uncorrected proofs

Nominal Determination

S F O PRO

D E ANY T P M C O RRE HING C

O BLIS U C P S N U N BENJAMIN © JOH

1st proofs

Studies in Language Companion Series (SLCS) This series has been established as a companion series to the periodical Studies in Language.

Editors Werner Abraham University of Vienna

Michael Noonan

University of Wisconsin-Milwaukee USA

Editorial Board Joan Bybee

Robert E. Longacre

Ulrike Claudi

Brian MacWhinney

Bernard Comrie

Marianne Mithun

William Croft

Edith Moravcsik

Östen Dahl

Masayoshi Shibatani

Gerrit J. Dimmendaal

Russell S. Tomlin

Ekkehard König

John W.M. Verhaar

University of New Mexico University of Cologne Max Planck Institute, Leipzig University of New Mexico University of Stockholm University of Cologne

Free University of Berlin

University of Texas, Arlington Carnegie-Mellon University University of California, Santa Barbara University of Wisconsin, Milwaukee Rice University and Kobe University University of Oregon The Hague

Christian Lehmann University of Erfurt

Volume 89

S F O PRO

Nominal Determination. Typology, context constraints and historical emergence Edited by Elisabeth Stark, Elisabeth Leiss and Werner Abraham



1st proofs

Nominal Determination Typology, context constraints and historical emergence

Edited by

Elisabeth Stark Free University Berlin

Elisabeth Leiss University of Munich

Werner Abraham University of Vienna

John Benjamins Publishing Company Amsterdam / Philadelphia

S F O PRO



1st proofs

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 Nominal determination : typology, context constraints, and historical emergence / edited by Elisabeth Stark, Elisabeth Leiss, Werner Abraham. p. cm. -- (Studies in language companion series, issn 0165-7763 ; v. 89) Includes bibliographical references and index. 1. Grammar, Comparative and general--Determiners. 2. Definiteness (Linguistics) 3. Grammar, Comparative and general--Nominals. I. Stark, Elisabeth. II. Leiss, Elisabeth. III. Abraham, Werner. P299.D48N66 2007 415--dc22 2007018285 isbn 978 90 272 3099 7 (Hb; alk. paper)

© 2007 – 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.

S F O PRO

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



1st proofs

Table of contents



Preface Introduction Werner Abraham, Elisabeth Stark & Elisabeth Leiss

1

1. Synchrony – and its implications for diachrony Discourse binding: DP and pronouns in German, Dutch, and English Werner Abraham

21

Gender, number, and indefinite articles: About the ‘typological inconsistency’ of Italian Elisabeth Stark

49

Covert patterns of definiteness/indefiniteness and aspectuality in Old Icelandic, Gothic, and Old High German Elisabeth Leiss

71

The definite article in Indo-European: Emergence of a new grammatical category? Brigitte L.M. Bauer

101

2. Synchrony – ontological and typological characteristics ‘No’ changes: On the history of German indefinite determiners in the scope of negation Agnes Jäger The functional range of bare singular count nouns in English Laurel Smith Stvan The definite article in non-specific object noun phrases: Comparing French and Italian Tanja Kupisch & Christan Koops

139 169

187

S F O PRO



1st proofs



Table of contents

Early functions of definite determiners and DPs in German first language acquisition Dagmar Bittner

213

3. Diachrony – universally unified characteristics? The discourse-functional crystallization of DP from the original demonstrative Werner Abraham

239

Determinerless noun phrases in Old Romance passives Anna Bartra-Kaufmann

255

On the structure and development of nominal phrases in Norwegian Terje Lohndal

285

The emergence of DP from a perspective of ontogeny and phylogeny: Correlation between DP, TP and aspect in Old English and first language acquisition Fuyo Osawa Demonstratives and possessives: From Old English to present-day English Johanna L. Wood

309 337

S F O PRO



1st proofs

Preface The papers in this volume represent a selection from the talks given at the workshop “Evolution and functions of nominal determination” at the XXVII annual meeting of the German Association for Linguistics (Deutsche Gesellschaft für Sprachwissenschaft, DGfS) held under the direction of Elisabeth Stark, Freie Universität Berlin, and Elisabeth Leiss, Ludwig-Maximilians-Universität München, in February 2005 at Cologne, Germany. The workshop was dedicated to a reopening of the discussion on fundamental linguistic categories and operations and their importance in any approach to language change, especially in the systems of nominal determination. The following papers (among others) were presented at the conference: Werner Abraham/Vienna, Dagmar Bittner/Berlin, Agnes Jäger/Jena, Tanja Kupisch/University of Hamburg) & Christian Koops (Rice University), Laurel Stvan/Arlington, and. These were supplemented by the following solicited papers also with a strong cline toward the problem of nominal determination in a typological and diachronic perspective: Anna Bartra-Kaufmann/Barcelona, Brigitte Bauer/ Austin, Elisabeth Leiss/Munich, Terje Lohndal/Oslo, Fuyo Osawa/Tokyo, Elisabeth Stark/Berlin, and Johanna Wood/Aarhus.

S F O PRO



1st proofs

S F O PRO



1st proofs

Introduction

.

Overview

The rich and inspiring literature on nominal determination of the last decades testifies the considerable interest in questions related to determination (see, e.g. Hawkins 1978; van der Auwera 1980; Lyons 1999; to name but a few). Heim (1988) and Kamp & Reyle (1993), among others, discuss different aspects of the semantic category of definiteness, mainly in a textual perspective. Work related to the names of Greenberg (1978), Seiler (1978) or, more recently, Rijkhoff (2002) and Coene & D’hulst (eds; 2003a and 2003b) represent well-known, successful insights into the diachronic and synchronic typology of nouns and nominals. Contributions by logical semanticists such as Longobardi (1994, 1996, 2005), Chierchia (1998) and Zamparelli (2000) combine comparative linguistics with theoretical models, especially in the area of generative linguistics. The evolution of determiner systems in single language families has been extensively discussed, for example, in the work of Selig (1992) and Vincent (1997) for definite and of Stark (2005, 2006) for indefinite determiners in Romance, and by Abraham (1997) and Leiss (2000) for German and the Germanic languages.

.

The notion of determination and definiteness

One first major point in all this work indicated above is that it has lead to many competing conceptions of nominal determination and definiteness/indefiniteness/ bare nominals. Notice, however, that definitions tend to overgeneralize. This can lead directly to erroneously mapping properties and indispensable characteristics of modern determination systems onto older language stages – i.e. onto languages which often do not possess overt determiners (and -systems) or are just on their way toward developing them. It is thus methodologically indispensable to distinguish between functional categories like ‘determination’ or ‘(in-)definiteness’ and their different forms (of overt expressions) in different languages and diachronic language states. Especially when it comes to research into the functions

S F O PRO



1st proofs



Nominal determination

of ‘nominal determination’, a historical approach to grammatical categories is crucial for any fundamental understanding of the former and the actual structural make-up in individual languages. “Determination of reference” can be seen as the explicit marking of nominals in order to provide information about the way the nominal predicate is to be mapped onto different sorts of (choices of sets of as well as text) referents (in the sense of Seiler’s (1978) notion of ‘perception of reality’). However, determination in this sense is not merely the overt indication of (in)definiteness. This shows most convincingly in historically early stages. When demonstratives or numerals emerge as nominal determiners, they do so because of their very meaning as indexicals (demonstratives) or quantifiers (numerals). Their prime task is not to indicate ‘givenness’ or ‘known to the addressee’, but conceptual categories and features such as ‘shape’, ‘singularity’, and ‘perceptibility’, among a few other characteristics. Let us give an example. Appreciating actual linguistic data uncovered from ancient text documents by means of formal semantic models of nominal determination, the functional reduction of Modern German articles to the ‘anaphoric-cataphoric opposition’ is in no way sufficient. Other characteristics such as thematic or definite/specific status may come into play as well or exclusively. Yet, what we find about the meaning of determiners of (in-)definiteness in modern text books is often vague or even ambiguous (see Lyons (1999), something that leaves the reader with two incompatible conceptions: For example, the quantificational (‘inclusive’ vs. ‘exclusive’, à la Hawkins 1978) or the (con)textual ‘given’ vs. ‘new’ (à la Kamp & Reyle 1993). We stress the point that the function marked by nominal determiners in one language may be marked by word order or paradigmatic case shift (as between accusative and genitive; see Abraham 1997; Leiss 2000) in another. This leaves undecided the problem of a clear-cut definition of ‘(in-)definiteness’ – unless one gives up and defines definiteness as a grammatical property only within one individual language – in other words, as the nothing-but-overt projection of DP as opted for by Lyons (1999). This is a choice that we do not advocate.

.

The question of DP

Many researchers assume that nominal determination as an overtly marked functional category D is not universal (recall, e.g. the work by Gil 1987; Vincent 1997 and Chierchia 1998) – whereas, no doubt, the semantic function of determination is universal. Marking determination makes use of different syntactic and morphological devices, a fact which complicates the language-independent description and definition and provides a challenge for modelling cross-linguistic variation. The positions among the contributors to this volume are divided: Bartra sides

S F O PRO



1st proofs

Introduction



with Boucher (2005), Dobrovie-Sorin (2001: 208), Delfitto & Schroten (1991), among others, in claiming that in languages without overt determiners, no functional projection D has to be established in order to derive the referential status of nominals. In these languages, the referential or definite value of the noun must be checked against other existing functional projections. However, there is a weighty partisanship in favor for the claim that even in the absence of overt (definite) articles, nominal reference as well as some properties of argument noun phrases can be only accounted for by positing a D projection. Progovac (1998) and Pereltsvaig (2006) are such theoretical protagonists, even for languages without overt determiners. Lohndal (in the present volume) decidedly opts for this position. BartraKaufmann (this volume) quotes Pereltsvaig (2006) taking the position that “complete” noun phrases or DPs contrast with small nominal. Assuming that there is no evidence or cue in those languages for acquiring D, it has to be universally present. The main question, then, is the following: Do we have to assume the existence of a basic D(P)-structure across all languages and language states even if and when, at some specific historical stage or across, no D-word is showing overtly? A good case in point is Modern Japanese (see below Section 3.2), another one is Classical Latin. This is then the main difficulty aligned with this question: Are there languages which never have shown N-to-D raising? Recall the claim above that this lack may be functionally compensated in other grammatical modules. In other words: Is D universal, either overtly or covertly? The contributions in this volume are far from definitely answering, or even touching explicitly upon, this issue. Although it is taken up by Leiss, Osawa, and, somewhat lightly, by Bauer, it comes under different perspectives and with varying results. Notice that discussing Old English without D-representatives and its path to modern English with an explicit definite article, we would have to posit one of the following two hypotheses:

“Out of nothing-to-D Hypothesis”: for example, a fundamental restructuring of Modern English has taken place, while under the “Covert-to-overt-D Hypothesis”: A silent DP might be invoked for a language introducing lexical determiners in D after changing the triggering grammatical and semantic features and thus satisfying singular reference and anaphoric, transclausal binding in due course. Hitherto undetected patterns of definiteness marking (covert DPs either in the linguistic sense or in the metalinguistic sense) are being replaced by an article system where D is less covert.

Expectably, there are predecessors to this distinction. As Bauer notes in her contribution (this volume) it was Benveniste (1974: 126–127) who distinguished two types of change: a grammatical category may be preserved, with possibly accompanying formal replacement (of case endings by prepositions). Or there may be (a) changes

S F O PRO



1st proofs




that lead to the loss of a grammatical category (e.g. the shift from a three-gender to a two-gender system in Indo-European); and (b) changes that lead to the creation of a new grammatical category Benveniste’s illustration being the definite article. Benveniste, thus, at least also opts for the first hypothesis above. Hermann Paul (1909/91975: 35–36), by contrast, sees no logical possibility for the first hypothesis, the Out of nothing-to-something Hypothesis.1 It can be shown that only under very specific distributional conditions in particular languages do we say that there is no such covert DP – thus, no final overt raising to D or SpecDP. .

Syntactic structure

Ever since Bloomfield (1933/65: 205), Abney (1987), and Longobardi (1994, 2005), among others, we take clausal argument nominals to be a projection of a D-head or a Num-head or an N-head. Thus, the structure of nominals is this (somewhat simplified). (1)

DP D

·theÒ my Peter’s

NumP Num

NP

·aÒ -

dog

. According to Paul (1975: 35f.), only the Covert-to-overt-something Hypothesis appears to be acceptable: “Wenn unsere Betrachtungsweise richtig durchgeführt wird, so müssen die allgemeinen Ergebnisse derselben auf alle Sprachen und auf alle Entwicklungsstufen derselben anwendbar sein, auch auf die Anfänge der Sprache überhaupt. Die Frage nach dem Ursprunge der Sprache kann nur auf der Grundlage der Prinzipienlehre beantwortet werden. Andere Hilfsmittel zur Beantwortung gibt es nicht. Wir können nicht auf grund der Überlieferung eine historische Schilderung von den Anfängen der Sprache entwerfen. Die Frage, die sich beantworten lässt, ist überhaupt nur: wie war die Entstehung der Sprache möglich. Diese Frage ist befriedigend gelöst, wenn es uns gelingt, die Entstehung der Sprache lediglich aus der Wirksamkeit derjenigen Faktoren abzuleiten, die wir auch jetzt noch bei der Weiterentwickelung der Sprache immerfort wirksam sehen. Übrigens lässt sich ein Gegensatz zwischen anfänglicher Schöpfung der Sprache und blosser Weiterentwickelung gar nicht durchführen. Sobald einmal die ersten Ansätze gemacht sind, ist Sprache vorhanden und Weiterentwickelung. Es existieren nur graduelle Unterschiede zwischen den ersten Anfängen der Sprache und den späteren Epochen.”

S F O PRO



1st proofs

Introduction



According to Longobardi (1994: 628; 2005: 24, 27), the difference between NP and DP is that NPs – determinerless, bare nominals – are inherently predicative and thus cannot occur in referential argument positions. Bare NPs are not referential, but classifying – in other words, they are close to adjectivals, i.e. property denotations. Referential nominals denote particular entities in the universe of discourse. Indefinitely determined nominals may, but need not, be classifying. Only DP can occur in argument positions, which need to refer to theta characteristics which in turn are contingent upon the semantics of the predicating verb. The role of picking out a particular referent is taken care of by a functional D. The role of a functional D is to change predicative nominals, bare NPs, into arguments, DPs, by identifying the referentiality of a nominal. This selection operation is best explained by the theory of theta-binding proposed by Higginbotham (1985). However, the function of identifying the referentiality of a nominal is not always taken care of by an overt D. It has been proposed (Abraham 1997; Leiss 2000; see also Leiss, this volume) that in the absence of an overt D-paradigm, morphological case on the head nouns can determine indirectly the referentiality of a nominal. In Old English, without overt D, morphological case performed the same task as overt D. Assuming the difference between NP and DP and assuming the role of a functional D, it can be argued (as Osawa does, this volume) that earlier D-less NPs changed into DPs via the emergence of a D-paradigm for nominals in the history of the specific language. Another position adopted for languages with indirect referencing means is that morphological case alternates contingent upon the choice of aspect determined definiteness versus indefiniteness (Abraham 1997; Leiss 2000; see Leiss, this volume). In other words, while there was no overt D-category present in such language states (Old English, Old High German, Gothic, Latin as opposed to their modern Germanic developments and modern Romance), yet referentiality was ascertained through the interaction of means other than direct lexical D-fillers. .

Typology

In languages like Chinese or Japanese, NPs, not DPs, are inherently argumental and can thus occur freely in the argument positions without determiners, while they are predicates in other languages (as the Romance languages) and cannot occur in argument positions without determiners, as argued by Longobardi (1994). By contrast, in the Germanic languages nominals are allowed to have these double natures. Thus, as has been argued by Chierchia (1998), there are parameterized differences in the nature of nouns sketched succinctly in Table 1 below.

S F O PRO



1st proofs




Table 1. The nominal mapping parameter (Chierchia 1998: 400) Languages Japanese, Chinese Slavic Germanic Italian French

Predicative mapping

Argumentative mapping

Mass reference

Count reference Article

−

+

+

−

−

+ + + +

+ + − −

+ + + +

+ + + +

− + + +

Bare arguments

+ + + −

Notice that the feature ‘predicative mapping’ (vs. ‘argumentative mapping’) accounts for the fact that Japanese predicates come in nominalized form. All this could lead to the hypothesis that Japanese has no D-projection at all. There are in fact several hints in the literature (cf. Gil 1987) at this interpretation for Japanese nominal expressions. But we believe that such a decision is based on imprecise and, therefore, inadvertent distributional ground. See 3.2.1 below. .. Pronouns Japanese has no personal pronouns – i.e. Japanese has no means to grammatically distinguish binding. This refers both to clausal binding by means of something that is ‘reflexive/reciprocal’ and extra-clausal binding in terms of personal pronouns. And the status of any findings to this effect in broader context is that, if Japanese does not know grammatical means of binding, then it will be plausible not to assume a covert DP in the first place. Now see Table 2 below, which is to show that Japanese pronouns are not DPs, but NPs. What are called pronouns according to a Euro-ethnic understanding are nothing but nouns. In other words, the distributional characteristics of pseudo-pronominals such as watashi “I”, anata “you”, kare “he” etc. behave like full nominals. Table 2. Pro-DP vs. Pro-NP (from Tanaka 2006: 9) English

Japanese

a little boy

chiisana otokonoko little boy chiisana kare little he hanashijouzuna otoko eloquent man hanashijouzuna anata eloquent you Kai-no- hon Kai- book Kai-no- sore Kai- that

*a little he the eloquent man *the eloquent you Kai’s book *Kai’s that

S F O PRO



1st proofs

Introduction



The overall conclusion is that there is no such closed paradigm as a third person of the pronominal in Japanese. Likewise, the 2nd person, the addressee form, has many expressions all depending on the way how well you know the addressee or how polite you feel you have to be: anata “that side” (neutral, but creating distance), anta (somewhat loose, not unfriendly and yet lightly condescending), kimi (condescending or friendly), omae (pejorative), kisama (obsolete and pejorative) etc. (Tanaka 2006: 8). Given the assumption that the semantic contribution of D is ‘singularity’ rather than ‘anaphoricity’ (cf. Heim & Kratzer 1998: 81), a language that has no (anaphoric) personal pronouns may still possess a covert DP-structure in nominals. This assumption can also be shown to hold diachronically: According to Leiss (2000), definiteness (in its wide, all-encompassing sense) is signaled in Old Icelandic by clause-inceptive position, while in all other linear positions definiteness marking occurs according to grammatical patterns. In other words, the left-edge position in the clause, reserved for unmarked thematic material by default, is the only position where the explicit definiteness marker does not occur. A language following this strategy exbraciates the non-anaphoric function of the definite article and restricts its appearance to the uniqueness/singulative function – just like Japanese. We claim that the anaphoric function is implied by the singulative/uniqueness function in the first place – which is what the historical data from Old Icelandic attest to in the first place. Essentially, the link that we have drawn has no parallel in the literature, to the best of our insight. Yet, it appears plausible, if not logical, to derive the following relation. Informally, this relation can be seen as follows (following essentially Heim & Kratzer 1998: 81). (2)

The uniqueness quantifier has a definiteness scope within the presupposition of the article which is restricted with respect to the contextually restricted set of discourse referents. This means that the meaning of the definite article falls into two components: the denotation of singularity and its presupposition of anaphoricity – i.e. singularity as a choice from a given unique referential world or a set of worlds.

According to (2), the definite article is a conglomerate of at least two components one of which is the core and the other the presupposition of this core. Clearly, the meaning of the definite article in English and German refers exclusively to the singularity, or uniqueness, with respect to the set choice. It simply says that the function of the, i.e. [[the]], is that term of the discourse set which selects one and only one from this set. The anaphoric relation is established through the choice from D and the set restriction 〈e,t〉 – i.e. λf: f ∈ D〈e,t〉 (see Heim & Kratzer 1998: 75). For the syntactic account and its observational motivation to the extent that bare plurality and classifiers are in one single structural node, see 3.4. below.

S F O PRO



1st proofs




.. Mass nouns: a caveat Languages with transnumeral nominals do not distinguish articles, nor do they possess personal pronouns, like Japanese. Transnumeral nominals are close, but not identical, to mass nouns. They are quantifiable, but never without classifiers and linkers, which denote sortal categories appropriate to the sort of nominal to be quantified over. Thus, nominal quantification in Japanese runs different from languages such as English and German on a number of criteria. Gil (1987: 235f.) claims therefore two types of languages: those with configurational NPs distinguishing between count and mass nouns, and those with non-configurational NPs treating all nouns as mass. Krifka (1989) draws the same typological conclusion. The specific question is whether the categorial identification as mass is felicitous or whether transnumerality is a different referencing property. Syntactically, it can be shown that Japanese ‘classifiers’ behave no more as classifiers in the traditional sense of the transnumeral language type, but, rather like lexical attributes2 (Chan Hok-Shing 1999; Kurita 2006) than functional morphemes, and that Japanese bare nouns are very likely to be transnumeral nominal constituents without any ‘mass interpretation’. Notice that Chierchia’s (1998: 400) distinction of mass-count typologies across languages (see Table 1 above) does not answer the questions that we have asked at the outset. The critical property of quantifications of transnumeral nominal constituents is the fact that, from the point of view of Indo-European, the quantified nominals – “books”, “students” – are neither sets of distributive entities nor collective subsets, but subsets of abstract nominal notions (‘bookhood’, ‘studenthood’, ‘waterhood’). Thus, the distinction between count nouniness and mass nouniness is blurred under the transnumeral referential strategy. In this sense, Chiercha’s typological distinctive feature opposition “predicativity” vs. “argumentativity”, as in Table 1 above, albeit necessary and plausible, is hardly sufficient and cannot lead to the conclusion that Japanese would be a language without DP. .

Diachrony

Received wisdom tells us that demonstratives are the ancestors of definite articles in the languages of the world. However, definite articles and demonstratives are not alike – thus, not to be listed under one identical category, D in DP. Following Jakobson (1957), we take demonstratives to be simple shifters. By contrast, definite articles are complex shifters. The opposition between demonstratives and definite

S F O PRO

. No doubt, one and the same element may turn out as either classifier OR as attribute. See Yukari (2006).



1st proofs

Introduction

articles spells out as follows: The deictic force of simple shifters such as demonstratives depends exclusively on the speaker’s viewpoint (Bühler’s Origo). Non-shifters such as the 18th of May imply no reference to the speaker. By contrast, shifters such as yesterday or the demonstrative this have relative reference since they are dependent on the Origo/location/viewpoint of the speaker. They are called simple shifters. Definite articles are yet more complex in as much as they, quite plausibly, involve speaker as well as hearer and, thus, are more complex as shifters. Demonstratives cannot replace definite articles, which have the functional status of complex, or double, shifters. The definite article creates a viewpoint shared between speaker and hearer. In other words, definite articles refer to information which is known information for both hearer and speaker, whereas demonstrative pronouns refer to information which is known to the speaker, but unknown to the hearer. Compare Abraham’s contribution in terms of Centering Theory (Abraham, this volume a): Definite articles (as well as personal pronouns) have an anaphoric component whose binding force reaches far beyond the antecedent (left) clause edge binding the erstwhile discourse thema (usually at the far left edge of the antecedent clause). By contrast, the binding force exercised by demonstrative pronouns is more local and deictic in the sense of ‘pointing-at’ in that it binds the closest discourse rhema (usually close to the end of the antecedent clause). The data from Old Icelandic and Gothic (cf. Leiss 2000 as well as Leiss, this volume) attest that the overt rise of the definite article starts in the rhema – for good reason. In order to answer the question why the demonstrative pronoun is the privileged source of overtly signaled definiteness we have to unfold the functional affinity between the demonstrative pronoun, the rhema, and definiteness. Quite plausibly, there is a strong binding affinity between the demonstrative pronoun, DemPro, and the preceding rhema (as has been shown by Abraham, this vol. a). This affinity aligns well with the fact that the rise of the definite article in Old Icelandic started out in the rhema the reason being that a functionally definite rhema is expected to be marked by a definite article in order to block the indefinite reading presupposed by rhematic objects. What all of this amounts to is the conclusion that categories like D (a complex shifter as argued above) essentially create shared viewpoints for both speaker and hearer – to be sure, if and only if, as we assume, D hosts definite articles, but not demonstratives (an assumption which is shared by many: cf. Szabolcsi 1994; Giusti 1997; Ritter 1995; Chan 1999, to name but a few). On the other hand, the viewpoint created by complex shifters such as the definite article (or past tense as opposed to present tense) is the anchoring point for simple shifters such as demonstrative pronouns and adverbials of time, place, and manner in the sense that simple shifters, or simple-layered indexicals, receive their value by the very opposition to the doubly layered complex shifters. Thus, assuming languages without

S F O PRO



1st proofs






a D-structure is at the risk of losing the quality of remote anchoring of local and temporal reference. One of the reasons cited for not assuming D in a language is that such languages have no personal pronouns with anaphoric binding force, in the first place, AND do not use a particular category for singling out individual referents. The bottom line of all that is that from the two hypotheses regarding the universal status of DP in any reference structure only the “Covert-to-overt-D Hypothesis” holds. We cannot assume that DP emerges out of nothing, diachronically (nor ontologically) speaking. Notice, however, that feature checking mechanisms (usually taken to be responsible for formal accounts) will have to formalize different feature-source and feature-goal sites to satisfy D-derivations on the basis of aspect and case alternatives (such as in Russian) or linear topicality (as for Old Icelandic and Russian; see Leiss 2000). But this is a totally different matter – one that any formal theory working with templates avoids in the first place.

.

The ontological acquisition of DP vs. NP

The decision between our two hypothesis towards the universality of D will no doubt be contingent also on the question what the ontological acquisition of the Determiner Phrase (DP) is like (see Bittner, in the present volume). To this end, we ask the following two questions: (a) What is the development of the acquisition of the DP like (in some specific language)? (b) How are predictions deriving from current theoretical analyses of the DP to be evaluated in the light of empirical findings on child development? Sub-questions may be the following: (c) When do we have evidence for the acquisition of the functional layers of the nominal domain? (d) Are the properties attributed to the DP acquired simultaneously or incrementally? (e) What is the relationship between the acquisition of syntax and morphology of the DP? (f) When are language specific properties of DP such as the use of multiple definite articles (in specific languages such as the Scandinavian ones or in Modern Greek) acquired? (g) When do complex DPs involving the possessive construction, Determiner Spreading and appositive constructions emerge in child Greek? And (h) Is the core system of the nominal domain acquired simultaneously with the Left Periphery of the DP? Let us take Modern Greek to clarify these issues. Marinis (2003: 140, 166, 193; see (14)–(16) below) investigates the acquisition of simple DPs consisting of articles and nouns as well as the acquisition of complex DPs, in order to answer the above questions. Concerning the simple DPs, the author investigates the acquisition of indefinite (ena vivlio ‘a/one book’) and definite articles (to vivlio ‘the book’); as far as the complex DPs are concerned, he investigates the acquisition of

S F O PRO



1st proofs

Introduction



the possessive construction (Examples 14a, b), Determiner Spreading (Examples 15a, b), and appositive constructions involving kinship terms and proper names (Examples 16a, b). (14)

a.

b.

(15)

a. b.

(16)

a.

b.

Pira to vivlio tu took the- the- ‘I took Nikos’ book’. Pira tu Nikou took the- Niku- ‘I took Niko’s book’. To the To the

Niku. Niku- to vivlio. the- book-

meghalo to petrino to spiti. big the stone-made the house meghalo to spiti to petrino. big the house the stone-made

O piitis Solomos ezise sti Zakintho. the- poet- Solomos- lived in-the Zakinthos ‘The poet Solomos lived in Zakinthos’. I nea siskevasia ton makaronion Melissa. the new package the- spaghetti- Melissa- ‘The new package of the spaghetti Melissa’.

To account for these empirical findings, the analysis of the DP may involve three functional layers:

the DP layer that hosts the definiteness feature; the FP layer that hosts the case feature; and the Num(b)P layer that hosts the number feature.

The syntax-semantics mapping focuses on Chierchia’s (1998) Nominal Mapping Parameter. According to this model, nouns in Modern Greek are of the Romance Type, i.e. they are of the predicative type. In order to be used in argument positions, a DP layer must be projected. MG allows bare nouns in argument positions, but in restricted environments. Bare arguments have the status of DPs with a null D0 (see specifically Guérin 2006 and, more generally, Vogeleer & Tasmowski (eds) 2006 as well as Section 3.4 above). The paradigms of definite articles and nouns are illustrated along with the degree of syncretism and the unmarked versus marked forms of each paradigm. Results of the investigation of the acquisition of the functional layers, namely DP, FP and NumP, the syntax-semantics mapping and morphological marking yield that the head of the DP layer hosts the indefinite article, and the head of the FP layer hosts the definite article and the case feature. Evidence for the acquisition of the NumP can be provided by the use of number marking. Definite articles

S F O PRO



1st proofs




emerge earlier than indefinite articles, and children use bare nouns a lot, which are grammatical with a non-specific interpretation, but ungrammatical with a specific one. This indicates that definite articles emerge earlier than indefinite ones in the speech of children acquiring Modern Greek. Marinis (2003) examines the status of definite articles as impostors, i.e. in a lexically based fashion, using the data of one child and arguing that children use determiners as impostors and not as target-like definite articles. This process lasts during Stage I, i.e. prior to the age of 2;0, from 1;8 to 1;11. Evidence for a productive use of definite articles is provided by the use of definite articles with several noun types after that period. Marina’s investigation furthermore shows that the data under discussion support the predictions deriving from Chierchia’s Nominal Mapping Parameter. The youngest child does not use any definite articles at all. All five children pass through a stage in which they omit definite articles. The fact that they are using definite articles (and the fact that they omit them) provides evidence that nouns have the [−arg, +pred] specification. Another piece of evidence that in MG nouns have the specification [−arg, +pred] is the early use of definite articles with proper names and kinship terms. As regards the acquisition of morphological marking, case and number marking are acquired simultaneously. Marinis (2003) analyses the data of one child, showing that this child has knowledge of plural marking on nouns from Stage II. As far as case marking on nouns is concerned, the analysis includes only nouns in contexts requiring marked forms and more specifically masculine and feminine nouns, since there were no marked contexts for neuter nouns in the data of the child. The results of the analysis show that nominative case is acquired prior to genitive in both definite articles and nouns and in nouns with both a two-way and three-way distinction. This supports the idea that this acquisition pattern does not reflect a gap in the morphological paradigm, but is due to syntactic reasons. These facts lend support to the Weak Continuity Hypothesis and to a minimalist view of the language faculty. At Stage II, finally, definite articles emerge and the target–like word order is observed. Incremental acquisition of CA shows that the DP is acquired gradually. Notice that, according to Clahsen, Eisenbeiss & Vainikka (1994), Clahsen, Eisenbeiss & Penke (1996), and Eisenbeiss (2000) for the acquisition of the German DP within the Weak Continuity Hypothesis and the Lexical Learning Approach, it is predicted that early child grammar may generate underspecified functional projections. The Lexical Learning Approach predicts that:

early determiners may be impostors and the U-shaped developmental curve should signal the change from the impostor status of determiners to the targetlike status of determiners;

S F O PRO



1st proofs

Introduction



if the DP is underspecified early determiners may not show case marking, given that D0 is the locus of nominal case features; lexical entries for the different determiners hosted in D0 may be acquired incrementally.

This is in line with the theoretical assumptions underlying the structural presentations in (12) and (13) above.

.

Issues in the contributions to this volume: An overview

The following theoretical and empirical points on the DP-issue have been raised and brought up for further clarification in the present book publication: Article category and its particular referential and anaphoric properties: Quite obviously, grammatical determiners of various sorts are differently distributed in individual languages. Abraham (“Discourse binding: DP and pronouns in German, Dutch, and English.”) lists and investigates briefly a few related and non-related languages to survey the lexical lexemes relating to Determiner (DemPro) status vs. Article status and pronominal anaphor (PersPro). The author points out, first, to which extent such determiners co-define anaphors in contexts reaching beyond the single clause. Second, he investigates typologically what the determinerdetermined features are where they are in interaction with aspect and morphological case. Third, and interlinking the synchronic and the diachronic chapters, since spoken-only codes use anaphoric determiners in ways strikingly different from their written(-only) standard varieties, parsing differences will be made responsible for such a variation. Not quite in line with Abraham on the status of DP, Bartra takes the existence of Bare Noun Phrases acting as subjects or displaced complements as evidence for the licensing properties of Functional Categories other that Determiner Phrase in earlier stages of Romance Languages such as Spanish and Catalan. The fact that in Spanish and Catalan subjects in passive sentences can be licensed inside the Verb Phrase also facilitates the use of BNPs in this context. Bartra takes Old Spanish and Old Catalan Bare Noun Phrases, mainly subjects of passive sentences and other BNPs moved from their basic position inside the VP, to show that in these grammars the grammar of the Determiner was still unstable and other nominal functional projections were able to license the NP, mainly noun markers such as Gender or nominalizing affixes and Number Phrase. She adopts a methodological minimalism in that she argues that from the inventory of Functional Categories in Universal Grammar, particular grammars activate only those for which there

S F O PRO



1st proofs




is formal and morphological evidence. Needless to say that Noun Phrases show semantic differences according to the active Functional Categories they have. Identification and adequate description of all functional nominal categories across languages: A generally overlooked and highly polysemous English noun phrase form, the bare singular, is looked at by Smith Stvan (“The functional range of bare singular count nouns in English”). The author takes this to be a null determiner with a singular count noun complement assuming a strict distinction between determination and (in)definiteness as well as between nominal determination and nominal classification. Occurring in all grammatical positions, this constituent shape is used in English for multiple functions. Examination of naturally occurring English data shows the conditions under which bare singulars are used as generics (meaning bare plurals), as components of a predicate conveying a stereotypical activity (with an indefinite meaning), and as markers of an identifiable referent (like nouns with definite articles, demonstratives, and possessive determiners).

Different stages of grammaticalization and different functional and semantic categories like specificity or negation: Such aspects are discussed in the contributions by Kupisch & Koops (“The definite article in non-specific direct object noun phrases: Comparing French and Italian”) for modern French and Italian, and by Jäger (“ ‘No’ changes: On the history of German indefinite determiners in the scope of negation”) for the history of German. According to Kupisch & Koops, Italian allows the definite article to occur in non-specific noun phrases forming part of verb+object constructions like mettersi la giacca ‘put on a jacket’ or comprare il pane ‘buy bread’. The corresponding constructions in French typically take the indefinite article, as in se mettre un blouson and acheter du pain, respectively. The authors interpret this phenomenon as indicative of the different degree of grammaticalization the definite article has attained in the two languages. Given that Italian makes much wider use of the definite article in these non-specific constructions we conclude that the Italian definite article is further grammaticalized than its French counterpart. This conclusion calls for a reconsideration of the widespread view that French has the most grammaticalized article system in all of Romance. Jäger’s paper investigates the evolution of nominal determination of a specific kind, viz. indefinite determination in the scope of negation. The changes within the system of indefinite determination in the history of German with respect to four patterns are described on the basis of their distribution in a corpus of several Old and Middle High German texts.

Description and typological explanation of the interaction of nominal and verbal determination at the sentence level: Compare the contributions by Abraham (“The discourse-functional crystallization of the historically original demonstrative”)

S F O PRO



1st proofs

Introduction



and Leiss (“Covert patterns of definiteness/indefiniteness and aspectuality in Old Icelandic, Gothic, and Old High German”) in the history of German. The findings of his synchronic discussion (see above) has a follow-up in Abraham’s second contribution on the emergence of the definite article in the history of German from the determiner homonym and how various steps in their grammaticalization paths invite certain generalizations of diachronic change. Leiss claims that the growth of the definite article is due to changes in the aspectual system of a language. Definiteness and perfective aspect are shown to be just two instantiations of an identical grammatical function. So are indefiniteness and imperfective aspect (see for this idea Krifka 1989). The central claim is that the definiteness effects of verbal aspect upon its ‘nouny syntactical neighbourhood’ suffice to create complex patterns of nominal determination. The complexity is created by combining aspect with a paradigmatic case system. One central claim with typological consequences is that paradigmatic case alternations are characteristic of aspect languages such as the Slavic languages. The same holds for older stages of the Germanic languages. There is converging evidence from linguistic typology that aspect languages tend to avoid article systems, and article languages tend to avoid aspect. The different stages of article development are analyzed in the light of Abraham’s Centering Theory, where grammatical determiners of variant semantic and syntactic purport are distinguished in contexts reaching beyond the single clause. Lohndal argues that double definiteness developed during ON as an instance of “downward” grammaticalization of the definite article. This entails the presence of a low definiteness head. He argues that this head developed alongside double definiteness. As to some movement puzzles in ON and MI, it is held that the two languages require different analyses. The MI cases appear to be straightforward handled as an instance of αP blocking nP, whereas this blocking presumably does not occur in ON Emergence of DP within a perspective of ontogeny and phylogeny as well as the correlation between DP, TP, and aspect in Old English and first language acquisition as by Osawa (“The emergence of DP from a perspective of ontogeny and phylogeny: correlation between DP, TP and aspect in Old English and first language acquisition”) and Bittner (“Early functions of definite determiners and DPs in German first language acquisition”) from the point of view of language-acquisition in German. Osawa aims at the English NP and its historical development into DP via the emergence from zero of a functional D-system due to a theta-binding mechanism. Furthermore, it is examined if there is a complementary distribution between DP/TP and aspect. Crucially, the correlation between DP/TP and aspect is observed in early child languages. This suggests that there might be a parallel between first language acquisition (ontogeny) and diachronic change (phylogeny).

S F O PRO



1st proofs




On the other hand, Bittner discusses the functional load of definite DPs in early child German regarding sentence-internal and sentence-external relations. It is argued that DPs exhibit functional load on both levels from the onset of production. Contradicting recent assumptions, (i) case-related distinctions are acquired prior to gender distinction, and (ii) children establish a functional distinction between pronominal DPs (continued or directly accessible reference) and noun-including DPs (disrupted or especially emphasized reference). The results allow the hypotheses that the noun-including DP is a functionally motivated extension of the pronominal DP and that properties relating the DP to other elements of sentence/text are the first to be acquired. Lohndal argues in his purely historical discussion that the definite suffixed article in Modern Norwegian developed from a clitic in Old Norse. Such a change creates interesting theoretical questions as to how we can account for this difference in phrase structural terms, and how such a change manifests itself. This paper discusses exactly this question and argues that this change can be viewed as grammaticalization “down the tree” from a high D head to a low n head. Furthermore, it argues that functional categories, like the definiteness category, are non-universal. That is, they are not part of Universal Grammar, but only arise when the child discovers them in the input. Bartra concludes that, may be in a very early stage, Gender Phrase or a nominalizing affix would suffice in some case to license an internal subject, but I have no concluding evidence in that sense. The relation between demonstratives and possessives emergent from Old English and developing to present-day English by Wood (“Demonstratives and possessives: From Old English to present-day English”). Three different nominal word orders in Old English through present-day English are investigated in order to determine whether English has an ‘adjectival’ possessive similar to modern Italian. It is argued that the orders of the demonstrative-possessive-noun and the possessivedemonstrative-noun represent different syntactic constructions, with different paths of development, much in line with Abraham’s findings.

The definite and indefinite articles in Indo-European and the question whether its emergence results in what Bauer (“The definite article in Indo-European: Emergence of a new grammatical category?”) calls a “new grammatical category”. Based on the distinction between weak and strong (short/long) adjectives in Germanic and Slavic, as well as to a correlation between case use on direct objects and aspect conveying definiteness, she concludes that definiteness did exist in Indo-European before the rise of the definite article. However, with several of the Near-Eastern daughter languages (among them Persian) and the Slavic languages not possessing articles, the definite article is concluded to be a complex innovation in Indo-European. Furthermore, not all daughter languages underwent the change or, conversely, some did, whereas the majority did not (Bulgarian did obviously due

S F O PRO



1st proofs

Introduction



to L-contact in the Balkan Sprachbund, while the other Slavic languages did not). Moreover, the functions of the definite article may vary cross-linguistically within a subgroup, possibly reflecting different degrees of grammaticalization. In contrast to earlier historical stages of Germanic languages, the connection between aspect, object case, and definiteness has not been attested for Latin despite the fact that Latin presumably had aspect and Aktionsart. It is important to note from this perspective that the definite article in Latin/Romance is attested to emerge primarily in combination with (topicalized) subjects (a claim that is explicitly contested by Selig 1992 on solid empirical grounds). The considerable variation in the emergence of nominal (in)definiteness in Indo-European as claimed by Bauer is in line with Stark’s contribution on the variation in the field of nominal indefiniteness in Romance (“Gender, number, and indefinite articles – about the ‘typological inconsistency’ of Italian”). Stark focuses on the central role of gender and number distinctions in Latin. She suggests that the loss of gender and number distinctions in the Romance languages catalysed the grammaticalization of the indefinite article which initially had partitive function. Her claim is based upon the observation of the functions of multiple gender in Latin. According to Stark, it was the demise of multiple gender distinctions (e.g. Late Latin caseus vs. caseum for “piece of cheese” vs. “cheesiness; of cheese”) which led to emergence of the indefinite article. Note, that in Germanic there has been meanwhile been documented a system of multiple gender which has been lost. As gender, number, and definiteness are closely related grammatical categories, Stark’s results have to be seen in the light of very similar earlier findings made for Germanic and a host of non-Indo European languages. As one additional important point of discussion the discussion of the Cologne workshop has shed light on questions of lack of, or seemingly redundant, nominal determination. Bare nouns in English are morpho-syntactically underspecified and therefore open to a manifold, yet not just any interpretation (cf. Smith-Stvan). By contrast, DPs in French and Italian with a surprisingly non-specific interpretation are apparently overspecified (cf. Kupisch & Koops). Although research for a finite list of indispensable functional categories inside nominals has not yielded a definite result shared by everyone, two decisive factors involved in the interpretation and grammatical evolution of nominals have been clearly identified:

the close, local syntactic context as amply demonstrated by Jäger for indefinite determiners in the history of German; and the role of previous (backtracked) and ensuing (focused by forward-looking) context, as discussed by Abraham for the synchronic and diachronic account of definite determiners and pronouns in German.

S F O PRO



1st proofs




It is claimed that this particular research for the fundamental interaction of determiners and bare nominals with other elements in the phrase, sentence, or text has been successfully dealt with. In particular, the interaction of nominal and verbal determination has been identified as crucial for the process of language change in the nominal structure as well as for the process of language acquisition and evolution.

References Abney, Steven Paul. 1987. The English Noun Phrase in its Sentential Aspect. PhD Dissertation, MIT. Abraham, Werner. 1997. The interdependence of case, aspect and referentiality in the history of German: The case of the verbal genitive. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 29–61. Cambridge: Cambridge University Press. Abraham, Werner. This volume a. The discourse-functional crystallization of the historically original demonstrative. Abraham, Werner. This volume b. Discourse binding: DP and pronouns in German, Dutch, and English. Auwera, Johan van der. 1980. Determiners. London: Routledge. Benveniste, Emile. 1974. Les transformations des catégories linguistiques. Problèmes de linguistique générale, 126–36. Paris: Gallimard. Bloomfield, Leonard. 1933/1965. Language. New York NY: Holt, Rinehart & Winston. Boucher, Paul. 2005. Definite reference in Old and Modern French: The rise and fall of DP. In Gramaticalization and Parametric Variation, Montserrat Batllori, M.-Lluïsa Hernanz, M. Carme Picallo & Francesc Roca (eds), 95–108. Oxford: Oxford University Press. Chan Hok-Shing, Brian. 1999. Classifers, demonstratives, and classifier-to-demonstrative movement. UCL Working Papers in Linguistics 11: 227–251. Chierchia, Gennaro. 1998. Reference to kinds across languages. Natural Language Semantics 6(4): 339–405. Clahsen, Harald, Eisenbeiss, Sonja & Vainikka, Anna. 1994. The seeds of structure: A syntactic analysis of the acquisition of case marking. In Language Acquisition Studies in Generative Grammar [Language Acquisition and Language Disorders 8], Teun Hoekstra & Bonnie D. Schwartz (eds), 85–118. Amsterdam: John Benjamins. Clahsen, Harald, Eisenbeiss, Sonja & Penke, Martina. 1996. Lexical learning in early syntactic development. In Generative Perspectives on Language Acquisition. Empirical findings, theoretical considerations, crosslinguistic comparisons [Language Acquisition and Language Disorders 14], Harald Clahsen (ed.), 129–159. Amsterdam: John Benjamins. Coene, Martine & D’hulst, Yves (eds). 2003a. From NP to DP. Vol. 1: The syntax and semantics of noun phrases [Linguistik Aktuell/Linguistics Today 55]. Amsterdam: John Benjamins. Coene, Martine & D’hulst, Yves (eds). 2003b. From NP to DP. Vol. 2: The expression of possessives in noun phrases [Linguistik Aktuell/Linguistics Today 56]. Amsterdam: John Benjamins. Delfitto, Denis & Schroten, Jan S. 1991. Bare plurals and the number affix in DP. Probus 3(2): 155–185. Dobrovie-Sorin, Carmen 2001. Génitifs et déterminants. In Typologie des groupes nominaux, Vol. II, Georges B. Kleiber, Brenda Laca & Liliane Tasmowski (eds), 205–234. Rennes: Presses Universitaires de Rennes.

S F O PRO



1st proofs

Introduction Eisenbeiss, Sonja. 2000. The acquisition of the DP in German child language. In Acquisition of Syntax. Issues in comparative developmental linguistics, M-A. Friedemann & Luigi Rizzi (eds), 26–63. London: Longman Gil, David. 1987. Definiteness, noun phrase configurationality, and the count-mass distinction. In The Representation of (In)definiteness [Current Studies in Linguistics 14], Eric Reuland & Alice G.B. ter Meulen (eds), 254–269. Cambridge MA: The MIT Press. Giusti, Guiliana. 1997. A unified structural representation of (abstract) case and article. In Studies in Comparative Germanic Syntax, Hubert Haider et al. (eds), 77–93. Dordrecht: Kluwer. Greenberg, Joseph H. 1978. How does a language acquire gender markers? In Universals of Human Language,Vol. 3, Joseph Greenberg (ed.). Stanford CA: Stanford University Press. Guérin, Jacqueline. 2006. Generic sentences and bare plurals. In Non-definiteness and Plurality [Linguistik Aktuell/Linguistics Today 95], Svetlana Vogeleer & Liliane Tasmowski (eds), 219–236. Amsterdam: John Benjamins. Hawkins, John A. 1978. Definiteness and Indefiniteness. A study in reference and grammaticality prediction. London: Croom Helm. Heim, Irene. 1988. The Semantics of Definite and Indefinite Noun Phrases. New York NY: Garland. Heim, Irene & Kratzer, Angelika. 1998. Semantics in Generative Grammar. London: Blackwell. Higginbotham, James. 1985. On semantics. Linguistic Inquiry 16: 547–593. Jakobson, Roman. 1957. Shifters, Verbal Categories, and the Russian Verb. Harvard: Harvard University Press. Kamp, Hans & Reyle, Uwe. 1993. From Discourse to Logic. Introduction to model-theoretic semantics of natural language, formal logic and Discourse Representation Theory. Dordrecht: Kluwer. Krifka, Manfred. 1989. Nominalreferenz und Zeitkonstitution: Zur Semantik von Massentermen, Pluraltermen und Aspektklassen. Munich: Fink. Kurita, Yukari. 2006. Nominale Quantifikation im Deutschen und Japanischen. Magisterarbeit im Fach Germanistische Linguistik, LMU München. Leiss, Elisabeth. 2000. Artikel und Aspekt: Die grammatischen Muster von Definitheit. Berlin: Mouton de Gruyter. Longobardi, Giuseppe. 1994. Reference and proper names: A theory of N-movement in syntax and logical form. Linguistic Inquiry 25: 609–665. Longobardi, Giuseppe. 1996. The syntax of N-raising: A minimalist theory. OTS Working Papers. Utrecht. Longobardi, Giuseppe. 2005. Toward a unified grammar of reference. Zeitschrift für Sprachwissenschaft 24(1): 5–44. Lyons, Christopher. 1999. Definiteness, Cambridge: Cambridge University Press. Marinis, Theodoros. 2003. The Acquisition of the DP in Modern Greek [Language Acquisition & Language Disorders 31]. Amsterdam: John Benjamins. Paul, Hermann 1909/91975. Prinzipien der Sprachgeschichte. Tübingen: Niemeyer. Pereltsvaig, Aissa 2006. Small nominals. Natural Language and Linguistic Theory 24: 433–500. Progovac, Liliana 1998. Determiner phrase in a language without determiners. Journal of Linguistics 34: 165–179. Rijkhoff, Jan. 2002. The Noun Phrase. Oxford: Oxford University Press. Ritter, Elisabeth. 1995. On the syntactic category of pronouns and agreement. Natural Language and Linguistic Theory 13: 405–443. Seiler, Hansjakob (ed.). 1978. Language universals. Papers from the Conference held at Gummersbach/ Cologne, Germany, October 3–8, 1976, Tübingen: Narr.

S F O PRO



1st proofs





Nominal determination Selig, Maria. 1992. Die Entwicklung der Nominaldeterminanten im Spätlatein. Romanischer Sprachwandel und lateinische Schriftlichkeit. Tübingen: Narr. Stark, Elisabeth. 2005. Explaining article grammaticalization in Old Italian. In Romanistische Korpuslinguistik II. Korpora und diachrone Sprachwissenschaft, Claus D. Pusch, Johannes Kabatek & Wolfgang Raible, 455–468. Tübingen: Narr. Stark, Elisabeth. 2006. Indefinitheit und Textkohärenz. Entstehung und semantische Strukturierung indefiniter Nominaldetermination im Altitalienischen. Tübingen: Niemeyer. Szabolcsi, Anna. 1994. The noun phrase. In Syntax and Semantics 27: The structure of Hungarian, Ferenc Kiefer & É. Katalin Kiss (eds), 179–274. New York NY: Academic Press. Tanaka, Shin. 2006. Satzstruktur und Textorganisation – deiktische und anaphorische Referenzstrukturen. Ms. Chiba/Munich. Vincent, Nigel. 1997. The emergence of the D-system in Romance. In Parameters of morphosyntactic change, Ans van Kemenade & Nigel Vincent (eds), 149–169. Cambridge: Cambridge University Press. Vogeleer, Svetlana & Tasmowski, Liliane (eds). 2006. Non-definiteness and Plurality [Linguistik Aktuell/Linguistics Today 95]. Amsterdam: John Benjamins. Zamparelli, Roberto. 2000. Layers in the Determiner Phrase. New York NY: Garland.

S F O PRO



1st proofs

Discourse binding DP and pronouns in German, Dutch, and English Werner Abraham Universität Wien, Austria

Grammatical determiners of various sorts are differently distributed in individual languages. We will list and investigate briefly a few related and non-related languages to survey the lexical lexemes relating to Determiner (DemPro) status vs. Article status and pronominal anaphor (PersPro). A synchronic line of discussion will be pursued. It will be crucial to see synchronically, first, to which extent such determiners co-define anaphors in contexts reaching beyond the single clause. Second, it will be investigated typologically what the determiner-determined features are where they are in interaction with aspect and morphological case. Third, and interlinking the synchronic and the diachronic chapters, since spoken-only codes use anaphoric determiners in ways strikingly different from their written(-only) standard varieties, parsing differences will be made responsible for such a variation.1

Deictic preamble In many a (not only generative) syntactic description and analysis, deictic pronouns such as the article pronoun (ArtPron) as in (1a) as well as the demonstrative pronoun (DemPro) as in (1b), together with the Personal pronoun (PersPro) illustrated by (1c), are the unloved and totally unnoticed foster siblings of the clause internal Reflexive anaphor (discussed under the binding generalization, Binding Principle A; Chomsky 1981). The main reason is that each of the deictics, ArtPro/ DemPro and PersPro, extend a binding reach beyond the simple clause. The following crucial assumptions about ArtPro, DemPro, PersPro, and the definite article (DefArt)+NP need to be made at this early point. PersPro and

. This paper has profited immensely from discussions with Elisabeth Leiss (Munich). Her input and unrelenting scrutiny is hereby gratefully acknowledged. Thanks are due also to Elisabeth Stark for scrutinously thinking along.

S F O PRO



1st proofs



Werner Abraham

DefArt(+NP) are on a par categorially as well as discourse-functionally: They represent full DPs and are themata – i.e. they are complex nominals, SpecDPs. ArtPro and DemPro, on the other hand, while sharing with PersPro maximal projection status, are rhemata. ArtPro occurs only with focus emphasis; DemPro with an empty NP-slot also does. This recourse to discourse functions is crucial for understanding the ArtPro phenomenon. Now, what exactly are form and function of ArtPro. See (1a–c). [  for clausal rhematic accent]. (1)

a. b. c. d.

Nicht alle, nur /*der  ein Gauner Not all, only the (= this one) is a rogue D/Dieser hat das auf dem Gewissen This (one) has it on the conscience E/Er hat das auf dem Gewissen He has this on his conscience Der Gauner is  The rogue is him

…

ArtPro

…

DemPro

…

PersPro

…

(definite) art(icle)

It is accepted knowledge in historical linguistics that the definite article (DefArt) in German, as in many other languages, derives historically from the demonstrative function of the ArtPron – i.e. from its earlier rhematic use. We shall unfold a grammaticalizing path for this modern emergence from its earlier homonymic forms in the last section of the present paper. But in order to do this beyond simply following well-blazen paths I shall develop an analytic theory encompassing binding beyond the simple sentence, both backward and forward looking. The background for this attempt is Centering Theory ; see Abraham 2002 based in turn on Grosz/Joshi/Weinstein 1986, Rambow 1993, Walker/Joshi/Prince 1998 as well as others. Our point of departure is the general idea that the diachronic predecessor of the modern German definite article and ArtPron arose by reducing original (Old High German, Middle High German) feature clusters, mainly co- and context deictic, to eventually yield the polyambiguous (highly underspecified) modern article form. Since, according to Heim (1988) and most other writers, the definite article and the PersPro share the main binding feature of anaphoricity, the two categories have to be analyzed as to their distinct differences. This will allow us to sketch more appropriately the diachronic feature reduction in the grammaticalizing process. In due course, the reduction of the definite article (DefArt) to an anaphoric function (Heim 1988) will be illuminated critically (see also Leiss 2000). Table 1 sketches, in necessary detail, what we shall be dealing with in the remainder of the paper.

S F O PRO



1st proofs

Discourse binding



Table 1. Projection status of article and pronominals with coreference reach beyond the single clause

DefArt Def(Art+)NP PersPro ArtPro DemPro

Thema status

Rhema status

SpecDP status

Status under contrastive Focus

+ + + − (+?) −

− − − + +

− + − + +

(D0 >) SpecDP (remains) SpecDP – raises to FocP* (remains) SpecDP – raises to FocP* (remains) SpecDP – raises to FocP* (remains) SpecDP – raises to FocP*

Notice the congruence between DefArt and PersPro in regard to the uptake of a prior Thema, on the one hand, and ArtPro and DemPro in all features, on the other hand. This is totally plausible. There are no distributional differences between ArtPro and DemPro. This is no luxury of the language in question: (contrastive) focussed article (+ NP) vs. ArtPro, which occurs only in focus position, vs. foused vs. unfocused DemPro. In a discourse-prominent language like German, there are normal (default) Rhema focus and extra-contrastive focus. The two foci are in different derived positions: Non-default Rhema (DemPro) as well as ArtPro are either in FocP above CP (Rizzi expansion: FocP*) or left of vP in the German middle field (between C0 and SpecvP; this is what “FocP*” is meant to signal; see Abraham, this volume). This paper divides into two parts: the synchronic part discussing not only the definite article, but, above all its solitary use as article pronoun (ArtPro) which is unique for German and Dutch. The results from this discussion will be used in the second, diachronic part to explain the emrgence of the definite article from ArtPro.

. .

Binding vs. Centering Text-related centering

The principled assumption (not wholly in line with Heim 1988) is that today’s definite article, as a subclass of determiners, Det, has a twofold function. According to the co- and context dependence of determiners, DefArt has a referential specification and, independently thereof, a context dependence, i.e. one of anaphoric contextualization. The latter, clause-unbound relation is shared by other categories such as the personal pronoun, while the former function is singular to the definite article.

S F O PRO



1st proofs



Werner Abraham

This may be motivated by the historically erstwhile deictic, non-anaphoric signal that the DefArt extended (van Benthem & ter Meulen (1984: 45–71), Leiss 2000, Westerståhl 2005). The overall perspective against which this distinction can be seen is, on the one hand, that of the discourse universe, which is constant over pieces of discourse, and, on the other hand, that of context sets, which are variable in discourse. Following this, determiners are seen as restricted to context-set variables (a relativisation that allows treating definites and partitives uniformly; see Westerståhl 2005). I restrict myself mainly to the anaphoric features and function of DefArt and will compare them with those of other, related anaphoric and deictic categories, among which PersPro, ReflPro, and ArtPro/DemPro. Counter to tradition, where nominals are the center of feature reference – i.e. where they are looked at under the perspective where and by what category they are coreferred (either anaphorically or cataphorically/as ante- or postcedents), pronouns and anaphors themselves can be taken to be the ‘center’ of attention. In the light of this novel approach we pursue the question where in the text chain, and by what category, they are fully feature-specified and where the distance and the linearly mediating structure do not permit coreference any longer, respectively. See the following inventory of relations in (2) (following essentially Kameyama (1998) and Walker, Joshi & Prince (1998). (2)

Un = for n appellative/common noun text elements as part of the narrative universe as well as 1  >      it this/that that NP the NP this NP an NP this NP PersPro DemPro Dem+NP defArt+NP Dem+NP indefinite Dem+NP NP

S F O PRO

It will be seen from concrete text illustrations that there are cross classifications at hand constantly invalidating partly the polarity cline and its lexical representations in (13).



1st proofs



Werner Abraham

(14)

The this-NP Hypothesis (for English; cf. Poesio/Modieska 2005: 431): -NPs are used to refer to entities which are activated. However, pronouns should be preferred to -NPs for entities  .

Since neither English this and that nor the can be used as DemPro referring to nominals and ArtPro as opposed to the Continental West Germanic languages (German, Dutch, West Frisian, Yiddish), (14) in any case does not apply to these languages. Notice that we said that the very fact that ArtPro, next to DemPro, occur makes a substantial component of the typology of a discourse prominent language (Abraham 2006).

.

Consequences of the ArtPro/DemPro-selection

According to (15a), ArtPro/DemPro selection interrupts any continuation of that thema that has not been introduced by PersPro. Question is: What is the discoursefunctional result of ‘ArtPro/DemPro1−ArtPro/DemPro2’: a new rhema or a ‘ArtPro/ DemPro-thema’? (15)

a.

Rhema continuation of DP2: (S1:) Hans1 traf Alfons2. (S2:) Der2 trug einen Regenmantel.5 (S3:) Er2 war bleich im Gesicht. No grammatical reference: b. *Hans1 traf Alfons2. Der2 trug einen Regenmantel. Der2 fror trotzdem. c. Hans1 traf Alfons2. Er1 trug einen Regenmantel. Er1/*Der1/2 fror trotzdem. d. Hans1 traf Alfons2. Es goß. ?Er ?Er

trug einen Regenmantel. /*Der 1 1/2 fror trotzdem. 1

Hans ran into Alfons The wore a raincoat He was pale in the face Hans ran into Alfons The wore a raincoat The was cold anyway Hans ran into Alfons The wore a raincoat He/The was cold anyway Hans ran into Alfons. It was raining hard He wore a raincoat He/The was cold anyway

(15b, c) present incoherent minimal discourses. ArtPro/DemPro cannot refer to any PersPro or ArtPro/DemPro. In (15c) the discourse chain is interupted in such

. The pronominal article (ArtPro) in German, der/die/das “the...”, and the demonstrative proper, dieser, diese, dies(es) “this...”, do not differ distributionally. In the ensuing distribution tests, ArtPro der/die/das stands for both anaphoric-deictic categories. Notice that neither goes without deictic emphatic focus.

S F O PRO



1st proofs

Discourse binding



a way that no rhema take-up can save S1: Der1 is out because of (11b), Der2 is out on the basis of (16). (16)

ArtPro/DemPro must refer to a DP; ArtPro/DemPro-reference to a PersPro-antecedent is impossible.

However, (16) must be further constrained to account also for (15d). (17)

a.

b.

Text coherence under the PersPro or ArtPro/DemPro-reference mechanism must ideally be monotonously dense. In other words, there must not be an intervention of any text-S such that the Pro-chain is interrupted. Our parlance will be that ArtPro/DemPro is ‘text-coherence deselective’. (17a) overwrites (16).

(17a) is supported and completed by the following illustrations. [  contrastive focus]. (18)

Alfons1/A2 trat ein. Alfons entered. Er1/Der2 trug einen Regenmantel. Es goß noch immer. He/The wore a raincoat. It kept raining hard. ?Er /*Der schlug sich den Regen von seinem nassen Mantel. 1 2 He/The beat the rain from his coat.

From (18) the following generalization derives: (19)

The text-coherence condition for antecedent-Pro-coreference is less limited for PersPro than for ArtPro/DemPro. PersPro is less text-coherence deselective than ArtPro/DemPro: Text-Coherence Deselection Hierarchy: PersPro < ArtPro/DemPro

As shown above, PersPro is stronger than ArtPro/DemPro with respect to creation of text-coherence. In other words, it is less susceptible to noise within the monotonous text densitiy. See (20) Recall that generative grammar considers only anaphoric reflexives for clause-internal binding and that it excludes PersPro, ArtPro (DemPro) and R-expressions from the binding relation. (20)

a.

Text Coherence Deselection Hierarchy: R-expression (common noun, proper noun) < PersPro < ArtPro/DemPro < Text reflexive

No doubt this is reminiscent of Givón’s topic-continuity scale (Givón 1983). But I refrain from comparing details and presuppositions. (20)

b.

 /  zero anaphora unstressed/bound pronouns or grammatical agreement stressed/independent pronouns

S F O PRO



1st proofs



Werner Abraham

Right-dislocated DEF-NPs neutral-ordered DEF-NPs Left-dislocated DEF-NPs Y-moved NPs (‘contrastive topicalization’) (=focussing topicalization) cleft/focus constructions referential indefinite NPs   ⁄  

As for violations of (20a) with respect to inceptive text portions the following appears to hold: From among all reference introducing pronouns, the reflexive pronoun is the least appropriate, and ArtPro/DemPro is the worst but one (both ungrammatical and referentially inappropriate). PersPro, to be true, can be referentially bound cataphorically, while ArtPro/DemPro can be referentially incorporated neither anaphorically nor cataphorically, on a par with ReflPro. Let us now investigate how far we get with incorporating the phenomena and generalization about them in the framework of Centering Theory.

.

Priority conditions within the perspective of Centering Theory

The discussion and distributional results presented so far gain a stronger systematic status if incorporated in the methodological framework developed within Centering Theory (as introduced by Grosz et al. (1986/1995) and Walker et al. (1998). Centering Theory allows for generalizations on text continuance and text coherence on the strength of anaphoric categories coreferring with DP-text participants (appellative/common nouns/common nouns, proper names). The inventory of terms in Section 2 will play a role in the ensuing discussion. Linear orders and priorities among appellative/common noun text participants, DPs, are unfolded from within the perspective of the new text participants (Cf, where {Cf}⊂Un). The antecedent DP plays a particular role in this perspective: i.e. all types of Pro-nominals taking up the reference (“coreferring”) of this antecedent and indicating the most appropriate discourse-activated text participant (Cb,6 where {Cb}⊂Un). The most prominent element, Cp, is referred to by the highest ranking Cb. Thus, Cp is that Cb

. Cb = backward-looking center (anaphor of some status: PersPro/ArtPro/DemPro) as member of some world (knowledge), Universe n. The terminology is derived from Greek grammatology: anaphor(a) from the verb anapherein (‘carry from below to above’ or ‘look ahead’) as opposed to kataphor(a) (‘carry from above to below’ or ‘look backward’). Notice the distinct use which is opposed to the generative terminology: anaphor(a) is the reflexive pronoun thoughout limiting its coreference relation to the borders of the single clause.

S F O PRO



1st proofs

Discourse binding



ranked highest, or preferred the most, among all Cbs.7 Cp mirrors the idea that every text participant once it has been introduced carries with it a specific persevering moment. It is priorized text re-referent as long as a new text participant has been introduced. See (21b) abbreviating this idea. (21)

Cp(Un) = Cb(Un) = Cb(Un-1)…; see (11a): “PersPro is thema continuant”

Cp imposes partial ordering onto the set of Cb. Let us briefly illustrate its applicability to the distributional distinctions between PersPro and ArtPro/DemPro. (22)

U1: Hans1 verließ das Zimmer. H. left the room U2: Er1 stieß auf Rolf2. He ran into R. U3: Er1 grüßte ihn2. He said hello to him

Cf = [Hans]; Cb is open Cf = [Hans>Rolf]; Cb1 = Hans Cf = [Hans>Rolf]; Cb1 = Hans = Cp

(23)

U1: Hans1 verließ das Zimmer. Cf = [Hans]; Cb is open H. left the room U2: Er1 stieß auf Rolf2. Cf = [Hans>Rolf]; Cb1 = Hans He ran into R. U3: E2 grüßte 1. (contrastive focus!) H said hello to  Cf = [Rolf>Hans]; Cb1 = Hans = Cp

(24)

U1: Hans1 verließ das Zimmer. H. left the room U2: Er1 stieß auf Rolf2. He ran into R. U3: D2 grüßte ihn1. T said hello to him

Cf = [Hans]; Cb is open Cf = [Hans>Rolf]; Cb1 = Hans = Cp Cf = [Rolf>Hans]; Cb1 = Hans = Cp

Since subjects function as unmarked themata (‘topics’) in German, it is plausible to assume that Subject is the Cb with highest priority (see also Grosz, Joshi & Weinstein 1986, 1995). This generalization has been found to hold also for Japanese (Kameyama 1985), English (Hudson-D’Zmura 1989, 1998), Italian (Di Eugenio 1998), Yiddish (Prince 1994), and Turkish (Hoffmann 1995, 1998). For Japanese, however, other constituting criteria were found beyond (25) (Kameyama 1985, confirmed by Walker, Iida & Cote 1994). The grammaticalized topic morpheme, wa, assigns a specific discourse role and speaker-individual saliency to the DP (‘argument empathy’ in Kameyama’s (1985: 141) terminology), where the unmarked locus of saliency is taken over by the subject, whereas the marked one is carried by the . Under this systematic approach, no principled difference is made between anaphor(a) (Cp(Un) – i.e. searching backward for the antecedent DP from the pronoun onwards) and kataphor(a) (Cp(Un+1) – i.e. searching forward for the antecedent DP from the pronoun onwards).

S F O PRO



1st proofs



Werner Abraham

indirect object. According to Turan 1998, in Turkish certain predicate-selected theta roles decide next to (25) which DP takes the highest Cb-role. Rambow (1993) has concluded for German that linear position – criteria of proximity and distance between the referred-to DPs and the referring pronoun have to figure prominently in the explanation. Thus, (25) cannot be considered any more than a first approach to a solution, which will have to be subject to stringent scrutiny. Let us pursue such a probe within the conceptual demarcations exploitable from Centering Theory. The generalization in (25) on relative Cb-priorities is reminiscent of Keenanb & Comrie’s (1977) clause-function hierarchy selecting antecedents for relative pronouns. (25)

Relative Cb-priorities of grammatical categories: Subject > Object > other clausal functions

The ensuing chapters make precise the conditions under which antecedent DPs and/or pronouns are referred to by either PersPro or ArtPro/DemPro. .

ArtPro/DemPro in the perspective of Centering Theory

.

Clausal function as antecedent for ArtPro/DemPro

ArtPro/DemPro assigns paradigm-specific case under verbal government. See (26a, b, d, e), where sich annehmen “care for” governs genitive object case. (26)

a.

b.

c.

d.

e.

Hänschen verletzte seinen Hund. Dessen*Hänschen/Hund/SeinerHänschen/??Hund nahm sich niemand an. Fritzii fürchtete ihre behinderte Turnlehrerinj. Derer*i/j/Ihreri/*j nahm sich niemand an. Fritzii fürchtete ihre behinderte Turnlehrerinj. Der*i/j/Ihri/*j half aber niemand. Die Mädcheni fürchten ihre behinderten Lehrerinnenj. Derer*i/j/Ihreri/*j nahm sich niemand an. Die Mädcheni fürchten ihre behinderten Lehrerinnenj. Den*(en)*i/j/Ihneni/*j half aber niemand.

H. hit his dog. This./his. no one cared for F. was afraid of her disabled gym teacher. This./his. no one cared for. F. was afraid of her disabled gym teacher. The./her. no one helped. Die Mädcheni fürchten ihre behinderten Lehrerinnenj These./They. no one cared for. The girls were afraid of their disabled gym teacher The../Them. no one helped.

S F O PRO

One cannot simply assume that ArtPro/DemPro is a nominalized definite article. Rather, I assume that it is a DP-ellipsis. ArtPro/DemPro obtains the status of Spec



1st proofs

Discourse binding



DP like an article morpheme, albeit in front of a phonologically empty N. This appears to be semantically legitimate since the article requires a quantifying variable in its scope. This accounts for the syntactic fact that ArtPro/DemPro – other than PersPro – can be modified by postposed adjectives and relative clauses. See (27). (27)

a. b.

Fritzii verehrte ihre junge Turnlehrerinj. F. adored her gym teacher. Die(se)*i/j Schöne*i/j … The/This beautiful … Fritzii verehrte ihre junge Turnlehrerinj. F. adored her gym teacher. Die(se)*i/j, [der keine Reckübung zu schwer war]*i/j … The/This to whom no bar exercise was too tough …

The syntactic structure in (28), (28)

[ [’ d(ies)e] [’ [/ Schöne/der keine Reckübung …] [0]]]

permits the direct reconstruction of the diachronic grammaticalizing step from ArtPro/DemPro to the definite article – to the best of our understanding not only in German, but also in the Romance languages (Late Latin ille to French le, albeit also to PersPro il; cf. Vincent 1997). We shall return to this. Our discussion in the distributional criteria between PersPro and ArtPro/ DemPro resulted in (29a, b)–(30). (29)

a. b.

PersPro PersPro

(30)

Art-/DemPro

∈ Cp(Un) as anaphor, i.e. with linear antecedent ∈ Cp(Un+1) as cataphor, i.e. with linear postcedent ∈ Cf(Un−1) only as anaphor, also only with linear antecedent

In other words, ArtPro/DemPro must corefer to an antecedent, just like PersPro (=Cf, i.e. a forward looking center). It cannot take up reference to a preferred appellative/ common noun antecedent of the actual text formation, Cp. For ArtPro/DemPro al defining components are valid except for the preferred PersPro. Referential access of ArtPro/DemPro as compared to PersPro is schiastically inverted. It is either more distal or more proximal than of PersPro, which is either more proximally defined or more distally: see (31a) and (31b). (31) a. [CP1 DP1-thema [VP DP2-rhema]]##[FocP ArtPro/DemPro2 [CP2 PersPro1 [VP V]]] b. [CP1 DP1-thema [VP DP2-rhema]]## [CP2 PersPro1 [VP ArtPro/DemPro2]]

PersPro is always in thema position unless it is displaced and becomes rhematic under refocus. ArtPro/DemPro is rhematic: either in the default CP-rhema position in VP or displaced under contrastive focus in FocP. No doubt, this distance criterion is to be taken both linearly and structurally. Centering Theory does not define the anaphoric binding properties of text portions in specific languages. It is assumed here that non-finite constructions such as infinitive constructions are jumped over when determining Cf. This is in

S F O PRO



1st proofs



Werner Abraham

line with the fact that the discourse functions of thema and rhema are grammatically anchored in finite, but not in non-finite constructions. (32)

Eri versprach/riet ihmj zu kommen. Aber eri/*j (≠ der*i/j) kam nicht.

He promised/suggested to him to come. But he (the) did not come.

Can PersPro ihm in (32) be taken up by ArtPro/DemPro in a discourse chain? (33)

a. Hansi versprach Albertj zu kommen. Der/Dieserj aber glaubte ihmi nicht. b. Eri versprach ihmj zu kommen. Der*i /Dieser*i aber glaubte ihm*j nicht. c. Ei versprach j zu kommen. Der/Dieseri aber glaubte ihmj nicht.

H. promised A. to come. The/This (one) did not believe him. H. promised A. to come. The/This (one) did not believe him. He promised  to come. The/This (one) did not believe him.

For one, (33a–c) demonstrates that centering is indeed restricted to finite expressions. Second, ArtPro/DemPro appears to be subject to constraints stronger than mere distal coreference. More narrowly, ArtPro/DemPro, other than PersPro, does not take up referentially a linearly secondary Cf, i.e. a Cf after another Cf or Cp. Thus, (34) holds. (34)

ArtPro/DemPro ∈⋅ [Cf(Un−1) ^¬ Cb/p(Un−1)]; meaning: ArtPro/DemPro presupposes an antecedent as well as non-thematicity of the antecedent.

In other words, the antecedent of ArtPro/DemPro must not itself be a PersPro, unless this antecedent has the status of a common noun, DP – which is what PersPro switches to categorially when used local-deictically and under contrastive emphasis. See Abraham (1995) and Molnárfi (2002). Rullmann (2001) concludes for Dutch that Cf-chaining is not a matter of grammatical clause function alone, i.e. exclusively of (28). Upon closer scrutiny, however, and focusing on discourse-oriented binding relations, it was demonstrated in Abraham (2002) that Dutch conditions the occurrences of PersPro and Art/DemPro in total consonance with German. See (35). (35)

(34) holds with ArtPro/DemPro ∈⋅ [Cf(Un−1) ^. ¬ Cb/p(Un−1)]; there may also be a Cb/p(Un−1) as an unambiguous antecedent, i.e. PersPro if it is in (contrastive) clausal focus.8

Following (38), also ArtPro/DemPro ∈⋅[Cf(Un−1) ^ Cb/p(Un−1)], if normal or focusinduced rhema status holds for Cb/p(Un−1). Thus, (35) needs to be reformulated. . Discourse functional rhema occurs either as clausal default focus or as contrastive accent. Default focus is identified by the Focus Null Hypothesis (Cinque 1993): clausal default accent is located on the head of the deepest embedded VP constituent. Thus, clause default accent is syntactically anchored and computable in accordance with phonetic accent (see Abraham 1993, 1995).

S F O PRO



1st proofs

Discourse binding

(36)



(34) holds for ArtPro/DemPro ∈ [Cf(Un−1) ^ Cb/p(Un−1)], if Cb/p(Un−1) has rhematic discourse function. (34) holds as long as Cb/p(Un−1) is proper discourse thema.

This is the most general result of our discussion: Rhema status overwrites clause function (subject) status as criterion for the choice of ArtPro/DemPro. In what follows we test whether this holds in purely linear terms – i.e. where no structural criteria have to be brought into play in the first place. .

Linear order as Cf-order condition for ArtPro/DemPro

The main question is which criteria are responsible for the priority serialization of Cf in Dutch. Van Gelderen (1998), Comrie (2000), and Rullmann (2001) have adopted clear stances, which have been discussed and refuted above. A major role has been attributed to the distinct, and different, centering-binding properties of PersPro and ArtPro/DemPro. The following discussion focuses on the question whether it is true that ArtPro/DemPro never take up the first mention of a DP in the proximate previous text sentence. This type of distribution test obviously works only in languages in which the subject can appear also in non-topic position. In Rullmann’s (2001: 5) Dutch corpus, the majority of ArtPro/DemPro occurrences are in close proximity to the Cf-DP in clause-last position. Given that there is a strong tendency toward right-extraposition in Dutch this Cf-finality criterion appears to be too superficial. See (37)–(40) for a selection of Rullmann’s illustration on which he has based his generalizations. (37)

Antecedent-DP in Last-position of the previous sentence: [We hadden vorige week logé’s uit Limburg] en die waren zo enthousiast over Elburg. Wir hatten vorige Woche Gäste aus Limburg, and die(se)/*sie waren so begeistert über Elburg. We had guests last week from Limburg, and the (ones) were so enthusiastic about E.

(38)

[Ik bemest regelmatig grond van andere mensen.] Maar die willen zich daar niet via een contract gedurende meerdere jaren op vastleggen. Ich dünge regelmäßig Land anderer Leute. Aber die(?se)/*sie wollen sich nicht mit einem Kontrakt for mehrere Jahre festlegen lassen. I constantly fertilize land for other farmers. However, the (ones) do not wish to be bound by contracts over several years.

(39)

Antecedent-DP in middle position of the previous sentence: In de vliegsportwereld is het gebruikelijk dat [de piloot soms andere mensen mee neemt op een vlucht.] Vaak betalen die er ook wat voor. Im Fliegersport ist es üblich, that der Pilot manchmal auf einem Flug andere Leute mitnimmt. Oft bezahlen die(se)/*sie auch noch dafür. Among sports aviators it is customary that the pilot taxis other people. Often the (ones) pay for it.

S F O PRO



1st proofs



Werner Abraham

(40)

[Iedereen wordt in de gelegenheid gesteld om verlangens aan het bestuur kenbaar te maken.] Die kunnen dan tijdens het overleg aan de gemeente overgebracht worden. Jedermann wird Gelegenheit gegeben, seine Wünsche bekannt zu geben. Die(se)/ *Sie können dann bei der Ratssitzung der Gemeinde vorgebracht werden. Everyone gets the opportunity to bring forth his requests. The (ones) can be brought to the fore in the council meeting.

For Rullmann (2001: 5) this permits the following provisional generalization: (41)

The determining criterion for Cf-serialization (antecedent position) is purely linear.

(42)

Given Cf(Un−1) as a singular element: its referential take up is by ArtPro/DemPro.

(43)

Given Cf(Un−1) with more than one element: ArtPro/DemPro cannot corefer to the highest ranking element, Cp(Un−1), which means that it cannot take up referentially a discourse-functional thema as antecedent.

(44)

There is a strong, but not unviolable preference for the antecedent, Cf, to be the last DP-member of the expression with respect to the coreferring ArtPro/DemPro.

However, the conclusion in (41–43) based on (37)–(40) cannot be shared – neither for German nor for Dutch. Notice that (43) follows the Grammatical Category Hierarchy for Cbs in (25) – which is counter to expectation given the purely linear linearization criteria. Moreover, (41)–(42) and (44) are dubious. (37), accounted for by (41), can be replaced by (45a) where reference access by ArtPro/DemPro is maintained. (45)

a.

Gäste aus L. hatten wir vorige Woche, and die(se)/*sie waren so begeistert über E. Guests from L. we had last week, and the (ones) were so enthusiastic about E.

But (44b) is not tenable either; cf. (45b) which is a variety about (37) yielding (41) and is compatible with (45a). (45)

b.

Unsere Gäste aus L. blieben eine ganze Woche. Sie/*Di(es)e waren ganz begeistert. Our guests from L. stayed all last week. The (ones)/These were so enthusiastic.

Finally, (44) is incompatible with (45c). (45)

c.

Wir hatten Gäste aus Limburgi und aus dem Institut for Blindej, and siei/*j/diei/*j blieben eine ganze Woche. We had guests from Limburg and from the institute for the blind, and they/the (ones) stayed for a whole week.

(45c), in particular, seems to force the conclusion that clause-function status in the sense of (25) is criterial for the decision between PersPro and ArtPro/DemPro. Before taking over this conclusion let us focus on clause-initial antecedent occurrences. Undoubtedly, they would make trivial Rullmann’s linearity option, since

S F O PRO



1st proofs

Discourse binding



the criterion ‘linearly last clause member’ would preempt the decision based on serialization. (46a, b) roughly presents structural analysis. R-T T (46) a. [ Nur wenige Reaktionenj [ tj registriertenk die Beamteni gestern [ ti tj tk]]]. Aber sie?i/??j/die(se)*i/j waren giftig. Only few reactions.. registered the officers.. yesterday. But they were venomous b. [ In kurzer Zeitm istk [ eine eigene Bevölkerungsgruppe entstanden tk]. Die/*Sie arbeitete in Fabriken, wo nur auf Knöpfe gedrückt werden mußte. In a short period a population in its own right emerged. The (one)/It worked in factories where they only had to press buttons.

ArtPro/DemPro links to the base position of a rhema, i.e. of a subject or object within vP. This is why in (46a) it is not the subject die Beamteni “the officers”, but the topicalized, yet rhematic object Nur wenige Reaktionenj “but few reactions” is taken up referentially by ArtPro/DemPro. The subject-NP in (46b) is a rhema solely due to its category-inherent indefiniteness – as we have claimed above, rhema has the referential takeup by ArtPro/DemPro. Both examples show beyond doubt that neither clause function (subject, object) nor linearly last position in the clause are determinant. What ArtPro/DemPro takes up referentially is determined solely in terms of a discourse-functional rhema in the immediately preceding clause. Examples like (46a), which deviate from the base order, demonstrate beyond doubt that the position of the clause rhema is defined at LF, i.e. as virtual accent position in the German and Dutch clause (recall Cinque’s Accent Null Hypothesis; Cinque 1983). See (47) below which is crucial for German. (47)

a. b.

ArtPro/DemPro is licensed only by an antecedent rhema. The antecedent rhema, which licenses ArtPro/DemPro, cannot be checked on the surface structure; licenser is the rhema element in base – underived – structure, i.e. the head of the most deeply embedded constituent inside VP (Cinque 1993, Abraham 1997a).

German is not only a discourse-functional language whose thema-rhema distribution as well as its clause accent positions are syntactically computed. German also determines discourse-functional links in terms of whether PersPro or ArtPro/DemPro have to be used for purposes of text coherence. In line with this, (48) distinguishes intra-clausal thema-rhema from inter-clausal distributions. (48)

The antecedent rhema licensing ArtPro/DemPro is defined on the level of clause structure. In particular, it is not defined on the derived level through scrambling defining refocused clause positions.9

S F O PRO

. No doubt this needs to be further confirmed since it is in contradiction with (38)–(39), where contrastive accent figures as antecedent requirement for ArtPro/DemPro.



1st proofs



Werner Abraham

(49)–(50) confirm the two discourse-functional principles constituent for German. (49)

Die jüngeren Generationen haben nichts gegen Traditionsverlust, wenn zuweilen auch [Anflüge von Nostalgie]i bei ihnenj spürbar sind. Di(es)ei/*siei aber richten sich eher konkret auf Omas alten Küchenstuhl und alte Mode als auf alte sittliche Werte. The younger generations. are not against loss of traditions, albeit, at times, there is a trace of nostalgia. The/These/They are aimed concretely at grandmother’s old kitchen chair and old fashion more than at old moral values.

(50)

Den dazugehörigen Regenbogenj hat unser Gast/eri tj nicht bemerkt; denn auf die Seite, wo d(ies)erj /*erj sichtbar sein sollte, hatte er keine Aussicht. The rainbow. our guest. has not noticed, since onto that side where it should be visible his vision was blocked.

In (49) and (50), ArtPro/DemPro – not, however, PersPro – takes up coreference. In either case the antecedent is in derived position, i.e. linearly distal to ArtPro/ DemPro: in (38) as subject before a pronoun in category-basic thematic discourse function; and in (39) as object in topicalized position before the subject. In either case surface linearity or clause function membership are excluded as license for coreference. As shown by (50), the antecedent for ArtPro/DemPro cannot be a discourse-functional thema, since discourse-functional PersPro bei ihnen “with them” is a thema irrespective of its position in the clause. See (51). (51)

Die jüngeren Generationen haben nichts gegen Traditionsverlust, wenn bei ihnenj zuweilen auch [Anflüge von Nostalgie]i spürbar sind. Di(es)ei/*siei aber richten sich eher konkret auf Omas alten Küchenstuhl and alte Mode als auf alte sittliche Werte. The younger generations. are not against loss of traditions, albeit, at times, there are traces of nostalgy. The/These/They are aimed concretely at grandmother’s old kitchen stool and old fashion than at old moral values.

(50) Disconfirms the position advocated by Grozs, Joshi & Weinstein (1986) that pronouns als Cf generally rank higher than appellatives/common nouns (“Rexpressions”). This would imply that from among all Cbs PersPro receives Cpstatus. This does not hold for German. It may be assumed that the type of German discourse prominence and anchoring in structural syntax play the determining role. It is not clear why both pronouns in (52), i.e. PersPro and likewise ArtPro/ DemPro, establish the required referential take up. (52)

[Die Herren aus der Zirkusverwaltung]i organisierten das dann. Siei/Di(es)ei gingen später allerdings auseinander. The gentlemen from the circus administration organized that then. Yet, they/the(se) later separated.

S F O PRO



1st proofs

Discourse binding



As (53a, b) demonstrate the antecedent takeups by means of PersPro and ArtPro/ DemPro create different text coherences. (53)

a.

b.

Die ganze Vorstellungstour wurde [von einer professionellen Firma]j-RhemaCf1 vermarktet. [Diese Herren]i-Themasub-Cf1 organisierten das Programm dann. Siei-Cf1/*Di(es)ei-Cf1 gingen später allerdings auseinander. The complete advertising tour was administered by a professional team. These gentlemen then organised the program. Yet, they/the(se) separated later. Die ganze Vorstellungstour wurde [von einer professionellen Firma]j-RhemaCf1 vermarktet. [Die Herren aus der Zirkusverwaltung]i−RhemaCf2 organisierten das Programm dann. *Siei-Cf2/Di(es)ei-Cf2 gingen später allerdings auseinander. The complete advertising tour was planned by a professional team. The gentlemen organised the program. Yet, they/the(se) separated later.

If PersPro is used in (53a), the two linearly separated Cf-antecedents, Cf1, are in a hyper-hypo-set relation to one another. On the other hand, (53b) illustrates the case of two referentially different antecedent sets. ArtPro/DemPro takes up only the linearly last rhema reference, PersPro, on the other hand, is left without any coreference in (53b). This leads to the generalization in (54a, b). (54)

a.

b.

ArtPro/DemPro restricts its antecedent relation to the latest mention of the set-referentially independent rhema, in contrast to PersPro, which presupposes a set-referentially dependent antecedent chain. This antecedent represents the clause-internal thema. PersPro does not find an antecedent in a pure rhema chain (cf. (53b) with *Siei-Cf2), independent of the fact which clause function(grammatical category) is concerned. [Die Herren aus der Zirkusverwaltung]i−RhemaCf2 is subject after all.

(54b) shows beyond doubt that the referential takeup by ArtPro/DemPro is not motivated through clause function (grammatical category) membership (contra Rullmann’s first motivating step). That solely linear serialization is not a successful motivation is linked to the fact that German scrambles thema- and rhema-bearing elements according to discourse-logical criteria. This sheds light again on the fact that motivations for distributions and coreference relations between ArtPro/DemPro and PersPro is in narrow interaction with the syntactically anchored discourse functions and their distributions (CF1-chain) and contrastive discourse functions and their distributions (Cf1 + Cf2; cf. Abraham 1997, Abraham/Molnárfi 2001; see also the independent investigation by Delisles 1993: 262ff.).10 . Gärtner (2001) takes the second clause in (i) to be a paratactic relative clause (whatever that syntactically may be). In other words, our ArtPro/DemPro phenomenon is analyzed as a purely (though not anaylzed in detail) syntactic derivation.

S F O PRO



1st proofs



Werner Abraham

.

Discourse forking – and conclusion

Rullmann (2001: 7) takes the following illustrations to generalize that every DP/NP in Cf referring to a discourse referent is ranked higher than any DP/NP introducing a new discourse referent. Exactly this line of argumentation has been pursued early on in the present discussion. Let us specify this with new material. (55)

Schlepplinien ziehen die historischen Landschaftszüge nach and setzen neue Konturen. Eine Kolonie Kiebitze fliegt verschreckt vor den heranrückenden mechanischen Geräten auf. Die(*se)/Sie kommen im nächsten Jahr nicht mehr zurück. Tow lines map the historical landscape characteristics and draw new contours. A colony of pewits flies up startled by the approaching mechanical tools. They/These will not return the next year.

(56)

Andererseits hat es diesen Einfluß natürlich sehr wohl gegeben. Elco Gelling begann in meiner Truppe, and der(*ser)/er hat später bei den Golden Earrings Furore gemacht. On the other hand, such an influence existed alright. E.C. began in my band, and the/this/he created quite a sensation later with the Golden Earrings.

Notice, first, that Cf consists of one identical NP in both cases. (55) again is a purely thetic (= rhematic) sentence. Although the indefinite subject Eine Kolonie Kiebitze “a colony of pewits” is topicalized the whole clause contains solely rhematic information. Again, this is the rhematic prequisite for ArtPro/DemPro as long as the discourse background, i.e. the general line of thema is represented by Schlepplinien “tow lines”, Landschaftszüge “landscape characteristics and neue Konturen “new contours”. This contrasts with the cluster Eine Kolonie Kiebitze “a colony of peewits”, as new thema/Cf2 – i.e. rhema before the old Cf1. PersPro Sie refers to Eine Kolonie Kiebitze in Cf1. In other words, it does not interrupt the thematic chain constituted by Cf1. This oscillating thema continuation yields the two different Pro-uptakes – as well as the different thema continuations not described here. Similarly the analysis for (56). With the thema line Cf1 with diesem Einfluß maintaining as well as natürlich sehr wohl gegeben hat, ArtPro/DemPro maintains its status as new, independent thema. By contrast, PersPro simply continues the

i.

Das Blatt hat eine Seite, (/) die ist ganz schwarz (Gärtner 2001: 98) the leaf has one side this is all black ii. Das Blatt hat eine Seite, (/) sie ist ganz schwarz (Gärtner 2001: 98) the leaf has one side it is all black See our (ii), which is totally equivalent to Gärtner’s (i). Since Gärtner does not discuss PersPro, which is totally equivalent to Gärtner’s (i) and relates fully to Art/DemPro it is tough to see what the deeper distributionally based motivation for a syntactic derivation of (i) would be.

S F O PRO



1st proofs

Discourse binding



thema chain Cf1. In other words, Elco Gelling and his professional history is maintained as subpart of Cf1. See the generalization in (57). (57)

Where Cf is represented by a simple NP the shift between ArtPro/DemPro and PersPro signals the change between narrative background and narrative foreground/figure. In other words, it signals a thema shift, Cf2 exchanging with Cf1. Only where Cf can be represented by more than one NP the representation by ArtPro/DemPro is forced distinctly distributed from PersPro – i.e. the division between Cb and Cp.

English has no option of referential takeup by means of ArtPro/DemPro – recall (8a) above. Thus, PersPro generally receives wider referential scope. Therefore, cross-sentential discourse status of PersPro in English is different, i.e. less restricted than in German and Dutch. ArtPro/DemPro, this/that (one), is not grammatical in text-coherent functions as in German and Dutch. Comrie’s (2000) claim is that ArtPro/DemPro in Dutch refers to ‘nontopical’ antecedents, in contrast to PersPro and possessive pronouns which take up ‘topicals’. Despite a different terminology, his analysis is at the core of the concept of Centering Theory: The demonstrative is used if two different antecedents occur for a single pronoun; DemPro is the marked option as opposed to PersPro; it represents the less-expected coreference (Comrie 2000: 54). Furthermore, Comrie concludes that languages differ on which antecedent is less expected and is represented by DemPro vertreten werde. Not in line with Centering Theory, however, Comrie defines the typological parameters in accordance with options of exclusion (Comrie 2000: 54). See the criteria in (58a–c). (58)

ArtPro/DemPro is excluded as binder for that occurrence of two antecedents which appears as a. subject (Criterion of Clause Function); b. the linearly leftmost NP (Linearity Criterion); c. the highest topic NP (Discourse Function Criterion).

References Abraham, Werner. 1993. Überlegungen zur satzgrammatischen Begründung der Diskursfunktionen Thema and Rhema. Folia Linguistica XXVI(1–2): 197–231. Abraham, Werner. 1995. Morphological case: No need for functional projections in German. In Language and Cognition 5, R. Jonkers, E. Kaan & A. Wiegel (eds), 1–12. Groningen. Abraham, Werner. 1997. The base structure of the German clause under discourse-functional weight: Contentful functional categories vs. derivative ones. In German: Syntactic problems – problematic syntax, Werner Abraham & Elly van Gelderen (eds), 11–43. T übingen: Niemeyer.

