypothesis  4 f DP parameterized  21 DP Structure  12 ff E Extraction out of DP  20, 24, 38 f, 101 f F Fission  145 f Functional Application  89, 104 f, 114 f Fusion  64, 149, 167 Futhark  34 f H Head Movement Constraint  23 f, 69 I Impoverishment  145 ff, 158 f, 163 f, 167, 169 ff Inflection see Paradigm, Principle of Monoinflection, Strong adjective endings, Weak adjective endings Intensifier  169 ff L Larsonian nP-shell  20 Last Resort  74 f, 96, 148, 162 Long-distance agreement  22, 26, 71, 73, 143, 148, 150 M Mirror Principle  19 Modal verbs  8 ff, 72, 113

Modifier see Non-restrictive modifier, Restrictive modifier Morphological Lowering  70, 149 Morphological Merger  69 ff, 74 ff, 86 f, 94 ff, 115 Movement see Extraction out of DP, Last Resort, N-raising, Head Movement Constraint, Relativized Minimality, Scope Freezing, Sideward movement, Split-NP, Successive-cyclic movement Multiple semantic spell-out  105 N Non-restrictive modifier  88, 90 f, 93, 96 ff Noun Phrase  12, 20 N-raising  6, 23 f, 43 ff, 46 f, 50, 67 ff, 73, 79, 86 Number Phrase  12, 19 P Paradigm strong, weak, “mixed” 120 ff Percolation  151 f, 154 f, 169 f Possessor  6, 12, 20, 24, 40, 42, 64, 69, 101, 127, 149, 160 f Predicate Abstraction  114 f Predicate Modification  89, 96, 103, 114 f Preposition Phrase  39, 95 Principle of Monoinflection  135 ff Pronoun as copulas  10 f Pronoun definite  31 f, 48, 64, 102 ff, 124, 130, 139, 166 ff, 172 f Pronoun indefinite  162 ff

  Subject index R Readjustment Rule  155 ff Reanalysis  2, 38, 44 ff Relative clause different semantics  90 ff, 112 f Relative clause different syntax  93 ff, 111 f Relativized Minimality  23 f, 68 Restrictive modifier  83 ff

S Schwa-insertion  131, 133 f, 148 f Scope Freezing  72, 87 Semi-lexical head  73, 95, 161 Sideward movement  153 f Split DP  42 Split-NP  13 f, 16, 125 ff, 160 ff Strong adjective endings  120 ff Successive-cyclic movement  66, 110 f, 117 f, 155, 169

U Underspecification  174 V Vocative  31, 55 ff, 63 W Weak adjective endings  120 ff

Linguistik Aktuell/Linguistics Today A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 142 Nunes, Jairo (ed.): Minimalist Essays on Brazilian Portuguese Syntax. vi, 236 pp. + index. Expected April 2009 141 Alexiadou, Artemis, Jorge Hankamer, Thomas McFadden, Justin Nuger and Florian Schäfer (eds.): Advances in Comparative Germanic Syntax. xv, 389 pp. + index. Expected April 2009 140 Roehrs, Dorian: Demonstratives and Definite Articles as Nominal Auxiliaries. 2009. xii, 196 pp. 139 Hicks, Glyn: The Derivation of Anaphoric Relations. xii, 306 pp. + index. Expected March 2009 138 Siddiqi, Daniel: Syntax within the Word. Economy, allomorphy, and argument selection in Distributed Morphology. xii, 138 pp. Expected February 2009 137 Pfau, Roland: Grammar as Processor. A Distributed Morphology account of spontaneous speech errors. 2009. xiii, 372 pp. 136 Kandybowicz, Jason: The Grammar of Repetition. Nupe grammar at the syntax–phonology interface. 2008. xiii, 168 pp. 135 Lewis, William D., Simin Karimi, Heidi Harley and Scott O. Farrar (eds.): Time and Again. Theoretical perspectives on formal linguistics. In honor of D. Terence Langendoen. 2009. xiv, 265 pp. 134 Armon-Lotem, Sharon, Gabi Danon and Susan Rothstein (eds.): Current Issues in Generative Hebrew Linguistics. 2008. vii, 393 pp. 133 MacDonald, Jonathan E.: The Syntactic Nature of Inner Aspect. A minimalist perspective. 2008. xv, 241 pp. 132 Biberauer, Theresa (ed.): The Limits of Syntactic Variation. 2008. vii, 521 pp. 131 De Cat, Cécile and Katherine Demuth (eds.): The Bantu–Romance Connection. A comparative investigation of verbal agreement, DPs, and information structure. 2008. xix, 355 pp. 130 Kallulli, Dalina and Liliane Tasmowski (eds.): Clitic Doubling in the Balkan Languages. 2008. ix, 442 pp. 129 Sturgeon, Anne: The Left Periphery. The interaction of syntax, pragmatics and prosody in Czech. 2008. xi, 143 pp. 128 Taleghani, Azita H.: Modality, Aspect and Negation in Persian. 2008. ix, 183 pp. 127 Durrleman-Tame, Stephanie: The Syntax of Jamaican Creole. A cartographic perspective. 2008. xii, 190 pp. 126 Schäfer, Florian: The Syntax of (Anti-)Causatives. External arguments in change-of-state contexts. 2008. xi, 324 pp. 125 Rothstein, Björn: The Perfect Time Span. On the present perfect in German, Swedish and English. 2008. xi, 171 pp. 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 42 41 40 39 38 37 36 35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7