S F O PRO



1st proofs



Werner Abraham Abraham, Werner. 2002. Pronomina im Diskurs: Deutsche Personal- and Demonstrativpronomina unter ‘Zentrierungsperspektive’. Grammatische Überlegungen zu einer Teiltheorie der Textkohärenz. Sprachwissenschaft 27(4): 447–491. Abraham, Werner. 2006. Topic, focus, and default vs. contrastive accent: Drawing typological differences with respect to discourse prominence. In On Information Structure, Meaning, and Form. The web of information structure across languages [Linguistik Aktuell/Linguistics Today 100], Kerstin Schwabe & Susanne Winkler (eds), 183–203. Amsterdam: John Benjamins. Abraham, Werner. This volume. The discourse-functional crystallization of the historically original demonstrative. Abraham, Werner & Molnárfi, Lázsló. 2001. German clause structure under discoursefunctional weight: Focus and antifocus. In On Formal German(ic) Typology. [Linguistik Aktuell/Linguistics Today 45], Werner Abraham & C. Jan-Wouter Zwart (eds), 1–43. Amsterdam: John Benjamins. Askedal, John Ole. 1995. Lexikalisierung and Grammatikalisierung im Bereich der Kontext- and Sprecherbezogenheit. Kontrastive Überlegungen zur Deixis im Deutschen, Englischen and Norwegischen. In Deutsch als Fremdsprache. An den Quellen eines Faches. Festschrift for Gerhard Helbig zum 65. Geburtstag, H. Popp (ed.), 575–596. München: Iudicium. Benthem, Johan van & ter Meulen, Alice (eds). 1984. Generalized Quantifiers in Natural Language. Dordrecht: Foris. Chomsky, Noam. 1981. Government and Binding. Dordrecht: Foris. Cinque, Guglielmo. 1993. A null theory of phrase and compound stress. Linguistic Inquiry 24(2): 239–298. Comrie, Bernard. 2000. Pragmatic binding: Demonstratives as anaphors in Dutch. In Proceedings of the Twenty-Third Annual Meeting of the Berkeley Linguistics Society, Matthew L. Juge & Jeri L. Moxley (eds), 50–61. Berkeley CA: Berkeley Linguistics Society. Delisle, Helga. 1993. Anaphora in German discourse. In Principles and Predictions. The analysis of natural language. Papers in Honor of Gerald Sanders, Mushira Eid & Gregory Iverson (eds), 243–268. Amsterdam: John Benjamins. Di Eugenio, Barbara. 1998. Centering in Italian. In M.A. Walker, A. Joshi & E.F. Prince (eds.), 115–138. Dixon, R.M.W. 2003. Demonstratives: A cross-linguistic typology. Studies in Language 27(1): 61–112. Gärtner, Hans-Martin. 2001. Are there V2 relative clauses in German? Journal of Comparative Germanic Linguistics 3: 97–141. Gelderen, Elly van. 1998. Demonstratives as Pronouns in Dutch. Paper GLAC 4, April 1998 (ms. Arizona State University). Givón, Tom. 1983. Topic continuity in discourse: An introduction. In Topic continuity in Discourse: A quantitative cross-language study, Tom Givón (ed.), 1–41. Amsterdam: John Benjamins. Grosz, Barbara J., Joshi, A.K. & Weinstein, S. 1986. Towards a computational theory of discourse. Ms. Harvard University. Grosz, Barbara J., Joshi, A.K. & Weinstein, S. (eds). 1995. Centering: A framework for modeling the local coherence of discourse. Computational Linguistics 21: 203–225. Gundel, J., Heidelberg, N. & Zacharski, R. 1993. Cognitive status and the form of referring expressions in discourse. Language 69: 274–307. Heim, Irene. 1988. The Semantics of Definite and Indefinite Noun Phrases. New York NY: Garland.

S F O PRO



1st proofs

Discourse binding



Hoffmann, Beryl. 1995. The Computational Analysis of the Syntax and Discourse Use of ‘Free’ Word Order in Turkish. PhD Dissertation, University of Pennsylvania. Hoffmann, Beryl. 1998. Word order, information structure, and centering in Turkish. In M.A. Walker, A. Joshi & E.F. Prince (eds), 251–272 Hudson-D’Zmura, Susan. 1989. The Structure of Discourse and Anaphor Resolution: The discourse center and the roles of nouns and pronouns. PhD Dissertation, University of Rochester. Hudson-D’Zmura, Susan. 1998. Assigning antecedents to ambiguous pronouns. The role of the center of attention as the default assignment. In M.A. Walker, A. Joshi & E.F. Prince (eds), 199–226. Kameyama, Megumi. 1985. Zero Anaphora: The case of Japanese. PhD Disseration, Stanford University. Kameyama, Megumi. 1998. Intrasentential centering: A case study. In M.A. Walker, A. Joshi & E.F. Prince (eds), 89–114. Kleiber, Georges. 2001. Regards sur l’anaphor et la Giveness Hierarchy. In Langage et référence. Mélanges offerts à Kerstin Jonasson à l’occasion de ses soixante ans, H. Kroning et al. (eds), 311–322. Uppsala: Acta Universitatis Upsaliensis. Leiss, Elisabeth. 2000. Artikel und Aspekt. Die grammatischen Muster von Definiheit [Studia Linguistica Germanica 55]. Berlin: Mouton de Gruyter. Lewis, David. 1986. Scorekeeping in a language game. Philosophical Papers, 233–249. Oxford: OUP. Molnárfi, László. 2002. Focus and anti-focus in Afrikaans and West Germanic. Linguistics 40(6): 1107–1160. Poesio, Massimo & Modjeska, Natalia N. 2005. Focus, activation, and this-pronoun phrases: An empirical study. In Anaphora Processing. Linguistic, cognitive, and computational modelling [Current Issues in Linguitic Theory 263], Antonio Branco, Tony McEnery & Ruslan Mitkov (eds), 429–442. Amsterdam: John Benjamins. Prince, Ellen F. 1994. Subject pro-drop in Yiddish. Paper at the Conference on Focus, Schloß Wolfsbrunnen, Germany, June 12–15, 1994. Rambow, Owen. 1993. Pragmatic aspects of scrambling and topicalization in German. Paper read Workshop on Centering Theory in naturally-occurring discourse. Institute for Research in Cognitive Science. University of Pennsylvania. Rullmann, Hotze. 2001. Dutch demonstrative pronouns in discourse. Paper read at GLAC 7, Banff-University of Calgary. Walker, Marilyn, Iida, Masayo & Cote, Sharon. 1994. Japanese discourse and the process of centering. Computational Linguistics 20(2): 193–223. Walker, Marilyn, Joshi, A. & Prince, E.F. (eds). 1998. Centering Theory in Discourse. Oxford: Clarendon Press. Westerståhl, Dag. 2005. Determiners and context sets. In Generalized Quantifiers in Natural Language, 127–151. Dordrecht: Foris.

S F O PRO



1st proofs

S F O PRO



1st proofs

Gender, number, and indefinite articles: About the ‘typological inconsistency’ of Italian Elisabeth Stark Freie Universität Berlin, Germany

This paper discusses some typologically significant correlations in nominal determination systems found in the family of Romance languages, specifically French, Italian, and Spanish. It proposes to reinterpret the complex system of indefinite nominal determination in French and Italian, which both feature an indefinite article and a partitive article, as devices of nominal classification in a broad sense, marking the conceptually important distinction between a single, delimited referent and a non-delimited substance. It is argued that this classification system arose when nominal declension in Latin, which differentiated these two referentially highly relevant cognitive concepts via overt gender and number affixes, got partially or completely lost. In contrast to modern central Romance languages, like French, which require rather obligatory (indefinite) determination in almost every argument position and have developed indefinite articles coding countability on the level of noun phrase, modern peripheral Romance languages like Spanish allow bare arguments to a larger extent and do not possess an explicit marker of non-countability. How to position Italian in this broad typology inside the family of Romance languages will be discussed in some detail and diachronically explained by its complex evolution of its nominal paradigms.1

.

The problem: The system of indefinite nominal determiners in modern Standard Italian and other Romance languages

From the perspective of ‘correlative typology’, modern Standard Italian is to be considered quite reluctant to any attempt to classify it. Körner’s attempt (1987a) to identify two morphosyntactic types inside the Romance languages, a more “ergative” type A with pro-drop properties, differential object marking, traces of object conjugation etc. (Portuguese, Rumanian, Sardinian, Spanish), and a more

S F O PRO

. This article is partly based on two papers already published/submitted and on my habilitation thesis (cf. Stark 2005, 2006 and forthcoming b).



1st proofs



Elisabeth Stark

accusative type B with the opposite properties (Catalan, French, Occitan), resulted merely in affirming the typological inconsistency2 of Italian: In its modern standard variety at least, it shows type A-like pro-drop characteristics, impersonal reflexives etc., but also the type B-like absence of ‘differential object marking’ and the existence of a ‘partitive article’. Especially when one tries to retrace and explain the emergence and present-day functional load of its indefinite determiner system, Italian seems to be half-way in between the strongly grammaticalized system of French indefinite determiners and the – in many positions – facultative indefinite article of Spanish. Thus, despite some well-known generalizations about the Romance system of nominal determination (e.g. Chierchia 1998; Longobardi 2001), the data in (1) demonstrate that there is considerable variation in the field of nominal indefiniteness: (1)

a.

b.

c.

d.

e.

f.

It.: Hai visto *(un) aquila? Fr.: As-tu vu *(un) aigle? Sp.: Has visto *(un) águila? “Did you see an eagle?” It.: Compro (del) pane. Fr.: J‘achète *(du) pain. Sp.: Compro pan. “I buy (some) bread” It.: Mi occorre (dell’)acqua. Fr.: Il me faut *(de l’)eau. Sp.: Me falta agua. “I need (some) water” It.: Dimostrò (*della) pazienza in questa situazione. Fr.: Elle montra *(de la) patience dans cette situation. Sp.: Demostró paciencia en esta situación. “She showed patience in this situation” It.: Vedo (degli) studenti nell‘edificio. Fr.: Je vois *(des) étudiants dans le bâtiment. Sp.: Veo (a unos) estudiantes en el edificio. “I see (some) students in the building” It.: Escono? (degli) studenti dall‘edificio. Fr.: Il sorte *(des) étudiants du bâtiment. Sp.: Salen estudiantes del edificio. “(Some) students leave the building”

In Italian, French and Spanish, indefinite nominals with an existential reading show quite heterogeneous characteristics in argument position. Table 1 presents

.

Cf. the German title of Körner 1987b.

S F O PRO



1st proofs

Gender, number, and indefinite articles – ‘typological inconsistency’ of Italian



an overview over the three most frequent and grammaticalized indefinite determiners used with indefinite nominals in argument position:3 Table 1. Distribution of indefinite nominal determiners in three Romance languages Italian

French

Spanish

‘zero’: abstract/ “mass-denoting” (non-specific): singular plural noun phrases (non-specific, mostly postverbally).

‘zero’: rarely with abstract nouns (only in more or less idiomatic expressions) no bare plural

‘zero’: abstract/“mass-denoting”, rarely: “entity-denoting” nouns (non-specific): singular plural noun phrases (mostly postverbally)

uno: singular countable noun phrases



del: “mass-denoting” in non-countable singular noun phrases plural noun phrases (specific)

du: abstract/“mass-denoting” in non-countable singular noun phrases plural noun phrases in argument position outside the scope of negation

no ‘partitive article’

Modern Standard Italian, on the one hand, shares some important characteristics with Spanish: Bare noun phrases in argument position occur in both languages under restricted grammatical conditions: in fact, only abstract nouns can appear freely in bare noun phrases in argument position even in the singular (cf. 1d); bare plurals surface post-verbally in subject and object position independently of the lexical category of the noun (with non-specific interpretation of the nominal, cf. 1e and 1f; for the interaction of the ‘prepositional accusative’ and specificity in Spanish cf. Leonetti 2003 and von Heusinger/Kaiser 2004; for non-specific interpretation of definite noun phrases in French and Italian see Kupisch/Koops, this volume). Bare singulars are possible with “mass-denoting nouns” in post-verbal subjects and objects (see 1b and 1c), again with non-specific interpretation of the nominal. Conversely, the only Romance language which almost never permits bare noun phrases in argument position is French. On the other hand, modern Standard Italian shares one peculiar element inside the paradigm of indefinite determiners with French, i.e. the so-called ‘partitive

S F O PRO

. Including ‘zero’ as a possible null determiner for the sake of a similar underlying syntactic structure.



1st proofs



Elisabeth Stark

(article)’ (cf. Renzi 31991). As for overt indefinite nominal determination, every Romance language possesses an indefinite article derived from the Latin numeral unus, ‘one’, which accompanies singular count noun phrases. The second indefinite determiner in Italian and French derived from the composition of Latin de and the definite article. It marks indefinite non-countable singular noun phrases, usually with “mass-denoting nouns”, in pre- and postverbal subjects and objects. So as for the paradigm of indefinite determiners, Italian and French are different from Spanish (and the majority of the other Romance languages) in that they both possess a ‘partitive article’. Yet, Italian seems to have grammaticalized this item to a lesser extent than French, given the fact that it is obligatory with abstract nouns in French, but only optional in Italian (see examples in 1b, 1c and 1d); the same holds for the morphological ‘plural partitive’, which functionally and distributionally represents the indefinite plural article4 and which is fully grammaticalized in French and optional in Italian. Table 2 gives an overview over the particular position of Italian between French and Spanish: Table 2. The Italian system of indefinite nominal determination in comparison to Spanish and French Italian

Spanish

French

1. 2. 3. 4. 5.

+ + + + −

− − − − + (obligatory)

−

+ (obligatory)

bare singulars bare singulars with “mass-denoting nouns” bare singulars with abstract nouns bare plurals (postverbally) optional ‘partitive article’ for non-countable NPs (“mass-denoting nouns”) 6. optional ‘plural partitive article’

. Correlations In order to explain the striking differences between French, Italian and Spanish concerning the possibility of permitting bare plurals or bare (‘mass’) singulars in argument position, the following correlation has often been observed (cf. e.g. Schroten 2001): the almost complete loss of overt morphological number marking in French nouns seems to correlate with the necessity of number marking via determiners for arguments. In Spanish, the number feature is considered interpretable

S F O PRO

. Just like uno, it combines preferrably with “entity-denoting nouns” and does not indicate non-countability, but rather specificity.



1st proofs




in certain contexts via consistent plural marking through s-affixation, thus allowing for bare nouns with indefinite interpretation. However, even if this correlation explains the degree of how obligatory explicit nominal determination in argument position is in French, it is not precise enough to explain the considerably different behaviour of Italian in this respect, as I have already argued elsewhere (cf. Stark 2005 and forthcoming a). Italian exhibits morphological number marking also in its spoken varieties, but has a quite restricted distribution of bare NPs (recall (1a–f)), and, like French but unlike Spanish, it possesses a ‘partitive article’. This is an important hint at the fact that there might be more to an adequate explanation than just the problem of overt morphological number marking in Romance. The first correlation mentioned above and frequently observed in the traditional literature (cf. Tekavčić 1972) relates nominal morphology and (indefinite) nominal determination, as has traditionally been done as well, in a diachronic perspective, in order to relate the loss of Latin case marking on nominals to the rise of (normally the definite) article in Romance (‘compensation hypothesis’).5 Now, this diachronic correlation is highly improbable, since the marking of syntactic functions is achieved via prepositional marking and word order in Romance, and it is empirically not corroborated (cf. the findings in Selig 1992 – on nominal determiners in Late Latin – occurring with equal probability in all syntactic functions, though preferably post-verbally). However, the morphosyntactic categories of number and, maybe even more so, of gender seem to play a major role not only synchronically, but also for the specific development of the different systems of Romance indefinite determiners. In fact, the second, but not the first phenomenon, i.e. the existence of a ‘partitive article’, seems to correlate with unambiguous plural marking via agglutinative affixes, whereas the tolerance for bare arguments correlates with the degree to which the Latin neuter (plural) is morphologically preserved in the different Romance languages (cf. Stark forthcoming a). The remainder of this article is structured as follows: Section 3 will give an overlook over the Latin and modern Romance (Italian, French, Spanish) nominal morphology with respect to gender and number marking. Section 4 will add data from Old Italian (Old Tuscan6 and partly Old French), retracing in particular the development of the ‘partitive article’. Section 5 will try to establish a functional explanation of the observed synchronic correlations and the diachronic development by focussing the central role of Latin gender- and number marking and

.

See Selig 1992, 23f., for a critical overview of the main accounts of the subject.

. Given the fact that modern Standard Italian is mainly based on the variety of Florence of the 14th century with its outstanding classics, the so-called “Tre Corone” Dante, Boccaccio and Petrarca.

S F O PRO



1st proofs



Elisabeth Stark

its loss for the grammaticalization of the indefinite and the ‘partitive article’ in some Romance languages and varieties, especially in Italian. Central to this functional explanation will be the notion of ‘countability’. In fact, the so-called ‘partitive article’ in French has repeatedly been categorized as a kind of ‘nominal classifier’, indicating non-countability as opposed to the indefinite article (cf. Herslund 1998). In order to account for the specific case of Italian and following Löbel (1993: 192ff.) and contrary to the conceptual system proposed by Herslund, I assume a fundamental difference between the lexical noun categories “mass-denoting”, “entity-denoting” and “abstract noun” (N), which derive from characteristics of the potential (extra-linguistic) referents (mainly additivity, divisibility, cf. Krifka 1991; Behrens 1995) and which are based on denotational properties of the head noun, and the countability or non-countability of entire noun phrases. This last opposition is a grammatical category or better a semantic feature interacting with the internal syntactic structure of the noun phrase and it is characterized by the (in)compatibility with certain indefinite determiners (French/Italian: uno vs. del). This assumption is justified by the fact that virtually any noun in Romance languages (like in any language with a grammaticalized countability distinction in this sense) can in principle appear in any kind of noun phrase. See (2a–b): (2)

a.

b.

It.: Hai mangiato (dell’) aquila? Fr.: As-tu mangé de l’aigle? Sp.: Has comido águila? “Did you eat (some) eagle (meat)?” It.: Compro un pane. Fr.: J‘achète un pain. Sp.: Compro un pan. “I buy one (a certain amount/piece of) bread”

Even if these examples seem semantically marked,7 due to prototypical affinities between “mass-denoting nouns” (like engl. bread) with non-countability, and between “entity-denoting nouns” (like engl. eagle) with countability (as discussed for English as early as in Allan 1980), they are grammatically well-formed and their ‘mass’ or ‘count’ readings derive from the prenominal indefinite determiners (‘zero’, ‘partitive’ or indefinite article).

. Cf. Behrens 1995: 47–50, Corbett 2000: 86f.; see also the sortal interpretation or “Artenplural” mentioned by Krifka 1991: 414f. for “mass-denoting nouns” in countable plural NPs and the unique meaning of the morphological plural in languages with grammaticalized countability: it is always understood as additive, “diskrete Gesamtheiten von Objekten derselben Art” (Link 1991: 418).

S F O PRO



1st proofs


. .



Latin and Romance nominal morphology Latin

Classical Latin possessed a complex declension system divided into 5 declension classes; all nouns required obligatory and overt marking of the morpho-grammatical categories case, gender and number. There were 5 (6) morphological cases, 3 genders, and 2 numbers, singular and plural. Even though clear-cut correspondences between gender, declension class, and ‘semantic’ or lexical noun class cannot be assumed (in contrast to the situation suggested for Proto-Indo-European; cf. Ralli 2002), there was some systematic ‘lexical motivation’ for nouns sharing the same lexical root, but differing in gender and/or number. See (3a–c): (3)

a. b. c.

caseus, ‘one single (piece of) cheese’, olea, ‘olive’/‘olive tree’ caseum, ‘cheese as a substance’, oleum, ‘oil’ acinus/acinum, ‘berry’, acina, ‘grape’ frumentum, ‘wheat’, frumenta, ‘corn’8 pilus, ‘one single hair’ pila, ‘hair, collective’9

(3a) shows lexical roots with masculine and feminine gender, resulting in “entitydenoting nouns”, whereas the nouns from the same root in (3b) with neuter gender are “mass-denoting nouns”. In addition, (3c) shows the well-known ‘collective’ semantics of the Latin neuter plural ending in -a (cf. Schön 1971; Windisch 1973; Tichy 1993). Although these oppositions do not cover the whole range of possible lexical denotations, the Latin neuter and especially the Latin neuter plural in -a – both unambiguously marked in spoken and written varieties – can be re-interpreted as a partly generalised ‘classification system’ denoting mainly the opposition between ‘single, delimited, individuated object’ (e.g. one piece of cheese, one olive, one berry, one single hair) and ‘non-delimited substance’ (e.g. cheese, oil) or ‘collective’ (grape, hair).10

. Cf. in detail Hofmann (21997: 7–10), Meisterfeld (1998: 56ff.), and for late Latin analogical neuter plurals following the same pattern cf. Morani (2000: 228). .

Klingenschmitt 1992, 90.

S F O PRO

. Compare similar observations concerning the loss of multiple gender in Germanic in Leiss 1994.



1st proofs



Elisabeth Stark

That this important semantic opposition is at least as much related to gender as to number is shown by the fact that, unlike the plural in modern Indo-European languages, including the Romance languages, the Latin plural is neither automatically interpreted as additive (cf. Link 1991) nor restricted to “entity-denoting nouns”. See (4a–b): (4)

a. b.

frigora caloresque, ‘an intense heat and cold’: plural indicating intensification acquae, ‘waters’, cerae, ‘wax tablets’: different appearances of a substance11

Although the Latin plural can have a sortal reading, bare plurals of abstract or “mass denoting-nouns” are not automatically re-categorized as for instance in modern Romance languages (compare Fr. huile, ‘oil’, des huiles, ‘different sorts of oil’). Virtually any Latin noun can be pluralized, and in fact frequent occurrences of plurals of “mass-denoting nouns” or abstract nouns, as in (4) above, are attested.12 This fact, together with the absence of compatibility restrictions for (optional) indefinite determiners with nouns (Lat. quidam or aliquis combine freely with abstract, “mass-denoting” and “entity-denoting nouns”) indicates that Latin had no grammaticalized “countability distinction” at the level of noun phrases (cf. Löbel 1993). If Latin inflectional affixes, which mark declension class, gender and number, indicate oppositions between ‘delimited single object’, ‘substance’ and ‘collective’, they can be considered as classification devices in the following sense: They are part of the universal dimension of nominal apprehension, which is a central universal operation of establishing reference : First of all, so it seems, one has to be able to express that something is a thing [= dimension of APPREHENSION, E.S.]. Only then can it be named: The dimension of NAMING [. . .] Following that, it can be referenced: The dimension of DETERMINATION. (Seiler 1986: 9) APPREHENSION is the universal operational dimension with corresponding subdimensions which explicate the grasping and representation of concepts corresponding to objects or things by means of language. (Seiler 1986: 145)

Consequently, apprehension concerns chiefly the classification of the denotation of the noun phrase as “an undifferentiated concept or as an individual” (Lehmann 1991: 206, see also Meisterfeld 2000: 328). .

Cf. Kühner/Stegmann (31955: 69, 73), Hofmann (21997: 18, 21).

. Cf. Iturrioz Leza (1986: 295f.): “This individualization strategy [= pluralization of abstract nouns, E.S.] is more widespread in the classical languages (Greek, Latin) than in modern German or any other European language; thus it is often difficult to translate an abstract [plural, E.S.] NP without changing its number: [. . .] Asperitates viarum et angustiae [. . .] ‘The roughness(es) and narrowness(es) of the ways.’”

S F O PRO



1st proofs


.



French, Spanish – and Italian

The main morphological changes in the nominal system from Latin to Romance are well-known and comprise the complete loss of morphological case (with the exception of Rumanian), a reduction of gender (especially the loss of the neuter, but see below), contrasted with a solid formal preservation of number, but with only the ‘additive plural meaning’ left. In addition, Modern Standard French shows the complete loss of the declension classes (already in Old French, cf. Delfitto/Schroten 1991: 180f.). Gender and number are usually marked (in the phonetic code) only by prenominal determiners. See (5): (5)

un ami/une amie – des ami(e)s [œnami/ynami – dezami] ‘a male friend’/‘a female friend’ ‘male or female friends’13

The French noun [ami] is thus not phonetically marked at all for gender or number. Just like Italian (see below, (6a–c)) and unlike French, Modern (European) Standard Spanish has 3 main declension classes, 2 overtly marked genders and overt number marking. Its morphological nominal morphology is “heterogeneous with respect to gender” (Harris 1992: 66ff.), but unambiguous with respect to number marking (“plurality is manifested consistently by the suffix /-s/”; Harris 1992: 67): (6) a. b. c.

Sg.: -o/ -a/ -e/; Pl.: -s: pas-o – pas-os (m.) ‘step’ – ‘steps’ pas-a – pas-as (f.) ‘raisin’ – ‘raisins’ jef-e – jef-es (m.) ‘chief ’ – ‘chiefs’

man-o – man-os (f.), ‘hand’ – ‘hands’; map-a – map-as (m.), ‘map’ – ‘maps’; nub-e – nub-es (f.), ‘cloud’ – ‘clouds’

However, there seems to be no ‘classification potential’ in nominal (declensional) endings in the morphological setup of Spanish nouns.14 The only slight systematic ‘classification potential’ left in Spanish is a kind of ‘neuter’ (deriving from the Latin neuter singular) in the pronominal system marking ‘abstract antecedents’, such as .

Illustration gleaned from Delfitto/Schroten (1991: 177ff.).

. A possible exception might be the -o-a-alternation with the same stem indicating size in el bolso (‘the handbag’) – la bolsa (‘the (bigger) bag’) or el cubo (‘der Eimer’) – la cuba (‘das Weinfaß’), but this is not or not clearly retracable to the functional load of the Latin neuter indicated in Section 3.1.

S F O PRO



1st proofs



Elisabeth Stark

quotations, matters of fact, etc. Spanish personal pronouns and demonstratives show a threefold morphological opposition, with forms ending in -e for masculine singular, -a for feminine singular and -o for the so-called ‘neuter’ (e.g. span.: este/ esta/esto: lo que me interesa es esto. . . ‘what I am interested in is the following:. . .’ vs. él que me interesa es este hombre ‘who I am interested in is this man’). By contrast, Modern Standard Italian is different from French and quite similar to Spanish in having preserved 3 main declension classes, 2 overtly marked genders as well as overt number marking. However, and this is a crucial point, the declensional endings -a and -e are far from being unambiguous markers of singular or plural, as they can either indicate feminine singular, (rarely) masculine singular or feminine plural (-a) or masculine singular or feminine plural (-e). The morpheme -o unambiguously indicates singular, but both masculine and (rarely) feminine gender. See (7a–c): (7) a. b. c.

Sg.: -o / -a / -e ; Pl.: -i / -a / -e : libr-o – libr-i (m.) ‘book’ – ‘books’ cas-a – cas-e (f.) ‘house’ – ‘houses’ can-e – can-i (m.) ‘dog’ – ‘dogs’

mano – mani (f.) bracci-o – bracci-a (m. – f.), ‘hand’ – ‘hands’ ‘arm’ – ‘arms’; poet-a – poet-i (m.), ‘poet’ – ‘poets’;

What is marked in bold characters in (7a) is a residue of the original Latin classification potential of the neuter plural in -a, as opposed to a regular plural form in -i (originating in Late Latin, cf. Hofmann (21997): 21) and reanalysed as feminine (but still plural!), always indicating a collective or at least ‘pair’ reading. Some nouns ending in -o (masculine singular), usually denoting concrete objects like body parts (It.: ginocchio ‘knee’, orecchio ‘ear’ and so on, also muro ‘wall’ etc.), have a plural form in -a when denoting a plurality, body parts or a ‘collective reading’. However, they form a plural in -i when used metaphorically to denote something similar in form, but without a collective denotation (e.g. It.: le braccia denotes both arms of an animate being, whereas i bracci denotes the arms of a river, It. le mura denotes the townwall, whereas i muri denotes the single walls of a building). Having a closer look at some Italian dialects as well as at older stages of Italoromance, especially in the central-southern area (parts of Lazio, in Campania, especially in Umbria and in the Marche), the Latin neuter plural in -a and even forms in -ora (e.g. castella, ‘castels’, locora ‘places’) are more lively and widespread than in Standard Italian or Northern Italian dialects, a fact which can easily be retraced from the medieval texts on (remember one of the oldest Italian documents, the wall-writings in the catacombs of Comodilla near Rome: non dicere illa secrita a bboce, ‘do not say these secrets aloud’, from the 9th century against

S F O PRO



1st proofs




Northern dialect forms such as castele instead of (Old) Tuscan castela, ‘castels’ etc.). Besides the pure formal continuation of the Latin neuter plural in -a, there is also still a ‘neuter singular’, for definite determiners, personal pronouns, and nouns, especially in Umbria (in addition partly in Asturias, cf. Hall 1968; Haase 2000), which is to indicate non-countability (cf. Haase 2000: 225). Thus, while lu pane in Central Italian dialects denotes an individuated, delimited piece or loaf of bread, lo pane in the same dialects denotes the substance of bread (cf. Haase 2000: 222). Formally, the second form of the article derives from the Latin neuter, but functionally, the systematic indication of non-countability via prenominal definite determiners15 is an innovation (cf. Haase 2000: 228). Important for our focus on indefinite determiners as potential classifiers is the following point: In Romance areas and dialects with a special form of the definite determiner for non-countability, there is no trace of any ‘partitive article’.

.

A short history of the ‘partitive article’

Typically, the main texts of the Old Tuscan period (traditionally set from the origins to 1375, i.e. Boccaccio’s death) show a considerably high tendency to preserve the old Latin neuter plural forms in -a, to a much larger extent than modern Standard Italian. Thus, we find anella much more often than anelli (‘rings’) in Novellino, Convivio, Decameron etc., castella or mulina (‘castels’, ‘mills’) instead of the rare forms castelli or mulini, today totally unmarked, in the Decameron etc., tempora (today: tempi) in the Decameron, luogora for modern luoghi (‘places’) in the Novellino, and so on and so forth (cf. for details Rohlfs 1968: 39f.). All these forms have a clearcut collective meaning, deriving more or less directly from the old Indo-European formations in -a (cf. already Brugmann 1897, Jespersen 1924) and the core of the Latin neuter plural forms (cf. for example Hofmann 21997, 9ff.). Additionally, some feminine nouns without this collective meaning (donna, ‘woman’ or ‘women’, capra, ‘goose’ or ‘geese’ etc.) had (and still have) a plural form in -a in (Old) Tuscan, being thus absolutely homophonous with their singular forms. The same holds for (phonetically regular) plural forms like i cane (‘the dogs’, deriving from lat. canes), identical to the masculine singular lo cane, or le chiave (‘the keys’, deriving from lat. claves), identical to the feminine singular la chiave, and rather widespread in Old Tuscan texts – just like the majority of French

S F O PRO

. With the exception of etymologically and formally feminine nouns like Umbrian la pajja, ‘straw’, cf. Haase 2000: 225.



1st proofs

 Elisabeth Stark

nouns from the 13th century on (cf. Rohlfs 1968: 28-f.). In sum, we find the same endings on nouns as in modern standard Italian (see above, Section 3.2), i.e. -o (masculine singular), -a (singular or plural, ‘collective’ or ‘individual’, feminine), -i (masculine and/or feminine plural), and -e (singular or plural, both feminine or masculine), but with an even increased heterogeneity and homonomy both according to number and gender marking. Comparing this “hereogeneous inflectional marking system” to Old French (or Old Occitan), we can still observe the above mentioned discrepancy in inflectional gender and number marking already realised in the Middle Ages. Old French could not distinguish, for example, between nominative singular (casus rectus) and accusative plural forms (casus obliquus) for more than 50 percent of its nouns (cf. Buridant 2000: 73), unless it used (definite) determiners, especially for masculine nouns (li murs – lo/le murs, ‘the wallNOM’ – ‘the wallsACC’, la flors – les flors, ‘the flowerNOM’ – ‘the flowersACC’). This contrasts sharply with the fact that contiguous Western Romance languages like Spanish had consistently reanalysed and preserved the originally case-marking s-ending as an unambiguous plural marker. We will not discuss here the problem of case-marking in Old French (cf. Schøsler 2001), but rather concentrate on gender and number marking according to the above mentioned correlations with the indefinite nominal determiners and in order to retrace their history. What is interesting in this context is the coincidence, in the history of French, of the loss of the phonetic realization of final s-endings and the rise of the ‘partitive article’ – both occur in the 12th and 13th century, the partitive (singular) with postverbal objects (cf. Buridant 2000: 119). The first attestations of the partitive as an indefinite determiner in Old French occur systematically with “mass-denoting nouns” in postverbal object position of verbs like avoir, ‘to have’, boire, ‘to drink’, donner, ‘to give’ mangier, ‘to eat’ etc. (cf. Englebert 1989) in non-generic sentences. In the same period, the indefinite article un is only found with “entity-denoting nouns” (cf. Carlier/Goyens 1998). Plural occurrences of the partitive are found from the 13th century on. The widening of the contexts in which the partitive article occurs starts in Middle French (from the 15th century on), showing the first occurrences in subject (usually with plural noun phrases), later also in predicative position.16 All these findings strongly corroborate Leiss’ (2000) hypothesis concerning the original motivation of article grammaticalization: Articles start out as indicators of the gestalt of the intended referent denoted

. For this and the following, please remember that the morphological plural of the partitive article is not functionally a partitive, but rather a normal indefinite plural article, cf. Section 1.

S F O PRO



1st proofs




by the respective nominal inside the VP, because the nature of the complements of V is crucial for the conceptualization of the whole event (durative, perfective, imperfective). If the complement NPs or DPs are not marked via gender and/or number as denoting ‘substance’ or an ‘individual’, ‘shaped entity’, the event they are involved in is less clearly perceivable either as perfective or imperfective (compare French nous avons mangé un fromage, ‘we ate a/one cheese’ with nous avons mangé du fromage ‘we ate cheese’). In order to compare this situation with Old Tuscan and to relate it to the restructuring of the Italian morphology in comparison to Latin, it is interesting to investigate the first attestations, the distribution and further development of the partitive article in two of the three above mentioned texts of the Old Tuscan/ Old Italian period, and additionally in a later, epigonal collection of novellas: The anonymous Novellino, a collection of 100 novellas from the end of the 13th century; Giovanni Boccaccios Decameron, written in its major parts from 1348 on; and Masuccio Salernitanos Novellino, written between 1440–1475/76. These three collections of novellas, though forming only a small corpus of about 430.000 words, guarantee a wide comparability as to text type, content, discourse traditional factors, and dialectal homogeneity (Tuscan, with slight Northern and Southern influences) and are at the same time well-known representatives of three subsequent centuries.17 The partitive article is distributed as follows in the three texts: Table 3. Occurrences of the partitive article in the three corpus texts

Novellino Decameron Masuccio

Singular

Plural

2 13 2

– 16 4

While the partitive is only attested twice in the Novellino, and only in the singular as a real non-countability indicator, it is Boccaccio who establishes its use, in the singular and plural form, and Masuccio seems to refrain again from this non-countability marker. Masuccio is from Campania, one of the dialectal zones having preserved and re-functionalized the Latin neuter (see above, Section 3.2) – and even if he tries to copy Boccaccios Tuscan style, his Southern origin might have inhibited the frequent use of an unfamiliar grammatical element.

. Narrative texts are chosen because the Romance ‘partitive article’ shows a considerable statistical affinity to this text type, cf. already Pestelli-Gori 1944/45 and Stark 2006, Chapter 7.5.3.

S F O PRO



1st proofs



Elisabeth Stark

Table 4 shows the distribution of uno, the partitive singular and plural and ‘zero’ according to the semantic class of the noun: Table 4. Uno, the partitive and ‘zero’ with “mass-denoting nouns”, abstract nouns and inanimate referents18 uno Singular Mass Abstract Inanimate Σ Plural mass abstract inanimate Σ

11 190 369 579

partitive 1,90% 32,82% 63,73%

‘zero’

13 2 17 17

76,47% 11,76% 100,00%

31 79 116 131

23,66% 60,31% 88,55%

– 6 19 20

– 30,00% 95,00%

– 49 109 142

– 34,51% 76,76%

These data indicate quite clearly that the partitive singular is specialized on “massdenoting nouns” (76,47% of its occurrences in the three texts of the corpus), while ‘zero’ typically marks abstract nouns (60,31%), and uno much more likely abstract (32,82%) than “mass-denoting nouns” (1,90%) in the singular. Uno has also the lowest percentage for inanimate referents, something which hints at its status as a specificity marker (see below) in these periods. Inanimateness is the main domain of the partitive singular again – non-countability would in fact be difficult to conceive for animate beings. The table shows furthermore the parallel distribution of uno and the partitive plural; especially the numbers for abstract nouns are almost identical (32,82% for uno, 30,00% for the partitive plural). The partitive plural shows a higher compatibility with inaminate referents, but is also conceivable with animates and even humans. “Mass-denoting nouns” are thus usually marked either by ‘zero’ or by the partitive article in all three texts, whereas abstract nouns do not show any significant preference for a particular indefinite determiner. One important exception here is a kind of ‘conversion’ in Masusccio with the partitive: (8)

Unde, a nui tornando, dico che non multo [lontano] da la cità, de la quale quanto

. The three semantic classes enumerated one below the other in the leftmost column are of course not orthogonal. Both the referents of “mass-denoting nouns” and abstract nouns are ususally inanimate, so that the numbers cannot be straightforwardly summed up to 100%.

S F O PRO



1st proofs




sia piacevole il luoco, il nome in parte lo demostra, [era una villetta], ne la quale non è gran tempo che essendo un preite, donno Battimo nominato, il quale, ancora che de villa fosse, pur del prattico e de l’intendente avea. . . “Where, while night is falling, I say that not very far from the city whose location was also very nice and whose name partly shows this, there was a little village, in which, not long ago, lived a priest, named donno Battimo, who, despite being of rural origin, had something practical and intelligent. . .” (Masuccio, V)

Masuccio effectuates in this example a nominalization of two abstract adjectives, prattico and intendente, via nominal determination through the partitive article. The effect is a ‘substance’ or ‘property reading’, whereas uno would have created an ‘individual reading’ (‘a pratical person’ or something similar). This can be considered a rather clear indication of the available function of the partitive as an indicator of non-countability of the respective nominal. Table 5 and 6 illustrate the interaction of the two fundamental denotational categories for article grammaticalization, i.e. countablity and specificity (more or less informally defined hereafter as reference to ‘a particular x’, usually highly ‘foregrounded referents’ in the text (cf. Stark 2002)). At the very beginning of its grammaticalization towards an indefinite article and countability indicator, the numeral unus/uno occurs only with specific referents in definite NPs in real partitive constructions (‘one of the x’). While becoming a countability marker on its own via implicature (if an element can be introduced as one out of a set, it is countable), uno can leave these highly specific contexts and appear also in nonspecific ones – and this is in turn an indicator of its high degree of grammaticalization, because in non-specific contexts, the assertion of a set containing only one element is usually irrelevant (consider If I meet a/one doctor, I will ask him. . .).19 At this point, the original implicational relation between countability and specificity has been reversed: While explicit nominal indefinite determination is at the beginning of article grammaticalization always limited to contexts of specificity, marking sometimes also (non-)countability, in languages with fully grammaticalized nominal determination, (non-)countability is, in the indefinite case, always marked, also in non-specific contexts. In order to identify then the grammaticalization stage our Old and Middle Tuscan texts have to be located in, with special respect to the partitive article and the ‘classification system’ in the singular via indefinite nominal determination, we

.

Cf. Carlier/Goyens 1998, 106, for Old French.

S F O PRO



1st proofs



Elisabeth Stark

now have a look at Table 5 and 6, which permits to explain particularly well the above mentioned unexpected occurrences of uno with “mass-denoting nouns” and abstract nouns: Table 5. Uno, partitive and ‘zero’ in singular nominals with central and non-central “mass-denoting nouns”, abstract nouns and inanimate referents Singular

uno

mass central abstract central inanimate central Σ

11 2 190 22 369 36 579

partitive 1,90% 18,18% 32,82% 11,58% 63,73% 9,76%

13 – 2 – 17 – 17

‘zero’ 76,47% – 11,76% – 100,00% –

31 – 79 – 116 – 131

23,66% – 60,31% – 88,55% –

Table 6. Uno, partitive singular and plural, ‘zero’ with singular and plural NPs with perfectivly marked verbs, in pre- and postverbal position, and in postverbal direct objects of prefectivly marked verbs20 uno perf. 246 prev.+perf. 75 postv.+perf. 171 po.+perf+dO 43 Σ 579

42,49% 12,95% 29,58% 17,55%

partsg.

partpl.

‘zero’sg

‘zero’pl

3 – 3 3 17

1 – 1 – 20

87 36 51 6 300

77 33 44 11 300

17,65% – 17,65% 100,0%

5,00% – 5,00% –

29,00% 12,00% 17,00% 6,90%

25,67% 11,00% 14,67% 14,29%

Table 5 shows clearly that part of the unexpected ocurrences of uno with “massdenoting nouns” (18,18%) or abstract nouns (11,58%) are due to the thematic centrality (important object in the history or else) of the respective referent, whereas neither the partitive nor ‘zero’ are able to assume this function in our early texts. Here, the opposition between ‘zero’ = non-specific/background and uno = specific/ foreground still seems to hold partly. Consider (9) as an illustration. (9)

Lo re mandò per maestri e fecela spezzare, e trovaro nella detta pietra un vermine. Allora lodò il greco d’oltremirabile senno, et istabiliò che un pane intero li fosse dato per giorno, alla spese di sua corte. “The king called for some experts and made them destroy the stone, and inside

. Again, these categories are not mutually exclusive; in Italian, perfect aspect is morphologically marked in the past tenses through the opposition passato remoto, perfective, and imperfetto, imperfective.

S F O PRO



1st proofs




the stone they found a worm. Therefore he praised the Greek man for his outstanding wisdom, and gave order that a whole loaf of bread would be given to him, every day, and he would pay it.” (Novellino, p.127)

Here, we can see the typical ‘recategorization effect’ already in the 13th century. Together with intero ‘whole’ pane has to be determined by uno – and its role in the novella is, additionally, of central importance. Table 6 shows furthermore the exclusive distribution of the partitive singular in postverbal direct objects in perfective contexts – another ‘specificity indicator’ (cf. Stark 2002). These findings both seem to allude to a quite early grammaticalization state of our indefinite determiners, where they would still be highly restricted to specificity contexts. ‘Zero’ would accordingly be the normal marking of the non-specific interpretation of nominals and would be in general possible for plural NPs or DPs due to rather clear inflectional number marking in Old and Middle Tuscan. However, the findings for ‘zero’ do not fully corroborate this diagnostics. A detailed analysis21 shows that, besides an almost identical distribution to Old French for the different syntactic functions of the partitive and especially the first occurrences of the partitive plural in subject rather than object position, ‘zero’ is almost equally distributed over argument and non-argument positions in the Old Tuscan texts with singular and plural nominals. The same holds for ‘zero’ in our tables: it occurs for example with inanimate referents with similar frequency in the singular (88,55%) and in the plural (76,76%, see above, Table 4). It occurs significantly less frequent with abstract nouns in the plural (34,51%) than in the singular (60,31%). Taken together, all of this excludes the possibility that the Old and Middle Tuscan system of indefinite nominal determination would still be in an early grammaticalization stage with typically less determination in the plural than in the singular. Number marking alone is no longer enough for nominals to appear in argument position, because the value of ‘zero’, especially in the singular, is no longer reduced to indicate non-specificity or non-referentiality. As a part of the new central Romance ‘classifcation system’, ‘zero’ in the singular now marks abstract nouns.

.

A functional explanation: Grammaticalizing countability

How can we now functionally relate these synchronic and diachronic morphosyntactic findings to the problem of the different indefinite determiner systems in the

.

Cf. Stark 2006, Chapter 7.5.3.

S F O PRO



1st proofs



Elisabeth Stark

Romance languages? Modern Standard French shows a complete re-analysis, a complete loss of the Latin neuter plural in -a, an evolution widely completed already in Old French: Lat. neuter plural folia becomes Fr. la feuille, feminine singular, just like Lat. feminine singular femina becomes Fr. la femme. While this happened already from Vulgar Latin to Old French, number marking via s-affixation gets reduced and lost until the 12th or 13th century. From then on, French nouns show almost no marking of gender and/or number. This loss of the Latin ‘classification system’ via noun morphology is compensated for by the simultaneous evolution of an obligatory ‘classification system’ via indefinite determiners (cf. Herslund 1998: 70ff.): ‘zero determination’ nowadays is practically excluded in argument position; the indefinite singular article, un, marks ‘contour’, ‘individualized referent’, and thus countability; the ‘partitive article’ du, appearing in the 13th century, marks ‘substance’, ‘diffuse’ (mass/abstract), and thus non-countability. See (10a–b): (10)

a. b.

Lat.: caseus Lat.: caseum

Fr. un fromage ‘one single (piece of) cheese’ Fr. du fromage ‘cheese as a substance’

This situation differs considerably from the situation in (Modern Standard) Spanish. Here, we find a partial preservation of the Latin neuter, unambiguous plural marking via s-affixation, but no ‘classification potential’ inside the indefinite determination system. ‘Zero’ can mean ‘abstract’/‘mass’, even (rarely) “entity-denoting”, besides the additional possibility of marking non-specificity (cf. Laca 1999). The indefinite article, un(o), less grammaticalized than in French or Italian, marks ‘contour’, ‘individualized referent’ and thus countability. However, there is no explicit marking of non-countability and therefore no unambiguous simple obligatory classification system (cf. Herslund 1998: 70–72). Modern Standard Italian shows some residue of the Latin neuter plural in -a with a certain ‘classification potential’ and overt, yet ambiguous, plural marking. However, it also has a French-like ‘classification system’ via indefinite determiners: ‘zero’ is partially permitted, but exclusively only for abstract/plural noun phrases; the indefinite article, uno, marks ‘contour’, ‘individualized referent’, and thus countability, just as in French. The ‘partitive article’, del, less grammaticalized than in French, marks ‘substance’ (‘mass’, as opposed to ‘abstract’) and thus noncountability. The diachronic data presented in Section 4 above corroborate the hypothesis that the establishment of a ‘classification system’ via indefinite determiners in central Romance languages, more precisely the grammaticalization of countability as a feature of whole noun phrases via determiner selection and its overt marking, is to be linked to the loss of the Latin number and gender marking system. Losing the unambiguous plural marking and partly an unambiguous gender marking

S F O PRO



1st proofs




system already in Old Tuscan, Early Italian starts out by grammaticalizing in the same period an element marking ‘substance’, non-countability. In contrast to (Old) French, it still possesses the possibility of phonetically marking singular and plural and is thus able to integrate ‘zero’ in its classification system in the singular. See Table 7: Table 7. Synopsis of the morphological and syntactic evolution of French and Italian indefinite nominals Latin/‘Old Romance’ Italian

French

> 13th century

13th – 15th century

15th century DP) (NP>DP) (NP>DP)

Often, complex coding patterns are not easy to detect. Especially native speakers remain unaware of complex patterning in general. Native speakers process these rules fast, automatically, and unconsciously in the Broca area (language area) of the left hemisphere of the brain. Therefore, in order to detect complex coding patterns it is necessary to process grammemes more slowly than a native speaker usually does. And it demands additional profound metalinguistic knowledge and observation. Therefore, it comes as no surprise that the patterns indicated above are due to Slavists who were non-native speakers of Russian (Birkenmaier 1977, 1979; Gladrow 1972, 1979). There is probably a vast amount of complex patterning in grammar that still awaits to be detected. But once such patterns are discovered the metalinguistic horizon opens wide. Similar patterns in other languages come into sight. The insights of Birkenmaier and Gladrow opened the way to the discovery of similar pattern in older stages of the Germanic languages (Leiss 1987/1990, 1994, 2000; Donhauser 1990, 1991; Abraham 1996, 1997; Philippi 1997) and of the Romance languages (Stark 2006, Fischer 2005). There is a lot of empirically founded evidence that the verb-governed genitive had a core function in the grammatical architecture of Old High German interacting with a still functional system of Old High German verbal aspect. The main assumption is that (in)definiteness effects are achieved in by combining verbal aspect with alternating morphological cases the main pattern being very similar to the pattern operating in Slavic languages. .

Complex encoding of verbal determination (perfectivity)

S F O PRO

Verbs in Finnish are aspectually neutral as to their form. Nevertheless, the reading of a verb is either imperfective or perfective the aspectual reading depending on



1st proofs



Elisabeth Leiss

the morphological case of the object. The partitive case has an imperfectivizing effect on verbs, whereas the genitive case implies perfectivity of the governing verb (Larsson 1983). There has been a lot of discussion whether the partitive case has an additional indefiniteness effect creating indefinite or partitive objects, whereas the genitive case provides definite object readings (Chesterman 1991). The discussion is very revealing as it shows that there are clear affinities between definiteness and perfectivity, on the one side, and between indefiniteness and imperfectivity, on the other side. Verbs and objects (internal arguments) are so close syntactic neighbours that they contaminate each other. Examples drawn from Wexler (1976: 52) demonstrate these effects. See (10)–(11). (10)

Ostan maidon buy:1.. milk: ‘I will buy all the milk’

(11)

Ostan maitoa buy:1.. milk: ‘I am buying (the) milk/some milk

With mass nouns such as milk the (in)definiteness reading is more evident. But looking at the temporal effects on the verb one will notice aspectual effects as well. If the object is marked in the genitive the verb has clear future time reference while being a formal present tense. This secondary temporal reading is characteristic only of perfective verbs. Thus, the definiteness effect of the genitive case has aspectualizing effects upon its verbal neighbourhood. The opposite effect imposes itself on the verb when the object is marked with the partitive case as in (11). There is a clear imperfectivizing effect on the verb which has clear present time reference in the present tense form. The dominance of definiteness/indefiniteness effects with secondary aspectual effects is typical of mass nouns. Once object count nouns are involved the readings seem to shift to a dominance of aspectual effects with secondary definiteness/ indefiniteness effects. See (12)–(13). (12)

Kirjotin kirjeen write: .1.. letter:. ‘I finished writing the letter’ (‘I wrote the entire letter’)

(13)

Kirjotin kirjettä write:.1.. letter:. ‘I was writing a/the letter’

These examples make evident why aspectualized verbs in Slavic languages translate into nominal definiteness/indefiniteness effects in article languages such as the Germanic languages. The following Czech example as well as its translation is taken from Lewandowska (2006: 3).

S F O PRO



1st proofs

Covert (in)definiteness and aspect

(14)

Včera večer dĕlala jsem domáci ukol Yesterday evening make:.. 1.. homework ‘Last night I was doing homework’

(15)

Včera večer udĕlala jsem domáci ukol Yesterday evening :make:.. 1.. homework ‘Yesterday evening I did my homework.’



Obviously nouns do have clear quantificational effects on verbs, and vice versa. Thus, nouns and verbs contaminate each other quantificationally and symbolize either DPs or AspPs depending on the syntactic category the quantifier enters into. The best tool to describe these effects is mereology. .

Mereology as a tool of unification over nominal and verbal categories

Mereology is defined as the theory of part-whole relations. Mereological features provide us with a tool to characterize nouns and verbs in a unified description. In order to understand the effects of mereological features on word classes we have to understand the mereological design of word classes themselves. Word class characteristics are the ‘first coat of grammar’ put on so-called bare lexical items or ‘archilexemes’. The notion of archilexeme has been introduced by Jean Marie Zemb (Paris) and is discussed in more detail in Leiss (1998). Lexical items are characterized by a matrix of lexical-semantic features and nothing else, particular no syntactic information. Each additional feature which does not qualify as a lexical-semantic feature is to be viewed as an additional concept-creating feature. Grammatical features such as word class characteristics, verbal categories, and nominal categories invest bare lexemes (archilexemes) with specific conceptual designs. The idea that bare lexical items are invested with coats of conceptual features is dating back to the Scholastic linguistic school of the ‘Modistae’, with Thomas of Erfurt being the best known amidst them. The Modistae view grammar as a ‘mode’ (means) of conceptualizing bare lexical items (Pinborg 1967, 1982, 1984; Eco/Marmo 1989). The precursor of the Modistae is Aristotle (cf. his ‘Metaphysics’), whose mereological theories were rediscovered by Central Europe during the reconquista of Spain. The modern representative of Modistic philosophy is Charles Sanders Peirce, who worked his whole life on a non-nominalistic theory of human concept formation. Modistic grammarians viewed grammar as a technique to construct perspectives. As far as we know the basic program of grammatical concept formation starts with the differentiation between homogeneity and heterogeneity. One simple bare lexical item can be viewed as being either homogenous or heterogeneous with respect to the concept evoked by it. The concept qualities of the lexical items cannot be derived from real objects in the world. Lexical items do not refer to real objects. Rather, they refer to classes of objects. Grammatical features do not refer to qualities of these classes. They refer to different views of these classes. See also Krifka (1986).

S F O PRO



1st proofs



Elisabeth Leiss

Table 3.  ↓ additivity divisibility

vs. vs.

 ↓ non-additivity non-divisibilty

mass nouns imperfective verbs

vs. vs.

count nouns perfective verbs

vs.

Modistic grammar was interested in the difference between lexical items such as help as a noun and help as a verb. Both items are characterized with the same matrix of semantic features (archilexematic characteristics). The difference between help and to help is a difference in concept formation. It is a difference in the constructional process involved. The noun refers to a concept which is bounded: a concrete shape or event of ‘help’. The boundedness of lexical items obviously does not refer to real-world qualities. The verb refers to non-boundedness of the same single archilexeme. The consequence of this approach to semantics is that the lexical item apple does not refer to a real bounded object its boundedness being nothing more but the effect of a grammatical technique. The archilexeme denotes neither ‘apple’ nor ‘appleness’ nor ‘to apple’ (as a verb). The function of grammar is to view archilexematic material under a certain perspective or a certain aspect in the sense of ‘point of view’. It is a technique of quantification, i.e. some sort of lens on the lexical items which refer to classes. Lexical classes are the result of finitizing an infinite universe of ‘objects’ which human cognition cannot afford to name by proper names. Human cognition is finite. By contrast, the universe is infinite. Language is a means of the human species to bridge and solve this conflict. As classes do not have the quality to refer, reference has to be guaranteed. The minimal unit which is able to refer to the outside world is the linguistic unit of a sentence. Therefore sentences have been viewed as a linguistic technique to create complex proper names. There are several layers of nominal quantification nominal characteristics such as mass vs. count. ● ● ●

number: singular vs. plural; distributive plural vs. collective plural grammatical gender definiteness vs. indefiniteness

Number is sensitive to nominal characteristics such as mass vs. count. Mass nouns do not pluralize. Grammatical gender itself is sensitive to number characteristics. According to Greenbergs universal Nr. 36 transnumeral languages (such as Japanese, Korean and Chinese) are not known to develop the category of grammatical gender.

S F O PRO



1st proofs




The features of homogeneity/heterogeneity represent the best tool to differentiate between mass nouns vs. count nouns (nominal aspect) and between verbal aspect as well (imperfectivity versus perfectivity). Let us first turn to nominal aspect. If you put a bed to pieces, no single one of the pieces is a bed. And if you add a bed to a bed, you have just two beds, not a bed as such. The reason is the heterogeneity of the concept which is evoked by bed. Neither heterogeneity nor its counterpart, homogeneity, is motivated by ontology. A lexeme with the feature [−] cannot be used to refer to parts of the conceptual representation evoked. Of course, ‘homogenous beds’ are conceptually feasible. You can think of languages that even prefer such non-nouny nouns, as for example Korean or Japanese. Nouny count nouns can be turned into mass nouns in non-transnumeral Standard German when inserted in specific syntactic slot such as in (16). (16)

Kaufen Sie mehr Bett/Auto/Laptop für weniger Geld! ‘Buy more bed/more car/more laptop for less money!’

The fact that perspectivizing techniques are not motivated by ontology is illustrated by Table 4 below (for the mereological conceptualization, but a slightly different terminology, cf. also Krifka 1986). Table 4. Mereological features of mass nouns vs. count nouns ● ● ● ●

If you put a sheep into pieces, none of the pieces is a sheep If you put mutton into pieces each single piece is mutton If you add a sheep to a sheep to you have two sheep If you add mutton to mutton you still can refer to it as mutton

... ... ... ...

[−] [+] [−] [+]

In order to detect the compositional effects of verbal aspect plus objects which are case-marked we have to understand that verbal aspect derives its characteristics from the same source as nominal aspect as is sketched in Table 5. Table 5. Mereological features of verbal aspect imperfective verbs . . . [+] > [+] + [+] perfective verbs . . . [−] > [−] + [−]

For this very reason verbal aspect is compared with techniques of nominal quantification the equation being: Mass nouns and imperfective aspect have identical mereological features. So do count nouns and perfective aspect. Maybe this equation is too simplified. There are different layers of nominal quantification (mass/count-characteristics, number, grammatical gender, and definiteness/indefiniteness): Quantification starts with word class characteristics: A

S F O PRO



1st proofs



Elisabeth Leiss

nouny noun is a count noun. A ‘verby’ noun is an infinitive. A typical verb shows just the opposite of the mereological characteristics of a noun: Table 6. Mereological features of nouns vs. verbs Typical verbs have the features . . . [+] > [+] + [+] Typical nouns have the features . . . [−] > [−] + [−]

These features are recursively used in order to create new categories such as number in the domain of nouns or iterativity in the domain of verbs. There are languages that do not have nouny nouns such as unmarked nouns in Korean. And there are languages that do not have typical verbs. There is English with its simple verbs, which are ‘count verbs’, or perfective verbs, which are imperfectivized by the progressive aspect. Similarly, Korean nouns become nounier when construed with numeral classifiers. English verbs behave more like verbs when put into the progressive aspect. The recursive use of mereological features creates nominal and verbal categories of increasing complexity. This makes it difficult to understand the parallels. There are yet too many simplifications in this domain and a lot of confusion about the different layers. From a functional point of view there is good reason to assume that perfective verbs are not just count verbs, but ‘definite verbs’. Perfective aspect creates definiteness effects and, vice versa, definite articles create aspectual affects. The same holds for the imperfective aspect, which is able to create indefiniteness effects. . Overt coding of nominal determination: the polysemy puzzle Having demonstrated that there are invisible DPs and that there are DPs created by the mereological interplay between aspect and objects in terms of paradigmatic case selection we have to ask next whether overt marking of DPs by mapping one form to one meaning is really the normal case. Overt grammatical marking is achieved by means of bound and non-bound grammatical morphemes (visible grammar): Definite, indefinite, and partitive articles in French are overt markings of (in)definiteness. In the light of covert and complex coding the technique of overt marking has to be reconsidered: The magic formula “one form – one function” leaves considerable room for doubt even when looking at seemingly clear cases of overt grammatical marking. In fact, cases of functional morphemes which are not polysemous are rare. Processes of grammaticalization regularly end up in considerable amounts of polysemy. There is a layering of several stages which survived the process of grammaticalization, for example: demonstrative pronoun, definite article, anaphoric article. Thus, polysemy

S F O PRO



1st proofs


seems to be the goal rather than some by-product of grammaticalization. As polysemy is just the result of underspecified morphology, there is reason to assume that ‘specified morphology’ (overt marking) is the worst case, usually avoided for reasons of cognitive economy. Let us give an example. Even in clear article languages there is no one to one pairing of definiteness functions and overt forms. This is due to the polysemy of the demonstrative pronoun and the definite article and, sometimes, the third person personal pronoun. As linguistic description tends to be hypnotized by overt morphological marking there are large areas of functional patterning still to be discovered. In the light of underspecified morphology one of the most important projects of future research will be syncretisms, the best known being the syncretism of morphological case. It is well known that case syncretism follows privileged pathways. Thus, there is good reason to assume that syncretism is motivated and guided by principles yet to be detected. This view dates back to the famous young-grammarian Delbrück (1907), who dedicated a whole book on case syncretism in Indoeuropean. In other words, syncretism suggests that the form under scrutiny is just an undetected case of underspecification. This seems to be true even for syncretism which transcend the paradigm of a single category and involve different grammatical paradigms. Thus, to give an example, the plural morpheme -s in English is at first sight completely unrelated to the s-affix of third person singular and to that of the possessive. A common denominator even of these seemingly unrelated functions is suggested in Leiss (1997). If you look at the data from the perspective that homonyms are in principle undetected polysemies, then there is no clear case of overt marking in the sense of a one to one mapping of form and function to be attested. At least I could not find a clear example to be inserted here to illustrate overt markings of DPs. The definite article seems to be a good candidate for overt marking of definiteness. However, the visible form shows only part of the meaning of the German (and Dutch) definite article, which shares its form with the demonstrative pronoun and, quite often, with the third person personal pronoun singular. Thus the meaning of the definite article is polysemous and, as a consequence, partly covert. Not until it receives its position in syntax does it bring forward the full meaning of the underspecified form. Thus, the definite article, the demonstrative pronoun, and the third person pronoun share one form, but they neither share their scopes nor their distributional behaviours in syntax. We see again that the semantics of syntax is involved in coding grammatical meanings. It would indeed be interesting to find out whether there exist article systems across languages which are overt in the sense that they create meaning without any participation of silent syntax. In other words, it would be interesting to know whether there are articles which are monosemous. I doubt that there are.

S F O PRO



1st proofs





Elisabeth Leiss

.

Underspecified morphology and the DP-question

There is a lot of uncertainty and discussion (even in the present volume, see Osawa) whether iconic coding and combinatory, indirect coding of definiteness in NPs can be qualified as DPs. The clear answer here is: Yes, they are DPs underlyingly. NPs in covert and semi-covert systems of coding of definiteness are DPs under the same conditions as definite article + NP are in quasi-overt systems (so-called ‘article languages’). We have seen that there is no single article system which exhibits clear overt marking. Underspecification is quite common in morphology. The inflectional paradigms of grammar are full of syncretisms which are assumed to serve a specific end. In the light of silent syntax, underspecification must be interpreted to be the most economic way of encoding functional categories. Underspecification (as formalized in Distributed Morphology) means that the patterns of grammatical coding are not fully spelled out by overt marking. Yet, they are spelled out by syntactic means (word order, scope) and/or by complex morphological patterning (compositional coding). Thus, DPs are always partly spelled out by syntax. In other words, morphology and syntax are natural partners in spelling out definiteness and indefiniteness. There is no reason to qualify the DP-generating contribution of silent syntax as less important or successful than the contribution of morphology. The covert coding of definiteness in the thema in Slavic languages such as in (1) (see 1.1 above) clearly has to be classified as DP. The same holds for combinatory symbolizing of definiteness by article and aspect which are semi-covert DPs, as illustrated in (8) and (9). If we would not accept them as DPs because of their underspecification (be it complete or just partial), we would not be in a position to accept DPs in article systems either. As argued above, articles are polysemous and therefore underspecified as well. What you hear in a sentence is not the whole picture of what you understand. Sentences are full of condensed functional information. Syntax is the means to unzip these bundles of presupposed (iconic) and combined (complex) functional information. . .

A historical view on aspect and definiteness Aspect languages and article languages

The working hypothesis guiding our historical investigation is that the growth of the definite article is connected to changes in the aspectual system of the verb. Definiteness and perfective aspect are two instantiations of the same grammatical function. So are indefiniteness and imperfective aspect (illustrated in more detail in Leiss 2000). Our starting point is the demise of the aspect system in the Germanic languages. When the system of verbal aspect was dissolving, first occurrences of the definite article can be accounted for. Converging evidence from linguistic typology supports

S F O PRO



1st proofs




this correlation: Aspect languages avoid article systems, and article languages avoid aspect. We argue in Leiss (2000) that the renewal of aspect in an article language indicates that the article system is eroding, the erosion being motivated by ‘overuse’ of article systems (hyperdetermination). Case, aspect, and (in)definiteness effects play also an important role at the eve of the rise of the indefinite article. There is much empirically founded evidence that the verbal genitive had a core function in the grammatical architecture of Old High German as part of a still functioning system of verbal aspect. The main assumption is that (in)definiteness effects are achieved in Gothic and early stages of the German language by combining verbal aspect with alternating cases, the main pattern being as in Russian and other Slavic languages: Table 7. Definiteness and indefiniteness effects in Gothic imperfective verb + accusative case = indefiniteness effect impf. fāhan ‘catch’ +  perfective verb + accusative case = definiteness effect pf. gafahan ‘catch’ + perfective verb + genitive case = partitive effect with mass nouns/indefiniteness effect with count nouns pf. gafahan ‘catch’ + 

With the gradual demise of a formerly functioning aspect system in German, the aspectually driven (in)definiteness effects came to be lost. The loss of verbal aspect induced important changes in the system of verbal and nominal quantification of German. The following four stages need to be distinguished (in more detail in Leiss (2000)): 1. The loss of perfective verbs and their indirect definiteness effect was compensated by the emergence and grammaticalization of the definite article in Gothic (4th century) and Old High German (750–1050). 2. The loss of the partitiveness effect of (ge-prefixed) perfective verbs governing the genitive case triggered the emergence of the ‘indefinite article’, whose grammaticalization started in Middle High German (1050–1350). Referential ‘indefiniteness’ started with the partitive meaning solely before gradually developing the indefinite function. Yet and unexpectedly, the set of verbs governing the genitive did not dwindle for lack of a function, but it rose massively. 3. As long as the ‘indefinite article’ was restricted to partitive effects, indefiniteness effects remained to be achieved by the imperfectivity of the remaining unprefixed verbs. Aspectually neutral verbs came to be marked by the verbal genitive only for the purpose of inducing imperfectivity. 4. After a short rise in frequency of the verbal genitive in Middle High German, the system of verbal quantification in terms of aspect oppositions was systematically

S F O PRO



1st proofs



Elisabeth Leiss

replaced by a system of nominal quantification yielding the paradigms of the definite and the indefinite articles. As a consequence, the verbal genitive based upon the system of verbal quantification completely lost ground and disappeared as a verbal case (except for a few fossilized verb valences). .

Hypodetermination vs. hyperdetermination: the case of Icelandic

Hypodetermination is at the very beginning of the process of grammaticalization of a definite article. Hypodetermination means that expressions which are inherently definite are not marked by an article. Inherent definiteness may comprise: ● ● ●

the thema which creates a natural space of definite readings proper names anaphoric reference (reference to known information)

Hypodeterminating languages such as Old Icelandic mark definite expressions indirectly in syntactic areas where definiteness is not to be presupposed: in the rhema. So-called idiosyncratic uses of the definite article in languages such as Old Icelandic or Gothic are completely regular as soon as non-overt marking strategies of definiteness are taken into consideration. Consider the Old Icelandic sequence of Sentences (17)–(20), cited in Heusler (1950: 125) as an example of a totally idiosyncratic use of the definite article. Heusler didn’t notice that the use of the definite article is fully regular as soon as the above mentioned distributional characteristics are taken into account: (17)

Þa verþr hann varr viþ griþung-enn,. . . bull:..    ‘Da wird er den Stier gewahr’ (German translation by Heusler 1950: 125) ‘Suddenly he perceives the bull’

(18)

griþungr snýr í móte . . . bull:.. ‘der Stier kehrt sich ihm entgegen’ ‘the bull turned against him’

 -  (German translation by Heusler 1950: 125)

(19)

griþungr stakk hornonom í síþo hestenom . . . bull:..  -  ‘der Stier stach die Hörner dem Pferd in die Seite’ (translation by Heusler 1950: 125) ‘the bull tossed the horns into the side of the horse’

(20)

konungs menn drópo griþung-enn. bull:.. ‘die Königsmannen erschlugen den Stier’ ‘the king’s men killed the bull’

   (translation by Heusler 1950: 125)

S F O PRO



1st proofs




The translations into German and English language clearly exhibit the features of hyperdetermination. Definite expressions are marked in natural environments of definiteness which are in no need of any overt marking of definiteness. Hypodeterminating languages avoid redundant marking. By contrast, hyperdeterminating languages mark definiteness redundantly. It is interesting to see that hypodeterminating languages signal the core functions of the definite articles only. They do not symbolize anaphoric reference. At first sight, hyperdeterminating languages exhibit an uneconomic way of redundant coding of definiteness. As redundancy is generally avoided in grammar, hyperdeterminating languages tend to transform the uneconomic redundancy of marking into a new function: the anaphoric use of definite articles. There are two known strategies of non redundant marking of definiteness in the rhema: ● ●

complex (indirect) marking: perfective aspect + case overt marking: definite article

When the Germanic languages did loose their aspect systems, they were compensating this loss by the emergence of the definite article from a prior demonstrative function. The interaction of aspect and case was a common pattern in the Gothic language whereas it was already lost in Old Norse. The loss of the aspectual system started in the north of the Germanic languages and expanded south with a delay of several centuries (Leiss 1992: 69). .

Gothic: interaction of verbal and nominal definiteness effects

In contrast to Old Icelandic (7th century and later), the aspectual system of the Gothic language (4th century) is elaborate and functioning. Table 8. Aspectual pairs (drawn from Krause 1987) 



E

G

andjan arman dailjan dauþjan digan drigkan taujan trauan waljan

ga-andjan ga-arman ga-dailjan ga-dauþjan ga-digan ga-drigkan ga-taujan ga-trauan ga-waljan

‘end’ ‘pity’, ‘to take pity’ ‘share’ ‘kill’ ‘form’ ‘drink’ ‘do’ ‘trust’ ‘elect’

‘enden’ ‘erbarmen’ ‘teilen’ ‘töten’ ‘bilden’, ‘aus einem Teig machen’ ‘trinken’ ‘tun’ ‘trauen’ ‘wählen’

S F O PRO



1st proofs



Elisabeth Leiss

We argue (along with Lehmann 1988) that aspect and aktionsart interact in a specific way. All aktionsart verbs belong to one of the two aspectual poles. They are either perfective or imperfective. Table 9. Aspect and aktionsart in Gothic simplex V

saihvan ‘see’

pf. prefix V

E

and-saihvan at-saihvan bi-saihvan ga-saihvan us-saihvan in-saihvan þairh-saihvan

‘look at’ ‘passim consider’ ‘look at’, ‘keep an eye on’ ‘look at’, ‘look around’ ‘see’, ‘perceive’ ‘look up’, ‘look at’ ‘look at’, ‘take care of ’, ‘glance up’ ‘look through’, ‘see in the mirror’

In Gothic as well as in Slavic languages, prefixed verbs generally are perfective verbs plus an additional semantic modification. There are ‘good’ aspect partners such as saihvan ‘to see (impf)’ and ga-saihvan ‘to see (pf)’, which are devoid of additional aktionsart semantics. Beyond that, there are peripheral aspect partners, which are clear aktionsart verbs. However, it is not excluded that there are contexts where a semantically modified aktionsart verb is preferred. According to Krause (1987: 209) and Leiss (2000: 152–155) the functional range of perfectivizing ga- is: ● ● ●

ga- perfectivizes the unmarked imperfective verb: renders aspectual function ga- symbolizes temporal completion in the past: renders temporal function ga- refers to definite and localizable events in opposition to recurrent (habitual) events: renders definiteness effect

There is a clear interaction between verbal aspect and nominal determination. In Gothic perfective verbs lose their perfective prefix when combined with explicitly marked definite objects. We conclude from this behaviour that the perfectivizing prefix ga- and the demonstrative pronoun (which is the source of the definite article) are functionally equivalent and interchangeable in verbal predicates. In (21) the meaning of the verb hausjan (impf. ‘hear’) is clearly perfective, the perfectivity being coded by the demonstrative pronoun þata ‘that’. (21)

jah fullai waurþun (Luke 4,28 in Streitberg 1908/2000) . full:.... became:3.. and full became allai modis in þizai all:... mood:..  :.. all of angriness in the

S F O PRO



1st proofs




swnagogein hausjandans þata. Synagoge:... hear:. :.. synagoge hearing that. ‘They were all filled with wrath in the synagogue, as they heard these things’

Table 10 shows the different possibilities of complex coding in the Gothic texts: Table 10. Patterns of complex coding in Gothic perfective verbs perfective verbs imperfective verbs

+ + ±

imperfective verbs + imperfective verbs imperfective verbs

+ +

complement (): renders definiteness effect complement (): renders partitiveness effect/indefiniteness effect complement (): renders indefiniteness effect

complement ():

complement (:): renders definiteness effect/aspectualizing effect complement (:): renders partitiveness effect

Only perfective verbs allow for paradigmatic case alternation (between  and ). Imperfective verbs normally do not select complements in the genitive case. But interestingly enough, they do so when combined with definite expressions. Definite expressions formally perfectivize imperfective verbs and thus allow for the selection of the genitive, which is otherwise selected by perfective verbs only (for more details see Leiss (2000: 152–155)). .

Old High German: the rise of hyperdetermination

Old High German shows patterns of article use that differ considerably from Old Icelandic and Gothic. Old High German primarily uses the definite article in the thematic part of the sentence (left margin of IP or Middle Field, essentially; see Abraham 2006). The definite article in Old High German is mostly used anaphorically. See (22). (22)

Latin Old High German In principio erat uerbum In anaginne uuas uuort & uerbum erat apud deum. ‘Inti thaz uuort uuas mit gote. & deus erat uerbum, Inti got selbo uuaz thaz uuort, [Tatian 65, 16–18 in the edition of Masser; Tatian 1, 1 in the edition of Sievers]

In the Latin text the definiteness of the prior-mentioned uerbum ‘word’ is coded covertly by the discourse semantics of linear syntax. In Old High German there is a different rule: Definiteness has to be coded overtly even when it is coded already implicitly by its linearly signaled thematicity. This overload of definiteness marking gave birth to the anaphoric function of the definite article. That means that the definite article is invested with transphrastic functions. The anaphoric functions were not the primary ones to emerge when the article system started. Anaphoric functions

S F O PRO



1st proofs



Elisabeth Leiss

emerge when there is hyperdeterminating usage of the definite article. But there are still traces of the older non-anaphoric article system. We find also the individualizing function of the definite article in Old High German. See (23). (23)

Latin Old High German In ipso uita erat; thaz uuas In imo lîb; & uita erat lux hominum. Inti thaz lib uuas lioht manno. & lux In tenebris Inti thaz lioht In finstarnessin luc&.’ & tenebrae liuhta.’ inti finstarnessi eam non comprehenderunt., thaz nibigriffun, [Tatian 65, 24–28 in the edition of Masser; 1, 3–4 in the edition of Sievers] The English translation is: In him was life, and the life was the light of men. The light shines in the darkness, and the darkness hasn’t overcome it. [Tatian 65, 24–28] [= S 1, 3–4] [= John 1, 4–5 in the Bible]

Besides the anaphoric use of the article in (24), a. lîb [+] - thaz lib [+] b. lioht [+] - thaz lioht [+] we find abstract nouns such as finstarnessi without the anaphoric article function. It is well known that abstract nouns avoid the use of the article. Nevertheless there are instances of finstarnessi with the definite article: (24)

Latin Si ergo lumen quod in te est tenebre sunt.{ipse tenebre quantæ erunt.

Old High German oba thaz lioht thaz thar in thir ist finstarnessi ist. thiu finstarnessiu vvuo mihhilu sint

The English translation is: If therefore the light that is in you is darkness, how great is the darkness. [Tatian 153, 26–28] [= S 37, 4] [= Matthew 6, 23]

The definite article refers to a specific darkness in contrast to darkness in general. This individualizing effect is well known with lexemes such as menigi ‘crowd’ and folc ‘people’. The definite article is used when the noun refers to a specific group of people. There are 5 exceptions to this rule which have been documented by Jäger (1917: 27), but were then left unexplained. An explanation offers itself, when the functional equivalence of definite article and perfective aspect is taken into account. The so-called exceptions are not exceptions at all. The abstract noun menigi with specific reference regularly is left without article because the verb involved is perfective, marked by the perfectivizing prefix gi-. See (25).

S F O PRO



1st proofs


(25)



OHG: gisah trumbara inti menigi sturmenta -saw (the) flute players and (the) wailing crowd.

Jäger (1917: 27) misses the definite article, because in this example, menigi refers to a specific, well-known crowd of people which in all the other 42 instances of a definite crowd are marked by the definite article. However, we have learned from Gothic that the perfective aspect in the above examples blocks the use of the definite article, being strong enough to create definite expressions. What we see is a hint to what is left from a former functioning system of verbal definiteness effects, which became opaque when the anaphoric use of the definite article opens the way for hyperdetermination. Obviously, Old High German has a hybrid system of article usage: There is still the former aspectual system which creates definiteness effects. Thus, the so-called irregular article use in Old High German is quite regular once the intricate interrelationships of aspect and article systems are taken into account. .

The complementarities of article paradigms and aspect systems

Once we take into account that the structural area of the discourse-functional thema and perfective aspect create definiteness effects then the so called irregularities of article use in Old Icelandic, Gothic, and Old High German will turn out to be part of a regular pattern of definiteness effects. These effects are created by the invisible effects of syntax (thematic space), by complex patterning (aspect plus case), and by overt morphological markers. They add up to the whole picture of DP-symbolizing means of a specific language. These different means are complementary, and they pattern differently in different languages. Essential to an understanding of article usage is that the process of generating definite reading must not be confused with the process of creating anaphoric reference. The frequency of article use in so called article languages such as English and German is due to the additional anaphoric function of the articles in these languages. However, the function to create text cohesion is but a secondary and derived use of the core function of the definite article in hyperdeterminating languages. By contrast, in hypodeterminating languages article use is restricted to create sentence internal definiteness effects. In these languages articles do not create effects of cohesion between sentences. They are used to create DPs, but not to reiterate the actual definiteness of a DP. By suggesting that there are DP-creating devices in sundry languages which have been neglected so far we do not aim at inflating the notion of DP. Quite the contrary is the case. There is good reason to assume that most of the research on DPs is hypnotized by overt morphology and is thus extending the notion of DP far too much. For instance, it is common practice to classify demonstrative pronouns

S F O PRO



1st proofs



Elisabeth Leiss

(DemPro) plus NP as DPs. In the next section arguments will be presented that show that demonstrative pronouns cannot create DPs. . .

The emergence of articles: demonstrative pronouns as predecessors Why demonstrative pronouns are no good at creating articles

There is an essential difference between demonstrative pronouns and articles which has hitherto gone unnoticed: Demonstrative pronouns are equipped with different deictic qualities compared with articles (see Abraham, this volume a). They are less endowed with deictic capacities than articles are, and they are less complex than articles. The lesser complexity is deducible from the fact that demonstrative pronouns are the privileged source of the grammaticalization of articles. In addition to this indirect evidence it is possible to provide functional reasons for the difference between demonstrative pronouns and articles. Demonstrative pronouns belong to the class of so-called shifters, a notion introduced by Jakobson (1957). Shifters are anchored in the viewpoint of the speaker. In other words, the viewpoint of the speaker has to be taken in consideration in order to work out the reference of the demonstrative pronoun and other shifters. To the class of shifters belong a large amount of adverbials such as here, yesterday. Shifters of this kind have simpler deictic force. The notion of simple deictic force means that deixis is dependent on the viewpoint of the speaker and that there is no other deixis involved. They differ from the class of complex shifters. Complex shifters are functional categories such as definiteness/indefiniteness (for instance articles), aspect and tense. In order to illustrate the difference of simple shifters such as the demonstrative pronoun and complex shifters such as the definite article a look at tense might be helpful. Time adverbials such as yesterday refer only if the viewpoint of the speaker is known. Quite different from these are tense grammemes such as the . They are endowed with double reference in the following sense. First, they localize the speaker (utterance time), and second they shift the point of the viewer (reference time) into the . When shifters such as yesterday are used in sentences with past time reference they are clearly anchored in the  and not at the point of view of the speaker. In other words, there is a double process of shifting. First, the anchoring of the viewpoint is shifted into the , and second, the temporal deixis of yesterday starts from the shifted point in the  and not from the utterance time. The difference between simple and complex temporal deixis can be also explained by looking at the localization of the speaker and the hearer. When using the time adverbial yesterday the speaker localizes himself in the past whereas the hearer remains localized in the present whereby, when using the past tense, speaker und hearer share the same viewpoint in the past (the reference point).

S F O PRO



1st proofs


Thus, tense is able to create a shared viewpoint, whereas simple shifters are devoid of this function. The difference between simple deixis and complex deixis has been introduced has been developed in more detail in Diewald (1991). We see clearly that time adverbials are unable to shift the reference point away from the utterance point. This is the reason why time adverbials can never compensate for the lack of tense in a language. Nor can modal adverbials for the epistemic or root readings of modal verbs (as in Germanic). The same holds for demonstrative pronouns. They cannot compensate for the lack of definite articles in a language because definite reference established by definite articles is independent of the viewpoint of the speaker. It is definite anyway as it refers to common knowledge which is not the privileged knowledge of the speaker only. In other words, the function of a definite article is to create a shared viewpoint. Definite articles refer to shared information. This piece of information is known by the speaker as well as by the hearer. Thus, definite articles refer to shared familiarity of the information, whereas demonstrative pronouns signal no more than the familiarity of the information on the side of the speaker only. Thus, demonstrative pronouns refer to information which is known to the speaker, but unknown to the hearer, whereas definite articles refer to information which is known information for both hearer and speaker. What this all amounts to is the conclusion that functional categories essentially create a shared viewpoint for both speaker and hearer. This viewpoint is the anchoring point for simple shifters such as demonstrative pronouns and time-, place- and manner-adverbials. The function of real grammatical (functional) categories is to create a place of shared viewpoint or shared perspective. Thus, functional categories contribute to the ability of building up a “theory of mind”, which is species-specific of the human kind. For this reason the importance of the difference between simple shifters and double shifters cannot be overestimated. Higher mammals such as dogs or primates are of course able to understand the use of a demonstrative gesture. However, they are unable to understand the deictic function of a definite article. There is good reason to believe that linguistic devices which create a shared view point and thus are essential for the development of a theory of mind are universals of language. For this reason it would be hazardous to assume languages without DPs. If we realize that demonstrative pronouns and other simple shifters are not equivalent to articles, we may understand why languages without articles do not display a higher amount of demonstrative pronouns than article languages. What we find in languages without articles are either invisible (iconic) or complex means of patterning definiteness/indefiniteness. Remains the question why demonstrative pronouns qualify as preferred source concepts of articles. This is what we tackle in the following section.

S F O PRO



1st proofs





Elisabeth Leiss

. How come that demonstrative pronouns are the universal source of definiteness? It is well established common opinion that the source of overt marking of definiteness is the demonstrative pronoun. Until now an explanation has not yet been offered for why the demonstrative pronoun is chosen as the best, and probably only, source of definiteness. The diachronic section above helps to answer this question. We have seen from data in Old Icelandic and Gothic that the rise of the definite article starts in the rhema section of the clause, for good reasons. In order to answer the question why the demonstrative pronoun is the privileged source of overt marking of definiteness we have to unfold the functional closeness between the demonstrative pronoun, the rhema, and definiteness. This seems to be impossible at first sight. Normally we would assume functional affinity between the rhema and indefiniteness. As there is indeed no way to find a functional link between the inherently indefinite rhema and definiteness we have to bridge this functional gap by turning to the functional capacities of the demonstrative pronoun. A rich source for our investigation is Centering Theory as presented in Abraham (2002) and in Abraham (this volume a). The central result of Abraham’s paper (this volume a) is that the anaphoric capacities of the demonstrative pronoun (DemPro) refer to the rhema, in contrast to the personal pronoun which refers to the thema of the preceding sentence. Thus, there is a strong affinity between DemPro and the preceding rhema. This affinity accords well with the fact that the rise of the definite article in Old Icelandic started in the rhema: A definite rhema is marked by a definite article in order to inhibit the indefinite reading presupposed by rhematic objects. Starting from this insight there are four phases of article development to be reconstructed. See the 4 phases of diachronic emergence below. Phase I: The demonstrative pronoun refers to rhematic arguments Demonstratives are universal and, according to Centering Theory, they refer to the rhema of the preceding sentence. In this function they refer to definite and indefinite objects equally. Now let us turn to the earliest documents of Old Norse (Runic inscriptions). In Old Norse indefinite and definite objects did not have to be marked overtly. There is good reason to assume that indefinite objects were encoded by the iconic technique of SVO-serialization, whereas definite objects were encoded by SOV-serialization as in Mandarin Chinese (see (5) and (6) above). In Runic inscriptions, two thirds of the sentences have SVO-word order, whereas one third has SOV (Braunmüller 1982: 126–140). As definite objects are marked occurrences, SOV is supposed to be less frequent than SVO. This is in agreement with the data in Old Norse (Leiss 2000: 56, Fn. 4).

S F O PRO



1st proofs




In this first iconic phase of serialization a demonstrative pronoun referred anaphorically to indefinite objects (in a SVO-serialization) as well as to definite objects (in a SOV-serialization). This process of anaphoric reference turns indefinite rhematic material into given information. Thus, indefinite objects become definite ‘one clause later’. The difference between definite and indefinite objects is not active any longer in clause x+1. Phase II: The demonstrative is introduced one clause earlier in order to create a definite object The logic unfolding from our starting point is evident. If demonstrative pronouns in clause x+1 create a definiteness effect even for indefinite objects referred to in sentence x, then why not recruite demonstrative pronouns in order to create definite objects from the start on, i.e. in clause x. This step initiates phase II where demonstrative pronouns appear in rhematic position in clause x in order to mark objects as definite. These pronouns do not refer anaphorically to any precedent information. They refer to shared knowledge of speaker and hearer. Thus, there is complex deixis involved and not just the simple deixis emanated by demonstrative pronouns. They signal a common point of view of the speaker and the hearer. In phase II it is impossible to use the new born article in order to refer anaphorically. In Old Icelandic definite objects which are marked by the definite article in clause x do not appear with definite articles when they reappear as thematic subjects in the following clause (x+1). This is in line with the logic of hypodetermination sketched in Section 2.2, where we observed that in the beginning of article development definiteness need not be marked in structural definiteness areas of a clause. Phase III: Hyperdetermination: Definiteness has to be marked in any case The next step of the emerging article is characterized by the overgeneralized use of the definite article. Hyperdetermination appears at the moment when definite articles are used whenever an argument is definite. At this phase even inherently definite subjects are marked by the definite article. Hyperdetermination generally leads to functional redundancy which otherwise is principally avoided in grammar. As a consequence of this avoidance strategy the redundancy of marking is functionalized: The open marking of inherently definite arguments provokes the rise of a new function – the anaphoric function of the article. Notice that there is no reason to mix up the anaphoric function of the article (in phase III) with the anaphoric function of a demonstrative pronoun (in phase I). Demonstrative pronouns are simple shifters with anaphoric reference, whereas definite articles are complex shifters with anaphoric reference (see 3.1). Complex anaphoricity is shared anaphoric reference whereas simple anaphoric reference depends on the viewpoint of the speaker only. The claim

S F O PRO



1st proofs



Elisabeth Leiss

that anaphoricity is implied and, as a consequence, must be derived from the singulative function, i.e. the core function of the definite article, is substantiated by more formal arguments by Abraham in the introduction to this volume (Stark/Leiss/Abraham this volume). The diachronic section of this paper made plausible that the anaphoric use of the definite article is clearly a derived one. It is derived from clause-internal definiteness which is characteristic of the early phases of article emergence. But how do we know which use of the article has to be classified as its core function? The answer is that it seems reasonable to restrict the core function to the clauseinternal function of a grammatical category. Textlinguistic functions are always derived. The same holds for the textlinguistic functions of grammatical categories such as tense, aspect, and mood. The fact that tense and, in principle, all grammatical categories can be used in order to provide textual cohesion is not reason enough to classify these textual uses as a core function. The same holds for the anaphoric use of the definite article (as well as for the cataphoric use of indefinite articles). In sum, the creation of textual cohesion by means of anaphoric articles cannot qualify for the core function of the article.

.

Conclusion

Our main aim was to show that the patterns of definiteness may be far less visible, more complex and more difficult to understand than is normally assumed. This amounts to the conclusion that whenever a language is said to lack DP, this is due to the fact that the full patterning of definiteness may yet have to be discovered. A caveat is appropriate in order to avoid the premature and ill-motivated claim of an existing DP-less language. In order to substantiate this danger some hitherto neglected forms of definiteness patterning were presented. Central to our investigation in definiteness patterns is the mereological insight that article functions and the functions of verbal aspect have a substantial intersection of properties. Thus, the focus of attention in definiteness research should be continually on aspect and related factors. There is another good reason to posit DP as a structural universal: DPs belong to the universal tools of functional categories which are characterized by the potential to create shared view points in linguistic communication. Shared viewpoints are the prerequisite for developing a theory of mind which is characteristic of human communication. Another important claim was that caution is required when comparing demonstrative pronouns with definite articles. Demonstrative pronouns cannot create DPs because they are not really functional categories. A functional category which deserves this name has complex deictic force. Complex deixis means that a viewpoint shared both by the speaker and the

S F O PRO



1st proofs




hearer is created. The position defended here is that DPs do have this quality in contrast to NPs modified by a demonstrative. Finally, we reconstructed three phases of article emergence, which starts from the demonstrative pronoun (phase I), develops via the core function of the definite article (phase II) to the anaphoric function of the article (phase III). The aim of the diachronic section has been to show that the core function of the definite article cannot be claimed to be identical to its anaphoric function. From a diachronic view the anaphoric function is clearly derived from the core function of the definite article. Our diachronic reconstruction demonstrates that the anaphoric function of the definite article emerges in hyperdeterminating languages, i.e. in languages which mark definiteness redundantly (i.e. according to phase III). In other words, redundant marking of definiteness is the very moment of the emergence of the anaphoric function of the definite article. In order to avoid redundant double marking of definiteness on the DP the hyperdeterminated DP is reinterpreted as an anaphor. That means that hyperdeterminating usages of the definite article are read as coindexes transgressing sentence boundaries. In other words, hyperderminating markings of definiteness are reinterpreted as coreferences extending beyond the sentence boundary. Thus, anaphoricity is not created by contextual information. Anaphoricity emerges by the process of sentence internal and hyperderminating marking of definiteness. The present results might help to open the way to a unified description of definiteness and anaphoricity without being forced to claim a common function of definite and anaphoric expressions. Heusinger’s claim “that anaphora and definiteness are two sides of the same familiarity principle” (Peregrin/Heusinger (2003: 2), who follow Heim 1982) are misguided in the sense that they are not able to explain why there are languages without the anaphoric function of the article. Another weakness of their approach is that they have to reconstruct the meanings of anaphoric article usages by referring to informational context, whereas the present approach enables one to reconstruct the anaphoric usages by referring to sentence-internal hyperdermination of DPs without extra contingencies. In other words, sentence internal double marking of definiteness suffices to create the anaphoric article. There is no need to refer to the context in order to detect the anaphoric function of an article.

References Abraham, Werner. 1996. Kasus, Aspekt und nominale Referenz: Komplexe kausale Zusammenhänge in der Diachronie des Deutschen und ihre formale Darstellung auf vergleichender typologischer Grundlage. In Language Change and Generative Grammar [Linguistische Berichte; Sonderheft 7], Ellen Brandner & Gisella Ferraresi (eds) 22–70. Opladen: Westdeutscher Verlag.

S F O PRO



1st proofs



Elisabeth Leiss Abraham, Werner. 1997. The interdependence of case, aspect and referentiality in the history of German. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 29–61. Cambridge: Cambridge University Press. Abraham, Werner. 2002. Pronomina im Diskurs: Deutsche Personal- and Demonstrativpronomina unter ‘Zentrierungsperspektive’. Grammatische Überlegungen zu einer Teiltheorie der Textkohärenz. Sprachwissenschaft 27(4): 447–491. Abraham, Werner. This Volume a. Discourse binding: DP and pronouns in German, Dutch, and English. Abraham, Werner. This Volume b. The discourse-functional crystallization of the historically original demonstrative. Abraham, Werner. 2006c. Topic, focus, and default vs. contrastive accent: Typological differences with respect to discourse prominence. In Information structure and the architecture of grammar: A typological perspective [Linguistik Aktuell/Linguistics Today], Kerstin Schwabe & Susanne Winkler (eds), Amsterdam: John Benjamins. Aristoteles. 1984. Metaphysik. Schriften zur Ersten Philosophie. Übersetzt und hg. von Franz F. Schwarz [Reclam Universalbibliothek 7913]. Stuttgart: Reclam. Baetke, Walter. 1965/1993. Wörterbuch zur altnordischen Prosaliteratur. 5., unveränderte Auflage [Sitzungsberichte der Sächsischen Akademie der Wissenschaften zu Leipzig, PhilologischHistorische Klasse; Vol 111, 1–2]. Berlin: Akademieverlag. Birkenmaier, Willy. 1977. Aspekt, Aktionsart und nominale Determination im Russischen. Zeitschrift für Slavische Philologie 39: 398–417. Birkenmaier, Willy. 1979. Artikelfunktionen in einer artikellosen Sprache. Studien zur nominalen Determination im Russischen [Forum Slavicum 34]. München. Braunmüller, Kurt. 1982. Sprachtypologische Studien zum Germanischen [Tübinger Beiträge zur Linguistik 197]. Tübingen. Chesterman, Andrew. 1991. On definiteness. A study with special reference to English and Finnish [Cambridge Studies in Linguistics 56]. Cambridge: Cambridge University Press. Delbrück, Berthold. 1907. Synkretismus. Ein Beitrag zur germanischen Kasuslehre. Strassburg: Trübner. Diewald, Gabriele M. 1991. Deixis und Textsorten im Deutschen [Reihe Germanistische Linguistik 118]. Tübingen: Niemeyer. Donhauser, Karin. 1990. Moderne Kasuskonzeptionen und die Kasussetzung im Althochdeutschen. Überlegungen zur Stellung des Objektsgenitivs im Althochdeutschen. In Neuere Forschungen zur historischen Syntax des Deutschen. Referate der Internationalen Fachkonferenz Eichstätt 1989 [Reihe Germanistische Linguistik 103], Anne Betten (ed.), 98–112. Tübingen: Niemeyer. Donhauser, Karin. 1991. Das Genitivproblem in der historischen Kasusforschung. Ein Beitrag zur Diachronie des deutschen Kasussystems. Ms, University of Passau. Eco, Umberto & Marmo, Constantino (eds). Glosses on the Medieval Theory of Signs [Foundations of semiotics 21]. Amsterdam: John Benjamins. Fischer, Susann. 2005. Interplay between reference and aspect. In Proceedings of the workshop “Specificity and the evolution/emergence of nominal determination systems in Romance”, K.V. Heusinger, G.A. Kaiser & Elisabeth Stark (eds), 1–16. Konstanz: Fachbereich Sprachwissenschaft der Universität Konstanz. Gladrow, Wolfgang. 1972. Sprachliche Mittel zum Ausdruck der Determiniertheit/Indeterminiertheit. Zeitschrift für Slawistik 17: 647–656. Gladrow, Wolfgang. 1979. Die Determination des Substantivs im Russischen und Deutschen; eine konfrontative Studie. Leipzig: VEB Verlag Enzyklopädie.

S F O PRO



1st proofs

Covert (in)definiteness and aspect Greenberg, Joseph H. 1966. Language Universals. With special reference to feature hierarchies [Janua linguarum. Series minor 59]. The Hague: Mouton. Hallberg, Peter. 1965. Om språklika författarkriterier i isländska sagatexter. Arkiv för nordisk filologi 80: 177–186. Jakobson, Roman. 1957. Shifters, Verbal Categories, and the Russian Verb. Cambridge MA: Harvard University. Jäger, Paul. 1917. Der Gebrauch des bestimmten Artikels bei Isidor und Tatian vergleichend dargestellt. PhD Dissertation, Leipzig. Krifka, Manfred. 1986. Massennomina mit einem Exkurs zu den Aktionsarten [Sonderforschungsbereich 99]. Konstanz: Universität Konstanz. Lehmann, Volkmar. 1988. Der russische Aspekt und die lexikalische Bedeutung des Verbs. Zeitschrift für slavische Philologie 48: 170–181. Larsson, Lars-Gunnar. 1983. Studien zum Partitivgebrauch in den ostseefinnischen Sprachen [Acta Universitatis Upsaliensis. Studia Uralica et Altaica Upsaliensia 15]. Uppsala. Leiss, Elisabeth. 1987/1990. Grammatische Kategorien und sprachlicher Wandel: Erklärung des Genitivschwunds im Deutschen. In Proceedings of the XIVth International Congress of Linguists (held in Berlin/GDR, 1987), Vol. 2, Werner Bahner, Johachim Schildt & Dieter Viehweger (eds), 1406–1409. Berlin: Akademie-Verlag. Leiss, Elisabeth. 1992. Die Verbalkategorien des Deutschen. Ein Beitrag zur Theorie der sprachlichen Kategorisierung [Studia Linguistica Germanica 31]. Berlin: Walter de Gruyter. Leiss, Elisabeth. 1997. Synkretismus und Natürlichkeit. Folia Linguistica 31: 133–160. Leiss, Elisabeth. 2000. Artikel und Aspekt. Die grammatischen Muster von Definitheit [Studia Linguistica Germanica 55]. Berlin: Walter de Gruyter. Krámsky, Jiři. 1972. The Article and the Concept of Definiteness in Language. The Hague: Mouton. Krause, Maxi. 1987. Sémantique et syntaxe des préverbes en Gotique. 4 Vols. Thèse. Paris: Sorbonne Paris IV [unpublished]. Leiss, Elisabeth. 1994. Die Entstehung des Artikels im Deutschen. Sprachwissenschaft 19: 307–319. Leiss, Elisabeth. 1997. Synkretismus und Natürlichkeit. Folia Linguistica 31: 133–160. Leiss, Elisabeth. 1998. Aristotelische Linguistik. Der Neubeginn einer Philosophischen Grammatik durch Jean-Marie Zemb. Sprachwissenschaft 23: 141–165. Leiss, Elisabeth. 2000. Artikel und Aspekt. Die grammatischen Muster von Definitheit [Studia Linguistica Germanica 55]. Berlin: Walter de Gruyter. Lewandowska, Ewelina. 2006. Tempussysteme europäischer Sprachen. Handout: Universität München 2006. Li, Charles N. & Thompson, Sandra A. 1981. Mandarin Chinese. A functional reference grammar. Berkeley CA: University of California Press. Masser, Achim (ed.). 1994. Die Lateinisch-althochdeutsche Tatianbilingue. Stiftsbibliothek St. Gallen Cod. 56. Unter Mitarbeit von Elisabeth De Filip-Jaud hg. von Achim Masser [Studien zum Althochdeutschen 25]. Göttingen: Vandenhoeck & Ruprecht. Peregrin, J. & Heusinger, K. von .2003. Dynamic semantics with choice functions. Paper. September 5, 2003. Pinborg, Jan. 1967. Die Entwicklung der Sprachtheorie im Mittelalter [Beiträge zur Geschichte der Philosophie und Theologie des Mittelalters. Texte und Untersuchungen XLII: 2]. Münster: Aschendorffsche Verlagsbuchhandlung. Pinborg, Jan. 1982. Speculative grammar. In The Cambridge History of Later Medieval Philosophy, Norman Kretzmann et al (eds), 254–269. Cambridge: Cambridge University Press.

S F O PRO



1st proofs



 Elisabeth Leiss Pinborg, Jan. 1984. Modus significandi. In Historisches Wörterbuch der Philosophie, Band 6, Joachim Ritter & Karlfried Gründer (eds), 67–71. Darmstadt: Wissenschaftliche Buchgesellschaft. Rieger, Gerd Enno. 1968. Die Spitzenstellung des finiten Verbs als Stilmittel des isländischen Sagaerzählers. Arkiv för nordisk filologi 83: 81–139. Sievers, Eduard (ed.). 1960. Tatian. Lateinisch und althochdeutsch mit ausführlichem Glossar. Nachdruck der 2., neubearb. Ausgabe [Bibliothek der ältesten deutschen Literatur-Denkmäler 5]. Paderborn: Schöningh. Stark, Elisabeth. 2006. Indefinitheit und Textkohärenz. Entstehung und semantische Strukturierung indefiniter Nominaldetermination im Altitalienischen, Tübingen: Niemeyer. Streitberg, Wilhelm. 1908/2000. Die gotische Bibel. Vol. 1: Der gotische Text und seine griechische Vorlage. Mit einem Nachtrag von Piergiuseppe Scardigli. 7th edn. Heidelberg: Winter. Thomas of Erfurt. 1972. De modis significandi sive grammatica speculativa. An edition with translation and commentary by G.L. Bursill-Hall. London: Longman. Van den Toorn, M.C. 1959. Zur Verfasserfrage der Egilsaga Skallgrímssonar. Köln: Böhlau. Wexler, Paul. 1976. On the non-lexical expression of determinedness (with special reference to Russian and Finnish). Studia Linguistica 30: 34–67.

S F O PRO



1st proofs

The definite article in Indo-European: Emergence of a new grammatical category? Brigitte L. M. Bauer The University of Texas at Austin, USA

The development of the definite article in Indo-European is a complex innovation, because if most modern languages have definite articles, there are important exceptions. In addition, within the Indo-European dialect groups the development of the definite article may not always seem consistent. Moreover, the definite article may trace back to different elements in the individual languages and its functions may vary cross-linguistically within a subgroup, possibly reflecting different degrees of grammaticalization. On the basis of patterns in early uses of definite articles in Greek and article-like uses of demonstratives in Latin, I will trace the change in question, evaluate the possible role of language contact and the possible connection with other phenomena, among them devices to express definiteness in Indo-European languages that do not have definite articles. The aim of the paper is to establish whether or not the definite article is a truly innovative feature in Indo-European or whether it is merely a formal innovation of a category that existed already.1

Introduction In his 1974 analysis of language development Benveniste distinguished two types of change. In so-called conservative changes (transformations conservantes) a grammatical category is preserved, but there may be a formal substitution as in the replacement of case endings by prepositions or the replacement of a synthetic comparative form by an [adverb + adjective] combination (1974: 127). Within the category of so-called “innovative transformations” (transformations innovantes) Benveniste distinguishes (a) changes that lead to the loss of a grammatical category, illustrated by the shift from a three-gender to a two-gender system in Indo-European (partial loss) or the disappearance of the dual; and (b) changes

S F O PRO

. I am obliged to the editors of this volume for their useful comments and suggestions. It goes without saying that shortcomings of the article are all mine.



1st proofs

 Brigitte L. M. Bauer

that lead to the creation of a new grammatical category. He specifically refers to the definite article and the development of adverbs out of compound-like combinations (English -ly ; Romance -ment[e]); 1974: 126–7). The question that comes up is indeed how to classify the creation of definite articles in Indo-European: is it an innovation as Benveniste proposed in his publication? The definite article is a wide-spread cross-linguistic phenomenon, geographically as well as linguistically. Yet it is not frequent nor predominant or typical in any language family. In any language group one may find languages without definite articles and languages with definite articles, which moreover often reflect different etymologies and represent different types. The definite article is found in Semitic languages such as Arabic, Hebrew, and Aramaic, but the place of occurrence varies: in modern Arabic and Hebrew it precedes the noun; in the earliest Aramaic texts it follows the noun, but in modern Eastern Aramaic dialects a newly developed “prefixed article” may be found (Greenberg 1978: 59). Elements conveying the function of a definite article are found in numerous languages in Africa. In Gurma, a group of Voltaic languages, for example, the definite article takes the form of a prefix, a suffix, or a more general preposed marker and conveys various levels of the (non-generic) article in Greenberg’s terminology (1978: 54–9). Other languages that feature definite articles are Khasi (Austroasiatic; prefixed article), several Austronesian languages on New Britain and New Ireland, and several native Australian languages. Amerindian languages do not offer many “certain instances of the non-generic article” although gender markers on the noun in several languages suggest a definite article at some stage in the language development (Greenberg 1978: 60–1). The definite article is attested in Indo-European as well, i.e. in modern IndoEuropean languages, primarily in the West. It is not found in today’s Indian languages, Modern Persian, Lithuanian, Latvian and the majority of Slavic languages. This observation brings up the question to what extent the definite article is a contact phenomenon. Bulgarian, for example, is the only Slavic language that has definite articles, sharing this characteristic with other languages of the so-called Balkan Sprachbund, Albanian and Rumanian (for an overview of explanations of this characteristic of the Sprachbund, see Sandfeld 1930: 167–8). The postposed definite articles in Albanian and Rumanian are similar and developed before the first written attestations of the languages; the development was later in Bulgarian. Sandfeld argued that the postposed article was a simultaneous development in Albanian and Rumanian and that Bulgarian developed it later in the Middle Ages, possibly as a contact phenomenon with Rumanian (1930: 165–73). If the development of the definite article is an areal phenomenon, it is not strong enough to have affected Serbian, which still has no definite article, or Greek that has had preposed articles from Antiquity on.

S F O PRO



1st proofs

The definite article in Indo-European 

Another possible example of contact-induced development may be the article in Latin/Romance, which according to Meader & Wölfflin (1900) and Salonius (1920), for example, traces back to the definite article in Greek. They argue that by way of Biblical texts Greek had an importance impact on the development of definite articles in Latin/Romance. From a historical perspective we know that definite articles generally are not original. They are absent in early Indo-European languages and in early Hebrew texts. In historically attested Indo-European languages the emergence of the element can be traced, as in Greek or Latin/Romance. In this article we will first examine the use of the definite article in Ancient Greek and then compare these patterns to uses in Latin/Romance. We subsequently will expand our observations to Indo-European languages that do not have definite articles, but that seem to have devices that mark definiteness. These processes traditionally have passed unnoticed. We thus will address the question to ask at a deeper level: does the emergence of the definite article as we know it in modern Western Europe reflect the emergence of a new grammatical category? .

The definite article in Greek

The definite article in Ancient Greek – masculine ‘ο, feminine ‘η, and neuter το – traces back to ‘ό, ‘ή, and τό, which were used primarily as demonstratives in Homer, having deictic function ‘this, that’. Article-like uses are attested (cf. also Section 1.1) in Homeric Greek, but there are numerous instances in Homer’s text without article where the article is obligatory in later documents (Schwyzer & Debrunner 1950: 22; 23). The later Greek texts have definite articles. Homer’s use of definite articles may therefore reflect a relatively advanced transitory stage in the development of articles. Further research into Homer’s usage may reveal patterns similar to those in other languages, such as Germanic, where article distribution in the earliest instances is related to rhematic position (see Leiss 2000). The demonstrative in Homeric Greek was characterized by the trichotomy that is found in other Indo-European languages as well, but in all probability is not inherited from the protolanguage (Brugmann 1911: 310–47; esp. 346–7), cf.: (1a)

ο ‘this’ (‘this, close to me’, 1st pers.) (< ο + δε) (2) ουτος ‘that’ (‘that, close to you’, 2nd pers.) εκεινος ‘that’ (‘that, close to him’, 3rd pers.)

. The following abbreviations have been used in the glosses: . = ablative; . = accusative; . = adverb; . = aorist; . = definite article; . = comparative; . = dative; . = genitive; . = imperfective; . = infinitive; . = middle; . = neuter; . = nominative;

S F O PRO



1st proofs

 Brigitte L. M. Bauer

Whereas the element ‘ό in Homer typically is used to contrast two objects or persons, or their actions, in later uses the element primarily defines the noun with which it is combined (Smyth 1956: 285–6). When ό became a definite article, the Greek demonstrative system did not disintegrate: a demonstrative particle −δε ‘here’ was added to the original ó and the demonstrative trichotomy was re-established, cf.: (1b)

οδε ‘this’ (‘this, close to me’, 1st pers.) (< ο + δε) ουτος ‘that’ (‘that, close to you’, 2nd pers.) εκεινος ‘that’ (‘that, close to him’, 3rd pers.)

In addition, an emphatic form emerged when −ι was added: (2)

οδι: οδι ανηρ ‘this man here . . .’

Demonstrative ο survives – as a residue – in combination with reinforcing elements, such as µεν, δε, especially in contrasting expressions with personal pronouns, cf.: (3)

οι µεν. . . οι δε . . . ‘some . . ., others . . .’

We find it also, again in combination with reinforcing elements, in contrastive adverbial expressions (4) and in adverbials with contrastive value (5), cf.: (4)

το µεν . . . το δε. . . ‘on the one hand . . . , on the other hand . . .’

(5)

το δε το και το

‘whereas’ ‘this and that’

The following pages will present an overview of uses of the definite article in Greek, its semantic motivation (Section 1.1) and its syntactic contexts (Section 1.2). For further details and examples of semantic and syntactic use, see Smyth (1956) and Schwyzer & Debrunner (1950). .

Semantic motivation

The definite article in non-Homeric Greek typically is used to refer to persons and objects that are known, as individual or group. It therefore specifically is used

. = participle; . = person; . = perfective; . = plural; . = present; . = preterite; . = particle; . = singular; 1. = first person singular; 3. = third person singular; 3. = third person plural; 1. = first person plural.

S F O PRO



1st proofs

The definite article in Indo-European 

with nouns referring to elements that one has in mind because they have been mentioned already, as in: (6)

Αετος. . . . . ., ο αετος . . . (Aes., Fab. 3.1) (3)(4) eagle-. A.-.. eagle-. ‘An eagle . . . , the eagle . . . .’

When the object, person, or event has not been mentioned yet, but is expected or is known, or is known to occur, the definite article is used as well, cf.: (7)

το µέρος των ψήφων ο A.-.. part-. A.-.. votes-. A.-.. διωκων ουκ έλαβεν prosecutor-. not get-.-3. ‘the prosecutor did not get the (requisite) part of the votes (that everybody knows he needs; Smyth 1956: 287)

Similarly, the definite article is obligatory when the noun refers to a person or object that is well-known, cf.: (8a) (8b)

η των Ελληνων ευνοια ‘the (well-known) goodwill of the Greek’ η σχολη ‘your (usual) idleness’ (examples from Smyth 1956: 289)

. In the bibliographical references the following abbreviations have been used: Authors and their works: Aes., Fab. = Aesop, Fables; Ant., DOC = Anthimus, De Observatione Ciborum; Arist., Pax = Aristophanes, Pax ; Aug., DCD = St. Augustine, De Civitate Dei; Aug., BV = St. Augustine, Beata Vita; Cato, R.R. = Cato, De Re Rustica; Cic., Fam. = Cicero, Ad Familiares ; CIL = Corpus Inscriptionum Latinarum; Her. = Herodotus; Hom., Od. = Homerus, Odyssee ; Hom., Il. = Homerus, Iliad; Hor., Carm. = Horace, Carmina; Pers. = Persius Flaccus; Petr. = Petronius; Pl., Amph. = Plautus, Amphitruo; Pl., Curc. = Plautus, Curculio; Pl., Stich. = Plautus, Stichus; Pl., Ep. = Plautus, Epidicus; Pl., Men. = Plautus, Menaechmi ; Pl., Mil. = Plautus, Miles Gloriosus; Pl., Rud. = Plautus, Rudens; Pliny, Ep. = Pliny, Epistulae ; Plu., Ant. = Plutarchus, Antonius; Ter., Phor. = Terence, Phormio; Ter., Heaut. = Terence, Heautontimorumenos ; Xen., A. = Xenophon, Anabasis; Xen., Hell. = Xenophon, Hellenica; Varro, R.R. = Varro, De Re Rustica. Biblical texts: Act. = Actus apostolorum; 1 Cor. = Ad Corinthios (1st letter); 2 Cor. = Ad Corinthios (2nd letter); Esr. = Liber Ezrae; Gal. = Ad Galatas; Gen. = Genesis; Math. = Gospel by Mathew; Mich. = Micha; Joh. = Gospel by John; Jud. = Judith; Ps. = Psalms; 2 Tim. = Ad Timotheum (2nd letter). Other: CdR = Chanson de Roland; Yv. = Yvain. . Examples have been taken from the Loeb collection, with the exception of: Chambry, Emile (ed). 1960. Esope, Fables. Paris: Les Belles Lettres; Moignet, Gerard (ed). 1969. La chanson de Roland. Paris: Bordas; Roque, Mario (ed). 1978. Yvain. Paris: Champion.

S F O PRO



1st proofs

 Brigitte L. M. Bauer

Moreover, generic uses of the noun also require the use of the definite article, as in the following examples from a Fable by Aesop, featuring two animals in which reference is being made to humans, who have not yet been mentioned: (9a)

αλεκτρυών οχληρος τοις ανθρώποις νυκτωρ cock-.. troublesome-.. A.-.. men-. by night κεκραγως (Aes., Fab. 12.3) croak-part.-.. ‘cock’s croaking at night is troublesome to human beings’

(9b)

προς ωφελειαν των δεσποτων (Aes., Fab. 12.8–9) for the sake of benefit-. A.-.. masters-. ‘for the benefit of the masters’

Similarly, the generic value of the noun in the next examples requires a definite article, cf.: (10)

δεοι τον στρατιώτην φοβεισθαι µαλλον must-3. A.-.. soldier-. fear-. more τον άρχοντα ή τους πολεµíους (Xen., A. 2.6.10) A.-.. commander-. than .-.. enemies-. ‘a soldier should fear his commander more than the enemy’ (Example from Smyth 1956: 287–8)

(11)

ο µυθος δηλοι ότι ο καμάτος A.-.. fable-. show-3. that A.-.. work-. θησαυρός εστι τοις ανθρώποις (Aes., Fab. 83. 10–1) treasure-. be-3. A.-.. men-. ‘this fable shows that work is a treasure for man’

Yet this use is not systematic, as adverbial expressions of time indicate or prepositional phrases with temporal value, as in the following examples, cf.: (12)

υπο νύκτα δια νυκτος δι’ημέρης

‘just at nightfall’ ‘in the course of the night’ ‘the whole day long’

Moreover, in given contexts a noun with a definite article may contrast with the same noun without definite article as in the opposition specific vs. generic, cf.: θεος ‘god, in generic sense’ vs. ο θεος ‘a specific god’. A similar use of the definite article is found in its combining with abstract notions. The rule of thumb is that abstract nouns require a definite article: (13a)

(ότι) τα χαλεπα των πραγμάτων (that) A.-.. difficult-.. A.-.. circumstances-.

(13b)

ο χρονος διαλύει (Aes., Fab. 30.10) A.-.. time-. resolve-3. ‘(that) time resolves difficulties’

S F O PRO



1st proofs

The definite article in Indo-European 

Yet abstract nouns that refer to virtues, vices, arts, and sciences do not combine with a definite article; neither do abstract nouns conveying space or measures, cf.: (14a) (14b)

τι μανια ‘what is madness?’ (example from Smyth 1956: 289) ημας περιφεύγειν δει φιλíαν ων . . . (Aes., Fab. 60.13) we-. avoid-. must-3. friendship-. those-. ‘we should avoid the friendship of those who . . .’

Already in Homeric Greek the definite article in anaphoric use combines with proper names when the name is repeated, cf.: (15)

Πελοπι. . . αυταρ ο αυτε Πελοψ . . .

At a first appearance the definite article may be used when the person is well known, e.g. ο Σολων ‘Solon’. Because of their “uniqueness of reference” (Lyons 1999: 21) proper names do not need a definite article. Although reflecting an advanced stage of development, this use in Greek does not reflect a fully overgeneralized use of definite articles in that language because the definite article occurs only if the person is known, in accordance with the use of common nouns. In late uses the definite article typically combines with (borrowed) proper names that are non-declinable, cf.: (16)

.

Αβρααμ εγέννησεν τον Ισαάκ (Math. 1.2) Abraham-. beget-.-3. A.-.. Isaac ‘Abraham begat Isaac’

Syntactic contexts

In addition to uses reflecting certain semantic characteristics, the definite article occurs in given syntactic contexts. First of all, in addition to nouns the element in Greek typically combines with adjectives; it then has nominalizing function, cf.: (17)

ο σοφος ‘the wise man’ (< σοϕος ‘wise’) του κακου ‘of evil’ (Hom.) (< κακος ‘bad’) τα μεγιστα ‘greatness’, ‘great deeds’ (< μεγιστος superlative of μεγας ‘big’)

In line with this usage, the definite article also combines with participles: here too the definite article has a strong straightforward nominalizing function, cf.: (18)

τα εσομενα το εσομενον εκ ο βουλεμενος

‘the things that are to be, future . . .’ (< εσσομενος part. middle of ειμι ‘be’) ‘future revenue from . . .’ (Plu., Ant. 24) ‘the one who wishes/anyone who wishes’ (< βουλομαι ‘wish’)

The combination may lexicalize: (19)

οι λογοποιουντες

S F O PRO

‘the newsmongers’ (< οι + λογον + ποιουντες)



1st proofs

 Brigitte L. M. Bauer

Similarly, the definite article nominalizes infinitives, cf. the example from Aesop: (20)

ο λογος δηλοι ότι πολλάκις το A.-.. fable-. show-3. that often A.-.. πεíθειν του βιάζεσθαι ανυτικώτερόν εστι persuade-. A-.. use-violence-. efficient-comp... be-3. ‘this fable shows that persuasion often is more efficient that violence’ (Fab. 73.12–3)

This linguistic process allows for nominal categories – case – to be expressed in verbal forms; consequently infinitives may function as full nouns and yet combine with a direct object where in Latin, for example, gerundive constructions are needed. The definite article in Greek further combines with many other grammatical categories as well, generally with nominalizing function, such as adverbs, as in (21), numerals, as in (22), or personal pronouns, as in (23), cf.: (21)

το πρíν ‘the past’ (< adv. πριν ‘before’, ‘formerly’) οι τοτε ‘the men of that time’ (< adv. τοτε ‘at that moment’) οι παλαι ‘the ancients’ (< adv. παλαι ‘formerly, in the past’) απο του νυν ‘from now on’ (< adv. νυν ‘now’) οι εκει ‘the dead’ (< adv. εκει ‘there’, ‘in the past’) (εκει was a euphemism for εν Αδου ‘in the underworld’)

(22)

ο των επτα σοφωτατος Σολων εις παρα τους δεκα (examples from Smyth 1954)

‘Solon, the wisest of the seven (Sages)’ ‘one man in ten’

(23)

τους υμας

‘our troops’

The definite article also combines with genitives and prepositions, but then the syntax is more complex because of ellipsis, cf.: (24)

τα πολιμου

‘matters of war’

(25)

οι επι των πραγματων

‘those in power’

The occurrence of definite articles in combination with proper names varies with registers: in the spoken language an article is included, cf.: (26)

Δημοσθενης ο Αλκισθενους

‘Demosthenes, son of Alkistenos’

whereas the official language does not have a definite article in this context, cf.: (27)

Περιδικκας Αλεξανδρου ‘Peridikkas, son of Alexander’ (examples from Smyth 1956: 290)

Although vocatives in Ancient Greek typically do not combine with definite articles, one may find instances of so-called “semi-vocative use”, a use comparable to what we also find in modern French, as in the following examples: (28)

οι εξ

‘you six . . .’

S F O PRO



1st proofs

The definite article in Indo-European 

Compare the use in modern French: (29)

bonjour les enfants, . . .

‘hello children, . . .’

Finally, the definite article in Greek typically combines with nouns that already have a determiner, either a demonstrative or a possessive; this pattern is often attested in Homer as well, cf.: (30)

.

ουτος ο ανηρ ‘this man’ τον εμον βιβλιον ‘my book’ vs. εμον βιβλιον ‘a book of mine’

Definite articles in Greek: preliminary conclusion

Evaluating the uses and contexts, we conclude that Ancient Greek is characterized by an advanced use of definite articles. An advanced stage both from Greenberg’s (1978) and Lyons’ (2000) perspectives. While Lyons is right in including in his “lifecycle” of definite articles phenomena that one may qualify as “pre-definite” (Lyons 1999: 277–281; 323–331), I focus in this paragraph on the actual use of definite articles as described by Greenberg. The use of the definite article in Greek does not precisely fit any of the three stages of article development as defined by Greenberg (1978). From purely deictic elements that may develop discourse function (Stage 0), demonstratives may become definite articles that combine with elements that are identified (e.g. mentioned previously) or generally known (e.g. unique elements, such as the sun) (Stage I). Stage II definite articles are fully grammaticalized and are used in definite determination and in “non-definite specific” contexts. While “the articulated form of the noun has become the normal form of the noun” (Greenberg 1978: 63), definite articles are not used in specific contexts (e.g. certain negative or generic contexts or contexts that are inherently definite, such as proper names). Stage III definite articles have become “mere markers” (Greenberg 1978: 61–74). Greenberg himself was the first to underscore the relativity of these stages, pointing out that languages will present “transitional phenomena such that certain languages are on the borderline between two stages” (1978: 61). Some structures will be “well advanced” in the development of the definite article, while others still present residues or in Greenberg’s terminology “in [some] instances it is clear that [the language] has only entered the stage recently” (1978: 61). Yet Greenberg also stresses that “the whole development is to be viewed as a single continuous process marked by decisive turning points” (1978: 61). With these caveats in mind we may summarize the use of definite articles in Greek as follows: the definite article is used both to mark definite determination and in generic instances, reflecting – mostly – the characteristics of a Stage-II

S F O PRO



1st proofs

 Brigitte L. M. Bauer

definite article. A typical feature of a Stage II definite article is not the (non-) existence of certain uses, but rather inconsistencies, the existence of contexts that show variation in the occurrence of a definite article, reflecting a difference of meaning. If the definite article, for example, occurs with generic nouns, there typically will be certain generic contexts in which the definite article will not occur. The definite article may be absent, for example, with generic nouns in negated contexts. Similarly, the absence in Greek of the definite article in many prepositional expressions conveying time or location reflects a Stage-II situation (see Greenberg 1978: 67–8). What does not correspond to this stage is the use of definite articles in Greek in combination with proper names or with nouns that are already determined by a demonstrative or a possessive. The occurrence of demonstratives and possessives with a definite article is qualified by Greenberg – albeit implicitly – as a progressive feature (1978: 66). Similarly, Lyons argues that Italian, which has these constructions, reflects a more advanced stage than French (Lyons 1999: 337). Yet data from Romance, where these combinations are found in early varieties (e.g. Old French), but not in all later languages (e.g. Modern French; for Italian dialects, see Rohfls 1969: 31–2), would suggest that it is not a progressive stage, but rather a transitional, albeit innovative, stage. Further cross-linguistic research is needed to substantiate this hypothesis and this is the topic of another paper (Bauer, In preparation a). In terms of syntactic effects, definite articles in Ancient Greek typically have a strong nominalizing function and combine with a wide range of grammatical categories that are being nominalized. Finally, another important finding in the history of Greek articles is their speedy development. Considering the fact that Homer basically had no article use, the functions and contexts in Ancient Greek are the result of a remarkably swift innovation. Greek data present a development, a complex of uses, and a combination of contexts not found elsewhere in IndoEuropean languages. Yet despite this advanced use, we also find many inconsistencies, comparable to what is found in Old French (e.g. Epstein 1993): certain well-defined patterns, but many exceptions.

.

Latin and Romance

Forerunners of the Romance definite article are found in Vulgar and Late Latin and, as said, according to some they trace back to Greek influence (for a discussion of some of these interpretations, see Leumann & Hofmann & Szantyr 1965: 191–2). Yet the data do not seem to support this scenario.

S F O PRO



1st proofs

The definite article in Indo-European

.



Short history

The Romance definite article traces back to a variety of Latin elements, i.e. demonstratives and an identative (ipse). Latin as well had a demonstrative system based on three distinctions, cf.: (31)

hic ‘this’ (1st pers., EGO) iste ‘that’ (2nd pers., TU) ille ‘that’ (3rd pers., ILLE)

Among these hic and iste typically refer to elements within the sphere of those communicating. Ille typically is a demonstrative that refers to “absens vel longe posita persona” (Priscian GL 2.577.20) ‘a person who is absent or far away’. Ille includes a particle followed by a demonstrative element -l-, which is found in a series of adverbial and prepositional formations conveying the notion of distance, e.g. Latin prepositions uls ‘beyond’, ultra ‘on the farther side, beyond’, the adverb olim ‘at that time (past or future)’, or the pronoun alius ‘another as opposed to . . .’. Similarly, hic comprises a particle followed by a former demonstrative -ce, marking proximity. This element is manifest in the short forms of the paradigm and in the long forms only in early Latin texts as in Plautus’s, harunc ‘these’ (genitive feminine). Like hic and ille, iste is a compound form including the particle is and the demonstrative element -te, which is inflected. Despite etymological correlates in other Indo-European languages (e.g. Skt. tá-; Gk τó; Go. Þata), the precise deictic value (proximity or remoteness) is not clear. In Armenian, which also has a tripartite system, the t- element is used in the same function, referring to someone or something that has just been mentioned and that is neither close (hic) nor remote (ille ; Ernout & Meillet 1959: 324). In addition, Latin had an anaphor, is, which refers to an element mentioned or that is about to be mentioned, without having deictic value. It typically refers to an antecedent. In the long run is disappeared as a grammatical marker, surviving longest in fixed expressions, such as id est. In the Romance languages it only survived in isolated lexical elements, as in Italian desso, which traces back to id ipsud. Finally, there was an identative ipse ‘same, -self ’. The pronoun is a combination of an element i- and a reinforcer -pse. Whereas in Classical Latin the declension is on the second element (e.g. ipse, ipsa, ipsum, ipsius, ipso, ipsum, etc.) there are instances in early documents of the type eapse, eopse, and so forth (Ernout & Meillet 1959: 322). There are instances of double inflection, external as well as internal, as in eapsa or eumpsum in Plautus, for example, reflecting the shift to external inflection (e.g. ipsum). The -pse suffix includes a reinforcing -p- (see also meopte ‘my very own-abl.’; Ernout & Meillet 1959: 323; for a different interpretation, see

S F O PRO



1st proofs

 Brigitte L. M. Bauer

Palmer 1954: 257). Like the demonstratives, ipse could be used as a pronominal and as an adjectival element. In its combining with nouns lies it syntactic potential to become a definite article. As an identative its original use was rather emphatic and it typically combined with focussed elements, often with contrastive value: ‘X himself, instead of Y’. From this situation a development took place at both formal and functional levels. Moreover, the development in fact is twofold: the emergence of the definite article in Latin coincides with the collapse of the demonstrative system. In Vulgar and Late Latin texts we observe a confusion of hic and iste, as in the following inscription where iste, or rather its derivative, clearly has the value of [here]: (32)

Ille mortuus quei istic sepultus est (CIL 1.1012) ‘the dead who is buried here’

When hic subsequently disappeared, iste survived as the sole EGO demonstrative, marking proximity. Hic only survived in a number of fixed expressions, some of which subsequently lexicalized, as in Sp. ahora (< hac hora, ‘this hour-abl., now’). This phenomenon already existed in Latin, as in hodie, which traces back to hoc die ‘this day-abl., ‘today’. In a parallel development ille came to be used as a TU demonstrative as well and alternated in this use with ipse (see below). In the Late Latin and Early Romance period the demonstrative value of the elements of the deictic trichotomy came to be reinforced by ecce, with the original meaning ‘behold’ (cf. also the etymology of Fr. voilà including the imperative of the verb voir ‘see’). In accordance with its etymology (ecce features the same -c- we find in the paradigm of hic), ecce originally was used to indicate that something was present, ‘here’ cf. ecce me ‘here I am’. In combination with demonstratives ecce had perhaps a reinforcing demonstrative effect, but no deictic value. In the spoken varieties of Latin we find instances of ecce + demonstrative, even in early Latin already, as in eccum, eccos, eccillim, eccillud, and so forth, cf: (33a)

Eccos exeunt ! (Pl., Mil. 1310) behold-they-. go.out-3. ‘Look, here they come’ (Ernout & Thomas 1964: 23)

(33b)

Certe eccistam video (Pl., Curc. 615) certainly behold-she-. see-1. ‘I certainly see her right before my face’

As a result the early Romance deictic system traces back to a situation that includes the following elements: (34)

1st pers. ecce + istum 2nd pers. ecce + ipsum and ecce + illum 3rd pers. ecce + illum

S F O PRO



1st proofs

The definite article in Indo-European 

In some of the Romance languages the ecce ipsum combination disappeared and with it the original trichotomy; this pattern is found in French, Provençal, standard Italian and most of the Italian dialects. In standard Italian, for example, we find: (35a)

[proximity] accu + istum > questo ‘this’ [distance] accu + illum > quello ‘that’

In others (e.g. Ibero-Romance) the threefold distinction is still valid today because ecce ille disappeared as a TU-demonstrative, cf.: (35b)

accu + istum > Sp. (aqu)este accu + ipsum > Sp. (aqu)ese accu + illum > Sp. (aqu)el

In dialects in Southern Italy the trichotomy survived but without ecce, which is exceptional: (35c)

1st pers. istum > stu, sta 2nd pers. ipsum > ssu, ssa 3rd pers. illum > ddu, dda (Rohlfs 1968: 247)

In the next pages I will further discuss the uses and occurrences of demonstratives in Latin. .

Demonstratives in Latin

In early Latin hic, iste, and ille were true deictic elements as well as pragmatic deictic elements, cf.: (36)

illa filia amici . . .? . . . hanc . . . mittimus? . . . Dem.-. daughter friend-. . . . Dem.-.. send-1. illa maneat? (Ter., Phor. 812) Dem-.. stay-3. ‘that daughter of our friend . . .? . . . Do we send her away? And keep the other?’

Important aspects of the grammaticalization process are the extension of function (true deictic > pragmatic deictic > other . . .) as well as semantic bleaching. While most studies on the (early) emergence of the definite article focus on the possible “articoloid” uses of demonstratives, they ignore the actual occurrence of the element. Yet the frequency of the elements in question may be perhaps more important for the actual grammaticalization process. The following pages focus on this aspect of the matter. In his extensive study of the use of adjectival demonstratives in a number of Latin authors by Fischer (1908) we find remarkable data concerning the actual occurrence of the various elements. Fischer used for his analysis selected writings

S F O PRO



1st proofs

 Brigitte L. M. Bauer

of Cicero (Epistulae and In Verrem IV), Pliny (Epistulae), Cassiodorus (Variae [letters]), Livy, Seneca, and Tacit. On the basis of his statistical data we notice first of all that of the four elements—hic, iste, ille, and is—is is by far the most frequent: is (645 instances), hic (582 instances), ille (294 instances), and iste (102 instances; Fischer 1908: passim). The element without deictic value therefore is most frequent. Moreover, the use of is decreases dramatically with time, whereas that of iste and ille increases and that of hic remains basically the same. From that respect comparison of the relative data from Cato and Petronius is telling: Table 1. Occurrence adjectival demonstratives, Cato and Petronius

Cato Petronius

Hic

iste

Is

ille

ipse

Not given 177

2 10

80 9

12 42

8 35

(Source, Fischer 1908: passim)

Among the demonstratives iste is least frequent in early and Classical Latin. Statistical data from Petronius further show that from least frequent in Cato (2 of a total of 94+) iste comes third in Petronius (10 of a total of 238), but is still rare. The persistence of the use of hic in Petronius’s text is remarkable as well, especially as compared to the dramatic decrease of is. In terms of word order the sequence [demonstrative + noun] clearly predominates: Table 2. Sequence [Demonstrative + Noun] vs. [Noun + Demonstrative]

Hic Iste Ille Is Totals

Demonstrative + Noun

Noun + Demonstrative

562 66 194 410 1232 (68,6%)

15 23 40 186 564 (31,4%)

(Source, Fischer 1908: passim)

Because of the varying length of the texts that have been analysed by Fischer it is not possible to tell whether there is a difference in demonstrative use between the different genres. The predominance of the sequence [Demonstrative + Noun] is a recurring phenomenon in analyses of individual authors as well, cf. Caesar’s De Bello Gallico, which features 610 instances of Dem N vs. 4 instances of N Dem, or Cicero’s (486 vs. 28), Augustine’s (596 vs. 82), Aulus Gellius’ (1427 vs. 453) writings (see Muldowney 1937: 73–4).

S F O PRO



1st proofs

The definite article in Indo-European 

Fischer’s statistical evidence is further supported by his analysis of the individual instances. He found that the sequence [Noun + Demonstrative] is motivated mainly because the noun in these contexts is strongly emphasized (1908: 6; 33; 66; 93; 114–5). The demonstrative forerunner and the later definite article in Romance – with the exception of Rumanian – therefore are characterized by the same sequence. The fifth element in the development of the demonstrative/definite article is ipse, which is attested in no less than 345 instances in the Fischer corpus (1908: 101–16). The use of adjectival ipse increased with time as well (compare e.g. Cato [8] vs. Petronius [35]). Because of its function as identative, ipse relatively often combines with proper names (e.g. ‘Caesar himself did this and this’), and it often has contrastive value. Whereas ipse more strongly is connected to the noun – hence the high incidence of juxtaposition – the preference for preposing is not as distinct as for the other elements: on a total of 345 instances, 203 have [ipse + Noun]; 134 have [Noun + ipse]. More than is the case for demonstratives, the use of ipse inherently is strongly emphasized (for statistical data, see Fischer 1908: 101–16). What is perhaps most remarkable about Fischer’s statistical data is the consistent and ongoing high frequency of non-deictic is, closely followed by hic. This incidence contrasts sharply with the ultimate fate of these two elements: disappearance, possibly related to their lack of phonological weight. With the loss of initial h- and the non-consistent occurrence of -c, the forms of the hic paradigm indeed had relatively little phonological body left in the later stages of Latin. Yet the occurrence of hic in Latin texts remains high throughout the Latin period also at times when some of the phonological changes referred to above had taken place. Moreover, from a typological perspective there seems to be a tendency of definite articles developing from distant, third person demonstratives, rather than those that mark proximity (Greenberg 1978: 61). So two factors were possibly at work, phonological and general linguistic ones. Another finding is the important increase of occurrence of adjectival demonstratives in absolute sense as well as the important increase of their occurrence with another demonstrative or – predominantly – with ipse or idem. Rare or absent in Cato, Varro, and the more literary documents, these instances are often attested in Pliny’s and Cicero’s letters (data: Fischer 1908: 117–36). Co-occurrence of these demonstratives underlines the process of semantic bleaching, which therefore must have been early. .

Article-like uses of demonstratives in Vulgar and Late Latin

S F O PRO

Perhaps the most remarkable phenomenon in later Latin texts is the apparent lack of consistency in article-like use. Vulgar and Late Latin texts show quite some



1st proofs

 Brigitte L. M. Bauer

variety in data: some documents, for example, have very few instances of ille, others have few instances of ipse. Moreover, studies like the one by Aebischer (1948) clearly show that ipse is well preserved if not predominant in documents covering a wide variety of genres and areas, including those where it did not survive as definite article (1948: 200– 201), e.g. the Peregrinatio, the Regula Benedicti, the Mulomedicina and other wellknown sources (see Selig 1992: 132 for an overview; Renzi 1976: 28). The rather limited distribution of the ipse article in today’s Romance therefore contrasts with its occurrence in terms of geographical spread and use in Vulgar and Late Latin documents all over the Empire.5 In addition to a rather inconsistent use, interpretation is another thorny question. Some scholars tend to interpret in a rather loose way any instance of ille + noun, for example, as “articoloid”, as forerunner of definite article + noun combinations. Others, such as Löfstedt (1956), warn against easy identification of definite uses. The precise value of demonstratives in concrete uses – the presence or absence of deictic value – is extremely difficult and open to discussion as Löfstedt (1956) clearly shows. Yet for the diachronic process that was taking place, it is important to note also that a demonstrative may occur in a Vulgar or Late Latin text which in earlier times or in more rigid style would not have been used at all. In Section 2.2 we already pointed out the increase in use of demonstratives in Latin. This tendency only became stronger. A case in point may be the next example from the Itala translation: (37)

Exiit ergo Petrus et ille alius discipulus (Joh. 20.3) ‘Peter went forth and that other disciple’

In this instance ille still has deictic value (Löfstedt 1956). Yet in documents in a style closer to the norms of Classical Latin, ille in this context would not occur at all. Another telling example is the emergence of demonstratives in temporal indications including the nouns annos, dies, or tempus, which in earlier time would have no demonstrative, cf.: (38)

ante hos dies (Jud. 5.22) (for further discussion of these constructions, see Löfstedt 1936)

. Ipse survives as definite article in Southern parts of Romania today, in Sicily, parts of Sardinia (Jones 1993), and in isolated dialects of Italy, the South as well as the North (e.g. Apenine dialects) (Rohlfs 1968: 248–51). In the West ipse is found in dialects in the Provence, Gascony, and Catalonia, where it is used as article or occurs in article-like uses, as well as in the Balneares. In some of these dialects ipse as article is only attested in early varieties. Among varieties that include ipse even today, Sardinian is best known, where it is attested from the earliest documents on (see Jones 1993).

S F O PRO



1st proofs

The definite article in Indo-European 

There is a vast literature on the development of the definite article in the transition from Latin to Romance, most of which focus on the early stages as assumed for Vulgar and Late Latin (e.g. Abel 1971; Bouvier 1972; Faingold 1996; Lapesa 1961; Löfstedt 1936, 1956; Löfstedt 1961; Orlandini 1982; Renzi 1976; Rosén 1994; Selig 1992; Werner 1998; Wolterstorff 1917, 1920). The outcome of these analyses is remarkably diverse. Some of these studies focus on one demonstrative element in a number of authors or just one author or focus on a group of demonstratives in one writer. Topics that are discussed in the existing literature on the development of the Romance definite article are the actual occurrence of article-like demonstratives in Latin, the deictic vs. non-deictic value of Latin demonstratives in specific contexts (Löfstedt 1956 vs. Woltertorff 1919), the origin of article-like uses of Latin demonstratives (e.g. Löfstedt 1956; Löfstedt 1961), and so forth. Yet others attempt to understand the “rivalry” between ipse and ille. Traditionally it has been assumed that adjectival ille and ipse had similar meaning and function and that there were sociolinguistic differences. More recent analyses, each with their own terminology and approaches, have indicated that the two elements had distinct functions and meanings, but that these may have overlapped. In an analysis covering a wide range of Latin texts among them the Peregrinatio, Renzi (1976: 29–31) found that ipse had specific textual reference, whereas ille was more general. Nocentini (1990) found that ille was a cataphoric element whereas ipse was anaphoric (Nocentini 1990: 146 and passim), and Selig (1992) on the basis of her analysis of a series of medieval documents (legal texts, 2nd–5th centuries AD and hagiographies, 580– 800 AD), came up with a functional distribution, whereby ille specifically combined with elements that are mentioned for the first time; ipse specifically is an anaphoric element (Selig 1992: 132 et passim). This function is in line with the original value of ipse, emphasizing the identity of a person already mentioned. In these uses presumably lies the beginning of the later development (see below). The findings and observations by Nocentini and Selig are tendencies, there is no precise information as to the distribution of ille and ipse, foretelling their grammaticalization in Romance. We know where ille and ispe survived, but we do not know why. The problem of evaluation and the inconsistency of occurrence explain why despite the extensive analyses there is no clear picture that emerges. There is apparently no straight-forward development that leads from a demonstrative element into a definite article with the various diachronic stages as reconstructed for processes of grammaticalization. What emerges however – despite the diverse results – is the long-lasting strong persistence of the hic paradigm, both in pronominal and adjectival use, the clear decrease with time in the use of is and its predominant use as a pronoun rather than adjectival element (362 pronominal vs. 152 adjectival

S F O PRO



1st proofs

 Brigitte L. M. Bauer

uses, numerical data from Wanner 1987: 97), the relatively infrequent occurrence of iste, even in later and more vulgar documents, the high occurrence of ipse in some of the documents (see also below), and the fact that when it comes to frequency nothing seems to predict the strong survival of ille as definite article in Romance. This last observation is not unique to the demonstrative/definite article scenario: I made similar comments pertaining to the development of –mente adverbs. The frequency of adverbial [adjective + mente] combinations in comparison to other adverbial [adjective + noun] sequences does NOT foretell the survival and wide-spread use of –mente as the Romance adverbial suffix (Bauer 2002). Later we will further discuss possible similarities between the development of –mente and that of ille. .. Syntactic contexts While most analyses focus on the semantics of the use of demonstratives/articles, relatively little attention is paid to the actual syntactic use of these elements. In the next few pages we will discuss aspects of the syntax of demonstratives/articles in several Late or Vulgar Latin documents. I will use data from earlier studies and my own data; I will refer to the element in question as demonstrative/article. As the heading of this subsection indicates, I will focus on combinatory features, but within each category semantic characteristics come into play as well. In Vulgar and Late Latin documents we find all three demonstratives in articlelike functions. Ille, for example, typically occurs with proper names and nouns referring to “well-known” entities, as in: (39)

ille Iupiter Dem.-.. Iupiter-. ‘the (well-known) Iupiter’ ille Hypocrates medicus (Varro, R.R. 1.4.5) Dem.-.. Hypocrates-. doctor-. ‘that (well-known) doctor, Hypocrates’

A similar use is attested with common nouns, cf.: (40)

Velleius Blaesus, ille locuples Vellius-. Blaeses-. Dem.-.. rich-.. consularis (Pliny, Ep. 2.20.7) person.of.consular.rank-nom. ‘Velleius Blaesus, a person of consular dignity’

In combination with nouns referring to concrete objects the demonstrative/article conveys the idea of ‘you know’ or ‘the one that everyone knows about’, cf.: (41)

S F O PRO

timet ne tua duritia . . . illa . . . be.afraid-3. your-.. severity-. Dem.-..



1st proofs

The definite article in Indo-European 

adaucta sit (Ter., Heaut. 435) increase-.-.. be-.-3 ‘he is afraid that that severity of yours has increased’ (42)

in illud subterraneum cubicullum (Pliny, Ep. 4.11.9) into Dem.-.. subterraneous-.. cell-. ‘into the subterraneous cell, . . .’ (a cell where prisoners on death row are brought)

Moreover, the demonstrative/article in combination with a common noun may express the notion of “inherently connected to”, as in: (43)

illa nocte Dem.-.. night-. ‘that/the same night’

The demonstrative/article combines also with adjectives and adverbs, reflecting a pattern already observed for Ancient Greek. Yet the phenomenon does not – by far – have the same extent as in Greek and the combination with participles is rare. Cf. the following example including an adjective: (44)

ille insanus (Pl., Men. 336) Dem.-.. insane-.. ‘that/the madman’

Ille with a nominalized adjective with the value ‘well-known’: (45)

Manabat enim illud malum urbanum spread-3.. because that-. bad-. of-town-. et ita corroborabatur cottidie (Cic., Fam. 12.1) and thus increase-3.. daily ‘that evil in town was spreading and thus it increased every day’

It is significant that Latin in its later or spoken varieties often uses nouns that nominalize the adjective, such as vir or, more frequently homo, as in the following examples from Plautus: homini libero (Pl., Rud. 114), vir scelestus (Pl., Rud. 1059), improbissumo homini (Pl., Rud. 662), and others (examples from Wolterstorff 1920: 80). Conversely, it is important to point out that Latin adjectives without demonstrative/article easily are nominalized, not only adjectives like malum or bonum, but also many others (e.g. sapienti [Pl., Rud. 432], miseram [Pl., Rud. 232], see Wolterstorff 1920: 80). With possessive pronouns – a common combination – the demonstrative/ article has nominalizing function, cf.: (46a)

illum tuum Dem.-.. your-.. ‘the one you own’

S F O PRO



1st proofs

 Brigitte L. M. Bauer

illo nostro Dem.-.. our--. ‘the one we own, our . . .’ (46b)

illa nostra scilicet ceciderunt that-.. our-... clearly fall-3.pf. ‘those projects of ours (mine) have fallen through’ (Cic., Ep. 13.47)

Yet the nominalizing function is not consistent in all pronominal contexts, as the next example illustrates, where the result is pronominal: (47)

Adolescens . . .. Quid ille alter (Petr. 81.5) young.man-.. what Dem.-.. other-.. ‘A young man who . . . What about the other?’

In Classical Latin already demonstratives typically combine with elements that are not declinable or rather that are syntactically isolated within the clause, such as quotations, cf.: (48)

Et hercule quousque illa vulgaria ‘quid agis ecquid commode vales? (Pliny, Ep. 3.20.11) ‘And really, how about the common questions ‘How do you do?’ and ‘I hope you are well?’

The phenomenon spread during Christian times, when numerous “exotic” names came into the language. Although it never became a wide-spread device as in Greek, the infinitive – another indeclinable – may be nominalized as well. Instances are attested in Persius, for example: (49)

Nostrum istud vivere triste (Pers. 1.9; Löfstedt 1956: 364) our-. this-. live-. sad-. ‘our sad existence’

Possessives may also have this function, as in: (50)

Meum . . . intelligere nulla pecunia vendo (Petr. 52.23) my-. . . . understand-. no-. money-. sell-1. ‘I do not sell my connaissseurship for any money’

According to Löfstedt (1956: 364) this feature owes its popularity partly to Greek influence. This assumption is supported by high frequencies in Cicero’s rhetorical but more precisely philosophical texts in which he followed the Greek masters closely; they used this device abundantly.

S F O PRO



1st proofs

The definite article in Indo-European 

Finally, there is a limited number of demonstratives in combination with genitives, a structure that was common in Greek, cf.: (51)

verum enim est illud Pacuvi true-... because be-3. that-... Pacuvius-. ‘what Pacuvius writes is true’ (Varro, R.R. 1.2.5)

Ille also occurs in noun phrases that already have a possessive element, as in: (52)

mos . . . illi fuit patri meo (Pl., Amph. 46) habit-. Dem.-.. be--3- father-.. my-.. ‘my father has the habit of . . .’

Instances are found in early Latin documents already: (53a)

placet ille meus mihi mendicus (Pl., Stich. 133) please-3. Dem.-.. my-.. I-. mendicant-. ‘that mendicant of mine is dear to me’

(53b)

servom . . ., qui illum sectari solet servant-. . . ., who-. Dem.-.. attend-. be.used.to-3. meum gnatum (Pliny, Ep. 486) my-.. son-. ‘that servant, [you know the one] who always attends my son’

(53c)

duravi interrogare illum interpretem meum (Petr. 41.2) endure-.-1. ask-. Dem.-.. informant-. my-.. ‘I ventured to ask that informant of mine’ (Löfstedt 1956: 370)

Examples with iste are attested as well, as the following example from Petronius illustrates, in which the pejorative connotation of iste is fully present: (54)

tuus . . . iste frater (9.4) your-.. Dem... brother-. ‘that brother of yours’ (example from Petersmann 1977: 136)

As said in Section 1.3 this pattern is attested in several (modern) Romance languages as well, cf.: Italian la mia macchina ‘my car’ vs. mio padre ‘my father’, Old French sa mulier (CdR 2576) ‘his wife’ vs. li nostre deu (CdR 2600) ‘our gods’, or with a demonstrative: cest mien anel (Yv. 2601) ‘this ring of mine’. Ille also combines with numerals, with or without nouns (see next section), as in: (55)

Tu in illis es decem sodalibus you-. in Dem... be-2- ten members-. ‘you belong to that fraternity of ten’

S F O PRO



1st proofs

 Brigitte L. M. Bauer

Ille may combine with prepositional phrases, but in these contexts the element is pronominal rather than adjectival, cf.: (56)

illi in Lydia (Varro, R.R. 3.17.4) Dem.-.. in Lydia-. ‘those in Lydia’

So far I have focussed on ille, but similar observations can be made for the other demonstrative elements as well as ipse. .. Predominant uses in individual texts In his De Observatione Ciborum, a long letter providing dietary suggestions written by Anthimus to Theodurus, king of the Franks (early 6th century), we count 132 instances of demonstratives, including ipse, which is most frequent, cf.: Table 3. Occurrence of demonstratives in Anthimus’ De Observatione Ciborum

Hic Iste Ille Is Ipse

Pronominal use

Adjectival use

Total

8 6 24 1 32

0 5 15 0 39

8 11 39 1 71

The low incidence of is fits the general pattern observed for the use of demonstratives in non-Classical Latin (Section 2.2). The continuing use of hic is noteworthy as well, albeit as pronominal rather than adjectival element. Finally, it is remarkable not only that ipse predominates, but that its adjectival use strongly prevails over that of ille. In this text as well, the sequence [demonstrative + noun] predominates, even for ipse (albeit to a lesser extent than for the other elements). The predominant case is the nominative singular (25 instances), followed by the nominative plural (10), and the accusative singular in prepositional phrases (5 instances). This case distribution corresponds to the general tendencies observed in Vulgar and Late Latin documents, where ille and ipse typically combine with subjects (for a discussion of the predominance of definite articles with subjects, see Meyer-Lübke 1899; Renzi 1976: 19–20; Vincent 1997: 161–2; 168–9). Finally, another important innovative characteristic of demonstrative use in Anthimus is the nominalization of adjectives, as in the following examples: (57)

Nam inter diversa commixta in prandio si unus cibus non congruus et crudior fuerit, illa alia bona dissipat et . . . (DOC 23) (underlying membra) ‘if among the various (ingredients) used in a meal one is not wholesome and uncooked, it ruins the other good ones’

S F O PRO



1st proofs

The definite article in Indo-European 

(58)

ita ut illa prima calda fundatur, et (DOC 67) ‘make sure that the first quantity of water is poured away . . .’

(59)

mela bene matura in arbore quae dulcia sunt, bona sunt, nam illa acida not sunt congrua (DOC 84) ‘apples that have ripened properly on the tree are good, but the sour ones are not healthy’

(60)

et mediolum illud tarde sentit (DOC 35) ‘and the yolk is heated later/gradually’

These instances of nominalization typically include the demonstrative ille ; there are no instances of this type with other demonstratives, with the exception of ista omnia simul trita (DOC 3) ‘ground these all together . . .’. In the Latin Bible translation Abel (1971) notes a clear predominance of hic, which however typically occurs in fixed expressions. He also points out that hic and iste no longer differ in terms of distal reference: hic as well as iste refer to [proximity] without distinguishing between TU and EGO and often translate the Greek demonstrative ουτος (TU demonstrative). Hic and iste typically do not render Greek εκεινος, which supports Abel’s assumption that hic and iste refer to proximity, while ille refers to distance. In contrast to hic, iste is especially frequent in direct speech and rarely is omitted. Although iste is still the least frequent element in the Latin Bible, its incidence is much higher than in earlier documents. These uses suggest that iste is on its way to oust hic. The element second in frequency is ille, which also typically is found in nonfixed expressions. Moreover, according to Abel’s findings it not only renders its Greek synonym εκεινος – like ipse – but often translates Greek definite articles as well. Significantly the other elements rarely do. Finally, ipse primarily renders its Greek synonym αυτος, αυτος ο and definite articles in combination with a proper noun. Yet as compared to other late documents where it predominates, ipse is not frequent in the Latin translation. Abel’s data therefore show that iste and hic still have a rather strong deictic value in the Latin Bible translation and that ipse is used mainly in its original sense. Ille, by contrast, lost some of its deictic value and was on its way to become a definite article. The individual examples from the Vulgata and Itala reveal patterns in use and an important difference in the development in Latin as opposed to that in Greek. In Greek one demonstrative element came to be used as a definite article and the original demonstrative ó came to be replaced in the demonstrative trichotomy. In Latin the entire demonstrative system formally collapsed and in some of the daughter languages the trichotomy gave way to a dichotomous system. In the Vulgata and elsewhere we find hic, iste, and ille in definite-article uses. The translation

S F O PRO



1st proofs

 Brigitte L. M. Bauer

from Greek into Latin is important because we see that – basically – all four forms render Greek ο and/or εκεινος. In Rönsch’s (1965) corpus of examples from the Latin Bible translation we find over 85 instances of hic, five of these combine with a proper name; mainly indeclinable names, cf.: (61a)

dabis veritatem huic Jacob, misericordiam huic Abrahae (Mich. 7.20) ‘you shall provide the truth to David, the mercy to Jacob’

(61b)

quia in te inventae sunt impietates huius Istrahel (Mich. 1.13) ‘. . . because the sins of Israel were found in you’

Moreover, there are instances in which hic combines with an adverb, cf.: (62)

ex hoc nunc ‘from this time on’ (Ps. 113.18)

Examples of this type – rare in Latin – are calques of the common Greek [article + adverb] combinations, cf.: (63)

απο του νυν ‘from now on’ (< adv. νυν ‘now’)

Hic is found with several regular nouns as well: (64a)

. . . huius saeculi (2 Cor. 7.10) ‘(the sorrow) of the world’

(64b)

. . . virum hunc cuius est zona haec (Act. 21.11) ‘(they shall bind) the man who owns this girdle’

Most widespread, however, are fixed expressions including the noun mundus ‘world’, of which more than seventy instances are found: (65)

in hunc mundum (e.g. Joh. 1.9) ‘to this world’

While noun phrases including ipse are close to those with hic in frequency (appr. 60 vs. 77 hic), over forty instances of ipse occur in combination with David, generally at the beginning of Psalms, e.g.: (66)

ipsi David (Ps. 7) ‘(a psalm) of David’

Another five instances combine with a proper noun: (67)

excitavit dominus satanan ipsi Salomoni ‘the lord stirred up an enemy unto Salomon’

S F O PRO



1st proofs

The definite article in Indo-European 

If combinations of ipse with an adjective are rare, there are several instances of common nouns in non-fixed expressions, cf.: (68)

ipsi principes ‘the leaders’

(3 Esr. 8.50)

It is remarkable that the Bible translation does not present many instances of ille. Yet among these only one includes a proper noun; in this instance ille has the value of “well-known”, cf.: (69)

Judas non ille Scarioth ‘Judas, not the (well-known) Iscarioth’

More importantly many instances include a common noun in non-fixed expressions, with or without adjective. These are the true fore-runners of definite NPs in Romance, cf.: (70a) (70b) (70c)

dicit ergo Petro illa ancilla (Joh. 18.17) ‘then the girl said to Peter’ in illa simulatione (Gal. 2.13) ‘in their dissimilation’ illi alii discipuli navigaverunt (Joh. 21.87) ‘the other disciples came in a boat’

The [demonstrative/article + noun] combination may have predicative value, cf.: (71)

quam mihi reddet dominus in illa die ille iudex iustus (2 Tim. 4.8) ‘which the Lord, the righteous judge, shall give me at that day’

It is noteworthy to point out the following instance, in which ille introduces a noun with relative clause construction: (72)

hic est discipulus ille qui testimonium perhibet (Joh. 21.24) ‘this is the disciple who testifies . . .’

In many Vulgar and Late Latin texts − translations and non-translations − one commonly finds [demonstrative ille + common noun + relative pronoun], most often in that sequence, cf.: (73)

O, si haberemus illos leones, quos ego hic inveni (Petr. 44) ‘If only we had the lions I found here’

The structure parallels the tendency of ille to combine with elements that are mentioned for the first time. Similarly, the combination may also refer to a segment that follows, cf.: (74)

verissima est enim illa sententia: Nam tu quod vitare possis, stultum admittere est (BV 4.25.17.35) ‘because this sentence is very true: what you can avoid, would be unwise to let happen’ (example from Muldowney 1937: 81)

S F O PRO



1st proofs

 Brigitte L. M. Bauer

In the Bible translation ille combines with numerals as well, a pattern we also observed for Ancient Greek, cf. the next example, where illis has nominalizing value: (75) et quia visus est Cephae et post hoc illis undecim (1 Cor. 15.5) ‘and that was seen by C. and after that by the eleven’

Finally, ille is attested in combination with a noun that already has a determiner, the possessive, cf.: (76) accipe filium tuum illum unicum (Gen. 22.2) ‘take you only son/take your son, your only son’

.

Demonstratives/articles in Latin: preliminary conclusion

This overview of the occurrence and the syntactic use of demonstratives/articles in Latin reveals a number of important patterns. There is an overall increase in the use of demonstratives: in later texts and in texts that represent spoken varieties we notice that demonstratives are used, which would not occur in documents closer to the classical norm. Yet the use of demonstratives in non-classical documents is not consistent: in some texts demonstrative X predominates, in others demonstrative Y does, without there being a special motivation for either choice. Despite this inconsistency we observe overall a remarkable persistence of the demonstrative hic, a high occurrence of ipse, the loss of is, and the relatively low but growing occurrence of iste and ille. In terms of actual use we notice that demonstratives/ articles typically combine with common nouns and proper names, conveying the notion of ‘well-known’. In contrast to Greek, their combining with elements other than nouns is very limited. The demonstrative/definite article in Latin therefore did not have the strong nominalizing function it had in Greek, except in some late texts where nominalized adjectives are not uncommon. The occurrence of demonstratives/articles with genitives overall is rare. Data from individual texts confirm the general tendencies. Anthimus’ text and the Bible translation show that among the demonstratives, ille in its articloid uses has most variety: it is ille that nominalizes adjectives in De Observatione ; it is ille in the Bible translation that is found especially with common nouns in non-fixed structures (as opposed to the very frequent hic). Consequently if ille is superseded by hic and ipse in terms of number, it seems to be the syntactic use of ille that foretells the later definite article. In my analysis of the emergence of adverbial mente constructions I came to similar conclusions: adverbial [adjective + mente] definitely was not the most common combination among the forerunners; yet it was the actual syntactic use of mente that may explain its survival (Bauer 2002). Latin demonstratives/articles do not combine with the variety of grammatical categories we observed for the Greek definite article. Moreover, in contrast to

S F O PRO



1st proofs

The definite article in Indo-European 

Greek, the slow emergence of the definite article in Latin/Romance coincides with the disintegration of the demonstrative system. Consequently if Latin had undergone Greek influence, the development of the definite article in Latin/Romance was still independent in terms of (a) demonstrative systems; (b) a rich variety of demonstratives/articles in Latin; (c) and actual uses: Greek had an exceptionally rich use, not found elsewhere in Indo-European languages, in the past or the present. Moreover, the Greek definite article had two important functions: defining and nominalizing, and it combined with a rich variety of grammatical categories. The Latin demonstrative/article does not share these characteristics. In addition, the rate of development was high in Greek. Examples from the various documents show that the emergence of the definite article in Latin/ Romance was a long-lasting development that started in the early period and came to completion only in the individual Romance languages. In terms of real time, Löfstedt on the basis of his analysis of a wide range of authors concludes that only during the 6th century in certain contexts one finds concrete and true instances of incipient definite articles; a true weakening of the value of ille takes place starting in the 8th century (1956: 373). Later analyses have supported a late rather than an early development (e.g. Selig 1992). The fact that several article-like uses of demonstratives in Latin show parallels with the definite article in Greek does not mean that the structure was borrowed. Greek influence would have been possible in Latin because of the influence of Greek civilization (e.g. the renaissance under Hadrian), slave education, and translation practices. Yet the definite article in Latin/Romance developed at its own pace, along its own lines. So if there indeed is some influence, there is no question of the definite article being borrowed from Greek. The systems in both languages were too diverse. What we can say, however, is that with the development of definite articles the category of definiteness becomes explicit in both languages; in Greek earlier and more extensively so than in Latin.

.

The definite article in other Indo-European languages

The definite article developed in many other Indo-European languages and the development that took place in Ancient Greek and Latin/Romance therefore is not an isolated phenomenon (cf. e.g. Heinrichs 1954 and Sauvageot 1929, for Germanic). Yet the scenario demonstrative > definite article may not always be accomplished. In Old Persian, certain phenomena suggest incipient uses of demonstratives as articles, such as the appositive combining of a demonstrative hya (tya) with a noun and adjective (Meillet 1915: 198). Subsequently the device disappeared and today only a suffix in certain circumstances may be used, marking a

S F O PRO



1st proofs

 Brigitte L. M. Bauer

degree of definiteness in direct objects (for a short description of this device, see Lyons 1999: 203–204). Moreover, some of the Indo-European languages seem to already have (had) a way to render a certain notion of definiteness before developing definite articles. Recent research involving various linguistic branches (e.g. Bader 1993, Abraham 1997, Philippi 1997) has shown that in languages without definite articles case use in combination with aspect may affect the degree of definiteness. In others two types of adjectival inflection reflect that characteristic. In yet other Indo-European languages we find derivational processes creating nouns with definite characteristics. .

Adjectival declension

In Germanic, which has definite articles today, the so-called weak adjectives in origin were definite. In the weak or nominal declension an element -n- followed the adjectival stem, adding defining and nominalizing value. The paradigm survives in modern German (e.g. gute Mann, guten Mannes; for the paradigms in Gothic, see Braune & Ebbinghaus 1961: 77–84). The so-called strong declension, which is later (e.g. Prokosch 1939: 261), includes the adjectival stem in combination with a pronominal element; with time the composite form has fully unified, as reflected in modern German (guter Mann, gute Männer). The original difference in definiteness is still manifest in today’s occurrence of weak adjectives in definite contexts (e.g. des guten Knaben ‘of the good boy’ vs. guter Wein ‘good wine’). A similar process is observed in Balto-Slavic, where in some of the languages the elements have become compound forms as well. In Lithuanian, for example, adjective-noun constructions either include a “short” adjective – the noun phrase is indefinite – or an “extended” adjective; the noun phrase then is definite. The extension of the adjective is the result of a suffix -ij- – possibly related to the Latin anaphor is – which has been added to the adjectival stem (Prokosch 1939: 260–261): (77)

Lith. gẽ ras žmõgus ‘a good husband’ geràsis žmõgus ‘the good husband’ gerà žmonà ‘a good wife’ geróji žmonà ‘the good wife’

The same pattern is found in Old Church Slavonic, where in addition the individual elements of the composite forms have kept their own declension, cf.: (78)

Lith. gẽ ras žmõgus ‘a good husband’ gerà žmonà ‘a good wife’

geràsis žmõgus ‘the good husband’ geróji žmonà ‘the good wife’

(79)

OCS dobrъ rabъ dobra žena

dobrъjъ rabъ dobraja žena

(80)

OCS: dobra-jego raba ‘of the good servant’ good-.-.-. servant-.

‘a good servant’ ‘a good woman’

‘the good servant’ ‘the good woman’

S F O PRO



1st proofs

The definite article in Indo-European 

(81)

dobru-jemu rabu good-.-.-. servant-. (examples from Prokosch 1939)

‘to the good servant’

Consequently, Balto-Slavic and Germanic include processes whereby adjectives are combined with a defining or nominalizing morphological element (cf. Prokosch 1939: 259–261). This device existed before definite articles developed. .

Aspect and case

The adjectival process discussed above is fully integrated in the language system but limited to a specific category. The same observation can be made about other processes in other languages. A group of Germanists, among them Leiss (1992; 2000) and Abraham (1997), have observed a possible connection between aspect, case use, and definiteness: “there exists an intricate interplay between the referential status of object NP and Aktionsart, or aspect, of the verb” (Abraham 1997: 34). Abraham’s observation is based on his analysis of case use in relation to the aspectual characteristics of the verb in early stages of German: in these stages perfective verbs could govern an accusative, which triggered a definite interpretation of the direct object, or a genitive, which then referred to an indefinite direct object. The distinction is not valid for imperfective verb stems, which simply govern an accusative. In line with the exclusive use of indefinite genitive with perfective verb forms, the genitive typically combines with verbs as well that have “perfectivizing verbal prefixes, affixoids, and adverbials” (Abraham 1997: 35). Conversely, verbs that typically govern a genitive as a default case (e.g. verbs referring to eating, drinking, and so forth), may govern an accusative in the early stages of German, if the context triggers definite reading or the verb had a prefix conveying perfective aspect or Aktionsart. Etan in early OHG, for example, governs a genitive, unless it has the meaning of ‘devouring’; it then governs an accusative. Similarly, whereas etan governs the genitive, perfective fraiten – with a perfective prefix – governs an accusative. Consequently these verbs govern a genitive as default case, but for perfective uses there is a choice between the accusative [+definite] and the genitive [−definite] (Abraham 1997: 35–8). Similarly, Philippi (1997) found a correlation in several languages between the accusative direct object and its definite referentiality on the one hand and on the other hand the non-accusative direct object and its indefinite referentiality in verbal constructions with perfective or perfective-like value. Finnish, for example, distinguishes between resultative verbs (event limited in time), which govern an accusative, and non-resultative verbs, which refer to a non-completed action and govern a partitive (Philippi 1997: 74). For Germanic she points out a correlation between genitive and non-countability vs. accusative and countability and the

S F O PRO



1st proofs

 Brigitte L. M. Bauer

perfective interpretation of the verb phrase. Moreover, imperfective forms of eison govern a genitive meaning ‘ask for’, whereas perfective forms govern an accusative, meaning ‘find out’. Consequently there seems to be a range of phenomena in early Germanic languages that involve the correlation between aspect or Aktionsart, choice of case, and the definite or non-definite interpretation of the verbal complement. Arguing that the choice of case of the direct object could convey the notion of definiteness or countability and pointing out that the genitive plays an important role in this process, Philippi identifies the loss of this case as the trigger of the simultaneous development of definite articles. Abraham (1997) points out a phenomenon in Russian – another language without definite articles – that shows similarities with the Germanic instances referred to in the previous paragraphs. In the following two examples the same transitive verb governs an accusative and a genitive respectively: (82)

on he-. on he-.

kolol drova ‘he split wood’ split-. wood-. raskolol drova ‘he split the wood’ split-.. wood-.

The accusative in the first instance combines with an imperfective verb and conveys an indefinite direct object. The second example includes an accusative as well, but because of the perfective form, the direct object has definite value (Abraham 1997: 42–45). In the next two examples, the same perfective verbal form governs either a genitive direct object – conveying indefiniteness – or an accusative, with definite value: (83)

on prinës he-. bring-. on prinës he bring-. (Abraham 1997: 44)

papiros ‘he brought cigarettes’ cigarettes-. papirosy ‘he brought the cigarettes’ cigarettes-.

Further research is needed to evaluate the age, possible origin, and the spread of the construction. I will therefore at the moment not go into further details of these phenomena. Independently from the results pertaining to the scope and origin of these case uses, it is however clear that languages without definite article may have devices to convey (a certain degree of) definiteness. Questions to ask are: (a) To what extent is the degree of definiteness as reflected in these verb phrases purely an extension of aspectual value? Or could it also be an extension of case syntax, especially of the genitive? and (b) Is it truly definiteness that is conveyed or rather does the phenomenon reflect a certain degree of definiteness or perhaps something similar

S F O PRO



1st proofs

The definite article in Indo-European 

to definiteness, which we render in English – a language without aspect as a morphological process – by a definite article? .

Case use

It has been argued that phenomena and their development as described in the previous section were absent in Latin: “another alternative would be to see the rise of articles as related to the loss of case as a means of signalling distinctions of count/ mass and definiteness. This view is plausibly developed for Germanic . . . but there is, to my knowledge, no evidence for an interaction of case and aspect in Latin . . . as a precursor to the rise of the article in Germanic” (Vincent 1997: 168). I am not aware of any research that would indeed point out such a correlation, but, to my knowledge, nobody has examined it either (as seems to be the case in Germanic before Leiss [1992; 2000], Abraham [e.g. 1997] and Philippi [1997]). It is interesting to note, however, that early Latin according to some may distinguish between genitive (indefinite) and accusative (definite) direct objects, cf.: (84a) vs. (84b)

aquae paulatim addito (Cato, R.R. 74) ‘add water gradually’ aquam ter mutato (Cato, R.R. 76.2) ‘change the water three times’

Yet this does not mean that the alternation of genitive/accusative systematically reflected definite and indefinite value; it means that in certain contexts the genitive possibly may have been used referring to an inherently indefinite direct object, reflecting partitive function. So far, there is no reason to assume there is a paradigmatic choice. Meillet & Vendryes 1924 argue that the partitive and the adnominal genitive were the original uses of genitives in Proto-Indo-European (Meillet & Vendryes 1924: 508–10). The partitive genitive could occur in a wide array of syntactic functions, in subjects function, for example (85) or direct object function (86), cf. (85)

επιπτον εκατερων (Xen., Hell. 4.2.20) fall-3.. each-of-two-.. ‘on both sides men died’

(86a)

πάσσε δ’αλος (Hom., Il. I.214) sprinkle-3.. salt-. ‘he sprinkles some salt’

(86b)

τυρων αινυμενους (Hom., Od. ι. 225) cheese-.. take-.. ‘taking cheeses’

S F O PRO



1st proofs

 Brigitte L. M. Bauer

(86c)

Αδρηστοιο δ’εγημε θυγατρων (Hom., Il. Ξ 121) Adrestes-. marry-3.. daughters-. ‘he took wife among the daughters of Adrestes’

Similarly, in adverbial functions the genitive conveys indefinite value, cf. (87a) vs. (87b): (87a)

νυκτος ‘at some point during the night’ (Hom., Od. ν278)

(87b)

νυκτα ‘during the [entire] night’ (Her., 1.181)

Compare also: (88a)

του λοιπου ‘the future-. > ‘at some point in the future’

(88b)

το

λοιπον ‘the future-. > ‘in the time to come’

The partitive indefinite direct object is referred with the genitive in (89a) vs. (89b): (89a)

ξυνετριβη της κεϕαλης (Arist., Pax. 71) crush-3.. Art.-.. head-.

(89b)

συντριβομεθα τας κεϕαλας (Lysias 3.18) crush-1.. Art-.. heads-. (examples from Meillet & Vendryes 1924: 509).

With time the genitive partitive has disappeared in favor of the accusative that came to mark any direct object. Residues of the early situation are found in the genitive use with verbs of thinking, verbs of perception (‘listen’, ‘see’, . . .), cf.: (90a)

αυδης κλυειν voice-. hear/listen [partitive value]

(90b)

αυδην κλυειν voice-. hear

[non-partitive]

In Latin the partitive genitive could function as subject or object as well; as object it originally combined with verbs of eating and drinking (e.g. potiri), subsequently with verbs of forgetting, remembering, desire, and from there with others as well (e.g. egere, indigere, carere, complere, abundare, levare, regnare, and so forth), cf.: (91)

regnavit populorum (Hor., Carm. 3.30.12) rule-3- peoples-. ‘he ruled of the people’

Instances of this type of genitive are found not only in Classical Latin, but in early, Vulgar and Late Latin texts as well. In Vulgar and Late Latin a prepositional phrase (de + ablative) assumed this function. The uses referred to in this section do not seem to involve aspect although Latin did have aspect. To address this problem systematic analysis is needed of

S F O PRO



1st proofs

The definite article in Indo-European 

devices in Latin other than demonstratives that mark definiteness. It should include analysis of genitive direct objects with a variety of verbs, their actual occurrence and their (inherent) value, both from a synchronic and diachronic perspective (Bauer In Preparation b). .

Derivation

The question whether case variation in combination with aspect truly reflects definiteness or rather something that we today render by using a definite article may also be asked about the following derivative process that conveys individualizing value as pointed out by Bader (1993). Bader argues that all (early) Indo-European languages include particles that are added to adjectives or nominal elements to individualize or personify: ‘the . . . one’. The process does not express deixis, but rather appurtenance, which is inherently definite: the genitive of the noun expressing appurtenance is turned into an adjective, which subsequently is turned into a noun again. Bader has identified several of such derivational suffixes, cf.: (92)

PIE *-de : Gk. Πριαμíδης (< Πρíαμος) ‘son of Priam, Trojan’;

(93)

PIE *-le : Hitt. auriyala (*auri-o-lo- ‘the (one) of the guarding > the guard’ (< auri ‘watch’); humilis (< humus) ‘the (one) from the earth’; Anthroponyms: La. Capri-lius (< caper, capris ‘goat’); Names of feast/months: Aprilis ‘the (month) of the boar (aper, apris)’; Palilia ‘the (feast) of Pales’;

La.

(94)

PIE *-se : the suffix *se is found in anthroponyms (e.g. Numa-sius [< Numa]) and toponyms, cf. Αρνισσα (< αρνος ‘lamb’) ‘the (one) of the lambs’;

(95)

PIE *-ne : the suffix *-ne is attested in Gaulish Epo-na ‘the (goddess) of the horse’, in names of persons, e.g. Μυρíνα (< μυρον) ‘the (one) of unguent’, in toponyms as in Αιγινα ‘the (one) of the goat/the (one) of the waves’ (< αιξ ‘goat’/αιγες ‘waves’), in common nouns as in La. dominus ‘the (one) of the house’. Cognate of adjectival –en in Germanic, the suffix is also found in a productive derivational process in Latin, for example, cf.: Cato (nom.) > Catonis (gen., . . .) ‘the sly one’ (< catus ‘intelligent, wise, sly . . .’), Rufo (nom.) > Rufonis (gen., . . . ) ‘the red head’ (< rufus ‘red’) (Overview based on data from Bader 1993: 22–7).

The examples above show that the derivational process was wide spread and that many of these suffixes were productive in names, proper names, place names and names of feasts, expressing the notion ‘the one of ’. .

Definiteness in Indo-European: preliminary conclusion

S F O PRO

The data gathered in the previous pages show that the absence of definite articles in Indo-European does not imply that there was no marking of (a certain



1st proofs

 Brigitte L. M. Bauer

degree of) definiteness. Most wide-spread in terms of geography, frequency, and age are derivational processes involving elements such as the ones Bader has identified that individualize morphological bases. It is legitimate to conclude that these indeed are old. It is striking that early instances of definiteness are related to nominalization, which parallels the importance of nominalization in the uses of the definite article in Ancient Greek. In addition, in many languages we find devices other than explicit definite articles that mark a degree of definiteness, as in Russian or early Germanic, suggesting that for Indo-European definite articles are not the sole device to convey definiteness and that the concept of definiteness may include different cross-linguistic and diachronic aspects.

.

The definite article in Indo-European: Emergence of a new category?

The definite article in Romance is the result of a long-lasting development the first attestations of which are found in early (spoken) Latin; the development came to completion in the individual Romance languages. On the basis of occurrences, uses, and related changes in the demonstrative system, we conclude that despite parallels with Greek article use, the change in Latin developed independently, at its own pace, and following its own functional and syntactic pathways. The definite article in Latin/Romance therefore is not a borrowing from Greek, even if some influence from Greek is not excluded. Yet in both languages the category of definiteness became explicit when it came to be marked in definite articles. Formulated in this way, the conclusion brings up the next question to address: did definiteness exist in Indo-European before there were definite articles? The distinction between weak and strong (short/long) adjectives in Germanic and Slavic has (had) defining function. Similarly, there is evidence of a correlation between case use on direct objects and aspect conveying definiteness as well. These important findings also show that if there were devices in these languages to express a certain degree of definiteness, it was limited to specific categories (aspect, adjectives) and contexts (adjective + noun combinations and direct objects). The connection between aspect, object case, and definiteness has not been attested for Latin despite the fact that Latin had aspect and Aktionsart. Further analysis (Bauer In preparation b) confirms this observation on the basis of extensive data. Moreover, it is important to note from this perspective the high percentage – if not predominance – of demonstratives/definite articles with nominative nouns in Latin/Romance (own findings, but also e.g. Meyer-Lübke 1899; Wolterstorff 1917; Trager 1932; Abel 1970; Rosén 1974; Vincent 1997); the article typically did not combine with objects even in early Romance languages. If there was no correlation between case, aspect, and definiteness, Latin did, however, have early

S F O PRO



1st proofs

The definite article in Indo-European 

derivational processes that had individualizing function. It is clear that further research into definiteness in Latin is called for to evaluate these and possibly other devices (Bauer In preparation b). On the basis of comparative and historical evidence, however, we can now already say in tentative conclusion that definiteness is a cross-linguistic phenomenon that existed much earlier in Indo-European than the definite article. The definite article is one device among many and it is spreading at the expense of other markers, which are limited to specific categories, such as aspect, adjectives, or syntactic function. In addition, comparative historical analysis strongly suggests that definiteness as marked by definite articles is the result of a development not only of form but also of concept. Languages that do not have definite articles at an early stage may have processes that mark a certain degree of definiteness (such as individuation or countability) and that may be considered forerunners of definiteness as marked by definite articles. This observation brings me back to my initial remarks about Benveniste’s characterization of the development of the definite article as an innovative change, creating a new grammatical category. Was Benveniste wrong? I do not think so if you define ‘new grammatical category’ in terms of a specific form in a unique combination with a new specific function. The definite article incorporates that new combination. But in my view the notion of definiteness was not new in IndoEuropean, even if it developed with time. In other words it existed before the emergence of definite articles. With definite articles definiteness becomes more explicit in semantic and formal terms and is characterized by a wider application, involving as it does a wider group of categories.

References Abel, Fritz. 1971. L’adjectif démonstratif dans la langue de la Bible latine. Étude sur la formation des systèmes déictiques et de l’article défini des langues romanes. Beihefte zur Zeitschrift für romanische Philologie 125. Tübingen: Niemeyer. Abraham, Werner. 1997. The interdependence of case, aspect, and referentiality in the history of German: The case of the verbal genitive. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 29–61. Cambridge: Cambridge University Press. Aebischer, Paul. 1948. Contribution à la proto-histoire des articles ille et ipse dans les langues romanes. Cultura Neolatina 8: 181–203. Bader, Françoise. 1993. Les génitifs-adjectifs déterminés et le problème de l’article: Comparaison typologique entre l’étrusque et les langues indo-européennes. In Indogermanica et Italica. Festschrift für Helmut Rix zum 65. Geburtstag, Gerhard Meiser (ed.), 12–45. Innsbruck: Institut für Sprachwissenschaft. Bauer, Brigitte L.M. 2002. The adverbial formation in mente in Vulgar and Late Latin. A problem in grammaticalization. In Latin vulgaire – latin tardif VI. Actes du VIe colloque international

S F O PRO



1st proofs

 Brigitte L. M. Bauer sur le latin vulgaire et tardif, Heikki Solin, Martti Leiwo & Hilla Hallo-aho (eds), 439–57. Hildesheim: Olms. Bauer, Brigitte L.M. In preparation a. Definite articles and possessives in NPs in Romance. Bauer, Brigitte L.M. In preparation b. Definiteness in Latin. Benveniste, Emile. 1974. Les transformations des catégories linguistiques. Problèmes de linguistique générale, 126–36. Paris: Gallimard. Bouvier, E. 1972. Le démonstratif latin ille et la formation de l’article défini des langues romanes. Cahiers de lexicologie 21: 75–86. Braune, Wilhelm & Ebbinghaus, Ernst A. 1961. Gotische Grammatik. 16th ed. Tübingen: Niemeyer. Brugmann, Karl. 1911. Vergleichende Laut-, Stammbildungs- und Flexionslehre nebst Lehre vom Gebrauch der Wortformen der indogermanischen Sprachen. II. 2. Lehre von den Wortformen und ihrem Gebrauch. Strassburg: Trübner. Epstein, Richard. 1993. The definite article: Early stages of development. In Historical linguistics 1991. Selected papers from the 10th international conference on historical linguistics, Jaap van Marle (ed), 111–34. Amsterdam: John Benjamins. Ernout, Alfred & Meillet, Antoine. 1959. Dictionnaire étymologique de la langue latine. Histoire des mots. Paris: Klincksieck. Ernout, Alfred & Thomas, François. 1964. Syntaxe latine, 2nd edition. Paris: Klincksieck. Faingold, Eduardo D. 1996. Demonstrative pronouns and the definite article in Latin and the Romance languages. Papiere zur Linguistik 54: 67–82. Fischer, Anton. 1908. Die Stellung der Demonstrativpronomina bei lateinischen Prosaikern. Tübingen: Heckhauer. Greenberg, Joseph H., Ferguson, Charles A. & Moravcsik, Edith A. (eds). 1978. Universals of Human Language. Stanford CA: Stanford University Press. Greenberg, Joseph H. 1978. How does a language acquire gender markers? In Joseph H. Greenberg, Charles A. Ferguson & Edith A. Moravcsik (eds), 47–82. Heinrichs, Heinrich M. 1954. Studien zum bestimmten Artikel in den germanischen Sprachen. Giessen: Schmitt. Jones, Michael. 1993. Sardinian Syntax. London: Routledge. Lapesa, Rafael. 1961. Del demonstrativo al articulo. Nueva revista de filologiía hispánica 15: 23–44. Leiss, Elisabeth. 1992. Die Verbalkategorien des Deutschen. Berlin: Walter de Gruyter. Leiss, Elisabeth. 2000. Artikel und Aspekt. Die grammatischen Muster von Definitheit. Berlin: Walter de Gruyter. Leumann, J., Hofmann, J.B. & Szantyr, Anton. 1965. Lateinische Syntax und Stilistik. München: Beck. Löfstedt, Bengt. 1961. Studien über die Sprache der langobardischen Gesetze. Beiträge zur frühmittelalterlichen Latinität. Stockholm: Almquist and Wiksell. Löfstedt, Einar. 1936. Vermischte Studien zur lateinischen Sprachkunde und Syntax. Lund: Gleerup. Löfstedt, Einar. 1956. Syntactica. Studien und Beiträge zur historischen Syntax des Lateins. I. Lund: Gleerup. Lyons, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press. Lyons, Christopher. 2000. The origins of definiteness marking. In Historical linguistics 1995. Selected papers from the 12th international conference on historical linguistics, John C. Smith & Della Bentley (eds), 223–241. Amsterdam: John Benjamins. Meader, Clarence & Wölfflin, Eduard. 1900–1901. Zur Geschichte der Pronomina demonstrativa. Archiv für lateinische Lexikographie und Literatur 23: 434–448. Meillet, Antoine. 1915. Grammaire du vieux perse. Paris: Guilmoto. Meillet, Antoine & Vendryes, Jean. 1924. Traité de grammaire comparée des langues classiques. Paris: Champion.

S F O PRO



1st proofs

The definite article in Indo-European  Meyer-Lübke, Wilhelm. 1899. Grammatik der romanischen Sprachen. III. Romanische Syntax. Leipzig: Reisland. Muldowney, Mary S. 1937. Word Order in the Works of St. Augustine. Washington DC: The Catholic University of America. Nocentini, Alberto. 1990. L’uso dei dimonstrativi nella Peregrinatio Egeriae e la genesi dell’ articolo romanz. Atti del convegnop internazionale sulla Peregrinatio Egeriae, 137–158. Arezzo: Accademia Petrarca di Lettere Arti e Scienze. Orlandini, Anna. 1982. Wesen und Entwicklung des Artikels vom Lateinischen zu den romanischen Sprachen. Indogermanische Forschungen 86: 223–247. Palmer, Leonard R. 1954. The Latin language. London: Bristol Classical Press. Petersmann, Hubert. 1977. Petrons urbane Sprache. Wien: Österreichische Akademie der Wissenschaften. Philippi, Julia. 1997. The rise of the article in the Germanic languages. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 63–93. Cambridge: Cambridge University Press. Prokosch, Edward. 1939. A Comparative Germanic Grammar. Philadelphia PA: Linguistic Society of America. Renzi, Lorenzo. 1976. Grammatica e storia dell’articolo italiano. Studi di grammatical italiana. 5: 5–42. Rohlfs, Gerhard. 1968. Grammatica storica della lingua italiana e dei suoi dialetti. II. Morfologia. Turin: Einaudi. Rohlfs, Gerhard. 1969. Grammatica storica della lingua italiana e dei suoi dialetti. III. Sintassi e formazione delle parole. Turin: Einaudi. Rönsch, Hermann. 1965. [1869]. Itala und Vulgata. München: Hueber. Rosén, Hannah. 1994. The definite article in the making, nominal constituent order, and related phenomena. In: Linguistic Studies on Latin. Selected papers from the 6th international colloquium on Latin linguistics, József Herman (ed.), 131–150. Amsterdam: John Benjamins. Salonius, Aarne H. 1920. Vitae Patrum. Kritische Untersuchungen über Text, Syntax und Wortschatz der spatlateinischen Vitae Patrum. Lund: Gleerup. Sandfeld, Kristian. 1930. Linguistique balkanique. Problèmes et résultats. Paris: Champion. Sauvageot, Aurélien. 1929. L’emploi de l’article en gotique. Paris: Champion. Schwyzer, Eduard & Debrunner, Albert. 1950. Griechische Grammatik, Bd. 2: Syntax und Syntaktische Stylistik. München: Beck. Selig, Maria. 1992. Die Entwicklung der Nominaldeterminanten im Spätlatein. Romanischer Sprachwandel und lateinische Schriftlichkeit. Tübingen: Narr. Smyth, Herbert W. 1956. Greek Grammar. Cambridge MA: Harvard University Press. Trager, George. 1932. The use of the Latin demonstratives (especially ILLE and IPSE) up to 600 AD as the source of the Romance article. New York NY: Publications of the Institute of French Studies. Vincent, Nigel. 1997. The emergence of the D-System in Romance. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 149–169. Cambridge: Cambridge University Press. Wanner, Dieter. 1987. Romance Clitic Pronouns. Berlin: Mouton de Gruyter. Werner, Heinz.1998. Artikelentstehung und Verlust der Kasusflexion in der Romania im Licht eines modifizierten Word-and-Paradigm Modells. Zeitschrift für romanische Philologie 114: 381–413. Wolterstorff, Gottfried. 1919. Artikelbedeutung von ille bei Apuleius. Glotta 8: 197–226. Wolterstorff, Gottfried. 1920. Entwickelung von ille zum bestimmten Artikel. Glotta 10: 62–93.