Featherston, Sam: Empty Categories in Sentence Processing. 2001. xvi, 279 pp. Alexiadou, Artemis: Functional Structure in Nominals. Nominalization and ergativity. 2001. x, 233 pp. Zeller, Jochen: Particle Verbs and Local Domains. 2001. xii, 325 pp. Hoeksema, Jack, Hotze Rullmann, Víctor Sánchez-Valencia and Ton van der Wouden (eds.): Perspectives on Negation and Polarity Items. 2001. xii, 368 pp. Gelderen, Elly van: A History of English Reflexive Pronouns. Person, Self, and Interpretability. 2000. xiv, 279 pp. Meinunger, André: Syntactic Aspects of Topic and Comment. 2000. xii, 247 pp. Lutz, Uli, Gereon Müller and Arnim von Stechow (eds.): Wh-Scope Marking. 2000. vi, 483 pp. Gerlach, Birgit and Janet Grijzenhout (eds.): Clitics in Phonology, Morphology and Syntax. 2001. xii, 441 pp. Hróarsdóttir, Thorbjörg: Word Order Change in Icelandic. From OV to VO. 2001. xiv, 385 pp. Reuland, Eric (ed.): Arguments and Case. Explaining Burzio’s Generalization. 2000. xii, 255 pp. Puskás, Genoveva: Word Order in Hungarian. The syntax of Ā-positions. 2000. xvi, 398 pp. Alexiadou, Artemis, Paul Law, André Meinunger and Chris Wilder (eds.): The Syntax of Relative Clauses. 2000. vi, 397 pp. Svenonius, Peter (ed.): The Derivation of VO and OV. 2000. vi, 372 pp. Beukema, Frits and Marcel den Dikken (eds.): Clitic Phenomena in European Languages. 2000. x, 324 pp. Miyamoto, Tadao: The Light Verb Construction in Japanese. The role of the verbal noun. 2000. xiv, 232 pp. Hermans, Ben and Marc van Oostendorp (eds.): The Derivational Residue in Phonological Optimality Theory. 2000. viii, 322 pp. Růžička, Rudolf: Control in Grammar and Pragmatics. A cross-linguistic study. 1999. x, 206 pp. Ackema, Peter: Issues in Morphosyntax. 1999. viii, 310 pp. Felser, Claudia: Verbal Complement Clauses. A minimalist study of direct perception constructions. 1999. xiv, 278 pp. Rebuschi, Georges and Laurice Tuller (eds.): The Grammar of Focus. 1999. vi, 366 pp. Giannakidou, Anastasia: Polarity Sensitivity as (Non)Veridical Dependency. 1998. xvi, 282 pp. Alexiadou, Artemis and Chris Wilder (eds.): Possessors, Predicates and Movement in the Determiner Phrase. 1998. vi, 388 pp. Klein, Henny: Adverbs of Degree in Dutch and Related Languages. 1998. x, 232 pp. Laenzlinger, Christopher: Comparative Studies in Word Order Variation. Adverbs, pronouns, and clause structure in Romance and Germanic. 1998. x, 371 pp. Josefsson, Gunlög: Minimal Words in a Minimal Syntax. Word formation in Swedish. 1998. ix, 199 pp. Alexiadou, Artemis: Adverb Placement. A case study in antisymmetric syntax. 1997. x, 256 pp. Beermann, Dorothee, David LeBlanc and Henk van Riemsdijk (eds.): Rightward Movement. 1997. vi, 410 pp. Liu, Feng-hsi: Scope and Specificity. 1997. viii, 187 pp. Rohrbacher, Bernhard Wolfgang: Morphology-Driven Syntax. A theory of V to I raising and prodrop. 1999. viii, 296 pp. Anagnostopoulou, Elena, Henk van Riemsdijk and Frans Zwarts (eds.): Materials on Left Dislocation. 1997. viii, 349 pp. Alexiadou, Artemis and T. Alan Hall (eds.): Studies on Universal Grammar and Typological Variation. 1997. viii, 252 pp. 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. Lutz, Uli and Jürgen Pafel (eds.): On Extraction and Extraposition in German. 1996. xii, 315 pp. Cinque, Guglielmo and Giuliana Giusti (eds.): Advances in Roumanian Linguistics. 1995. xi, 172 pp. Gelderen, Elly van: The Rise of Functional Categories. 1993. x, 224 pp. Fanselow, Gisbert (ed.): The Parametrization of Universal Grammar. 1993. xvii, 232 pp. Åfarlí, Tor A.: The Syntax of Norwegian Passive Constructions. 1992. xii, 177 pp.