S F O PRO



1st proofs

S F O PRO



1st proofs

‘No’ changes: On the history of German indefinite determiners in the scope of negation Agnes Jäger Johann-Wolfgang-Goethe-Universität, Frankfurt, Germany

This paper investigates the evolution of nominal determination of a specific kind, viz. indefinite determination in the scope of negation. Four basic syntactic patterns of indefinite nominal determination in the scope of negation are distinguished. The changes within the system of indefinite determination in the history of German with respect to these four patterns are described on the basis of their distribution in a corpus of several Old and Middle High German texts. More specifically, the development and distribution of dehein/kein is investigated. While the original n-word determiner nehein (‘no’) and the second NPI determiner einig (‘any’) were virtually lost, dehein/kein changed from a weak NPI (negative polarity item) comparable to any and licensed in various non-affirmative contexts into an n-word.

. A typology of indefinite determination under negation For the purpose of this paper1 I will distinguish four basic syntactic patterns concerning the determiner of indefinite, non-specific NPs2 in the scope of negation, i.e. in clauses including sentential negation. These four patterns A through D are exemplified in (1) with the help of English.

. I thank the audience at the DGfS annual conference 2005 and the two reviewers for their comments. . In this paper, I will use the label NP. However, my approach is entirely compatible with a DP analysis. Some of the phrases under investigation, viz. pattern A, are indeed bare NPs. The term NP will be used here as a general label to refer to these as well as those phrases including determiners.

S F O PRO



1st proofs

 Agnes Jäger

(1)

Syntactic patterns of indefinite, non-specific NPs in the scope of negation: A bare NP/no Det

B indef. Det that is also used in affirmative contexts (e.g. Engl. a)

C special Det that is used in nonaffirmative contexts (e.g. Engl. any)

D special Det that is only used in negative contexts (e.g. Engl. no)

[–non-aff, –neg]

My neighbours have got dogs.

My neighbours have got a dog.

—

—

[+non-aff, –neg] (weak NPI contexts)

If my neighbours had dogs … Do your neighbours have dogs?

If my neighbours had a dog … Do your neighbours have a dog?

If my neighbours had any pets … Do your neighbours have any pets?

—

[+non-aff, +neg] (strong NPI contexts)

My neighbours don't have dogs.

My neighbours don't have a dog.

My neighbours don't have any pets.

My neighbours have no pets.

The patterns are defined in terms of the kind of determiner used and the type of contexts that the respective NP occurs in. As can be seen in the leftmost column of Table (1), I differentiate three basic types of contexts on the basis of the features [±non-affirmative] and [±negative], where, according to underspecification theory, the positive value corresponds to the marked option. Note that there is a feature hierarchy in so far as [+neg] presupposes [+non-aff] but not vice versa. In order to capture the relevant typological patterns and the changes throughout the history of German concerning indefinite determination in the scope of negation, it is not sufficient to distinguish between negative and non-negative clauses or even simply negative and positive clauses, which is not a complete disjunction at any rate, as which Danielsen (1968) treats it. It is crucial to differentiate the group of [+non-aff, −neg] contexts including conditionals, questions, the standard of comparison, the syntactic contexts of ‘without’, ‘before’ etc. and clauses dependent on negated or on so-called adversative matrix predicates such as ‘to forbid’, ‘to deny’, ‘to fear’, ‘to doubt’.3 In the terminology of van der Wouden (1997) these contexts are called weak NPI contexts, whereas [+non-aff, +neg] contexts, that is negated clauses, are referred to as strong NPI contexts. The term NPI (negative polarity item) generally refers to elements that are sensitive to the

S F O PRO

. Ladusaw (1979) identified the semantic property of downward-entailment as the common property of the contexts here referred to as [+non-aff].



1st proofs

German indefinite determiners in the scope of negation 

polarity of a sentence, the prototypical example being any. These items are typically licensed by negation, but some of them may also appear in other non-affirmative contexts such as the ones just listed. Accordingly, NPIs may be subdivided into strong and weak NPIs. As is illustrated in Table (1), indefinite NPs in the scope of negation can firstly appear without a determiner and thus take the form of a bare NP (pattern A). This type of indefinite NP, such as English dogs is found in all three types of contexts. Secondly, the same indefinite determiner that is also used in all other types of contexts may be used in the scope of negation, consider English a as in a dog (pattern B). Thirdly, a special determiner may be used that is restricted to [+non-aff, ±neg] contexts (pattern C) and can thus be classed as a (weak) NPI. This pattern is exemplified by English any as in any pets. Finally, a special determiner may be used that is restricted to [+non-aff, +neg] contexts (pattern D), e.g. English no as in no pets.4 The English determiner no can be classed as an n-word (negative word) in the terminology of Laka (1990). N-words are indefinite pronouns such as ‘nothing’ or ‘nobody’, or adverbs such as ‘never’ or ‘nowhere’ that are often morphologically marked for negation, take part in Negative Concord (NC) in many languages and give rise to a negative interpretation for instance in one-word answers, which distinguishes them from NPIs. In English, all four patterns of indefinite nominal determination occur in the scope of negation. With patterns A through C, negation is marked by some other constituent in the clause, whereas a pattern D determiner may either be the only negative marker in the clause or co-occur with others in a concordant reading in NC languages. In the remainder of this paper, the situation and historic changes in German regarding indefinite determination in the scope of negation will be discussed against the background of this typology. A brief description of the situation in Modern German will be contrasted with accounts of indefinite determination in the scope of negation in Old and Middle High German, discussing the proportions and syntactic distribution of each pattern. Particular emphasis is put on the determiner dehein(ig)/kein that (i) occurs in quite a number of different syntactic contexts and (ii) shows a significant shift in its distribution. These observations are explained by the assumption that this determiner developed from a weak NPI into an n-word, replacing the original n-word determiner. An important change in the diachrony of indefinite determination in German thus consists in the development of the determiner meaning ‘no’.

S F O PRO

. My approach differs fundamentally from Haspelmath’s (1997: 199) who explicitly opts not to distiguish ‘special’/NPI indefinites and n-words, i.e. my patterns C and D.



1st proofs

 Agnes Jäger

. .

Modern German Standard pattern D – negative determiner kein

In Modern German, sentential negation is generally expressed by the verbindependent negation particle nicht (‘not’). If a non-specific, indefinite full NP occurs in a negated clause, negation is identified instead by the special determiner kein (‘no’) (unless there already is a preceding n-word).5 ●

(2)

pattern D – negative determiner kein a. Erwin hat einen Dackel. Erwin has a dachshund b. Erwin hat keinen Dackel. Erwin has no dachshund.

[−neg] [+neg]

Kein is restricted to [+non-aff, +neg] contexts and is thus an instance of pattern D as defined above. This is the standard pattern in Modern German. It holds for NPs that would be accompanied by the determiner ein (‘a’) in [−neg] contexts cf. (2a), as well as for those that would appear as bare NPs, e.g. mass nouns and indefinite plurals cf. (3), NPs incl. andere ‘other’ and solche ‘such’ cf. (4), and certain nounverb collocations cf. (5): (3)

a.

b.

(4)

a.

b.

(5)

a.

Erwin mag {Fencheltee/Hunde}. Erwin likes fennel tea/dogs ‘Erwin likes fennel tea/dogs’ Erwin mag {keinen Fencheltee/keine Hunde}. Erwin likes no fennel tea/no dogs ‘Erwin does not like fennel tea/dogs’

[−neg]

[+neg]

Erwin hat auch andere Haustiere. Erwin has also other pets ‘Erwin also has other pets’ Erwin hat auch keine anderen Haustiere. Erwin has also no other pets ‘Erwin doesn’t have any other pets, either’

[−neg]

Erwin hat {Hunger/Angst} Erwin has hunger/fear ‘Erwin is hungry/afraid’

[−neg]

[+neg]

. By contrast, indefinite NPs with a specific reading are not accompanied by the determiner kein. Instead, the standard indefinite determiner is used and negation is marked by nicht, e.g.: Erwin mag EINEN Dackel nicht. ‘There is one dachshund that Erwin does not like’

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

b.

Erwin hat {keinen Hunger/keine Angst} Erwin has no hunger/no fear ‘Erwin isn’t hungry/afraid’

[+neg]

The n-word kein is the only neg-marker in the clause in all these cases in Modern Standard German, which is a non-NC language. In German NC dialects such as Bavarian, additional neg-markers are possible:6 (6)

wai’e koa Geid ned kod han because-I no money N had have ‘Because I have had no money’

As far as the semantics of kein is concerned, it is worth noting two points. First, kein-NPs identify sentential negation, not constituent negation as is sometimes claimed. Thus the meanings of (2b) and (3b) can be paraphrased as in (2') and (3'): (2')

‘It is not the case that Erwin has a dachshund’ (≠ ‘Erwin has a non-dachshund’)

(3')

‘It is not the case that Erwin likes {fennel tea/dogs}’ (≠ Erwin likes non-fennel-tea/non-dogs)

Second, instead of analysing kein as an inherently negative quantifier of the form λP λQ ¬ ∃x (Px ∧ Qx) or λP λQ ∀x (Px → ¬ Qx), I take kein to be semantically equivalent to ein, but licensed by (in Modern German non-overt) sentential negation. Evidence for this position comes from the fact that in the interpretation of a clause containing kein, supposedly a portmanteau form of ¬ and ein, various kinds of operators may in fact intervene between ¬ and ein. This phenomenon is known as Neg-Split (Jacobs 1991, Penka/Stechow 2001): (7)

Du musst keinen Anzug tragen. you must no suit wear (i) ‘It is not necessary that you wear a suit’ (ii) ‘It is necessary that you wear no suit’ ≠ (iii) ‘There is no suit for which it is necessary that you wear it’

In line with Penka/Stechow (2001: 273) and others, I therefore assume that kein does not contribute the negation to the clause, but is licensed by and identifies the semantic negation that is positioned higher up in the clause and may remain nonovert as in Modern German. What is important for the purpose of this paper is that kein is not morphologically negative, either. Although purely synchronically one may arrive at an analysis .

Example from Weiß (1998: 186).

S F O PRO



1st proofs

 Agnes Jäger

such as Eroms (1993) of kein as a combination of the indefinite article ein and the negation element k, in the diachronic perspective, however, it will become clear that k did not originate as a neg-morpheme: the predecessor of kein was the morphologically non-negative NPI determiner dehein. .

Other patterns

Besides pattern D, patterns A and B are also occasionally used in the scope of negation in Modern German in combination with the negative particle nicht. Pattern A, a bare NP, is used instead of pattern D with some noun-verb collocations (cases of noun stripping) cf. (8),7 optionally with predicative NPs cf. (9). ●

(8)

(9)

pattern A – bare NP/no determiner: Jetzt mal kurz nicht Luft holen Now once briefly not air get ‘Don’t breathe for a moment’ Erwin ist {nicht/kein} Rentner, (sondern (ein) Frührentner) Erwin is not/no pensioner (but (an) interim pensioner) ‘Erwin isn’t a pensioner (but (an) interim pensioner)’

Pattern B is used with an emphatic reading of ein as a numeral cf. (10). ●

(10)

pattern B – determiner ein: Erwin hat in seinem Leben nicht EINEN Schluck Fencheltee getrunken. Erwin has in his life not a sip fennel-tea drunk ‘Erwin has not drunk a (single) sip of fennel tea in his life.’

With topicalisation (inverse scope reading) cf. (11), so-called light negation (nonasserted negation) cf. (12) and in some PPs cf. (13), (14), pattern A or B is used with the same distribution as in non-negated clauses (mass nouns and plural indefinite count nouns pattern A, singular indefinite count nouns pattern B). ●

(11)

pattern A – bare NP/no determiner or pattern B – determiner ein: {Haustiere/Einen Dackel} hat Erwin nicht. Pets/ A dachshund has Erwin not ‘Pets/A dachshund, Erwin doesn’t have.’

. Further collocations where a bare NP occurs under negation are Schlange stehen (‘to cue’) or Wort halten (‘to keep/stick to one’s word’). Other noun-verb collocations such as Luft bekommen (‘to get air’/‘to be able to breathe’) or Hunger/Angst haben (‘to be hungry/afraid’, cf. 5) follow the standard pattern D.

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

(12)

wenn Erwin sich nicht doch noch {Haustiere/einen Hund} zulegt If Erwin himself not yet still pets/ a dog gets ‘Unless Erwin is getting pets/a dog after all’

(13)

Dieser Hund sieht mir nicht [PP nach einem Dackel] aus. This dog looks me not after a dachshund VPart ‘This dog doesn’t look much like a dachshund to me.’

(14)

Erwin hört nicht [PP auf Ratschläge] von anderen Erwin listens not to suggestions of others ‘Erwin doesn’t take advice from others’

The fact that PPs form exceptions to the standard pattern suggests that syntactic locality constraints play a role. Unlike in Modern English, there is no pattern C determiner in Modern German. In other words, there is no NPI-determiner that is restricted to [+non-aff] contexts.8

.

Old High German

Negation in OHG is generally expressed by the proclitic negative particle ni on the finite verb (the head Neg0 of a corresponding functional projection in the domain of verbal inflexion, cf. Jäger 2005). N-words, that is neg-marked indefinite pronouns and adverbs such as nioman (‘nobody’), niwiht (‘nothing’), nio (‘never’) etc. are possible in addition: OHG is an NC language. Narrow focus of negation is marked by the verb-independent neg-particle nalles. The investigation of indefinite determination in the scope of negation in OHG is based on a corpus from the OHG translation of Tatian’s gospel harmony (before 830 AD), the OHG translation of Isidor’s “De fide catholica” (around 810 AD), Otfrid’s gospel book (around 860–870 AD) and Notker’s “Psalter” (around 1020 AD). All four syntactic patterns of indefinite nominal determination that I introduced above are attested in the scope of negation in the OHG corpus: ●

(15)

pattern A – bare NP/no determiner: (& non erat illis filius.) Inti niuuard In sun. and N-was them son ‘and they had no son’ T 26, 6

. This does of course not mean that there are no NPIs in Modern German. For instance in the adverbial domain there is je (‘ever’). For a list of Modern German NPIs compare Kürschner (1983: 301–327). Yet there is no indefinite NPI-determiner comparable to English any.

S F O PRO



1st proofs

 Agnes Jäger ●

(16)

●

(17)

(18)

●

(19)

pattern B – determiner ein: (& ipsi uno digito/uestro non tangitis sarcinas.) Inti ir mit einemo fingare/íuuueremo niruoret thia burdin. And you with a/one finger yours N-touch the burden ‘and do not touch the burden with a single finger of yours’ T 246, 26f. pattern C – non-affirmative determiners einig and dehein/thehein(ig): (& neque catenis/iam quisquam eum poterat ligare) noh mit k&inon/giu mohta ín einig mán gibintan nor with chains ever could him any man bind ‘nor could any man bind him with chains’ T 87, 12f. Ni mag thiu worolt, wizit thaz,/haben in iu theheinan haz, N can the world, know that, have in you any hatred ‘You must know that the world cannot bear any hatred against you’ O III. 15, 29 pattern D – negative determiner nihein(ig): ni si mán nihein so úeigi N be man none so cowardly ‘no man shall be so cowardly’ O I. 11, 10

All four determiners, ein, einigig, dehein/thehein(ig) and nihein(ig), could be used both adnominally and pronominally, that is with an empty N0, throughout their history. (20)

(Videte ne condemnatis/unum ex his pussillis) gisehet thaz ir ni uornidaret/einan fon thesen luzilon see that you N despise one of these little ones ‘Take heed that you do not despise one of these little ones’ T153,1f.

(21)

(quia nullus in regno israhel cyrus est dictus.) In andra uuiis ni uuardh eo einic in israhelo riihhe cyrus chinemnit. in other way N was ever any in Israel kingdom Cyrus called ‘Otherwise, noone had ever been called Cyrus in the kingdom of Israel’ Is III, 3

(22)

daz si deheinen wolde ze truotenne han that she any-/noone wanted for husband have ‘that she wanted noone for a husband’ Nib III. 49, 1–3

(23)

(& precepit/illis. ne cui dicerent.) gibot her/in tho thaz sie niheinagamo nisagatin told he them then that they none N-said ‘He told them to tell noone’ T 130, 15f.

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

These kinds of examples were also included in the data. .

Proportion of patterns A, B, C, D

The proportions of all patterns of indefinite determination in the scope of negation for each text are given in (24).9 (24)

Proportions of patterns A, B, C, D in OHG: 60

57

50 A: bare NP

40

B: ein 30

C: einig

20 10 0

5

9 3

0

Tatian

.

C: dehein(ig)

16 8

6

5 1

0 0

Isidor

D: nihein(ig)

6

00 1

00 1

Otfrid

Notker

5

Patterns A, B: no determiner and determiner ein

Pattern A, the use of indefinite bare NPs under negation, is very frequent in contrast to Modern German where the standard is pattern D. In Tatian, Otfrid and Notker, pattern A is the most frequent, most strikingly so in Tatian. On the other hand, this fact is easily accounted for: bare NPs were generally more common in OHG. The rise of articles only took place in later OHG and MHG (cf. Oubouzar 1992; Leiss 1994; Philippi 1997).10 Pattern B with the determiner ein in the scope of negation, which is comparatively common in Tatian and especially Isidor, is also attested in cases where it is not possible today (cf. Ex. (16) above). Both patterns are distributed across all three main types of contexts distinguished above.

. Tatian and Isidor were analysed entirely, of Otfrid and Notker the first 100 negated clauses were included in the corpus. In the case of Otfrid this means approximately 50 pages of text. The last example is OI. 15, 7. In the case of Notker it is approximately 30 pages of text, the last example being NPs 12, 1 (= 37, 24f.). The proportions for each text are given in absolute numbers of occurences in the corpus. . The question of whether the common use of a bare NP (pattern A) instead of the other patterns may be related to verbal aspect due to a link between negation and imperfectivity or due to accusative/genitive case alternations (cf. Leiss 2002; Abraham 1997) would require a comparative investigation of indefinite determination outside the scope of negation. I have to leave this issue to future research.

S F O PRO



1st proofs

 Agnes Jäger

.

Pattern D: negative determiner nihein(ig)

The ratio of pattern D with the determiner nihein(ig) increases in OHG from Tatian via Otfrid to Notker. It is not attested in Isidor. With nihein(ig), OHG in contrast to Modern German possesses a clearly morphologically neg-marked determiner.11 Nihein(ig)-NPs are mostly used in addition to the neg-particle ni on the finite verb or other neg-markers in the same clause, cf. (19) and (23). In other words, nihein(ig) mostly occurs in constructions including NC, cf. Table (27). However, nihein(ig) occasionally occurs as the only neg-marker in the clause (vs. Donhauser 1996) and thus suffices to identify negation, by which criterion it is to be classed as an n-word: (25)

(quia nemo/proph&a acceptus est) thaz nihéin/uuizago antphengi ist that none prophet accepted is ‘that nobody is accepted as a prophet’ T 114, 18

(26)

daz siê îro nehêinen loûgen getórsten háben. that they them no lie dared have ‘that they did not dare to deny them’ N Ps 3, 8 (=14, 7)

The distribution of nihein(ig) in clauses with and without NC in all four texts is given in Table (27). (27)

T Is O N

OHG nihein ± NC: nihein + further neg-marker (NC)

nihein as only neg-marker

6 – 7 4

3 – 1 1

Within the indefinite NP, nihein is placed before or – less often – after the noun. The position relative to the noun makes no difference for the inflexion of nihein: apart from use with an empty noun, there is no ending for singular nominative

. The etymology of nihein(ig) has been subject to some discussion in the literature. Braune/ Reiffenstein (2004: 254, 151) postulate a composition of *nih+ein, where *nih is a neg-element. However, this element is not attested anywhere in isolation. According to Bech (1964: 213), Danielsen (1968: 93f.), and more recently Lloyd et al. (1998: 562f.), nih-/neh- can be linked to Indogermanic *ne-kue ‘and not’. Lloyd et al. further assume that noh- is linked to OHG noh ‘and not/nor’ with semantic change from ‘until now’ to ‘and not’.

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

of all genders and accusative neutre. Otherwise, nihein inflects according to the strong adjectival inflexion: (28)

nihéinigemo níde (‘no envy/hatred’ D M)

(29)

lídes nihéines (‘no fruit wine’ G M) OI. 4, 35

(30)

ne-hêin zît (‘no time’ N F)

(31)

man nihéin (‘no man’ N M) OI. 9, 31f.

OI. 2, 22

NPs 2, 7 (= 12, 1f.)

Nihein is restricted in its distribution to [+non-aff, +neg] contexts, i.e. direct sentential negation. .

Pattern C: non-affirmative determiners einig and dehein(ig)

With einig and dehein(ig), OHG in contrast to Modern German possesses specific determiners for non-affirmative contexts. As will be shown in this section, they are both weak NPIs. With respect to einig, this is all the more plausible in view of the fact that the OHG form einig corresponds directly to Old English wnig, the predecessor of the Modern English NPI any. Furthermore Dutch enig in the meaning of ‘any’ is still restricted to NPI contexts (Hedde Zeijlstra, p. c.): (32)

Heb jij enig idee wat dat zou kunnen betekenen? – ?/*Ja, ik heb enig idee. Have you any idea what that shall can mean Yes I have any idea ‘Do you have any idea what that could mean? – Yes, I have some idea.’

A few comments in the literature already hint at the NPI character of OHG einig. Braune/Reiffenstein (2004: 253) state that this indefinite occurs only in negative clauses, questions and several, not further specified kinds of dependent clauses. Karg-Gasterstädt et al. (1971: 191–194) mention as syntactic contexts for einig : negated clauses, ‘negated phrases’ (phrases with āna ‘without’ and noch ‘nor’), questions, conditional clauses, dependent interrogative clause, clauses dependent on negated matrix clauses, ‘modal clauses’, ‘hypothetic clauses’ (contrafactuals), certain attributive clauses (restrictors of universal quantifiers). Furthermore they report that einig is used to translate Latin ullus in glosses, which also corraborates the hypothesis that einig is an NPI since ullus is an NPI in Latin.12 The contexts listed in the literature match those referred to as weak NPI contexts in van der Wouden (1997).

. According to Karg-Gasterstädt et al. (1971: 193), einig is rarely found in non-negated matrix clauses. The only example they give is: einig [est denique] quoddam (Ms. aliquod). However, this example is inconclusive, especially since the manuscript apparently has aliquod rather than quoddam.

S F O PRO



1st proofs

 Agnes Jäger

In order to elucidate further the syntactic distribution of einig, I analysed all occurences of einig in the entire texts of Tatian, Isidor and Otfrid.13 For a correct interpretation of the results it is crucial to differentiate the indefinite einig meaning ‘any’ from the adjective einig meaning ‘unique/alone/single’.14 Isidor has only indefinite einig, Tatian also adjectival einig, but Otfrid only adjectival einig (compare also Kelle 1881: 94f.). The number of occurrences of einig on the whole is extremely low. The following graphs illustrate the distribution of all occurrences of the indefinite determiner einig across the three main types of contexts in Tatian and Isidor.15 (33)

Distribution of OHG einig einig in Isidor

einig in Tatian 7

4

6 5

3

4 2

3 2

1

1 0

0 [–nonaff, –neg]

[+nonaff, –neg]

[+nonaff, +neg]

[–nonaff, –neg]

[+nonaff, –neg]

[+nonaff, +neg]

. In Notker’s Psalter, I checked the first 50 pages of text in the edition by Tax (1979): until Ps 18, 13 (page 58, line 26) there are no occurrences of einig at all. . Whereas the former corresponds to Old English wnig, the latter is the equivalent of Old English ānga (cf. Karg-Gasterstädt et al. 1971: 171, 191). ther gotes einigo sun. ‘God’s only son’ (O II. 3, 23) Note that Modern Dutch einig preserves both uses. In OHG translations of Latin texts, the indefinite einig occurs as a translation of Latin quisquam or quis, in combination with the neg-particle ni on the finite verb in OHG also as a translation of Latin nullus, Adjectival einig translates Latin unicus or unigenitus and is used almost only in affirmative contexts. Since the 16th century, adjectival einig is attested with the Modern German meaning ‘in agreement/of one opinion’ (Pfeifer 1989: 340f.)

S F O PRO

. The distribution is given in absolute numbers of all occurrences of einig and its variants in the entire texts.



1st proofs

German indefinite determiners in the scope of negation 

The indefinite einig is indeed restricted to non-affirmative contexts, both negative and non-negative ones. The following [+non-aff, −neg] contexts for the indefinite determiner einig are found in these two texts: ●

(34)

●

(35)

●

(36)

(37)

questions (numquid aliquis ex principibus/credidit In eum aut ex pharisæis?) enoni ening fon then heriston/giloubta In Inan odo fon then phariseis? P any of the lords/ believed in him or of the pharisees ‘And did any of the lords or the pharisees believe in him?!’ T 213, 6f. conditionals (Quodsi de cyro persarum rege quis hoc crediderit prophetatum . . .) Ibu dhanne einic chilaubit, dhazs dhiz fona cyre persero chuninge if then anyone believes that this by Cyrus Persian king sii chiforabodot . . . was predicted . . . ‘If anyone believes that this was predicted by the Persian king Cyrus . . .’ Is III 3 the complement position of āna/ūzzan (‘without’) (et antequam tempus uirginis parturiendi ueniret, genuit eum sine tempore pater.) enti aer danne dera magadi ziit biquami za gaberanne, and before than the maiden time came to give-birth gabar inan fater ano einigero ziteo bigin. gave-birth to-him father without any time beginning ‘and before the time came for the virgin to give birth, he was born by the father without/beyond any time’ Is II 1 (procul dubio [. . .] cognoscitur) buuzssan einigan zuuiuun ist dhanne archennit, dhazs . . . without any doubt is then recognized that ‘It is recognized without any doubt that . . .’ Is VI 5

These contexts are of course typical weak NPI contexts so that, just as its English cognate any, einig ought indeed to be classed as a weak NPI. As the figures in (33) show, the majority of occurences of the indefinite einig is however found in proper negative contexts, that is in the semantic scope of sentential negation. In the OHG corpus, there is a second pattern C determiner viz. dehein(ig). This indefinite determiner is even rarer than einig. Interestingly, it is not attested in the earlier texts that contained einig but only in Otfrid (in the form of thihein, thehein and theheinig) and in Notker (in the form of dehein and dohein). This might suggest that einig is an earlier NPI determiner than dehein(ig), which only starts to appear in the course of the OHG period. This prooves Behaghel’s (1913: 181) statement wrong that

S F O PRO



1st proofs

 Agnes Jäger

einig only stepped in later for deheinig with the meaning equivalent to Latin ullus. With respect to dehein(ig) on the other hand, my findings coincide with Behaghel (1913: 178) who says that thihein/dehein is found only starting with Otfrid. Yet according to Lloyd et al. (1998: 562f.), dehein(ig) is already attested as early as the ninth century.16 There are a few remarks in the literature regarding the distribution of dehein(ig). Kelle (1881) notes that it is used in clauses dependent on negated matrix clauses. Behaghel (1913: 179) already gives quite a comprehensive list of contexts. He mentions negated clauses, conditionals, the standard of comparison and questions, but also a few ‘affirmative’ clauses. Danielsen (1968: 96, 116f.) states that dehein(ig) is “decidedly positive” and “restrictive”. According to Lloyd et al. (1998: 562), it occurs in negative clauses, conditional clauses, clauses after comparatives, and questions. Braune/Reiffenstein (2004: 254) also list negative clauses and questions as licensing contexts, as well as “several kinds of dependent clauses” that they do not specify any further. A closer look at the data from my corpus against the background of the three types of contexts I distinguished above, i.e. [−non-aff, −neg], [+non-aff, −neg] and [+non-aff, +neg] contexts, reveals the following picture for the distribution of dehein(ig):17 (38)

Distribution of OHG dehein(ig) and variants: Notker: dehein/dohein

Otfrid: thihein/thehein(ig) 20

20

15

15

10

10

5

5

0

0 –nonaff, –neg

+nonaff, –neg

+nonaff, +neg

–nonaff, –neg

+nonaff, –neg

+nonaff, +neg

. Again, the etymology is controversial: The hypothesis of a composition of OHG dih/deh and ein cf. Braune/Reiffenstein (2004: 253) is to be rejected since dih/deh is otherwise not attested in OHG (cf. Bech 1964; Lloyd et al. 1998). Bech (1964) and Danielsen (1968) assume a combination of the demonstrative *þes and the numeral *ain-, conjoined according to Bech by the copulative particle *-h, according to Danielsen by the interrogative or generalizing *hw-. However, the presupposed development -sh- > -hh is unparalleled according to Lloyd et al. (1998: 563). Lloyd et al. assume that doh- in dohein can be linked to OHG doh. Dihein and dehein were probably formed under the influence of the parallel n-words nohein/nihein/nehein.

S F O PRO

. The distribution is given in absolute numbers of all occurrences of thehein and its variants in the entire texts.



1st proofs

German indefinite determiners in the scope of negation 

Dehein(ig) is evidently only licensed in [+non-aff] contexts. In both texts, it is predominantly used in [+non-aff, −neg] contexts, less often in [+non-aff, +neg] contexts. In this respect dehein(ig) differs from einig, which is used mostly in [+non-aff, +neg] contexts. On the other hand, dehein(ig) behaves just as einig in that it is licensed in non-affirmative contexts and therefore is to be considered a weak NPI. Dehein(ig) is also lacking in [−non-aff, −neg] contexts.18 Thus Danielsen’s (1968) characterisation of dehein(ig) as decidedly positive is not adequate. This is partly due to his failure to differentiate [±non-aff] contexts. Dehein(ig) cannot be said to be generally licensed in positive contexts. The basic contexts that Kelle (1881), Behaghel (1913), Lloyd et al. (1998) and Braune/Reiffenstein (2004) mention are all non-affirmative contexts.19 Among the [+non-aff, −neg] licensing contexts for dehein(ig) in Otfrid and Notker there are: ●

(39)

●

(40)

●

(41)

the complement position of āna (‘without’) theist ouh fésti ubar ál/ ána theheinig zwíval that-is also fix above all without any doubt ‘That is definitely the case without any doubt’ O V. 11, 14 conditionals War imo sulih man thihein so quami wisheiti heim was him such man any so came wisdom home ‘if he had any such man, wisdom would come home’ O II. 4, 13 the standard of comparison Vuanda rêinerun/lîchamen/sêla kîbest dû mir. danne dehêin corpus because purer body soul give you me than any body uuerden múge. become may ‘because you give me a spiritual body purer than any body may be’ N Ps 50 (= 177, 12–14)

. However, there is one possible counterexample in Otfrid: Sin drút thehein ther wúrti/er síneru gibúrti, // firnám thaz scolti wérdan tház ‘One of his friends who was born before him had heard that that was going to happen’ (O IV 5, 63). Behaghel (1913: 179) also quotes this sentence as a “quite singular” (“ganz vereinzelt”) example. . In fact, Behaghel’s (1913) rare examples of dehein in ‘affirmative’ contexts also constitute weak NPI contexts that would license any in English (e.g. daz du den iemer hazzen muost, deme dehein ere geschiht Iw. 141 ‘that you must always hate him, who any honour is given to’).

S F O PRO



1st proofs

 Agnes Jäger ●

(42)

clauses dependent on negated matrix clauses or on adversative matrix predicates Nist ther io gihogeti in alleru worolti,/thaz kuning thihein fuari mit N-is who ever remembered in all world that king any went with sulicheru zieri such adornment ‘There is noone in all the world who would remember that any king ever went with such adornment’ O IV. 4, 24

Again, these contexts are obviously weak NPI contexts. Thus the apparently random range of licensing contexts for dehein(ig) observed in the literature can be captured in a unified account of dehein(ig) by stating that it is a weak NPI. Accordingly, it is not surprising that, just as for einig cf. (36), one major licensing context for dehein(ig) is also the complement position of āna (‘without’) – a fact that seems to have gone unnoticed in the literature so far. The respective equivalents of ‘without’ are known to license weak NPI elements in various languages. Compare for instance English: (43)

I would like my chips without {any/*some} salt

In contrast to nihein(ig), einig and dehein(ig) are never used as the only means of identifying negation in OHG and are thus not n-words but NPIs corresponding to English any.20 .

The role of the Latin original in translations

Regarding those OHG texts that are direct translations of Latin texts, one always has to consider the possibility of linguistic influence of the Latin original. This holds for Tatian and Isidor, but only partly for Notker, who mostly interprets and explains the Latin original rather than literally translating it. The Latin original was taken into account for each sample sentence from these texts. For the issue of this investigation, it is important to ascertain to which extent the choice of the particular syntactic pattern for indefinite NPs in the scope of negation was determined by the Latin text. There are correspondences between the Latin original and the OHG translations concerning pattern A, i.e. the use of bare NPs, as well as pattern B with OHG ein rendering Latin unus, quis or quisquam. The high ratio of pattern A in Tatian may accordingly be partly due to Latin influence. However, Otfrid’s gospel book, which is not a direct translation of a Latin text, also has a majority of pattern A, suggesting that this was indeed a genuine OHG pattern. . The correspondance of OHG einig and dehein(ig) to English any is also briefly noted in Lockwood (1968: 79) who, coming from an English background, was more sensitive to what I refer to as pattern C/NPI determiners.

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

Significantly, the OHG patterns B, C and D are used even against the Latin original. One can conclude that these patterns were definitely a genuine part of OHG grammar. Examples of the OHG translation diverging from the Latin original in using pattern B, C or D, respectively, are given in (44) through (48). ●

(44)

●

(45)

●

(46)

●

(47)

●

(48)

Latin D (nullus): OHG B (ein) (Quia iam sicut nullum templum [. . .]remansit) Huuanda so selp so im noh ein tempel ni bileiph [. . .] Because as same as him neither a temple N remained ‘Because just as neither a temple remained to him. . .’ Is VIII, 3 Latin D (nullus): OHG C (einig) (quia nullus in regno israhel cyrus est dictus.) In andra uuiis ni uuardh eo einic in israhelo riihhe cyrus chinemnit. in other way N was ever any in Israel kingdom Cyrus called ‘Otherwise, noone had ever been called Cyrus in the kingdom of Israel’ Is III, 3 Latin A (bare NP): OHG C (einig) (nec apostolus dicit nec propheta conperit nec angelus sciuit nec creatura cognouit) Dhazs ni saget apostolus noh forasago ni bifant noh angil gotes ni uuista That N says apostle nor prophet N found-out nor angel god N knew noh einic chiscaft ni archennida. nor any creature N acknowledged ‘No apostle said this nor did any prophet find it out nor did any angel of God know it nor did any creature acknowledge it’ Is II, 3 Latin A (bare NP): OHG D (nihein(ig)) (Ego enim non inuenio in eo causam) ih nifindu in imo niheiniga sahha I N-find in him no thing ‘I do not find any thing in him’ T 308, 28 Latin B/C (ullus/aliquis/quis): OHG D (nihein(ig)) (& non est tibi/cura de aliquo.) Inti nist thir/suorga fon niheinigemo and N-is you worry of none ‘and you do not worry about anyone’ T 207, 7f.

S F O PRO

The second pattern C determiner, dehein(ig), does not occur in the OHG translations of Tatian or Isidor.



1st proofs

 Agnes Jäger

.

Middle High German

In MHG, negation is marked by the clitic negative particle (Neg0) in the phonetically reduced form of ne or en immediately before the finite verb21 and/or a second, verb-independent neg-particle niht (SpecNegP, cf. Jäger 2005), which was grammaticalised from OHG niwiht (‘nothing’). N-words are possible either in addition to the negative particle or as the only neg-markers in the clause. Just as OHG, MHG was an NC language. The following discussion of indefinite determination in the scope of negation in MHG is based on a corpus from Nibelungenlied (1190–1200 AD), ProseLancelot (around 1250 AD), and Berthold of Regensburg “Four German Sermons” (around 1275 AD). Again, all four syntactic patterns distinguished above are to be found in the scope of negation: ●

(49)

●

(50)

●

(51)

●

(52)

pattern A – bare NP/no determiner: Es enwart nye frauw die also sere mynnet gejegcz zu wald als sie det. It N-was never woman who quite-so much loved hunting at wood as she did ‘There was never a woman who loved hunting in the wood quite so much as she did’ Lanc 24, 94 pattern B – determiner ein: wir heten ninder einen zagen. We had nowhere a hesitation ‘We did not hesitate at all’ Nib IV. 231, 422 pattern C – non-affirmative determiner dehein/kein: wann Claudas [. . .] enwolt im dheynen dienst geben because Claudas N-wanted him any/no service give ‘because Claudas did not want to serve him’ Lanc 10, 7 pattern D – negative determiner nehein(ig): daz man helt neheinen so schoenen nie gesach. that man hero none so beautiful never saw ‘that one never saw any such beautiful hero’ Nib V. 292, 423

. Ne/en may appear either proclitically on the finite verb or – secondarily after V-to-C movement – enclitically on the element immediately before the finite verb. In any case it is left-adjacent to the finite verb and moves with it. .

= ms. A; vs. mss. B/C: cheinen zagn/dheinen zagen (= pattern C).

.

= ms. A; vs. mss. B/C: helt deheinen (= pattern C).

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

. Proportion of patterns A, B, C, D The proportions of the four patterns of indefinite determination in the scope of negation for each MHG text are given in (53).24 (53)

Proportions of patterns A, B, C, D in MHG: 25 20

22 17

17

16

A: bare NP

15

5 0

C: dehein/kein D: nehein

4 5 0 Nibelungenl.

.

B: ein

10

10

1

0

Lancelot

0

0

Berthold

Pattern A: bare NP/no determiner

With the rise of articles in the MHG period (cf. Oubouzar 1992; Leiss 1994; Philippi 1997; Paul 1998) fewer bare NPs are used in general. Since the writers often resort to n-words as determiners for indefinite NPs under negation, a secondary effect of the rise of determiners is the increase of NC. The following parallel examples of the indefinite NP with N0 stein/steyn (‘stone’) in the scope of negation once in OHG and once in MHG are quite illustrative of the development away from bare NPs: ●

(54)

(55)

OHG pattern A → MHG pattern C: (& non relinquent In te/lapidem super lapidem) Inti niforlazent In thir/stein obar steine and N-leave in you stone above stone ‘and they do not leave inside you one stone above the other’ T 192, 6f. und wart das lant allenthalben so zurfuret das nye keyn steyn off and was the country in-every-way so destroyed that never any/no stone above dem andern beleib; the other remained ‘and the country was so much destroyed that no stone remained above the other’ Lanc 12, 13

. The first 100 negated clauses from each text were included in the corpus. In Nibelungenlied these are contained in the text up to V, 290, 3, in Lancelot up to sentence 266, page 52, and in Berthold up to Sentence 299, page 48. The proportions for each text are again given in absolute numbers of occurences.

S F O PRO



1st proofs

 Agnes Jäger

Despite of this development, pattern A is still quite frequent in MHG. In Nibelungenlied and Berthold, it is the even most frequent pattern. In Lancelot it amounts to almost 40 percent of the indefinite NPs in the scope of negation contained in the corpus. Bare NPs are often used where this would be ungrammatical in Modern German, compare Ex. (49) above. . Patterns B, D: determiner ein, and negative determiner nehein In the MHG corpus, there are hardly any instances of the pattern B determiner ein in the scope of negation any more. The n-determiner ne-/enhein (from OHG nihein) is only found in later parts of some manuscripts of Nibelungenlied (but not in the first 100 negated clauses that form the basis for Figure (44)). Altogether there are six examples of nehein in manuscript A and one each in manuscripts B and C, as well as one example each of enhein in manuscripts A and C and two examples in manuscript B. There is no case in which all manuscripts of Nibelungenlied coincide in using ne-/enhein. The n-determiner ne-/enhein is not contained at all in Lancelot or Berthold. One may conclude that the pattern D determiner nehein is becoming extinct in the MHG period. Behaghel (1913: 179) assumes that nehein was entirely driven out and replaced by dehein around 1200. According to Lloyd et al. (1998: 563), the extinction of nehein is complete by the 16th century. Only in certain dialects of German it is still preserved, compare Swiss German variants of nekein/ekein: (56)

as de Bueb Charakter hed ond e kän Plauderi ischt that the boy character has and no chatter-box is ‘that the boy has character and is no chatter-box’ (www.christentum.ch/gebet.htm)

The system of indefinite determiners thus became ‘defective’ in comparison to the general system of MHG indefinites that included various n-words. This loss of the original n-word determiner probably caused the ensuing change of dehein/kein. .

Pattern C: non-affirmative determiner dehein/kein

As compared to the OHG situation, there is a drastic increase of the pattern C determiner dehein in MHG. This goes mostly at the expense of pattern D and partly also of pattern A, cf. Ex. (54) vs. (55) above. In all texts, variants both with and without de- can be found: dehein/dechein/ dhein vs. chein/kein. One can observe a diachronic shift from a majority of forms with de- to a majority of forms without de-. Dehein is the direct ancestor of Modern German kein (cf. Behaghel 1913: 180; Eroms 1993: 2; Lloyd et al. 1998: 562f.). Kein may partly follow on from nehein also (cf. Pfeifer 1989: 823; Danielsen 1968: 92/99; Lloyd et al. 1998: 563), but this pattern D determiner had already largely become extinct before the rise of dehein and kein. Furthermore, there is a straightforward

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

phonetic development from dehein to kein. Behaghel (1913: 180) therefore concludes that dehein formed the only source for kein.25 The following phonetic development can be assumed: starting from deh.ein, the syllable boundary shifted (cf. Braune/ Reiffenstein 2004: 151) producing de.hein. Ensuing ә apocope (cf. Lloyd et al. 1998: 562) led to the form dhein/dchein which is also evidenced in writing: (57)

wann dhein man enmöcht nymer so großen jamer erdencken because no/any man N-may never so great sorrow think-up ‘because no man may ever think up such a great sorrow’ Lanc 54, 287

Eventually, a place assimilation of the initial stop took place rendering kein. The same kind of place assimilation of an initial dental to a following fricative can be observed for instance in Swiss German where the combination d’chind (‘the child’) of the definite article /t/ and the noun /xind/ (‘child’) is pronounced as /kxind/. The distribution of dehein and its variants in the three MHG texts in the corpus across the three main types of contexts, viz. [−non-aff, −neg], [+non-aff, −neg] and [+non-aff, +neg], is given in (58).26 Forms with and without d(e)- are differentiated in order to capture any differences in distribution. For Nibelungenlied, which also contains a few occurences of ne-/enhein, the distribution of this item is depicted in addition. (58)

Distribution of MHG dehein and variants (as well as nehein in Nibelungenlied): Nibelungenlied

Lancelot

Berthold

20

20

20

15

15

15

10

10

10

5

5

5

0 –non- +non- +nonaff, aff, aff, –neg –neg +neg de(c)hein chein/kein ne/enhein

0 –non- +non- +nonaff, aff, aff, –neg. –neg. +neg. dheyn/dhein keyn/kein

0

–non- +non- +nonaff, aff, aff, –neg. –neg. +neg. dehein/dekein kein

The figures in (58) show that the various forms of dehein occur mostly in negative contexts. This is particularly so for the forms kein/keyn/chein without de- as compared to dehein, which indicates a development that eventually led to Modern German kein as .

For the controversy concerning kein see also Gerring (1927), Lloyd et al. (1998).

S F O PRO

. The distribution was analysed for all up to the first 20 examples of each variant of dehein in each text. Close variants such as dehein and dechein were then grouped together in the graph.



1st proofs

 Agnes Jäger

an element entirely restricted to [+non-aff, +neg] contexts. The determiner nehein/ enhein, that only appears in Nibelungenlied, is restricted entirely to negative contexts. The different MHG variants of dehein are still fairly common in [+non-aff, −neg] contexts, notably the forms with de-, but not attested at all in [−non-aff, −neg] contexts. This last result contradicts Pensel (1981), who explicitly speaks of negative and affirmative dehein. Dehein does not occur in affirmative contexts in any of the three texts. Paul (1998) states that dehein is used in “hypothetic clauses”, meaning clauses dependent on negated matrix clauses or on questions. These are indeed two of the contexts that dehein is attested in my corpus, but again there are several other [+non-aff, −neg] licensing contexts that the various forms of dehein are found in: ●

(59)

●

(60)

●

(61)

●

(62)

conditionals Nu dunckest du mich als wise, sol kein man radt darzu geben, das thust Now seem you me as wise, should any man advice to-that give, that do auch du. also you ‘Now you seem to me just as wise. If any man should give advice on this, so should you’ Lanc 88, 521 standard of comparison Du bist milt und gut, me dann ye kein konig wart, beide gegen got und You are mild and good, more than ever any king was, both towards god and gegen der welt. towards the world Lanc 138, 872 occasionally in clauses dependent on negated matrix clauses or adversative predicates such as furhten (‘to fear’), verbieten (‘to forbid’) etc. Wan ez wart niemer gehört von anegenge der werlte, daz ie Because it was never heard from beginning of-the world that ever dehein mensch sô bittern tôt ie erlite any man such bitter death ever suffered ‘Because it was never heard since the beginning of the world that any man ever suffered such a bitter death’ Bert 4, 1098 in the syntactic context of selten (‘rarely’), ee (‘before’), kûme (‘hardly’) etc. ich zeih iv einen wirt,/ daz ir ce hvse selten so wol bechomen birt/in I show you a landlord that you to house rarely so well come are in {[ms. A:] deheime/[ms. B:] deheinen/[ms. C:] keinem frömden} lande, als any/ any/ any foreign country as iv hi mach geschehen you here may happen

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

‘I will show you a landlord such that you have rarely come to a house so well in any (foreign) country as may happen to you here’ Nib XXVI. 1711, 1–3

Comparing the figures in (58) to those for OHG given above in (38), there is a diachronic shift in distribution from predominantly [+non-aff, −neg] contexts in OHG towards predominantly [+non-aff, +neg] contexts in MHG, or, employing the terminology of van der Wouden (1997), from weak towards strong NPI contexts for thehein/dehein/kein. In the course of the historic development of German, kein slowly shifts from a pattern C determiner to a pattern D determiner. While it started out as an element that was licensed in various non-affirmative contexts, it finally became restricted to negative contexts.27 These findings contradict Eroms (1993), who proposes that dehein/kein was originally a ‘neg-strengthener’. My investigation of the distribution of this element in OHG and MHG texts reveals that instead, it originally was a special indefinite determiner for non-affirmative contexts comparable to the Modern English NPI determiner any. It served to ‘strengthen’ negation just as little as any does in English. Over the course of time the distribution of dehein/kein shifted more and more to [+non-aff, +neg] contexts. Eventually it sufficed to identify the sentential negation on its own without any additional neg-marker present in the clause. According to Paul (1998: 379/400) this was the case since the 12th century, according to Lloyd et al. (1998: 562) since around 1200. In other words, dehein/kein was eventually reanalysed as an n-word (or negative item/NI in the sense of Weiß 2002). Note that the fact that dehein/kein still also occurs in [+non-aff, −neg] contexts does not by itself speak against its categorisation as an n-word. N-words are also licensed in these contexts in several other languages, e.g.: (63)

Pedro duda que venga nadie. Pedro doubts that come nobody ‘Pedro doubts that anybody will come’

(Spanish)

(64)

Natalka pryshla bez zhodnoj prodjapyny zi sholy. Natalka came without no scratch out school ‘Natalka came out of school without a scratch’

(Ukranian)

(65)

O Paulo sabe isso melhor do que ninguém. the Paulo knows that better than nobody ‘Paulo knows that better than anybody else’

(Portuguese)

. Erben (1950: 202/214) also notices in passing the fact that dehein is more and more restricted to negative contexts without however investigating its exact distribution and diachronic changes.

S F O PRO



1st proofs

 Agnes Jäger

The grammar of the MHG n-word dehein/kein resembles that of the n-words in these languages in so far as, while it mostly occurs in negative contexts, it is still occasionally used in [+non-aff, −neg] contexts. Alternatively, one may assume a temporary ambiguity between or a lexeme split into an NPI and an n-word dehein/kein of which only the latter was preserved. N-words differ from NPIs in that they are associated with a formal neg-feature that may suffice to identify sentential negation. Indeed, whereas OHG dehein(ig) was never used as the only means of identifying negation in a clause (compare also Donhauser 1996), in the MHG corpus there are frequent cases of dehein/kein as the only neg-marker in the clause, so that at least in those cases if not in general one ought to class it as an n-word: (66)

ovch hoere iuch selben der degenheite iehen,/daz man kvnich deheinen also hear you self the bravery tell that man king any/none chuoner habe gesehen. more-courageous has een ‘I also hear yourself being praised of such bravery that one has not seen any more courageous king’ Nib III. 112, 1f.

(67)

das sol das buch vil wol hernach gesagen, wann wir haben es yczo keyn that shall the book very well later tell, because we have it now no/any stadt place ‘The book shall tell that later, because we don not have any room for it now’ Lanc 10, 3

(68)

roub unde diepheit daz mac kein amt gesîn. robbery and theft that may no/any profession be ‘Robbery and theft cannot be a profession’ Bert 1, 94 (p. 16)

Besides cases where it is the only neg-marker, dehein/kein also appeares in [+non-aff, +neg] contexts together with other neg-markers, i.e. in NC constructions, just as do other MHG n-words. The ratio of dehein/kein in [+non-aff, +neg] contexts with and without NC is given in Table (69). (69)

Nib Lanc Bert

MHG dehein/kein ± NC: dehein/kein + further neg-marker (NC)

dehein/kein as only neg-marker

1 14 6

4 8 4

S F O PRO



1st proofs

German indefinite determiners in the scope of negation 

The development of dehein/kein can be summed up in the following way: (70)

Development of dehein/kein: weak NPI: in various [+non-aff] contexts ↓ strong NPI: mostly or entirely restricted to [+non-aff, +neg] contexts ↓ n-word: suffices to identify a context as [+non-aff, +neg]

Similar diachronic developments have taken place within the system of indefinites in other languages, too. Compare for instance the distribution of aliquis and its successors in various Romance languages (Elisabeth Stark, p. c.): Classical Latin aliquis was restricted to [−non-aff, −neg] contexts. In Mediaeval Latin its distribution shifted so that it could also be used in [+non-aff, −neg] contexts, i.e. weak NPI contexts. Italian alcuno (< aliquis + unus) is a weak NPI. French aucun, on the other hand, is used only very occasionally in [+non-aff, −neg] and mostly in [+non-aff, +neg] contexts. It even gives rise to a negative interpretation in isolation in one-word answers. (71)

Quels gâteaux as-tu achetés? – Aucun. which cakes have you bought – any/none ‘Which cakes did you buy? – None.’

Generally, NPIs have a tendency to change into n-words, diachronically. Elements that are licensed by a specific feature such as negation may over time become associated with that feature. This has for instance also been observed for the English expression Jack shit that used to be licensed by negation but by now may suffice on its own to identify negation: (72)

You (do not) know Jack shit. ‘You do not know anything’

With the great increase of dehein in MHG, the second indefinite NPI determiner einig is virtually lost. It is not even mentioned in the chapters on indefinites in standard MHG grammars such as Penzl (1989), Mettke (1989), Paul (1998), or de Boor/Wisniewski (1998). In the MHG dictionary by Benecke/Müller/Zarncke (1854: 424b, 15), there is only an entry for adjectival einig, or einec respectively. Here, they explicitly state that the second use that this word had in OHG, viz. that of an indefinite pronoun, is not found in MHG and only ‘returned’ in Modern German einige. On the other hand, Lexer (1872: 525, 27) in his MHG dictionary has a seperate entry for einic with the meaning ‘irgendein’ (‘any(one)’). Thus the indefinite einig was presumably not entirely lost in MHG but hardly ever appeared at all in contrast to the omnipresent dehein/kein.

S F O PRO



1st proofs

 Agnes Jäger

In my MHG corpus of negated clauses, einig is found only once and in arguably adjectival use cf. (74). As with dehein/kein, I therefore additionally checked the entire texts of Nibelungenlied, Prose-Lancelot and Berthold’s “Four Sermons”. There are only very few examples of einig. The clear cases are all instances of adjectival einig ‘unique/single’ cf. (73), so that in the ambiguous cases, cf. (74), einig is also more likely to be an adjective: (73)

daz ir iemer ein einigez wort von in gelernet. that you never a single word from them learn ‘that you never learn a single word from them’ Bert 3, 770

(74)

und hâst aller wâren riuwe niht sô vil als einigen tropfen. and have all true remorse not as much as single drop ‘and you feel not as much as a single drop of remorse’ Bert 1, 198

My data thus support the picture that the second original NPI determiner einig was lost or extremely marginalised by the MHG period. This development is surprising in so far as, of the two pattern C determiners, einig was mainly found in [+non-aff, +neg] contexts in OHG, whereas dehein(ig) was predominantly used in [+non-aff, −neg] contexts. Accordingly, einig might have been more prone to develop into an n-word than dehein. It was probably due to the support from the parallel form nehein (that may even partly constitute another source for kein as discussed above) that dehein could supercede einig. However, remains of the indefinite NPI determiner einig are still found as late as the Early New High German cf. (75) and even the Early Modern German period cf. (76):28

.

(75)

es weisz Reinhart meher von streits übung wan einicher ritter der ganzen welt it knows Reinhart more of fight practice than any knight the whole world ‘Reinhart knows more about fighting practice than any knight in the world’ Hutten (quoted after Grimm 1862: 210)

(76)

dasz sie weder an leibe noch vermögen einige kränkung erdulden sollen that they neither at body nor property any loss suffer shall ‘that they shall not suffer any loss concerning their body or property’ Goethe (quoted after Grimm 1862: 210)

Conclusion: Diachronic development

Compared to Modern German, the system of indefinite nominal determination was more differentiated in earlier stages of the language. In OHG and MHG all

S F O PRO

. The Modern German einige ‘a few/several’ is evidenced with that meaning since around 1700 cf. Pfeifer (1989: 341). It is not restricted to [+non-aff] contexts.



1st proofs

German indefinite determiners in the scope of negation 

four syntactic patterns of indefinite NPs in the scope of negation that are evidenced typologically are attested: ● ● ● ●

A: bare NP/no determiner B: ein C: einig, dehein(ig)/kein D: nihein(ig)/nehein

restricted to [+non-aff, ± neg] contexts restricted to [+non-aff, + neg] contexts

In particular, OHG and MHG differ from Modern German in that they possessed specific pattern C determiners viz. einig and dehein(ig)/kein. The classification especially of dehein(ig)/kein as an original NPI explains why it appears to be ‘negative’ at times and ‘positive’ at others, a fact that has been little understood so far. This classification makes possible a unified account of this indefinite determiner. Einig and dehein(ig)/kein are licensed by a number of contexts of which sentential negation is one and which are known to license also other (weak) NPIs. In so far, their distribution is quite regular and expected. It is not a random collection of ‘positive’ contexts that they occur in, but only weak NPI contexts. With the grammaticalization of determiners, pattern A decreases diachronically. The original pattern D determiner nihein/nehein, which is also morphologically neg-marked, becomes extinct. The increasingly used pattern C determiner dehein/ kein takes over, supercedes einig, and eventually turns into a pattern D determiner: dehein/kein changes from an NPI, for which negation obviously already formed one important licensing context, into an n-word. Thus Modern German, unlike Modern English, ended up with a system of indefinite nominal determination in the scope of negation without a pattern C determiner, that is a specific non-affirmative or NPI determiner. One may think of the development that took place once dehein had superceded einig as a kind of pull chain: as the pattern D determiner disappears from the system, it pulls the former pattern C determiner dehein/kein after it from left to right in terms of Table (1). Through the common use in a specific context, viz. in the scope of negation, an element, in this case the NPI dehein/kein, becomes diachronically associated with the respective functional feature, i.e. [+neg]. Here, we are dealing with a second kind of grammaticalization phenomenon in relation to negation that is quite independent from that of Jespersen’s cycle: the existence of NPI- or n-word determiners is not linked to the phrasal or head status of the negative particle. One similarity is of course that also through Jespersen’s cycle, former NPIs, especially minimal unit expressions, may become associated with a [+neg] feature, viz. turn into neg-particles, as in the well-known case of French pas. The described change from NPI- to n-word determiners appears to be unidirectional just as other cases of grammaticalization: the licensing contexts of an item are narrowed down until only negation remains. At the same time, that lexical item is enriched with formal features.

S F O PRO



1st proofs

 Agnes Jäger

The fact that determiners may change from NPIs to n-words, on the one hand constitutes a challenge to this classification: is the respective determiner still an NPI or already an n-word? The decisive criterion here should be whether it can give rise to a negative interpretation in isolation. However, there may well be a period of ambiguity of an item between both categories. On the other hand, the described kind of change supports this classification because without it, one would not be able to capture diachronic developments within the systems of nominal determination such as the ones discussed here for German.

Primary sources Berthold von Regensburg, Vier Predigten Mittelhochdeutsch/Neuhochdeutsch. Werner Röcke (ed.) Stuttgart: Reclam, 1983. TITUS text entry by H. Knaus et al., TITUS version by J. Gippert, Frankfurt/M., 1998–2003. http://titus.fkidg1.uni-frankfurt.de/[= Bert]. Das Nibelungenlied. Paralleldruck der Handschriften A, B und C nebst Lesarten der übrigen Handschriften. Michael S. Batts (ed.) Tübingen: Niemeyer, 1971. TITUS text entry by B. Karner & H. Reichert. TITUS version by J. Gippert, Frankfurt/M., 1998–2003. http://titus.fkidg1.unifrankfurt.de/texte/etcc/germ/mhd/nibelung/nibel.htm [= Nib]. Der althochdeutsche Isidor. Nach der Pariser Handschrift und den Monseer Fragmenten. Hans Eggers (ed.) Tübingen: Niemeyer, 1964. TITUS text entry by P. Fernández Alvarez et al. TITUS version by J. Gippert, Frankfurt/M., 1997–2000. http://titus.fkidg1.uni-frankfurt. de/texte/etcs/germ/ahd/isidor/isido.htm [=Is]. Die lateinisch-althochdeutsche Tatianbilingue Stiftsbibliothek St. Gallen Cod. 56. Achim Masser (ed.) Göttingen: Vandenhoeck & Ruprecht, 1994. [=T]. Lancelot und Ginover I. Prosalancelot I. Nach der Heidelberger Hs. Cod. Pal. germ. 147. Reinhold Kluge (ed.), erg. durch Hs. Ms. allem. 8017–8020 Bibl.de l’Arsenal Paris. Hans-Hugo Steinhoff (ed.) Frankfurt/Main: Deutscher Klassiker Verlag, 1995. TITUS text entry by H. Knaus et al., TITUS version by J. Gippert, Frankfurt/M., 1999–2003. http://titus.fkidg1. uni-frankfurt.de/texte/etcc/germ/mhd/proslan1/prosl.htm [= Lanc] Notker der Deutsche: Der Psalter. Petrus Wilhelmus Tax (ed.) Tübingen: Niemeyer 1979. [=N]. Otfrid von Weissenburg: Evangelienbuch. Johann Kelle (ed.) Regensburg: Manz, 1856. TITUS text entry by R. Schuhmann et al. TITUS version by J. Gippert, Frankfurt/M., 1997–2003. http://titus.fkidg1.uni-frankfurt.de/texte/etcs/germ/ahd/otfrid/otfri.htm. Otfrid von Weissenburg: Evangelienbuch. Wolfgang Kleiber (ed.) Vol. 1. Tübingen: Niemeyer, 2004 [=O]. Tatian: lateinisch und altdeutsch. Eduard Sievers (ed.). 2nd ed., Paderborn et al.: Schöningh, 1892. TITUS text entry by P. Fernández Alvarez et al. TITUS version by J. Gippert, Frankfurt/M., 1997–2003. http://titus.fkidg1.uni-frankfurt.de/texte/etcs/germ/ahd/tatian/tatia.htm

References Abraham, Werner. 1997. The interdependence of case, aspect and referentiality in the history of German: The case of the verbal genitive. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 29–61. Cambridge: Cambridge University Press.

S F O PRO



1st proofs

German indefinite determiners in the scope of negation  Bech, Gunnar. 1964. Zur Etymologie des althochdeutschen Pronomens dehhein. Studia Neophilologia 36: 211–216. Behaghel, Otto. 1913. Von deutschen Bindewörtern. Wissenschaftliche Beihefte zur Zeitschrift des allgemeinen deutschen Sprachvereins 5/36: 165–182. Benecke, Georg Friedrich, Müller, Wilhem & Zarncke, Friedrich. 1854. Mittelhochdeutsches Wörterbuch. Vol. I. Leipzig: Hirzel. Braune, Wilhelm & Reiffenstein, Ingo. 2004. Althochdeutsche Grammatik. 15th edn. Vol. 1: Laut- und Formenlehre. Tübingen: Niemeyer. Danielsen, Niels. 1968. Die negativen unbestimmten Pronominaladjektive im Althoch- und Mittelhochdeutschen. Zeitschrift für deutsche Sprache 24: 92–117. de Boor, Helmut & Wisniewski, Roswitha. 1998. Mittelhochdeutsche Grammatik. 10th edn. Berlin: Walter de Gruyter. Donhauser, Karin. 1996. Negationssyntax in der deutschen Sprachgeschichte. Grammatikalisierung oder Degrammatikalisierung? In Deutsch – typologisch, Ewald Lang & Gisela Zifonun (eds), 201–217. Berlin: Walter de Gruyter. Erben, Johannes. 1950. Syntaktische Untersuchungen zu einer Grundlegung der Geschichte der indefiniten Pronomina im Deutschen. Beiträge zur Geschichte der deutschen Sprache und Literatur 72: 193–221. Eroms, Hans-Werner. 1993. Der indefinite Nominalnegator kein im Deutschen. In Studien zur Syntax und Semantik der Nominalgruppe, Marcel Vuillaume et al. (eds), 1–18. Tübingen: Narr. Gerring, Hugo. 1927. Die unbestimmten Pronomina auf –ein im Alt- und Mittelhochdeutschen bis zum Anfang des 14. Jahrhunderts. Uppsala: Almquist & Wiksell. Grimm, Jacob & Grimm, Wilhelm. 1862. Deutsches Wörterbuch. Vol. III. Reprint 1984. München: Deutscher Taschenbuch Verlag. Haspelmath, Martin. 1997. Indefinite pronouns. Oxford: Oxford University Press. Jacobs, Joachim. 1991. Negation. In Semantik: Ein internationales Handbuch der zeitgenössischen Forschung, Arnim von Stechow & Dieter Wunderlich (eds), 560–596. Berlin: Walter de Gruyter. Jäger, Agnes. 2005. Negation in Old High German. Zeitschrift für Sprachwissenschaft 24: 227–262. Karg-Gasterstädt, Elisabeth, Frings, Theodor & Große, Rudolf. 1971. Althochdeutsches Wörterbuch: Auf Grund der von Elias von Steinmeyer hinterlassenen Sammlungen. Vol. 3. Berlin: Akademie-Verlag. Kelle, Johann. 1881 Glossar der Sprache Otfrids. Regensburg: Manz. Kürschner, Wilfried. 1983. Studien zur Negation im Deutschen. Tübingen: Narr. Ladusaw, William Allen. 1979. Polarity Sensitivity as Inherent Scope Relations. PhD Dissertation, University of Texas, Austin (published in 1980 by Garland Press, New York). Laka, Miren Itziar. 1990. Negation in Syntax: On the nature of functional categories and projections. PhD Dissertation, MIT. Leiss, Elisabeth. 1994. Die Entstehung des Artikels im Deutschen. Sprachwissenschaft 19: 307–319. Leiss, Elisabeth. 2002. Explizite und implizite Kodierung von Deontizitaet und Epistemizitaet: über die grammatische Musterbildung vor der Entstehung von Modalverben. Jezykoslovlje 3: 69–98. Lexer, Matthias. 1872. Mittelhochdeutsches Handwörterbuch. Vol. I. Leipzig: Hirzel. Lloyd, Albert Lancaster, Springer, Otto & Lühr, Rosemarie. 1998. Etymologisches Wörterbuch des Althochdeutschen. Vol. 2. Göttingen: Vandenhoeck & Ruprecht. Lockwood, William Burley. 1968. Historical German Syntax. Oxford: Clarendon Press.

S F O PRO



1st proofs

 Agnes Jäger Mettke, Heinz. 1989. Mittelhochdeutsche Grammatik. 6th edn. Leipzig: Bibliographisches Institut. Oubouzar, Erika. 1992. Zur Ausbildung des bestimmten Artikels im Althochdeutschen. In Althochdeutsch: Syntax und Semantik: Akten des Lyonner Kolloquiums zur Syntax und Semantik des Althochdeutschen, Yvon Desportes (ed.), 71–87. Lyon: Univ. Lyon. Paul, Hermann 1998. Mittelhochdeutsche Grammatik. 24th edn. Tübingen: Niemeyer. Penka, Doris & Stechow, Arnim von. 2001. Negative Indefinita unter Modalverben. In Modalität und Modalverben im Deutschen, Reimar Müller & Marga Reis (eds), 263–286. Hamburg: Buske. Pensel, Franzjosef. 1981. Die Satznegation. In Zur Ausbildung der Norm der deutschen Literatursprache auf der syntaktischen Ebene (1470–1730): Der Einfachsatz, Gerhard Kettmann & Joachim Schildt (eds), 287–326. Berlin: Akademie-Verlag. Penzl, Herbert. 1989. Mittelhochdeutsch. Eine Einführung in die Dialekte. Bern: Peter Lang. Pfeifer, Wolfgang. 1989. Etymologisches Wörterbuch des Deutschen. Berlin: Akademie-Verlag. Philippi, Julia. 1997. The rise of the article in the Germanic languages. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 62–93. Cambridge: Cambridge University Press. Weiß, Helmut. 1998. Syntax des Bairischen: Studien zur Grammatik einer natürlichen Sprache. Tübingen: Niemeyer. Weiß, Helmut. 2002. Indefinite pronouns: Morphology and syntax in cross-linguistic perspective. In Pronouns: Grammar and representation, Horst Simon & Heike Wiese (eds), 85–107. Amsterdam: John Benjamins. Wouden, Ton van der. 1997. Negative contexts: Collocation, polarity and multiple negation. London: Routledge.

S F O PRO



1st proofs

The functional range of bare singular count nouns in English Laurel Smith Stvan University of Texas at Arlington

One overlooked and highly polysemous English noun phrase form is the bare singular, i.e. a null determiner with a singular count noun complement. Occurring in all grammatical positions, this constituent shape is used in English for multiple functions. Examination of naturally occurring English data shows the conditions under which bare singulars are used as generics (with a meaning like bare plurals), as components of a predicate conveying a stereotypical activity (with an indefinite meaning), and as markers of an identifiable referent (like nouns with definite articles, demonstratives, and possessive determiners). While generic and indefinite meanings are well documented for bare forms, the reading that picks out an identifiable referent is unexpected for a bare nominal. This range of meaning-based distinctions suggests additional theoretical consequences for cross-linguistic noun interpretation and DP-internal syntactic structure.1

Introduction Languages may allow several different articles within the determiner slot of a noun phrase constituent—typically these could be a definite, an indefinite, or a null/zero form. But these three choices of grammatical shape map onto more than three noun phrase functions. One illustration of this is the overlooked and highly polysemous nominal type in English of the bare singular count noun, i.e. a null determiner with only a singular count noun in its complement. In this chapter, my goal

. Earlier versions of the present material on generic expressions and semantic classes were presented at the Linguistic Society of America annual meetings in Los Angeles in 1999 and in Chicago in 2000, and at the Nominal Incorporation and Its Kind conference in Ottawa in 2006.

S F O PRO



1st proofs

 Laurel Smith Stvan

is to attest to the existence of this form in English, to point out its distribution within a clause, and then, most importantly, to examine the range of functions displayed by this unexpected form. My focus here is on the semantics and pragmatics of the form, leaving open the issue of the nominal’s internal syntax. Section 2 details the location count nouns found in bare singular form. Section 3 discusses the syntactic distribution of bare singulars, noting their occurrence as subjects, direct objects, and objects of prepositions. Section 4 points out the differences in meanings created in those cases where bare singular uses are found, and also shows tests for contrasting these different implicated meanings. Section 5 discusses the distribution of nouns used for the different meanings. Section 6 suggests consequences of these meaning differences, first as they might be explored cross-linguistically, and then by considering some ways that syntactic theories of nominals might be drawn upon to account for the different readings.

.

The absence of articles in English noun phrases

Bare noun forms have been previously discussed (e.g. Pelletier 1974, 1975; Pelletier and Schubert 1989; Carlson 1977; Longobardi 1994, 2003; and Zamparelli 2002), but such works typically focus on plural and mass nouns. Bare singular count forms, however, have been mostly unnoticed, or even ruled to be impossible in many descriptions of English nominals (e.g. Christopherson 1939; Vangsnes 2001; Krifka 2003). Some works, however, have pointed out the sharply limited occurrence of bare count forms in English (Stvan 1993, 1998; Soja 1994; Ross 1995, and Carlson 2003). Ross (1995: 403) for example, notes the unexpectedness of these forms in English when he includes them as one kind of “defective noun phrase.” Indeed, in traditional English grammars, outside of set, idiomatic expressions, the zero form of the article was thought to occur only with plurals, mass or abstract nouns, or proper names, but not with singular count nouns (cf. Azar 1981; Quirk et al. 1985; Biber et al. 1999). This can be seen by the typical uses of the zeroform in (1), here appearing in prepositional phrases (PPs). While the bare use is acceptable with the forms in (a) through (d), the bare singular count noun in (e) is ungrammatical. (1)

a. “In space no one can hear you scream.” b. in sickness and in health c. I have put stickers on cars for six people so far. d. The bee landed on Janine. e. *I put the book on desk.

(mass) (abstract) (plural count) (proper name) (singular count)

S F O PRO



1st proofs

Bare singular count nouns in English 

One set of location noun phrases, however, is unusual, for although they contain count nouns, they regularly appear in singular form, yet without any article. Though limited in range, this bare singular form shows productivity in certain domains, including recording expressions (on record, on video, on film), framing expressions (on camera, out of frame, on stage) and temporal interruptions (on break, on vacation, at lunch, at recess). The largest set in my data, however – and the set that shows distinct meaning contrasts when in bare form – is that involving PPs naming social or geographic places (e.g. at camp, in church, at school, in prison, at home). The current set of attested locations from this latter group is illustrated in Table 1. Table 1. Bare singular social and geographical places on/off base in/to/out of bed at/to/from/in/into/toward camp close to/off/on/to campus down cellar at/in/to chapel at/in/from/to church in to class in clinic at/from/in/out of/to college in country at/in out of court [legal] at/in court [royal] at/in/to daycare on deck in district in dock in hall in harbor

Up hill at home In hospital up/down island in/to/into jail in/to kindergarten in kitchen in/on line to market in/after/to meeting after/out of office at pasture off planet in/into port off post from/in/out of/to prison on property down/up river at/in/from/outside/to/ toward school

at/to sea from/off/on shore on site at/in/to seminary up slope down/on stage in/down/out of state down/up stream in studio in/to synagogue at/to table to temple in theater across/around/down/in/ into/out of/outside/ through/to/toward/up Town at/to university at/from/to work off world in yeshiva

Data used as illustration here have been opportunistically gathered from naturally occurring uses of American and British written English from the 19th to 21st centuries. (The same categories are found in both regional varieties, although a couple of bare location forms, notably in hospital and at university, are specific to British dialects.) Though bare singulars are most frequently found as objects in locative PPs, these nominals can also appear in subject and object positions. Examples of bare forms in each of these three grammatical positions are discussed in the next section.

S F O PRO



1st proofs

 Laurel Smith Stvan

. .

Syntactic distribution of bare singulars Bare singulars as objects of prepositions

The irregular distribution of bare singulars has led some authors to claim that English bare singular count forms do not occur at all in certain syntactic positions (cf. Werth 1980: 251; Behrens 1995: 48). Indeed, their most frequent occurrence is within a prepositional phrase, as seen in the Examples in (2). Even here, however, the position is not fully productive. The (a) and (b) Examples in (2) are typical of the many such uses that are considered ill-formed, while (c) and (d) show that grammatical uses are found as well: (2)

a. *The children were in park. b. *The customers waited in restaurant. c. The rep arrived in town with a heavy pack of excuses but no artwork. (William Fisher, “When a Firm Unreels Secrets of Soviet Joint Ventures,” Wall Street Journal, Aug. 28, 1989) d. The following week should work. Which days will you be on campus? (G. Ward, email correspondence, Aug. 25, 1996)

In PPs, these bare forms are most often found following in, on and at, but other prepositions are attested as well, as seen above in Table 1. .

Bare singulars as subjects

Bare singulars do not show up only in PPs. In (3), examples of bare singulars are shown in subject position. Again, although location nouns are frequently used in bare singular form, (a) and (b) show that not all such nouns are acceptable in these positions; yet those sentences with bare singulars in (c) and (d) are quite well-formed. (3)

a. *Library was busy today. b. *Store closed early on Mondays. c. Well, your winter of ‘35 was a mild winter . . . School never closed and I don’t believe we got more than forty inches of snow. (Garrison Keillor, Lake Wobegon Days, New York: Viking, 1985, p. 209) d. I started going to church for the first time in my life about four years ago. Since then, church has become very important in my life. (Stephen Collins quoted in article by Bart Mills, “All That Love Leaves Cast in Hug ‘Heaven,’” Chicago Tribune, Dec. 24, 1996, Sect. 5, p. 5)

The occurrence of bare noun forms is more limited in this position. Of the 56 institutional locations collected, for example, only the following 10 were found as subjects in my data: bed, camp, church, college, court (legal), home, jail, prison, school, and town.

S F O PRO



1st proofs

Bare singular count nouns in English 

.

Bare singulars as direct objects

As a third position, bare singulars can also appear as direct objects – as shown in Example (4). Again, (a) and (b) show that many count nouns cannot occur here in bare form, but (c) and (d) show examples of those that can. (4)

a. *Do you need help setting up tent? b. *Do you plan to join club? c. Need help setting up camp? (Bill Amend, Fox Trot, comic strip, Aug. 15, 1996) d. They – whoever they is – think I’ve left town and I want to keep it that way. (Tony Hillerman, The Fly on the Wall, 1971, New York: Harper Collins, p. 171)

Those verbs that take such bare objects are limited. Fifteen attested English examples of bare direct objects showed up in my data, some collocating with more than one verb. These are shown in (5). (5)

a. b. c. d. e. f. g. h. i. j. k. l. m. n. o.

break/pitch/set up camp leave campus attend/cut short/skip church survive court save downtown start high school leave home leave hospital break/leave jail enter kindergarten enter/face/flee/mention prison attend/hate/invent/like/miss/start/teach school enter seminary leave/visit town leave work

Collectively, the range of grammatical positions illustrated in Examples (2) through (5) shows that bare singulars have the distribution of a full constituent – an NP or DP – rather than simply that of a count noun. However, because the bare singular count noun is so marked as a noun phrase type, parsing programs need explicit guidance to properly mark the P+N structures. Baldwin et al. (to appear) suggest ways to do this based on the semantics of the bare singular nouns involved. A subset of work has also noted the presence of bare singulars in many languages in coordinated constructions—which can then appear, when paired up, as subject, object or PP as well. As early as 1946, bare constructions in English were observed consisting of “groups of nouns, either in enumerations or when the words are

S F O PRO



1st proofs

 Laurel Smith Stvan

arranged in pairs connected by a conjunction or a preposition” (Ahlgren 1946: 190). More detailed studies have been done by Heycock and Zamparelli (2003) and Roodenburg (2004), that look at data from a range of Romance and Germanic languages, focusing especially on data in Italian and French, though English also shows this construction, as seen in Example (6). In (6), (a) shows conjoined forms as subject, (b) shows conjoined predicate nominals, and (c) shows serial PPs. A contrast is shown in (d), whereby one bare form alone is ill-formed, although the string of them in (c) is fine. (6)

a.

[Goblet and spoon] were set on the right of the plate. (Heycock and Zamparelli (2003) Ex. (6) p. 445) b. Tom was both father and mother to the child. c. Since 1987, it’s been a Performance Today tradition to journey through the country, from church to auditorium to town hall, sharing in celebration of the holidays. (http://www.npr.org/programs/pt/command/CDs.html) d. *It’s been a tradition to journey through the country to auditorium.

Aside from conjoined count nouns, although certain locative PP forms may more easily allow bare singular objects, there are still constraints on what count nouns can occur as bare singulars. Yet, even when focusing only on locative expressions, the high number of constructions found containing bare singulars suggests that they are more than just idiomatic. For example, in examining several semantic domains of PP expressions, I found 82 types of bare singulars in a collection of over 800 tokens of naturally occurring uses. It is the semantics of the nouns and the pragmatics of the entire utterance however, that constrain when such bare forms are allowed. While Section 5 will discuss the semantic limits on the nouns, in the next section, I will first examine those contexts under which bare singular count forms are felicitously used.

.

Bare singular uses

Many authors, in both syntactic theory and language pedagogy, have suggested that there should be a one-to-one correlation between a noun phrase’s morphosyntactic form and its referential uses. For example, Christophersen (1939: 81) notes that it is often believed that “the shifting between zero-form, a-form, and the-form is as regular as a case-distinction.” If the form predicts the function, this approach would indicate that the choice of intended referring use is all that is required in deciding among article forms. Likewise, the article itself is often taken to indicate how a nominal can be interpreted. Vangsnes (2001: 253) claims that “definite determiners entail that the noun phrase is ‘discourse anaphoric,’” that is,

S F O PRO



1st proofs

Bare singular count nouns in English 

that its referent is specific, and identifiable to the speaker (Vangsnes 2001: 254). But overtly determined noun phrases may be an insufficient set of forms to account for all the nominals that introduce a referent. Likewise, much research reflects a well-established difference that nominal expressions used as arguments are (i.e. subjects or complements) are D-projections, while nonarguments nominals (such as those with a predicative use) would be N-projections that lack a determiner (Stowell 1991; Longobardi 1994; Radford 1997: 156). In a similar vein, (Krifka et al. 1995: 67) observes, “If a common noun has to be used like a proper noun, as in the kind-referring interpretation, it must somehow be transformed into a member of the NP category” which would be via a plural form or a definite determiner. Such distinctions imply that English referent-evoking nominals must have definite determiners and will not show up in bare forms. Claims of this type, however, underpredict the actual forms of English referring expressions. As I show below, some of the undetermined noun phrases do show the property of identifying a particular referent. Conversely, as will be seen in the next examples, there is no one function matching up with those times when a bare singular form is used. Like many nominal forms, this constituent shape is used in English for multiple functions: examination of naturally occurring English data shows the conditions under which bare singular count nouns are used indexically (with uses much like nouns with definite articles, demonstratives, and possessive determiner present), as components of an adjectival predicate to convey a stereotypical activity (similar in use to nouns with an indefinite article present), and as generics (with uses like bare plurals). Though pragmatically achieved, these are not implicated meanings conveyed via exploiting Grice’s Cooperative Principle in context; instead, as conventional implicatures, their meaning is tied to the bare singular form (Stvan 1993). Examples of the three readings are discussed in (7)–(17). .

Familiarity reading

One meaning conveyed by bare singular noun phrases shows the nominal being used much like a definite referring expression. As shown in (7), Examples (a) through (d) contain a noun phrase that is identifiable to the hearer, and could be paraphrased, as illustrated below each example, by including a definite, possessive, or demonstrative determiner: (7)

a.

b.

Look out America, the world’s smallest con artist is in town. (Billboard for the movie Curly Sue, 1991) = your town/this town/the town of the reader My dad was in town the weekend before my birthday at the DEMA Show at the Anaheim Convention Center.

S F O PRO



1st proofs

 Laurel Smith Stvan

c.

d.

(http://www.net-quest.com/~chasbo/) = my town/your town/my father’s town I work at home, and I have found that this arrangement has a tremendous potential for personal growth, because nobody will notice I if you eat as many as 20 lunches per day. (Dave Barry, usenet group clari.feature.dave_barry, May 9, 1992) = my home Mr. Thompson [the principal] sometimes prevails upon the parents to come to school and do the paddling, which is limited to two licks with a nine-inch long oval wooden paddle, kept in the school office. (Susan Sheehan, “Kid, Twelve,” The New Yorker, Aug. 19, 1996, p. 56) = this school/the principal’s school

As illustrated in (7), this first interpretation, which I refer to as a familiarity reading, involves the noun phrase being used not merely to indicate a location, but used more precisely to pick out a particular, known location. This meaning is created by a conventional (rather than conversational) implicature that keys into shared knowledge about how each participant in the discourse (speaker, hearer or locatum) is related to the location (Stvan 1993); thus, while in both Examples (7a) and (7b), for example, the referent of the subject phrase can be interpreted as being located in some town, in (7a) the town referred to is, in particular, the town of the reader; while in (7b) it could be that of the speaker or of the 3rd person subject as well. Beyond sharing a meaning equivalent to that related by the boldfaced examples containing definite determiners, further verification that the reading in (7) involves definite reference comes from anaphora. The referent evoked by the bare form can be previously introduced with an indefinite form, showing that the bare forms function as second-mention referring expressions, just as noun phrases with definite determiners do. For example, contrast Examples (a)–(d) in (7) with (a)–(d) in (8), respectively: (8)

a. b. c. d.

[Once you arrive in [a town]i, you find that] the world’s smallest con artist is in [town]i. [I just moved to [a new town]i last year.] My dad was in [town]i the weekend before my birthday at the DEMA Show at the Anaheim Convention Center. [I have both an office and [a private residence]i.] I work at [home]i. Mr. Thompson began work as a principle at [a school with lots of discipline problems]i. Mr. Thompson sometimes prevails upon the parents to come to [school]i and do the paddling.

Additional evidence of the ability of bare singular count forms in the familiarity reading to pick out a known referent is seen in the way that they are interpreted as having the definiteness of a proper name. One possible explanation might be

S F O PRO



1st proofs

Bare singular count nouns in English 

that this use represents a diachronic transition from being a proper noun. This suggested origin is supported by the orthography of a few examples in which it is possible, even into the 21st century, to find bare singular tokens in capitalized form – as seen in the three quotes in (9). (9)

a.

b.

c.

Knocking again at Miss Lydgate’s door shortly before Hall, she suddenly remembered that the English Tutor has said she was leaving before lunch to spend a couple of nights in Town. (Dorothy L. Sayers, Gaudy Night, 1936, p. 84) This was the news with which Miss Hillyard had had to face Miss Lydgate when the latter returned to College immediately after breakfast on the Monday. (Dorothy L. Sayers, Gaudy Night, 1936, p. 85) If the installation of in-street pedestrian crossing signs is approved by the Council tonight, we will use these guidelines to evaluate conditions at other unsignalized crosswalks in Town. (Chapel Hill, North Carolina town council memo, 23rd of May, 2005 http://townhall.townofchapelhill.org/agendas/ca050523/)

This capitalization of the nouns seems to be one way for writers to convey the knownness of the location referent, signaling that its indexicality is anchored off either the reader or the locatum. While most bare singular nouns used for the familiarity reading are not capitalized, they do still show this definiteness. .

Activity reading

In addition to the familiarity reading seen in Examples (7) through (9), bare singulars are found with a second interpretation, which I call the activity reading, in which there is no assertion of locative placement. Unlike the Examples in (7), for example, where the boldfaced noun phrase is used only to refer to a place, in the Examples in (10), it is information about the subject’s activity that is conveyed. That is, the speaker of (10a) asserts that the father is actually serving time as a prisoner, in (10b) the companies are asserted to have been engaged in a legal trial, etc. (10)

a.

b.

Two are currently in foster care – one girl because her father is in prison for murdering her mother; another girl spent last year in foster care. (Susan Sheehan, “Kid, Twelve,” The New Yorker, Aug. 19, 1996, p. 54) = serving time as a prisoner ≠ in the/his prison Off and on since then, the companies have skirmished in court. (Jeffrey A. Tannenbaum, “Tiny Firm Faces Legal Might of Wrathful Multinational,” Wall Street Journal, Oct. 16, 1989) = taking part in a trial ≠ in the/their court

S F O PRO



1st proofs

 Laurel Smith Stvan

c.

d.

So we know that the Emperor began by treating the English with benign condescension, but in August, 1793, while the embassy was still making its way to the capital – it had landed in June, after nine months at sea – his attitude became hostile. (Naomi Bliven, The New Yorker, January 11, 1993, p. 105) = sailing/traveling the ocean ≠ at this/my sea My husband is in prison – I visit him every day. (Ken Follet, Pillars of the Earth, 1989, p. 370) = being held for committing a crime ≠ in the/his prison

As with bare singulars used as direct objects (seen in 4), the PP examples showing an activity reading conveyed by bare singulars involve the whole PP constituent as a predicate. Evidence that the noun is functioning within a predicate includes the synonymous replacement by a VP, shown in boldface. Further, in many cases synonyms of versions of some of the P + N predicates are available as single words, though some are now found only in more archaic English usage, e.g. abed, asea, encamped, hospitalized, imprisoned, etc. In addition, nouns in PPs with this activity reading show an inability to serve as the antecedent for a pronoun co-referring to a particular specified place: (11)

a. “Her father is in prison” . . . b. and he has been there[pro=PP] for six years. c. #and he has been in it[proNP] for six years.

This predicative function of the activity reading is similar to noun incorporation, creating a “fused” multiword predicate that includes the location noun as part of the predicate. The verb plus bare singular object constructions in (4) are also typical of noun incorporation, which commonly involves a V + morpho-syntactically marked direct object form (cf. Mithun 1984; Van Geenhoven 1998). However, due to the paucity of morphological agreement shown by English nouns, it is in their semantic markedness and discourse opaqueness that parallels to incorporation are called to mind for the activity reading (cf. Farkas and De Swart 2003). .

Generic reading

A third use of the bare singular forms in PPs is as a kind-referring expression. This reading is distinct from the activity reading since it does not predicate a state or action of the locatum, yet the generic reading is also distinct from the familiarity reading in not linking the named location to a specific discourse participant anchor. For instance, in the Examples in (12), the writer using each PP makes a claim about the kind of place named by the noun. The genericity of this reading

S F O PRO



1st proofs

Bare singular count nouns in English 

can be verified by replacing each singular with a bare plural form, another typical construction that has a generic referring use. As seen in (12), the meaning remains the same after such a substitution: (12)

a.

b.

c.

An amendment to the Senate’s anti-drug bill would have barred alcohol companies from sponsoring any sort of event at all on campus. [= on campuses] (Joanne Lipman, “Alcohol Firms Elude Senate Move To Restrict Presence on Campus,” Wall Street Journal, Oct. 6, 1989) “Free speech,” “Question authority,” and “Leave us alone” are now conservative and libertarian battle-cries on campus. [= on campuses] (Alan Charles Kors, “It’s Speech, Not Sex, the Dean Bans Now,” Wall Street Journal, October 12, 1989) The contrast is apparent on campus, too. Unlike their predecessors in the Vietnam era, many ROTC students today can be seen wearing their uniforms and boots – not just to military classes, but all day long. [= on campuses] (Jill M. Bullock, “ROTC Regains Respect on Campuses As Graduates Fare Well in Workplace,” Wall Street Journal, Sept. 11, 1989)

With this substitution test, a contrast is also revealed with bare singular forms that are used for the familiarity reading, such as those in (13). In (13), where the noun phrase is used to indicate a particular, known campus, it is not possible to substitute a bare plural and keep the same meaning. This is illustrated by the examples in square brackets below each example: (13)

a.

b.

c.

Have you been back to campus lately? (UIC Alumni Association membership flyer) [≠Have you been back to campuses lately?] If you are interested in the game, are in a group, are looking for someone on campus to play with, or know someone who is . . . give me a call. (Ad recruiting people to play Magic, posted in a university student union building) [≠If you are interested in the game, are in a group, are looking for someone on campuses to play with, or know someone who is . . . give me a call.] The following week should work. Which days will you be on campus? (G. Ward, email correspondence, Aug. 25, 1996) [≠The following week should work. Which days will you be on campuses?]

As (12) and (13) show, the generic reading of bare singulars exhibit the traditional number neutrality of other generic forms, while the familiarity readings do not. For activity readings, however, depending on the action being predicated of the

S F O PRO



1st proofs

 Laurel Smith Stvan

locatum, the location noun may show number neutrality, as in (14a), or, as in (14b), it may not: (14)

a. b.

Off and on since then, the companies have skirmished in court. [= in courts] My husband is in prison – I visit him every day. [≠ in prisons]

In addition, it should be pointed out that deletion of an article is typically not optional. Different meaning are convey when an article is present, as shown by the familiarity use in (15a), the activity use in (16a) and the generic use in (17a):

.

(15)

a. b.

Is your mother at home? Is your mother at the home?

(16)

a. b.

John is in prison. John is in the prison.

(17)

a. b.

Cigarettes are sold in prison. Cigarettes are sold in the prison.

Count nouns that convey the three readings

The Examples in (7) through (17) show that noun phrases in the bare singular form can be used to convey three kinds of information: (1) they are used to specify the particular location connected to either the speaker, hearer, or located entity; (2) instead of merely naming a location, these bare forms are used to evoke the locatum’s activity as being the one that is associated with that place; or, (3) they are used generically. Not all nouns are found with all of these readings, however. A selection of the nouns found in only the familiarity reading, only the activity reading, or in all three, is shown in Table 2. Table 2. Sample distribution of bare singulars in PPs Activity A

B

C

campus home town work court jail prison sea bed camp church class college school

Familiarity

Generic

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

S F O PRO



1st proofs

Bare singular count nouns in English 

Table 2 reveals that certain locations appear more closely associated with a discourse participant (speaker, hearer, or located third person); when mentioned in the bare form, these places are not used to convey any particular action on the located person’s part. This is seen by the locations in group A, whereby referring to a locatum as on campus, at home, in town or at work only conveys the idea that one of the participants belongs to that place. Group B, on the other hand, shows some of the locations that are not connected to a particular participant, but are instead used to convey what the locatum is doing there. In group C, in contrast, we see that some words can be found with both of these uses. Furthermore, as shown in the rightmost column, all the nouns in bare singular form could be used generically. Thus, by this ability of bare singular forms to convey one, two, or all three of these readings, it becomes clear that English lacks sufficient distinctive morpho-syntactic forms to suggest a one-to-one mapping of nominal syntax and semantics. That is, a generic reading can be conveyed with a bare singular or with a bare plural. At the same time, definite reference can be conveyed with a definite article or without an article, and so on. The presence of the article alone, therefore, does not determine the function of the noun phrase. In English, the bare singular form, especially, is used in a variety of ways (though see Carlson and Sussman (in press) for a discussion of indefinite readings of definite forms). Traditionally, nouns are considered to have denotations, while NPs are used to refer to particular entities. Predicates, however, contrast with referring expressions by naming states or relationships instead. The bare singular noun phrases involving social/geographical expressions are used in both predication and reference. The familiarity use is individual referring; the generic use is kind-referring; while the activity use is not referential at all, but consists of a PP predicate describing the state of the locatum. These different uses of bare singular NPs are displayed to show a continuum of specificity in Table 3. Table 3. Different specificity uses of bare singulars Familiarity Generic Activity

Individual-Referring Expressions (DPs) ∧ Kind-referring Expressions Predicates (PPs) ∨

Most Specific Referent Less Specific Referent Non-Specific Referent

The likelihood of a count noun being found in the bare singular form is partly related to the semantic domain of the noun. The domains listed in (18) for example contain clusters of location count nouns found in bare singular form. (18)

controlled zones: religious settings: educational settings: nautical settings: landscape features:

base, campus, country, district, home, island, planet, post, property, site, state, theater, town, world. chapel, church, seminary, synagogue, temple. campus, class, college, kindergarten, school, university, yeshiva. deck, dock, harbor, port. hill, river, road, sea, shore, slope, stream.

S F O PRO



1st proofs

 Laurel Smith Stvan

However, many count nouns that denote similar places to those shown in (18) are not found in bare singular forms, as seen by the ill-formed PPs in (19): (19)

.

*in pool *in stadium *at theater

*at resort *at office *at station

*at museum *at store *in park

*in apartment *at library *at gymnasium

Cross-linguistic and syntactic ramifications of the different readings

Distinctions equivalent to the interpretations seen in English bare singular noun phrases are available in other languages by means of morpho-syntactic marking other than article deletion. Preliminary examination of translations of the English locative constructions with bare singular count nouns shows disambiguation of activity readings from regular locative constructions (which in English involve determined nouns). For example, the choice of preposition alternations in French and in Dutch (e.g. en prison ‘in prison’ vs. dans la prison = ‘in the prison’) gives options to show the activity reading. Similarly, optional contraction of the preposition and article, in Portuguese and in some dialects of German (e.g. zur Kirche gehen ‘to go to church’ vs. zu der Kirche gehen ‘to go to the church’), provides a way to talk about the activity at a place rather than the place itself. Verb choice can indicate when a PP is predicative rather than locative in Danish (e.g. på skole ‘away at school’ vs. på en skole ‘on a school’s grounds’) with the verbs go or be, but no contrast is seen when used with verbs like see; and case marking in languages such as Korean supplies alternatives that would show the activity vs. a regular locative sense of some nouns (e.g. hakkyo-lul ka-ta school- go- ‘to go (to) school as a vocation, e.g. as student or teacher’ vs. hakkyo-e ka-ta school- go- ‘to go to a school’). Further work is needed to explore the full variety of forms reflecting these meaning differences, but the above selections show some of the ways by which it is possible to reflect this range of noun interpretations. In short, speakers have a number of relationships to set out between location and locatum referents, but in English, a limited set of determiner forms to use when signaling these via noun phrases. Since the familiarity reading shows the nominal functioning as an argument (seen, for example, in (3) by their ability to occur in subject position), while the activity reading functions like a predicate, questions arise as to whether all bare singular count nouns could be said to be syntactic constituents with the same internal structure. I have used the neutral term “noun phrase” here for the highest level of nominal structure, without committing to a structure for the nominal as either NP, small N, or DP, but there are consequences yet to be explored concerning the ways that this range of functionality of the bare singular count forms might

S F O PRO



1st proofs

Bare singular count nouns in English 

shed some light on the structure of nominal constituents. For example, depending on the claim that one makes about the status of a single-word noun phrase, bare singular count nouns might be said to be nouns raised to Det (cf. Longobardi 2003) or they might be considered nouns functioning as complements to a covert determiners with a singular feature. In addition, some researchers suggest a distinction between the referential functions of NPs and DPs (Vangnes (2001), for example, attempts to align KP, NumP, and DP heads with different referential uses; Stowell (1991), Longobardi (1994; 2003) and Zamparelli (2000), too, have looked at parametric distinctions in sub-levels within the noun phrase to predict separate functions for each). Bare singular count nouns, with their distinct meanings, can supply further data to help differentiate noun phrase structures once an acknowledgment of bare singular count nouns in English and their varied referential uses is taken into account.

.

Conclusion

The existence of English count nouns occurring in singular form with no article opens several additional lines of research. Besides clarifying the structural theories mentioned above, these forms are significant in second language teaching – both in order to offer a complete paradigm of noun forms, but more significantly, to illustrate how the choice of the bare form with certain nouns will be taken to convey one of three distinct readings separate from the meaning when an article is present (as Examples (15) through (17) show). To summarize these interpretations, casting location count nouns in the bare singular can be a way for a speaker to predicate the locatum’s involvement in the activity associated with such a place; it can be used in discussing the characteristics of a kind of place; or, it can be used to refer to a particular, known place. While these first two functions are in keeping with known uses of English bare forms, the last use is unexpected for a bare nominal.

References Ahlgren, A. 1946. On the Use of the Definite Article with ‘Nouns of Possession’ in English. Uppsala: Appelbergs. Azar, B.S. 1981. Understanding and Using English Grammar. Englewood Cliffs NJ: Prentice-Hall. Baldwin, T., Beavers, J., van der Beek, L., Bond, F., Flickinger, D. & Sag, I.A. In press. In search of a systematic treatment of determinerless PPs. In Computational Linguistics: Dimensions of syntax and semantics of prepositions, P. Saint-Dizier (ed.). Dordrecht: Kluwer. Behrens, L. 1995. Categorizing between lexicon and grammar: The MASS/COUNT distinction in a cross-linguistic perspective. Lexicology 1(1): 1–112.

S F O PRO



1st proofs

 Laurel Smith Stvan Biber, D., Johansson, S., Leech, G., Conrad, S. & Finegan, E. 1999. Longman Grammar of Spoken and Written English. Harlow: Longman. Carlson, G.N. 1977. A unified analysis of the English bare plural. Linguistics and Philosophy 1: 413–456. Carlson, G. 2003. No lack of determination. In The Second Glot International State-of-the-article Book, L. Cheng & R. Sybesma (eds), 147–178. Berlin: Mouton de Gruyter. Carlson, G. & Sussman, R. In press. Seemingly indefinite definites. In Linguistic Evidence: Empirical, theoretical, and computational perspectives, S. Kespar & M. Reis, (eds). Berlin: Mouton de Gruyter. Christophersen, P. 1939. The Articles: A Study of their Theory and Use in English. Copenhagen: Einar Muskgaard. Farkas, D.F. & De Swart, H. 2003. The Semantics of Incorporation: From argument structure to discourse transparency [Stanford Monographs in Linguistics]. Stanford CA: CSLI Publications. Heycock, C. & Zamparelli, R. 2003. Coordinated bare definites. Linguistic Inquiry 34: 443–469. Krifka, M. 2003. Bare NPs: Kind-referring, indefinites, both, or neither? Paper presented at the Semantics and Linguistic Theory (SALT) 13, Seattle. Krifka, M., Pelletier, F.J., Carlson, G.N., ter Meulen, A., Chierchia, G. & Link, G. 1995. Genericity: An introduction. In The Generic Book, G.N. Carlson & F.J. Pelletier (eds), 1–124. Chicago IL: University of Chicago. Longobardi, G. 1994. Reference and proper names. Linguistic Inquiry 25(4): 609–665. Longobardi, G. 2003. Determinerless nouns: A parametric mapping theory. In From NP to DP, Vol. 1: The syntax and semantics of noun phrases, M. Coene & Y. D’hulst (eds), 239–254. Amsterdam: John Benjamins. Mithun, M. 1984. The evolution of noun incorporation. Language 60(4): 847–894. Pelletier, F. J. 1975. Non-singular reference: Some preliminaries. Philosophia 5: 451–465. Pelletier, F. J. 1974. On some proposals for the semantics of mass nouns. Journal of Philosophical Logic 3(1/2): 87–108. Pelletier, F.J. & Schubert, L.K. 1989. Mass expressions. In Handbook of Philosophical Logic, D. Gabbay & F. Guenthner (eds), 327–407. Boston MA: D. Reidel. Quirk, R., Greenbaum, S., Leech, G. & Svartvik, J. 1985. A Comprehensive Grammar of the English Language. New York NY: Longman. Radford, A. 1997. Syntactic Theory and the Structure of English: A minimalist approach. Cambridge: Cambridge University Press. Roodenburg, J. 2004. French bare arguments are not extinct: The case of coordinated bare nouns. Linguistic Inquiry 35: 301–313. Ross, H. 1995. Defective noun phrases. In Papers from the 31st Regional Meeting of the Chicago Linguistic Society, A. Dainora, R. Hemphill, B. Luka, B. Need & S. Pargman (eds), 398–440. Chicago IL: CLS. Soja, N.N. 1994. Evidence for a distinct kind of noun. Cognition 51: 267–284. Stowell, Tim. 1991. Determiners in NP and DP. In Views on Phrase Structure, K. Leffel & D. Bouchard (eds), 37–56. Dordrect: Kluwer. Stvan, L.S. 1993. Activity implicatures and possessor implicatures: What are locations when there is no article? In Papers from the 29th Regional Meeting of the Chicago Linguistic Society, K. Beals, G. Cooke, D. Kathman, S. Kita, K.-E. McCullough & D. Testen (eds), 419–433. Chicago IL: CLS.

S F O PRO



1st proofs

Bare singular count nouns in English  Stvan, L.S. 1998. The Semantics and Pragmatics of Bare Singular Noun Phrases. Ph.D. Dissertation, Northwestern University. Van Geenhoven, V. 1998. Semantic Incorporation and Indefinite Descriptions: Semantic and Syntactic Aspects of Noun Incorporation in West Greenlandic. Stanford, CA: CSLI Publications. Vangsnes, Ø. A. 2001. On noun phrase architecture, referentiality, and article systems. Studia Linguistica 55: 249–299. Werth, P. 1980. Articles of association: Determiners and context. In The Semantics of Determiners, J. van der Auwera (ed.), 250–289. Baltimore & London: University Park Press & Croom Helm. Zamparelli, R. 2000. Layers in the Determiner Phrase. New York NY: Garland. Zamparelli, R. 2002. Definite and bare kind-denoting noun phrases. In Romance Languages and Linguistic Theory 2000: Selected papers from ‘Going Romance’ 2000, C. Beyssade, R. BokBennema, F. Drijkoningen & P. Monachesi (eds), 305–342. Amsterdam: John Benjamins.

S F O PRO



1st proofs

S F O PRO



1st proofs

The definite article in non-specific direct object noun phrases Comparing French and Italian Tanja Kupisch University of Hamburg, Germany

Christian Koops Rice University, Houston Texas, USA

Italian allows for the use of the definite article in non-specific direct object NPs (mettersi la giacca ‘put on a jacket’, avere il gatto ‘have a cat’). However, in French, the corresponding constructions typically take only the indefinite article (se mettre un blouson, avoir un chat). We present a corpus analysis and a questionnaire study which both establish a striking contrast between French and Italian on this point. We argue that the more widespread use of the Italian definite article in this non-specific function shows that it is further grammaticalized than its French counterpart. This conclusion calls for a reconsideration of the widespread view of French as the language with the more grammaticalized article system.1

Introduction This paper is concerned with the use of the definite article in a particular type of direct object noun phrase.2 For example, consider how the English noun phrase a car in (1) translates into Italian and into French, in (2a) and (2b). (1)

If Mario had a car, he could drive to Spain.

. This paper was made possible by the following people who assisted us in the data collection and who provided us with acceptability judgments: Mélanie Blome, Katja Cantone, Karin Campagnoli, Ev-Daniela Felis, Marc Hinzelin and his family, Michel Achard, and especially Moïra Berger and Ferdinando Giammichele. We also would like to thank Werner Abraham, Elisabeth Leiss, and especially Elisabeth Stark for their comments on earlier drafts of this paper. All remaining errors are our own. . Throughout this paper we use the term noun phrase (NP) as a theoretically neutral term to cover what is today referred to as noun phrase or, since Abney’s (1987) proposal of the DP-hypothesis, determiner phrase.

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

(2)

a. b.

It. Se Mario avesse una/la macchina, potrebbe andare in Spagna. Fr. Si Mario avait une/*la voiture, il pourrait aller en Espagne.

Due to the counterfactual context in (1), the NP a car is interpreted as non-specific. That is, no particular car is being referred to; any car would allow Mario to drive to Spain. The translations in (2) bring out an interesting contrast between indefinite and definite article use in Italian and French. While both languages permit use of the indefinite article for this particular function, only Italian allows the definite article here as well. In this paper, we are specifically interested in the possibility of using the definite article in non-specific direct object NPs, as seen in (2a). For expository convenience, we will refer to such NPs as “non-specific definite direct object NPs” or “non-specific definite direct objects”, thus leaving implicit the fact that we are concerned only with NPs which include the definite article (rather than definite NPs in general). We approach this phenomenon from two vantage points. First, we provide a grammaticalization-oriented perspective. Our analysis is based on the observation that indefinite and definite articles show partially parallel effects with respect to their occurrence in specific and non-specific NPs during grammaticalization. It is well known that the diachronic development of indefinite markers is characterized by an early restriction to specific NPs followed by a gradual expansion to non-specific NPs (for example, Givón 1981; Heine 1997; see also Blazer 1979 and Stark 2002 for studies of French and Italian). In the case of definite markers, this development has not been attested to the same extent, but Greenberg’s (1978) classic study of the “cycle of the definite article” suggests an analogous grammaticalization scenario. It shows that definite markers may also, eventually, spread to non-specific contexts. We therefore interpret the option of using a definite article with non-specific direct objects, as seen in Italian in (2a), as an indicator of an advanced degree of grammaticalization. Second, we specifically contrast Italian with French in regard to the availability of the definite article in non-specific direct object NPs. We choose these particular languages because while this phenomenon appears to be more common in Italian than in French (cf. [2a]), French is widely seen as the language with the more developed article system (Harris 1980; Longobardi 1999 inter alia). The reason for the latter view is the high degree of obligatoriness of articles (and determiners in general) and the almost complete absence of bare noun contexts in French. Indeed, there are grammatical contexts which can remain bare in Italian but are obligatorily determined in French. However, we argue that these contrasts are irrelevant concerning the distribution of the definite article. Where such contrasts exist, bare nouns in Italian correspond to indefinite marked NPs in French, and therefore reflect only the lower degree of obligatory indefinite mark-

S F O PRO



1st proofs

The definite article in non-specific object NPs 

ing in Italian. If each article type is considered independently, the Italian definite article appears further grammaticalized than its French counterpart. The crucial piece of evidence for this claim comes from its greater use with non-specific direct objects. We begin our examination of non-specific definite direct objects with a discussion of their referential status, their relation to generic NPs, their productivity, and their general occurrence in Italian and French in Section 2. Section 3 outlines aspects of the grammaticalization of articles, the role of the specificity distinction in this process, and the relevant grammatical differences between the French and Italian article systems. In Section 4 we present two empirical studies which test article use in non-specific direct objects in French and Italian. In Section 5 the results are further discussed, and Section 6 concludes with a revised view of the article systems of French and Italian.

.

Non-specific definite direct object NPs in Italian and French

The non-specific reading of the Italian definite article can also be observed in the ambiguity of (3). Note how the definite NP la macchina ‘the car’ can have two English translations. (3)

Luca vuole farsi la macchina.

(4)

a. b.

Luca wants to buy the car. Luca wants to buy a car.

The two translations of (3) reveal two contrastive readings. Under the first reading, seen in (4a), la macchina is understood as referring to a particular car (for example, the car which the speaker has already been talking about). Under the second reading, seen in (4b), the meaning of the sentence can be understood without identifying a particular car; la macchina is understood as non-specific. It refers to the category of cars in general, as seen earlier in (2a). Additional examples of the non-specific use of the definite article in Italian direct object NPs are given below. (5)

Dopo la guerra, molte famiglie con un salario basso tuttavia avevano il gatto. ‘After the war, many low income families still had a cat.’

(6)

Ogni giorno dopo la scuola Maria si compra il gelato. ‘Every day after school, Maria buys ice cream.’

(7)

Quando arrivai alla festa, tutti i linguisti portavano la cravatta. ‘When I arrived at the party, all linguists were wearing ties.’

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

The NPs il gatto, il gelato, and la cravatta in (5), (6), and (7) designate a type of pet, snack, and attire, respectively. A specific interpretation is excluded in each case. Interpretations in which the low income families all owned the same cat, Maria bought the same ice cream every day, and all the linguists wore the same tie, are nonsensical. The examples discussed so far have demonstrated that the definite article in Italian object NPs is associated with two clearly distinct functions, which we have designated as either specific or non-specific. This choice of terminology may appear unorthodox because the specificity distinction is traditionally restricted to indefinites. For indefinite NPs, analysts have long distinguished a specific and a non-specific reading, as shown in the ambiguity of classic examples like (8). (8)

Sally wants to marry a Norwegian, a. . . . his name is Per. (specific) b. . . . but she hasn’t found one yet. (non-specific)

Definite NPs, by contrast, are not generally thought of as subject to the specificity distinction. It is probably fair to say that, in practice, many linguists assume a three-way distinction between definite, specific indefinite, and non-specific indefinite NPs, and would consider non-referring definites a phenomenon confined to generic sentences (to be discussed below). This mode of classification can also be found in descriptive grammars. In the Grande Grammatica Italiana di Consultazione, Renzi (1988) divides discussion of the indefinite article into separate sections corresponding to specific and non-specific NPs, whereas nonreferring definite NPs are discussed only in connection with generic sentences. The existence of non-specific definite NPs has, in fact, been questioned altogether. For example, Enç (1991) argues against such a category on theoretical grounds. Thus, there is no consensus about the theoretical status of non-specific definites. Lyons (1999) presents a recent exception to the trend of restricting nonspecificity to indefinite NPs. Lyons argues that parallel specificity effects can be seen in indefinite and definite NPs. He proposes to identify the distinction between specific and non-specific definites with the distinction between the “referential” and “attributive” use of definites, which has been independently discussed in the philosophy of language since Donnellan (1966). We follow Lyons in assuming that the dimension of definiteness cross-cuts the dimension of specificity.3 In fact, we regard the Italian constructions discussed above as

S F O PRO

. Note that this does not imply that non-specific definite and indefinite NPs must be synonymous. As we will discuss in Section 5, they can, and typically do, contrast in meaning.



1st proofs

The definite article in non-specific object NPs 

clear evidence that non-specificity can, indeed, be observed in definite NPs.4 Finally, a practical motivation for us to use the labels specific and non-specific is to draw attention to the parallel course of grammaticalization of indefinite and definite articles (see Section 3). As alluded to above, non-referring definite NPs have been discussed extensively in connection with generic sentences (see Krifka et al. 1995 for a review). We therefore want to point out how the definite direct object NPs which we have called non-specific differ from definite NPs in generic sentences (for a more detailed discussion of this issue, see Korzen 1996, 1998). Using the terminology of Krifka et al., the difference is that the former behave like object-referring rather than kind-referring NPs. This can be observed in upward entailing contexts, where object-referring NPs, but not kind-referring NPs, can be replaced by semantically superordinate NPs. (9)

a. b.

(10)

Berber lions escaped from the zoo Lions escaped from the zoo. (object-referring) Berber lions are extinct // Lions are extinct. (kind-referring) (Krifka et al. 1995: 13)

Ho la Volvo Ho la macchina. (non-specific) ‘I have a Volvo’ ‘I have a car’ (Korzen 1996: 623)

Note that the same test distinguishes non-specific object NPs, as in (10), from kind-referring (generic) object NPs, as in (11). (11)

Odio il Chianti. // Odio il vino. (generic) ‘I hate Chianti’ ‘I hate wine’

. The existence of non-specific definite NPs has also been acknowledged in the discussion of inalienable constructions. For example, Vergnaud & Zubizaretta (1992) propose that the French NP la main in (i) can be interpreted as non-specific in the sense of denoting a type rather than a token (see also Guéron 2003). i.

Les enfants ont levé la main. ‘The children raised their hands’ (Vergnaud & Zubizaretta 1992: 596)

Alexiadou (2003) compares inalienable and alienable possessed nouns in Greek, where they are definite, arguing that the former do not behave as independent DPs syntactically but rather form a complex predicate together with the possessor, while the latter refer independently and form an independent syntactic DP. Although our paper is not directly concerned with the syntax of possessive constructions in this paper, we want to point out that our Italian data do not show a distinction along the lines of alienable and inalienable possession.

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

We therefore distinguish between non-specific and generic definite direct objects. In the remainder of this paper, we will exclusively be concerned with the former.5 Another preliminary issue involves how productively the definite article is used in its non-specific function in Italian. The exact extent of this use is difficult to ascertain. Korzen (1996, 1998), whose examples are based on a large survey of usage, shows that non-specific definite direct object NPs occur in a great number of possessive expressions, such as those with avere ‘have’ in (12), as well as with other verbs, as shown in (13). (The intended readings are, of course, non-specific.) (12)

a. b. c.

avere il pianoforte avere la parrucca avere le spalle larghe

‘have a piano’ ‘have a wig’ ‘have broad shoulders’

(13)

a. b. c.

Vado a comprare il pane. ‘I’ll go and buy bread.’ Ti faccio il caffè. ‘I’ll make coffee.’ Si mette la giacca. ‘He is putting on a jacket.’ (Korzen 1998: 103, 105)

Taking Korzen’s examples to be broadly representative, it appears that nouns which appear in non-specific definite direct object NPs include different types of possessions (incl. inalienable possessions, e.g. body parts), food, clothing, and accessory items.6 The verbs with which they occur reflect these semantic domains. Within these domains, the phenomenon appears to be somewhat productive. Korzen specifically attributes a certain degree of productivity to cases like those in (13). On the other hand, it is easy to show that the non-specific reading is not entirely productive. . It is interesting to relate our observations to Krifka et al.’s (1995: 83ff.) comments on definite marked object NPs such as the grizzly in (i). i.

We filmed the grizzly in Alaska.

Since, as Krifka et al. point out, the grizzly in (i) can be understood as a representative of the whole kind, the authors speak of a “representative object interpretation”. Interestingly, such NPs pattern with object-referring rather than kind-referring NPs in the upward entailment test. Furthermore, Krifka et al.’s suggestion that “languages … seem to differ in their readiness to employ this interpretation” (p. 83) is borne out in the striking degree to which the “representative object reading” has grammaticalized in Italian, as we show in this paper. . As another type of non-specific object NP which takes the definite article, Korzen lists expressions of profession, function, or behavior with the verb fare ‘do’: fare il contadino/il padre/il galante ‘be a farmer/a father/courteous’ (1998: 103). We are not concerned with this type of NP in this paper.

S F O PRO



1st proofs

The definite article in non-specific object NPs 

(14) ??Vado a comprare i kiwi. intended reading: ‘I’ll go and buy kiwis.’

(definite article i)

According to Korzen, what licenses cases like (13) is the “habitual occurrence of the category in question in the described situation” (1998: 105). This formulation would rule out cases like (14), because kiwis are not widely consumed in Italy and therefore do not habitually occur in the buying scenario of (14). The factors which constrain non-specific definite NPs constitute an interesting topic from a semantic and pragmatic perspective, which is clearly in need of elucidation. We will not discuss them in further here because our main concern is not the motivation of the phenomenon per se, but rather the degree to which it exists in Italian and French and the implications of this for the grammaticalization of articles. Nevertheless, we will return to the issue in Section 5 when we are in a position to compare Italian and French on this point. Finally, we want to address the question of whether non-specific definite direct objects should be considered arguments or predicates. The examples show that such NPs occur as direct objects of a variety of transitive verbs (‘have’, ‘buy’, ‘wear’, etc.) rather than occurring with copular verbs. Thus, they are clearly distinct from predicate nominals. This leaves the possibility that such NPs and the verbs with which they are associated constitute unitary, fixed expressions. Under such an analysis, the NPs could be argued to be predicates by virtue of forming part of a larger, predicative unit. However, such an argument would imply that the verb+object complex is not productively constructed. Since the question of productivity has not been resolved, as argued in the previous paragraph, we will not take a stand on the issue here. Having characterized the Italian constructions, let us turn to French. Where Italian allows the definite article to occur in non-specific direct object NPs, this is not always an option in French. Instead, non-specific NPs typically take only the indefinite article. This leads to contrasts like that seen in (15) (parallel to [2]). (15)

a. b.

It. avere un/il pianoforte (indefinite article un/definite article il) Fr. avoir un/*le piano (indefinite article un/definite article le) intended reading: ‘have a piano’

In some cases, however, French also allows the definite article. This is most clearly the case with certain body parts, as in (16a), and with food items that form part of social routines, as in (16b). Some speakers also use the definite article with possessions that constitute status symbols, as in ((16c); again, the intended readings are non-specific). (16)

a. b. c.

avoir les épaules larges prendre le café avoir la BM

‘have broad shoulders’ ‘have coffee’ ‘have a BMW’

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

Cases like (15) and (16) raise the following question: How far-reaching is the contrast between French and Italian? Individual examples clearly point to a difference between the two languages, but they do not provide a clear view of its extent. A systematic analysis is needed to determine whether non-specific direct object NPs constitute a major difference in the article systems of French and Italian. Thus, the purpose of this paper is to demonstrate the significance of this contrast empirically.

.

The grammaticalization of articles

In this section, we discuss the grammaticalization of definite and indefinite articles and review the evidence that has been presented to support the view that French has a more developed article system than Italian. We argue that this evidence points to a difference in the grammaticalization of indefinite marking, but that it is silent on the question of the definite article. We then focus on the role of the specificity distinction, which, when adduced as an indicator of grammaticalization, suggests that Italian has a more grammaticalized definite article. Researchers working in two different theoretical and methodological traditions have arrived at the conclusion that the French article system is further developed than that of Italian. The first of these traditions is that of grammaticalizationoriented studies. Here definite and indefinite articles are viewed in terms of their diachronic development, which follows quite similar paths across languages. Greenberg’s (1978) 3-stage model of the development of the definite article describes the grammaticalization path leading from demonstrative to definite marker and, eventually, to gender marker. Givón (1981) and Heine (1997) provide similar models for the development of indefinite markers, which typically have their diachronic origin in the cardinal numeral ‘one’. What these studies show is that, in general terms, the grammaticalization of articles is a process of gradual expansion towards an ever wider range of grammatical contexts of use. This expansion of usage contexts goes along with greater obligatoriness, and, by the same token, the disappearance of contexts in which bare nouns are still admitted. It follows that the degree of grammaticalization of articles in a particular language can be read off their range of application to NPs of different types and, at the same time, the degree to which there still exist grammatical contexts in which they are not obligatory. In his survey of languages with definite articles at different developmental stages, Greenberg (1978: 64, 66) briefly mentions French and Italian. He cites French as a Stage II language and Italian (with English) as a Stage I language (where Stage I is the least grammaticalized stage). Applying Greenberg’s typology to several Romance

S F O PRO



1st proofs

The definite article in non-specific object NPs 

languages, Harris (1980) claims that the modern French article system is actually even closer to Stage III than Stage II. Both Greenberg’s and Harris’ arguments in regard to French are based on the strikingly low number of bare nouns in French (see below). However, neither author explicitly contrasts French with Italian. Nor do they distinguish the roles played by the definite and indefinite articles and the possibility that they may show different effects. The contrast between French and Italian has also been described from the perspective of synchronic syntactic theory (Longobardi 1994, 1999; Chierchia 1998 inter alia). Longobardi presents the following typology. (17)

a. b. c. d. e.

languages with no bare nouns (French) languages with stricter bare nouns (apparently the rest of Romance: Spanish, Italian . . .) languages with freer bare nouns (English and perhaps most of Germanic) languages with indefinite bare singulars (and only a definite lexical article: Icelandic, Celtic, Hebrew . . .) languages with ambiguous bare singulars (i.e. articleless languages: Russian, Czech, Latin . . .) (Longobardi 1999: 584)

Longobardi’s typology is based on the degree to which a language allows bare nouns in argument position. Both Longobardi and Chierchia point to indefinite plural and mass nouns to show that Italian differs from French. An article is considered optional in Italian, but must be present in French.7 (18)

a. b.

It. Mangio patate. Fr. Je mange *Ø/des pommes de terre. ‘I eat/am eating potatoes.’

(Longobardi 1994: 613)

The contrast in (18) clearly demonstrates different degrees of grammaticalization, with obligatoriness as the relevant parameter (cf. Lehmann 1995). However, note that this contrast pertains to indefinite NPs only. If an article was obligatory in Italian, as it is in French, one would expect the indefinite (partitive) article delle to appear in (18a). Thus, (18) is revealing with respect to the degree of obligatory indefiniteness marking. However, it is irrelevant as far as the definite article is concerned, revealing nothing about its degree of grammaticalization. This is not to deny the relevance of obligatoriness. What we are pointing out is that on a view of grammaticalization that considers each article separately, we cannot draw any conclusions about the definite article on the basis of contrasts like (18). Thus, either

S F O PRO

. But see Epstein (1995) and Roodenburg (2004) for examples of remaining contexts of bare nominals in French.



1st proofs

 Tanja Kupisch and Christian Koops

the two languages show no differences with respect to the definite article, or evidence needs to be sought elsewhere.8 Let us now turn to the role of specificity in the grammaticalization of articles. The specific/non-specific distinction has been shown to constitute an important parameter in this grammaticalization process. Diachronically, both the indefinite and the definite article are first used with specific reference only. This may be followed by an extension to contexts of non-specific (as well as generic) reference. Hence, the degree of grammaticalization of a given article can be determined with reference to its use in contexts beyond specific reference. In the case of indefinite markers, there is ample evidence for the diachronic extension from specific to non-specific (and generic) contexts. Cross-linguistic accounts are provided by Givón (1981) and Heine (1997); language-particular evidence includes historical studies of French (Blazer 1979; Selig 1992 and Carlier 2001), and Italian (Stark 2002; see also Maiden 1995). In the case of definite markers, the extension to generic contexts is often reported, while the extension to non-specific contexts is not so well attested. However, the wide range of languages cited by Greenberg (1978) shows that definite markers can also take on the function of marking nonspecificity. (While this does not occur as frequently as is the case with indefinite markers, this is possibly because the existence of an indefinite article in a language can prevent the definite article from developing further; see, for example, Harris 1980.) Today, Italian and French both use articles in syntactic constructions which do not constitute cases of specific reference. Since, as discussed above, these are the contexts most resistant to the presence of articles; they can be used to reveal the degree to which an article has grammaticalized. In our case, they should reflect any existing differences between Italian and French. (19)

a b.

Fr. Le lion est le roi des animaux. It. Il leone è il rè degli animali. ‘The lion is the king of the animals.’

(20)

a. b.

Fr. Ferdinand est Ø jardinier. It. Ferdinando è Ø giardiniere. ‘Ferdinando is a gardener.’

. Lyons (1999: 337) proposes a typology in which the determining factor is the number of contexts in which the definite article appears. He concludes that Italian has a more grammaticalized definite article than French. Italian, unlike French, requires the article in attributive possessive constructions (la mia macchina ‘my car’) and also makes limited use of it with proper names (la Maria ‘Maria’). However, the occurrence of articles before possessive pronouns does not necessarily provide evidence because Italian possessive pronouns behave like adjectives (Lyons 1986; Giorgi & Longobardi 1991; Haspelmath 1999) and the occurrence of the definite article before attributive adjectives is not unusual.

S F O PRO



1st proofs

The definite article in non-specific object NPs 

(21)

a. b. c. d.

Fr. aller au cinéma It. andare al cinema ‘go to the movies’ Fr. sans Ø sucre It. senza Ø zucchero ‘without sugar’

Examples (19) through (21) show no contrast between French and Italian. Generic subject NPs show the same choice of article. In predicate nominals (here, the names of professions) an article may be absent in both languages. In the case of complex adverbials, the picture is less obvious since both languages exhibit cases which require the article, and others which do not allow it, as seen in (21a–d). Hence, no systematic contrast emerges that would warrant the conclusion that either French or Italian make greater use of articles here. We therefore conclude that these contexts show no differences in the applicability of articles (whether definite or indefinite) to non-specific and generic NPs. However, we do find a contrast pertaining to definite articles in the construction type discussed at the outset of this paper. Recall that certain non-specific direct objects in Italian allow the use of the definite article where this is not an option in French. Thus, with regard to the definite article, the extension to nonspecific NPs does point to a difference. To the extent that it constitutes a major difference between French and Italian, the possibility of using the definite article with non-specific NPs suggests different degrees of grammaticalization for the two definite articles. The remainder of this paper will demonstrate this contrast empirically.

.

Analysis

In this section, we present two studies of article use in non-specific direct object NPs. In each study, we performed parallel analyses of French and Italian in order to determine whether the contrast seen in the examples discussed at the outset of this paper is borne out in a larger, more representative sample. If Italian does, in fact, make more use of the definite article in non-specific direct objects, there should be higher proportions of definite articles in parallel samples of such NPs in French and Italian. To test this hypothesis, we quantified and compared the number of definite articles used in this function in each language. Given that the non-specific function of the definite article appears to be characteristic of colloquial, spoken language use, purely intuition-based analyses run the risk of being affected by normative preconceptions. We therefore combined two methodologies, first analyzing a selection of spoken language corpora, and

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

then conducting a questionnaire study. By combining the advantages of these two independent methodologies, we minimized any methodological bias. .

A Corpus study

.. Methodology Our first study addressed the question of article use in non-specific NPs in spontaneous, spoken language. We begin by discussing certain methodological problems connected with this approach. Ideally, a parallel, corpus-based approach would take into account verb+object combinations that have counterparts in both languages. For example, one would collect instances of the verb avere followed by the object macchina in Italian, and instances of the verb avoir followed by the object voiture in French, and compare the determination patterns. However, two problems arise. First, it is unlikely that a sufficient number of parallel examples will be included even in large corpora of spoken language. The variety of possible combinations is simply too great, and the size of the available corpora too small. Second, even if parallel examples are found, these may not instantiate non-specific NPs because language use includes both specific and non-specific NPs. Our solution to the first problem was to depart from the ideal of considering only fully parallel verb+object NP combinations. Instead, we held the verb constant and took account of all object NPs. We chose a set of seven verbs, which we hypothesized to be used with non-specific object NPs. As noted in Section 2, these verbs include verbs of possession, the preparation and consumption of food, and the wearing of clothes and accessories. The set is given in (22). (22)

a. b. c. d. e. f. g.

avoir/avere ‘have’ acheter/comprare ‘buy’ faire/fare ‘make, do’ manger/mangiare ‘eat’ prendre/prendere ‘take’ mettre/mettere ‘put’ porter/portare ‘carry, wear’

Regarding the second problem, we included both non-specific and specific NPs rather than determining specificity on a case-by-case basis and excluding all cases of specific object NPs. The rationale for doing so was that we can expect, on theoretical grounds, that the cases of specific reference do not constitute a significant source of error. To explain this, we need to discuss exactly what the effect of such specific NPs would be. Note first that in representative samples, the relative frequency of specific NPs – i.e. the proportion of specific NPs out of all NPs – can be expected not

S F O PRO



1st proofs

The definite article in non-specific object NPs 

to vary significantly. In other words, we have no reason to assume that, on average and in similar situations, speakers of Italian talk more or less about specific referents than speakers of French. Given that we drew our samples from the same discourse genre and used large sample sizes (see below for details), the impact of the specific NPs should be equally great in both languages. This leaves the possibility of a qualitative effect stemming from the different determination patterns in specific NPs. However, note that this does not affect NPs that include the definite article. As argued in Section 3, differences in obligatoriness pertain only to indefinite NPs. These must be indefinite marked in French, but can be both indefinite or zero marked in Italian (cf. [18]). Specific definite NPs, by contrast, can be expected to receive the same marking because any specific NP that takes a definite article in French also takes a definite article in Italian. Thus, the inclusion of specific object NPs should lead to no overall differences in the rate of occurrence of definite articles. Both the French and the Italian data should be equally affected by such cases. Any remaining differences in the rate of occurrence of definite articles can be attributed to non-specific NPs. At the same time, we would expect specific object NPs to affect the number of indefinite and zero-marked objects, given the different determination patterns found in such NPs. In our statistical interpretation of the results (in the following section), we therefore compare the rate of definite NPs to the combined rate of indefinite and zero marked NPs. The discourse genre we investigated was child-directed adult speech, i.e. the language used by adults when interacting with small children. We obtained all data from the CHILDES database (MacWhinney & Snow 1985), a resource aimed primarily at child language researchers.9 Because child-directed speech is clearly a specialized genre of language use, it may be legitimate to question to what degree the results obtained from this genre can be generalized to language used in adultadult interaction. Although we cannot exclude the possibility of a genre effect, we consider an analysis of this particular genre to be a very useful starting point due to the advantage of relatively large collections of transcribed recordings available for both French and Italian. This allows a maximally parallel analysis, in which genre effects would be equally present in both languages, and therefore neutralized.

. Our samples of the French verbs avoir ‘have’, faire ‘make, do’, mettre ‘put’ are taken from the Champaud corpus, and those of the remaining 4 verbs from the Rondal corpus (Rondal 1985). We extracted all instances of the Italian verbs avere ‘have’, fare ‘make, do’, mettere ‘put’, and prendere ‘take’ from the Martina part of the Calambrone corpus (Cipriani et al. 1989), and all instances of the verbs comprare ‘buy’ and mangiare ‘eat’ from the Guglielmo part of the Calambrone corpus. The instances of the verb portare ‘carry, wear’ we obtained by searching the entire Calambrone corpus.

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

We extracted all tokens of the seven verbs in (22) occurring with lexical direct object NPs and counted the number of instances of three categories of determination: (i) object NPs including a definite article, (ii) object NPs including an indefinite article, and (iii) bare object nouns. Partitive articles (Fr. du, des, etc., It. dei, delle, etc.) were subsumed under the category of the indefinite article, as were occurrences of Fr. de in negation (On ne mange pas de viande. ‘We don’t eat meat.’) and before adjectives with plural nouns (Elle a de longs ongles. ‘She has long fingernails.’).10 We excluded cases of definite NPs with a possessive marker (It. la mia macchina ‘my car’) and definite NPs with proper names (It. la Maria ‘the Maria’) as these two phenomena occur only in Italian and, furthermore, can be predicted to have specific reference. As noted above, we considered only adult utterances and excluded the commentary included in the transcripts and direct repetitions of child utterances by adults. .. Results Figure 1 shows the relative and absolute frequencies of each of the three categories of determination. The ratio of definite NPs to the combined indefinite and zero marked NPs is significantly higher in Italian than in French (df = 1, χ2 = 204.5, p < 0.001). 82% (403)

41% (302)

48% (354)

DEF INDEF ZERO 11% (80)

FRENCH

11% (53)

7% (34)

ITALIAN

Figure 1. Overall frequency of three determination types.

. The former de is the variant of the indefinite article used under negation; de before adjectives followed by plural nouns is used in formal registers and commonly appears as des in spoken language (Klein & Kleineidam 1999: 40).

S F O PRO



1st proofs

The definite article in non-specific object NPs 

Table 1 gives the relative and absolute frequencies of the three types of determination for each verb individually. The frequencies in Table 1 show that no one verb alone is responsible for the general effect. The contrast seen in Figure 1 is present, with minor variations, in each individual verb. Table 1. Frequency of three determination types with each verb FRENCH DEF ‘have’ 16 (12%) (avoir, avere) ‘make/do’ 18 (16%) (faire, fare) ‘buy’ (acheter, 27 (38%) comprare) ‘eat’ (mangrer, 45 (42%) mangiare) ‘take’ (prendre, 144 (67%) prendere) ‘put’ (mettre, 19 (53%) mettere) ‘carry/wear’ 33 (69%) (porter, portare) TOTAL

ITALIAN INDEF

ZERO

89 (65%)

32 (23%)

99 (72%)

9 (6%)

31 (22%)

58 (47%)

48 (39%)

107 (84%)

20 (15%)

1 (1%)

44 (62%)

0 (0%)

17 (100%)

0 (0%)

0 (0%)

61 (48%)

0 (0%)

21 (100%)

0 (0%)

0 (0%)

70 (33%)

0 (0%)

29 (81%)

7 (19%)

0 (0%)

17 (47%)

0 (0%)

70 (84%)

13 (16%)

0 (0%)

15 (31%)

0 (0%)

60 (91%)

4 (6%)

2 (3%)

80 (11%)

403 (82%)

53 (11%)

34 (7%)

302 (41%) 354 (48%)

DEF

INDEF

ZERO

A general inspection of the corpus data suggests that non-specific constructions greatly contribute to the striking asymmetries seen in Figure 1 and Table 1. There are indeed cases of semantically and referentially equivalent uses. For example, in (23), the indefinite article in French corresponds directly to the definite article in Italian. In (24) the verb manger/mangiare ‘eat’ is combined with noun phrases denoting similar food items, also exhibiting the hypothesized contrast.11 (23)

.

a.

Fr. (Grégoire 2; 0, 5) ADULT: Qu’est ce qu’il a sur sa tête le cheval? ‘What does it have on its head, the horse?’ CHILD: Cheval. ‘Horse.’ ADULT: Qu’est ce qu’il a le cheval sur sa tête? ‘What does it have, the horse, on its head?’ Il a des cornes! ‘It has horns!’

S F O PRO

The transcription has been slightly edited to ensure greater readability.



1st proofs

 Tanja Kupisch and Christian Koops

(24)

b.

It. (Martina 1; 11, 20) ADULT: Con le scarpe, sì, questa? [asking child to name an object] ‘With the shoes, yes, this one?’ CHILD: Lallo. [= cavallo] ‘Horse.’ ADULT: No, non è un lallo, guarda. ‘No, it’s not a horse, look.’ Ci ha le corna. ‘Here it has horns.’

a.

Fr. (Grégoire 1; 9, 28) CHILD: Soussours. [= nounours] ‘Bear.’ [child points at a bear] ADULT: Oh mais il mange du miel avec sa patte. ‘Oh but he eats honey with his paw.’ It. (Martina 2; 1, 12) [Mother and child are talking about a doll sitting on a chair.] ADULT: È sopra la seggiolina, è come la tua, vedi, piccolina, piccolina, rossa, e cosa mangia? ‘It’s on the baby chair, it’s like yours, you see, very small, red, and what does she eat?’ CHILD: Dillo te! ‘You say it!’ ADULT: Eh, questa è difficile, mangia la marmellata. ‘Ehm, this is difficult, she is eating jelly.’

b.

Light verb constructions with ‘do’ and ‘take’ show similar contrasts. (25)

a. b.

Fr. faire de la musique, prendre un bain, faire des grimaces It. fare la musica, fare il bagno, fare le smorfie ‘make music’, ‘take a bath’, ‘make faces’

Although cases like (25) were not part of our hypothesis, such constructions demonstrate the same general effect. Non-specific NPs tend to occur with the definite article in Italian, and with the indefinite article in French. Another interesting result is the rate of occurrence of bare nouns. Overall, bare nouns are less frequent in Italian than in French, although the discussion of obligatoriness in Section 2 would suggest the opposite. Table 1 shows that most verbs actually show no difference in the degree of bare nouns. The asymmetry stems almost entirely from the verbs fare and faire ‘do’, which occur in idiomatic expressions that show zero marking in French and definite marking in Italian. (26)

a. b.

Fr. faire Ø dodo, faire Ø mal, faire Ø pipi, It. fare la nanna, fare la bua, fare la pipì, ‘sleep’, ‘hurt’, ‘pee’

S F O PRO



1st proofs

The definite article in non-specific object NPs 

Cases like (26) deserve mentioning because they show that the Italian definite article covers even contexts which show zero determination in French. In summary, we find a significant contrast between French and Italian. The data strongly suggest that the definite article is much more common in non-specific object NPs in Italian than in those in French. In addition, bare direct object NPs are surprisingly rare in both languages. .

A questionnaire study

.. Methodology Our second study was designed to test native speakers’ intuitive judgment of article use with non-specific direct objects. French and Italian have many analogous constructions, such as avere il pianoforte and avoir un piano, which can be used to test the language contrast. We compiled a list of nouns which we expected to be used in non-specific object NPs, such as ‘car’ in ‘have a car’ or ‘bread’ in ‘buy bread’, and constructed verb+object pairs for each noun. The verbs used in these pairs were the seven verbs examined in the corpus study, semantically near-synonymous verbs (e.g. It. farsi, the colloquial counterpart of comprare ‘buy’) and the verb boire/bere ‘drink’. The nouns were selected from five semantic groups, which are listed in (27). The total number of items was 54. (27)

a. b. c. d. e.

body parts (e.g. ‘big eyes’, ‘a goatee’) alienable possessions (e.g. ‘cell phone’, ‘computer’) “semi-alienable” possessions (e.g. ‘piercing’, ‘wig’) clothing and accessories (e.g. ‘skirt’, ‘earrings’) food items (e.g. ‘wine’, ‘bread’)

(9 items) (19 items) (5 items) (5 items) (16 items)

Instead of having native speakers rate a list of decontextualized phrases or sentences, we made the task more natural by creating a context for the items. Adding context was also a way to ensure a non-specific interpretation. We created a single, connected narrative with the 54 items embedded in which a first person narrator talks about her friend using a colloquial style of language. For each item, the text included a pre-nominal slot where the subjects were asked to indicate which article they heard most frequently in everyday language. The narrative was constructed to suggest a non-specific reading of the relevant NPs. We created the narrative in English and had native speakers of Italian and French with good knowledge of English translate them. In this way, the linguistic contexts were kept exactly parallel. An extract of the narrative is given in (28).

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

(28)

It. L’altro giorno Gianni mi ha chiesto cosa ne pensassi se si fosse fatto {un/il} piercing. Gli ho risposto: “Se devi metterti delle cose artificiali nel tuo corpo, almeno aspetta una dozzina d’anni, quando a quel punto avrai {la/una} protesi, {un/il} pacemaker e {la/una} dentiera.” Allora decise di farsi {un/il} tatuaggio. Fr. L’autre jour Jean m’a demandé ce que je penserais si il se faisait {un/le} piercing. Je lui ai dit: “Si tu veux avoir des choses artificielles dans le corps, attends une douzaine d’années et tu verras. Tu auras alors {la/une} hanche artificielle, {un/le} pacemaker et {les/des} fausses dents.” Alors, il s’est décidé à se faire {un/le} tatouage. ‘The other day John asked me whether he should get a piercing. I told him, “If you want to have artificial things in your body, just wait a few dozen years. Then you’ll have an artificial hip, a pacemaker, and false teeth.” So he decided to get a tattoo.’

The example shows that we consistently gave definite and indefinite articles as options. We also gave the option of choosing both articles, if the subjects felt that they were used equally often. However, we did not give “zero” as an option because we wanted to reveal the difference between the definite and the indefinite article, and one of the two was always possible. It was inevitable that, with some additional assumptions, a few contexts still allowed for a specific interpretation of the object NP. Thus, in some cases the choice of the definite article could have been due to a specific reading. However, since the texts had been translated from English, where the NPs were clearly non-specific, the non-specific interpretation was always possible. Moreover, if ambiguity existed, it did so equally in both Italian and French. 29 native speakers of French and 44 native speakers of Italian participated in the study. The age span was 13 to 58 years for the French subjects (mean 34, median 28), and 19 to 60 years for the Italian subjects (mean 38, median 40). Of the 29 French subjects, 23 spoke a Northern variety of French. The remaining 6 subjects included one speaker of a Southern French variety, 4 speakers of Canadian French, and one speaker of Belgian French. All 44 Italian subjects spoke the standard variety of Italian, but they came from different regions in Italy: 29 speakers were from Northern Italy, 9 from Central Italy and 6 from Southern Italy. .. Results We determined the total number of times that subjects chose the definite article, the indefinite article, or both the definite and the indefinite article. As in the earlier study, we counted partitive articles and French de (under negation and in adjectivenoun sequences) as indefinite. Figure 2 shows the proportion of each choice in the two languages. The relative frequency of definite articles is significantly greater in Italian (df = 2, χ2 = 1408.8, p < 0.001).

S F O PRO



1st proofs

The definite article in non-specific object NPs  84%

55% DEF INDEF BOTH

27% 18% 9%

7% FRENCH

ITALIAN

Figure 2. Article choice in the questionnaire study.

Several of the French subjects indicated that they sometimes hear younger speakers of French using definite articles in these or similar contexts. Therefore, we also analyzed article choice in relation to age. To do so, we devised a scoring system which assigns each subject a single score reflecting his or her overall preference for the definite or indefinite article. Three points were given for each choice of the definite article; two points were given if both articles were considered equally likely; one point was given for each choice of the indefinite article. Given the total number of 54 items, the minimum score was 54 and the maximum score was 162. The results of this analysis are shown in Figure 3. 160 140 120 ITALIAN FRENCH

100 80 60 40 10

20

30

40 AGE

50

Figure 3. Article choice by age.

60

70

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

Figure 3 shows considerable variation in article choice overall, but no correlation with age. Although our study was not designed to capture the effect of regional dialectal differences, we also tested for this factor.12 Our data show no regional differences. Italian subjects speaking a Northern variety do not show different preferences than speakers of other varieties. The same is true for the French subjects of different regional origins. Of course, in order to come to definite conclusions about the effect of dialectal differences, this parameter would have to be tested more systematically. Finally, we want to point out the effect of particular verb+object combinations within the overall pattern. Most of the items fall in line with the main effect: a strong preference for the indefinite article in French and a general preference for the definite article in Italian. However, there are some interesting cases of parallel preferences. To measure the effect of individual items, we assigned each item a score as was done for the individual subjects. However, due to the unequal number of participants from each language, the French and Italian scores are not directly comparable. We therefore indicate each score as the percentage of participants who chose the definite article, either as one option or as the only option. The examples in (29) are typical in that they follow the overall pattern seen in Figure 2. This effect is found with nouns of all semantic types (cf. [27]). (29)

a.

b.

c.

Fr. porter des lentilles de contact It. portare le lenti a contatto ‘wear contact lenses’ Fr. avoir un double menton It. avere il doppio mento ‘have a double chin’ Fr. manger du poisson It. mangiare il pesce ‘eat fish’

(0% definite) (96% definite) (3% definite) (98% definite) (0% definite) (80% definite)

In other cases, speakers of both languages showed a preference for the definite article, or at least a tendency to choose it. The two groups of nouns for which this is most obvious are possessions tied to particular institutions, as in (30a), and body parts, as in (30b). . Because Southern varieties of Italian do not employ indefinite plural and partitive articles (e.g. del, dei, delle etc.) (cf. Rohlfs 1968; Maiden 1995), it may be argued that the absence of these articles leads speakers to use the definite article more widely than speakers of other varieties. (cf. Harris 1980, who argues that the presence of an indefinite article can prevent the definite article from spreading further). Our data do not support this view, because we find no such regional differences. Furthermore, the corpus analysis revealed that speakers of Northern Italian varieties do not use indefinite plural and partitive articles either, at least not in colloquial, spoken language, and in the contexts investigated.

S F O PRO



1st proofs

The definite article in non-specific object NPs 

(30)

a.

b.

Fr. avoir le permis de conduire/le cable/l’ADSL It. avere la patente/la TV via cavo/l’ ADSL ‘have a driver’s license/cable TV/DSL’ Fr. avoir les mains poilues/les cheveux bruns It. avere le mani pelose/i capelli scuri ‘have hairy hands/dark hair’

(69/89/89% definite) (100/99/99% definite) (84/81% definite) (93/98% definite)

Such examples show that French sometimes shows the same effect as Italian. Furthermore, in all cases where we found a tendency towards the definite article in French, there was an equal or stronger effect in the same direction in Italian. In other words, cases of this kind in French are a subset of those in Italian. A particularly striking example of parallel preferences is the item ‘buy bread’, which is given in (31) together with the context it appeared in on the questionnaire. (31)

a. b.

Fr. Le problème c’est que Jean ne fait jamais les courses. C’est toujours moi qui dis: “Je vais acheter le pain.” (33% definite) It. Il problema è che Gianni non fa mai la spesa. Sono sempre io che dico: “Vado a comprare il pane.” (93% definite) ‘The problem is that John never buys groceries. It’s always me who says: “I’m going to get bread.” ’

As seen in the above percentages, one out of three French speakers allowed the definite article in this case. However, ‘bread’ was the only food noun tending toward the definite article in French. Others showed weaker effects in the same or similar verb context, though not rejecting the definite article entirely: acheter les oeufs (17% definite), acheter le lait (17% definite), chercher les pommes (10% definite). In summary, there is a clear overall contrast between French and Italian. Non-specific object NPs are much more regularly marked with the definite article in Italian. However, at the same time, we find a small number of parallel cases in which both languages take, or allow, the definite article. The cases where this occurs in French are a subset of the cases where it occurs in Italian.

.

Discussion

The two analyses presented in Section 4 have revealed clear differences between French and Italian. The definite article is indeed more widespread in non-specific direct object NPs in Italian than it is in French. At the same time, the questionnaire study in particular has brought out some interesting parallels. In some cases, French also allows both articles. Moreover, the constructions where this occurs in French are a subset of those in Italian.

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

These results raise the following questions. First, if two alternative ways of marking non-specific direct objects are available, is there a difference in interpretation? Second, if the articles lead to different interpretations, do we find the same meaning contrasts in both languages? These questions have interesting implications from a grammaticalization perspective. If parallel meaning contrasts exist in French and Italian, this would indicate that we are witnessing the same phenomenon in both languages. The two articles do lead to different interpretations. For instance, consider the French pair in (32). (32)

Fr. a. b.

Je vais acheter le pain. Je vais acheter du pain. ‘I am going to buy bread.’

(definite article le) (partitive article du)

Recall that in the case of pain ‘bread’ a relatively large number of French respondents chose the definite article as an option. In (32), the definite article additionally evokes the idea of buying bread as a routine, day-to-day activity. These components of meaning are absent when the indefinite article is used. Another example from French is given in (33). (33)

Fr. a. b.

Jean a la BM. Jean a une BM. ‘John has a BMW.’

(definite article la) (indefinite article une)

Although not all speakers of French report using (33a), those who accept it agree that the definite article emphasizes certain properties of BMWs as a type of car, especially the fact that they are expensive and status symbols. Again, there is no such effect with the indefinite article. Quite parallel contrasts are found in Italian. The pair of sentences in (34) is analogous to that in (32). (34)

It.

a. Vado a comprare il pane. b. Vado a comprare del pane. ‘I am going to buy bread.’

(definite article il) (partitive article del)

While for some speakers (34b) is not an option at all, for those who use it and who sense a meaning difference, the interpretation imposed by the definite article is that of purchasing bread as a daily routine, exactly as in (32). This supports Korzen’s analysis of such examples as reflecting the habitual occurrence of the situation (cf. Section 2). Moreover, note that the definite article is not used with uncommon food items, which are unlikely to be purchased routinely, as shown in (35) (repeated from [14]). (35)

S F O PRO

It. ??Vado a compare i kiwi. (definite article i) intended reading: ‘I’ll go and buy kiwis.’



1st proofs

The definite article in non-specific object NPs 

Next, consider the pair in (36), which parallels (33). Again, the same contrast arises. As in French, the definite article invokes the idea that BMWs are a status symbol, while the indefinite leads to a neutral interpretation. (36)

It.

a. Gianni ha la BMW. (definite article la) b. Gianni ha una BMW. (indefinite article una) ‘John has a BMW.’

Interestingly, the definite article cannot be used with reference to a type of car that is not considered prestigious. For instance, Fiats do not have a reputation of being reliable in Italy. (37)

It. ??Gianni ha la Fiat. (definite article la) intended reading: ‘John has a Fiat.’

Overall, these examples show that the non-specific reading connected to the definite article is quite different from that of the indefinite article. The two differ in subtle, culture-specific ways. Clearly, we have only scratched the surface of this phenomenon. However, the examples suffice for our purposes, demonstrating that, to the extent that non-specific definite NPs also exist in French, they closely resemble those in Italian. It appears that we are, indeed, dealing with essentially the same phenomenon in both languages. The extension of the definite article from specific to non-specific NPs affects similar nouns and leads to similar interpretations. In other words, the grammaticalization process we are describing is not unique to Italian, but has merely progressed much further in Italian than in French.

.

Conclusion

Non-specific definite direct object NPs are significantly more widespread in Italian than in French. Thus, the function of non-specificity marking clearly sets apart the Italian and French definite articles. Given that the extension of definite markers from specific to non-specific NPs is an independently attested grammaticalization phenomenon, and that the two articles show no other obvious differences in their range of applicability or obligatoriness, we conclude that the Italian definite article is further grammaticalized than its counterpart in French. It is important to note that this conclusion would be lost if article grammaticalization is viewed strictly as a process in which obligatory determination increases, while the number of contexts in which bare nouns are still admitted decreases. On such a view, contexts in which the use of an article is already obligatory, but where two articles compete, will be left out of consideration. It is this view which underlies the analysis of French as having the more developed article system. If, however,

S F O PRO



1st proofs

 Tanja Kupisch and Christian Koops

the distribution and function of each article is taken into account separately, an important difference between French and Italian emerges, which calls for a more differentiated analysis: While indefinite marking is less grammaticalized in Italian, definite marking is more grammaticalized. Ultimately, a grammaticalization account must also provide a motivation for the extension of definite markers from specific to non-specific NPs, which is an interesting question for further research. The results of this study apply to this question because they show that, to the degree that this process has affected French as well, the resulting semantic and pragmatic effects are quite similar. This suggests that the phenomenon in Italian is of wider significance and opens a window to a functional extension that has not previously been studied in detail.

References Abney, Steve. 1987. The English Noun Phrase in its Sentential Aspect. PhD Dissertation, MIT. Alexiadou, Artemis. 2003. Some notes on the structure of alienable and inalienable possessors. In From NP to DP, Vol. II [Linguistics Today 56], Martine Coene & Yves D’hulst (eds), 168–188. Amsterdam: John Benjamins. Blazer, Emily Deering. 1979. The Historical Development of Articles in Old French. PhD Dissertation, University of Texas at Austin. Carlier, Anne. 2001. La genèse de l’article un. Langue Française 130: 65–88. Chierchia, Gennaro. 1998. Reference to kinds across languages. Natural Language Semantics 6: 339–405. Cipriani, Paola, Pfanner, Pietro, Chilosi, Anna, Cittadoni, Lorena, Ciuti, Allessandro, Maccari, Anna, Pantano, Natalia, Pfanner, Lucia, Poli, Paola, Sarno, Stefania, Bottari, Piero, Cappelli, Giuseppe, Colombo, C. & Veneziano, E. 1989. Protocolli diagnostici e terapeutici nello sviluppo e nella patologia del linguaggio [1/84 Italian Ministry of Health]. Stella Maris Foundation, Pisa. Donnellan, Keith S. 1966. Reference and definite descriptions. Philosophical Review 75: 281–304. Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22: 1–25. Epstein, Richard. 1995. The later stages in the development of the definite article: Evidence from French. In Papers from the 11th International Conference on Historical Linguistics 1993, Henning Andersen (ed.), 159–175. Amsterdam: John Benjamins. Giorgi, Allessandra & Longobardi, Giuseppe. 1991. The Syntax of Noun Phrases. Cambridge: Cambridge University Press. Givón, Talmy. 1981. On the development of one as an indefinite marker. Folia Linguistica Historica 2: 35–53. Greenberg, Joseph H. 1978. How does a language acquire gender markers? In Universals of Human Language, Vol. III, Joseph H. Greenberg, Charles A. Ferguson & Edith A. Moravcsik (eds), 47–82. Stanford CA: Stanford University Press. Guéron, Jaqueline. 2003. Inalienable possession and the interpretation of determiners. In From NP to DP, Vol. II [Linguistics Today 56], Martine Coene & Yves D’hulst (eds), 189–220. Amsterdam: John Benjamins.

S F O PRO



1st proofs

The definite article in non-specific object NPs  Harris, Martin B. 1980. The marking of definiteness: A diachronic perspective. In Papers from the 4th International Conference on Historical Linguistics, Elizabeth Closs Traugott, Bernd Heine, Rebecca Labrum & Susan Shepherd (eds), 75–86. Amsterdam: John Benjamins. Haspelmath, Martin. 1999. Explaining article-possessor complementary: Economic motivation in noun phrase syntax. Language 75(2): 227–243. Heine, Bernd. 1997. Cognitive Foundations of Grammar. Oxford: Oxford University Press. Klein, Hans-Wilhelm & Kleineidam, Hartmut. 1994. Grammatik des heutigen Französisch. Stuttgart: Klett. Korzen, Iørn 1996. L’articolo italiano fra concetto ed entità, I–II [Etudes Romanes 36]. Copenhagen: Museum Tusculum Press. Korzen, Iørn 1998. On nominal determination – with special reference to Italian and comparisons with Danish. In Nominal Determination [Copenhagen Studies in Language 21], Gyde Hansen (ed.), 67–132. Copenhagen: Samfundslitteratur. Krifka, Manfred, Pelletier, Francis J., Carlson, Greg N., ter Meulen, Alice, Link, Godehard & Chierchia, Gennaro. 1995. Genericity: An introduction. In The Generic Book, Greg N. Carlson & Francis J. Pelletier (eds), 1–124. Chicago IL: Chicago University Press. Lehmann, Christian. 1995. Thoughts on Grammaticalization: A programmatic sketch. Munich: Lincom. Longobardi, Giuseppe. 1994. Reference and proper names: A theory of N-raising in syntax and logical form. Linguistic Inquiry 25: 609–665. Longobardi, Giuseppe. 1999. The structure of DPs: Some principles, parameters, and problems. In The Handbook of Contemporary Syntactic Theory, Mark Baltin & Chris Collins (eds), 562–603. Oxford: Blackwell. Lyons, Christopher. 1986. On the origin of the Old French strong-weak possessive distinction. Transactions of the Philogical Society 84: 1–43. Lyons, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press. Mac Whinney, Brian & Snow, Catherine. 1985. The child language data exchange system. Journal of Child Language 5: 337–345. Maiden, Martin. 1995. Storia Linguistica dell’Italiano. Bologna: Il Mulino. Renzi, Lorenzo. 1988. L’articolo. In Grande Grammatica Italiana di Consultazione, Lorenzo Renzi, Giampaolo Salvi & Anna Cardinaletti (eds), 357–424. Bologna: Il Mulino. Rohlfs, Gerhard. 1968. Grammatica storica della lingua italiana e dei suoi dialetti. II. Morfologia. Torino: Einaudi. Rondal, Jean A. 1985. Adult–child interaction and the process of language understanding. New York NY: Praeger. Roodenburg, Jasper. 2004. Pour une approche scalaire de la défiance nominale: La position du français dans une théorie des “noms.” PhD Dissertation, University of Amsterdam. Selig, Maria. 1992. Die Entwicklung der Nominaldeterminanten im Spätlatein. Romanischer Sprachwandel und lateinische Schriftlichkeit. Tübingen: Narr. Stark, Elisabeth. 2002. Indefiniteness and specificity in Old Italian texts. Journal of Semantics 19: 315–332. Vergnaud, Jean-Roger & Zubizaretta, Maria Luisa. 1992. The definite determiner in French and in English. Linguistic Inquiry 23: 595–652.

S F O PRO



1st proofs

S F O PRO



1st proofs

Early functions of definite determiners and DPs in German first language acquisition Dagmar Bittner ZAS-Berlin, Germany

The paper discusses the functional load of definite DPs in early child German regarding sentence-internal (argument structure) and sentence-external relations (text coherence). The data consist of adult-child dialogues of two children in the period 1;11–2;11. It is argued that DPs exhibit functional load on both levels from the onset of production. Contradicting recent assumptions, (i) case-related distinctions are acquired prior to gender distinction, and (ii) children establish a functional distinction between pronominal DPs (continued or directly accessible reference) and noun-including DPs (disrupted or especially emphasized reference). The results allow the hypotheses that the noun-including DP is a functionally motivated extension of the pronominal DP and that properties relating the DP to other elements of sentence/text are the first to be acquired.1

.

The problem

It is well known that determiners are related to a wide range of grammatical functions in German. In addition to the morpho-syntactic properties of the noun (gender, case, and partially number), they symbolize certain degrees of definiteness and contribute to text structure and coherence. The present paper focuses on definite determiners, more specifically on those forms that can be equally used as demonstrative pronouns and definite articles. Research on the emergence of these forms in German first language acquisition has so far merely focused on the morpho-syntactic features (amongst others, Mills 1986; Müller 1994; Bittner forthcoming). Their capacities on the text level, especially their contribution to text coherence, are regarded as a matter of later developmental processes, only

. For fruitful criticisms and comments on a preliminary version of this paper I am thankful to Elisabeth Stark, Werner Abraham, and the participants of the AG 10 “Entstehung und Funktion von Nominaldetermination” at the 27th annual conference of the DGfS in Köln in February 2005.

S F O PRO



1st proofs

 Dagmar Bittner

emerging after age 5. The relevant evidence is supposed to arise from difficulties of younger children in producing coherent narratives on the base of picture stories or in retelling video-clips. It is argued that young children do use definite determiners not to establish anaphoric relations, but merely as deictic elements (for a summary of studies along these lines see Hickmann 2003). This state of the art is doubtful for several reasons. With respect to the morpho-syntactic categories, most analyses have dealt with the acquisition of either case features or gender features. The interrelation of both functional domains in the acquisition process has not been a matter of debate. As a consequence, the emergence of the pronominal paradigms has remained a black box so far. Furthermore, a first investigation on the acquisition of the paradigms (Bittner, forthcoming) provided evidence that some of the conclusions drawn from the separate treatments of either case or gender do not present an adequate description of the course of acquisition. With respect to anaphoric capacities of definite determiners, the assumption of relatively late acquisition might be an underestimation of the child’s early abilities for the following reasons. First of all, definite determiners are acquired very early, i.e. around the second birthday. They belong to the first grammatical forms children produce. No difficulties to use these forms in the appropriate contexts have been reported. Of course, children frequently omit determiners and when they use them they have some difficulties to choose the appropriate morpho-syntactic form over a longer period. Nevertheless, they have no difficulties to distinguish the grammatical contexts allowing or disallowing a definite determiner. Secondly, use of definite determiners typically starts with demonstrative pronouns. In the beginning, these forms exclusively point to objects given in the actual situation. However, as I will show, after a short period, they become related to the preceding linguistic context. The same happens with personal and interrogative pronouns, which emerge shortly after demonstrative pronouns. In what follows I will argue that the acquisition of definite determiners (henceforth: d-determiners) and, hence, the acquisition of definite noun phrases (DPs) starts – beside the property of definiteness – with properties which determine sentence-internal and sentence-external relations, i.e. morpho-syntactic properties and properties of reference specification. In other words, d-determiners are acquired as grammatical means used “um den Bezug eines (virtuellen oder aktuellen) Zeichens zu begrenzen, zu präzisieren und zu orientieren” (Coseriu 1975: 261)2 (emphasis by the author).

S F O PRO

. Translation: “in order to restrict, determine, and orientate the reference of the (virtual or actual) sign.”



1st proofs

DPs in the acquisition of German 

The early phases in the acquisition of the morpho-syntactic, i.e. sentenceinternal properties of d-determiners are described in more detail in Bittner (2002a and forthcoming). Section 3 briefly presents the results of these studies. Section 4 discusses the sentence-external properties of DPs, i.e. their deictic and anaphoric capacities. Finally, Section 5 presents the conclusions on the functional load of d-determiners on both utterance and text level at the onset of grammatical development. However, to begin with, the overall development in the emergence of d-determiners and the resulting structural differentiations will be summarized in Section 2. The acquisition data underlying this study originate from longitudinal corpora of two girls: Anna and Simone. The corpora contain regularly recorded spontaneous parent-child dialogues in everyday situations in the age range of 1;8 to 2;10 (Anna) and 1;9 to 4;0 (Simone).3 For the present study, the use of definite pronouns and articles was analysed from the onset of production through the end of the third year of life.

.

Overall development: From bare nouns to DPs

In German, as in all languages, the acquisition of NPs starts with the production of bare nouns. The first d-determiner that emerges in German child language is the pronoun das ‘this/that’. It is used as a deictic element in the strongest sense, i.e. as a linguistic sign for a pointing gesture towards an object given in the actual situation. Initially, this form constitutes a one-word utterance, later on it is combined with modifiers like hier ‘here’ and da ‘there, over there’ (2a) followed by application in more complex but infinite utterances (2b). The definite pronouns der, die, and (later on) den ‘this/that--’ become productive after or simultaneously to the acquisition of finite verbs. At first they are used in phrases containing the copula ist ‘is’ (3a). Then they get combined with finite forms of lexical verbs (3b). The acquisition of the finite verb seems to be a prerequisite for the acquisition of these forms of definite pronouns (Bittner 2002a). The ability to produce complex DPs, i.e. to use demonstrative pronouns as definite articles of nouns, is the fourth step in the emergence of the DP.

. The data of Simone are available from the CHILDES database (http://childes.psy.cmu. edu/data/Germanic/German/). The data of Anna were recorded by the author. The age of the children is coded in following way: year; month(:day). 2; 11:16 means the child was two years, 11 months, and 16 days old.

S F O PRO



1st proofs

 Dagmar Bittner

Complex DPs firstly emerge in stereotyped utterances containing the copula ist ‘is’ (4a). Finally, children start to produce complex utterances containing finite lexical verbs and complex DPs (4b). The general line of development is summarized in (1) to (4): (1)

1st step:

bare nouns

(2)

2nd step: pronoun das a. pre-propositional – one- or two-word utterances (declarative and interrogative): i. das ii. das hier? iii. das auch this/that this/that here this/that too b. propositional – non-finite utterances: i. das auch kreis ii. eine uhr das iii. das auffressen this/that too circle a clock this/that this/that eat up ‘this/that is a circle too’ ‘this/that is a clock’ ‘eat up this/that’

(3)

3rd step: pronouns der, die, (den)4 a. finite utterances with copula i. da ist der ii. weg ist die iii. der ist es, den there is he/it gone is it/she it/he is it, this/that-acc ‘he/it is there’ ‘it/she is gone’ ‘it/he is it, this/that’ b. finite utterances with modal and lexical verbs i. der auch losfährt ii. die anziehen will iii. ich trag den it/he too drives off it dress want I carry it ‘it/he drives off too’ ‘want to dress it’ ‘I am carrying it’

(4)

4th step: definite article + noun5 a. finite utterances with copula i. wo ist die ente? ii. da ist der Mäcki iii. die Oma war das where is the duck there is the Mäcki the grandmother was it ‘where is the duck?’ ‘Mäcki is there’ ‘it was grandmother’ b. finite utterances with modal and lexical verbs i. möchte die kette habe ii. der vogel macht? want the necklace have the bird does ‘I want to have the necklace’ ‘what is the bird doing?’

DPs usually appear between the 26th and 30th month of life in the acquisition of German (Clahsen 1984; Mills 1986; Müller 1994; Bittner 1998, 2002a). From this

.

In parallel to these forms also personal pronouns are acquired.

. Because this paper focuses on DPs, the emergence of indefinite noun phrases like ein haus ‘a house’ is not mentioned here. Indefinite NPs occur at approximately the same time as definite pronouns (3rd step), i.e. before the acquisition of complex DPs (Bittner 1998).

S F O PRO



1st proofs

DPs in the acquisition of German 

time on, the child’s grammar includes different forms of d-determiners: das, der, die, den and different types of DPs: pronominal DPs – complex DPs. The question to be pursued in the next sections is as to whether the different forms of d-determiners and the different types of DPs have different functions in early child grammar.

. .

Sentence-internal functions Distribution of d-determiners over syntactic positions

One possibility to unravel the functional difference between certain forms or structures is the analysis of their distribution over syntactic positions. The early data of the two children has been analysed according to (i) the verbal type of the utterances containing a d-determiner: no verb (−V), infinite verb (+Vinfinite), finite verb (+Vfinite), and – as a special type of finite verbs – copula; and (ii) the syntactic position of the d-determiner in utterances with infinite and finite lexical verbs: subject or object position. Tables 1 and 2 present the results of the analysis for each child.

Table 1. Simone – syntactic positions of the pronouns das, der, die, and den (tokens) +Vinfinite age: 2;2–2;5

−V

das der die den

77 5 32 24

+Vfinite

subject position

object position

subject position

object position

with copula sein ‘be’

2

5

54

2

3

26 43 20

48 19 13 1

2 2

Table 2. Anna – syntactic positions of the pronouns das, der, die, and den (tokens) +Vinfinite age: 2;1–2;4

−V

das der die den

71 12 14 11

+Vfinite

subject position

object position

subject position

object position

with copula sein ‘be’

2 4

12

28 40 73

64

112 10 25 2

3 4

17 9

S F O PRO



1st proofs

 Dagmar Bittner

Certain preferences in the distribution of the four forms can be registered. In general, they are rarely used in utterances with infinite verbs (das aufhängen ‘this/that hang.up-’, der schlafen ‘this/that (=he) sleep-’). This underlines the existence of a correlation between verbal finiteness and nominal definiteness suggested by the parallel emergence of these structures. The form das preferably occurs in verbless utterances (das ‘this/that’, das auch ‘this/that also’) and in utterances where the finite verb is the copula sein ‘be’ (das ist Mama ‘this/that is Mummy’, wo ist das? ‘where is this/that (=it)?’, meine bonbons sind das ‘this/that (=these) are my sweets’). Over time, das not only refers to concrete entities given in the situation of speaking, but also to complex propositions, for instance, ich will das nicht ‘I don’t want this/that’ can be used for ‘I don’t want to go to bed’ or ‘… to visit uncle Fritz’. In these cases, das occurs in utterances with finite lexical verbs and more frequently in object position than in subject position.6 In all positions, the main characteristic of das is its deictic function. The form der shows a strong preference to be used in utterances with finite lexical verbs. Furthermore, the absolutely preferred position is the subject position. Apparently, der is specified for subject arguments of finite verbs in early child grammar. The form die most frequently occurs in the subject position of finite utterances but at the same time, it occurs in object position. The impression of a subject preference might be caused by the fact that, due to its obligatory occurrence, subject phrases are much more frequent than object phrases.7 Taking this in consideration, die appears not to be specified for a certain argument position. It symbolizes an argument status of the DP, but leaves unspecified (under-specified) the type of the argument. The form den is not as frequent in the data as the other forms discussed so far, depending on its later acquisition and, furthermore, its preference for the object position, which – as has been said above – has not obligatory to be filled in an utterance. In all cases in that den appears in utterances containing a lexical verb,

. Children at the onset of the acquisition of syntactic structures can be assumed not to have acquired the syntactic categories of subject and object. More likely, they first acquire certain verb-dependent sentence frames and the distribution of semantic roles in those frames (HirshPasek and Golinkoff 1996). For this reason the terms ‘subject position/argument’ and ‘object position/argument’ are used here.

S F O PRO

. This is a circumstance that strengthens the above registered preference for object positions of das when it refers to complex propositions.



1st proofs

DPs in the acquisition of German 

it takes the object position. Despite its non-frequent use, it seems to symbolize object arguments. We have mentioned in Section 1 that, according to previous research, children show no difficulties to use d-determiners in appropriate contexts. The present analysis confirms these findings. The four forms are exclusively used in contexts and positions where d-determiners would be appropriate in adult language. Thus, the deictic and demonstrative capacities of these pronouns are in place from the onset of production. The functional properties of the four pronouns derivable from the above analyses are summarized in Table 3. Table 3. Functional properties of das, der, die, and den at the onset of production

das die der den

+deictic/ demonstrative

+verb argument

+subject argument

+ + + +

+ + +

+

+object argument

+

(5)–(8) present examples for the proposed functional specifications. (5)

+deictic i. das aufhängen? this/that hang.up ‘hang up this?’

ii. das ist ein krankenhaus this is a hospital ‘this is a hospital’

iii. ich mach das I do that ‘I do that’

(6)

+deictic/demonstrative pronoun – +argument i. die kommt gelauf ii. Mone zieht die auf she/it comes run-PP Mone winds this/that up ‘she comes running’ ‘Mone winds it up’

(7)

+deictic/demonstrative pronoun – +argument – +subject argument i. der spuckt hier immer aus ii. wird der immer nass he spits here always out will.get he/it always wet ‘he always spits out here’ ‘he/it will always get wet’

(8)

+deictic/demonstrative pronoun – +argument – +object argument i. du kennst den noch nicht ii. hier hab ich den you know him/it yet not here have I him/it ‘you don’t know him/it yet’ ‘I have him/it here’

Further evidence for the assumed functional structure is derivable from patterns of non-appropriate use of the four d-determiners with respect to case and gender specification.

S F O PRO



1st proofs

 Dagmar Bittner

.

Overgeneralization of das, der, die, and den

We have mentioned in Section 1 that young children show some difficulties to choose the appropriate case/gender form of definite pronouns and articles over a longer period. The results of the analysis presented in Section 2 allow predictions about the syntactic contexts in which each form might be overgeneralized. It has been found that the form das is not restricted to a certain syntactic position. Therefore, overgeneralization of das might occur in all syntactic contexts. The forms die, der, and den seems to be specified for argument positions of finite verbs. Therefore, overgeneralizations of these forms should occur primarily within finite utterances. The form die is not specified for a certain argument position, i.e. overgeneralizations in subject and object position are to expect. The form der is specified for subject position and the form den for object position, i.e. overgeneralizations should occur in exactly these positions. Table 4 presents the distribution of the nonappropriate case/gender use of definite pronouns in the early data of Anna. Table 4. Anna – overgeneralizations and uncertainties8 in the use of definite pronouns (tokens) 1;11.30–2;3.8

das die der

Vfin

Vinfin

total

subject position

object position

subject position

object position

21 7 8

8 2 8

11 5

1

1

Overgeneralizations of the pronoun den only occur after age 2;3. However, they show the expected distribution. Examples for all types of overgeneralizations are given in (9) and (10). The overgeneralized d-determiners show a mismatch in their gender specification with the gender of the noun. (9)

subject position i. der (krokodil) ist böse it/this- (crocodil-) is mean ‘it/this is mean’ ii. die (stift) geht gar nicht this- (-) goes at.all not ‘it/this is not working’

S F O PRO

. ‘Uncertainties’ means use of more than one pronoun in the same syntactic position, cf. (9ii) and (10i).



1st proofs

DPs in the acquisition of German 

iii. wie heißt das/puppe da? how is.called it/this-/doll- over.there ‘what’s the name of it/doll over there?’ (10)

object position i. die fang(en) die/de (hasen) they catch it-/it-  (rabbit-) ‘they catch it’ ii. ich hab die (krokodil) weggeschickt I have it/this- (crocodil-) sent.away ‘I have sent it/this away’ iii. nimm den (angel) take the- fishing.rod- ‘(I) take the fishing rod’

Although the number of overgeneralizations is small, the distribution is as expected. Children are aware of functional differences between the pronominal forms from early on. The overgeneralizations are based on hypotheses on the functional meaning of each form. However, the meaning remains underspecified in early child grammar. Overgeneralizations happen because the children have not yet learned that d-determiners are additionally specified for case and gender. Definite pronouns begin to be used as definite articles in the fourth step of the acquisition of DPs (cf. Section 2, especially (4)). In grammars on German, definite pronouns and articles are treated separately. It is worth noting that they are formally identical in most instances and show the same syntactic distribution in child as well as in adult grammar. (11)–(13) present examples from the children’s data. (11)

das krokodil frisst dich auf the- crocodile eats you up ‘the crocodile eats you up’

(12)

der Erik hat mich (ge)schubst the- Erik has me shoved ‘Erik has shoved me’

(13)

hol die lokomotive raus take the- train out.of ‘(I) will take the train out of ’ (a box)

der (hamster) frisst dich auf it/this- (hamster) eats you up ‘it eats you up’ die (Kosima) hat mich (ge)schubst she/the- (Kosima) has me shoved ‘she has shoved me’ ich mach die (nuss) auf I open it/this- (nut) up ‘I will open it’

Those pronominal and article forms that are not identical, i.e. the forms of genitive singular and plural as well as of dative plural,9 are acquired in a later phase of

S F O PRO

. Compare: gen.sg.fem: der (article) – deren (pronoun; gen.sg.msc/neut: des (article) – dessen (pronoun); gen.pl.: der (article) – deren (pronoun), dat.pl.: den (article) – denen (pronoun).



1st proofs

 Dagmar Bittner

the acquisition process. In the period considered here, only the forms of nominative, accusative and the dative singular are acquired. Thus, the pronominal DP consisting of a definite pronoun and the complex DP consisting of a definite article plus (at least) noun are directly related to each other. The complex DP can be viewed as an extension of the pronominal DP by adding the noun (and other elements). Therefore, overgeneralizations of definite articles are expected to show the same distribution as found with definite pronouns. Table 5 presents the distribution of Anna’s and Simone’s non-appropriate use of the definite articles der, die, and den in finite utterances. The longitudinal data has been analysed up to the end of the third year of life, i.e. up to 2;10 with Anna and 2;11 with Simone. The form das has been excluded from the analysis because it is hard to decide in most cases whether pronominal or article use is intended by the child; consider das Mama ‘this/that Mummy’ could be a case of omission of the copula (das ist Mama ‘this is Mummy’) or the wrong form of the definite article, i.e. das (the-) instead of die (the-). Table 5. Distribution of overgeneralized definite articles in the third year of life (tokens) subject position

Anna Simone

der

die

den

3 24

1 10

1

direct object position

indirect object position

der

die

den

der

die

den

3

1 6

2 5

3

3 12

4 9

As expected, overgeneralizations of der occur predominantly in subject position (14), overgeneralizations of den in object position (15), and die is distributed over all syntactic positions (16). (14)

overgeneralization of der i. psps macht der kanne psps does the-. coffee-pot-. ‘the coffee pot is saying psps’ ii. gehört denn der Fahrrad? owns then the-. bicycle-. ‘(who) owns the bicycle, then?’

(15)

overgeneralization of den i. die Mone den kopfkisse(n) braucht the- Mone the-. pillow-. need-3.. ‘the Mone needs the pillow’ ii. den Uhr habe the-. clock-. have ‘(I) have the clock’

S F O PRO



1st proofs

DPs in the acquisition of German 

(16)

overgeneralization of die i. is die fisch (weg) is the-. fish-. (gone) ‘the fish is (gone)’ ii. wer hat die fernseh(er)? who has the-. TV-.? ‘who has the TV?’

The tendencies found in overgeneralizations of definite articles also confirm the assumptions on the functional properties of each form given in Table 3. It is worth noting that there is no evidence for motivation of the overgeneralizations by semantic or phonological properties of the nouns as suggested by Mills (1986), Müller (1990, 1994), Koehn (1994) and others. .

Conclusion: Sentence-internal properties of d-determiners

Recall firstly that children do not show any difficulties in using d-determiners in the appropriate contexts, i.e. they produce a DP in contexts allowing a DP, and do not use a DP in contexts disallowing a DP. This means that children acquire d-determiners as a special class of determiners, distinguished from the other classes of determiners. So far, the common terms ‘definite pronouns’ and ‘definite articles’ have been used to label the class and the terms ‘deictic’ and ‘demonstrative’ to characterize the general properties that are shared by all members of the class. In the literature, however, there is some debate on the property or property cluster constitutive for the class of d-determiners. Depending on the theoretical framework, [+definite], [+individualized], [+known], [+bounded], [+totality], or [+specificity] or combinations of them are suggested as the constitutive properties (compare among others Heim 1982/1988; Löbner 1985; Leiss 2000). The present analyses do not provide a base to decide which is the most appropriate assumption. Nevertheless, the analyses provide clear evidence that children acquire these class-constituting properties before they start to produce d-determiners. The first determiner form children produce is the deictic pronoun das. The acquisition of the pronouns der, die, and den is related to the emergence of the finite verb. The three forms specify the argument status of the DP. Nevertheless, also these pronouns point to something given in the actual situation. The difference between child grammar in the first half of the third year of life and adult grammar lies in the realization of case and gender features. The analysis of the overgeneralizations of d-determiners showed that children differentiate subject arguments and object arguments from early on. However, the two types of object arguments (direct/accusative and indirect/dative) remain undistinguished during the third year of life. The overgeneralizations also reveal that gender classes

S F O PRO



1st proofs

 Dagmar Bittner

are generally not distinguished in the early period.10 Only at the end of the third year of life, children distinguish feminines from non-feminines (see also Müller 1994; Bittner forthcoming). The distinction of masculines and neuters emerges even later in development. Hence, it can be argued that in the acquisition of the inflectional paradigms of d-determiners, case (distinction) precedes gender (distinction).11 Children firstly establish a uniform case paradigm allowing the distinction of the argument status of the DPs. In this paradigm, die specifies that a DP is an argument, without saying what type of argument. The form der specifies subject arguments and den object arguments. The form das is not included in this early paradigm. The development from das used as a deictic pronoun in infinite and copula constructions to das productively used in argument position takes about four months. This explains the late emergence of das in complex DPs and the comparatively high amount of overgeneralizations with neuters. The order of acquisition exhibits a stepwise differentiation of the grammatical properties. The development starts with the most general class-constituting property/ies shared by all d-determiners. Other properties present in adult grammar remain unspecified in the early period. In the next step, case-related properties emerge, but again certain properties remain unspecified (cf. the die form and the gender features in general). The pattern of underspecification found in the acquisition data supports considerations on Gesamtbedeutung and underspecification in adult grammar (cf. among others, Jakobson 1936/1971; Marantz 1984; Bittner 2002b). Certain forms, as for instance die, will not be further specified in the later acquisition process. The overgeneralization of those forms disappears because of the acquisition of the adult like properties of forms which are appropriate in the contexts of – for instance – die overgeneralization. The property [+definite] or [+determined] which is associated with the die form is the only property or feature of this form in adult grammar too (e.g. Bierwisch 1967; Bittner 2002b). Case or gender properties remain underspecified in the die form.

. .

Sentence-external functions Goal and method of analysis

The use of definite pronouns as articles marks the point in development when two different types of DPs exist in the child’s grammar: pronominal DPs and .

See also Wegener (2000) for second language acquisition.

S F O PRO

. The appearance of case distinctions prior to gender distinctions is discussed in more detail in Bittner (forthcoming).



1st proofs

DPs in the acquisition of German 

complex DPs. Both types show the same constraints in syntactic distribution and morpho-syntactic behavior of the determiner. The question is: are there functional differences between the two DP types in early child language? An exhaustive answer to this question demands analyses of a wide range of different aspects, such as word order, noun class of the overt/covert noun, syntactic structure of the related utterances, the interplay between the different types of definite determiners, amongst others. The analysis presented here concentrates on the deictic and demonstrative capacities of the two DP types. Again, the analysis is based on the longitudinal data of the two girls, Anna and Simone, which are described at the end of Section 1. It is worth noting that the data constitute a specific type of text. They consist of adult-child dialogues in everyday communication. All kinds of entities spoken about are typically present in the situation of speaking and therefore present in the actual knowledge of speaker and hearer. Additionally, the dialogues are characterized by frequent changes of the speech topic. Our hypothesis is that there is a functional difference between the two types of DPs in early child language. Assumptions about the character of these differences can be derived from theories on discourse structure and text coherence. Givón (1995), for instance, stated:12 The dichotomy between grammatical devices that signal continued reference vs. those that signal disrupted reference … coincides, in the main, with the dichotomy between purely grammatical referential devices, and those that combine grammarcuing with lexical information. The most obvious source of lexical information in the latter group is the lexical noun itself. (Givón 1995: 375; emphasis by the author)

Accordingly, the functional difference between pronominal and complex DPs should be related to continued vs. disrupted reference. A pronominal DP is expected to signal that the referent foregrounded by the previous utterance(s) is the referent of the actual DP, whereas the complex (noun-including) DP is expected to signal that a new referent is introduced. Applied to our adult-child dialogues, this can be viewed as the difference between anaphoric and deictic reference. Because all potential referents are given in the situation of speaking, all of them are available for deictic reference. However, only some of them will be available for anaphoric reference at a given time in the dialogue. Karmiloff-Smith (1981, 1985) proposed that children younger than age 6 have not acquired the difference between anaphoric and deictic reference. In her study of narratives produced on the base of picture-stories, younger children used DPs for

.

See also Grosz et al. (1995); Levinson (2000).

S F O PRO



1st proofs

 Dagmar Bittner

deictic reference irrespective of the structure of the DP and the linguistic context. Particularly, consecutive pronominal DPs often referred to distinct referents. These findings contradict our hypothesis on a functional difference between both types of DP in early child language. Moreover, they imply that the functional differences proposed by Givón for adult language are not in place in early child language. The following analysis examines the use of pronominal and complex DPs with respect to anaphoric and deictic reference. Because the form das refers to concrete entities as well as to complex propositions, it has been excluded from the analyses. Also excluded are all utterances resembling complete imitations of the preceding adult utterance, immediate repetitions of the child’s own utterances, utterances containing parts of songs and rhymes and, finally, unanalysed holophrases.

.

Anaphoric and deictic use of DPs

The two girls show a slightly different onset of the production of DPs. Pronominal DPs emerge earlier and are more frequent than complex DPs in the data of Anna. In the data of Simone, the proportion between the two types is nearly balanced in the beginning. The later emergence of DPs in Simone’s data corresponds with its overall smaller amount during the analysed period. Table 6 gives the proportions of anaphoric vs. deictic use of both types of DPs.

Table 6. Anaphoric and deictic use of pronominal and complex DPs (tokens)

age

total number of utterances

pronominal DP (der, die, den) anaphoric use

deictic use

complex DP (der, die, den + noun) anaphoric use deictic use

ANNA 2;0.29 2;3.8 2;3.29 2;4.19 total

524 513 441 377 1855

11 63 32 22 128

4 17 9 5 35

2 14 11 3 30

5 23 21 6 55

SIMONE 2;2.21 2;4.17 2;4.21 2;5.13 total

420 472 343 457 1692

5 18 5 31 59

1 2 1 – 4

5 13 6 6 30

1 6 10 12 29

S F O PRO



1st proofs

DPs in the acquisition of German 

Both children preferably use the pronominal DP for anaphoric reference. 78,5% (Anna) and 93,6% (Simone) of the pronominal DPs fall in this category:13 (17) ADL: rechts rum läuft  (ente) CHI:  kommt gelauf

‘this/that (=it, duck) turns right’ ‘this/that (=it) comes running’

(18) ADL: da oben soll  sein? (puppe) CHI: ja –  schläft

‘this/that (=it, doll) is on the top, you say?’ ‘yes, this/that (=it) is sleeping’

(19) ADL: erzähl mal der Mama, ‘tell your Mum was die K gemacht hat what Kosima has done’ CHI:  hat mich schubst [: geschubst] ‘this/that (=she) has shoved me’

The distribution of the complex DP is less strictly determined. Anna shows a preference for deictic use. This holds for all of the four recordings; in total, for 67,5% of her complex DPs. In the data of Simone, the proportion of deictic and anaphoric use is balanced if one considers the total numbers. However, comparing the data of age 2;2.21 and 2;4.17 with that of age 2;4.21 and 2;5.13, a change from anaphoric to deictic use is observable. Anaphoric use is comparatively frequent in the first two recordings but decreases in the last two recordings. In the latter, deictic use dominates to 64,7%, which is about the same proportion as found in Anna’s data: (20) ADL: mit was spielen wir denn? CHI: baukloetzer – mit’n baukloetzer spielen wir act:14  springt zur kiste mit den baukloetzen ADL: mit’n baukloetzen solln wir spieln? say?’ CHI: ja CHI: hol  L raus act:

CHI holt eine lokomotive aus der kiste

(21) ADL: da steht ein bisschen saft für dich und ne banane ADL: is(t) das gut? CHI: ja ADL: sollen wir das holen? CHI: ja

‘what will we play with?’ ‘bricks, with the bricks we will play’ ‘CHI goes to the box with the bricks’ ‘we will play with the bricks, you ‘yes’ ‘(I) take the engine out (of the box)’ ‘CHI takes the engine out of the box’ ‘there is a bit juice for you overthere and a banana’ ‘is that good?’ ‘yes’ ‘shall we fetch it?’ ‘yes’

.

In the following examples, the abbreviation ADL means ‘adult’ and CHI means ‘child’.

.

The ‘act’-line provides a description of the situation.

S F O PRO



1st proofs

 Dagmar Bittner

ADL: komm, wir gehn mal in die küche raus CHI: ja ADL: kannste alleine raus aus’m bett? CHI: ADL: CHI: act:

 P soll da auch xxx15 was sagste?  P nimmt die puppe aus ihrem bett

‘come on, we will go into the kitchen’ ‘yes’ ‘can you come out of the bed by yourself?’ ‘the doll shall also xxx’ ‘what are you saying?’ ‘the doll’ ‘CHI takes the doll out of the bed’

Table 7 summarizes the observed correlations. It can be argued that there is a tendency to use the pronominal DP for anaphoric reference and the complex DP for deictic reference. Table 7. Preferred correlation of DP type and reference type type of the DP

preferred type of reference

pronominal DP

anaphoric

complex DP

deictic

referent is mentioned in the directly preceding linguistic context referent is not mentioned in the directly preceding linguistic context, but is given in the actual situation

However, the correlation of the DPs with either anaphoric or deictic reference is not strong enough to infer a clear functional difference. The correlations rather resemble preferred or unmarked use of each type. In order to further clarify whether there is a clear functional distinction or not, the use of the DPs in the non-preferred or, henceforth, marked contexts has to be analysed. .

DPs in marked contexts

According to the findings of the previous section, a pronominal DP occurs in a marked context when there exists no antecedent in the linguistic context, i.e. when it is used deictically. On the contrary, a complex DP occurs in a marked context when there is an antecedent in the linguistic context, i.e. when it is used anaphorically. Deictic use of pronominal DPs is comparatively rare (cf. Table 6). The single instances are often accompanied by a pointing gesture: (22) CHI: ADL: CHI: act:

das ist ein (k)rankenhaus und wer geht da rein, ins kranhkenhaus? mmm,  hier (den = der16) zeigt eine duplo-figur

. ‘xxx’ signals unintelligible speech.

‘this/that is a hospital’ ‘and who will go into the hospital?’ ‘mmm, this/that one’ ‘the child shows a duplo-figure’

S F O PRO



1st proofs

DPs in the acquisition of German 

Gestures similar to pointing appear when speaker and hearer play with the same objects. Pronominal DPs may refer to objects which are not mentioned before but set in action by child or adult immediately before or at the moment of speaking about them. The pronominal DP, thus, refers to the actually manipulated (pointed at) object. However, there are certain instances for which no pointing gesture is reported and it is to expect that indeed there was no one. This seems to be typical when a newly appearing situation contains one especially prominent character. For instance, in picture books, the situation changes with each new page or picture, either because there are different objects in each picture or because of the fact that young children do not identify characters in different pictures as being identical. Change of situation (or topic) is a typical property of the type of dialogues analysed here. In (22), such change is initiated by the insertion of the new referent ein krankenhaus ‘a hospital’ and the question of the adult. In (23), an incoming phone call brings about a new situation. (23)

telefon klingelt,  geht ans telefon, telefoniert mit der mutter von  ADL: gut, ok, ja, gut ADL: willst du deine Anna mal haben? act:

act: act: CHI: act: CHI:

frage an die mutter  bekommt das telefon hallo, mhm, mhm, tschuess! beendet das Gespräch da ist  weg

‘the phone rings,  takes the phone, talks to the ’s mother’ ‘fine, ok, yes, alright’ ‘do you want to speak to your Anna?’ ‘addressee ist the ’s mother’ ‘ gets the phone’ ‘hallo, mhm, mhm, bye-bye’ ‘finishes the call’ ‘now  is gone’

Both adult and child were talking to the mother, thus, the referent of the DP die in the last sentence of (23) is present to both of them. In sum, 27 of the 35 instances of deictic use of pronominal DPs in Anna’s data, i.e. 77%, coincide with a pointing gesture or the onset of a new situation containing a most prominent character. For two of the remaining instances of pronominal DPs it can be assumed that the child has still not acquired the appropriate noun. In three other instances the pronominal DP is immediately followed by a complex DP (compare der schläft nicht, der teddy ‘he/it is not sleeping, the teddy-bear’).

. Example (22) is taken from the data of Simone. Simone shows several instances of using  ‘this/that-’ as deictic pronoun. The occurrence of den instead of der ‘this/that-’ in (22) is not a counter-example to the assumption that den symbolizes object arguments and der subject arguments of finite lexical verbs proposed in Section 3.1 and 3.2, since den here occurs in a verbless utterance.

S F O PRO



1st proofs

 Dagmar Bittner

Marked use of complex DPs, i.e. anaphoric use, is documented to a higher amount for both children (cf. Table 6). The analysis of the respective contexts allows for the classification of certain types of contexts provoking anaphorical use of complex DPs. A first type is: the child’s utterance follows a question expressing misunderstanding of the previous utterance(s). (24)

ADL: ADL: CHI: CHI: ADL: CHI:

mal sehn, ob’s dir gut schmeckt is(t)  gut, Mone? ja kalt – kalt kalt? W ()   M is(t) kalt

(25)

CHI: mein Nuckel ADL:  ()  CHI:  N

‘Let’s see if you like the taste (melone)’ ‘Is  (=melon) good, Mone?’ ‘yes’ ‘cold, cold’ ‘cold?   ?’ ‘  is cold’ ‘my dummy’ ‘  ?’ ‘ ’

A second and closely related to the first type is: emphatic repetition of utterances because the adult shows no reaction. (26)

CHI:   act: CHI will stifte haben ADL: (zu 2. ADL) du, diese sachen gefallen mir ADL: (zu 2. ADL) die sind so phantasievoll CHI:   (=stifte) ADL: ja CHI:   da habe

‘ ’ ‘CHI wants to get crayons’ ‘(to 2nd ADL) I like these things’ ‘(to 2nd ADL) they are so imaginative’ ‘ ’ ‘yes’ ‘(I) want   over there’

The third type is: utterances containing a negation like nein ‘no’ or nicht ‘not’ or a contrastive conjunction like aber ‘but’. (27)

ADL:  (nüsse) können wir alle noch essen ADL: nicht die Schale, die Schalen nicht CHI: aber  ü

‘we can eat all of  (nuts)’ ‘not the shells, the shells not’ ‘but  ’

Finally, there are utterances which either contain a complex DP in an immediate repetition of a less complex one (28) or in post position, i.e. at the right edge of the utterance (29). (28)

ADL: CHI: ADL: CHI:

kannst du Fische fangen? ja ja? M auch –  M auch

‘can you catch fishes?’ ‘yes’ ‘yes?’ ‘M too –  M too’

S F O PRO



1st proofs

DPs in the acquisition of German 

(29)

CHI:  kommt gelauf (ente) ADL: oh, gucke mal CHI: nich(t) putt mache   (=subj.) act:

die ente rennt gegen die klötze, so dass der stall kaputt zu gehen droht, CHI läuft zur ente

‘/ (=it, duck) comes running’ ‘oh, look’ ‘  should not destroy (barn)’ ‘the duck is running against the bricks so that the barn seems to break down, CHI rushes to the duck’

In all these types of contexts, certain emphasis is put on the complex DP. It marks either a contrast to another object, an expressive repetition, or a later specification of a DP or NP uttered before. In the data of Simone, 22 of the 30 instances occur in one of these contexts. In the data of Anna, these are 21 of 30. In sum, about 70% of the anaphorically used complex DPs are related to contrastive or emphatic use. . Conclusion: sentence-external properties of DPs The distribution of pronominal and complex DPs in the first months after the onset of their production reveals a clear tendency towards distinct functions of the two types of DPs. Pronominal DPs are preferably used when the referent of the DP was already mentioned in the preceding utterance(s). Complex DPs are preferably used when the respective referent was not mentioned in the preceding utterance(s) but is present in the situation of speaking. According to that pattern, the functional difference between both types of DPs is related to the opposition of anaphorical vs. deictic reference. In contrast to the results of Karmiloff-Smith (1981, 1985), our longitudinal data provide clear evidence for anaphorical use of pronominal DPs at an early age. 80% to 90% of them have an antecedent in the preceding utterance(s). The opposition between anaphorical and deictic use correlates with Givóns distinction of continued vs. disrupted reference. Anaphorical use of pronominal DPs causes or symbolizes continued reference, deictic use of complex DPs disrupted reference. In other words, children use the DP structure in order to guide the hearer’s search for the referent, either towards the preceding utterances (pronominal DPs) or to the extra-linguistic context (complex DPs). However, the functional difference can not be related to the mere opposition of deictic vs. anaphoric use. Occurrences in the non-preferred reference type are well motivated in most cases too. In other words, even so called marked uses follow a certain pattern of distribution. Deictic use of pronominal DPs exhibits two patterns: (a) the newly introduced referent is identified by non-linguistic

S F O PRO



1st proofs

 Dagmar Bittner

means, typically a pointing gesture,17 or (b) a new situation emerging in the discourse includes one entity which is more prominent than other entities in the same situation (e.g. pictures/situations showing only one main or a certain prominent character). Both patterns (a) and (b) have something in common: the referent of the pronominal DP is the most prominent referent in the actual situation, either it is made prominent by a pointing gesture or its prominence is grounded in the structure of the situation. Consequently, anaphorical and deictic use of pronominal DPs have something in common: the referent is (or belongs to) the most prominent referent(s) in the situation, either because of linguistic means (beforementioned) or because of non-linguistic means (pointing gestures or situational prominence). Anaphorical use of complex DPs can be subsumed under two patterns too: (a) giving special emphasis on the referent, often including a contrast to other entities of the situation, and (b) repetition of the already mentioned DP because the hearer shows no (or not enough) attention on the referent. Again, both patterns have something in common: the speaker emphasizes that the referent is prominent in the situation, he demands the attention of the hearer on that certain entity. In order to recognize that there is a correspondence also between deictic and anaphoric use of complex DPs, recall that deictic use is defined here as having no antecedent in the preceding utterance(s). Deictic use includes the introduction of a new prominent or central referent in the actual linguistic discourse. Having this in mind, it can be argued that there is certain identity in deictic and anaphorical use of complex DPs: the referent is explicitly marked to be the actually most prominent referent in the situation. The complex DP is used in order to signal that the hearer should foreground the referent in its actual representation of the situation and the discourse. The correspondence in the unmarked and marked uses allow for the hypothesis of a general functional distinction of pronominal and complex DPs in child grammar from early on. Table 8. Functional opposition of pronominal vs. complex DPs in early child grammar type of the DP

function

pronominal DP

symbolizing that the referent is one of the entities which are (already) most prominent in the actual situation symbolizing that the referent is an entity which has not been most (or enough) prominent so far, at least in the attention of the hearer

complex DP

S F O PRO

. Children can even try to direct the attention of the hearer by head or eye movement. These types of gestures are not registered in the transcripts of our longitudinal data.



1st proofs

DPs in the acquisition of German 

In the theoretical approach of Givón (1995), continued and disrupted reference are related to the symbolization of the topic of the actual part of the discourse. Possibly, the notion of topic is involved in early child language too. However, more detailed research is required to determine topic-focus relations in early child language. For the time being it can be stated that the children use the two DP types in order to guide the hearer’s search for the appropriate referent (Table 8). Children do that by applying universal constraints on the structural complexity of DPs in dependence on the prominence or salience18 of their referents. Referents already prominent in the actual discourse or situation can be referred to by a minimal complex DP, i.e. a pronominal DP. Referents intended to be made prominent have to be referred to by a more complex DP, specifically there is a need of lexical information to identify the appropriate referent. The structural correlation is iconic: less content – less formal complexity, more content – more formal complexity. Iconic relations are supposed to be determined by universal cognitive constraints on language structure (Peirce 1960; Posner 1980; Mayerthaler 1981; Givón 1985; Dressler 1995) and as such they are expected to emerge early in language acquisition (Dressler 1997).

.

Summary

In summarizing the results of the analyses it can be argued that: 1. Children have command of the class-constituting properties of d-determiners from early on, i.e. these properties seem to be acquired prior to the onset of the production of d-determiners. The exact nature of these properties is, however, a matter of debate. What can be said is that they are clearly related to the individualization of the referent. The vitality of the respective properties in early child grammar is evidenced by the fact that children never use d-determiners or DPs in contexts disallowing a DP in adult language. Specifically indefinite and definite NPs exhibit strong distributional opposition (see Section 3.3). 2. Children acquire case-related properties of d-determiners during the first two to three months of their production. The sentence-internal function of d-determiners is seen in differentiating the argument status of the DPs. Overgeneralizations of determiner forms reveal children’s hypotheses on their functional load: der is specific

. For salience as the resulting value out of a diversity of criteria determining the prominence of an NP for intersentential resolution of anaphora, compare recent works in Centering Theory (e.g. Grosz et al. 1995).

S F O PRO



1st proofs

 Dagmar Bittner

for subject arguments, den for object arguments, whereas die is underspecified with respect to a determined argument position, it only assigns ‘argument-being’. The relation of the different forms to argument specification is underlined by the fact that they are acquired when finite lexical verbs becomes productive in child speech. There is no evidence for gender-related properties of d-determiners in the third year of life (see Section 3.3). 3. Children acquire discourse-related properties of DPs during the first two to three months of the production of pronominal and complex (i.e. noun-including) DPs. The functional opposition between the two DP types is related to the prominence of the referent in the actual part of the discourse. Pronominal DPs symbolize continued or directly to access reference, i.e. they symbolize that the referent which is (already) in the shared focus of attention is the actual referent. Complex DPs symbolize disrupted or especially emphasized reference, i.e. they symbolize that the hearer is intended to direct her/his attention to a referent which has not (or not especially) been in the shared focus of attention so far (see Section 4.4). These results challenge recent assumptions on the functional load of DPs in early child grammar. On the contrary to assumptions of, for instance, Mills (1986); Müller (1994); Clahsen (1984) and Tracy (1986), case-related properties are assigned to d-determiners prior to gender-related properties. On the contrary to assumptions of Karmiloff-Smith (1981, 1985), children have command of anaphoric use of DPs very early. With respect to the latter, it can be assumed that the diverging findings are related to the different types of texts analysed. Karmiloff-Smith analysed narratives on the base of picture stories. It is well known that children have difficulties to interpret pictures as related to each other for a longer time. Instead, they tend to consider each picture as a separate story with characters not identical to the characters on the preceding and following pictures. This leads to the impression that children are not capable of expressing continued reference, i.e. anaphorical relations by linguistic means. However, adult-child dialogues during joint actions with a certain set of objects provide evidence that the children are capable to differentiate between referents which are already in the focus of shared attention and which are not, and, furthermore, to relate certain linguistic structures to that opposition. The results of our study provide empirical material for the theoretical discussion of various aspects of sentence-internal and sentence-external properties of DPs. The most central are in our view: (a) the character of the functional properties established in early child language, (b) the role of salience and shared focus of attention in early acquisition, and (c) the “genetic” relation of pronominal and complex DPs.

S F O PRO



1st proofs

DPs in the acquisition of German 

a) character of the early acquired functional properties The acquisition processes summarized above (cf. 1–3) are part of the overall process of acquiring grammatical means for the determination of linguistic signs in the sense of Coseriu (1975), i.e. in order to restrict, determine, and orientate the reference of the (virtual or actual) sign. The early acquired properties of DPs concern relations of a DP or its referent to other parts of the utterance(s): Sentenceinternally these are verb-argument or argument-argument relations; sentenceexternally these are degrees of prominence of referents within the actual discourse situation. Properties affecting the noun concept itself like gender and number distinctions are seemingly not subject to early hypotheses on DP properties.19 As Coseriu (1975: 264) pointed out, definite articles are the means par excellence to actualise a nominal concept. However, definiteness not merely affects the interpretation (Lesart) of the nominal concept as such but has a more general grammatical scope. The emergence of pronominal DPs before noun-including DPs and the fact that these pronominal DPs can even be used when the child still has not acquired an appropriate noun for the intended referent highlight that d-determiners are primarily means to direct the attention of the hearer to a certain individual referent irrespective of an existing nominal concept. The use of a d-determiner is an advice to search for the appropriate referent. They do not provide lexical-semantic features to narrow the reference space. The actualisation is more or less hypothetical as long as the hearer has not identified the intended referent. The identification process is determined by the relational information given by the d-determiners, basically the range of prominence of the entities in the described situation. Other types of relations are actualised in dependence of the language-specific pattern, i.e. which further properties beside definiteness are carried by d-determiners. It can be concluded that the process to restrict, determine, and orientate the reference by d-determiners is an act of determination which relies on the actualisation of relations, not of internal specification of the referent. b) the role of salience and shared focus of attention Although the present study is at a very preliminary stage, it gives rise to the conclusion that children are sensitive for some basic mechanisms of ensuring text coherence from early on. It has been shown that the use of the two types of DPs is guided by the opposition of continued and disrupted reference (Givón 1995). The decision when which type of reference is appropriate is based on a hypothesis

. Although children produce plural forms in the analysed period, it has to be assumed that the respective forms are learned by heart. Analyses of non-appropriate plural forms reveal a tendency to omit plural suffixes during the 3rd year of life. The child is still not aware of certain regularities of plural marking (Bittner 2000).

S F O PRO



1st proofs

 Dagmar Bittner

whether the intended referent is prominent enough in the representation of the situation by the hearer. According to this hypothesis, the child uses the less or the more complex DP type. Obviously, the choice of the DP type is determined by the principles of saliency and iconicity (cf. Lewis 1970; Levinson 2000): formally less complex structures refer to high salient referents and formally more complex structures refer to less salient referents. It is worth pointing out the cognitive base of this functional opposition. Obviously, the child is able to distinguish between objects lying within the shared focus of attention and such that are not. This means, s/he has some knowledge of the conditions of human interaction involving the generation of hypotheses whether the hearer shares the focus of attention or not. This is evidence for early abilities to take into consideration the perspective of the hearer (cf. Tomasello 2003: 3–4). Furthermore, the ability to decide which DP type is appropriate in which context reveals that the child has some basic knowledge of so-called pragmatic or conversational implicatures (Levinson 2000), at the very onset of text construction. Very likely, parts of this basic knowledge belongs to the genetic endowment of human beings, e.g. the strong focus of getting and directing attention and shared focus of attention. Other parts, especially those related to conversational implicatures, seem to become acquired in the pre-linguistic phase. Assumptions on a late(r) acquisition of grammatical distinctions guiding text coherence, especially later acquisition of anaphoric use of definite pronouns (Karmiloff-Smith 1981; Hickmann 2003) have to be refined. The type of text and the situation of text production (e.g. shared knowledge or not) play a decisive role in the choice of linguistic means. c) “genetic” relation of pronominal and complex DPs Pronominal use of d-determiners precedes article use, i.e. the children have command of pronominal DPs before they acquire complex noun-including DPs. This is in accordance with assumptions on the status of complex DPs in the grammatical system as well as with the order of temergence of the DP types in language history (compare Abraham 2006, this volume). The noun-including DP is considered the most complex kind of definite construction both formally and functionally (see studies on definiteness and reference, Heim 1982/1988; Löbner 1985). The pronominal DP refers to entities already individualized and in the focus of attention. The complex noun-including DP has the function to bring about the status of an individual referent. The pronominal DP is exclusively based on the linguistic and non-linguistic context, no special linguistic effort to solve reference is necessary, it points to the referent seen as the undoubted most salient entity. The course of development and the structural relations between the two types of DPs suggest that the formally less complex type, the pronominal DP, is at the

S F O PRO



1st proofs

DPs in the acquisition of German 

base of the development of DPs. Probably, the complex DP has to be viewed as a contextually determined extension of the pronominal DP. The emergence of the complex DP signals a step from exclusively situation-bound deixis to linguistic/ grammatical deixis in the domain of the DP.

References Abraham, Werner. 2006 This volume. Anaphoric deictics in German: present – and diachronic changes: Non-clause binding pronouns under discourse perspective. Bierwisch, Manfred. 1967. Syntactic features in morphology: General problems of so-called pronominal inflection in German. In To honor Roman Jakobson. Essays on the occasion of his seventieth birthday 11 October 1966, Vol. I, 239–270. The Hague: Mouton de Gruyter. Bittner, Dagmar. 1998. Entfaltung grammatischer Relationen im NP-Erwerb: Referenz. Folia Linguistica XXXI(3–4): 255–283. Bittner, Dagmar. 2000. Sprachwandel durch Spracherwerb? Pluralerwerb. In Angemessene Strukturen. Systemorganisation in Phonologie, Morphologie und Syntax, Andreas Bittner, Dagmar Bittner & Klaus-Michael Köpcke (eds), 123–140. Hildesheim: Olms. Bittner, Dagmar. 2002a. Emergence of grammatical complexity and markedness in the acquisition of verb and noun phrases in German. In Future Challenges for Natural Linguistics, Katarzina Dziubalska-Kolaczyk & Jarek Weckwerth (eds), 25–56. Munich: Lincom. Bittner, Dagmar. 2002b. Semantisches in der pronominalen Flexion des Deutschen. Zeitschrift für Sprachwissenschaft 21(2): 196–233. Bittner, Dagmar. Forthc. Case before gender in the acquisition of German. Folia Linguistica. Clahsen, Harald. 1984. Der Erwerb von Kasusmarkierungen in der deutschen Kindersprache. Linguistische Berichte 89: 1–31. Coseriu, Eugenio. 1975. Determinierung und Umfeld. In Sprachtheorie und Allgemeine Sprachwissenschaft. 5 Studien, Eugenio Coseriu, 253–290 München: Fink. Dressler, Wolfgang U. 1995. Interactions between iconicity and other semiotic parameters in language. In Iconicity in Language [Current Issues in Linguistic Theory, 110], Raffaele Simone (ed.), 21–37. Amsterdam: John Benjamins. Dressler, Wolfgang U. 1997. Studies in pre- and protomorphology. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Givón, Talmy. 1985. Iconicity, isomorphism, and non-arbitrary coding in syntax. In Iconicity in Syntax, John Haiman (ed.), 187–219. Amsterdam: John Benjamins. Givón, Talmy. 1995. Functionalism and Grammar. Amsterdam: John Benjamins. Grosz, Barbara, Joshi, Aravind & Weinstein, Scott. 1995. Centering: A framework for modeling the local coherence of discourse. Computational Linguistics 21: 203–225. Heim, Irene. 1982/1988. The Semantics of Definite and Indefinite Noun Phrases. New York NY: Garland. Hickmann, Maya. 2003. Children’s Discourse. Person, space, and time across languages. Cambridge: Cambridge University Press. Hirsh-Pasek, Kathy & Michnik Golinkoff, Roberta. 1996. The Origins of Grammar. Evidence from early language comprehension. Cambridge MA: The MIT Press. Jakobson, Roman. 1936/1971. Beitrag zur allgemeinen Kasuslehre. Gesamtbedeutungen der russischen Kasus. Roman Jakobson: Selected Writings II, 23–71. The Hague: Mouton de Gruyter.

S F O PRO



1st proofs

 Dagmar Bittner Karmiloff-Smith, Annette. 1981. The grammatical marking of thematic structure in the development of language production. In The Child’s Construction of Language, Werner Deutsch (ed.), 121–147. New York NY: Academic Press. Karmiloff-Smith, Annette. 1985. Language and cognitive processes from a developmental perspective. Language and Cognitive Processes 1: 61–85. Koehn, Caroline. 1994. The acquisition of gender and number morphology within NP. In Bilingual First Language Acquisition. French and German grammatical development, Jürgen M. Meisel (ed.), 29–51. Amsterdam: John Benjamins. Levinson, Stephen C. 2000. Presumptive Meanings. The theory of generalized conversational implicature. Cambridge MA: The MIT Press. Lewis, David. 1970. General semantics. Synthese 22: 18–67. Leiss, Elisabeth. 2000. Artikel und Aspekt. Die grammatischen Muster von Definitheit. Berlin: Mouton de Gruyter. Löbner, Sebastian. 1985. Definites. Journal of Semantics 4: 279–326. Marantz, Alec. 1984. On the Nature of Grammatical Relations. Cambridge MA: The MIT Press. Mayerthaler, Willi. 1981. Morphologische Natürlichkeit. Wiesbaden: Athenaion. Mills, Ann. 1986. The Acquisition of Gender. A study of English and German. Berlin: Springer. Müller, Natascha. 1990. Developing two gender assignment systems simultaneously. In Two Tirst Languages – Early grammatical development in bilingual children, Jürgen M. Meisel (ed.), 194–234. Dordrecht: Foris. Müller, Natascha. 1994. Gender and number agreement within DP. In Bilingual First Language Acquisition. French and German grammatical development, Jürgen M. Meisel (ed.), 53–88. Amsterdam: John Benjamins. Peirce, Charles S. 1960. Collected Papers. Vol. II: Elements of logic, Charles Hartshorne & Weiss, Paul (eds). Cambridge MA: Harvard University Press. Posner, Roland. 1980. Ikonismus in den natürlichen Sprachen. Zeitschrift für Semiotik 2: 1–6. Tomasello, Michael. 2003. Constructing a Language. A usage-based theory of language acquisition. Cambridge MA: Harvard University Press. Tracy, Rosemarie. 1986. The acquisition of case morphology in German. Linguistics 24: 47–78. Wegener, Heide. 2000. German gender in children’s second language acquisition. In Gender in Grammar and Cognition, Barbara Unterbeck & Matti Rissanen (eds), 511–544. Berlin: Mouton de Gruyter.

S F O PRO



1st proofs

The discourse-functional crystallization of the historically original demonstrative Werner Abraham Universität Wien, Austria

In Abraham (this volume), grammatical determiners of various sorts were discussed in regard of the extent to which such determiners co-define anaphors in contexts reaching beyond the single clause. Making use of its findings, the present discussion shows how the definite article emerged in the history of German from the determiner homonym and how various steps in their grammaticalization paths invite certain generalizations of diachronic change.1

.

The synchronic argument for the diachronic emergence of the definite article in German

In Abraham (this volume), the synchronic basis for the ensuing diachronic discussion on the emergence of Art in Old High and Middle High German up to Modern standard German was laid. We found the article pronoun der/die/das, ArtPro, the predecessor of the definite article function, to be crucially discourse bound. See (1) (the original (54) in Abraham this volume). (1)

a.

b.

ArtPro/DemPro restricts its antecedent relation to the latest mention of the set-referentially independent rhema, in contrast to PersPro, which presupposes a set-referentially dependent antecedent chain. This antecedent represents the clause-in-ternal thema. PersPro does not find an antecedent in a pure rhema chain, independent of the fact which clause function (grammatical category) is concerned.

This is what we take as a point of departure for discussing the historical material. Notice that we have the handicap in that the historical material allows no concrete

. This paper has profited immensely from discussions with Elisabeth Leiss (Munich), the author of the important Leiss 1994 and 2000. Her input and unrelenting scrutiny is hereby gratefully acknowledged. Thanks are due also to Elisabeth Stark for scrutinously thinking along.

S F O PRO



1st proofs

 Werner Abraham

conclusions about accent distributions and, consequently, discourse functions – particularly also because much of the material occurs in origin-true translations from Latin (Tatian). Thus, we may be bound to relate to conclusions that we have drawn for Modern German above. The following discussion crucially hinges on the diagnostic criteria at the hands of which the early formal occurrence of the determiner or its non-occurrence can be identified as a definite article or not. The main division needs to be made between the referential use of what is to become a definite article (Def) as opposed to ArtPro, on the one side, and its anaphoric usage (i.e. thematic back-/forward linking across the single sentence; often simply referred to as ‘identifiability’, which no doubt is ambiguous), on the other side. Clearly, if the Art can be omitted as a morphemic form we expect thematic linking to be executed by topical position. This bore out for Gothic in a convincing number of argument positions (Leiss 2000). The function of referential deixis (selected singularity) appears to be of secondary importance since neither generics nor uniques appear with the topical Det form, nor do proper names as in the spoken varieties of German (*(der) Hans “the John”, *(die) Mutter “the mother” strongly stigmatised, however, in Modern Standard, written German).2 The clearest prediction as to which criteria should be expected to play a role in the emergence of the definite article have been suggested by Greenberg (1978: 61–64). Greenberg’s grammaticalization path comprises 4 stages: In Stage 0 a demonstrative identifies an element previously mentioned in discourse; in Stage I, this demonstrative (+NP) has already emerged as a compulsory definite article with a general identifying function and determined by syntactic construction. However, it is not yet Stage II, with an “automatic definite” largely grammaticalized and determined by syntactic construction (Greenberg 1978: 64). Finally, at Stage III, cooccurs with inherently determined proper names and generics, including negatives (Je n’ai pas d’eau), adverbials (on + *(the) foot − à *(le) pied − zu

. Omissibility of the Art form, and thus zero-representations of Det, appears to be the most convincing criteria against accepting a D-function in the language. Cf. modern Finnish se (Laury 1997: 251). However, the two distinct functions of Reference vs. Thematic Linking are often not kept distinguished with sufficient care (cf. Laury 1997). Furthermore, as we have argued with some emphasis in the present article, the fact that Det words do not surface is in itself not sufficient proof for the non-existence of a determiner/article function. See Leiss 1994, 2000 and Abraham 1997a,b arguing that paradigmatic case and perfectivity alternates serve such a purpose in Russian and Old High German. We have argued that failure to consider all pronominal paradigms in the attempt to gauge the status of Dem as Art fails to even satisfy the methodological requirements of bare Structuralism, let alone more advanced methodological demands.

S F O PRO



1st proofs

The discourse-functional crystallization 

*(dem) Fuss, at *(the) night − dans *(la) nuit − nachts), predicate nominals (Il est tailleur), and incorporated nouns (babysitter). Notice that, all along above in the present article, we have argued that such a grammaticalization path can hardly do justice to the empirical data and the typological differences since it does not include the entirety of pronominals and their distinct anaphorizing and referencing tasks. Thus, a fixed grammaticalizing path fails to consider the differences between languages as close as English and French: viz. Il est tailleur – He is *(a) tailor – Er ist (ein) Schneider ; Je n’ai pas d’eau – I want (*a/*of the) water – Ich möchte (*ein/*von) Wasser/vom Wasser. The most general distributional parameter, it is claimed, for a language to have a covert, yet functional DP is the presence of personal pronouns binding anaphorically themata across the sentence border is evidence enough. Thus, there is a very general cross-linguistic correlation between the existence of PersPro and DP, on the one hand, and the non-existence of PersPro and NP, on the other hand. The latter correlation is realized, to all probabilities, in Japanese (see, for this general claim, Leiss, this volume, and, for Japanese, Osawa, this volume). In other words, Greenberg’s (1978) grammaticalization path (4 stages with rising replenishment of distributional D-satisfaction) is to be completed by at least this general PersPro-correlation. The present methodological assumption was quite different in that it developed an account of the partial commonalities and distributional and discoursefunctional differences between all pronominals (ArtPro, DemPro, PersPro, PossPro, Art(+NP). The point of departure was that morphology plays a particular role for the derivation of syntactic forms. Within this perspective this means that ArtPro/DemPro derives from the composition of an indeclinable deictic stem dand an instance of the PersPro paradigm. The inflection ending is accounted for by a functional category F where case and number agreement are checked. The definite article, der/die/das “the.//” can be seen as grammaticalized, defocused, and, subsequently, recategorized ArtPro. In other words, ArtPro, SpecDP, was the point of departure, and its diachronic endpoint was Art, D0. Recategorization implied restructuring and, probably, feature reduction. In order to sketch the diachronic processes in more detail in 6.2, we first have to discuss the descriptive prerequisites in 6.1. .

The syntax of ArtPro/DemPro and PersPro

In what follows we try to provide a positive answer to the question whether semantic properties of noun phrases can be related to their syntactic properties. This assumption requires that partial interpretation of noun phrases is allowed and, in fact, I argue in favour of it. See Abraham (this volume, 5.1.) for prior assumptions. The syntactic structure in (2),

S F O PRO



1st proofs

 Werner Abraham

(2)

[DP [D’ /die(se)] [NP [AP/NP Schöne/, der keine Reckübung . . .] [N 0]]]

permits the direct reconstruction of the direct grammaticalizing step from ArtPro, SpecDP, to article, D0, in German (not all Germanic, since English does not possess ArtPro; nor do the Romance languages: [D’ the/le [NP x]], but not *[SpecDP the/le [NP 0]]). Consider Latin [SpecDP ille/illa [NP 0]] > French [D’ le/la [NP x]], and, likewise PersPro [SpecDP il/elle [NP 0]]).3 I assume that DemPro had minimally the features [+existential], [+definite] or [+deictic], while ArtPro/DemPro at least encoded the interpretable feature [+deictic] (Keenan 1987, 1995 as well as Kaplan 1989).4 Following Silva-Villar & Giutiérrez-Rexach (2001: 330) I assume that either only the deixis feature, [+deictic], or the definiteness feature, [+def], alone or [+deictic] + [+def]5 together were subject to processes of syntactic core grammar. . For our purposes it is sufficient to point only briefly at the Latin-Romance development. Romance took its point of departure from the Latin tripartite system of paradigms hic-iste-ille. See Silva-Villar 1996 and Vincent 1997. . According to Keenan (1995), an article word (‘Dem’) satisfies the condition of existentiality if it is definable in terms of an intersective generalized quantifier, D, of type 〈e,t〉,〈e,t〉,〈t〉 and, furthermore, if for all words E and A, B, A’, B’ and E holds: if A_B=A’_B’, then D(A)(B)=D(A’)(B’). This is to say: Definiteness normally implies existentiality. Notice that it would be wrong to judge from the non-occurrence of a definite article morpheme that the language in question lacks the feature of definiteness, which may be there all along expressable through other links such as by paradigmatic change between accusative and genitive (as, roughly, in Russian; see Leiss 1994; Abraham 1997a) or by linear position alone (as in SOV languages such as Japanese (topicalized -wa- morpheme expressing thematicity and, by implication, d-coreferentiality in Keenan’s sense above) and German (Abraham 1993, 1995, 1997a). Possibly, the best correlation to look for in a covert-article vs. a non-article language is to see whether it has PersPro as a paradigmatic category. PersPro always refers to a thematic/definite antecedent lexical. Japanese does not have PersPro – it has to repeat proper names and common nouns to express coreferentiality – and, consequently, would have the status of a no-article language (despite its thematicity link through –wa) and, consequently, might be a candidate for a non-DP-language – albeit still an uncertain one. We still do not have full metalinguistic awareness of the whole range of patterns of definiteness/indefiniteness effects. After all, Russian is a non-article language also, but, clearly, has (in)definiteness implications through aspect and case alternatives. . Both features are activated if, as is the case in Spanish or Modern Greek, both ArtPro/DemPro and definite article occur paradigmatically. See (i)–(ii) (from Silva-Villar & Giutiérrez-Rexach (2001: 331).

i. ii.

Greek:

afto to vivlio to oreo this the book the good Spanish: el libro bueno este the book good this

S F O PRO

ArtPro/DemPro and article occur in these languages as well as others in serializations even richer than in (i)–(ii).



1st proofs

The discourse-functional crystallization 

The asymmetric distribution dieser er, but *er dieser leads one to assume the following syntactic structure. See (3)–(4). DP

(3) ArtPro/DemPro

N'

d(ies)er

N'/AP gute

N Tee/0

(4)

DP D N' [+def] PersPro

N 0

The German DP takes the Spec-position if we follow the main thrust in the literature (Müller 2001: 3 Fn. 5); N is phonologically empty. PersPro is usually taken to be raised in order to check the definiteness feature. The pronominal article, Art, and the independent DemPro occupies this structural position already and consequently need not be raised. Obligatory case assignment at the left periphery of DP (Müller 2001) undermines the assumption that ArtPro and DemPro can be considered as non-governing DPs with empty NPs. This accounts for the fact that ArtPro/DemPro can quantify semantically as well as that adjectival atrribution and relative modification under the pronoun are possible. See (5), which connects with (3)–(4) above. (5)

. . .] CP. Das/Dieses (Gute)/, was gefallen hat, . . . .... The/This (good)/, that has pleased, . . .

However, there are other distributions that should be accounted for in terms different from (3)–(4) above. See (6)–(8) where necessary or optional focus accent is signalled by caps. (6)

i. ii.

[SpecDP // [NP 0]] [SpecDP der/die/das [NP 0]] read: ArtPro may be focus accentuated; there is also an unaccentuated ArtPro triggering identical rhema coreference the difference being that no referential subset selection is triggered

S F O PRO



1st proofs

 Werner Abraham

(7)

i. ii.

(8)

i. ii.

[D’ der/die/das [NP x]] as well as [?SpecDP/?D’ // [NP x]] read: Art may be focus accentuated

. . . discourse unmarked . . . discourse marked

[SpecDP //() [NP 0/x]] . . . discourse marked as well as [SpecDP dieser/diese/dies(es) [NP 0/x]] . . . discourse unmarked read: DemPro may, but need not, be accentuated, in which case it has a discourse-functional status different from the unaccentuated occurrence.

Diachronically, all we can conclude from the status of Modern German is that seeming grammatical (6ii) was in fact (6i). Accent was never encoded, after all. Thus, Old High German had only (7i), not (7i/ii). The grammaticalizing path from OHG to MHG to Modern Standard German, MStG, must have gone from (6i) via (7ii) to (7i). See (9) for this. (9)

stage I (reaching onto OHG) stage II (OHG – MHG) [SpecDP // [NP 0]] > [?SpecDP/?D’ // [NP x]] > stage III (MHG – MstG) [D’ der/die/das [NP x]]

The synchronic question remains how the structural status of [?SpecDP/?D’ //  [NP x]] vs. [D’ der/die/das [NP x]] can be decided. Pursuing the methodological virtue of a syntax that mirrors semantic distinctions throughout, I assume an expansion of DP by two discourse-functional categories on top of DP.6 (10)

[FocDP ai [ThemaDP b [FocDP ti [SpecDP c [D’ d [NP e]]]]]]7

The reason for two FocPs in (10) is that one would like to distinguish between two types of focus accents/rhemata: One clausal default rhema (in terms of Cinque’s . See similar proposals for DP by Bernstein 2001, Brugè 2002, and Aboh 2005, albeit for nonGermanic languages. . These focus projections generalized for DPs are supported by necessary extensions for PPs as (i) and (ii) suggest. i. ii.

sofort zum Mitnehmen “to-the-immediately-take-along” instead of non-focus moved zum sofort Mitnehmen “to-the-immediately-take-along” genau von hinter dem Jungen “precisely-of-behind-the-boy” instead of von genau hinter dem Jungen “of-precisely-behind-the-boy”

However, this does not appear to work for the local Ps auf “on”, unter “under”, neben “next to” etc., where the silent purely localizing function of / or the purely relational function of von conflates with a space compartment. I owe this observation to Daniel Hole (Munich).

S F O PRO



1st proofs

The discourse-functional crystallization 

Null Focus Hypothesis) and another one for derived, base-dislocated positions (and refocusing).8 See Abraham (1997a) as well as Abraham & Molnárfi (2001). (10) then takes care of the partly non-distinct structures in (8) in an empirically satisfactory way. (11)

a.

b.

c.

d.

e.

instead of (6i): [FocDP //i [ThemaDP 0 [FocDP ti [SpecDP t [D’ 0 [NP 0]]]]]] read: ArtPro may be focus accentuated for contrastive opposition (singling out a subset); coreference is with a rhema instead of (6ii): [FocDP 0 [ThemaDP 0 [FocDP der/die/dasi [SpecDP ti [D’ 0 [NP 0]]]]]] read: ArtPro on the left edge of the German middle field may be unaccentuated triggering identical rhema coreference the difference with (11a) being that no referential subset selection is triggered instead of (8i): [FocDP //() [NP 0/x]]; i.e. ArtPro outside/to the left of its default position instead of (7ii): [FocDP //i [ThemaDP 0 [FocDP ti [SpecDP 0 [D’ 0 [NP 0]]]]]]; i.e. contrastive discourse function DefArticle: [CP C [SpecDP [Thema der/die/das [NP N] [object]] [adv*] [VP [Rhema DP*] V]]]; definite DPs are in the thematic period (left edge) of the middle field just left-short of pronominal clitics and personal anaphors, which attach directly to the right of C0 (Abraham 1997a, 2006, this volume).

ArtPro as well as DemPro, irrespective of their focus accents, corefer to rhemata, albeit with different references: Focus accent singles out referentially next to rhema anaphorization, while lack of accent only anaphorizes a rhema. This was taken care of in German – but not English and the Romance languages – by a discoursefunctionally expanded DP as in (12a–d). In this structure chain, the definite article occupies a clausal discourse-thema category in the left middle field, while ArtPro as well as DemPro raise a node higher into a discourse rhema category. Notice that this is fully in line with the Centering Theoretical rules established for ArtPro/ DemPro vs. PersPro in MStG, both accentuated and unaccentuated. Accentuated always implies raising into a FocusP above DP, while unaccentuated Do raises into ThemaP above DP – for reasons that have to do with the clear accent and, consequently, thema-rhema distributions.

. This is reminiscent of Grohmann’s (2003) CP-expansion for German dislocations. However, it is not clear what hanging topics have to do with refocused topics. No doubt, this merits extra future attention.

S F O PRO



1st proofs

 Werner Abraham

.

The diachronic periods

.. Earliest period Definite DP-determiner functions cannot be distinguished formally from demonstratives. This is the diagnosis: The constituent [definite determiner + NPi] provides no anaphoric coreference with a lin-early proximate NPi – i.e. any occurrence of such NPi as well as Det + NPi in immediately adjacent clauses has the reading, ‘Det-Form = Dem-Function’. Compare Philippi (1997: (56a–d), p. 84f.): (12) a.

G-Wulfila (NT, Lk., 2.25) (Þaruh was manna in Jairusalem, Þizei namo Swmaion,) jah sa manna was garaihts [. . .] G: (. . .) kai ho anthropos houtos dikaios [. . .] L: (. . .) et homo iste iustus erat [. . .] (There was a mani in Jerusalem.) ?The/This/i mani was righteous b. O H G-Tatian 1.1 c (in anabeginne was uuort) inti thaz uuort uuas mit gote L: (In principio erat verbum) et verbum erat apud deum In the beginning was (the) word, and this// word was with God c. O S-Heliand.3256–7 (tho thar te theme lerende quam en jung man) [ . . .]tho sprak eft the jungo man When he preached a young man stepped up to him and this/ young man said

Similarly for Old English. Notice that Gothic and Old Saxon are different from OHG and partly from Latin. What we do not know is whether we may assume that the complete DemPro-form standing for DP – equivalent to Modern German ArtPro/DemPro or not? – can be emphasized as in Modern German. Latin has iste, clearly a deictic DemPro, while OHG as well as Old Saxon have the janus-faced Dem/Art. In Modern German one can use the article form under focus to achieve Dem-status: Cf.  Mann “ man” in (12c). Since we have no clue as to focus marking in this early stage and since we do not find any D-form in isolation, this is what we may conclude in line with Centering Theory: (13) a. in those syntactic contexts that require ArtPro/DemPro in MStG, the early predecessors of German were to posit Cfi+1 – as in Modern Japanese. See note 11 above. The linear sequence is thus: Cfi ## Cfi+1. b. notice that (1269a–c) had antecedents which were identified above as rhemata. Consequently, (1370a) is valid only for rhematic antecedents. In other words, thematic anaphoricity (where Cb=PersPro) is excluded.

.

Old High German

S F O PRO

In Old High German, Old Saxon (and Old English), not, however, in Gothic, one can find formally definite determiners also in indirectly anaphoric contexts.



1st proofs

The discourse-functional crystallization 

The formally definite NP refers to an individual, only implicable from a prior pronominal occurrence. This disallows a reading as definite article. Rather, it has the reference of a demonstrative. Compare (14a) (from Philippi (57a), (58b), p. 85). (14)

a.

b.

O H G (Tatian.147.6) (Samaso man, Þer elilento farenti forliez sin hus [. . .] inti themo duiruuarte gibot Þaz her uuahhteti . . .like a mani who left hisi house, and *the/this. mani/. was ordered to wake O H G (Tatian.230.6) inti mittiu her thaz qad, arougta [0] in henti, fuozi inti sita and as soon as he said that (he) showed them (the/his) palms, (the/his) feet and (the/his) flank.

Formally demonstrative determiners usually do not cooccur with abstract NPs. The fact that they cannot be found in Old High German texts attests that formal determiners have to be read as demonstratives and not as definite articles. See (15) (from Philippi (59) S. 86). (15)

a.

b.

G-Wulfila swaei nu dauÞus in uns waurkeiÞ, iÞ libains in izwis So is now *this/(the) death in you, (but) *that///(the) life in him. O E-Beowulf.466 . . ., gif mec deađnimeđdh. [0] Byređ blodig wæl. . . . if me the death takes. (He) carries away (a) bloody corpse.

From this follows for the early period of German in Centering terms that text coherence with ArtPro/DemPro in the sense of (13) above (MStG) cannot be excluded. But we do not have any clear evidence. This leads us to the conclusion that the ArtPro/DemPro-rule did not yet have validity. Possibly, ArtPro/DemPro could not complete is rhematic-anaphoric function until: i.

ii.

either the text-deictic distance to the article function was established; in fact, this occurred only later and was co-dependent upon the later emergence of the indefinite articles; and/or the discourse-structural separation of thematic material (left of/before VP) and rhematic material (inside of VP) had been completed.

There is no conclusive evidence as to when one can assume (ii) to have been established. There is reason to assume that it did not take place before the radical restructuring of Late Middle High German (around 1450) – i.e. co-temporal with, or following, the loss of lexical aspect (paradigmatically fixed Aktionsart) and the emergence and expansion of a tense systematics and, as a consequence, the sequence of tenses (Consecutio temporum). At the same time, the indefinite

S F O PRO



1st proofs

 Werner Abraham

article emerged as well as the paradigmatically stabilization of the strong and weak adjectival inflection. Determiners do not occur with NPs with only a generic reading. See (16) (expanded from Philippi (60), p, 86). (16)

a.

b.

c.

G-Wulfila sabbato in mans warÞ gaskapans, ni manno in sabbato (the) Sunday is for (the) man made, not vice versa. O H G-Tatian.100.2 ist arloubit manne zi uorlassana sina quênun is (it) permitted (the.) man to leave his wife. O S-Heliand.1300-1 [sagde im thô te sôđan,] quađ that thie sâlige uuârin man an thesero middilgardun (he) told him then that blessed were (the) men in this world.

The function of article in the sense of Modern Standard German had not been reached. The generic use of ModStGerman der Löwe ist der König der Tiere “the lion is the king of the animals” is not yet possible. Instead of the definite article, Gothic regularly inserts PersPro – see Wulfila’s in mans “dem Menschen/the. man” (reminiscent of (South) Bavarian eam Mō “he. man” accountable from the definite article and PersPro sharing features of discourse thematicity (‘Thema’ as opposed to ‘Rhema’). .. Definite article It is common assumption (Leiss 2000) that the definite article (not, however, DemPro) occurs in thematic function in Old High German (in Tatian in Old High German, not following the Latin original). See (17a) (correction of Philippi (1997: 87), the author’s (65)). (17b) shows the Old High German Tatian section with thereunder the Latin original.9 (17)

a.

G-Mt.XXVI.69-72 [iþ Paitrus uta sat ana rohsnai, jah duatiddja imma aina þiwi qiþandei: jah þu wast miþ Iesua þamma Galeilaiau. iþ is laugnida faura þaim allaim qiþands: ni wait hwa qiþis. usgaggandan þan ina in daur, gasahw ina anþara jah qaþ du þaim jainar: jah sa was miþ Iesua þamma Nazoraiau.] jah sa was miÞ Iesua Nazoraiou. [jah aftra laugnida miþ aiþa swarands þatei ni kann þana mannan.] And also (this///*the/*he) was together with Jesus of Nazareth

. Thanks to Irmengard Rauch/Berkeley for help on literature.

S F O PRO



1st proofs

The discourse-functional crystallization 

b.

c.

O H G-Tatian CXI.2 [Thie tho her sie gisah, quad: gét inti arouget íúuih then bisgofon. Inti uuard thó, mittiu sie fúoron, thaz sie gisubrit uurdun. Ein fon then, so her gisah thaz her gireinit uúas, uuarb uuidar mit michilera stémmu michilsonti grot. Inti fiel in sin annuzi furi sine fúozi thánca tuonti,] inti ther was Samaritanus. [Tho antuurtita ther heilant inti quad: . . .] [Quos ut vidit, dixit: ite, ostendite vos sacerdotibus. Et factum est, du mirent, mundati sunt. Unus autem ex illis, ut vidit quia mundatus est, regressu est cum magna voce magnificans deum. Et decidit in faciem ante pedes eius gratias agens,] et hic erat Samaritanus. [Respondens autem Ihesus dixit: . . .] “. . . and (also) ()(=.hic!) “the/this (one)” was (a) Samaritan/from Samaria. . .” O S-Heliand 264-6 . . . Thu scalt uses drohtines uuesan môdar mid mannun endi scalt thana magu fôdean, thes hôhon he_ancuninges suno. The scal Heliand te namon êgan mid eldiun . . . You will our. master. mother be under mankind and will eventually (a) son bear, the. high heavenly king’s son. (T/T will Saviour as name bear among (the-) men.

Fact is, first, that formal [Art+NP] occurs in OHG. While occasional such clusters have been noted for Gothic, its frequency was so low that Leiss (2000) assumed that it signalled a definite rhema. In the majority of occurrences, the position at the left periphery was sufficient for signalling thematicity. The rareness of the definite article has led the researchers to assume that formal Art(+NP) functioned like DemPro – which takes up a rhema in the prior sentence. Leiss (2000) took specific issue with this canonical position arguing, on the basis of visible and invisible, but implied categories, that concepts of hypo-determining article languages (such as Gothic, Old Icelandic, and still Old High German which signal definiteness overtly in indefinite regions such as the rhema) vs. hyper-determining article languages (signalling referentiality and thematicity overtly even when definiteness is already implied by the thematic position) did better justice to the entirety of empirical data. Second, synchronically, Art+NP take up reference from material in thematic or rhematic position in the earlier sentence. See the generalizations at different stages of the discussion of the corpus data in (12) and (14)–(17). This appears to hold for as early as Old High German where the article had an anaphoric function. This is often, but not always in line with the use of ArtPro/DemPro in Modern German – not, however, with the definite, and therefore thematic, use of the article. Indication for such Article-ArtPro/DemPro uses very early on is provided by selected translations collected by Philippi (1997: 87): Latin ille, hic, is, Ancient Greek. houtos, hoi

S F O PRO



1st proofs

 Werner Abraham

have all been taken up by different determiners (such as article, demonstratives, pronouns) in Old High German. Each of these categories must have been ambiguous in this old stage of German dependent on the specific text, writer, and accent distributions, which remained unencoded. Notice that, in order to arrive at this non-canonical insight, we have made use of the discussion of Modern German in Sections 1–5 above. In order for the modern definite article usage to emerge, earlier focus (in the sense of ArtPro) had to yield to non-focus and, consequently, thematic reference. One has to keep in mind that the generalizations above are still in need of further analytic explanation. It has not been spelled out with sufficient clarity how the features of thematicity (anaphorics) and definiteness (reference) interact and in what order, and on what encoding motivation, they become overt in the historical grammaticalization process. What appears to be clear is that thematicity has been signalled by linear order (sentential left edge) early on. Furthermore, definiteness seems to be secondary as it is implied by thematicity. And, third, to separate referential definiteness and the discourse function of thematicity as overt codes is but a later coding need. No doubt, to say that the definite article has but an anaphoric function misses not only the diachronic point completely. All of the above confirms the non-canonical insight that the distribution of definite determiners in the early linguistic periods of Germanic converges with the distribution of demonstratives, DemPro. Furthermore, the anaphoric take-up of rhematic antecedents by means of ArtPro/DemPro cannot be registered without ambiguity. Recall that paradigmatically manifest aspect (ge-verbs versus simple homonymic stems: wit gescrîban “complete writing” vs. scrîban “write”) was in demise by the end of MHG (around 1400). Compare the originally manifest relations between article reference and (im)perfectivity in Gothic, Old High German, and classical MHG according to (18). (18)

[+def]/DO/*IO/*SU ÷ [+perf] as opposed to [−def]/DO/*IO/*SU ÷ [−perf)

(18) reads as ‘Any definite DP in DO clause function correlates strongly with perfectivity of the governing predicate’, while ‘Any indefinite DP in DO clause function correlates strongly with imperfectivity of the governing predicate’. For a semantic-syntactic and typological classification see Leiss (1994) and Abraham (1997b). In sum, Fig. 1 appears to be a realistic path of grammaticalization of the definite article in German.

S F O PRO



1st proofs

The discourse-functional crystallization 

[MF = Middle Field = between Co and SpecvP; MStG = Modern Standard German]. G zero > −focus in S-top position

Stage I ArtPro > +focus in S-rear position

Stage II ArtPro > +focus right in MF

Stage III Art(+NP) > −focus in left MF

MSG Art(+NP) −focus left MF

thema −reference

rhema −reference

rhema −reference

thema −reference

thema +reference

Figure 1. Diachronic stages of the emergent definite article according to position, Art/Dem representation, and focus realization – see also (10) above.

.

Conclusion

Synchronically, article pronoun (ArtPron) and demonstrative pronoun (DemPro) in Modern German and Dutch have been shown to be distinguished typologically from PersPro and possessive−pronoun (PossPro). We have seen that the criterion of part of speech (grammatical category; Cf = subject) is attributed the main criterial weight in centering theoretical work on English. The present discussion demonstrated that this could not confirmed for German or Dutch. No doubt, the typological status of the language under discussion plays a determining role: In Japanese, grammatical, or zero, topic ranges higher as Cf-representant than the subject (Walker, Iida & Cote 1994). This was shown to be valid for German and Dutch also. In particular, for NPbinding beyond the clausal border lines German displays clear discourse-prominent characteristics. No doubt, this is linked to the fact that the subject need not be thema and that German provides ample means to make topic/thema just any other grammatical category by means of displacement from the base order, or scrambling, and subsequent refocusing. Modern English does not have this means. As for the results of the present discussion, see the summary in the following table. Compare (7)–(9) as well as (12a–d) above. Thema binding ArtPro AP DemPro DP PersPro PP Art(+DP) A(+DP)

+ − + − + − + −

Rhema binding − + + + − + − +

Contrastive

Singling out reference

− + − + − + − +

− + − + − − − +

S F O PRO

Figure 2. Discourse binding: Anaphoricity and singling out reference.



1st proofs

 Werner Abraham

See also (20) above. The fact that contrastive accent/refocusing occurs with linear relocation from the base order (see 12a–d) and change of D-binding and subset singling out is in line with the discussion of the specific type of discourse prominence of German (Abraham 1997a, 2006). Focusing on the historical emergence of the definite article, German proved to be a typical example of covert, indirect referential disambiguation: Morphological case and verbal aspect (perfective vs. imperfective) interacted to denote definiteness, and thereby thematicity/givenness, as opposed to indefiniteness/ rhematicity. Mereologically, this is plausible: Verbal perfectivity and definiteness are characterized by the event features [−additive] and [+terminative], whereas verbal imperfectivity and indefiniteness bear [+additive] and [−terminative]. Accusative has a definiteness triggering role under just any verbal aspect, while the genitive limits its indefiniteness trigger only under imperfective aspect. Despite the lack of morphological aspectual distinctions, and although the speaker of late MHG was still sensitive to aspectual distinctions, the genitive was taken to signal verbal imperfectivity in the novel absence of the previous aspectual marking. It may be assumed that, as soon as the sensitivity toward aspectual distinctions declined (since unsupported by aspectual morphology), the causal link between case (genitive) and imperfectivity faded, with the eventual result that either the accusative, as the structurally more versatile case, could take over, or prepositional cases took over due to their semantic motivation (Leiss 1994). What we have not shown in any detail is why perfectivity has a unique syntactic mapping in terms of predicatives (secondary predication; small clauses). It is beyond the scope of this paper to explain this (see Abraham 1990, 1991, 1994 for details). Once this syntactic mapping of an otherwise and traditionally semantically perceived category is accepted, then a host of formerly semantic features can be distinguished uniquely in syntactic terms. This is an important result in that it gives credit to a principle which appears to have wide linguistic import: ‘one meaning, one form’. The present paper is an instantiation of the varying linguistic scenarios that this principle entails in the diachronic development of German.

Primary sources Die gotische Bibel. Band 1. Der gotische Text und seine griechische Vorlage ed. by Wilhelm Streitberg. Heidelberg: C. Winter. 72000. Heliand and Genesis ed. by Otto Behaghel. Halle (Saale): M. Niemeyer. 61948. Tatian. Lateinisch and Althochdeutsch mit ausführlichem Glossar ed. by Eduard Sievers. Paderborn: Ferd. Schöning. 21960.

S F O PRO



1st proofs

The discourse-functional crystallization 

References Aboh, Enoch Oladé. 2005. Topic and Focus within D. LIN 21: 1–12. Abraham, Werner. 1990. A note on the aspect-syntax interface. In Progress in Linguistics, M. Mascarò & Marina Nespor (eds), 1–12. Dordrecht: Foris. Abraham, Werner. 1991. Aktionsartsemantik and Auxiliarisierung im Germanen. In Neue Fragen der Linguistik. Vol. 1 [Linguistische Arbeiten 270], E. Feldbusch, R. Pogarell & C. Weiβ (eds), 125–133. Tübingen: Niemeyer. Abraham, Werner. 1993. Überlegungen zur satzgrammatischen Begründung der Diskursfunktionen Thema and Rhema. Folia Linguistica XXVI(1–2): 197–231. Abraham, Werner. 1994. Ergativa sind Terminativa. Zeitschrift für Sprachwissenschaft 12(2): 157–184. Abraham, Werner. 1997a. The base structure of the German clause under discourse-functional weight: Contentful functional categories vs. derivative ones. In German: Syntactic problems – problematic syntax, Werner Abraham & Elly van Gelderen (eds), 11–43. Tübingen: Niemeyer. Abraham, Werner. 1997b. The interdependence of case, aspect, and referentiality in the history of German: The case of the genitive. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 29–61. Cambridge: Cambridge University Press. Abraham, Werner. 2006. Topic, focus, and default vs. contrastive accent: Drawing typological differences with respect to discourse prominence. In On Information Structure, Meaning and Form. The web of information structure across languages, Kerstin Schwabe & Susanne Winkler (eds), [Linguistik Aktuell/Linguistics Today]. Amsterdam: John Benjamins. Abraham, Werner. This volume. Discourse binding: DP and pronouns in German, Dutch, and English. In Nominal determination. Typology, context constraints, and historical emergence, Elisabeth Stark, Elisabeth Leiss & Werner Abraham (eds). Amsterdam: John Benjamins. Abraham, Werner & Molnárfi, Lázsló. 2001. German clause structure under discourse-functional weight: Focus and antifocus. In On Formal German(ic) Typology, [Linguistik Aktuell/ Linguistics Today 45], Werner Abraham & C. Jan-Wouter Zwart (eds), 1–43. Amsterdam: John Benjamins. Bernstein, Judy B. 2001. Focusing the ‘right’ way in Romance determiner phrases. Probus 13: 1–29. Brugè, Laura. 2002. The positions of demonstratives in the extended nominal projection. In Functional structure in DP and IP. Vol. 1, Guglielmo. Cinque (ed.), 15–53. Oxford: Oxford University Press. Greenberg, Joseph H. 1978. How does a language acquire gender markers? In Universals of Human Language, Vol. 3. Word structure, Joseph H. Greenberg (ed.). Stanford: Stanford University Press. Grohmann, Kleanthes. 2003. Prolific Domains. On the anti-locality of movement dependencie. [Linguistik Aktuell/Linguistics Today, 66]. Amsterdam: John Benjamins. Kaplan, David. 1989. Demonstratives. In Themes from Kaplan, J. Almog et al. (eds), 481–564. Oxford: Oxford University Press. Keenan, Edward. 1987. A semantic definition of indefinite NPs. In The Representation of (In)definiteness, Eric J. Reuland & Alice ter Meulen (eds), 286–317. Cambridge MA: MIT Press. Keenan, Edward. 1995. The semantics of determiners. In Handbook of Contemporary Semantic Theory, Sharon Lappin (ed.). 41–64. Oxford: Blackwell. Laury, Ritva. 1997. Demonstratives in Interaction. The emergence of a definite article in Finnish [Studies in Discourse and Grammar 7]. Amsterdam: John Benjamins.

S F O PRO



1st proofs

 Werner Abraham Leiss, Elisabeth. 1994. Die Entstehung des Artikels im Deutschen. Sprachwissenschaft 19: 307–319. Leiss, Elisabeth. 2000. Artikel und Aspekt. Die grammatischen Muster von Definiheit [Studia Linguistica Germanica 55]. Berlin: Mouton de Gruyter. Leiss, Elisabeth. This volume. Covert patterns of nominal determination in Old Icelandic, Gothic, and Old High German. In Elisabeth Stark, Elisabeth Leiss & Werner Abraham (eds), Nominal Determination. Typology, context constraints, and historical emergence. Amsterdam: John Benjamins. Müller, Gereon. 2001. Verb-second. A new approach. Paper IdS-Mannheim. Osawa, Fuyo. This volume. The emergence of DP from a perspective of ontogeny and phylogeny: Correlation between DP, TP and aspect in Old English, Japanese and first language acquisition. In Elisabeth Stark, Elisabeth Leiss & Werner Abraham (eds), Nominal Determination. Typology, context constraints, and historical emergence. Amsterdam: John Benjamins Philippi, Julia. 1997. The rise of the articles in the Germanic languages. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 62–93. Cambridge: Cambridge University Press. Silva-Villar, Luis 1996. Enclisis in Northwestern Iberian languages. A diachronic theory. PhD Dissertation, University of California, Los Angeles. Silva-Villar, Luis & Gutiérrez-Rexach, Javier. 2001. Demonstratives in a feature-based theory of syntax. In The Minimalist Parameter [Current Issues in Linguistic Theory 192], Galina Alexandrova & Olga Arnaudova (eds), 325–344. Amsterdam: John Benjamins. Vincent, Nigel. 1997. The emergence of the D-system in Romance. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 149–169. Cambridge: Cambridge University Press. Walker, Marilyn, Iida, Masayo & Cote, Sharon 1994. Japanese discourse and the process of centering. Computational Linguistics 20(2): 193–223.

S F O PRO



1st proofs

Determinerless noun phrases in Old Romance passives1 Anna Bartra Kaufmann Universitat Autònoma de Barcelona

In this paper we take the existence of Bare Noun Phrases acting as subjects or displaced complements as evidence for the licensing properties of Functional Categories other that Determiner Phrase in earlier stages of Romance Languages such as Spanish and Catalan. The fact that in Spanish and Catalan subjects in passive sentences can be licensed inside the Verb Phrase also facilitates the use of BNPs in this context.

.

Introduction

The main goal of this paper is to argue that a comprehensive and integrated theory of grammar permits to relate some lexical, morphological, and syntactic properties that are apparently unrelated. More specifically, we want to show that in the framework of a theory that envisages the syntactic projection of the relevant morphological properties of lexical elements, a non-stipulative account of some syntactic facts can be elaborated. We will focus on the properties and distribution of some bare noun phrases (BNPs),2 specifically those appearing as subjects of . I wish to thank the audiences at the III Simposi Internacional Vers una Sintaxi Històrica del Català (La Nucia, Alacant, Spain 17–19 November, 2005); the VII Congreso Internacional de Historia de la Lengua Española (Mèrida, Yucatán, México, 4–8 September, 2006, and the Workshop on New perspectives on Morphological Change (Freie Universität Berlin, Berlin, Deutschland, 27–28 Oktober, 2006), specially to Werner Abraham, Montserrat Batllori, Jan Terje Faarlund, Georg Kaiser, Natasha Pomino, Gemma Rigau, Elisabeth Stark for comments and discussions on previous partial versions of this paper. I owe special thanks to Werner Abraham, Maria Teresa Espinal, and Elisabeth Stark for lots of comments and encouragement. All remaining errors are mine. This work has been supported by grants BFF2003-08364-C02-01 from the Ministerio de Educación y Ciencia and 2005SGR-00753 from the Generalitat de Catalunya. . There is a huge amount of literature on BNPs. Since we can not refer to it here, see the reference volumes Coene & D’Hulst (2003); Lingua, 115.6 (June 2005) monographic issue on “The structure of (in) definiteness: issues in the form and interpretation of noun phrases”; Heycock & Zamparelli (2003); Vogeleer (2006); Vogeleer & Tasmovsky (2006).

S F O PRO



1st proofs

 Anna Bartra Kaufmann

passive sentences in Old Romance, basically in Old Catalan and Old Spanish, in order to make some hypothesis about the morphological and functional structure of Old Romance noun phrases. As a corollary, we hope to sketch a correlation between lexical, morphological and syntactic-distributional properties of both noun phrases and (passive) verb phrases. We will take the structures of the type in (1) as a starting point and indicators of the functional properties of noun phrases. (1)

a.

b.

c.

d.

Old Catalan (Epistolari de València Medieval: letter, 66) e arrest és estat fet per l’arcediano and arrest be-3-- be- make-  by the=arcediano ‘And arrest has been made by the Arcediano’ Old Catalan (Epistolari de València Medieval: letter, 17) manament és estat feit a vós de part warrant be-3-- be- make- to=you by part d’En Bertomeu of .=Bertomeu3 ‘You have been ordered by Bertomeu’ Old Spanish (Ordenamiento de las cortes celebradas en Alcalá de Henares, 1348, CORDE) non fue fecha stipulaçion que quier dezir not be-3- make- stipulation that want-3-- say- prometimiento promise ‘It wasn’t made stipulation, it means promise. . .’ Old Spanish (El Nuevo Testamento según el manuscrito escurialense I-j.6, 1260, CORDE) Question fue fecha de los disciplos de Iohan con los iudios discussion be-3- make- of=the =disciples of Johan with the Jews del baptismo of=the baptism ‘Johan’s disciples argued with the Jewish about baptism’

In Modern Spanish and Modern Catalan, as well as in other Modern Romance languages, the corresponding structures are ungrammatical: (2)

a.

Modern Catalan: *detenció ha estat feta arrest has be- make--.-

. We gloss as . the special determiner used in Old and Modern Catalan to introduce human proper names and which does not share all properties with usual definite determiners. These determiners will not be part of our analysis.

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

b.

Modern Spanish: *pregunta ha sido hecha question has be- make--.-’

The explanation of the ungrammaticality of sentences in (2) is usually the one in (3): (3)

Subject arguments need a Determiner to be licensed as referential entities.

It is reasonable to ask what the licensing conditions of BNP subjects in Old Spanish and Romance are, and to what extent the morphological properties of noun phrases and of passive sentences play a determining role. We take as a theoretical framework the Principles and Parameters development of grammatical theory, with the role of DP and other Functional Projections in the Extended Noun Phrase Projection as a central point in our discussion. As for the verbal projections in sentential structure, we will pursue the little v hypothesis originating from Hale & Keyser (1993) and pursued in Chomsky (2001) and subsequent work.

.

The role of functional categories in the syntax-morphology interface

In the eighties of the XXth. century, the extension of X’ theory to the (abstract) members of morphological classes such as INFL(ection) or T(ense) in the verbal domain, Comp(lementizer) in the event and discourse domain, D(eterminer) in the nominal domain, or Deg(ree) in the adjectival extended projection, proved to enhance the explanatory power of grammar. Syntactic properties such as the order of words and constituents, the content and distribution of empty categories or some semantic relations received a more elegant, general explanation; briefly, the grammar gained in explanatory power. Functional Categories (FCs) are syntactic categories headed by (abstract) morphemes or bundles of features. Functional categories are deprived of descriptive content and contain only information relevant for the grammar internal computational purposes. In a strong lexicalist framework, in which lexical items are selected form the lexicon provided with a complete inventory of properties or features, such as category features, φ-features (person, number, gender), Case, tense, aspect, etc. the functional categories need to check their formal features against the features of lexical items in order to make representations readable at the interfaces, in the output of the computational grammatical component. The properties of functional categories are determined by the lexical properties of the categories they select. Thus, for instance, for a derivation to converge, the category Tense has to be provided with the same features as the verb that will

S F O PRO



1st proofs

 Anna Bartra Kaufmann

finally figure in this position.4 Functional categories have contingent properties. Therefore, we can interpret the proposal of functional categories as an attempt to integrate the morphological properties into the syntactic, i.e. grammatical analysis. By contrast, lexical categories are (projections of) lexical roots with their inherent features, not contextually determined. As mentioned before, the universally admitted FCs are D(eterminer) in the nominal domain, T(ense), in the verbal predicational domain, and C(omplementizer) in the eventive and discourse domain (cf. Chomsky 1995, 2001). Several proposals have been made that develop the functional projections of each domain. The most outstanding ones have been the cartographic views of sentence structure, in which a complete mapping of – semantically relevant – features and properties gives rise to a universal hierarchy of functional projections (cf. Cinque 1999; Rizzi 2004). Rizzi (1997, and subsequent work) develops the C domain into several categories hosting finiteness, topic, focus and mood. In the verbal domain, several projections hosting adverbs and the aspectual and mood properties of predicates have been proposed, in the same line of reasoning, by Cinque (1999). In the nominal field, since Abney (1987), the syntactic and semantic differences between nominal projections with or without a (definite) Determiner are accounted for in a specific projection, D. (4)

[SD D [SN N]] (Abney 1987; Longobardi 1991, 1994; Szabolcsi 1994).

Later on, other functional projections in the nominal domain, such as Ge(Gender), Num(Number), and K(Case) among others, have been proposed (see Bernstein 1993a/b; Picallo 1991, 2002, 2005, 2006; Ritter 1993, 1995; Stark 2006a/b and references therein). As expected, several properties in the different sentence domains can be related. Our present purpose is to relate the properties in the DP with the morphological and functional properties of passive sentences, with the main purpose of using passive sentences as a cue in determining the functional structure of nominal projections in Old Catalan and Old Spanish, and, conversely, to cast some more light on the properties of romance passives. First of all, we will present a standard accepted account of passive structures, focusing on the way passive subjects are licensed. .

The properties of morphological passives

Since Chomsky (1965), several attempts have made to account for the restrictions on passive sentences from the very basic structure of the sentence. As seen in (5),

S F O PRO

. As for our present purposes, it doesn’t matter the technical mechanism by which the features of the lexical category match or check their analogous ones in the functional category.



1st proofs

Determinerless NPs in Old Romance passives 

the base structure rule which develops passive sentences is directly contingent on the possibility of the active sentence of hosting Manner Adverb. Clearly enough, the passive Participle is one of the rewriting possibilities of the constituent Manner: (5)

Manner ---> by passive 5 NP – Aux – V- . . . – NP- . . . – by passive- . . .

An elemental transformation substitutes the dummy element passive by the first NP and puts the second NP in the place of the first one. In that way, (5) can restrict the application of the passive rule to the verbs subcategorised for a Manner Adverb (“. . . it accounts automatically for the restriction of passivization to verbs that take Manner Adverbials freely.” (Chomsky 1965: 104)). Given transitivity as a conditio sine qua non6 for the passive construction to be available in Romance, the class of verbs subcategorised for manner adverbs seems to be equivalent to the class of verbs that take an external AGENT argument.7 In more recent frameworks, the agentive properties of verbs are reflected in the verbal domain projections. Hale & Keyser (1993) restated the traditional view that the external argument was not present in the verbal projection. Against formulations such as Koopman & Sportiche’s (1991) VP-Internal Subject Hypothesis, Hale & Keyser argued convincingly that the external argument is not part of the VP projection in the lexical argument structure: it merges into the Spec position of a functional abstract light verb category vP. Chomsky (2001) posited the functional category v in order to merge the external argument of unergative and transitive verbs. Given this structure, a unergative or transitive verb raises to v0, prior to value the interpretable features in T.8 The subject DP rises to Spec,T in overt syntax to erase the D0 feature in T. .

Chomsky (1965: 104).

. As a well known fact, German, Dutch and other languages admit passives with unergative verbs (Es wurde die ganze Nacht entlang getanzt 〈It was all night long danced〉 ‘{One/people/we} danced all night long’). . The equation does not work on both senses. It is true that only verbs that take manner adverbials can passivize, but it is not clear at all that manner adverbials are only possible with transitive agentive verbs, as the Spanish sentences in (i), with unaccusative predicates and –alleged-manner adverbials show:

i.

a. b.

La lluvia caía pausadamente the rain fall- slowly El espía murió violentamente the spy dye- violently

In the second place, there is not a one to one connection between agentivity and transitivity, as unergative verbs expressing activities (run, jump, speak) show. . We adopt a simple standard view. See Chomsky (2001).

S F O PRO



1st proofs

 Anna Bartra Kaufmann

See the representation in (6):9 (6)

vP DPExtArg

v'

… v0

VP

to Spec,T V

DPIntArg

to T0

In contrast to this, there is no external argument available in passive sentences, and the internal argument is the one which deletes the uninterpretable D-features on the T projection. Since the PP has nominal (i.e. gender and number) φ-features, it can be supposed no to be able to delete the uninterpretable features of T. We can hypothesize that PP is a neutralized A/V category that does not project a small v.10 Instead, an Agreement Projection with uninterpretable features of Gender and Number is needed. No external argument can be merged in this structure. We assume further that the auxiliary verb be – or its equivalents such as Spanish ser – has to be merged into the structure in order to satisfy the verbal features of T.11 As for the internal argument, it is licensed by Internal Merge into the specifier of the agreement projection in the extended projection of the PP (the A/V category).

. In (6) we adopt a simple structure, with enough descriptive power for our purposes. In a “Larsonian” structure, successive VP shells host the verbal arguments, the main idea being that in order to derive the passive sentence Mary was sent a letter from a ditransitive double object structure, the Indirect Object PP has to be analysed as the internal argument of the VP whereas the Direct Object occupies the [Spec,VP]. As our main purpose is to investigate how BNPs can be licensed as subjects in transitive sentences, as well as in other structures where it can be argued to have moved, the alternatives that have been suggested for analysing the argumental structure of the VP are not of primary interest to us. .

For the sake of simplicity, we do not take into consideration the difference between v and v*.

. There are different technical ways to develop the main idea that the predicate of passive sentences is defective in what concerns verbal properties. The verb be can be viewed as a morpheme that merges onto T to delete its verbal unvalued features or as a functional verb that rises into T for these purposes. These alternatives do no affect our arguments. Furthermore, the minimal FCs needed to license the properties of the Past Participle and its internal argument is an AgrP. Nevertheless, perfective AspP has been proposed to govern the participial small clause (see Abraham (1997), Cinque (1999)).

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

A brief sketch of the (possible) derivation of a passive sentence is given in (7). Notice that if no auxiliary verb ser would be introduced, we would obtain an absolute participial construction.12 vP

(7) Ø

v'

…

v0 to Spec,T

AgrP

be

Agr0

V/AP (PPP)

to T V/A0

DPInterArg

(b) to Spec TP (a) to Spec AgrP

Therefore, we obtain the structure in (8): (8)

TP DPIA

la ley

T' T fue

vP la ley

v fue

AgrP Agro

la ley dada

V/AP (PPP) dada

DPIA

la ley

S F O PRO

. We abstract away from the possible functional categories responsible of the licensing of the aspectual properties of participles. See, for instance Borer (1994) or, for a slightly different proposal relying on VoiceP, Kratzer (1996).



1st proofs

 Anna Bartra Kaufmann

. The functional categories of nominal projections in Old Romance: a sketchy overview Assuming the DP hypothesis,13 some new problems have to be envisaged with reference to Latin and Old Romance Languages. Following Longobardi (1994), the Determiner is the locus of the referential properties of NPs. This is the reason why, in his proposal, proper names have to rise to D in order to obtain referential properties (put otherwise, in order for D to erase its referential unvalued features). The question we can ask then is whether Latin noun phrases have to be analysed as DPs or NPs. This is not a clearly assessed point in the literature. In the line of reasoning of Longobardi, one could hypothesize that in a language without overt determiners, the noun rises (covertly) to D in order to check its referential properties. From the point of view of acquisition, however, it could also be said that Latin learning children did no have any cues to posit the existence in the language of a D projection.14 The same question can be posited for Old Romance Languages. Boucher (2005), in his work about the evolution of the determiners in Old French, asserts, following Dobrovie-Sorin (2001: 208), Delfitto & Schroten (1991) among others, that in languages without overt determiners, no functional projection D has to be established in order to derive the referential status of nominals. In these languages, the referential or definite value of the noun could be checked against other existing functional projections. Other authors like Progovac (1998) and Pereltsvaig (2006) argue that even in languages without overt determiners, some properties of argument noun phrases can be only accounted for by positing a D projection. For Pereltsvaig (2006) “complete” noun phrases or DPs contrast with small nominals. Assuming that there is no evidence or cue in those languages for acquiring D, it has to be universally present. Let us suppose with Boucher (2005) that on minimalist assumptions only those Functional Categories are activated for which there exists evidence in the Primary Linguistic data available for the child. Then, how come is the referential value of arguments licensed in the absence of D? Is it possible that other functional projections can do this task? We briefly examine the different possibilities offered to us in the literature. .

See the Introduction of this volume.

. If there is no D, the question arises of how reference and discourse linking properties of noun phrases can be interpreted in that language. As for reference, we will discuss the licensing properties of other functional projections later on in the text. As far as discourse linked interpretation is concerned, it could be associated to the order of constituents, but we can not go into this question here.

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

.

Kase phrase

It has been proposed that in languages with morphological Case, a functional projection (frequently represented as KP) should be activated. We cannot go into the discussion of this category, but we represent some possible structures in (9). (9)

a. b. c.

[KP K [DP D [NP N]]] (Löbel, 1994: 51) [DP D [CaseP2 Case2[Acc] [QP Q [CaseP1 Case1[Gen] [N]]]]] (Abraham 1997: 50) a. [. . . K-of admiring [John a picture]] → b. . . . [admiring [ti a picture ]]j of Johni K-of tj

As can be seen by the two structures in (9), Case Phrase can be envisaged either as an external ‘link’ of the DP to relational properties of the sentence (Case assignment) or as a category accounting for the internal morphological and syntactic properties of the nominal. Some authors propose that this projection should also be extended, with a different lexical content, to languages that do not exhibit case morphology. We leave the question of the extension of KP to Romance languages open, since this point merits further investigation. .

Number phrase

In languages like English, a noun phrase without an overt determiner can appear in an argumental position in the plural, but not in the singular: (10)

a. Beavers build dams b. *(A) beaver builds dams

This data can be interpreted in several ways, but the main alternative views are the presence vs. absence of an empty determiner in (10a). Boucher (2003, 2005), adopts a principle established in previous work he relies on, (11)

The restricted Quantification Constraint (RQC): The existential interpretation of DPs is only available if two ‘logically’ different elements are present, filling distinct syntactic positions and respectively providing the domain of quantification and the ‘quantifying in’ operator. (Boucher 2005: 97)

The assertion in (11) has to be interpreted in the following line of reasoning. Lexical items (N, V, A, P) are first order predicates having scope respectively over objects, situations, attributes, and locations. Each lexical head projects a ‘pure’, semantic projection and assigns (one or more) theta roles. All nouns predicate a quality of some referent, which we call R (Higginbotham 1986), and place in [Spec, NP]: (12)

[NP R [N’N. . .

S F O PRO

‘Semantic’ functional heads are operators that must bind an argument in the lexical projection they c-command. Specifically, D binds (R) in the NP projection.



1st proofs

 Anna Bartra Kaufmann

Following Boucher, in the case of Bare NPs, other extended functional projections may do so in absence of DP. Restating Boucher’s claims, we can hypothesize as in (13) that NumP can act as a quantifying operator in a language with overt plural morphology: (13)

The universal condition on restricted quantification can be satisfied by NPs if dominated at least by NumP, which is included in the set of ‘operator’ heads.

In Boucher’s analysis, two other conditions are to be met. In first place, number affixation has to be strong (and therefore it rises to D for checking purposes). As for subjects, another condition has to be met: NumP has to be dominated by KP. (14)

[NUMP [NUM’ [Ni-Num] [NPR [N’ti]]]]

The structure in (14) would reflect the properties of a nominal like the English one in (10a), represented in (15): (15)

[NUMP [NUM’ [beaver-s] [NPR [N’beaver]]]]

However, these structures are not possible in Modern French, Modern Spanish or Catalan,15 because in these languages the number affix is not strong, and there is no KP available. In Boucher’s analysis, therefore, for a BNP to be used in subject position, the noun must adjoin also to K0. Germanic languages have maintained a case position, while Romance languages have not. This position is justified because Modern German determiners, nouns, and pronouns bear overt case morphology, and adjectives have a ‘strong’ declension in the absence of an overt determiner, as opposed to a ‘weak’ declension in presence of D. This would mean that KP is still an active projection in Germanic and that either the determiner or the (first) adjective rises overtly to K, as in Modern German and perhaps Middle English, or covertly at LF, as in Modern English. Boucher claims that the same was the case in Old French. Once this possibility is lost, BNPs can only appear in VP governed (internal) position. We would like to claim that morphological properties of Modern Romance Noun Phrases play a role in the behaviour of DPs. Since there is an active DP projection with φ features (number and gender), these features cannot be checked

. We leave aside the well known special cases, like newspapers’ headings in (i), complex BNPs or BNPs with complements, which have special licensing conditions: i. ii. iii.

Huelguistas toman el aeropuerto (MS) strikers take-6- the airport Mujeres, niños y ancianos merecen atención especial (MS) women, children, and old people need-6- care special Niños de todas las edades recibiran la vacuna children of all-.- the-.- ages-.- receive-6- the vaccine

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

if no nominal rises into that position. The English D has no such features; therefore no rising of NP is needed overtly. What affects our present reasoning is that in some cases to be discussed below, NumP can be a sufficient Functional Category in order to license referential arguments in absence of D.16,17 Summing up, we have considered a proposal that states that in the absence of D, NumP and KP can be the projections responsible for licensing the referential properties of nouns in Old Romance. In the following section, we will review the properties of (definite) determiners in Old Romance, focusing on Spanish. .

The grammaticalization of determiners in Old Romance

As is a well-known fact, the Romance definite article results from the grammaticalization of Latin demonstratives.18 We briefly review the main properties of such process.

. The role of Number has been shown to be relevant also at a more basic level to control properties that traditionally had been considered lexical. Further exploiting this line of reasoning, Stark (2005, 2006) and Stark and Pomino (2006) point out that in Latin, “the basic notional distinction between an individualized object and some uncontoured substance would be coded using gender and number” (caseus ‘one single (piece of) cheese’, vs. caseum ‘cheese as a substance’). In Romance, where the distinction between neuter and masculine/feminine has been lost, some nouns have a mass interpretation when used in the singular, but are reinterpreted as count in the plural, as we see in (a) versus (b): a.

queso, pan, café, vino, papel, melón . . . cheese, bread, coffee, wine, paper, melon ‘some cheese, some bread, some coffee, some wine, some (quantity of) paper, some melon, . . . quesos, panes, dos cafés, dos vinos, unos papeles, tres melones cheeses, breads, two coffees, two wines, some papers, tree melons ‘Some pieces of cheese, some pieces of bread, two cups of coffee, two different sorts of wine, certain papers, tree melons, . . .’

Furthermore, the authors assume that nouns do not have a specific interpretation with regard to ‘count’ or ‘mass’. The interpretation of a nominal as being ‘semantically pluralized ’ is created by merge of a noun with a functional head ‘Pl*’ which operates semantic pluralisation. See also Bosque (1996). . Picallo (2005, 2006) assesses the existence of another nominal functional category, , which in Romance Languages like Catalan and Spanish encodes the binary feature [+/− fem]. For this author, nouns that are selected only by the  or gender category name kinds or types. To name instantiations of types, c () needs to merge with number, hosted in a superordinate functional projection.

S F O PRO

. See Giusti (2001), Roberts & Roussou (2003) and references therein for an analysis on formal terms.



1st proofs

 Anna Bartra Kaufmann

Formal grammar has adopted the classical concept of grammaticalization (see Traugott & Heine (1991), Harris & Campbell (1995); Hopper & Traugott (1993) and references therein), which basically implies the loss of semantic descriptive content, the loss of phonological weight and the loss of distributional freedom, and has reinterpreted and restated it in the internal theoretic terms in (16): (16)

Grammaticalization a. Reanalysis of a lexical category into a functional category (semantically impoverished). b. Direct merge of the functional element in the head of the corresponding functional projection (→ no Move).

The two properties in (16) can be formally represented as in (17): (17)

[XP Y + X [YP tY]] > [XP Y = n X [YP Y]]

Functional categories act as probes for the formal unvalued features of the lexical categories they govern. Therefore, the process of grammaticalization is always a bottom-up process, i.e. functional categories are hierarchically higher than the lexical ones they stem from.19 Following Roberts & Roussou (2003); Batllori & Roca (2000) or Giusti (2001) the grammaticalization of Romance demonstratives can be represented as in (18): (18)

[DP D [DemPDem [ . . . [XP] . . . [NP] . . . ]]]20

The representation in (18) means, following these authors, that Demonstratives had to climb to D in order to check their definiteness properties, whereas definite articles merge directly into D. Batllori & Roca (2000) argue that in Old Spanish, there was a period in which the process represented in (19) was not yet completed: in

. In a different theoretical framework, namely a functionalist one, Company (2004) also argues that forms grammaticalized by the effects of a subjectivization process are usually located at the leftmost position in the utterance, usually at the very beginning, so that their meaning affects the whole utterance in a global way. . This structure would be supported by DPs like Spanish: El niño este (Lit.: 〈The boy this〉, i.e. ‘This boy’). Nevertheless, this order could also be obtained by post posing a focalised Demonstrative. As our work focuses basically on determinerless NPs, nothing hinges specifically on adopting the structure in (18). See also Cinque (2005), for a detailed discussion about different cross-linguistically possible orders within the DP.

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

that period, Old Spanish had two different subsystems of grammar for ille descendants. In the etymological grammar, el, la, los, las still behave as deictic elements, i.e.: demonstratives, and discourse linking properties are not marked by means of these elements. Therefore, examples like those in (19), in which both the noun phrases can receive a linked interpretation, would be commonplace. In the innovative one, ille descendants had already been reanalysed as discourse anaphors, and those examples would not be possible with the same meaning: (19)

a.

b.

Old Spanish: Fazienda: 73 E quando alçava Moisés sos manos, vencia Israel. and when rais- Moses his hands, win- Israel. Manos de Moysen eran pesadas Hands of Moses be-3-- heavy-.- ‘And when Moses raised his hands, Israel won. Moses hands were heavy.’ Old Spanish: Cid, 712 Moros le reciben Arabians he-=meet-6- ‘Some Arabians meet him’

We would like to argue that, at least from the XII C. on, there was an interpretive difference between noun phrases with or without overt definite determiners. We also claim that, in absence of such definite articles, other functional categories could bind the referential properties of noun phrases. This fact would mean that reference is a property distinct from discourse linking and definiteness.

.

Determiners in Old Spanish and Catalan

We briefly review some generalisations about the need of determiners in Old Spanish and Old Catalan in order to show that determinerless subjects in passive sentences are not naturally derivable and need to be carefully examined. It will be argued that BNP subjects can be taken as evidence for the functional architecture of NPs. In (20), the properties that intersect to account for the distribution of determiners are shown: (20)

a. b. c. d. e. f. g.

Mass vs. count properties. Referential properties of the head N ([+/− unique]). Syntactic function and position of the DP/NP. VP Aktionsart. Thematic structure of the VP. Mood and modal features. Discourse value (topicality) of the DP/NP.

S F O PRO



1st proofs

 Anna Bartra Kaufmann

Countable entities denoting a unique element21 need a D from very early stages on in Romance: (21)

a.

b.

c.

d.

Old French: Lancelos, Artu ; cit, Dufresne, Dupuis, Tremblay (2003: 33) quant li rois entent ceste parole, . . . when the king hear-3- this- discourse-, . . . ‘As the King heard this discourse, . . .’ Old Catalan: Llull, Mer., VIII, LIV). Lo rei d’aquelles gents era hom molt bo e acustumat the king of=tose people- be-3- man very good and well-mannered ‘The King of those people was a good and well-mannered man’ Old Italian, Novellino, 53, rr. 3–9, cit. Italant Lo’ mperadore donò una grazia a un suo barone The=emperor give-3- a present to one his lord (. . .) Il barone (. . .) mise alla porta un suo passagere (. . .). The lord put to=the door one his traveller(. . .) Il pedaggere li domandò un danaio. The traveller him- ask-3- a coin ‘The emperor made a present to one of his lords. The lord sent away one of his guests. The guest asked him for a coin. . .’ Old Portuguese, apud E assi escapou o comde Joham Fernandez de nom seer morto and so escaped the count Joham F. from not being death ‘And this way escaped the count J.F. from death’

As is clear from (21), nouns denoting unique entities such as the king (21a, b), the emperor (21c), or a count from which the name is previously given in the text (21d), need a definite determiner from early times on. In (21c) we also see the effects of discourse linking: clearly enough, thematic or discourse linked instances need the definite article. Abstract mass nouns with predicate value acting as direct objects of a semantically impoverished or light verb22 need no D in general. Such constructions are still found in Modern Spanish and Catalan: (22)

a.

O.S.: Fazienda UM, §2. e vio el ninno que plorava e ovo piedad del and saw the- boy which cry-3- and had pity of=him ‘He/she saw the boy crying and took pity on him’

.

This semantic value is highly dependent on encyclopaedic knowledge.

.

For a general view and references, see Mohanan (2006).

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

b.

Modern Spanish: No tiene piedad de sus enemigos not has pity of his enemies ‘He/she takes no pity on his enemies’

As expected, these bare NPs (BNPs) cannot move outside the VP in Modern Spanish (23) or Modern Catalan (24). This fact means that they can be only licensed as being part of the predicate:23 (23)

a. *cola sera traída por esta decision tale (=consequences) will be brought by this decision ‘This decision will bring (bad) consequences PASS’ b. *provecho pues sacaste de la situación benefit therefore took--2 from the situation ‘You took benefits from this situation’ c. *tiene el chiste gracia has the joke fun ‘The joke is fun’

(24)

a. *enemistat fou declarada entre elles enemy was declared between them-. ‘They became enemies’ b. *salvament va ser demanat des de la barca salvation was demanded from-of the boat ‘There came a  from the boat’

Compare the situation in (23) and (24) with the possibilities in Old Spanish (25) and Old Catalan (26), where BNPs are found in displaced positions: (25)

a.

b.

O.S.: Fuero de Zorita de los Canes, §6, pág. 126. Si los fijos al padre o a la madre sospecha ouieren if the children to=the father or to the mother suspect have-6- que alguna cosa les çele en particion that something them--Plur hid--3 in partaging ‘If the children suspect from the father or mother that he or she hid something in the part aging’ O.S.: Fazienda UM, §41 darta el to Sennor Dios alçamiento sobre give--2 the your Lord God superiority over

. There are several accounts found in the literature for this fact. Jayaseelan (1984) put forward an account following which the internal NP in light verb constructions would have a predicative value. The theta roles assigned by this NP would “percolate” up inside until the VP, and from here on they would be assigned to the other arguments and finally to the subject. In more recent accounts, BNPs inside a VP have been analyzed as properties; see f.i. Dobrovie Sorin, Bleam and Espinal (2006) and references therein.

S F O PRO



1st proofs

 Anna Bartra Kaufmann

c.

(26)

a.

b.

c.

todas las yentes de la tierra-. all the-.- people of the earth ‘God will make you superior to all people in the world’ CORDE, 1260, Anón. El Nuevo Testamento según el manuscrito escurialense I-j.6 Ca ley fue dada por Moysen; because law was given by Moshe gracia e uerdat fu fecha por Ihesu Cristo mercy and truth was made by Jesus Christ ‘Cause the Law was given by Moshe, mercy and truth were made by Jesus Christ’ Llull, Blanq., II, LX volien que al senyor abat fos retut guardó wanted--6 that to=the sir abbot were given honor ‘They wanted that honour would be given to the abbot’ Llull, Blanq., II, XXXIV E fo feta concordança e amistat enfre aquelles gents e and was made agreement and friendship between those people and los cathòlichs the catholics ‘And those people and the catholic ones entered into friendship and peace’ Vides, 176.17 per què dematinent fo fet gran terratrèmol because early in the morning was made big earthquake ‘There was an earthquake early in the morning’

In general, the possibility of appearance of BNPs in Old Spanish and Old Catalan obeys clear semantic restrictions, and becomes more and more restricted as time proceeds. As for the description of the evolution of determiners in Old Spanish and Old Catalan, we build our analysis on Company (1991a/b) and Batllori y Roca (2000). Company (1991a, b) establishes the filter in (27): (27)

No se permiten nominales escuetos si el nombre designa entidades definidas referenciales, concretas, tangibles. (‘No BNPs are allowed if the noun designates definite, referential, concrete, physical entities’).

This filter is absolute in the case of subjects. Some BNPs are permitted as direct objects -or in a position governed by a preposition-denoting kinds or types (as opposite to tokens). In the singular only kinds are allowed as Ns complement of a P. The BNPs allowed belong to one of the categories listed in (28): (28)

BNPs in Old Spanish and Old Catalan; a. abstract/deverbal predicative Ns (in construal with an Auxiliary or light verb); b. mass NPs; c. unique Ns; d. indefinite kinship Ns.

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

Some Old Spanish examples are given in (29) and some Old Catalan examples in (30): (29)

a.

b.

c.

c.

d.

(30)

a.

b.

c.

d.

Fazienda UM, §46, p. 105 Cras avredes salvamyento tomorrow have--2 rescue 〈deverbal N〉 ‘You will be rescued tomorrow’ Fuero Zorita, p. 36 Del que quisiere petiçion fazer. of=the which like--3 request make ‘From the one you would made the request’ Cid, 553 Açerca corre Salon, agua no=l puedent vedar near flows Salon, water not=him.DAT can.6 prohibit ‘Since near from there flows the Salon, they cannot forbid water to him’ Hita, 294d por ello en infierno, desque morio yacia, because of that in hell, since die...3, lie...3 ‘Because of that, (he) was lying in hell since his death’ Crónica General, 644/37a et que tomare marido con and that take..3 husband ‘And that she would marry’ EVM, carta 90.45, p. 029.12 E aytampoch del dit fet havem haüda provisió and nor of=that mentioned fact have..4 evaluation of you alcuna de vós anyone of you ‘Also, we have no evaluation from your part of that fact’ Feits,2 feu resposta al bisbe e als nobles make..3 answer to=the bishop and to=the aristocrats qui uengueren ab ella who come..6 with her ‘She answered the bishop and the aristocrats who came with her’ Llull, Mer., XVIII Fill, bonea d’hom està en membrar son, goodness of=man is in remember. ‘Son, the goodness of man is to remember (God)’ Lull, Blanq., II, LIX prudència li donà intel·ligència prudence him. give..3 intelligence ‘The prudence gave intelligence to him’

S F O PRO



1st proofs

 Anna Bartra Kaufmann

e.

SVF, Ser., XCVII matrimoni fo honrat per Jesuchrist marriage be..3 honored by Jesus Christ ‘The marriage was honoured by Jesus Christ’

Some pairs can be found that illustrate very clearly the fact that the type or kind reading needs no determiner, whereas the token reading usually takes it from very early stages. (31)

a.

b.

Crónica General, 118/26a E a su mugier otrosi mataron= la a espada and to his wife also killed=her to sword ‘They put his wife to the sword’ Crónica General, 118/7a et dio muy gran ferida con el espada and give..3 very big wound with the sword ‘He hurt him severely with his sword’

In (31a) it is meant that the wife was killed with a sword or similar weapon, whereas in (31b) it is told a specific event of hurting with a specific sword. Similarly, in (32a), a concrete battle in a concrete camp is narrated; in (32b) it is meant that Arabians do not begin to fight: (32)

a.

b.

Cid, 755 los moros nos van del campo the arabians no= go from=the camp ‘Arabians don’t leave the camp’ Cid, 2354 los moros non ficaran en campo the arabians not enter in camp ‘Arabians do not begin the battle’

Taking into account the use of determiners in the noun phrases of the Poema de Mio Cid, Company (1991a, b) establishes the generalisation in (33):24 (33)

Company (1991a, b: 95) −Determiner +Determiner

Referential Near − − + +

Metaphorical + −

Known − +

Abstract + −

. This author also establishes three stages in the introduction of the definite article in Old Spanish, evolving as schematized in (i): i.

S F O PRO

A. subject kind names (XII–XIII Cent.) → B. nouns of unique reference (XIV Cent.) → C. abstract mass nouns (XV Cent.). (Company, 1991/b: 101–103).



1st proofs

Determinerless NPs in Old Romance passives 

.

Functional categories in BNP in Old Spanish and Old Catalan

As we have already seen, some semantic interpretations permit BNPs in Old Spanish and Old Catalan. What remains to be explained is the syntactic mechanism by which this noun phrases are licensed. Predicative noun phrases, such as the ones in (29a/b) and (30a/b) could be licensed inside the VP, probably by means of an incorporation mechanism. There are, nevertheless, other cases in which the NP is interpreted as an argument outside the VP, and, presumably, functional categories of the nominal domain play a role in the licensing process. Still the question remains: (34)

Which functional projections are present in the extended nominal projection in Old Spanish and Old Catalan?

We have considered passive sentences with an abstract (deverbal) BNP to have a heuristic value. Assuming any standard analysis of passive structures as the one presented before, or, we have to assume that the moved BNP needs referential properties to be interpreted as an argument outside the VP and not as a part of the predicate.25 We follow (Batllori y Roca, 2000) in their assumption that the grammaticalization process of determiners was still unachieved or unsteady in some sense in Old Romance. Therefore different possibilities were at hand to license the referential properties of noun phrases, depending on the properties of the nominal itself and of the properties of the structure they were merged in. Moreover, we assume that from the universal set of functional categories only those are “activated” for which there is overt morphology in the lexicon of the language. We also assume the principle in (35), following Roberts & Roussou (2003): (35)

Grammaticalization is an upward process.

This means that grammaticalized linguistic objects are merged higher in the structure than the lexical items they stem from. Following Batllori and Roca (2000) or Roberts and Roussou (2003), we would obtain (36) as a representation of the grammaticalization of Demonstratives into Determiners in Old Romance:26 (36)

[SD D [SDEM D [SDEM Dem [ . . . [Pos] . . . [SN] . . . ]]] (Batllori & Roca (2000: 253))

. For the semantic interpretation of the NPs licensed inside the VP, see Dobrovie-Sorin, Bleam & Espinal (2006). . A similar view is presented in Giusti (1999), who asserts that demonstratives are XPs merged in the structure lower than D and further moved to [Spec, DP] −SpecFPmax in Giusti’s terms- to check their referential features. As for the grammaticalization of the determiner, see also Giusti (2001).

S F O PRO



1st proofs

 Anna Bartra Kaufmann

Reviewing (some of) the different functional projections that have been proposed to constitute the extended projection of the NP, we obtain the sequences in (37a/b): (37)

a. b.

[ D ([Kase]) [ Dem [ (Q) ([Kase]) [ Num [ Gen [N]27, 28 [ D ([Kase]) [ Dem [ (Q) ([Kase]) [ Num [n [/√] . . .

In what follows we are going to abstract away from K(ase)P. It is likely that such a projection should be needed in Old French and Old Occitan, but we think that we have no compelling evidence for it in Spanish and Catalan, and this is a point that needs further investigation.29 As for the lower part of the structure (37) presents two alternate possibilities that in fact do not amend the main purposes of our reasoning. (37a) presents a structure in the spirit of Picallo (2005, 2006) where the Gender is the functional element with classifying properties. (37b) follows the spirit of Distributed Morphology (see Halle & Marantz (1993); Marantz (2000)),30 by asserting that roots are unlabelled as for lexical category; therefore, a functional element such as n or v is the responsible of the categorical status of the root. Both proposals assert that there is a classifying or categorizing nominal feature above the root (be it N or √). From a morphological point of view, Gender or nominalizer marks are sometimes implicit in the stem of the noun (árbolMASC, ‘tree’), sometimes present in a fossilized (semantically irrelevant) morphological marker (caball-oMASC ‘horse’, libr-oMASC ‘book’, ventan-aFEM ‘window’). If in Latin the way of classifying Ns were nominal . We cannot revise here the huge bulk of literature concerning nominal functional categories (see, for instance, Bernstein (1997); Löbel (1989), (1994); Giusti (1999), (2001), (2002); Picallo (2005), (2006); Ritter (1993); Stark (2006); Watanabe (2006) and references therein). . As mentioned before, there is no general agreement with respect to the relative order between Dem and D. The structure we propose aims to capture the sequence of the diachronic evolution. Dem and D are usually in complementary distribution in pronominal position. I leave the question open. (See also Section §3, and Footnote 16). .

The preposition de introduces mass or plural denoting Direct Objects in Old Catalan: i.

ii.

Llull, Blanq., II, LVIII A la taula hach de diverses viandes at the table (he) had of several meats ‘He had several meats at lunchtime’ Jaume I, Feits, 225 E quan fom aqui nos haguem de conseyl ab Edon ·Bñ· ·G· Dentença and when were..4 here we had of council with 

‘As we arrived we asked EBGD for advice’ This preposition has been considered a case marker, but there are also other possible analyses. (See Kayne (2002) and Cardinaletti & Giusti (1992) for different perspectives). .

S F O PRO

In fact, this proposal follows the spirit of Chomsky’s (1968) “Remarks on Nominalization.”



1st proofs

Determinerless NPs in Old Romance passives 

declensions, it is reasonable to argue that the residues of that declension (Gender) are word markers in Romance. As for abstract deverbal nouns, the nominalizing suffix clearly shows the nominalizing properties associated with Gender features. In (38) we present some examples of the different forms of marking Gender or nominalizing features of the stem: (38)

a. b. c.

árbol-ØMASC, pared-ØFEM tree wall caball-oMASC, ventan-aFEM horse window salva-çionFEM, ayunta-mientoMASC salvation joining together

In the remaining of this section we are going to argue that the several types of BNPs found in Old Spanish and Old Catalan project different functional nominal projections, which license the BNP. We keep separately the types presented in (28). . Abstract denominal verbs as Direct Objects of light verbs These BNPs have been said to have a predicative value;31 therefore they should not be referential. Therefore they have to stay inside the VP and incorporate into the V.32 In (39) some such cases are presented: (39)

a.

b.

Old Spanish: Libro de Buen Amor, 71b–d Por aver mantenençia; (. . .) Por aver juntamiento con fenbra placentera. to have nutrition; to have joining with female pleasant ‘to keep going; (. . .) to live together with a pleasant female’ Old catalan: Jaume I, Furs, R. 63, F.9 hauran fet testimoni. have..6 made witness ‘They will testify to’

The fact that these nouns were not completely lexicalised at that time is an explanation of the huge amount of cases of a semantically impoverished (light) verb and a deverbal nominal (in front of the non periphrastic, synthetic construction used in modern language). In these constructions, the light verb is responsible for the Aktionsart (together with aspectual, temporal and eventive features) of the predicate, whereas the root of the BNP is responsible of semantic and lexical features. The nominalizing suffix suffices to allow the nominal to be displaced. In (40) .

For the interpretation as a property see Dobrovie-Sorin, Bleam & Espinal (2006).

S F O PRO

. See Hale & Keyser (2002), or Espinal (2001, 2002, 2004), for different views of the incorporation process.



1st proofs

 Anna Bartra Kaufmann

we represent the relevant part of the structure once the root has reached the nominalizing or Gender category:33 (40)

[VP [Vaya [GenP ayunta-miento [NP ayunta]]]]

In (40) the noun, together with its Gender projection, can be incorporated into the light verb in order to give rise to a complex predicate and to provide the compositional meaning of the predicate (daua castigo, fiziere partición). Besides this possibility, in Old Spanish and Old Catalan, the BNP can be moved outside the VP in passive sentences or can be focalised: (41)

a.

b.

Old Catalan: Clams, F 27r, R 274 a. 1306 demanà (. . .) que fos donat jurament ask..3 (. . .) that be..3 given oath ‘He asked that they sweared’ Old Spanish: CORDE, 1348, Ordenamiento de las cortes celebradas en Alcalá de Henares non fue fecha stipulaçion que quier dezir prometimiento not was made stipulation that want..3 say promise ‘It was not stipulated, and that means that it was not promised’

We hypothesize that in absence of any other functional category, the nominalizing affix suffices to singularise the noun. This possibility could mean that in a period in which the Determiner was not yet completely steady the nominalizing affix sufficed to license syntactically an unincorporated BNP.34 . BNPs with a kind or unique meaning Other different semantic interpretations are possible. Some BNPs could receive an indefinite non-specific interpretation, as in (42),35 and some can even receive a unique interpretation as in (43a/b): (42)

LBA, 23c Tróxote mensaj’ divino bring..3=to you message divine ‘He brought to you a divine message’

. We represent the classical alternative. In a DM framework, the structure could be represented as in (i): i.

[VP [Vaya [nP ayunta-miento [√ ayunta]]]]

. Nevertheless, this BNPs are not able to raise to Spec,T. M.T. Espinal (p.c.) suggests that the fact that the subject of passive sentences can remain inside the VP could permit to analyse them as internal arguments. We would like to suggest that, even remaining inside the VP, these subjects have to agree with the functional projections above the Participial Phrase.

S F O PRO

. Example (43) is ambiguous, even taking the whole context into consideration. The BNP can have a unique or and indefinite non-specific reading.



1st proofs

Determinerless NPs in Old Romance passives 

(43)

a.

b.

LBA, p. 11 escogiendo é amando (. . .) salvaçión e gloria del parayso choosing and loving (. . .) rescue and glory of heaven ‘Choosing and loving the glory of Paradise’ LBA, 319a–b Otrosy con açidia traes ypocresía moreover with indolence bring-2- hypocrisy ‘With indolence you bring the hypocrisy’

These same values can be found in passives, as the example in (44) show: (44)

Old Catalan: Tir, CXCIII lla hon és trobada rahó e bondat there where be-3- find-PP reason and goodness ‘Where reason and goodness are found’

In passive sentences, we have to assume the referential value of the BNP. There are different possibilities to license the BNP. Some linguists hold that the indefinite article was explicit in the case of specific indefinites, but was null in the case of nonspecific indefinite noun phrases. See (Foulet (1990)) for Old French.36 There are, nevertheless, some problems in assuming a null indefinite determiner only for the cases of non-specific NPs. First, the reading of some non-specific indefinite NPs is very similar to a definite NP: (45)

Fuero de Soria, §27, p. 210–211 Si alguno leuare esposa agena por fuerça if someone take..3 wife somebody else’s by forc ‘If someone would kidnap somebody else’s wife’

(46)

LBA, 473 molyno andando gana/ mill working improves ‘A mill which keeps working improves’ Huerta mijor labrada da la major mançana/ vegetable garden better worked gives the better apple ‘The best you take care of your garden, the best the apples from it are’ Muger mucho seguida, sienpre anda loçana. woman very much followed always goes beautiful ‘Loved women look beautiful’

. [A null indefinite article is present] “si l’individualité ne ressort pas nettement, si nous avons affaire à un type plutôt qu’à un individu, ou si l’individu nous est présenté comme devant satisfaire à telles ou telles conditions qui pourront être ou ne pas être remplies.” (Foulet 1990: 56). Roberts & Roussou (2003: 152) assume this as a hypothesis.

S F O PRO



1st proofs

 Anna Bartra Kaufmann

(47)

Fazienda UM, §41 Maldito el baron que fiziere ymagine damned the man who would made image ‘The man who would made an image will be damned’

The preceding sentences are clear donkey sentences, others are aphorisms. Most of them contain modality elements or indefinite operators (Maldito el. . ., si, qui quier que, conjunctive tense) that trigger a generic or existential interpretation of the BNP. Some BNPs have a semantically ambiguous meaning, between a specific and a non-specific interpretation: (48)

Old Catalan: Clams, F 20v, R 243 Fo demanat si donà obra ni consel ni ajuda was asked if give..3 work nor advise nor help ‘He was aked wheter he gave {any/the }advise or help to. . .’

We suppose that these kinds of sentences are contextually disambiguated. We argue, for the sake of simplicity, that the absence of a definite determiner is enough to trigger an indefinite interpretation and to derive specificity properties from contextually driven properties. It could be argued that GenP suffices to license the reference of the BNP: (49)

.

[VP [V [escogiendo [GenP salva-çion [NP salva]]]]]

Plural BNPs

Old Spanish and Old Catalan behaved like Modern English in allowing plural BNPs as subjects: (50)

a.

b.

c.

Cid, 712 Moros le reçiben arabians him wellcome ‘Some Arabians welcome him’ Fernán González, 87c tovieron castellanos el puerto bien guardado had castilians the harbour good preserved ‘People of the castle kept the harbour safe’ Cid, 335 pastores te glorificaron shepherds youAC glorified ‘Shepherds glorified you’

In these cases the plural morphology is transparent evidence for the existence of a full NumP and we suggest, in the line of B (2005) that this functional category is the licensor of the argument value of the NP.

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

(51)

[NumP [castellan-o-s-[GenP castellan-o [NP castellan]]]]

It is interesting to note that, in a time when overt determiners were already present in the grammar of Old Catalan and Old Spanish, the presence or absence of determiner was significant, as we observe in (52): (52)

a.

b.

c.

d.

Cid, 1143 Moros son muchos, ya quieren reconbrar Arabians are many, already want to recover ‘There are many Arabians, and they want to retrieve’ Cid, 2390 Los moros son muchos, derredor le cercaban the arabians are many, around himAC = surround..6 ‘Many arabians surrounded him’ Cid, 712 Moros le reciben por la seña ganar arabians him wellcome for the sign to win ‘Some Arabians welcome him to know the sign’ Cid, 618 Los moros yazen muertos, de vivos pocos veo the Arabians lie dead, of alive few see..1 ‘(The) Arabians are dead, I see very few ones alive’

With this observation we mean that one should distinguish between the licensing and the semantic properties of NumP. The fact that NumP suffices in order to have the R position of NP licensed does not mean that it provides the NP with exactly the same semantic and discourse features as the definite article filling this position. (53)

[NumP [mor-o-s [GenP mor-o [NP mor]]]]

Besides the plural morpheme, coordination, adjectival complementation, generic operators, and other quantifying elements also trigger NumP activation. .

A difference between Old Catalan and Old Spanish?

The number of passive sentences with BNPs, mainly abstracts ones, seems to decrease quickly in Old Spanish. From the second half of XIV Cent, we find them almost only in law documents. In Catalan, the rate of abstract deverbal BNP subjects seems to be higher and last near a half a Century or a century later. This fact could be related to different related properties. One of them could be the endurance in Catalan of the above mentioned de as a possible Case marker, and therefore, the possibility to have another functional element (KP) in this language. Another one could be the frequent position of the derived subject in post verbal position in Catalan, as the examples in (55) show.

S F O PRO



1st proofs

 Anna Bartra Kaufmann

(54)

a.

b.

c.

d.

Llull, Blanq., II, LX Longament de temps m’ és estat fet honrament per vosaltres longly of time to me is been made honour by you. ‘During a long time you honoured me’ Ll. Cort Cocentaina, a. 1250–1299 éser fet conpliment de dret to be made compliance of law ‘The law to be respected’ SVF, Ser., XCII fo moguda quesito entre ells was caused discussion between them ‘It was caused a discussion between them’ Eximeno, Crim, carta 2, a. 1475–1499 que sia fet protest ab scripturathat be..3 made protestation with document ‘That you lodge a written appeal’

We can hypothesise that the BNP in these structures remains in the verbal or participial domain, were it checks an objective (or inherent partitive) Case. (55)

[TP T [vP v [AgrO/PP Agr [VP/PP [BNP] [PP]]]]] *

In active sentences, the DP subject argument checks a D feature in T. The passive post verbal subjects in Old Catalan have no D feature, since they only project a Num element. Therefore, they check their Case in the VP domain against the participial agreement. As the discontinuous lines show, these subjects cannot raise further. We propose that, given morphological evidence, (at least) the functional categories in (38a) or (38b) are part of the functional architecture of Noun Phrases. Probably, the elements in these categories have been grammaticalized in an upward progression. If D is missing, the other functional elements are responsible of the licensing of the referential properties of Ns – the R property. We take passive sentences and structures in which the noun has been moved as evidence for the referential value of the NP. .

Concluding remarks

In this paper, we have tried to bring some new evidence to support recent proposals on the functional properties of the nominal domain. We have taken Old Spanish and Old Catalan Bare Noun Phrases, mainly subjects of passive sentences and other BNPs moved from their basic position inside the VP to show that in these grammars the grammar of the Determiner was still unsteady and other nominal

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives 

functional projections were able to license the NP, mainly noun markers such as Gender or nominalizing affixes and Number Phrase. We relied on present accounts of passive sentences, allowing for the possibility of BNPs to stay inside the VP to agree with the nominal agreement features of the Past Participle. We adopted a methodological minimalism in the sense that we argue that from the inventory of Functional Categories in Universal Grammar, particular grammars activate only those for which there is formal and morphological evidence. We have brought further evidence for current hypothesis that assess that in the nominal domain (at least) the following categories are needed: Gender Phrase or a classifying nominalizing morpheme, Number Phrase, (Kase Phrase), and Determiner. We said nothing about other elements inside the DP such as Possessors or Quantifiers. Furthermore, we hypothesized that the external subject position requires at least an active FC NumP. May be in a very early stage, Gender Phrase or a nominalizing affix would suffice in some case to license an internal subject, but I have no concluding evidence in that sense. We have also argued that Noun Phrases show semantic differences according to the active Functional Categories they have. Some important points need further investigation. First, a quantificational statistical analysis is needed in order to check if the diachronic evolution mirrors the hierarchy of Functional Categories presented. Second, we need a principled explanation of the differences between Old Spanish and Old Catalan presented. Third, we have to check if the properties of pronominal passives and impersonal sentences parallel those investigated so far. And forth, the relevant evidence has to be brought to distinguish between indefinite unspecific and specific value of the BNPs.

References Abney, Stephen. 1987. The English Noun Phrase in its Sentential Aspect. PhD Dissertation, MIT. Abraham, Werner. 1997. The interdependence of case, aspect and referentiality in the history of German: The case of verbal genitive. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 29–61. Cambridge: Cambridge University Press. Batllori, Montserrat & Roca, Francesc. 2000. The value of definite determiners from Old Spanish to Modern Spanish. In Diachronic Syntax. Models and Mechanisms. Susan Pintzuk, Georg Tsoulas & Anthony Warner (eds), 241–254. Oxford: Oxford University Press. Bernstein, Judy. 1993a. Topics in the Syntax of Nominal Structure across Romance. PhD Dissertation, CUNY. Bernstein, Judy. 1993b. The syntactic role of word markers in null nominal constructions. Probus 5: 5–38. Bernstein, J.B. 1997. Demonstratives and reinforcers in Romance and Germanic languages. Lingua 102: 87–113.

S F O PRO



1st proofs

 Anna Bartra Kaufmann Borer, Hagit. 1994. The projection of arguments. In Functional projections, Elena Benedicto & Jay Runner (eds). University of Massachusetts Occasinal papers 17: 19–48. Bosque, Ignacio (ed.). 1996. El sustantivo sin determinación. La ausencia de determinante en la lengua española. Madrid: Visor. Boucher, Paul. 2003. Determiner phrases in Old and Modern French. In From NP to DP. Vol. 1: The syntax and semantics of noun phrases, Martine Coene & Yves D’hulst (eds), 47–69. Amsterdam: John Benjamins. Boucher, Paul. 2005, Definite reference in Old and Modern French: The rise and fall of DP. In Gramaticalization and Parametric Variation, Montserrat Batllori, M.-Lluïsa Hernanz, M. Carme Picallo & Francesc Roca (eds), 95–108. Oxford: Oxford University Press. Cardinaletti, Anna & Giusti, Giuliana. 1992. Partitive ne and the QP-Hypothesis. A case study, Proceedings of the XVII meeting of Generative Grammar, Trieste, February 22–24, 1991, 121–142 Volume presented to G. Francescato on the occasion of his seventieth birthday, Elisabetta fava a cura di. Torino: Rosenberg & Sellier. Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge MA: The MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge MA: The MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: A life in language, Michael Kestowicz (ed.), 1–52. Cambridge MA: The MIT Press. Cinque, Guglielmo. 1999. Adverbs and functional heads. A cross-linguistic perspective. Oxford: Oxford University Press. Cinque, Guglielmo. 2005. Deriving Greenberg’s universal 20 and its exceptions. Linguistic Inquiry 36(3): 315–332. Coene, Martine & Yves d’Hulst (eds). 2003. From NP to DP. (2 Vols). Amsterdam: John Benjamins. Company Company, Concepción. 1991a. La extensión del artículo en el español medieval. Romance Philology 44(4): 402–424. Company Company, Concepción. 1991b. La frase sustantiva en el español medieval. Cuatro cambios sintácticos. Universidad Nacional Autónoma de México. Publicaciones del centro de lingüística Hispánica 32. Company Company, Concepción. 2004. Gramaticalización por subjetivización como prescindibilidad de la sintaxis. Nueva Revista de Filología Hispánica LII(1): 1–27. Delfitto, Denis & Schroten, Jan. 1991. Bare plurals and the number affix in DP. Probus 3(2): 155–185. Dobrovie-Sorin, Carmen. 2001. Génitifs et déterminants, In Typologie des groupes nominaux, Georges Kleiber, Brenda Laca & Liliane Tasmowski (eds), 205–234. Rennes: Presses Universitaires de Rennes 2. Dobrovie-Sorin, Carmen, Bleam, Tonia & Espinal, Teresa M. 2006. Bare nouns, number and types of incorporation, In Svetlana Vogeleer & Liliane Tasmovsky (eds), 51–79. Dufresne, Monique, Dupuis, Fernande & Tremblay, Mireille. 2003. Preverbs and particles in Old French. In Yearbook of Morphology 2003, Geert Booij & Jaap van Marle (eds), 33–60. Dordrecht: Kluwer. Espinal, Maria Teresa. 2001. Property denoting objects in idiomatic constructions. In Romance languages and Linguistic Theory 1999, Yves D’Hulst, Johan Rooryck and Jan Schroten (eds), 117–141. Amsterdam: John Benjamins. Espinal, Maria Teresa. 2002. Idiomatic constructions vs. light verb constructions. In Current Issues in Generative Grammar, M. Leonetti, O. Fernández Soriano & V. Escandell (eds), 69–81. Univerdidad de Alcalá, UNED y Universidad Autónoma de Madrid. Espinal, Maria Teresa. 2004. Lexicalization of light verb structures and the semantics of nominals. Catalan Journal of Linguistics 3: 15–43.

S F O PRO



1st proofs

Determinerless NPs in Old Romance passives  Foulet, Louis. 1990. Petite syntaxe de l’ancien français. Paris: Champion. Giusti, Giuliana. 1999. The functional structure of noun phrases: A bare phrase structure approach. University of Venice Working Papers in Linguistics 1999: 105–160. (Reprinted as Giusti 2002). Giusti, Giuliana. 2001. The birth of a functional category: From Latin ILLE to the Romance Article and personal pronoun In Current Studies in Italian Syntax. Essays offered to Lorenzo Renz, Guiglielmo Cinque & G. Salvi (eds), 157–171. Amsterdam: North-Holland. Giusti, Giuliana. 2002. The functional structure of noun phrases: A bare phrase structure approach. In Functional Structure in DP and IP, Guglielmo Cinque (ed.), 54–90. Oxford: Oxford University Press. Hale, Ken & Keyser, Jay S. 1993. On argument structure and the lexical expression of syntactic relations. In The View from Building 20: Essays in honor of Sylvain Bromberger, Ken Hale & S. Jay Keyser (eds), 53–108. Cambridge MA: The MIT Press. Hale, Ken & Keyser, Jay. 2002. Prolegomenon to a Theory of Argument Structure. Cambridge MA: The MIT Press. Halle, Morris & Marantz, Alec. 1993. Distributed morphology and the pieces of inflection. In The view from Building 20: Essays in linguistics in honor of Sylvain Bromberger, Ken Hale & S. Jay Keyser (eds). Cambridge MA: The MIT Press. Harris, Alice C. & Campbell, Lyle. 1995. Historical Syntax in Cross-Linguistic Perspective [Cambridge Studies in Linguistics 74]. Cambridge: Cambridge University Press. Heine, Berndt, Claudi, Ulrike & Hünnemeyer, Frederieke. 1991. Grammaticalization: A Conceptual Framework. Chicago IL: The University of Chicago Press. Heycock, Caroline & Zamparelli, Roberto. 2003. Friends and colleagues: Plurality, coordination, and the structure of DP Ms. University of Edinburg & University of Bergamo. Downloadable at http://semanticsarchive.net Higginbotham, James. 1986. Indefiniteness and predication. In The Representation of Indefiniteness, Eric Reuland & Alice ter Meulen (eds), 43–70. Cambridge MA: The MIT Press. Hopper, Paul J. & Traugott, Elisabeth Closs (eds), 1993. Grammaticalization. Cambridge: Cambridge University Press. Italant, available at http://geocities.com/gpsalvi/konyv Jayaseelan, Karattuparambil A. 1984. Complex Predicates and the Theory of θ-Marking [CIEFL Working papers in Linguistics 1.1]. Hyderabad: Central Institute of English and Foreign Languages. Kayne, Richard. 2002. On some prepositions that look DP-internal: English of and French de. Catalan Journal of Linguistics 1: 71–115. Kemenade, Ans van & Nigel Vincent (eds). 1997. Parameters of Morphosyntactic Change. Cambridge: Cambridge University Press. Koopman, Hilda & Sportiche, Dominique. 1991. The position of subjects. Lingua 85(1): 211–258. Kratzer, Angelika. 1996. Severing the external argument from the verb. In Phrase structure and the lexicon, Johan Rooryck & Lauri Zaaring (eds), 109–137. Dordrecht: Kluwer. Lingua 115(6). 2005, June. The structure of indefiniteness: Issues in the form and interpretation of noun phrases. Monographic issue. Löbel, Elisabeth. 1989. Q as a functional category. In Syntactic Phrase Structure Phenomena, Crista Bhatt, Elisabeth Löbel & Claudia Maria Schmidt (eds), 133–187. Amsterdam: John Benjamins. Löbel, Elisabeth. 1994. KP/DP-syntax: Interaction of case-marking with referential and nominal features. Theoretical Linguistics 20: 37–70.

S F O PRO



1st proofs

 Anna Bartra Kaufmann Longobardi, Giuseppe. 1991. Proper Names and the Theory of N-movement in Syntax and Logical Form [University of Venice Working Papers in Linguistics 9]. Venezia. Longobardi, Giuseppe. 1994. Reference and proper names. Linguistic Inquiry 254: 609–665. Marantz, Alec. 2000. Words. Ms, MIT. Mohanan, Tara. 2006. Grammatical verbs with special reference to light verbs. In The Blackwell Companion to Syntax, Martin Everaert & Henk van Riemsdijk (eds), 459–492. Oxford: Blackwell. Munn, Alan & Schmitt, Christina. 2004. Number and indefinites. Lingua 115: 821–855. Pereltsvaig, Aissa. 2006. Small nominals. Natural Language and Linguistic Theory 24: 433–500. Picallo, M. Carme. 1991. Nominals and nominalization in Catalan. Probus 3: 279–316. Picallo, M. Carme. 2002. Abstract agreement and clausal arguments. Syntax 5: 116–147. Picallo, M. Carme. 2005. Some notes on grammatical gender and l-pronouns. In Proceedings of the Workshop Specificity and the Evolution/Emergence of Nominal Determination Systems in Romance [Arbeitspapier 119], Klaus von Heusinger, Georg A. Kaiser & Elisabeth Stark (eds), 107–121. Konstanz: Fachbereich Sprachwissenschaft der Universität Konstanz. Picallo, M. Carme. 2006. On gender and number. Ms, UAB. Pomino, Natasha & Stark, Elisabeth. 2006. Discretness and the case of the Spanish neuter demonstratives. Ms, Freie Universität Berlin. Progovac, Liliana. 1998. Determiner phrase in a language without determiners. Journal of Linguistics 34: 165–179. Ritter, Elisabeth. 1993. Where’s gender? Linguistic Inquiry 24: 795–803. Ritter, Elisabeth. 1995. On the syntactic category of pronouns and agreement. Natural language and Linguistic Theory 13: 405–443. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Elements of Grammar, Liliane Haegeman (ed.). Dordrecht: Kluwer. Roberts, Ian & Roussou, Anna. 2003. Syntactic Change. A minimalist approach to grammaticalization. Cambridge: Cambridge University Press. Stark, Elisabeth. 2005. Typological correlations in nominal determination in Romance. In Proceedings of the Workshop Specificity and the Evolution/Emergence of Nominal Determination Systems in Romance [Arbeitspapier 119], Klaus von Heusinger, Georg A. Kaiser & Elisabeth Stark (eds), 123–138. Konstanz: Fachbereich Sprachwissenschaft der Universität Konstanz. Stark, Elisabeth. 2006a. The role of the plural system in Romance. In Explanations of grammatical change. Selected papers from the workshop at the 29th biannual meeting of the German Romanists at Saarbrücken in September 2005, Ulrich Detges & Richard Waltereit (eds). Stark, Elisabeth. 2006b. Gender, number, and indefinite articles – About the ‘typological inconsisitency’ of Italian. This volume. Szabolcsi, Anna. 1994. The noun phrase. In The Syntactic Structure of Hungarian [Syntax and Semantics 27], Ferenc Kiefer & É. Katalin Kiss (eds), 179–274. San Diego CA: Academic Press. Traugott, Elisabeth Closs & Heine, Berndt (eds). 1991. Approaches to Grammaticalization. Amsterdam: John Benjamins. Vogeleer, Svetlana (ed.). 2006. Bare plurals, indefinites, and weak–strong distinction [Belgian Journal of Linguistics 19]. Amsterdam: John Benjamins. Vogeleer, Svetlana & Tasmovsky, Liliane (eds). 2006. Non-definiteness and plurality [Linguistik Aktuell/Linguistics Today 98]. Amsterdam: John Benjamins. Watanabe, Akira. 2006. Functional projections of nominals in Japanese: Syntax of classifiers. Natural Language & Linguistic Theory 24: 241–306.

S F O PRO



1st proofs

On the structure and development of nominal phrases in Norwegian Terje Lohndal University of Oslo

The suffixed definite article in Modern Norwegian developed from a clitic in Old Norse. Such a change creates interesting theoretical questions as to how we can account for this difference in phrase structural terms, and how such a change manifests itself. This paper discusses exactly this question and argues that this change can be viewed as grammaticalization “down the tree” from a high D head to a low n head. Furthermore, it argues that functional categories, like the definiteness category, are non-universal. That is, they are not part of Universal Grammar, but only arise when the child discovers them in the input. The paper also addresses some movement puzzles emerging in Old Norse and Modern Icelandic which have remained a theoretical puzzle. I will propose an analysis of this where I argue that we need to separate Modern Icelandic and Old Norse and thus give two separate analyses.

. Introduction1 There is an interesting difference between Old Norse and Modern Norwegian concerning the structure of the DP and the realization of definiteness.2 In ON, the definite article was a clitic whereas in MN it is a suffix. The clitic in ON developed

. Thanks to the audience at Grammar in Focus in Lund 2007, and to the following persons for valuable feedback and comments: Werner Abraham, Cedric Boeckx, Elly van Gelderen, Helge Lødrup, Kjartan Ottósson and Halldór Ármann Sigurðsson. Special thanks to Jan Terje Faarlund for most valuable comments and discussion of these matters, and to the editors for wanting to include this piece in the present volume. . The following abbreviations are used in the text:  = definiteness;  = feminine gender;  = masculine gender;  = Modern Icelandic;  = Modern Norwegian;  = nominative;  = Old Norse;  = plural;  = singular.

S F O PRO



1st proofs

 Terje Lohndal

from a demonstrative, and the difference between ON and MN is illustrated in (1a)–(1d).3 (1)

a.

b.

c.

d.

sá inn gamli hestr that the old horse ‘the old horse’ hestrinn horse. ‘the horse’ den gamle hesten that old horse. ‘the old horse’ hesten horse. ‘the horse’

We see that the definiteness marker inn in ON could either be attached to the noun like in (1b), or stand on its own like in (1a). The reason for the latter possibility is that the article originally developed from the demonstrative pronoun hinn ‘that’. In MN, the definiteness marker needs to be attached to the noun itself, cf. (1c) and (1d). In addition, a new phenomenon has developed in Norwegian and Swedish, that is, the use of double definiteness. ON did have instances of this use, but the phenomenon is much more pervasive in MN. Perhaps not surprisingly, Modern Icelandic (MI) does not have double definiteness either (Sigurðsson 1993, 2006: 200, 205–206).4 The question we face is how to account for this change. It will be argued below that the definite article was a clitic in ON, which became grammaticalized into a suffix in MN.5 Since the same feature is expressed by different means in the two languages, there is reason to believe that this correspond to a difference in

. The Modern Norwegian examples in this paper are given in Nynorsk, which is one of the two written standards of the language. In the cases at hand, the choice of written standard makes no difference to the argumentation. . I say surprising, because there is a very intimate relationship between ON and MI. ON is usually used as a cover term for both Old Norwegian and Old Icelandic, which shows us that originally MN and MI were one common language. I have chosen the neutral term to avoid entering a discussion regarding the choice of term, as the latter also is very much a political choice (see e.g. Faarlund 1990: 10–11; Haugan 2000: 4–8). The important point to bear in mind is this close relationship between ON and MI. . In this paper I will only focus on what happens internal to the DP. Leiss (2000) argues that the article/determiner develops in the rhematic position as soon as aspect disappeared. Unfortunately, a careful study on the relationship between Leiss’s proposal and the one developed here will have to remain for future research.

S F O PRO



1st proofs

Nominal phrases in Norwegian 

syntactic structure. Such a stand raises several questions, which touches on very large and disputed subjects within generative grammar. Let me briefly mention one of these issues, namely the universality of functional projections. Some people believe that syntactic structure is entirely uniform across all languages. Every language has the same functional projections as any other language, though each functional projection does not have to be overtly filled in each language, perhaps being guided by a parameter or something similar to that. This is essentially the approach taken by Cinque (1999, 2005) and his followers. Cinque (1999: 106) says that we find the following universal sequence of functional heads in relation to adverbials: (2)

[frankly Moodspeech act [luckily Moodevaluative [alledgely Moodevidential [probably Modepistemic [once T(Past) [then T(Future) [perhaps Moodirrealis [necessarily Modnecessity [possibly Modpossibility [usually Asphabitual [again Asprepetitive(I) [often Aspfrequentative(I) [intentionally Modvolitional [quickly Aspcelerative(I) [already T(Anterior) [no longer Aspterminative [still Aspcontinuative [always Aspperfect(?) [just Aspretrospective [soon Aspproximative [briefly Aspdurative [characteristically(?) Aspgeneric/progressive [almost Aspprospective [completely AspSgCompletive(I) [tutto AspPlCompletive [well Voice [fast/early Aspcelerative(II) [again Asprepetetive(II) [often Aspfrequentiative(II) [ completely AspSgCompletive(II) ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]

A radically different hypothesis is the one put forward by Bobaljik & Thráinsson (1998) and Thráinsson (1996, 2003) (see also Bobaljik 1995; Iatridou 1990; Ouhalla 1991, and van Gelderen 1993) who argue that we only find those functional projections that the child has evidence for. Hence there is no universal structure across all languages, and children are not born pre-wired with the entire sequence of functional projections. This question is actually quite important for our present purposes, so I will return to it below after the data have been presented and discussed. I will argue that the data do not point to a clear conclusion, but that theory internal considerations provide a rather straightforward conclusion. Before moving on, a short comment on the status of head movement is in order as it is commonly claimed that head movement occurs frequently within nominal phrases (e.g. N-to-D movement). Recently, the traditional head movement analysis (e.g. Baker 1988) qua adjunction has come under heavy attack by Boeckx & Stjepanović (2001) and Chomsky (2001) who claim that head movement is a purely phonological process. Since I will use the framework in Julien (2005) as my point of departure, I will nevertheless continue to assume that head movement belongs to narrow syntax. This choice is also further corroborated by various researchers claiming, contra Chomsky (2001), that head movement has semantic effects (Lechner 2005; Mohr 2004; Roberts 2005, 2006). The question is tangential to our present concerns, so I will leave the matter here.

S F O PRO



1st proofs

 Terje Lohndal

The paper is structured as follows. Section 2 outlines and discusses the phrase structural representation of nominal phrases in MN and ON. Section 3 contains a discussion of some movement differences between ON and MI, and provides a new account of the differences between these two languages. In Section 4, an account of the change is provided, where it is argued that functional structure is non-universal and that the development of double definiteness in ON and MN is an instance of “downward” grammaticalization. Section 5 concludes the paper. .

The DP in Old Norse and Modern Norwegian

In this section, I will give an overview of the common DPs in ON and in MN. I shall also compare the structure proposed for Old Norse by Faarlund (2004) and the one for Mainland Scandinavian by Julien (2005). These works are the most comprehensive studies of the DP in ON and MN respectively. I will show that the structure proposed by Faarlund (2004) needs to be enriched towards the structure proposed by Julien (2005), but that ON and MN differ with respect to whether there is a low definite phrase or not. The main difference between ON and MN was presented above in (1a)–(1d), and the data are repeated here for expository convenience. (3)

a.

b.

c.

d.

sá inn gamli hestr that the old horse ‘the old horse’ hestrinn horse. ‘the horse’ den gamle hesten that old horse. ‘the old horse’ hesten horse. ‘the horse’

Double definiteness was not very common in ON, although some examples exist. Note that they are from the latter part of the ON period (approximately 1270– 1330). A couple of examples are provided in (4): (4)

a.

b.

ins versta hlutarins the worst part. ‘of the worst part’ hinir beztu menninir the best men. ‘the best men’

(Faarlund 2004: 58)

S F O PRO

(Faarlund 2004: 58)



1st proofs

Nominal phrases in Norwegian 

Lundeby (1965) is an extensive study of the development of double definiteness in Norwegian and related languages. Lundeby demonstrates its development through the texts and shows how it gains frequency. As is well known, ProtoGermanic and Proto-Nordic did not have any definite article, thus no double definiteness, so it must have developed during the ON period. Two other studies have also been concerned with the development of the definite article in Norwegian. Both Dyvik (1979) and Holm (2002) approach the issue from a semantic point of view. They elaborate on the foundation created by Lundeby (1965) but do also develop their own hypothesis. Since they are mostly concerned with how we got the article in the first place and what the semantic contribution amounts to, I will not discuss these studies any further at this point. The interested reader is referred to their work. It has been argued convincingly that the definite article in ON was a clitic (Faarlund 2007, cf. also Faarlund 2004). The two most important reasons for this are that clitics may have free word counterparts, whereas affixes do not, and clitics may occur outside affixes. Both of these conditions are born out. See Faarlund (2007) for a discussion of other relevant criteria and for evidence that they confirm the clitic status. In order to get a better understanding of the structure of the DP in ON, let us look at some other different DPs in addition to those already mentioned above. (5) contains structures where the article is non-cliticized whereas (6) shows some typical uses of the cliticized article. (5)

a.

b.

c.

(6)

a.

b.

c.

it fyrra sumar the former summer ‘last summer/the summer before’ (Faarlund 2004: 56) en þó var hann hit mesta afarmenni but still was he the greatest outstanding-man ‘But still he was a most outstanding man’ (Faarlund 2004: 56) þitt hitt milda andlit your the mild face ‘your mild face’ (Faarlund 2004: 60) stýrimaðr fyrir skipinu helmsman for ship. ‘helmsman of the ship’ ok ekki þykkir fjándanum jamillt and nothing seems devil. even-bad ‘and nothing seems equally bad to the devil’ þat er eigi at réttu mannsins óðal that is not at right man. legacy ‘That is surely not the purpose of mankind’

(Faarlund 2004: 58)

(Faarlund 2004: 58)

S F O PRO

(Faarlund 2004: 58)



1st proofs

 Terje Lohndal

There is nothing particularly special about these DPs. However, there are some structures that better illustrate some of the variation that we find: (7)

hafit þat it djúpa sea that the deep ‘the deep sea’ b. fé þat allt money that all ‘all that money’ c. ??sá hestrinn gamli that horse. old ‘that the old horse’

a.

(Gordon 1957: 312)

(Faarlund 2004: 84)

(7a) and (7b) show that the noun can move above the article and the demonstrative. The question is, however, whether a structure such as (7c) was allowed or not. It seems not to be mentioned in the literature (e.g. not in Faarlund 2004 or Nygaard 1906, the most comprehensive overviews we have of the data), which it most likely would be if it was encountered. Interestingly, (7c) does not exist in MI either (Kjartan Ottósson, p.c.), though the conditions which make it ungrammatical appears to be slightly different from the ones applying in ON. I will return to this in Section 3, where I also discuss more closely whether it is likely to assume that (7c) is ungrammatical or not. Faarlund (2004: 83) proposes the following structure for the ON DP: (8)

DP D' D

RP R' R

NP N' AP

N' N

RP stands for a Referential Phrase, and as such resembles what Julien has decomposed into αP and nP (cf. below). Faarlund also suggests that the noun is attached to the article by way of movement from N to R (Faarlund 2004: 57). Thus the sentence in (9) will get the structural representation in (10) (Faarlund 2004: 82–83).

S F O PRO



1st proofs

Nominal phrases in Norwegian 

(9)

(10)

þau in stóru skip those the big ships ‘those big ships’ DP D' D Þau

RP R' R in

NP N' AP stóru

N' N skip

For the cliticized article, he provides the structure in (12) for the sentence in (11) (Faarlund 2004: 82–83): (11)

(12)

sá Ärninn that eagle. ‘that eagle’ DP D' D sá

RP R' R Ärni+inn

NP N ti

This structure evidently has its merits, as it seemingly accounts for the structure of the ON DP. Let us now look at the structure of the MN DP to see the differences. Julien (2005) is a major investigation of Scandinavian nominal phrases from a contemporary perspective, building on important previous studies of Norwegian and Mainland Scandinavian (Taraldsen 1990; Delsing 1993; Kester 1993; Santelman 1993; Sandström & Holmberg 1994; Vangsnes 1999; Zamparelli 2000; Vangsnes,

S F O PRO



1st proofs

 Terje Lohndal

Holmberg & Delsing 2003; Svenonius 2006). Recently, Anderssen (2005) has also developed Julien’s structure somewhat further and related it to the acquisition of Norwegian DPs. Julien (2005) proposes the following maximally expanded structure (Julien 2005: 281): (13)

[DP D [PossP Poss [CardP Card [αP α [nP n [NumP Num [NP N]]]]]]]

To make this more explicit, consider the structure in (15) for the Modern Norwegian DP in (14) (Julien 2005: 11). Note that she adopts a very strong formulation of the nonlexical approach to morphology (cf. Baker 1988; Marantz 1997; Cinque 1999; Julien 2002), which appears to be supported by the findings in Anderssen (2005). I will assume this to be correct without further justification. (14)

dei to gaml-e teikning-a-ne mine av by-en . two- old drawing-- my. of town-.. 6 ‘my two old drawings of the town’

(15)

DP D dei

CardP Card'

WQP to

αP

Card

α'

AP gamle α

nP n

Num

teikn

n ne

Num a

N

NumP Num

NP N'

mine N

N ing

PP av byen

Concerning double definiteness, Julien (2005: 28) assumes that a Norwegian phrase like (16) has the structure in (17). . This sentence is glossed according to the glosses in Julien.  = weak inflection. I assume the other glosses to be self evident. In general, I will not detail the glosses to such a degree that Julien does, but simplify them according to the purposes of this paper. This also applies to all the data taken from Faarlund (2004), where glosses are simplified as much as possible in order to focus on the issue at hand.

S F O PRO



1st proofs

Nominal phrases in Norwegian 

(16)

skjort-a shirt-.. ‘the shirt’

(17)

DP nP n Num N skjorta

D' NumP D

n Num a

Num

nP

NP N

In other words, nP moves to SpecDP, and she argues that D agrees with n. As Julien points out, this is in accordance with the Agree system of Chomsky (2000, 2001, 2004, 2005, 2006) where Move can be part of Agree. She further takes NP, NumP, nP and DP to be present in every Scandinavian DP. Note that she also derives the impossibility of nP moving to SpecDP in cases where an AP intervenes, cf. (18). (18) *teikningane gamle drawings. old

An AP that is merged in SpecαP will agree with α, which in turn agrees with n. This means that when an AP is present, this AP will be a closer goal for D, and thus we derive the impossibility of nP moving above AP (Julien 2005: 29). This could probably also have been derived along the lines of relativized minimality (Rizzi 1990; Starke 2001), though I will not pursue that alternative here. Julien argues as mentioned that nP is present in every Scandinavian DP. This means that not every language can spell this head out overtly. Danish is one such instance, as it only has a pronominal article (adapted from Julien 2005: 65): (19)

a.

b.

et stort hus a big house ‘a big house’ det store hus(*-et) the big house (*) ‘the big house’

Interesting support in favor of her argument that n is present covertly, is the following sentence (Julien 2005: 66): (20)

Ved du ikke det, stor-e pige! know you not that big- girl ‘Don’t you know that you big girl!’

S F O PRO



1st proofs

 Terje Lohndal

Here, the nominal phrase store pige ‘big girl’ is vocative and definitely definite as the adjective has weak inflection: “Consequently, there must be a definite n head in this phrase, although it is not spelled out” (Julien 2005: 66). This is interesting also with respect to ON, as ON is similar to Danish in this respect (and so is MN with respect to this particular structure). In fact, we do find such phrases as (21a) and (21b) in ON: (21)

a.

b.

til þess helg-a húss to that sacred- house ‘to that sacred house’ sam-a haust same- autumn ‘the same autumn’

(Faarlund 2004: 68)

(Faarlund 2004: 68)

In (21a) there is a demonstrative, which we can interpret as making the NP definite. However, in (21b) there is no demonstrative, though it is common in ON to use the definite form with comparatives and superlatives. Although these examples do not mirror (20) exactly, they indicate that we have a definite interpretation that cannot be due to the presence of a definite article. Note, however, that the examples of this kind quoted in Faarlund (2004) are fairly old, thus from the latter part of the ON period (approximately 1200–1330, judging from the age of the manuscripts). I take this to mean that at this stage, n is probably present due to the development of double definiteness and the ongoing change of the definite article. Given the fact that we also find instances of double definiteness in ON, cf. (4), repeated here as (22), we can conclude that n develops already during the ON period. (22)

a.

b.

ins versta hlutarins the worst part. ‘of the worst part’ hinir beztu menninir the best men. ‘the best men’

(Faarlund 2004: 58)

(Faarlund 2004: 58)

However, since we know that double definiteness develops during this period of time (Lundeby 1965; Dyvik 1979; Holm 2002), and that we seldom find instances of double definiteness in the early writings, it is most likely to assume that n was not present in Proto-Nordic. I will return to a formalization of this change in Section 4. We have now seen accounts of the ON nominal phrase and the MN nominal phrase, in addition to some considerations of the Modern Scandinavian varieties. Obviously, the structure proposed for ON in (12) and for MN in (15) are very different when it comes to complexity. The structures are repeated here as (23) and (24).

S F O PRO



1st proofs

Nominal phrases in Norwegian 

(23)

[DP D [RP R [NP N]]]

ON

(24)

[DP D [PossP Poss [CardP Card [αP α [nP n [NumP Num [NP N]]]]]]]

MN

We have to bear in mind that Faarlund (2004) and Julien (2005) work within quite different assumptions. Whereas Julien adopts a very strong formulation of the mirror principle (Baker 1988), Faarlund does not assume any such principle. Thus it is not that easy to compare the structures in (23) and (24). Evidently, the lower part of Julien’s structure, from nP and downwards, is clearly a reflex of assuming the mirror principle.7 ON nouns inflect for number, as is illustrated in (25).8 (25)

a. b.

hest-r horse-. hest-ar horse-.

Thus the question is more pertinent for the phrases above nP. However, looking at the structures discussed in Faarlund’s monograph, we do find clear indications favoring (24) instead of (23). The following postverbal nominal phrase contains a CardP, a αP, a NumP and a NP (and possibly a covert nP, cf. the discussion above). (26)

ok því næst fann han einn gamlan munk. and that next met he one old monk ‘And next he met some old monk.’

(Faarlund 2004: 74)

Here einn ‘one’ is the CardP and gamlan ‘old’ is the αP. In (27) we find a PossP: (27)

var þeira dóttir Húngerðr. was their daughter Hungerd ‘Their daughter was Hungerd.’

(Faarlund 2004: 60)

The structures in (26) and (27) then serve as evidence for the child growing its ON grammar, and the child will thus encounter the rich structure in (24) and not the structure in (23). One important difference between ON and MN to bear in mind is that ON has a clitic article in D (Faarlund 2007) instead of a suffix article in n. In Section 4, I will propose an analysis of this difference between ON and MN. Before turning to that, let us examine some patterns of movement within the nominal phrases in ON and MI.

. Julien argues that n is a phase head (Julien 2005: 4, 73), but since the N moves to this head it is within the phase edge and can as such move on to D (Chomsky 2000, 2001). . Due to the fact that the ON morphology is heavily flexional, we face problems when it comes to segmentation. I will not address these questions here, but see the debate between Enger (1993, 1997) and Papazian (1995).

S F O PRO



1st proofs

 Terje Lohndal

.

Movement in Old Norse and Modern Icelandic nominal phrases

Sigurðsson (1993, 2006) discusses the structure of the MI noun phrase. Interestingly, and perhaps not surprisingly, the noun phrases in MI exhibit some of the same properties as the ones on ON. However, it is possible to discuss the issues in far more depth concerning MI as it is a living language and one can rely on judgments by native speakers. Here I will try to deal with the movement issue that (7c) poses, repeated here for expository convenience as (28). My reason for doing so is that it can tell us something about the structure that Julien has proposed and its validity. As will be argued, her structure captures the differences between ON and MI. (28) ??sá hestrinn gamli that horse. old ‘that the old horse’

MI clearly disallows (28), and perhaps also ON, but apparently for different reasons. For ON, the reason is maybe that the presence of the demonstrative pronoun sá ‘that’ and the adjective gamli ‘old’ blocks movement, whereas for MI it is the adjective only that blocks movement. That would explain the grammaticality of the following two sentences in ON: (29)

a.

b.

á Orminum langa on Serpent. long. ‘on board “The Long Serpent” ’ (Faarlund 2004: 71) í eilífri dýrð fÄður ok sonar ok andans helga in eternal glory father and son and spirit. holy. ‘in the eternal glory of the Father, the Son, and the holy Spirit’ (Faarlund 2004: 71)

However, as Faarlund (2004: 71) says, these are less common. The common pattern in ON is the following: (30)

a.

b.

inn mesti vinr Erlings the greatest friend Erling ‘a very good friend of Erling’s’ (Faarlund 2004: 70) Þjódólfr inn fróði ór Hvini var skáld Haralds ins hárfagra Thjodolf the learned. of Hvin was poet Harald the hair-fine. ‘Thjodolf the learned of Hvini was Harald the Fine-haired’s poet’ (Faarlund 2004: 70)

We see that the common way of using the definite article is to let it stand alone in D (30a). When we have proper names, as in (30b), we often get movement. Faarlund says that “this is presumably movement to the specifier position of R, which prevents

S F O PRO



1st proofs

Nominal phrases in Norwegian 

cliticization of the article” (Faarlund 2004: 70). Hence an entire phrase may precede the article: (31)

Hákonar jarls ins ríka Hakon earl the mighty. ‘of Earl Hakon the mighty’

(Faarlund 2004: 70)

Let us now turn to the MI nominal phrase. One difference between ON and MI is that the definite article is a suffix and not a clitic (though see Sadock 1991: 113–116 for a different view).9 In other words, we find the same situation as in MN. However, one can find a variant which resembles ON, though this variant has a more literary flavor (Sigurðsson 1993: 180). The general pattern in MI can thus be illustrated as follows (Sigurðsson 1993: 179): (32)

Hið fræga verk. the famous work b. *Hið fræga verkið. the famous work. c. Fræga verkið. famous work. a.

(32a) is the one closest to ON, while (32c) is the common one today. As we see in (32b), MI does not have double definiteness. When it comes to different movement patterns, MI is rather similar to ON. Consider the following structures: (33)

a. ?Alla þrjár greingarnar eru réttar.10 all three analyses. are correct ‘All three analyses are correct.’ b. Allar [greiningar]nar þrjár [__] eru réttar. all analyses. three are correct ‘All three analyses are correct.’ c. [Greiningar]nar þrjár [__] eru réttar. analyses. three are correct ‘The three analyses are correct.’

(Sigurðsson 2006: 204)



. One argument in favor of this is that the definite article in MI must appear suffixed to the noun (Sadock 1991: 113): i.

a.

hesturinn horse. ‘the horse’ b. *hinn hestur

S F O PRO

. (Sigurðsson 2006: 204) comments in a footnote that “this word order is degraded for me, but is accepted by some speakers (see Vangsnes 2004)”.



1st proofs

 Terje Lohndal

d.

Allir bílarnir þínir þrír [__] eru rauðir. all cars. your three are red ‘All your three cars are red.’


We see that it is unproblematic to move above the numeral in CardP (33b–33d). However, the following movement in (34b) is illicit (Halldór Sigurðsson, p.c.): (34)

Allar gömlu greiningar þrjár [__] eru réttar. all old. analyses. three are correct ‘All three old analyses are correct.’ b. *Allar greiningar gömlu [__] eru réttar. all analyses. old. are correct ‘All old analyses are correct.’ a.

(34b) shows that it is not possible to move above an adjective. We do find some instances where movement appears to be allowed, cf. (35a)–(35b), but it is mostly disallowed (35c) (Halldór Sigurðsson, p.c.). # indicates that the sentences are highly marked. (35)

a. #stúlkan fagra. girl. beautiful. ‘the beautiful girl’ b. #dagurinn bjarti day. bright. ‘the bright day’ c. *bókin nýja. book. new. ‘the new book’

Halldór Sigurðsson (p.c.) informs me that the instances in (35a)–(35b) are lexically or stylistically limited. First, they sound archaic and second they have a very limited use. This is probably also the case for ON. So a valid conclusion appears to be that it actually is the adjective that blocks movement in MI. The question is then how we can give an analysis of this phenomenon for both MI and ON. We noticed above that in Julien’s (2005) structure the AP block movement of nP as AP is a closer goal than nP. This appears to give the right predictions for MI, but the wrong ones for ON when it comes to proper names and movement of phrases which do not undergo cliticization.11 The relevant examples are repeated here for expository convenience in (36). Adjectives are put in boldface. . There is an exception when it comes to proper names in Icelandic. They preferably undergo movement; cf. (ia)–(ib). This is probably due to some feature which makes it a more attractive goal and hence the probe looks beyond AP. However, that proposal is rather tentative, and deserves to be worked out more completely.

S F O PRO



1st proofs

Nominal phrases in Norwegian 

(36)

a.

b.

Hákonar jarls ins ríka Hakon earl the mighty. ‘of Eral Hakon the mighty’ (Faarlund 2004: 70) Þjódólfr inn fróði ór Hvini var skáld Haralds ins hárfagra Thjodolf the learned. of Hvin was poet Harald the hair-fine. ‘Thjodolf the learned of Hvini was Harald the Fine-haired’s poet’ (Faarlund 2004: 70)

The problem is how one should explain these cases if the adjective count as an intervener for this apparently restricted class of nouns. Note, however, that all cases like these involve a definite adjective. Sadly, this generalization does not give us anything, as a proper name may precede an indefinite adjective as well: (37)

Hákon herðibreiðr. Hakon broad.shouldered ‘Hakon the broad-shouldered.’

(Faarlund 2004: 71)

To make things even more complicated, even common names can precede the independent article (cf. also (29a)): (38)

engi maðr mátti nefna hann annan veg, en jarl inn illa. no man could mention him other way than earl the evil. ‘Nobody was allowed to refer to him in any other way than as “the evil earl.” ’ (Faarlund 2004: 70)

What should be clear from these examples is that it is unfeasible to give a unified account of the possible movements in both MI and ON DPs. In MI, adjectives appear to block movement in most cases, whereas in ON this blocking seems to be conditioned by the presence or non-presence of a demonstrative together with an adjective.12 The adjective itself does not count as an intervener in ON (36), and cliticization is also not probihibited when just a demonstrative is present (39) (repeated from (11)).

i.

a.

[DP Anna litla] fékk dúkku. Anna little got doll ‘Little Anna got a doll.’ b. ?[DP Litla Anna] fékk dúkku. little Anna got doll

(Julien 2005: [19])

‘Little Anna got a doll.’

(Julien 2005: [19])

. Since we have cliticization and not suffixation, we have head movement in ON. Thus an AP does not count as an intervener since it is a phrase and not a head. Thus this appears to favor a relativized minimality account of the data.

S F O PRO



1st proofs

 Terje Lohndal

(39)

sá Ärninn that eagle. ‘that eagle’

Thus it seems rather unlikely that the structure sá hestrinn gamli ‘that the old horse’ (25) is ungrammatical, since it is merely a combination of the two alternatives. Due to the lack of native speaker judgments, this is however impossible to confirm. Despite the fact that neither Faarlund (2004) nor Nygaard (1906) have listed this as a possible structure, I think the above data show that we have every reason to assume the structure to be grammatical. If that is the case, then the adjective (αP) does not count as the closest goal in MN, but it does in MI. An important point to bear in mind here is that I have suggested that the n is not present in ON (see also next section). Julien (2005: 12) assumes that every functional element introduced into the structure will agree with the elements that are already there (cf. Sigurðsson 2004a). Since there is no n, there is no definiteness feature for α to agree with. Hence there is no such feature that makes α(P) a closer goal than Num(P), and thus Num(P) can move to D(P). From this we can conclude that the αP is predicted not to be an intervener in ON, and as such derive the grammaticality of the examples above. Notice that it also derives the MI cases: Since n is present in MI, movement crossing an adjective is predicted to be impossible. Thus Julien’s structure appears to be rich enough to capture the essentials of the ON, MN, and MI nominal phrase, and thus suit as a tool for accounting for definite article’s change from a clitic to a suffix. I turn to that in the next section. .

An account of the change

We have seen examples of how the structure of nominal phrases was both in ON and MN. At the end of the previous section, it was shown that we should assume a quite rich structure for ON, too. It has also been established that the definite article changed from a clitic in ON to a definite suffix in MN. One question that we have not answered is how exactly we should represent this change in phrase structural terms. In this section I shall provide an answer to that question, which also touches upon larger issues having to do with how we understand the role of functional categories within Universal Grammar. In the introduction, we briefly mentioned the two opposing views within generative grammar concerning the universality of functional projections. Let us elaborate somewhat on these. The main proponent of the view which says that all functional projections are present in all languages, is Cinque (1999) (see also Sigurðsson 2004b, Borer 2005). He argues that there is a universal structure within the functional domain (see Section 1). Cinque points to a relevant point concerning

S F O PRO



1st proofs

Nominal phrases in Norwegian 

the opposite view, the view that functional projections exist only in the presence of overt morphological material: Since, in this case, (most) adverbs would not be systematically related to a functional head, UG would have to countenance two distinct conditions (one ruling over the hierarchy of heads, the other over the hierarchy of AdvPs), basically yielding (duplicating) the same information on the relative scope of what are essentially identical functional notions (Cinque 1999: 107).

He also points to the problem of acquisition, and says that the child will have less left to acquire on his hypothesis than on the opposite view. This view has been supported my much research into language acquisition, e.g. Poeppel & Wexler (1993); Harris & Wexler (1996); Guasti (1993/1994); Hoekstra & Hyams (1998); Hyams (1996); Schütze (1997); Wexler (1998); Borer & Rohrbacher (2003). One especially debated issue is the one concerning root infinitives (also known as optional infinitives); those cases where the child omits inflecting the verb (see Radford 1990 for English and Wexler 1994 for other languages). I will, however, not have the opportunity to review the debate here. Thráinsson (1996, 2003) has argued in favor of the other main position, and suggested what he calls “The Real Minimalist Principle”, stated in (40) (Thráinsson 1996: 261). (40)

Assume only those functional categories that you have evidence for

Theoretically this principle has been defended by various people, e.g. Fukui (1986, 1995, 2005); Bobaljik & Thráinsson (1998); Thráinsson (1996, 2003); van Gelderen (1993); Vangsnes (1999); Fukui & Sakai (2003). It has also been argued on the basis of acquisition studies (Lebeaux 1988, Radford 1990, 1996, 2000). A possible conjecture as to how this happens is provided by Thráinsson (1996: 260): The Functional Projection Alibi Let F be a functional head and Φ the set of ϕ-features (and/or other features) which F contains. Now assume that the different values for the ϕ-features in Φ are systematically represented in overt morphology in language L in clauses of type A but never in clauses of type B. Assume further that there is evidence for overt syntactic movement to F and/or to Spec-FP in language L in clauses of type A but no such evidence can be found in clauses of type B. Then we may hypothesize that F is absent in clauses of type B in language L although it is present in clauses of type A.

His main goal is to account for the variation we find regarding verb movement between the Mainland Scandinavian languages the insular languages like MI and Faroese. Vangsnes (1999) takes a somewhat similar approach as Thráinsson

S F O PRO



1st proofs

 Terje Lohndal

(1996, 2003) does. He argues in favor of a principle of identification, which he defines as follows: A functional category must be identified by having a constituent containing one or more relevant morphological features either in its specifier or head position. The constituent must be merged within the extended projection of which the functional category is a part (Vangsnes 1999: 4, 47).

Vangsnes assumes that “all functional categories are headed by abstract heads” (p. 47), and that the fixed order within a noun phrase is semantically ordered. This ordering may very well be universal, and thus both the inventory of functional categories and their relative ordering is fixed (Vangsnes 1999: 47). This latter argument shows that it is possible to assume that functional categories have a universal ordering, but that they must be identified in each single language. A third alternative might to say that the Initial State involves some functional structure, but as Anderssen (2005: 168–169) points out, “the task of determining which projections should be considered universal is complicated”. Such a stand seems to be doomed to fail as it appears to be impossible to determine what the core functional categories would be and what the more peripheral categories would be. On purely conceptual grounds it seems quite unlikely that we have more or less hundred functional projections between the verb projection and the complementizer, and that we only use some of them.13 If true, that means that we need to attribute very much to things we know rather little about or at least have a hard time confirming. It is not implausible at any rate that we come pre-equipped with some structure, e.g. a clausal spine including the main projections (whatever they turn out to be), but it is hardly likely that we have all structure present. The principle (48) takes the opposite stand: we only have evidence for what we discover. However, it would be nice if the data itself could have anything to say about this. Anderssen (2005: 149) says that “the possibility that there is no functional structure present cannot be excluded based on child language data”. That seems to be correct given for instance the large debate and disagreement concerning root/optional infinitives. Still, perhaps some data from the languages themselves can tell us something. In addition, theoretical points of views should also be considered. In the following, I will consider one example of each using the material above. Hopefully this can also shed some light on the nature of the change we have discussed.

. As Newmeyer (1998: 364) puts it, “there is no logical necessity for the set of parameters of variation to be innate; one can imagine the possibility of a theory in every respect like recent principles-and-parameters models, but in which the parameters are arrived at inductively by the child”.

S F O PRO



1st proofs

Nominal phrases in Norwegian 

Above we discussed evidence in favor of a complex nominal phrase in ON. One example that we discusses was the one where a structure has definite meaning but indefinite syntax. The two relevant examples are repeated here as (41). (41)

a.

b.

Ved du ikke det, stor-e pige! know you not that big- girl ‘Don’t you know that you big girl!’ af fyrr-a konungi of former- king ‘of the former king’

(Julien 2005: 66)

(Faarlund 2004: 68)

We took these structures to indicate the presence of nP. This was further supported by the fact that ON also developed double definiteness. When this change happened, the child must somehow get a cue as to where in the phrase structure this second marker of definiteness should appear. If we take a universal structure of the Cinque kind to be the correct answer, then the solution appears to be simple: If the structure of the DP has a universal structure, a second marker of definiteness has its natural place, namely low in the structure. However, this argument is probably too weak. First, languages vary with respect to the ordering within the DP layer (Cinque 2005), so there is not any reason for expecting there to exist a universal ordering. Second, there is no a priori reason why syntax itself is not sufficient, that is, why syntax itself does not suffice as a trigger. In fact, given the variation found, we are lead to expect that syntax needs to play some role in order to determine the final landing place in cases of movement. So far both approaches seem to be compatible with the data. Let us therefore turn to a more important theoretical argument. In an interesting paper, Abels & Neeleman (2006) argue that assuming universal base structures gives us a non-restrictive theory of movement. The latter is of course a bad result for minimalism and its inclination towards restrictiveness. Abels & Neeleman discuss Cinque’s (2005) paper in particular and put forward another suggestion as to how we can maintain the insights of Greenberg’s Universal 20 without lowering the standards for a theoretically restrictive theory. They argue in favor of several base structures and against the view that there only exists one single base from where every movement departs. Given that they are correct, this makes a strong case for the view that we only assume what is present. If there is some kind of universal structure, this probably needs a universal ordering too (Cinque 1999, 2005; see also Vangsnes 1999). Where this universal ordering comes from, remains less clear. I think this theoretical argument lends support to Thráinsson’s argument, i.e. that phrase structure develops according to presence (either semantically or overtly) in the input. Thus my answer to the development of n during the ON period is that the n emerged in the phrase structure. Why it occurred at its specific

S F O PRO



1st proofs

 Terje Lohndal

place must have been determined by the syntactic input, by itself there is not any reason why definiteness should be so low in the structure, and in particular why it resides between αP and NumP. This change can be accounted for by assuming grammaticalization “down” the tree, as the article goes from D to n, that is, from being a clitic to being an inflectional affix (as in the traditional literature, represented by the grammaticalization cline as presented in e.g. Hopper and Traugott 2003: 7; see also Faarlund 2007). It has been argued convincingly by van Gelderen (2004a, 2004b) that grammaticalization, as the result of economy principles (see also Roberts & Roussou 2003), overwhelmingly goes “up” the tree.14 An example of this is the preposition for in Old English developing into a complementizer in Middle English; compare (42a) and (42b) (van Gelderen 2004b: 30): (42)

a.

b.

þæt he for eaxlum gestod. that he before shoulders stepped ‘that he stood in front of…’ Locrin 7 Camber to þon scipen comen. for to habben al þa æhte. Locrin and Chamber to the ships came for to have all the goods ‘Locrin and Camber came to the ships to take all the goods.’

However, the development of definiteness suggests very clearly that the article grammaticalized downwards, going from a clitic to an inflectional affix. If the present analysis is on the right track, it would be an interesting example entailing that grammaticalization can proceed both upwards and downwards. Van Gelderen’s (2004b) account is guided by what she dubs “The Late Merge principle”, which says “Merge as late as possible” (van Gelderen 2004b: 28). Note, however, that this instance of upward grammaticalization does not go against her principle. Recall that this change happened during a time at which double definiteness developed, and as such a need to express definiteness twice within the nominal phrase developed. When a definite pre-nominal article occurs, this blocks the movement of the post-nominal article, hence it is impossible for the latter to move further up the tree. Thus no conflict arises, and the post-nominal article has merged as late as possible. To conclude, we have seen that the empirical data does not in itself provide any clues as to whether the functional structure is universal or not (as mentioned

. There has been a huge amount of literature on grammaticalization recently, and, obviously, grammaticalization from a functional and a formal point of view is two different things. I will not discuss the concept of grammaticalization more closely here, but in addition to the cited work in the main text, see Lohndal (2006) and in particular Faarlund (in press) for a compelling argument that grammaticalization is not an independent process, but should be studied from the point of view of the child growing its language.

S F O PRO



1st proofs

Nominal phrases in Norwegian 

by Anderssen 2005 too). I have, however, argued that on theoretical grounds we ought to prefer the developmental story. This is also in line with current minimalist tenets, where theoretical machinery is sought to be kept at a minimum.

.

Conclusion

In this article I have argued that double definiteness developed during ON as an instance of “downward” grammaticalization of the definite article. This entails the presence of a low definiteness head, and I have argued that this head developed alongside double definiteness. I have also discussed some movement puzzles in ON and MI, arguing that the two languages require different analyses. The MI cases appear to be straightforward handled as an instance of αP blocking nP, whereas I suggested that this blocking does not occur in ON.

References Abels, Klaus & Neeleman, Ad. 2006. Universal 20 without the LCA. Ms., University of Tromsø and University College London. Anderssen, Merete. 2005. The Acquisition of Compositional Definiteness in Norwegian. PhD Dissertation, University of Tromsø. Baker, Mark C. 1988. Incorporation: A theory of grammatical function changing. Chicago IL: The University of Chicago Press. Boeckx, Cedric & Stjepanović, Sandra 2001. Head-ing toward PF. Linguistic Inquiry 32: 345–355. Bobaljik, Jonathan David. 1995. Morphosyntax: The syntax of verbal inflection. Doctoral dissertation, MIT. Bobaljik, Jonathan David & Thráinsson, Höskuldur. 1998. Two heads aren’t always better than one. Syntax 1: 37–71. Borer, Hagit. 2005. Structuring Sense: In name only, Vol. I. Oxford: Oxford University Press. Borer, Hagit & Rohrbacher, Bernhard. 2003. Minding the absent: Arguments for the full competence hypothesis. Language Acquisition 10: 123–175. Chomsky, Noam. 2000. Minimalist inquires. In Step by Step. Essays on minimalist syntax in honor of Howard Lasnik, Roger Martin, David Michaels & Juan Uriagereka (eds), 89–155. Cambridge MA: The MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: A lifein language, Michael Kenstowicz (ed.), 1–52. Cambridge MA: The MIT Press. Chomsky, Noam. 2004. Beyond explanatory adequacy. In Structures and Beyond – The cartography of syntactic structure, Vol. 3, Adriana Belletti (ed.), 104–131. Oxford: Oxford University Press. Chomsky, Noam. 2005. On phases. Ms., MIT. (To appear in Robert Freidin, Carlos P. Otero & Maria-Luisa Zubizaretta (eds), Foundational Issues in Linguistic Theory. Cambridge MA: The MIT Press.

S F O PRO



1st proofs

 Terje Lohndal Chomsky, Noam. 2006. Approaching UG from below. Ms., MIT. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A cross-linguistic perspective. Oxford: Oxford University Press. Cinque, Guglielmo. 2005. Deriving Greenberg’s Universal 20 and its exceptions. Linguistic Inquiry 36: 315–332. Delsing, Lars-Olof. 1993. The internal structure of noun phrases in the Scandinavian languages: A comparative study. PhD Dissertation, University of Lund. Dyvik, Helge. 1979. Omkring fremveksten av artiklene i norsk (About the development of the articles in Norwegian). Maal og Minne, 40–78. Enger, Hans-Olav. 1993. Morfemet i beskrivelse av bøyingsmorfologien til norrøne substantiv – utfordring eller hindring? (The morpheme in the description of inflectional morphology in Old Norse nouns – challenge or obstacle?) Maal og Minne, 12–27. Enger, Hans-Olav. 1997. Morfemet: En replikk (The morpheme: A reply). Maal og Minne, 91–103. Faarlund, Jan Terje. 1990. Syntactic Change. Toward a theory of historical syntax. Berlin: Mouton de Gruyter. Faarlund, Jan Terje. 2004. The Syntax of Old Norse. Oxford: Oxford University Press. Faarlund, Jan Terje. 2007. From clitic to affix: The Norwegian definite article. Ms., University of Oslo. Faarlund, Jan Terje. In press. A mentalist interpretation of grammaticalization theory. In Grammatical change and linguistic theory: the Rosendal papers, Thorhallur Eythòrsson (ed.). Amsterdam: John Benjamins. Fukui, Naoki. 1986. A theory of Category Projection and its Applications. PhD Dissertation, MIT. (Reprinted as Fukui 1995). Fukui, Naoki. 1995. Theory of Projection in Syntax. Stanford CA: CSLI Publications. Fukui, Naoki. 2005. Theoretical Comparative Syntax. Studies in macroparameters. London: Routledge. Fukui, Naoki & Sakai, Hiromu. 2003. The visibility guideline for functional categories: Verbraising in Japanese and related issues. Lingua 113: 321–375. (Reprinted in Fukui 2005, 289–336). Gelderen, Elly van. 1993. The Rise of Functional Categories. Amsterdam: John Benjamins. 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. Gordon, E.V. 1957. An Introduction to Old Norse. Oxford: Clarendon Press. Guasti, Maria Teresa. 1993/1994. Verb syntax in Italian child grammar: Finite and non-finite verbs. Language Acquisition 3: 1–40. Harris, Tony & Wexler, Kenneth. 1996. The optional-infinitive stage in child English: Evidence from negation. In Generative Perspectives on Language Acquisition, Harald Clahsen (ed.), 1–42. Amsterdam: John Benjamins, Haugan, Jens. 2000. Old Norse Word Order and Information Structure. PhD Dissertation, Norwegian University of Science and Technology. Hoekstra, Teun & Hyams, Nina. 1998. Aspects of root infinitives. Lingua 106: 81–112. Holm, Karen B.F. 2002. Utviklingen av den bestemte artikkel i norsk (The development of the definite article in Norwegian). Cand. Philol. thesis, University of Oslo. Hopper, Paul J. & Traugott, Elizabeth Closs. 2003. Grammaticalization. Cambridge: Cambridge University Press.

S F O PRO



1st proofs

Nominal phrases in Norwegian  Hyams, Nina. 1996. The underspesification of functional categories in early grammar. In Generative Perspectives on Language Acquisition, Harald Clahsen (ed.), 91–127. Amsterdam: John Benjamins. Iatridou, Sabine. 1990. About Agr(P). Linguistic Inquiry 21: 441–577. Julien, Marit. 2002. Syntactic Heads and Word Formation. Oxford: Oxford University Press. Julien, Marit. 2005. Nominal Phrases from a Scandinavian Perspective. Amsterdam: John Benjamins. Kester, Ellen-Petra. 1993. The inflectional properties of Scandinavian adjectives. Studia Linguistica 47: 139–153. Lebeaux, David Seth. 1988. Language Acquisition and the Form of the Grammar. PhD Dissertation, University of Massachusetts. Lechner, Winfried. 2005. Interpretive effects of head movement. Ms., University of Tübingen. Leiss, Elisabeth. 2000. Artikel und Aspekt: Die grammatischen Muster von Definitheit. Berlin: Mouton de Gryter. Lohndal, Terje. 2006. Grammaticalization as a science universal. Ms., University of Oslo. Lundeby, Einar. 1965. Overbestemte substantiv i norsk og de andre nordiske språk (Nouns with double definiteness in Norwegian and the other Nordic languages). Oslo: Universitetsforlaget. Marantz, Alec. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. In University of Pennsylvania working papers in linguistics 4.2, Alexis Dimitriadis, Laura Siegel, Clarissa Surek-Clark, & Alexander Williams (eds), 201–225. Mohr, Sabine. 2004. Clausal Architecture and Subject Positions: Impersonal constructions in the Germanic languages. PhD Dissertation, University of Stuttgart. Newmeyer, Frederick J. 1998. Language Form and Language Function. Cambridge MA: The MIT Press. Nygaard, Marius. 1906. Norrøn syntax (Old Norse syntax). Oslo: Aschehoug. Ouhalla, Jamal. 1991. Functional Categories and Parametric Variation. London: Routledge. Papazian, Eric. 1995. Morfemet og substantivbøyinga i norrønt (The morpheme and the inflection of nouns in Old Norse). Maal og Minne, 129–149. Poeppel, David & Wexler, Kenneth. 1993. The full competence hypothesis of clause structure in early German. Language 69: 1–33. Radford, Andrew. 1990. Syntactic Theory and the Acquisition of English Syntax. Oxford: Blackwell. Radford, Andrew. 1996. Towards a structure-building model of acquisition. In Generative perspectives on language acquisition, Harald Clahsen (ed.), 43–89. Amsterdam: John Benjamins. Radford, Andrew. 2000. Children in search of perfection: Towards a minimalist model of acquisition. Essex Research Reports in Linguistics 34. Rizzi, Luigi. 1990. Relativized Minimality. Cambridge MA: The MIT Press. Roberts, Ian & Roussou, Anna. 2003. Syntactic Change: A minimalist approach to grammaticalization. Cambridge: Cambridge University Press. Roberts, Ian. 2005. Principles and Parameters in a VSO Language: A case study in Welsh. Oxford: Oxford University Press. Roberts, Ian. 2006. Clitics, head movement and incorporation. Ms., University of Cambridge. Sadock, Jerrold M. 1991. Autolexical Syntax. Chicago IL: The University of Chicago Press. Sandström, Görel & Holmberg, Anders. 1994. Adjective incorporation and the syntax of the Scandinavian noun phrase. Department of general linguistics, University of Umeå, Report, 35: 81–97. Santelmann, Lynn. 1993. The distribution of double determiners in Swedish: Den support in D. Studia Linguistica 47: 154–176.

S F O PRO



1st proofs

 Terje Lohndal Schütze, Carson T. 1997. INFL in child and adult language: Agreement, case and licensing. Doctoral dissertation, MIT. Sigurðsson, Halldór Ármann. 1993. The structure of the Icelandic NP. Studia Linguistica 47: 177–197. Sigurðsson, Halldór Ármann. 2004a. Agree and agreement: Evidence from Germanic. In Focus on Germanic typology, Werner Abraham (ed.), 55–99. Berlin: Akademie Verlag. Sigurðsson, Halldór Ármann. 2004b. Meaningful silence, meaningless sounds. Linguistic Variation Yearbook 4: 235–259. Sigurðsson, Halldór Ármann. 2006. The Icelandic noun phrase: Central traits. Arkiv för nordisk filologi 121: 193–236. Starke, Michael. 2001. Move dissolves into Merge: A theory of locality. PhD Dissertation, University of Geneva. Svenonius, Peter. 2006. The position of adjectives and other phrasal modifiers in the decomposition of DP. Ms., University of Tromsø. Taraldsen, Knut Tarald. 1990. D-projections and N-projections in Norwegian. In Grammar in progress, Joan Mascaró & Marina Nespor (eds), 419–431. Foris: Dordrecht. Thráinsson, Höskuldur. 1996. On the (non-)universality of functional categories. In Minimal Ideas: syntactic studies in the minimalist framework, Werner Abraham, Samuel David Epstein, Höskuldur Thráinsson & C. Jan-Wouter Zwart (eds), 253–281. Amsterdam: John Benjamins. Thráinsson, Höskuldur. 2003. Syntactic variation, historical development and minimalism. In Minimalist syntax, Randall Hendrick (ed.), 152–191. Malden: Blackwell. Vangsnes, Øystein Alexander. 1999. The Identification of Functional Architecture. PhD Dissertation, University of Bergen. Vangsnes, Øystein Alexander. 2004. Rolling up the Scandinavian noun phrase. Ms., University of Tromsø. Vangsnes, Øystein A., Holmberg, Anders & Delsing, Lars-Olof. 2003. Dialektsyntaktiska studier av den nordiska nominalfrasen (Dialect syntactic studies of the Nordic nominal phrase). Oslo: Novus. Wexler, Kenneth. 1994. Optional infinitives, head movement and economy of derivation. In Verb Movement, Norbert Hornstein & David Lightfoot (eds), 305–350. Cambridge: Cambridge University Press. Wexler, Kenneth. 1998. Very early parameter-setting and the unique checking constraint: A new explanation of the optional-infinitive stage. Lingua 106: 23–79. Zamparelli, Roberto. 2000. Layers in the Determinative Phrase. New York NY: Garland.

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny Correlation between DP, TP and aspect in Old English and first language acquisition Fuyo Osawa Tokai University, Japan

The aim of this paper is to discuss the English NP and its historical development into DP via the emergence of a functional D-system. This is related to the mechanism of theta-binding (Higginbotham 1985). The languages of the world could be described in terms of a D-system and morphological case. In addition, drawing on Abraham (1997) and Leiss (2000), I will examine if there is a complementary distribution between DP/TP and aspect. The correlation between DP/TP and aspect is observed in early child languages. This suggests that there might be a parallel between first language acquisition (ontogeny) and diachronic change (phylogeny) (cf. Osawa 2003a).1

Introduction In this paper I will discuss the development of nominal structures, focusing on the history of English. In doing so, I will examine the relation between DP and aspect or TP in languages. More precisely, drawing on Abraham (1997) and Leiss (1999; 2000), I will seek a correlation between the absence of DP and TP, and the presence of aspect. Child language is also the target of the examination.

. Acknowledgements: I would like to thank many people for their useful comments and suggestions. Especially, I am very grateful to the editors of this volume for their reviewing of my paper. They spent considerable time reading my drafts and pointing out inadequacies of my argumentation. Their insightful observations and invaluable advice resulted in numerous improvements of this paper. Needless to say, all remaining errors are my own responsibility.

S F O PRO



1st proofs

 Fuyo Osawa

My claim is that in Old English (OE2) a functional category D is absent, that is, OE has only NPs, and hence, the related syntactic phenomena are absent. There are many pieces of evidence for the absence of a functional D-system, as discussed below. Later NPs develop into DPs via the emergence of a functional category D within former nominal structures. This is also supported by historical facts, for example, by the appearance of new nominal phrases such as group genitive constructions. The question that arises here is whether this direction of language change from NP to DP is universally applicable. With regard to the Germanic languages and the Romance languages, this direction has been confirmed by a few researchers. Not only Abraham (1997), but Philippi (1997: 62–93) argues that the emergence of articles is a relatively recent development; languages like Gothic, Old High German (OHG), Old Saxon and OE do not have a definite or an indefinite article. Vincent (1997: 14–169) also argues that functional categories like D, C and I, which are amply instantiated in all the modern Romance languages, play a limited role or are simply absent in Latin. Then, neither of articles and clitic pronouns was attested in Latin and hence it follows that they emerged in the Romance languages later. I propose OE nominal structures which are different from Present-day English (PDE) ones, assuming parallelism between nominal and clausal structures. In the generative framework, the parallel internal structure of noun phrases and clauses has been widely accepted since Abney’s (1987) DP hypothesis. The parallel internal structure between noun phrases and clauses in PDE is supposed to account for its many syntactic phenomena. Thus, paralleling V-to-I or C movement, there is N-to-D raising in overt syntax in some languages, which is an instance of head movement (cf. Ritter 1991; Longobardi 1994 among others). Genitives and possessives are paralleling the clausal subject. I have proposed that OE clauses are a projection of a verb and have a partially flat structure without a functional T-system (cf. Osawa, 2000b; 2003a). Then, an

.

The following is the list for abbreviations used in this paper. . . e E position (E role) . ME . OE OHG PDE R role

accusative dative empty Event position, Event role genitive Middle English nominative Old English Old High German Present-day English Referential role

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

OE nominal structure is expected to have a structure paralleling an OE flat clause structure against the analysis that an OE noun phrase contains a DP layer (cf. Wood 2003; Alexiadou 2004). I propose that an OE noun phrase has a partially flat structure. I claim that what triggered the emergence of a functional D-system is, first of all, the demise of case morphology. I also argue that the languages of the world might be described in terms of a D-system and morphological case, assuming that a Referential role of nominals must be bound by either a functional D or morphological case. Furthermore, I will examine the hypothesis that there is a correlation between DP, TP and an aspect system. Drawing on Abraham (1997) and Leiss (1999, 2000), I examine if a morphologically realized aspectual system is involved in determining the referential status of nominals. I will show that there is some parallelism between first language acquisition (i.e. ontogeny) and diachronic language change (i.e. phylogeny) in that a functional D emerges later, and the emergence of a D-system brings about the change from NPs to DPs in both domains. The role of aspect and the absence of DP or TP also concern this discussion. My general background assumption is that functional categories are not operative from the beginning of languages, but over a period of time a functional category emerges and the new, emergent functional category brings about new syntactic phenomena. Following the Minimalist Program, I assume that language variation is due to differences in the degree to which functional features are codified as grammatical functional categories in the clause structure (cf. Chomsky 1995; van Gelderen 1993, 2004; Osawa 2000b, 2003a, 2003b). I claim that this applies both synchronically and diachronically; that is, the theory of functional category emergence also accounts for synchronic variations which are observed in many languages. This paper is organized as follows. In Section 2, I look through the structure of nominals in PDE and describe the difference between NP and DP. I show that the theory of theta-binding proposed by Higginbotham (1985) is the most relevant theory. In Section 3, I turn to the OE period and show that there are no DPs in OE by examining historical data. I refer to some related works on a D-system such as Wood (2003) and Alexiadou (2004) etc. Next, I discuss the emergence of a D-system and argue that the demise of case morphology and the demise of the OE aspectual system triggered this emergence of a D-system. The emergent D made new nominal constructions, such as group genitives possible in Middle English (ME), which were not available in OE. In Section 4, I try to classify languages of the world in terms of a D-system and morphological case, assuming that a Referential role of nominals must be bound by either a functional D or morphological case. Since NPs are inherently predicative, they cannot occur in argument positions.

S F O PRO



1st proofs

 Fuyo Osawa

Furthermore, I will discuss the possibility of the role that an aspectual system has in identifying the referential status of argument nominals. See Leiss (1999, 2000) for the definiteness effects of aspect. Interestingly, early child grammars around 20 months lack functional categories including DP and TP. However, data from several languages show that child grammars contain aspectual information at this pre-functional stage. This will be dealt with in Section 5. Finally I conclude my discussion in Section 6.

.

Structure of nominals: NP versus DP

According to the DP analysis (Abney 1987; Longobardi 1994), nominals in Present-day adult English are assumed to be a projection of a head D, not a head N. The structure of nominals is the following: (1)

DP D the/a/my..

NP dog

The difference between NP and DP is described as follows. As Longobardi (1994: 628) argues, NPs are inherently predicative and thus cannot occur in argument positions. NPs are not referential; referential nominals may be paraphrased as “those that are understood as denoting a particular entity in the universe of discourse” (Rapoport 1995: 154). This is exemplified in the following sentences: (2)

a. Bill is captain. b. *I met captain yesterday. c. *Captain donated a lot of money to the Red Cross.

As is claimed in Longobardi (1994), NPs are inherently predicative and not referential. Only DP can occur in argument positions. As Longobardi (1994: 628) argues, a common noun is kind-referring, not referential. (3)

DP can be an argument, NP cannot.


The role of picking out a particular referent is assumed to be taken care of by a functional D in PDE; the role of a functional D is to change predicative nominals into arguments. There is another approach to nominals. Chierchia (1998) argues that there are parameterized differences in the nature of nouns. In some languages like Chinese, NPs are inherently argumental and can thus occur freely in the argument positions without determiners, while they are predicates in other languages (Romance languages) and cannot occur in argument positions without determiners. In Germanic

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

languages NPs are allowed to have these double natures. If the claim that Chinese NPs are inherently argumental is along the right lines, this may give some solution to the question of Chinese, which will be touched upon in Section 4.1. This picking out operation is best explained by the theory of theta-binding proposed by Higginbotham (1985). According to Higginbotham (1985), a simple noun like dog has an open place in it and so denotes each of the various dogs. This open place is a referential argument in the argument structure of the word dog, which is called “Referential role”, or R role. This position must be bound for an NP to be an argument as discussed above. That is, a nominal must be specified as either definite or indefinite for interpretation. This binding mechanism is illustrated below: (4)

[DP [D' D NP]] DP D' D the

NP

N' N dog

Theta-grids shown in angle brackets are projected from lexical items and are carried over by every node in the tree. The Referential argument position 1 is theta-bound by D, that is, the position is discharged by theta-binding. The asterisk indicates that the position closes or is discharged. When every theta role in an associated theta grid is discharged, we can say that a constituent is saturated. The complete phrase DP is saturated, i.e. all positions are discharged and the phrase is thematically complete (cf. Higginbotham 1985: 561). The bare noun milk is a DP headed by a null determiner: DP

(5) D ∅

NP milk

Nouns like milk, water, or wine are mass nouns and have an indefinite existential interpretation. Plural count nouns which can occur without overt determiners are also understood as belonging to this category. This reading comes from the principle stating that an empty determiner is subject to the universal constraint that it has an existential interpretation by default: (6)

[D e] = default existential interpretation

S F O PRO




1st proofs

 Fuyo Osawa

Since it is a default, this interpretation may be overruled by the presence of other elements like quantifiers or adjectives. Pronouns are assumed to be determiners used without a complement. DP

(7)

D we/you/they

NP ϕ

That is, the person properties of the DP (first person, second person, etc.) are determined by those of its head determiner. Thus, we can account for why themselves or they are allowed to occur in the following sentences. (8)

a. b.

We syntacticians take ourselves/*yourselves/*themselves too seriously, don’t we/*you/*they? Syntacticians take themselves/*ourselves/*yourselves too seriously, don’t they/*we/*you? (Radford 1997: 96) DP

(9) D we/ϕ

NP syntacticians

According to Abney (1987), genitive -’s in English is supposed to be a head determiner which takes a D projection as its specifier, and an N projection as its complement: (10)

DP DP John

D' D -’s

N car

There is no agreement about the status of this/that in PDE among the researchers. They are used either as prenominal determiners or pronominally without any complement. The task of a functional D is, as we have observed, to change predicative nominals into arguments by identifying the referentiality of a nominal. However, the function of identifying the referentiality of a nominal is not always taken care of by D. I propose that in the absence of a functional D-system, morphological case on the head nouns can determine the referentiality of a nominal. In OE, without D, morphological case performed the same task as D. Case affixes attached to head nouns can bind the Referential role as in (11).

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

(11)

OE stanum

NP NP N stan

case affix um (, plural)

(Osawa 2000a: 63)

The derivation of these forms is a purely morphological process, with no syntactic implications, that is, the addition of case affixes to nominals is done in the lexicon. In OE morphological case was assigned to a thematically related NP. All NP arguments of a V must have one thematic role and carry a case related to its thematic role. Morphological case marking was sufficient for a nominal constituent to be an argument. Assuming the difference between NP and DP, and assuming the role of a functional D, I will argue that earlier NPs changed into DPs via the emergence of a D-system within nominals in the history of English in Section 3. . .

The emergence of a D-system in the history of English Absence of a D-system in Old English

In this Section, I turn to the OE period and first show that there are no DPs in OE. I will refer to some related works on a D-system such as Wood (2003) and Alexiadou (2004). As touched upon above, I claim that there was no D-system and morphological case bound the R-role of a nominal in OE. The possible counterargument against this claim is that there were two demonstratives se (seo/þæt), þes (þis/þeos), which functioned as determiners. Can these words be a functional D? We must be careful when we decide whether some lexical item in a given language is qualified as functional category or not. As Abney (1987: 64f.) argues, the nature of functional categories is multi-faceted; functional elements are generally phonologically and morphologically dependent; functional elements are usually inseparable from their complements; functional elements lack descriptive content; the semantic contribution is second-order, etc. However, none of these properties are crucial in deciding whether an element is lexical or functional. Each of these properties constitutes a tendency. Not all of these properties need to be shared by all functional categories. Therefore, I would like to suggest, in addition to the forementioned properties, that the presence of syntactic effects which are dependent on the presence of a relevant functional category in a given language is crucial. As Philippi (1997: 62) discusses, even if “we find in all these languages (i.e. the Germanic languages) demonstratives used in a way similar to the article of the modern Germanic languages, we cannot label them as articles, the latter (i.e. articles)

S F O PRO



1st proofs

 Fuyo Osawa

acting as obligatory definiteness markers in modern Germanic languages.” A functional D is syntactically required in some nominal structures in PDE. Wood (2003) argues that OE had DP since there is N-to-D movement of proper names, although there was no dedicated definite article, and demonstratives were DP specifiers. She claims that there must be some functional structure before the noun, although there was no article as such. Again care is needed when we posit the invisible or inactive elements in languages. I follow the proposals made by Thráinsson (1996) and Fukui and Sakai (2003). Fukui and Sakai (2003: 329) argue that “if the functional categories are present in a language, but they are not active, what does their existence mean exactly?” Along similar lines with Thráinsson (1996), they propose “The Visibility Guideline for Functional Categories” (Fukui and Sakai 2003: 327): (12)

The Visibility Guideline for Functional Categories A functional category has to be visible (i.e. detectable) in the primary linguistic data.3

Thráinsson (1996) argues that languages may vary with respect to the functional categories they have, and proposes the real Minimalist Principle: “Assume only those functional categories that you have evidence for”. Drawing on these discussions, and the enumerated properties of functional categories, I will show that overt evidence observed in English suggests the absence of DPs in OE and that demonstratives in OE are not determiners. .

Historical facts

In OE, we can find examples of bare NPs in which determiners would be required in PDE. (13)

Her Martianus and Valentinus on-fengon rice here Mauricius-. and Valentinian-. seized kingdom-. (AS. Chronicle Parker MS, from Sweet 1953: 73) ‘At this point Mauricius and Valentinian seized the kingdom’

(14)

He gefeng He (The warrior) grasped Þæt hire wið halse that her with neck-.

þa fetelhilt . . . yrringa sloh then linked hilt-. angrily stroke heard grapode banhringas bræc strongly grasped vertebras-. broke

. This view is strongly repudiated in Leiss (in this volume) as well as Abraham (in this volume). A language may have a covert functional category like an empty D in PDE in the presence of visible D heads. However, it is implausible to say that a language has an invisible functional category, which has no visible counterpart.

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

bil eal ðurhwod fægne flæschoman heo on flet sword-. all went through fated body-. she on floor-. gecrong sweord wæs swatig. . . fell sword-. was bloody ‘He (the warrior) drew a linked hilt . . . and brought it down angrily to take her by the neck strongly, breaking the bones; the sword went through the death-doomed body and she fell to the ground and the sword was bloody’ (Beowulf 1563–1569)

In (13), rice ‘kingdom’ usually would need a determiner in PDE. In (14) fetelhilt is an internal argument of gefeng ‘grasped’ and in PDE a determiner would be required. Likewise, bil and sweord are arguments of the predicate verbs and determiners would be required in PDE. These examples clearly show that OE had no syntactic D-system, contrary to PDE. As mentioned above, functional elements often cannot occur independently of their complements and they are generally phonologically and morphologically dependent in PDE. (15)

a. I saw the/a *(boy). b. *The is a great king.

However, demonstratives in OE were not dependent on the noun or nominal elements, but they were independent lexical elements. The evidence to show this comes from the fact that they were used as demonstrative pronouns without the company of nominals, or as an antecedent to a relative, meaning that (man) that, he that etc. as in (16): (16)

a.

b.

c.

d.

þæt wæs god cyning! that was good king-. ‘that was a good king’ (Beowulf 11) 7 se swiþe gewundad wæs and he very badly wounded was ‘and he was very badly wounded’ (AS. Chronicle Parker MS.48, 10.755) Se wæs feorða eac fiftegum from Augusto He was forth also fifty from Augustus-. ‘He was the fifty-fourth from Augustus’. (Bede 54. 22–4) Đes is forðon ðe ðe 3ecuoeden wæs ðerh esaias this is for that man spoken was by Isaiah-. ‘For this is he that was spoken of by Isaiah’ (Lindisf. Gosp. Matt. Iii. 3, OED)

The demonstrative se/seo/þæt was used as a relative pronoun, which inflected for gender, number and case: (17)

S F O PRO

Æþelswiþ cuen, sio wæs Ælfredes sweostor Æthelswith queen-. who was Alfred’s-. sister-.



1st proofs

 Fuyo Osawa

cyninges forþ ferde king’s-. away passed ‘Queen A., who was King Alfred’s sister, passed away’ (AS. Chronicle Parker MS 82. 1. 888, from Ono and Nakao 1980: 325)

As Abney (1987) points out, functional elements are assumed to lack descriptive content and their semantic contribution is subsidiary. However, the demonstratives mentioned above made an important semantic contribution of their own. These demonstratives were used to denote a person or a thing pointed out or presented deictically, and attention was drawn onto them. All the above examples suggest that demonstratives in OE have the status of N, not D and they were used to modify a head noun. Concerning the numeral an, a number of examples to show that it is not an article are available. In the following examples, an is used alone as a nominal, while PDE a/an cannot occur alone ‘*A of them stood by’, or ‘*A who was a lawyer. . .’: (18)

a.

b.

c.

Soðlice an of þam þe ðar embe-uton stodon his Truly one of them-. that there around stood his swurd abræd sword-. drew ‘And/truly one of them that stood by drew his sword’ (Mark 14. 47, from Ono and Nakao 1980: 354) 7 an þe wæs þære æ-ys lareow axode hyne and one who was there law doctor asked him-. ‘And one of them, who was a lawyer, asked him.’ (Matthew 22. 35, from Ono and Nakao 1980: 354) oð ðæt an ongan fyrene fre(m)man feond on helle until one began sin-. do enemy-. on hell-. ‘until one, an enemy from hell, began to compass deeds of malice’ (Beowulf 100)

Although the use of an or sum is rare, they can mean something more than just one when they are used: (19)

(20)

þæt wæs an cyning æghwæs orleahtre that was a king-. in every respect blameless ‘that was a peerless king’, blameless in every respect’

(Beowulf 1885–1886)

eower sum Your-. certain one ‘a particular one among you’.

In (19) the meaning ‘peerless, blameless’ comes from the numeral an, and ‘particular’ is ascribed to sum in (20) (cf. Rissanen 1967: 200–201, 210; Mitchell and Robinson 1992: 107).

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

I argue that OE personal pronouns occupy the N position. The genitive form of personal pronouns like min or his served as an argument of a predicate verb or a predicate adjective like other substantives, since in OE there are verbs or adjectives which require genitive-marked nominals as their arguments: (21)

Nu þu his [i.e. þæs leohtes] hrinan meaht Now you it (the light) (neuter .) touch can ‘Now you can touch it.’ (Cædmon’s Genesis 616(gr.) OED)

This original function of genitives, however, shrank and they came to modify the head nouns. Like this genitive forms of personal pronouns have lost their direct relation to the predicative verbs. See Abraham (1997). Alexiadou (2004: 39–41) argues that they are like adjectives, giving several criteria that support her analysis: they carry case, gender and number information (cf. Demske 2001; Wood 2003). The overall argument of Alexiadou (2004) that possessive pronouns become possessive determiners in English due to the breakdown of Agreement morphology (cf. Alexiadou 2004: 54) is basically consistent with my claim. However, it is doubtful whether they were established as adjectives. The criteria which she has provided meet with counter-examples. For example, the word order is crucial for this analysis: possessive pronouns precede nouns much like other adjectives; when possessive pronouns appear together with other adjectives, possessive pronouns precede them. (22)

a. b.

þin rice your-. kingdom-. (Matthew 6, 10) eowerne gyngstan broðor your-. youngest-. brother-. (Genesis 42, 34, cited from Alexiadou 2004: 40)

However, as Alexiadou (2004) argues, the reversed word orders are also possible. Possessive pronouns may follow the nouns such as fæder min ‘my father’4 and there are nominal phrases in which possessive pronouns follow adjectives. (23)

of [inneweardre his heortan] from the more inner-. his-. heart-. ‘from the more inner of his heart’

(Alexiadou 2004: 40)

I keep open the question of whether OE possessive pronouns have the nominal status or adjectival status. In my view, they are nouns and they are . This pattern is said to be marginal in OE (see Alexiadou 2004, citing. Pintzuk: 2002). However, as pointed out by Alexiadou (2004), the presence of a few examples, nevertheless, suggests that this word order is marginally possible in OE and that it is not totally excluded from the grammar of the language.

S F O PRO



1st proofs

 Fuyo Osawa

used like adjectives. Not all of genitive forms of personal pronouns inflected; his, hiere, hiera did not inflect. These behaviours of possessive pronouns which Alexiadou (2004) has given do not constitute strong supporting evidence for a DP layer in OE in their discussion. I will turn to the question of a DP layer in Section 3.4. Finally, the absence of a D-system is also supported by the syntactic evidence, i.e. the lack of anaphor binding in OE. In OE, simple personal pronouns were used as anaphors and then, the meaning of the sentence “He killed him” was indeterminate whether the object referred to the subject or not. Since a D-system is the locus of binding properties of nominals and pronouns, this absence will follow easily if we assume the lack of a D-system in OE. See van Gelderen (2000) for further details. . A DP layer in Old English? As observed in the previous Section, my position is not far from Wood (2003) and Alexiadou (2004); there was no dedicated definite article and OE only had demonstratives (Wood 2003), or the grammaticalization of possessive adjectives as possessive determiners took place in the history of English since its result is the appearance of new functional material (Alexiadou 2004). Alexiadou argues that this grammaticalization took place under two conditions: (a) the establishment of D0 and (b) the loss or rather weakening of the agreement features of weak adjectival possessors (Alexiadou 2004: 54–55). However, their positions differ from mine in that they assume a DP layer with an empty D-slot for OE. My position on invisible functional heads has already been shown above in Section 3.1. Evidence for the assumed DP layer in OE seems to come from word order in nominals. For example, Wood (2003) gives a very detailed analysis of word order in OE nominals and claims that demonstratives, possessives and adjectives are strictly ordered, and then, there must be some functional layers above NP in OE. Thus, when possessive pronouns appear together with other adjectives, possessive pronouns usually precede them. However, there are examples of different word orders available for most of the cases. So, the possessive + noun order is observed, while the noun + possessive order is also found. The demonstrative + possessive + adjective + noun order co-existed with the possessive + demonstrative + adjective + noun order (cf. Alexiadou 2004: 41). For that matter, we should note that, although the adjective + noun order is dominant in late OE, the noun + adjective order is also found in the OE texts. What I could suggest here is that the presence of a few pronominal elements before the head nominal and the word order do not give crucial evidence for the presence of a DP layer and, especially, for the presence of a D-head.

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

.

The emergence of a D-system

The leveling of inflectional endings had already begun in OE, and by the early ME period many OE inflectional distinctions were lost. Morphological case could not perform the task of identifying the R role of nouns and turning them into arguments any more, and subsequently a functional D-system has developed to do the same job in English. There is a view that case morphology on adjectives can perform the task of theta-binding or the identification of referentiality of nominal phrases instead of decayed nominal case inflection. Then, the emergence of determiners may be triggered by the fact that even adjectives gave up their case inflections (see Alexiadou 2004). Apparently, it is not implausible to refer to the decay of case inflections of adjectives as one of the triggering factors. However, this argumentation meets with counterevidence. Abraham (1997: 29) argues that the claim that adjectival, prenominal agreement inflection suffices for the syntactic marking of valency meet with typological counterevidence. Modern Greek has had throughout its history, both definite articles and nominal and adjectival agreement inflection. Thus, it appears that the rise of the article categories cannot be linked directly to inflectional morphology in the attributive adjective. I would like to clarify that the demise of case inflections on nouns, not on adjectives, is directly linked to the emergence of determiners. Abraham (1997) argues that in the case of German at its earliest stage cases on object nouns selected by verbs with morphologically marked aspect could identify the referential status of objects. However, in the diachronic development German loses its aspectual verbal paradigms and then, this mechanism will be lost too. The loss of specific perfectivity as marked morphologically and identified paradigmatically led to the shrinking of verbally governed genitive and to the dominance of accusative or nominally-selected genitives. The interaction of case and aspect was gradually obliterated and, instead, the definite article emerged. In my discussion of English the decay of case inflections is understood in a wider sense, that is, it is not limited to genitives. However, the tenet is the same in that the mechanism of the definiteness effects is through case on nominal arguments of predicatives. It follows that the demise of case on adjectives is not directly related to the emergence of a D-system. I will return to this issue later in Section 4.1. Thanks to the emergent D, new nominal constructions were made possible. For example, group genitive constructions such as the king of England’s wife are not possible without a D-system, since in this construction another DP occurs in [Spec, DP] position. If we assume that the genitive form was reanalyzed as a D-head, we can easily explain this innovation. This issue will be discussed in the next Section.

S F O PRO



1st proofs

 Fuyo Osawa

. Further evidence: Group genitive constructions In this Section I take up the issue of group genitive constructions as independent syntactic evidence for the non-presence of a D-system in OE, and its subsequent appearance. I will show that the DP analysis accounts for the appearance of group genitive constructions nicely, drawing on Nakao (1972), Ono & Nakao (1980) and Hamasaki (1993). The group genitive is a construction where the genitive ending -’s is apparently affixed to the last element of a noun phrase. (24)

the king of England’s hat DP

(25) DP D' D the

N king

D' D ’s

NP

NP

N'

N'

N hat PP of England

As mentioned above, genitive -’s in PDE is supposed to be a head determiner according to the DP analysis. The genitive marker -’s is not an inflection on the last noun, but is attached to the whole DP in the specifier position (cf. Roberts 1997: 67). Then, the structure of a group genitive construction (24) is (25). This group genitive construction is not observed in OE; instead OE used the following constructions to express the corresponding notion: i. (26) ii. (27)

[N1-. + N-Head + N2-.] Ælfred-es sweostor cynning-es ‘King Alfred’s sister’

(AS.Chronicle Parker MS 82. 2. 888)

[N1-. + N2-. + N-Head] Ælfred-es cynning-es godsunu ‘King Alfred’s godson’

(AS.Chronicle Parker MS 82. 10. 890)

In pattern (i), which is called a split genitive, a head noun is placed between two genitive-marked nouns, and in pattern (ii) two genitive-marked modifying nouns precede a head noun. It is said that the split type construction is not so rare and is used more often than the (ii) type (cf. Ekwall 1943: 2). In later OE, the genitive inflection on N2 in pattern (i) began to drop, while in pattern (ii) the genitive ending on N1 began to drop and this form became common in ME (see Nakao 1972: 220–221; Ono and Nakao 1980: 291–292).

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

i’. (28) ii’. (29)

[N1-. + N-Head + N2-Ø] þurh Iulianes heste ðe amperur by Julian-. command the emperor [N1-Ø + N2-. + N-Head] Davið kinges kinn

(The Ancrene Riwle 109, 11) (Ormulum, from Nakao 1972: 221)

Then, the latter pattern is replaced by the following one: [N1 of N2] -.+ Head

iii.

In (iii), an of-N is placed before a head noun, and the genitive inflection is attached to the last noun of a phrase, which is supposed to be a preceding form of the group genitive construction. This construction first appeared in Chaucer and became common in the 15th century. (30)

the god of slepes heyr

(Chaucer Book of Duchess: 168)

This genitive ending was reanalyzed as a head determiner and the group genitive construction was established as such around the middle of the 15th century. Supposedly, the establishment of group genitive constructions took place after the emergence of a D-system. This explains the temporal delay of the appearance of group genitive constructions. Thus, we can conclude that group genitive constructions have been made possible due to the emergence of a D-system (cf Nakao 1972: 221; Hamasaki 1993). .

Old English nominal structures

I propose the following structure for the nominal phrases such as (26) Ælfred-es sweostor cynning-es as one possibility: (31)

NP N1.

N

N2.

Although the flat structure looks strange, Fischer and van der Leek (1983) propose a partially flat structure for OE impersonal constructions. Thus, I claim, although tentatively, that the OE nominals have a partially flat structure in parallel with the clause structure (cf. Miyamae 2005). I propose that the noun phrase in general has a partially flat, non-configurational structure like this. Note that there is a parallelism between the development of the clause structure, i.e. the development of embedding and a DP structure: just as another clause is integrated into a higher clause, so too another DP can be integrated into the specifier of one DP.

S F O PRO



1st proofs

 Fuyo Osawa

.

Absence of a T-system in Old English

It would be better to refer to the non-establishment of a functional T-system in OE. If I say this, a question of how to analyse a past tense form in OE arises. Is that not enough to posit TP in OE? My answer is that we need other evidence for the presence of TP than temporal suffixes on verbal stems. Verbs, as well as nominals, have an open position, which must be bound for temporal interpretation, too. This Event position (E position or E role) is the hidden argument place for events or situations. In PDE, the binder of this E position is a functional T. However, the binder of this E position is not limited to a functional T. There are many other devices to bind the E position. The role of temporal adverbials in temporal interpretation has been long recognized (cf. Kiparsky 1968). A second binder is aspect. See the discussion in Section 4. In OE the binder is provided by temporal affixes attached to verbal stems. The tense morpheme on the verb is only related to giving a semantic temporal location to the event described by the predicate and does not constitute a syntactic functional category TP. Many syntactic effects are due to the presence of TP in the clause structure such as the subject requirement known as the Extended Projection Principle, modal auxiliaries and do-support. None of them are observed in OE, while all of them are attested in PDE. Thus, we must give syntactic evidence for a T projection. The absence of these phenomena implies the absence of syntactic TP. The affixation is a purely morphological process with no syntactic implication. A few words on the issue of morphology are in order. In the Minimalist Program (Chomsky 1995: 20, 21) morphology is almost reduced to feature checking in relevant functional categories. The past tense form, for example, is already in the lexicon as such, and the inflectional feature [tense] is specified as [past]. This inflected form is drawn from the lexicon and is checked by an appropriate functional head (T). Then, the presence of inflections is supposed to prove the presence of functional categories in the Minimalist framework. Morphology is integrated into syntax. However, my idea of affixation mentioned above is dissociated from this idea of morphology. A morphological process, which is independent of syntax, should be posited for earlier languages. Then, the presence of inflectional realizations does not always lead to the presence of a corresponding functional projection.

. .

Case, D and aspect Introduction

S F O PRO

We have observed that the task of turning an NP into an argument is done by either morphological case or a syntactic D-system. Given these two possibilities,



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

the languages of the world might be described in terms of the following combinations (Osawa 2003a): (32)

The classification of the world languages Morphological case

+

+

−

−

Syntactic D system

+

−

+

−

Example

Modern Greek

Old English Present-day English

*Modern Chinese

Although the co-occurrence of DP and morphological case might appear to go against the main claim, this is not problematic. The co-occurrence might be possible at a certain stage of development, assuming a reallocation of duties from morphology to syntax (cf. Osawa 2003a). However, there is no language in which both a syntactic D-system and a morphological case system are developing together. Some languages have only morphological case, but no D system. Others have only a D system. The asterisk above means that there is no language in which neither morphological case nor a syntactic D system is operative, whereas morphological case can co-occur with a syntactic D-system. Modern Chinese is an apparent counterexample, which has neither morphological case nor determiners. As touched upon in Section 2, Chierchia (1998) argues that Chinese NPs are inherently argumental and can thus occur freely in the argument positions without determiners. Although this may give some hint to this question, this analysis will need greater substantiation. In the case of Chinese, the word order and aspectual information may have an important role in giving (in)definiteness to nouns. I will not go into further details. My hypothesis can also account for the fact that case distinctions on adjectives cannot be sufficient for identifying the referentiality of nouns. According to this hypothesis, the following is impossible: (33)

*NP AP Adjective

NP Case N

The case on adjectives cannot bind the open position in a noun, since this configuration does not meet the structural requirement of theta-binding that a binder and a bindee must be sisters. That is, case on adjectives cannot identify the referential

S F O PRO



1st proofs

 Fuyo Osawa

status of a noun. This is already mentioned above in 3.4. Thus, it appears that the rise of the article categories cannot be linked directly to inflectional morphology in the attributive adjective. Modern Greek makes a good example of this situation. Greek noun phrases have a sequence of both adjectives with case and nouns with case. Case realizations on nouns, not those on adjectives, bind the R role position. Case on adjectives may be an instance of the percolation of case features. As I have touched upon above, the co-occurrence of DP and morphological case might be possible. The same argument goes for this case, too. A sequence of definite articles with case affix and nouns with case affix as in Modern Greek is not impossible. Case morphemes on articles may be an instance of the percolation of case features. Case is first realized on head nominals and percolates onto the other elements in phrases. However, case on nouns, not case on determiners can take part in determining the referentiality. Although I do not exclude the possibility of a different explanation for this, I prefer this one for the moment, since a binder should be limited to one, if we follow Chomsky’s (1982), prohibition of vacuous quantification. An apparent counter-example is, as touched upon above, the presence of the strong declensions of adjectives in OE. However, there are many examples in which case on nominals is present as well as strong adjective declensions: (34)

a.

b.

ðurh wise wealhstodas through wise-. translators-. ‘by learned translators’ (CP7, 4) butan anum Bryttiscum gisle except for one British (Welsh)-. hostage-. ‘except for one Welsh hostage’ (AS.Chronicle Parker MS 48, 10.755)

Furthermore, weak adjective inflections are used without accompanying demonstratives, while strong declensions co-occur with demonstratives: (35)

a.

b.

deoran sweorde good-. sword-. ‘with (my) good sword’ (Beowulf 561) æt þæm ælmihtigum Gode from that-. almighty-. God-. ‘from the almighty God’ (Orosius, from Ono & Nakao 1980: 348)

Then, it is implausible that case on adjectives can identify the referentiality of nominals. .

A link between referentiality and aspect

S F O PRO

Invoking Leiss (1999, 2000) and Abraham (1997) I discuss how a morphologically realized aspectual system is involved in determining the referential status of



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

(verbal) arguments. It has been pointed out by a few researchers that there is a correlation between the definiteness of objects and the aspectual meaning of the predicate. Mahajan (1990: 68–106) points out that there is a correlation between specificity of objects, case assignment and the aspectual meaning of the predicate verb in Hindi. Leiss (1999, 2000) argues that in Germanic languages the combination of alternating cases and verbal aspect had the (in)definiteness effects on object nominals. The growth of the definite article is connected to the loss of aspectually driven (in)definiteness effects of verbal aspect. Abraham (1997) argues that there is a link between aspect, referentiality and nominal case. There appears to be a complementary distribution between languages with strong (morphologically and paradigmatically identified) aspectual systems and the surface representation of determiner forms: if some language makes a systematic formal distinction between perfectivity and non-perfectivity, it can (but need not) do without surface articles. Examples are almost all Slavic languages like Russian, the Indic languages Chinese and Gothic and OHG. Case distinctions are adequate substitutes for referentiality distinctions on NP under verbal government as long as certain aspect and/or Aktionsart conditions are met. Thus, if a language loses its aspectual system, it develops articles in compensation for the loss. This nicely explains what happened in OHG. It is probable that this also pertains to English, too, although the evaluation of aspect in PDE is difficult. We cannot say that PDE has completely lost its aspectual system. However, it is well known that the prefix ge- used with verbs signified the perfective meaning in OE, and English lost this prefixal device for expressing perfectivity in the ME period (cf. Marchand 1969: 130–131; Strang 1970: 190, 351). PDE has also developed its elaborate functional T-system, which was not present in OE as mentioned above. At least, we can say that there has been a shift in the temporal system of English. English has changed from an aspect-dominant to a functional TP-dominant language. It is said that languages which have lost aspect develop articles and a class of modals with deontic and epistemic meanings. This is indeed the case with English. The overall picture I have depicted for English nicely matches this. English developed articles later in its history, compensating the demise of case morphology and the demise of morphologically identified aspectual system. .

Aspect and tense

Here a few words on the nature of aspect may be in order. I argue that aspect is, unlike tense, not a functional category and perhaps belongs to a substantive category (i.e. lexical category like N or V). One piece of evidence for this claim comes from data in first language acquisition. The data shows that children acquire aspect earlier than the acquisition of functional categories. As discussed in Section 6 early

S F O PRO



1st proofs

 Fuyo Osawa

child grammars around 20 months lack functional categories including DP, and TP. However, data from several languages show that child grammars contain aspectual information at this pre-functional stage. This suggests that aspect is not a functional category. It is also widely accepted that the earlier Indo-European languages originally had no grammatical category to express tense (cf. Kuryłowicz 1964; Lehmann 1974). The verbal system of Proto-Indo-European was based on aspect with a basic contrast between imperfective and perfective. Given the above facts, I assume that aspect belongs to a substantive category, which is part of the mental lexicon. Then, aspect may be more deeply rooted in human cognition. When aspect has a morphological realization in a clause structure, the aspect can bind the E(vent) position without TP. See Osawa (1999, 2000b). It needs further research to know how this nature of aspect is intermingled in the current system of referentiality determination.

.

First language acquisition

.

Absence of a D-system

In this Section, I will turn to first language acquisition and show that the later emergence of a D-system is observed in the acquisition data (mainly English). Many properties which are observed in diachronic data are shared by child grammars. I assume that the process of first language acquisition is described as the acquisition of functional categories according to Radford (1990) and Tsimpli (1996). Then, there is some parallelism between first language acquisition (i.e. ontogeny) and diachronic language change (i.e. phylogeny). At the one-word stage around 12 months to 18 months, children can name the object requested in the immediate environment. However, what they acquire is simple sound-meaning associations, and the words are acategorial. Maratsos (1982: 248) argues that “there is no good reason to credit children with having captured formal categories such as ‘noun’ and ‘verb’ in their early speech, even if they use terms such as dog and want”. One piece of morphological evidence for this is that there are no occurrences of the noun plural +s inflection, or of the verbal inflection (Radford 1990: 31). Plurality is related to the lexical category N. (36)

a. b.

What do you want? (Adult) Away. (Alison 16 months)

(37)

a. b.

what are you doing? (Adult) Truck. (Alison 19 months)

S F O PRO (Radford 1990: 32)



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

Children proceed to the next early multi-word stage (aged 18–24 months). This stage consists of lexical categories such as NP, VP and AP only. Children begin to acquire categorization of words. The morphological evidence for this comes from the occurrences of inflected forms observed in the acquisition data. The noun plural forms like cookies, toys are regularly used, alongside a singular form cookie. The number feature is related to a lexical category N. The -ing verb inflected forms are used alongside the uninflected forms. Importantly, although children have acquired lexical categories and their projections, there is no evidence that they have acquired any functional categories at this stage. Therefore, the absence of DPs and the absence of related phenomena are expected in the acquisition data. The examples support this prediction (Radford 1990: 83–84): (38)

a. b. c.

Where helicopter?/Here helicopter/Where bee? Open door/Want ball/ Want car Open can/Open box/Eat cookie

(Stefan 17 months) (Stefan 19 months) (Allison 22 months)

Nouns or NPs can occur in places where DPs are expected in the corresponding adult English. The absence of determiners in the above examples leads to ungrammaticality in the adult grammars. Like this, children acquire semantic properties related to the lexical category N, while their grammars lack phenomena involving DP. Referential properties of determiners are not observed in the child utterances at this stage. The non-presence of pre-nominal determiners such as a/the has already been discussed and has been confirmed to a great extent. However, there are examples which show that children at this stage use the demonstratives this and that, as Radford (1990: 100) points out. (39)

Want that / Want this

(Daniel 19 months)

(Radford 1990: 100)

If this and that are analyzed as pronominal DPs (e.g. DPs headed by a head determiner D), it would follow from this fact that the child had acquired a D-system at this age. However, the child who utters (39) never combines demonstratives with nominals. No examples like this dog/that kitten are observed in the data. As mentioned before, one of the defining properties of functional categories is that they take a specific type of complement. Therefore, there is no plausible reason to analyze this/ that as determiners taking NP complements. Although there is a hot debate about the status of PDE demonstratives among the researchers, following Giusti’s (1997: 110) proposal and taking the historical facts in Section 3.2. into consideration, I argue that demonstratives like this/that should be distinguished from determiners like a/the. They make some semantic contribution of their own and they are also deictic. Hence, the examples like (39) do not provide a strong counter-argument.

S F O PRO



1st proofs

 Fuyo Osawa

The non-acquisition of the genitive determiner -’s, which is assumed to be a head determiner D in the adult English, is illustrated in the following examples: (40)

Mommy’s/Mommy key

(Gia 20 months, holding mother’s key) (Bloom 1970: 93)

In other words, the child who utters (40) does not produce Mommy’s key. This suggests that this ’s form is not a genitive determiner as in adult speech. Maybe, this Mommy’s is better analyzed as an inseparable nominal NP as a whole, which corresponds to Mama-no ‘Mommy’s’ observed in the Japanese acquisition data at the same stage. The non-acquisition of the referential properties of nominals is also expected in early child speech since the binding properties of pronominals are determined by a D-system. Indeed, the reflexive binding is not observed in early child speech. (41)

Kendall see Kendall. ‘I can see myself ’

(Kendall 23 months, looking at a picture of herself) (Radford 1990: 97)

In adult speech, the second Kendall should be a reflexive myself. This absence is also parallel to the situation observed in OE. Concerning the use of pronominals like you, the following dialogue which is from Jespersen (1922: 58) shows that the child is not aware of the variable reference property of pronouns and the child refers to himself by the same pronoun you, which suggests that the child misanalyses you as name (cf. Radford 1990: 104): (42)

ADULT: Shall I carry you?

(43)

CHILD: Carry you (Frans 21 months)

(from Jespersen 1922: 58)

Clark (1978) suggests that the nature of the person error is that pronouns are misanalysed by the child as a type of name, so that I = adult and you = child. Maratsos (1979: 235) characterizes the problem in similar terms, suggesting that “The child hears himself addressed as you and the person talking to him is I. If the child believed these were names, his name would be you and the listener’s I”. Furthermore, look at the following examples (Radford 1990: 96): (44)

a. b.

Mommy pick-up . . . Kendall. Mummy smack Jem.

(Kendall 23 months) a speaker = Kendall (Jem 23 months) a speaker = Jem

The examples show that both speaker and addressee are referred to by nouns. In adult speech, (44a, b) should be “You pick me up” and “You smack me” respectively. All the facts suggest that children have not acquired referential properties of nominals and pronominals.

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

.

TP and aspect in early child English

Child English at the early multi-word stage is also assumed to lack a T-system and the associated grammatical features.5 Indeed, child utterances lack inflected verb forms such as +s, indicating a third person singular present tense form, or +d indicating a past tense form. The verbs produced by young children are either uninflected bare forms, or +ing forms, or participial forms with +en: (45)

a. b.

Hayley draw it/Me talk./Him gone. Baby do it/Daddy coming.

(Radford 1990: 148)

There is much literature supporting the aspectual status of +ing and +en. See, for example, Comrie (1976), Dahl (1985) and C. Smith (1991). I claim that these morphemes in child data occur not as the result of a syntactic operation, but as the result of lexical affixation. See the discussion in 3.7. Further, I claim that verbs in their base forms have an aspectual reading, in particular, imperfective, and crucially aspect does not project as a functional category. As discussed above, TP is a syntactic category and its presence has many syntactic effects. However, child English at this early multi-word stage lacks all of syntactic effects: no auxiliaries, no subject-verb agreement and no EPP. I argue that aspect rather than tense is involved in a number of examples in early child languages. Data from several languages show that child grammars contain aspectual information at this pre-functional stage. In the traditional literature early verbal forms at this stage are said to exhibit distinctions between aspectual categories in a consistent way (cf. Tsimpli 1996: 50). Brown (1973) points out that the -ing form in early child English is consistently used with non-stative verbs from the very beginning. According to Antinucci and Miller (1976), the distinction between stative and non-stative verbs is one of the earliest to appear in early child Italian, where the stative/non-stative distinction is a manifestation of situation type aspect according to C. Smith (1991). Stephany (1986: 379), examining data from Greek children, claims that perfective and imperfective verb stems are already formally distinguished at this stage. Tsimpli (1996) has also examined the cross-linguistic data from Modern Greek, German, French, Irish, Spanish and English, and has concluded that tense marking is absent in general from early child grammars.6 Hence, I claim that aspectual distinctions are operative at the pre-functional stage, while tense distinctions are missing. Aspect morphemes attached to the verb stem are not the result . This position, i.e. the maturation theory of language acquisition, which I am adopting, is rivaled by the continuity hypothesis. See Clahsen (1996). .

S F O PRO

There is a hot debate on this issue among the researchers of language acquisition.



1st proofs

 Fuyo Osawa

of syntactic affixation, but of a morphological rule such as lexical affixation and aspect binds the E role. After 24 months, children proceed to the next stage where functional categories are involved in addition to lexical categories. Genitive determiners and pre-nominal determiners such as a/the begin to appear in the data. A D-system emerges in child utterances.

.

Conclusion

In this paper, I have discussed the English NP and its development into DP via the emergence of a functional D-system in the history of English, assuming the DP analysis (Abney 1987; Longobardi 1994), and the difference between NP and DP. This is related to the mechanism of theta-binding (Higginbotham 1985): either a functional D or case morphology binds the R role. I have shown that historical facts strongly suggest the absence of DP in OE and hence, morphological case bound the R role of nominals in OE. The demise of case morphology has triggered the emergence of a D-system. Thanks to this emergent D, new nominal phrases such as group genitive constructions were made possible in ME. I also have shown that the languages of the world could be described in terms of a D-system and morphological case. At the same time, I have examined the possibility that an aspectual system is also involved in determining the referential status of (verbal) arguments, drawing on Abraham (1997) and Leiss (1999, 2000). Under their hypothesis, there appears to be a complementary distribution between aspectual systems and determiner systems. If some language makes a systematic formal distinction between perfectivity and non-perfectivity, it can do without DP. Interestingly, almost the same correlation between the absence of DP/TP and the presence of aspectual information is observed in early child languages at the pre-functional stage (aged 18–24 months). This suggests that there might be a parallel between first language acquisition (ontogeny) and diachronic change (phylogeny) (cf. Osawa 2003a). In both domains, a functional D emerges later, and the emergence of a D-system brings about the change from NP to DP. In both domains, the correlation between DP, TP and aspect is also observed. There are many issues yet to be addressed: the correlation between DP/TP and aspect may concern the more deeply rooted differences between aspect and tense: what motivates the connection between child language and earlier language, etc. Still, my discussion in this paper has made it clear that this link between the presence of an aspectual system and the absence of DP, or TP, that had been argued for, is confirmed not only diachronically but also synchronically.

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny 

Primary Sources The Ancrene Riwle. The English text of the Ancrene riwle (Cotton MS. Nero A. XIV). (ed.) M. Day. (1952). EETS. AS. Chronicle Parker MS. Two of the Saxon chronicles, parallel: with supplementary extracts from the others. (eds) John Earle & Charles Plummer (1892–1899). Oxford: Clarendon. Bede. The Old English version of Bede’s Ecclesiastical history of the English people. (ed.) T. Miller. (1890–8). EETS. Beowulf. Beowulf and the fight at Finnsburg. Klaeber, F. (ed.) (1950). Lexington: Heath. Chaucer. The Book of the Duchess. The Riverside Chaucer. 3rd (ed.) Larry D. Benson. (1988). Oxford: Oxford University Press. CP King Alfred’s West Saxon version of Gregory’s Pastoral care (ed.) Henry Sweet. (1871). EETS. Matthew. The Gospel according to Saint Matthew. (ed.) W. W. Skeat. (1887). Cambridge: Cambridge University Press.

References Abney, Steven Paul. 1987. The English Noun Phrase in its Sentential Aspect. PhD Dissertation, MIT. Abraham, Werner. 1997. The interdependence of case, aspect and referentiality in the history of German: The case of the verbal genitive. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent. (eds), 29–61. Cambridge: Cambridge University Press. Alexiadou, Artemis. 2004. On the development of possessive determiners. In Diachronic Clues to Synchronic Grammar, Eric Fuß & Carola Trips (eds), 31–58. Amsterdam: John Benjamins. Antinucci, Francesco & Miller, Ruth. 1976. How children talk about what happened. Journal of Child Language 3: 167–189. Bloom, Lois. 1970. Language Development: Form and function in emerging grammars. Cambridge MA: The MIT Press. Brown, Roger. 1973. A First Language: The early stage. London: George Allen & Unwin. Chierchia, Gennaro. 1998. Reference to kinds across languages. Natural Language Semantics 6(4): 339–405. Chomsky, Noam. 1982. Some Concepts and Consequences of the Theory of Government and Binding. Cambridge MA: The MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge MA: The MIT Press. Clahsen, Harald (ed.). 1996. Generative Perspectives on Language Acquisition. Amsterdam: John Benjamins. Clark, Eve V. 1978. From gesture to word: On the natural history of deixis in language acquisition. In Human Growth and Development, Jerome S. Bruner & Alison Garton (eds). London: Oxford University Press. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Dahl, Östen. 1985. Tense and Aspect Systems. Oxford: Blackwell. Demske, Urlike. 2001. Merkmale und Relationen: Diachrone Studien zur Nominalphrase des Deutschen. Berlin: Walter de Gruyter. Ekwall, Eilert. 1943. Studies on the Genitive of Group in English. Lund: Gleerup. Fischer, Olga & van der Leek, Frederike. 1983. The demise of the Old English impersonal construction. Journal of Linguistics 19: 337–368.

S F O PRO



1st proofs

 Fuyo Osawa Fukui, Naoki & Sakai, Hiromu. 2003. The visibility guideline for functional categories: Verb raising in Japanese and related issues. Lingua 113(4–6): 321–375. Gelderen, Elly van. 1993. The Rise of Functional Categories. Amsterdam: John Benjamins. Gelderen, Elly van. 2000. A History of English Reflexive Pronouns. Amsterdam: John Benjamins. Gelderen, Elly van. 2004. Grammaticalization as Economy. Amsterdam: John Benjamins. Giusti, Giuliana. 1997. The categorical status of determiners. In New Comparative Syntax, Liliane Haegemann (ed.), 95–23. London: Longman. Hamasaki, Koichiro. 1993. Eigoshi ni okeru DP kouzou no kakuritsu ni tsuite (On the establishment of a DP structure in the history of English). In Studies in Modern English Editorial Board (ed.) Kindai Eigo no Shosō (Aspects of Modern English), 213–223. Tokyo: Eichosya. Higginbotham, James. 1985. On semantics. Linguistic Inquiry 16: 547–93. Jespersen, Otto. 1922. Language: Its nature development and origin. New York NY: Macmillan. Kiparsky, Paul. 1968. Tense and mood in Indo-European syntax. Foundations of Language 4: 30–57. Kuryłowicz, Jerzy. 1964. The Inflectional Categories of Indo-European. Heidelberg: Carl Winter. Lehmann, Winfred Philipp. 1974. Proto-Indo-European Syntax. Austin TX: University of Texas Press. Leiss, Elisabeth. 1999. Gender in Old High German. In Gender in Grammar and Cognition, Barbara Unterback & Matti Rissanen (eds). 237–257. Berlin: Mouton de Gruyter. Leiss, Elisabeth. 2000. Artikel und Aspekt: Die grammatischen Muster von Definitheit. Berlin: Mouton de Gruyter. Longobardi, Giuseppe. 1994. Reference and proper names: A theory of N-movement in syntax and logical form. Linguistic Inquiry 25: 609–665. Mahajan, Anoop. 1990. The A/A-bar Distinction and Movement Theory. PhD Dissertation, MIT. Maratsos, Michael. 1979. Learning how and when to use pronouns and determiners. In Language Acquisition: Studies in first language development, P. Fletcher & M. Garman (eds), 225–240. Cambridge: Cambridge University Press. Maratsos, Michael. 1982. The child’s constriction of grammatical categories. In Language Acquisition: The states of the art, Eric Wanner & Lila Gleitman (eds), 240–266. Cambridge: Cambridge University Press. Marchand, Hans. 1969. The Categories and Types of Present-Day English Word Formation: A synchronic-diachronic approach. München: Beck’sche. Mitchell, Bruce & Robinson, Fred C. 1992. A guide to Old English. Oxford: Blackwell. Miyamae, Kazuyo. 2005. Why has a functional D emerged? Tsuda Journal of Language and Culture 20: 86–98. Nakao, Toshio. 1972. Eigoshi II (A history of English II). Tokyo: Taishukan. OED Second edition on CD-ROM. 1994. Oxford: Oxford University Press. Ono, Shegeru & Nakao, Toshio. 1980. Eigoshi I (A history of English I). Tokyo: Taishukan. Osawa, Fuyo. 1999. The relation between tense and aspect: The emergence of the T system. UCL Working Papers in Linguistics 11: 521–543. Osawa, Fuyo. 2000a. The historical emergence of DP in English. English Linguistics 17(1): 51–79. Osawa, Fuyo. 2000b. The Rise of Functional Categories: Syntactic parallels between first language acquisition and historical change. PhD Dissertation, University College London. Osawa, Fuyo. 2003a. Syntactic parallels between ontogeny and phylogeny. Lingua 113(1): 3–47. Osawa, Fuyo. 2003b. The rise of IPs in the history of English. In Historical linguistics 2001, Barry J. Blake & Kate Burridge (eds), 321–337. Amsterdam: John Benjamins.

S F O PRO



1st proofs

The emergence of DP from a perspective of ontogeny and phylogeny  Philippi, Julia. 1997. The rise of the article in the Germanic languages. In Parameters of Morphosyntactic Change, Ans van Kemenade & Nigel Vincent (eds), 62–93. Cambridge: Cambridge University Press. Pintzuk, Susan. 2002. The syntax of Old English poetry: The position of heads in noun phrases and prepositional phrases. Talk presented at the 8th Germanic Linguistics Annual Conference, Bloomington IN. Radford, Andrew. 1990. Syntactic Theory and the Acquisition of English Syntax. Oxford: Blackwell. Radford, Andrew. 1997. Syntax: A minimalist introduction. Cambridge: Cambridge University Press. Rapoport, Tova R. 1995. Specificity, objects, and nominal small clauses. In Small Clauses [Syntax and Semantics 28], Anna Gardinaletti & Maria Teresa Guasti (eds), 153–178. New York NY: Academic Press. Rissanen, Matti. 1967. The Uses of One in Old and early Middle English [Mémoires de la Société Néophilologique de Helsinki 31]. Helsinki: Société Néophilologique. Ritter, Elizabeth. 1991. Two functional categories in noun phrases: Evidence from Modern Hebrew. In Prospectives on Phrase Structures [Syntax and Semantic 26], S. Rothstein (ed.), 37–62. New York NY: Academic Press. Roberts, Ian. 1997. Comparative Syntax. London: Arnold. Smith, Carlota. 1991. The Parameter of Aspect. Dordrecht: Kluwer. Stephany, Ursula. 1986. Modality. In Language Acquisition, Paul Fletcher & Michael Garman (eds). Cambridge: Cambridge University Press. Strang, Barbara. 1970. A History of English. London: Routledge. Sweet, Henry. 1953. Sweet’s Anglo-Saxon Primer, revised throughout by Norman Davis. Oxford: Clarendon. Thráinsson, Höskuldur. 1996. On the (non)-universality of functional categories, In Minimal ideas: Syntactic studies in the Minimalist framework, Werner Abraham, Samuel D. Epstein, Höskuldur Thráinsson & C. Jan-Wouter Zwart (eds), 253–281. Amsterdam: John Benjamins. Tsimpli, Ianthi-Maria. 1996. The Prefunctional Stage of Language Acquisition. New York NY: Garland. Vincent, Nigel. 1997. The emergence of the D-system in Romance, In Parameters of morphosyntactic change, Ans van Kemenade & Nigel Vincent. (eds), 149–169. Cambridge: Cambridge University Press. Wood, Johanna. 2003. Definiteness and Number: Determiner phrase and number phrase in the history of English. PhD Dissertation, Arizona State University.

S F O PRO



1st proofs

S F O PRO



1st proofs

Demonstratives and possessives From Old English to present-day English Johanna L. Wood University of Aarhus, Denmark

Three different nominal word orders in Old English through present-day English are investigated, in order to determine whether English has an ‘adjectival’ possessive similar to modern Italian. It is argued that the orders a) demonstrative, possessive, noun and b) possessive, demonstrative, noun represent different syntactic constructions, with different paths of development. It is concluded that the a) order represents three different constructions: i) apposition, ii) a possible ‘adjectival’ possessive, no longer found in Middle English, iii) an Early modern English focus construction using the proximal. The b) order represents a demonstrative in form, functioning only as a definiteness marker.1

Introduction As is well known, some languages (e.g. Italian, Portuguese, Russian, Czech, Bulgarian) permit the prenominal co-occurrence of articles and possessive determiners, while in others (e.g. Irish, English, French, Dutch, German, Swedish, Danish, Norwegian) the co-occurrence is ungrammatical. Following Lyons (1986) these are known as adjectival-genitive (AG) and determiner-genitive (DG) constructions (while Schoorlemmer (1998) calls them type 1 (AG) and type 2 (DG) languages). Recent discussions include Lyons (1986, 1999), Giorgi and Longobardi (1991), Plank (1992), Cardinaletti (1998) and Haspelmath (1999). The difference is illustrated below with examples from Italian and English: (1)

a. my book b. *the my book c. *a my book

DG

. Thanks are due to Werner Abraham and Elizabeth Stark for their helpful reviews and suggestions that resulted in significant improvements to the paper, and for helpful comments from the audience at ICEHL 13, (Vienna, Austria, August 2004) and from Sten Vikner. All remaining errors are my own.

S F O PRO



1st proofs

 Johanna L. Wood

(2)

a. *mio libro b. il mio libro c. un mio libro

(3)

a. *un libro de/a me b. a book of mine

AG

In the DG construction, as in (1), a prenominal genitive forces a definite interpretation on the noun phrase and the possessive may not co-occur with articles, while the presence of a possessive in the AG construction does not necessarily result in a definite nominal. If the language has articles they co-occur with the possessive in the AG construction, as in the definite (2)b and the indefinite (2)c. The usual strategy for expressing indefiniteness in languages which use the DG construction is with a post-nominal prepositional phrase as in (3)b. As Lyons (1999: 133) points out, it is necessary to use the term AG and DG “constructions” rather than “languages” because Spanish, for example, permits both constructions, the definite (4)a and (4)b and the indefinite (4)c. It is also apparent that the AG construction is not characteristic of the Romance language family as a whole since French, like English, only permits the DG construction, shown in (5), although Old French (OF) uses the AG construction, as in (6) with a demonstrative, and (7) with an article. Both OF examples are from Arteaga (1995: 69). (4)

a. b. c.

mi libro el libro mio un libro mio

(5)

a. mon livre b. *le ma livre c. *un mon livre d. un livre à moi

(6)

par ceste mie barbe by this--- my--- beard--- ‘by this beard of mine’ (Roland 1719; Jensen 1990 §373)

(7)

Dieu, par le tuen glorioz God, by the--- your--- glorious--- non name--- ‘God, by your glorious name’ (Ste. Eustache 1937; Togeby 1974 §100)

Although present-day English (PDE) is a DG language, it has long been noted that in earlier English demonstratives and possessives2 can co-occur.

S F O PRO

. Sometimes the Italian type is called the possessive “adjective” and the English type is called the possessive “determiner”. The philological literature refers to possessive “adjectives” and “pronouns”,



1st proofs

Demonstratives and possessives 

(Heltveit 1977; Mitchell 1985; Traugott 1992; Demske 2001; Rosenbach 2002; Wood 2003; Allen, in press). In Old English (OE) the three possible orders of demonstrative, possessive, and noun can be found, as discussed by Mitchell (51–56):3 (8)

a. b. c.

demonstrative possessive noun (this my book) possessive demonstrative noun (my this book) demonstrative noun possessive (this book my)

(9)

a.

in þis user circlice stær in this-- our ecclesiastical history- his þa æfestan tungan his that-- pious tongue-- ðone halgan his that- saint his Latin: sanctum suum

b. c.

(Dem Poss N) (Poss Dem N) (Poss N Dem)

(Bede 282.23) (Bede 342.17)

(Vespasian Psalter)

Within the determiner phrase (DP) hypothesis (e.g. Abney 1987), these data have been used to support two different claims about the structure of earlier English. First, there are claims that OE has no DP; both the possessive and the demonstrative are said to be “adjectives” and can occur in either order. As Allen (in press: 151–152) points out, most treatments of Dem Poss and Poss Dem treat them as free variants, not as different syntactic structures. Scholars who assume they are free variants go on to use this as evidence to argue that there is no determiner position (e.g. Yamamoto1997; Osawa 2000). I will not directly address that claim in this paper, though I will argue, as does Wood (2003) and Allen (in press), that Dem Poss and Poss Dem are two quite different syntactic constructions, not freely ordered “adjectives.” Further arguments in support of a DP in OE may be found in Ackles (1997) and Wood (2003, to appear). Secondly, there are claims of diachronic development in the history of English from an AG construction in OE to exclusive use of a DG construction in Early Modern English (EModE). It is generally supposed that EModE constructions such as this my book are evidence that Dem Poss in English occurs up until the 17th or 18th century, after which it starts to decline, as suggested, for example, by

depending on whether they are dependent or independent, inflected or uninflected. To avoid ambiguity and because the adjective/determiner status is under investigation here, I follow Mitchell (1985: 102) and use “possessive” as a neutral term. . All OE examples are from the Helsinki Corpus unless otherwise indicated. Examples from Pope Gregory’s Dialogues and the Blickling Homilies were taken from the editions indicated in the reference list.

S F O PRO



1st proofs

 Johanna L. Wood

Denison (1998) and Rissanen (1999). The focus on EModE has led those scholars who argue that the OE possessive construction resembles the Italian possessive to assume that the change from an AG language to a DG language was not completed until the 17th century. See Alexiadou (2004: 48), for example. Although there is evidence that the possessive in OE is not definite, as first suggested by Heltveit (1977: 49), I will argue that indefinite possessives do not occur as late as the 17th century and that there is nothing remarkable about the 17th century as far as the decline of the AG construction is concerned. I will discuss the constructions in (8)a, b, and c separately, emphasizing that they represent different structures with different textual distribution, and follow different paths of development. In the discussion I emphasise two points. First, I maintain that in order to fully understand the changes that have taken place in the Dem Poss order, (8)a, it is essential to consider the entire history of English up to and including the present. Second, it is important to not only consider whether a demonstrative co-occurs with a possessive, but also whether that demonstrative is the distal, the source of the definite article, or the proximal. I argue that Poss Dem and Dem Poss are different constructions and that there is no evidence to support English having an indefinite possessive later than the OE period and certainly not until the 17th century. In Section 1, I discuss my background assumptions about nominal structure and the history of OE demonstratives and articles. In Section 2, I discuss the demonstrative first order, Dem Poss, starting with PDE. Section 3 focuses on the possessive first order, Poss Dem, both its co-occurrence with an adjective and its occurrence in one particular text, Pope Gregory’s Dialogues. Section 4 briefly discusses the third order, that with a postnominal possessive. Section 5 is a conclusion.

.

Nominal structure

The structure I assume for PDE nominals is as in (10) below. I assume, following Radford (2000) that possessors are the nominal equivalent of verbal subjects, merged lower in the structure. I will adopt the spirit of the analysis pursued by Lyons (1999), who argues that D is a grammatical category encoding definiteness. In this approach, definiteness of possessors is defined structurally, not lexically, that is, possessive determiners are definite if they move to DP; indefinite if they move to a lower position or remain in situ. Unlike, for example, Delsing (1998), Schoorlemmer (1998), and Julien (2005), I do not assume a dedicated Poss(essive)P for possessives. Although not crucial for this analysis, I also assume that the English indefinite article expresses

S F O PRO



1st proofs

Demonstratives and possessives 

cardinality, not “indefiniteness” and heads NumP, giving the following structure for PDE: (10)

DP possk demj

D' NumP

the

numerals

Num' a

DemP tj

Dem' nP tk

n' NP

The co-occurrence of articles and possessives in languages of the Italian type requires an analysis of nominals in which the possessive occurs in a structural position below the article. In the AG construction, instead of moving to DP, possessives are said to move to an Agr(eement) projection below DP. The indefiniteness is a result of not being in a structural definiteness position. This is the line of analysis adopted by, among others, Cardinaletti (1998) and Alexiadou (2004). (11)

DP il

D' AgrP AdjP mio

Agr' a

DemP

NP

S F O PRO

If earlier English has the AG structure, as in (11), we would expect to see this reflected in the word order. In Section 2 below, I argue that even though it appears



1st proofs

 Johanna L. Wood

to be generally accepted in the literature that EModE allows a structure such as (11), with an adjectival type of possessive, the evidence being examples of the form this my book, these examples are not convincing evidence of a AG construction. In the history of English, formal and functional changes have taken place in both the demonstrative and the possessive. In OE, possessives are usually prenominal. The possessive in 1st and 2nd person inflects to agree in person, number and gender with the head noun, but third person is not declined. A possible hypothesis is that Dem occurs only with third person Poss in order to provide required agreement features. Therefore, I will take into account whether co-occurrences of Poss and Dem are found with all persons. With respect to the PDE demonstrative, I assume that DP is a universal category and demonstratives, which have the features [+Def] [+Dem] [±Proximal], move to Spec-DP as shown in (10). It is generally accepted that there is no dedicated definite article in OE and the PDE definite article developed from the distal demonstrative that. The grammaticalization path demonstrative > article is well known, and especially well documented for Romance languages, for example: Heine and Kuteva (2002), Lyons (1999), Roberts and Roussou (2003). Although Lyons (1999: 299) argues that free form articles, including PDE the, are specifiers, I will assume an analysis in which the grammaticalization of the article involves a lower specifier becoming a higher head. That is, there is loss of the [+Dem] feature and loss of Dem > D movement. This follows the economy principles in Van Gelderen (2004) “Head over Spec” and “late Merge” and is a change from Move > Merge as suggested by Roberts and Roussou (2003). In PDE, then, the definite article is the head of DP and is merged there. Although we do know that the PDE article developed from the masculine distal demonstrative through the forms: se > þe > the, the features of demonstratives and articles overlap, and it is not always clear whether surface forms are demonstratives or developing articles. In this respect it is important to remember that the article developed from the distal demonstrative, not the proximal, and it will be seen that there are important distributional differences that differentiate the orders in (8)a and b. As Allen (in press: 158) points out, the Poss Dem order is only found with the distal demonstrative. In terms of the tree structure in (10) this would be consistent with the possessive being Spec-DP and the distal demonstrative (in form) functioning as the article, as the head of DP. On the other hand, the Dem Poss construction is found with both proximal and distal, and, as will be seen in Section 2 below, the proximal demonstrative is found with far greater frequency than the distal, especially in later English. The structure of Dem Poss, however, need not be that of the Italian type in (11). We will see that often the Dem Poss construction may be analysed as two nominals in apposition, rather than the AG construction, which has a determiner-like element and a possessive in the same nominal.

S F O PRO



1st proofs

Demonstratives and possessives 

. The Dem Poss construction In 2.1 below, I discuss the demonstrative first order in PDE and EModE. I argue that the expression is found in PDE and may be analysed not as a single nominal but as apposition. In 2.2, I discuss the construction in OE and Middle English (ME) and conclude that in these periods also, apposition is a possible analysis. I also suggest that the demonstrative-first order that we see from late ME on is a new construction, a focus nominal using the proximal demonstrative, and need not be a continuation of the OE Dem Poss construction. .

Post 17th century Dem Poss

I will start by discussing the status of Dem Poss in PDE. There is a prevalent view in the literature that Dem Poss is frequent in EModE and has been on the decline ever since, although statements about its exact status in PDE are somewhat contradictory. Kytö and Rissanen (1993: 258) report that “In the course of the seventeenth century, this expression [dem + poss + noun] seems to become obsolete, and it only occurs in non-standard varieties of Present-day English,” whereas Denison (1998: 114–115), in discussing determiners such as the, this, a, his, and our, says: “All these items are in contrastive distribution; colloquial present-day English does not permit NPs **this my chapter, possible until the beginning of our period (1776–1997) and later still in literary and legal usage.” Denison’s view is supported by Rissanen (1999: 324) who says that “the order [dem + poss + noun] “occurs in present-day written English in archaic contexts.” First, I want to challenge the claim that the construction is obsolete, archaic, literary, legal, and non-standard in PDE. Since change occurs in speech before it is seen in the written record, older forms tend be preserved in writing. If a frequent expression is leaving the spoken language, it would be consistent that it is found only in archaic or formal contexts as Denison and Rissanen remark. In a search of the British National Corpus (BNC) I found several examples of Dem Poss, both with and without an accompanying adjective; Examples (12)–(14) as well as the well known phase, this, my last will and testament, are indeed archaic. (12)

You know the sort of thing I mean: For as much as ye do it unto one of these my children (G02 Cathedral. Maitland, I.)

(13)

Watson notes some forty examples, including Jer 31.21: Return, O Virgin Israel, Return to these your cities (EUX Directions in biblical Hebrew poetry)

(14)

In witness thereof we have caused these our letters to be made patent (G0S Indigo. Warner, M.)

S F O PRO



1st proofs

 Johanna L. Wood

However, Examples (15)–(19) below show that, in PDE, Dem Poss is not restricted to archaic expressions. The following five representative examples show the demonstrative co-occurring in PDE with 1st, 2nd and 3rd person pronouns. Furthermore, none of these examples are archaic, literary, or legal, least of all (17) and (18), which are from the sports Sections of newspapers. (15)

She was an apparition of dreams and yet I was permitted to whisper intimately to this my dream, to this my vision. (EC8 The bright face of danger. Fisher, Margery)

(16)

on this your glorious second birthday (HA1 Underground: the London alternative press. Fountain, Nigel)

(17)

He must have created some kind of record already with this his ninth world title fight in under two years. (CH3 The Daily Mirror. London: Mirror Group Newspapers)

(18)

Hill’s confidence has grown throughout this his first full season as a front-line racer in Formula One since he replaced Nigel Mansell in the Williams team. (K3X [Liverpool Daily Post and Echo])

(19)

I’d like to welcome Jean-Claude to this his first A G M as a full member of the board. (HM7 Pearson interim results: meeting (Business))

How, then, do we analyze the modern examples above? Are they evidence that the AG construction did not die out in the 17th century but continues into PDE? This seems hardly likely, given that neither definite articles nor indefinite articles cooccur with PDE possessives. A better analysis would be that suggested by Denison (1998: 115) for one of the 19th century examples he quotes, (20), below, that the demonstrative and possessive belong to “parallel NPs in apposition rather than jointly filling a single determiner slot.” However, he does not adopt this analysis for another 19th century example (21), which appears quite similar. (20)

As brisk as bees . . . did the four Pickwickians assemble on the morning of the twenty-second day of December, in the year of grace in which these, their faithfully recorded adventures, were undertaken and established. (1836–7 Dickens, Pickwick xxviii. 408 [Poutsma]) (Denison’s 66c)

(21)

which have already been highly approved of in this their new form by my daughters (1864 Gaskell, Letters 545 p. 723 (1 Jan.)) (Denison’s 66d)

Of course, in (20) the punctuation is a clue to the possible structure, and to native speakers these constructions “feel” like apposition. Allen (in press: 154) quotes a PDE spoken example from ABC news, which she too analyses as apposition: (22)

on this, his third NASA assignment

S F O PRO



1st proofs

Demonstratives and possessives 

Syntactic evidence for apposition comes from examples like (23) below. In this example there is no question that the possessive is part of the same DP as these, since His Majesty’s must be a phrase not a head. (23)

and a squadron of smaller ships in the North Seas were the only secure guardians to these His Majesty’s kingdoms against invasions. (BNB Island fortress. Longmate, N.)

A further observation that may be made about examples (15)–(19) is that they all use the proximal demonstrative. I found no examples with the distal in the BNC. Similar expressions showing a preference for the proximal demonstrative occur in contemporary German, according to Plank (1992), who points out that, while some speakers prefer (24), (25) is also quite natural for some speakers. (26), with the distal demonstrative and not the proximal, is less frequent and (27), while similar to the BNC example in (23), with a full phrase for the possessive, is ungrammatical in German. (24)

dieses Land von uns ‘this country of us’

(25)

dieses unser Land ‘this our country’

(26) ?jenes unser Land ‘that our country’ (27) *diese Kanzler Kohls Rede ‘this Chanzler Kohl’s speech’

The preference for the proximal is apparent also in EModE according to Rissanen (1999: 206) who points out that “the combination of this (or rarely that) and the possessive pronoun is a common construction that was superseded by the end of the seventeenth century by the expression of the type ‘this X of mine (yours etc.).’” It seems then, that in the PDE and EModE data we are looking at a particular use of the demonstrative restricted to the proximal. According to Dixon (2003: 80) the proximal is used in English typically to introduce new information and when there is no spatial contrast. Notice that unlike definite descriptions and universal quantifiers which are excluded from the predicates of existential sentences, as shown in (28)a and b, this in this particular use does not display the definiteness effect (Milsark 1974), as the grammatical (29)a shows. The grammaticality of (29)a is consistent with its function, to introduce new information as does (29)b (cf. Example (15).4

S F O PRO

. Schoorlemmer, in her analysis (1998: 59), “chooses to ignore” expressions such as these friends of mine which, she points out, do not show the definiteness effect.



1st proofs

 Johanna L. Wood

(28)

a. ?there was the/that vision in my mind b. ?there was my vision in my mind

(29)

a. b. c.

there was this vision in my mind there was a (certain) vision in my mind there was this, my vision in my mind

I suggest, then, that when a demonstrative co-occurs with a possessive in PDE the proximal demonstrative acts as a focus marker to emphasise the following nominal. It appears that in EModE and PDE this is undergoing grammaticalization from a demonstrative to a focus marker. This is not an unprecedented change. Heine and Kuteva (2002: 112) give examples of a grammaticalization chain DEMONSTRATIVE > (PERS-PRON>COPULA>) FOCUS, which can proceed directly from demonstrative to focus marker without the intermediate stages. They suggest it is possibly part of the grammaticalization path demonstrative > focus marker> expletive, illustrated in French in (30): (30)

C’ est lui que j’ ai vu This is 3:- whom 1: have seen ‘He is the one that I saw’

The conclusion is that Dem Poss in PDE need not be analysed with the demonstrative as part of the same DP but with the structure as in (31) below: (31)

[DP this ], [DP myi [nP [n ti [NP letter]]]]

Having established that the PDE examples above may be analysed as appositions, what can be said about the supposedly now obsolete 17th century expressions? Rissanen (1999: 206), in his discussion of English syntax 1476–1776, gives the three examples below (his 102–104): (32)

This his goodness stood not still in one or two

([HC] Ascham, 280)

(33)

your Highness will be as good a Lord to that your monastery, as your noble Progenitors have been ([HC] Wolsey, 19)

(34)

So far from complaining from this their inclination (Fielding Tom Jones I.ix 73)

While (33) does look a little different from the modern examples in that it uses the distal, not the proximal, demonstrative, many of, or perhaps all, the EModE examples could also be analysed as appositions. Although the frequency of Dem Poss in EModE appears to be ‘received wisdom’, I know of no systematic study that compares the frequency and function of this expression in EModE with PDE. There is, as yet, no conclusive evidence that the AG construction is found in English up until the EModE period.

S F O PRO



1st proofs

Demonstratives and possessives 

It is important to establish whether or not earlier English has the AG construction at all, and if it does, when it is no longer used. Some researchers appear to accept the conclusion that the EModE Dem Poss construction represents the cooccurrence of two determiners, and consequently something similar to the Italian structure. For example, Rosenbach (2002: 26) cites Rissanen’s examples (32) and (33) above and concludes: “Again, we must either assume that the possessive pronoun was not yet a definite determiner, or, alternatively, this co-occurrence of two definite determiners points to the fact that the ModE situation has not yet been reached.” Also, Alexiadou (2004: 45), who argues for diachronic development in English from an AG to a DG language, cites these same two examples when she suggests that “ME possessives are ambiguous between being clitic and weak elements”, clitic, in this case, meaning the DG construction, and weak meaning the AG construction. But, as was shown above for the PDE examples, the mere co-occurrence of these two elements is not evidence that they must be analyzed as the AG construction. Therefore, in the next Section I will consider Dem Poss in the ME and OE periods and whether the AG construction is represented there. If the Dem Poss construction in EmodE represents an AG expression, common in OE and gradually leaving the language, we would expect the expression to be less frequent in ME than in OE, less frequent in EModE than in ME and even less frequent in PDE. Declining frequency in the modern language is what Plank (1992) predicts for the expression in modern German. However, gradually declining frequency in the history of English does not appear to be the case. Rissanen (1999: 324) points out that the Dem Poss order is “frequent in the 17th century “common” in Old English and “scantily attested in Middle English texts.” It seems that the well documented co-occurrence in EModE is so noticeable because it starts to become more frequent at that time after being relatively infrequent in ME. The following examination of Dem Poss in ME and OE considers the number of co-occurrences in ME when we would expect Dem Poss to be more frequent than in PDE and in EModE, if it is a disappearing AG construction. .

Pre-17th century Dem Poss

In his comprehensive investigation of OE syntax, Mitchell (1985: 53) points out, as has been discussed above, that Dem Poss is common in EModE. He comments: “Whether it has had a continuous existence since Old English and whether it was in origin a calque on Latin remains to be established.” These two questions, the relationship this expression in OE has to Latin and its existence and frequency in OE and early ME, will be addressed in this Section. A pioneering study in this area is that of Kytö and Rissanen (1993), who observe that there is very uneven textual distribution in OE, with most of the examples of

S F O PRO



1st proofs

 Johanna L. Wood

Dem Poss occurring in Latin translations. Allen (2004) investigates the Latin influence further. She compares the OE Gospel of St. Matthew with the Vulgate text from which it is translated and finds that with each of 4 examples of Dem Poss, a form of OE þes, ‘this’, translates a form of Latin hic, ‘this’, as in (35) below (her (8) and her gloss). (35)

Eornustliec ælc þæra þe ðas mine word gehyrð Truly, each those- that these my words hears ‘Truly, each of those who hears these words of mine’ Latin: Omnis ergo, qui audit verba mea haec (Matthew 7.24)

However, Kytö and Rissanen point out that the influence need not necessarily be that of direct translation from Latin, but the influence of a Latin model and Allen (2004: 16) supports this suggestion. She finds a strong correlation between the presence of Dem Poss in a text and the existence of a Latin source, even though the OE constructions are used in places that do not always have corresponding constructions in Latin. She concludes that Dem Poss is part of a “Latinate register which a translator might use.” So, to answer Mitchell’s second question, Dem Poss is not exactly a Latin calque, but nevertheless particularly likely to be found in texts influenced by Latin. It has also been widely observed that not all OE examples need be analysed with Dem and Poss as part of the same nominal, (i.e. an AG construction). For example, both Allen and Kytö and Rissanen discuss (36) below which is from native poetry and hence a counter-example to the suggestion that Dem Poss is Latin influenced. (36)

Dreogeð se mine wine micle modceare suffers that--- my--- friend-- much sorrow ‘that friend of mine suffers great sorrow at heart.’

Allen, however, suggests that the best translation of se mine wine might be ‘he, my lover’, especially since in OE the demonstrative was used as a 3rd person pronoun. That is, the expression could be analysed as involving two nominals, just as I argued in the previous Section with respect to post 17th century examples. Both Traugott (1992) and Mitchell (1985) also indicate that examples like these may involve a pronoun. In discussing (37) below (her 11), Traugott (1992: 173) comments: “se is probably a pronoun in a topicalized construction because the adverb ða follows the subject rather than being verb-initial . . . It is possible that other instances of demonstrative preceding possessive . . . are also to be interpreted as pronominal.” (37)

Se heora cyning ongan ða singan that--- their--- king began then sing ‘He, their king, then began to sing’

(Or 56.31)

S F O PRO



1st proofs

Demonstratives and possessives 

Mitchell takes a similar position in discussing (38), pointing out that Skeat’s translation reads, “What is he, your god, who . . .” (38)

Hwæt is se eower god þe. . . (ÆLS 36.367) what is that--- you-- god who. . . ‘what kind of a god is this, who . . .’

In other words, the structure of the above examples in OE is not demonstrative + possessive + noun, but pronoun + [possessive + noun], two nominals in apposition similar to the modern construction. Similarly (39) and (40) below, ME examples from the Helsinki corpus, also could also be analysed as apposition: (39)

And for as moch as be the summyns and commensyng of this your present parlement. (Petition of Commons 1450)

(40)

wherby he trustith not only to bring this his Realme to the auncien fame and honour (Statutes of the Realm 1420–1500)

So we see that, throughout the history of English, in OE, ME, EModE and PDE, the Dem Poss construction need not be analysed as involving a demonstrative and determiner-like possessive in the same nominal. This does not mean that the AG construction is completely ruled out in OE, just that a more in-depth analysis is called for. Finally, in this Section, I address Mitchell’s first question, that of continuity. Is it possible that the Dem Poss we see in OE has a continuous existence up until EModE? Again some early work was carried out by Kytö and Rissanen (1993) who count the number of examples of Dem Poss in the Helsinki corpus and find the totals shown in Table 1: Table 1. Dem Poss in the Helsinki Corpus (from Kytö and Rissanen 1993: 255)

OE ME EModE

this

that

17 34 40

37 1 11

It appears from table 1 that even though there are fewer examples of Dem Poss in the ME period the expression has continuity. However, Allen (2004) breaks down the ME data into separate time periods and argues convincingly that Dem Poss is absent from English in the two early Helsinki Corpus periods, 1150–1250 and 1250–1350. Even when she extends Kytö and Rissanen’s Helsinki Corpus data to additional early ME texts, including texts of the genres that had Dem Poss in the OE period, she finds no convincing examples. It is only in the M3 Helsinki period, 1350–1420, that examples start to appear. It will also be noted from Table 1 that in ME there is a high frequency of the proximal (34 examples) in comparison with

S F O PRO



1st proofs

 Johanna L. Wood

the distal (1 example) as was observed in the preceding discussion of later periods (Section 2.1 above). It seems, then, that Dem Poss is present in OE Latin-influenced texts and both the proximal and distal demonstratives are used. The construction dies out in early ME only to be replaced in late ME by a Dem (proximal) Poss construction that has focus properties. On the question of whether English was once an AG language, as evidenced by Dem Poss, I conclude that, if at all, it is only likely to have been so in early OE. What is sometimes argued to be an AG type of construction, frequent in EModE and now on the decline, is, in fact, a focus construction restricted to the proximal demonstrative, which starts to emerge in the late 14th century and is still used in PDE. Additionally, referring back to the Italian examples in (2), we see that an article is a requirement in the AG construction, whereas in OE the demonstrative, the emerging article, is certainly not obligatory preceding a possessive. In terms of the structures, I suggest that it is possible for the AG construction to be used in OE, but more work needs to be done in this area, with respect to whether Dem Poss constructions are unambiguously AG or involve a pronoun as do (36)–(38). The new focus construction that appears in late ME does not have the AG structure but involves two nominals. .

Poss Dem in OE

In this Section, I discuss the order in which the possessive precedes the demonstrative as in (9)b, repeated here as (41): (41)

his þa æfestan tungan his that-- pious tongue--

(Bede 342.17)

In this discussion I emphasize that Poss Dem is not equivalent to, or mere variation in the word order of, Dem Poss. It is found in different text types from those that show Dem Poss and, as mentioned earlier, it only occurs with the distal demonstrative, not the proximal. It will be recalled from the previous Section that the proximal is the favoured demonstrative for the Dem Poss order, at least in later English. Additionally, there are two other differences to be observed between the types. As will be discussed in 3.1, Poss Dem does not occur without an adjective, and as will be discussed in 3.2, Poss Dem appears to be ungrammatical by late OE, while, as has been shown above, Dem Poss is found in OE, late ME, EModE and PDE. .

Poss Dem and adjectives

S F O PRO

The most important (and intriguing) point about the Poss Dem construction is that it almost certainly does not occur without an adjective. Heltveit’s (1977) investigation



1st proofs

Demonstratives and possessives 

of texts written before 1200, reports only 12 occurrences without an adjective (see Table 2) as opposed to 80 with an adjective; Mitchell (1985) and Traugott (1992) both report that Poss Dem is more frequent with an adjective. However, Allen’s investigation of OE texts reports 215 examples in OE of Poss Dem with an adjective and 0 without. There thus appears to be a discrepancy between her figures and Heltveit’s. However, my investigation of Heltveit’s 12 examples supports Allen. On closer inspection, these 12 examples are all examples of nominalized adjectives, with an elided noun, or an empty noun head, similar to PDE the poor. These examples will be discussed below. In Table 2 below I have tabulated the figures reported in narrative form by Heltveit (1977: 67–69). Table 2. Breakdown of Heltveit’s figures Text

Poss Dem Adj N

Poss Dem N

Chronicles Orosius Gosp Matth Gosp Luke Blickling Alfred CP Aelfric Hom Aelfric Saints I Aelfric Saints II

1 11 3 45 2 3 8 7

10

Total

80

12

Dem Poss Adj N

Dem Poss N

1 2

1 1 17 6 1

4

2 5

28

First, consider the two examples from Orosius, (ðin sio swiðre), which are close to each other in the text and adjacent to an example of Poss Dem Adj N (ðin sio winestre hond): (42)

gehæleme ðin sio swiðre. open you- that-- right-- Ne cwæð he no ðin sio winestre hond,  said he  you- that-- left- - hand- ac ðin sio swiðre, but you- that-- right-- ‘open your right. He didn’t say your left hand, but your right’

In this example, when the full context is considered, a strong case can be made that swiðre is not a noun but an adjective followed by the elided noun hond. Mitchell (1985: 52 Fn. 28) points out that the word hond occurs 7 times in the preceding 10 lines and again in the quote. Now consider the other 10 examples that Heltveit reports, all from the Blickling Homilies. I found four more examples in the text in addition to his 10, all four with

S F O PRO



1st proofs

 Johanna L. Wood

the same phrase, his þam halgum. The examples all contain nominalised adjectives followed by an elided noun. They all refer to animate (or inalienably possessed) items. In PDE these are the only adjectives that can be used without a noun as in the poor, the brave etc, when they always refer to a class of people defined by the adjective. Representative examples from the Blickling Homilies are in (43) and (44) below: (43)

min saul bletsaþ Drihten; my soul blesses Lord & ealle mine þa inneran his þone halgan naman (89.1) and all- my- that- inner his that--- holy name ‘my soul blesses the Lord and my whole spirit blesses his holy name’

(44)

ure Drihten his þæm halgum sægde (119.8) our Lord his that-- holy-- said ‘our Lord said to his holy people’

A slightly different example is shown in (45) below. (45)

Matheus min se leofa beheald on me (229.30) Matthew my that-- beloved look on me ‘Matthew my beloved look at me’.

The phrase min se leofa/leofsta appears to be fairly common in late West Saxon prose texts including Alexander’s letter to Aristotle and Ælfric’s Catholic Homilies. It will be discussed with reference to (48) below. The first major difference then, between Poss Dem and Dem Poss is that Poss Dem is never found without an adjective. According to Allen (2004: 14), the only two known examples without an adjective are from the late 11th and early 12th centuries “from a period when the construction was dying out” and could have been produced by a scribe who did not fully understand the construction but was trying to achieve an archaic style. This brings us to the other major difference between Poss Dem and Dem Poss, that Poss Dem is only found in OE. In order to further investigate the disappearance of Poss Dem from the language, I turn to one particular Old English text, Gregory’s Dialogues (GD), which, because of the history of the manuscripts, is a particularly useful text for analyzing syntactic change. .

Demonstratives and possessives in two versions of the Dialogues

In the ninth century, between about 870 and 890, Gregory’s Dialogues (GD), were translated from Latin into English by Bishop Wærferth of Worcester on King Ælfred’s command. Around 100 or 150 years later, an anonymous reviser, working with both the Latin original and Wærferth’s translation, produced a revised

S F O PRO



1st proofs

Demonstratives and possessives 

edition, systematically changing the spelling, vocabulary and syntax. The translation of all four books of the Dialogues survives in two manuscripts, Corpus Christi College, Cambridge 322 (C) and British Library Cotton Otho C.i Vol. 2, Fols. 1–137 (O), which were copied in the eleventh century. The revision survives in one manuscript, Bodleian Library, Hatton 76 fols 1–54 (H) which has three quarters of the revision of books I and II. Thus parts of (C) and (O) overlap with (H) (cf. Yerkes 1979 and 1982). Both the original and the revision must contain forms that were understandable to the listeners and readers of the time, but the differences between them give a unique insight into what variation is possible. These texts are particularly interesting in the examples of co-occurring demonstrative and possessive determiners. In the first row of Table 3 below, I show the number of occurrences of Poss Dem in the C text. In the second row, I show the number of examples that are found in both the C and H texts. Since the C and H texts do not overlap in their entirety, only 16 of the 56 examples found in the C text have equivalents in the incomplete revision, the H text, as the second row shows. As may be seen from the third row, there is only one example of Poss Dem in the revised text (H). The reviser has changed 15 of the co-occurrences in such a way that the possessive no longer precedes the demonstrative. The data have been sorted by person and number because OE 1st and 2nd person possessives are inflected for case whereas 3rd person possessives are not. As was mentioned above it is possible that Dem occurs only with third person Poss in order to provide required agreement features. However, it may be seen from the table that there is co-occurrence with all persons and numbers. The high number of 3rd person examples seems to be a result of 3rd person being more frequent in this particular text type. Table 3. Poss Dem in the Dialogues

Poss Dem in C C and H overlap Poss Dem in H

1sg

2sg

3sg

1pl

2pl

3pl

total

4 3 1

1 0 0

38 10 0

2 0 0

1 0 0

10 3 0

56 16 1

When making changes, the reviser’s most frequent strategy is to remove the demonstrative leaving only the possessive, which he does in 11 of the examples, as, for example, in (46) below: (46)

a. b.

þæt seo eorðe gehæfde his þone onfangenan lichaman his that-- received body þa geheold seo eorðe his underfanzenan lichaman (GD:H 155.9)

S F O PRO



1st proofs

 Johanna L. Wood

In one case he merely changes the order so that the possessive follows the demonstrative: (47)

a. b.

uneaðe Petrus, his seo gemæne speæc his that- false speech uneaðe Petrus seo his gemæne spræc

(GD: H 150.32)

In the remaining four examples, the construction is reworded; in three of them the possessive and the demonstrative are in different nominals. In only one example does the demonstrative still follow the possessive, as shown in (48) below: (48)

a.

b.

ða com to me min sunu Petrus 7 then came to me my-- son Peter and min se leofesta diacon my-- that-- dearest deacon þa com to me min se leofesta sunu Petrus diacon

(GD: C 3.26) (GD: H 3.25)

In (48)b, the coordination has been removed. In (48)a there is parallel structure. The two co-coordinated nouns, Petrus and diacon, are both introduced by the possessive min. In (48)b, there are two nouns in apposition with my having scope over the whole nominal: [my [the dearest son] [Peter deacon]]. Why might the reviser leave only this example of the Poss Dem construction? It has already been pointed out in reference to (45) that min se leofesta/leofa is a frequent phrase in OE, often used in the vocative, and could have remained in the language longer because it is a set phrase or formula. Similar phrases occur in Gothic: (49)

þu is sunus meins sa liuba you are son my the beloved

(Mark 1: 11)

These 16 examples, which were all changed by the reviser, support the hypothesis that Poss Dem is a construction required by the grammar of earlier OE, but which was ungrammatical in later OE, in this case mid 11th century. (Yerkes (1982: 9) dates the revision (H text) between 950 and 1050). However, we might wonder why Bishop Wærferth, when translating the Latin text into the English C text in the late 9th century, would use Poss Dem at all. One possibility might be that he was influenced by demonstratives in the Latin exemplar. It will be recalled from the previous Section (see Example (35)) that the Dem Poss order occurs in Latin influenced texts and often corresponds to a demonstrative in Latin. However, with Poss Dem there does not appear to be a Latin influence. According to Yerkes (1982: 20) who examines 10 examples in which the reviser has removed the demonstrative leaving the possessive alone, the Latin original never supports the Old English C text. The Latin original either has a possessive alone, as in (50)c, or no corresponding word.

S F O PRO



1st proofs

Demonstratives and possessives 

(50)

a. b. c.

min þæt ungesælige mod (GD: C 4.9) my- that-- unhappy spirit min ungesælige mod (GD: H 4.9) infelix. . . . animus meus (Latin)

These results correlate with observations made by Allen (to appear: 155). She points out that in the West Saxon Gospels, examples of Dem Poss always have a corresponding demonstrative in the Latin text whereas examples of Poss Dem do not.5 Since Bishop Wærferth was not following the Latin exemplar when using Poss Dem, it is likely that the co-occurrence is required by the grammar of earlier OE. I suggest that the demonstrative here is purely a definiteness marker, the emerging article, required because the possessive at this time is not definite.6 However the structure of Poss Dem is not that of the AG construction with the possessive merged lower in the structure, but with Poss in Spec-DP and Dem as head of DP. It appears that the Poss Dem construction is already ungrammatical by the time of the revision of GD, at the latest by the 11th century. This conclusion is supported by Allen’s data. She points out that in her investigation of texts that were probably composed in early ME, (the year 1100 or later) there are only 8 examples of Poss Dem, seven of them in a text that might have been composed earlier. The conclusion is that Poss Dem did not survive the OE period. In this Section I have argued that the Poss Dem construction is very different from the Dem Poss construction. First, neither Allen’s nor my investigation have found a convincing example without an adjective; second, as the investigation of GD and Allen’s search of early ME texts show, Poss Dem did not survive the OE period; thirdly, only examples of Poss Dem with the distal demonstrative have been found, and finally it does not occur only in Latin influenced texts but in native OE texts too. I have suggested that the construction has a demonstrative in form which functions as a definite article, structurally the head of DP. .

Dem N Poss in OE

The third order, (8)c above has not yet been discussed. Here the possessive follows the noun and the demonstrative precedes, as in (51) and (52) below. According to . GD also has examples of Dem Poss, not unexpected since it is a Latin translation. There are 26 examples of the order, Dem Poss in the early version of the Dialogues. The reviser also changes these, with only three examples remaining in the H text. . Why this structure is conditioned by an adjective is not at all clear. There appears to be a similarity with Scandinavian languages, where a definiteness marker in the form of a demonstrative has to be inserted in nominals only when adjectives are present.

S F O PRO



1st proofs

 Johanna L. Wood

Mitchell (1985: 53), this order is only found in one text, the Vespasian Psalter. He suggests that the construction is very restricted and is probably not a “normal” OE construction, unlike the patterns (8)a and (8)b which occur in a wide variety of texts. (51)

mont þone halgan his mountain that- saint his Latin: montem sanctum eius

(Ps (A) 2.6)

(52)

ðone halgan his that- saint his Latin: sanctum suum

(Ps (A) 4.4)

These examples appear to be an attempt to follow the Latin exemplar very closely, almost word for word, with a post nominal possessor as in Latin. However, in (51) and (52), it is interesting to note that even though the word order keeps to the Latin word order and the possessive follows the noun, there is no demonstrative in the Latin. Possibly the demonstrative was added in these cases because the possessive alone was not enough to indicate a definite nominal.

.

Conclusion

What evidence is there, then, to suppose that OE has the AG construction? One difference, it will be recalled, between AG and DG constructions is that in the DG construction the possessive is always definite; in the AG construction the possessive is “neutral” to definiteness and a definite article must be added for it to be definite. In OE, this would mean the addition of the demonstrative, the emerging article. In the preceding discussion there is evidence that the possessive is not definite in OE. As discussed in Section 3, with the Poss Dem order, consider why the original translator of GD would add a demonstrative following a possessive even when, as we have seen in (50), the Latin does not have a demonstrative. One explanation could be that in early OE the possessive is not definite so the translator adds a distal demonstrative, the emerging article as the head of DP to make definiteness visible. In terms of the structure in (10), the demonstrative would be head of DP and the possessive the specifier. In other words, the demonstrative has demonstrative form but is functioning as the article. In the later version of GD, 100 or 150 years later, the PDE situation is reached. The reviser removes the demonstratives since in his grammar the possessive is definite. In the Poss Dem construction we have additional evidence for the diachronic change of the demonstrative from specifier to head. A child, hearing the sequence Poss Dem, would conclude that the demonstrative is head of DP, even though it is

S F O PRO



1st proofs

Demonstratives and possessives 

still fully inflected for case, number and gender. The reanalysis as article precedes the loss of case and reduction in form. However the obligatory presence of the adjective has not been explained and remains for further research. Finally, more evidence that the possessive is not definite in early OE comes from the Dem N Poss examples in the Vespasian Psalter glosses discussed in Section 4. Here again, in the OE gloss the possessive alone is not enough to express the definiteness of the nominal and a demonstrative, the emerging article is in the head of DP. As for the other order, Dem Poss, it was established in Section 2 that Dem Poss need not represent an AG construction in ME and EModE even though the word order is similar to the word order of the Italian type. It was also established that an analysis of these constructions as apposition is possible throughout the history of English. A more detailed analysis is required here as it may well be that that there are two different Dem Poss constructions in OE. The first would be one with the Italian type structure that is indeed an AG construction with the distal demonstrative as head (emerging article) and the possessive in a lower AgrP. A second construction would be an appositive with either a distal or proximal demonstrative (a pronoun in traditional grammar terms). In late ME a third construction emerges, the focus construction with the proximal. The first two constructions would be expected to decline in frequency from OE on and the last one to start to rise in frequency. Further research will establish whether this scenario is correct.

Primary Sources Hecht, Hans. 1965. Bischofs Wærferths von Worcester Übersetzung der Dialoge Gregors des Grossen. Darmstadt: Wissenschaftliche Buchgesellschaft. Morris, Richard. 1967 (1874–1880). The Blickling Homilies. EETS 58, 63. London: Trübner.

References Abney, Steven. 1987. The English noun phrase in its sentential aspect. PhD diss, MIT. Ackles, Nancy. 1997. Historical syntax of the English articles in relation to the count/non-count distinction. PhD diss, University of Washington. Allen, Cynthia L. (in press). Possessives and determiners in Old English. Types of variation: diachronic, dialectal and typological interfaces, Terttu Nevalainen, Juhani Klemola and Mikko Laitenen (eds), 149–170. Amsterdam: John Benjamins. Allen, Cynthia, L. (2004). These our letters: The dem poss construction from old to early modern English. An international master of syntax and semantics. Papers presented to Aimo Seppänen on the occasion of his 75th birthday, Gunnar Bergh, Jennifer Herriman & Mats Mobärg (eds), 11–19. Gothenburg: Acta Universitatis Gothoburgensis.

S F O PRO



1st proofs

 Johanna L. Wood Alexiadou, Alexis. 2004. On the development of possessive determiners: Consequences for DP structure. Diachronic clues to synchronic grammar. Eric Fuß & Carola Trips (eds), 31–58. Amsterdam: John Benjamins. [Linguistik Aktuell/Linguistics Today 72]. Arteaga, Deborah. 1995. Strong and weak determiners in Old French. Language Quarterly 33(1–2): 67–80. Cardinaletti, Anna. 1998. On the deficient/strong opposition in possessive systems. Possessors, predicates and movement in the determiner phrase. Alexis Alexiadou & Chris Wilder (eds), 17–53. Amsterdam: John Benjamins. Delsing, Lars-Olof. 1998. Possession in Germanic. Possessors, predicates and movement in the determiner phrase. Alexis Alexiadou and Chris Wilder (eds), 87–108. Amsterdam: John Benjamins: Amsterdam. Demske, Ulrike. 2001. Merkmale and Relationen: Diachrone Studien zur Nominalphrase des Deutschen. Berlin: Walter de Gruyter. Denison, David. 1998. Syntax. The Cambridge history of the English language. (Vol. IV), Suzanne Romaine (ed.), 92–326. Cambridge: Cambridge University Press. Dixon, Robert M.W. 2003. Demonstratives: A cross-linguistic typology. Studies in Language 27(1): 61–112. Gelderen, Elly van. 2004. Grammaticalization as economy. Amsterdam: John Benjamins. Giorgi, Alessandra & Giuseppe Longobardi. 1991. The syntax of noun phrases. Cambridge: Cambridge University Press. Haspelmath, Martin. 1999. Explaining article-possessor complementarity: Economic motivation in noun phrase syntax. Language 75.2: 227–243. Heine, Bernd & Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge University Press. Heltveit, Trygve. 1977. Aspects of the syntax of quantifiers in Old English. Norwegian Journal of Linguistics 31: 47–94. Julien, Marit. 2005. Nominal phrases from a Scandinavian perspective. Amsterdam: John Benjamins. Kytö, Merja & Matti Rissanen. 1993. ‘By and by enters [this] my artificiall foole. . . who, when Jack beheld, sodainely he flew at him’: Searching for syntactic constructions in the Helsinki Corpus. Early English in the computer age: Explorations through the Helsinki corpus. Matti Rissanen, Merja Kytö & Minna Palander (eds), 253–266. Berlin: Mouton de Gruyter. Lyons, Christopher. 1986. The syntax of English genitive constructions. Journal of Linguistics 22: 123–143. Lyons, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press. Milsark, Gary. 1974. Existential sentences in English. Ph.D. MIT. Published 1979 by Garland Press, New York. Mitchell, Bruce. 1985. Old English syntax : Concord, the parts of speech and the sentence (Vol. I). Oxford: Oxford University Press. Osawa, Fuyo. 2000. The historical emergence of DP in English. English Linguistics 17: 51–79. Plank, Frans. 1992. Possessives and the distinction between determiners and modifiers (with special reference to German). Journal of Linguistics 28: 453–468. Rissanen, Matti. 1999. Syntax. The Cambridge history of the English language. (Vol. III), Roger Lass (ed.), 187–326. Cambridge: Cambridge University Press. Radford, Andrew. 2000. NP Shells. Essex research reports on linguistics 23: 2–20. Roberts, Ian. G. & Anna Roussou. 2003. Syntactic change: A minimalist approach to grammaticalization. Cambridge: Cambridge University Press. Rosenbach, Annette. 2002. Genitive variation in English. Berlin: Mouton.

S F O PRO



1st proofs

Demonstratives and possessives  Schoorlemmer, Maike. 1998. Possessors, articles and definiteness. Possessors, predicates and movement in the determiner phrase. Alexis Alexiadou & Chris Wilder (eds), 55–86. Amsterdam: John Benjamins. Traugott, Elizabeth C. 1992. Syntax. The Cambridge history of the English language. (Vol. I), Richard M. Hogg (ed.), 168–289. Cambridge: Cambridge University Press. Wood, Johanna L. 2003. Definiteness and number: Determiner phrase and number phrase in the history of English. PhD Arizona State University. Wood, Johanna L. (to appear). Is there a DP in Old English? Proceedings of the XVIIth international conference on historical linguistics. Madison, WI. Yamamoto, Keiko. 1989. The historical development of determiners: A parametric approach. English Linguistics 6: 1–17. Yerkes, David. 1979. The two versions of Wærferth’s translation of Gregory’s dialogues: An Old English thesaurus. Toronto: University of Toronto Press. Yerkes, David. 1982. Syntax and style in Old English: A comparison of the two versions of Wærferth’s translation of Gregory’s dialogues. Binghampton, New York: Center for medieval and early renaissance studies.

S F O PRO



1st proofs

S F O PRO



1st proofs